Iif - KGPM - Geeraerts D. Words and Other Wonders

Words and Other Wonders
Cognitive Linguistics Research

33
Editors
Dirk Geeraerts
Rene Dirven
John R. Taylor
Honorary editor
Ronald W. Langacker
Mouton de Gruyter
Berlin New York
Words and Other Wonders
Papers on Lexical and Semantic Topics
by
Dirk Geeraerts
Mouton de Gruyter
Berlin New York
Mouton de Gruyter (formerly Mouton, The Hague)
is a Division of Walter de Gruyter GmbH & Co. KG, Berlin
Printed on acid-free paper
which falls within
the guidelines of the ANSI
to ensure permanence and durability.
Library of Congress Cataloging-in-Publication Data
Geeraerts, Dirk, 1955
Words and other wonders : papers on lexical and semantic topics /
by Dirk Geeraerts.
p. cm. (Cognitive linguistics research ; 33)
Includes bibliographical references.
ISBN-13: 978-3-11-019042-7 (hardcover : alk. paper)
ISBN-10: 3-11-019042-7 (hardcover : alk. paper)
1. Semantics. 2. Lexicology. I. Title.
P325.G43 2006
4011.43dc22
2006018773
Bibliographic information published by Die Deutsche Bibliothek
Die Deutsche Bibliothek lists this publication in the Deutsche Nationalbibliografie;
detailed bibliographic data is available in the Internet at http://dnb.ddb.de.
ISBN-13: 978-3-11-019042-7
ISBN-10: 3-11-019042-7
ISSN 1861-4132
Copyright 2006 by Walter de Gruyter GmbH & Co. KG, D-10785 Berlin
All rights reserved, including those of translation into foreign languages. No part of this book
may be reproduced or transmitted in any form or by any means, electronic or mechanical,
including photocopy, recording, or any information storage and retrieval system, without
permission in writing from the publisher.
Printed in Germany
Preface
Granunar and lexicon being traditionally separated in mainstream twentieth-
century linguistics, one of the great things about Cognitive Linguistics is the
opening it has created towards a renewed study of the lexicon and an inte-
gration of lexical semantics into theoretical granunar at large. In fact, the
study of the lexicon plays a crucial role in Cognitive Linguistics: questions
about conceptual structure and semantic mechanisms that lie at the heart of
Cognitive Linguistics, belong to the time-honored expertise of lexical seman-
tics. In the present collection, I bring together a representative sample of my
own contributions to lexicology and lexical semantics within the framework
of Cognitive Linguistics (including the extension of insights from lexical
semantics to other areas of linguistics).
In the fifteen years covered by this collection, from 1988 to 2003, the
central topics of my research led to a number of monographs: Diachronic
Prototype Semantics (1997) synthesizes the research into historical semasi-
ology that dominated much of my work in the 1980s, and two co-authored
monographs, The Structure of Lexical Variation (1994) and Convergentie
en Divergentie in de Nederlandse Woordenschat (1999) embody the cor-
pus-based studies of onomasiological variation and change that I developed
with Dirk Speelman, Stefan Grondelaers and the other members of my Leu-
ven research group in the 1990s, and that continue to dominate our present
research efforts. While these two central fields of research are only indirectly
represented in the present volume, this collection predominantly brings to-
gether a number of papers devoted to lexicological and semantic topics that I
dealt with more sporadically or in parallel to these major foci.
Thematically speaking, I have consistently tried to choose papers that
represent a variety of issues and topics within the broadly defined field of
lexicology and lexical semantics. Accordingly, the chapters are organized in
thematic groups. The first section deals with prototypicality as a theoretical
and practical model of semantic description. The second section discusses
polysemy and criteria for distinguishing between meanings. The third section
tackles questions of meaning description beyond the level of words, on the
level of idioms and constructions. The following section casts the net even
wider, dealing with the cultural aspects of meaning. Moving away from the
theoretical and descriptive perspective towards applied concerns, the fifth
section looks at lexicography from the point of view of Cognitive Linguis-
VI Preface
tics. The final section has a metatheoretical orientation: it discusses the his-
tory and methodology of lexical semantics.
Theoretically, the papers are interconnected by a number of accents that
mark their specific position within the varied landscape of Cognitive Lin-
guistics: a continued attempt to provide solid empirical foundations for
semantics, an emphasis on the multivariate nature of semantic structure, and
an insistence on the social and cultural background of language. In the
research goals that we are currently pursuing in Leuven, these accents com-
bine in multifactorial corpus-based studies in the field of cognitive sociolin-
guistics - studies, that is, in which corpus materials provide the firm empiri-
cal basis for research, in which linguistic phenomena are statistically
analysed with the help of multivariate techniques, and in which social and
cultural variation is explicitly included into the multifactorial model.
Except for necessary corrections, small additions, and some typographi-
cal reorganization (and unless otherwise stated), the chapters are reprinted in
their original form. All chapters are reprinted with permission of the original
publishers. Each paper is preceded by a brief introduction that situates the
text against the period in which it was first written, but that also points to
further developments, in my own research or in Cognitive Linguistics at
large. In this way, the collection should give the reader an idea of the dyna-
mism of Cognitive Linguistics in the crucial period when it developed from a
more or less marginal approach to one that is firmly situated in the main-
stream of contemporary linguistics.
Knowledge is not produced in isolation, and it befits any researcher to
thank those who have provided guidance and motivation in the course of his
career. In gratitude, I dedicate this collection to the memory of Brygida
Rudzka-Ostyn (1939-1998). Brygida's impact on the international expan-
sion of Cognitive Linguistics can hardly be underestimated. With her infinite
enthusiasm and her sense for initiative, she played a crucial rol in the second
half of the 1980s in bringing Cognitive Linguistics from its Californian
homeground to Europe and beyond. Without the warm stimulus she then
provided for a young scholar to place his work in a wider context, none of
the papers collected here might ever have been written.
Table of contents
Preface
Publication sources
Section 1. Prototypicality and salience
1 Prospects and problems of prototype theory
2 Where does prototypicality come from?
3 The semantic structure of Dutch over
4 Salience phenomena in the lexicon. A typology
Section 2. Polysemy
5 Vagueness's puzzles, polysemy's vagaries
6 Classical definability and the monosemic bias
Section 3. Constructions and idioms
V
IX
3
27
48
74
99
149
7
8
The semantic structure of the indirect object in Dutch
The interaction of metaphor and metonymy in
composite expressions
175
198
Section 4. Meaning and culture
9 Looking back at anger. Cultural traditions and
metaphorical patterns 227
10 Beer and semantics 252
11 Cultural models of linguistic standardization 272
12 Caught in a web of irony: Job and his embarassed God 307
Section 5. Lexicography
13 The lexicographical treatment of prototypical polysemy 327
14 The definitional practice of dictionaries and the
cognitive semantic conception of polysemy 345
Section 6. Theory and method in lexical semantics
15 Cognitive grammar and the history oflexical semantics 367
16 The theoretical and descriptive development of lexical
semantics 398
17 Idealist and empiricist tendencies in cognitive semantics 416
Vlll Table ofcontents
References
Index of names
Index of subjects
445
479
485
Publication sources
Chapter 1 - Geeraerts, Dirk. 1989. Prospects and problems of prototype theory.
linguistics 27: 587-612.
Chapter 2 - Geeraerts, Dirk. 1988. Where does prototypicality come from? In
Topics in Cognitive Linguistics, Brygida Rudzka-Ostyn (ed.), 207-229. Amster-
damJPhiladelphia: John Benjamins.
Chapter 3 - Geeraerts, Dirk. 1992. The semantic structure of Dutch over. Leu-
vense Bijdragen 81: 205-230.
Chapter 4 - Geeraerts, Dirk. 2000. Salience phenomena in the lexicon. A typol-
ogy. In Meaning and Cognition, Liliana Albertazzi (ed.), 125-136. Amster-
Chapter 5 - Geeraerts, Dirk. 1993. Vagueness's puzzles, polysemy's vagaries.
Cognitive Linguistics 4: 223-272.
Chapter 6 - Geeraerts, Dirk. 1994. Classical definability and the monosernic bias.
Rivista di Linguistica 6: 189-207.
Chapter 7 - Geeraerts, Dirk. 1998. The semantic structure of the indirect object in
Dutch. In The Dative 11. Theoretical Approaches, Willy Van Langendonck and
Wiliam Van Belle (eds.), 185-210. AmsterdamJPhiladelphia: John Benjamins.
Chapter 8 - Geeraerts, Dirk. 2002. The interaction of metaphor and metonymy in
composite expressions. In Metaphor and Metonymy in Comparison and Contrast,
Rene Dirven and Ralf Parings (eds.), 435-465. BerlinINew York: Mouton de
Gruyter.
Chapter 9 - Geeraerts, Dirk, and Stefan Grondelaers. 1995. Looking back at an-
ger. Cultural traditions and metaphorical patterns. In Language and the Construal
of the World, John Taylor and Robert E. MacLaury (eds.), 153-180. BerlinINew
York: Mouton de Gruyter.
Chapter 10 - Geeraerts, Dirk. 1999. Beer and semantics. In Issues in Cognitive
Linguistics, Leon De Stadler and Christoph Eyrich (eds.), 35-55. BerlinINew
Chapter 11 - Geeraerts, Dirk. 2003. Cultural models of linguistic standardization.
In Cognitive lvfodels in Language and Thought. Ideology, Metaphors and Mean-
ings, Rene Dirven, Roslyn Frank, and Martin Piitz (eds.), 25-68. BerlinINew
x Publication sources
Chapter 12 - Geeraerts, Dirk. 2003. Caught in a web of irony: Job and his emba-
rassed God. In Job 28. Cognition in Context, Ellen Van Wolde (ed.), 37-55. Lei-
den: Brill.
Chapter 13 - Geeraerts, Dirk. 1990. The lexicographical treatment of prototypical
polysemy. In Meanings and Prototypes. Studies in Linguistic Categorization,
Savas L. Tsohatzidis (ed.), 195-210. London: Routledge.
Chapter 14 - Geeraerts, Dirk. 2001. The definitional practice of dictionaries and
the cognitive semantic conception of polysemy. Lexicographica 17: 6-21.
Chapter 15 - Geeraerts, Dirk. 1988. Cognitive grammar and the history oflexical
semantics. In Topics in Cognitive Linguistics, Brygida Rudzka-Ostyn (ed.), 647-
677. Amsterdam/Philadelphia: John Benjamins.
Chapter 16 - Geeraerts, Dirk. 2002. The theoretical and descriptive development
of lexical semantics. In The Lexicon in Focus. Competition and Convergence in
Current Lexicology, Leila Behrens and Dietmar Zaefferer (eds.), 23-42. Frank-
furt: Peter Lang Verlag.
Chapter 17 - Geeraerts, Dirk. 1999. Idealist and empiricist tendencies in cognitive
linguistics. In Cognitive Linguistics: Foundations, Scope, and Methodology, Theo
Janssen and Gisela Redeker (eds.), 163-194. BerlinINew York: Mouton de
Gruyter.
Section 1
Prototypicality and salience
Chapter 1
Prospects and problems of prototype theory
Originally published in Linguistics 1989,27: 587-612.
Parallel to the research that I did on the application of prototype theory to ques-
tions of semantic change (which eventually resulted in my Diachronic Prototype
Semantics monograph of 1997), I paid attention on a number of occasions to the
theoretical clarification of the notion of prototypicality as such (and of the related
notion of polysemy: see the second section of this collection). In the paper re-
printed here, I try to bring some systematicity into the many uses of the concept
'prototypicality' by distinguishing between two cross-classifying dimensions.
First, I make a distinction between two crucial structural phenomena underlying
prototypicality effects: flexibility (i.e. the absence of clear boundaries and demar-
cations) and salience (i.e. differences of structural weight). Second, I suggest that
both prototypicality phenomena may be found on an intensional level (the level of
definitions) and on an extensional level (the level of referents). The cross-
classification of the two dimensions defines four basic types of prototypicality
effects.
The chapter originally appeared as the introductory paper of a thematic issue of
the journal Linguistics. A section of the original paper describing the various
contributions to the thematic issue has been omitted from the present reprint.
1. Prototype theory within linguistics
The starting-point of the prototypical conception of categorial structure is
sununarized in the statement that
when describing categories analytically, most traditions of thought have
treated category membership as a digital, all-or-none phenomenon. That is,
much work in philosophy, psychology, linguistics, and anthropology as-
sumes that categories are logical bounded entities, membership in which is
defined by an item's possession of a simple set of criterial features, in
which all instances possessing the criterial attributes have a full and equal
degree of membership. In contrast, it has recently been argued ... that some
natural categories are analog and must be represented logically in a man-
ner which reflects their analog structure (Roscll and Mervis 1975: 573-
574).
4 Prototypicality and salience
As we shall see in section 2, the exact definition of the concept of prototypi-
cality as used in linguistics is not without problems. The major part of this
introduction to the prototypicality-based studies collected here will, in fact,
consist of an attempt at clarification of some of the problematic aspects of
the way in which the notion of prototype has been used in linguistics. To
begin with, however, we shall be concerned with a brief overview of the state
of the art in linguistic prototype theory. I
The theory originated in the mid 1970s with Eleanor Rosch' s research
into the internal structure of categories. (Overviews may be found in Rosch
1978, 1988, and Mervis and Rosch 1981; the basic research is reported on
mainly in Heider 1972, Rosch 1973, 1975, 1977, Rosch and Mervis 1975,
Rosch, Simpson and Miller 1976, Rosch et al. 1976.) From its psycholin-
guistic origins, prototype theory has moved mainll in two directions. On the
one hand, Rosch's findings and proposals were taken up by formal psy-
cholexicology (and more generally, information-processing psychology),
which tries to devise formal models for human conceptual memory and its
operation, and which so, obviously, borders on Artificial Intelligence. Excel-
lent overviews of the representational and experimental issues at stake here
are Smith and Medin (1981), and Medin and Smith (1984); an interesting
sample of current research may be found in Neisser (1987). On the other
hand, prototype theory has had a steadily growing success in linguistics
since the early 1980s, as witnessed by a number of recent monographs and
collective volumes in which prototype theory and its cognitive extensions
play a major role (Wierzbicka 1985, Lakoff 1987, Langacker 1987, Craig
1986, Holland and Quinn 1987, Rudzka-Ostyn 1988, Lehmann 1988a,
Hullen and Schulze 1988, Tsohatzidis 1989, Taylor 1989). It is with the
latter development that we shall be concerned with here.
Against the background of the development of linguistic semantics, pro-
totype theory may be defined primarily in contrast with the componential
model of semantic analysis that was current in transformational grammar
and that is stereotypically associated with Katz and Fodor's analysis of
bachelor (Katz and Fodor 1963); in an early defense of a prototypical ap-
proach, Fillmore (1975) called this the 'checklist theory' of meaning. The
prototypists' reaction against this featural approach had, however, the nega-
tive side-effect of creating the impression that prototypical theories rejected
any kind of componential analysis. This is a misconception for the simple
reason that there can be no semantic description without some sort of de-
compositional analysis. As a heuristic tool for the description and compari-
son of lexical meanings, a componential analysis retains its value (a value
Prospects and problems ofproto(vpe theory 5
that, incidentally, it did not acquire with the advent of componential analysis
as an explicit semantic theory, but which had been obvious to lexicographers
from time immemorial). Rather, the difficulties with the neostructuralist kind
of feature analysis that grew out of structuralist field theory lie elsewhere; it
is not the use of decomposition as a descriptive instrument that causes con-
cern, but the status attributed to the featural analysis. Two important points
have to be mentioned.
In the first place, as suggested by the quotation at the beginning of this
introduction, featural definitions are classically thought of as criterial, i.e. as
listing attributes that are each indispensable for the definition of the concept
in question, and that taken together suffice to delimit that concept from all
others. In contrast, prototype theory claims that there need not be a single set
of defining attributes that conform to the necessity-cum-sufficiency require-
ment.
3
In the second place, prototype theory is reluctant to accept the idea that
there is an autonomous semantic structure in natural languages which can be
studied in its own right, in isolation from the other cognitive capacities of
man. In particular, meaning phenomena in natural languages cannot be stud-
ied in isolation from the encyclopedic knowledge individuals possess; it is
precisely the presupposition that there exists a purely linguistic structure of
semantic oppositions that enables structuralist and neostructuralist semantics
to posit the existence of a distinction between semantic and encyclopedic
knowledge. Prototype theory tends to minimize the distinction primarily for
methodological reasons: because linguistic categorization is a cognitive phe-
nomenon just like the other cognitive capacities of man, it is important to
study it in its relationship to these other capacities. More specific arguments
have also been formulated to show that the distinction between an encyclo-
pedic and a semantic level of categorial structure is untenable.
4
For instance,
given that the flexible extendibility of prototypical concepts is a synchronic
characteristic of linguistic structure, and given the fact that these extensions
may be based indiscriminately on allegedly encyclopedic or on allegedly
semantic features, the distinction between both kinds of information loses its
synchronic relevance. Take the case of metaphor: before lion acquires the
meaning 'brave man', the feature 'brave' is not structurally distinctive
within the semasiological structure of lion, and hence, it has to be considered
encyclopedic according to structuralist theories. But if it can be accepted
(and this is of course the crucial point) that the metaphorical extension of
lion towards the concept 'brave man' is not just a question of diachronic
change, but is merely an effect of the synchronic flexibility of lexical items,
the feature clearly acquires semantic status. If, furthermore, the argument
can be repeated in the sense that such synchronic metaphorical extensions
may be based on any allegedly encyclopedic attribute, the distinction be-
tween semantic and encyclopedic concepts as a whole falls.
5
The matter need not, to be sure, be settled here. What is important for our
introductory purposes is rather to see what exactly prototype theory objects
to in componential theories of the Katzian type. First, the suggestion that
lexical concepts are criterial in the classical sense, and second, the sugges-
tion that there exists a purely linguistic level of conceptual structuring that is
neatly separated from other, 'encyclopedic' forms of conceptual information,
and that may thus be studied autonomously, in methodological isolation from
other kinds of cognitive research. As against these points of view, prototype
theory defends a non-criterial conception of categorial structure, and an in-
terdisciplinary methodological perspective that takes into account relevant
research from the other cognitive sciences. (The very transposition of the
prototypical approach from experimental psychology to linguistics derives
from this attitude.)
But this historical positioning of prototype theory with regard to its im-
mediate predecessors within the field of lexical semantics clearly does not
explain why it has turned out to be such a successful alternative. Why did
(and does) the prototypical approach appeal to a sizeable part of the linguis-
tic community? On the one hand, the historical development of generative
grammar had raised a considerable amount of interest in semantic matters. It
should not be forgotten, in fact, that it was only after the incorporation of a
semantic component into the transformational framework that Chomskyan-
ism became internationally populac the universal appeal of the generative
Standard Theory was at least partly due to the promises held by its Katzian
semantic component. On the other hand, the promises were not fulfilled.
Within the generative paradigm, Generative Semantics (which most strongly
embodied the semantic approach) withered in favor of Autonomous Syntax,
in which semantics hardly played a role worthy of note. Outside the genera-
tive approach, formal semantics of the Montagovian kind was too narrowly
restricted to sentential meaning to be able to hold the attention of those who
were interested primarily in the internal structure of natural language catego-
ries (and not primarily in the way these categories combine into larger uni-
ties).6 In short, as far as semantics was concerned, there was a gap in the
linguistic market of the early 1980s that was not filled by the major ap-
proaches of the day.
7
Prospects and problems ofprototype theory 7
But again, recognizing that there was an interest in the semantics of natu-
ral language categories to which prototype theory could appeal does not tell
the whole story. Why didn't people simply stick to the componential theory
popularized by Katz, or to the rival axiomatic method of representation -
even if these gradually moved out of the centre of the linguistic attention as
Autonomous Syntax and Formal Semantics took over? In general, there are
a number of methodological requirements people nowadays expect of lin-
guistic theories: descriptive adequacy (mainly in the form of a broad empiri-
cal scope), explanatory depth, productivity, and formalization. Although
prototype theory rates much lower on the formalization scale than either the
axiomatic or the featural approach, its assets with regard to the other three
points are considerable.
In the first place, it tackles a number of semantic phenomena that had
been swept under the rug by the more structurally minded approaches. The
fuzzy boundaries of lexical categories, the existence of typicality scales for
the members of a category, the flexible and dynamic nature of word mean-
ings, the importance of metaphor and metonymy as the basis of that flexibil-
ity - these are all intuitively obvious elements of the subject matter of se-
mantics that were largely neglected by structural semantics. It is true that
they were occasionally pointed at as an indispensable aspect of any full-
fledged semantic theory: think, for instance, of Weinreich's remark (1966:
471) that a semantic theory should be able to deal with 'interpretable devi-
ance', or Uhlenbeck's plea (1967) for a dynamic conception of word mean-
ing.
8
These remarks did not, however, have much effect as far as theory
formation was concerned. In particular, it is only with the advent of proto-
type theory that contemporary linguistics developed a valid model for the
polysemy of lexical items. This is perhaps the single most appealing charac-
teristic of prototype theory: here at last is a descriptive approach to lexical
meaning in which our pretheoretical intuitions about gradedness, fuzziness,
flexibility, clustering of senses etc. receive due attention.
In the second place, prototype theory appears to be a productive theory
not just in the sense that its insights into the structure of lexical categories
can be easily applied in various fields of the lexicon, but also in the sense
that it may be extended towards other aspects of linguistics. Whereas proto-
type theory started with being descriptively fruitful in lexical semantics, it
soon became theoretically fruitful in the sense that other areas of linguistics
were taken into consideration. A few recent examples of such extensions
may suffice: phonology (Nathan 1986), morphology (Bybee and Moder
1983, Post 1986), syntax (Van Oosten 1986, Ross 1987), historicallinguis-
tics (Winters 1987, Aijmer 1985), markedness theory (Van Langendonck
1986), theoretical lexicography (Geeraerts 1985b). Through these and simi-
lar extensions,9 prototype theory has become one of the cornerstones of
Cognitive Linguistics, which tries to account for the interaction between
language and cognition on all levels of linguistic structure: one need only
have a look at the prominent place attributed to a prototypical conception of
categorial structure in Langacker (1987) (one of the basic works of the Cog-
nitive Linguistic approach) to appreciate its importance. IQ In this sense, the
development of prototype theory into Cognitive Linguistics contains exciting
promises of a unified cognitive theory of linguistic categorization.
In the third place, the explanatory depth of prototype theory resides
partly in its generalizable character, but also in its interdisciplinary nature.
The importance of its genetic link with psycholinguistics can only be fully
appreciated against the background of the Chomskyan requirements with
regard to theories of grammar. Chomsky' s methodology is, in fact, in the
awkward position of declaring linguistics a cognitive science, but refusing to
deal directly with the findings of the other sciences of the mind. Roughly
stated, Chomskyan linguistics claims to reveal something about the mind,
but imperviously prefers a strictly autonomist methodology over the open
dialogue with psychology that would seem to be implied by such a claim.
Prototype theory's linguistic application of psycholinguistic findings, on the
other hand, takes the Chomskyan ideal of cognitive explanatory depth to its
natural consequences, viz. of giving up the methodological autonomy of
linguistics in favor of an interdisciplinary dialogue with the other cognitive
sciences.
l1
Prototype theory takes the cognitive claims of Chomskyanism
methodologically seriously by its interdisciplinary openness. This is all the
more important at a moment when Cognitive Science is emerging as an in-
terdisciplinary cluster of psychology, neuroscience, Artificial Intelligence,
and philosophy. It is probably one of the reasons for the appeal of prototype
theory that its interdisciplinary connections hold the promise of linking lin-
guistics to the most important development that the human sciences are cur-
rently witnessing.
2. Definitional problems, first series: 'Prototype' as a prototypical
notion
The appeal of prototype theory should not, however, obscure the fact that
the exact definition of prototypicality is not without problems. The purpose
of this section (and the following) is to analyze the sources of the confusion
by making clear that prototypicality is itself, in the words of Posner (1986),
a prototypical concept. As a first step, we shall have a look at four charac-
teristics that are frequently mentioned (in various combinations) as typical of
prototypicality. In each case, a quotation from early prototype studies is
added to illustrate the point.
(i) Prototypical categories cannot be defined by means of a single set of cri-
terial (necessary and sufficient) attributes:
We have argued that many words ... have as their meanings not a list of
necessary and sufficient conditions that a thing or event must satisfy to
count as a member of the category denoted by the word, but rather a psy-
chological object or process which we have called a prototype (Coleman
and Kay 1981: 43).
(ii) Prototypical categories exhibit a family resemblance structure, or more
generally, their semantic structure takes the form of a radial set of clustered
and overlapping meanings: 12
The purpose of the present research was to explore one of the major struc-
tural principles which, we believe, may govern the fonnation of the proto-
type structure of semantic categories. This principle was first suggested in
philosophy; Wittgenstein (1953) argued that the referents of a word need
not have common elements to be understood and used in the nonnal func-
tioning of language. He suggested that, rather, a family resemblance might
be what linked the various referents of a word. A family resemblance rela-
tionship takes the form AB, BC, CD, DE. That is, each item has at least
one, and probably several, elements in common with one or more items,
but no, or few, elements are common to all items (Rosch and Mervis 1975:
574-575).
(iii) Prototypical categories exhibit degrees of category membership; not
every member is equally representative for a category:
By prototypes of categories we have generally meant the clearest cases of
category membership defined operationally by people's judgments of good-
ness of membership in the category ... we can judge how clear a case some-
thing is and deal with categories on the basis of clear cases in the total ab-
sence of information about boundaries (Rosch 1978: 36).
(iv) Prototypical categories are blurred at the edges:
New trends in categorization research have brought into investigation and
debate some of the major issues in conception and learning whose solution
10 Proto(vpicality and salience
had been unquestioned in earlier approaches. Empirical findings have es-
tablished that ... category boundaries are not necessarily definite (Mervis
and Rosch 1981: 109).
As a first remark with regard to these characteristics, it should be noted
that they are not the only ones that may be used in attempts to define the
prototypical conception of categorization. Two classes of such additional
features should be mentioned.
On the one hand, there are characteristics that do not pertain (as the four
mentioned above) to the structure of categories, but that rather pertain to the
epistemological features of so-called non-Aristotelian categories.
13
For in-
stance, the view that prototypical categories are not 'objectivist' but 'experi-
ential' in nature (Lakoff 1987) envisages the epistemological relationship
between concepts and the world rather than the structural characteristics of
those concepts. In particular, it contrasts the allegedly classical view that
'categories of mind ... are simply reflections of categories that supposedly
exist objectively in the world, independent of all beings', with the view that
'both categories of mind and human reason depend upon experiential aspects
of human psychology' (Lakoff 1982: 99). Such an epistemological rather
than structural characterization of natural concepts also has a methodologi-
cal aspect to it; it entails that prototypical categories should not be studied in
isolation from their experiential context. While such an epistemological or
methodological conception of prototypical categorization is extremely valu-
able, we shall take a structural point of view in the following pages; we shall
try to determine whether it is possible to give a coherent, structurally intrin-
sic characterization of prototypical categories.
On the other hand, there are structural characteristics of prototypical
concepts that can be reduced to the four basic structural features mentioned
above. For instance, in my own work on prototypical categorization, I have
repeatedly stressed the flexibility of prototypical concepts (1983a, 1985a),
together with the fact that a distinction between semantic and encyclopedic
components of lexical concepts cannot be maintained in the case of proto-
typical concepts (l985b). But the flexibility of prototypical categories is
linked in a straightforward manner with the fourth characteristic: uncertain-
ties with regard to the denotational boundaries of a category imply that it
need not be used in a rigidly fixed manner. Similarly, the absence of a clear
dividing line between encyclopedic and purely semantic information follows
from this very flexibility together with the first and second characteristic. As
illustrated in the previous section, the possibility of incorporating members
into the category that do not correspond in every definitional respect with the
eXlstmg members entails that features that are encyclopedic (non-
definitional) with regard to a given set of category members may turn into
defmitional features with regard to a flexibly incorporated peripheral cate-
gory member. The resemblance between central and peripheral cases may be
based on allegedly encyclopedic just as well as on allegedly 'semantic' fea-
tures. In short, features of prototypicality that are not included among the
ones mentioned in (i)-(iv) may often be reduced to those four, and this in
turn justifies a preliminary restriction of the discussion to the latter.
A second remark with regard to the four characteristics is concerned with
the fact that they are systematically related along two dimensions. On the
one hand, the third and the fourth characteristic take into account the refer-
ential, extensional structure of a category. In particular, they have a look at
the members of a category; they observe, respectively, that not all referents
of a category are equal in representativeness for that category, and that the
denotational boundaries of a category are not always determinate. On the
other hand, these two aspects (centrality and non-rigidity) recur on the inten-
sional level, where the definitional rather than the referential structure of a
category is envisaged. For one thing, non-rigidity shows up in the fact that
there is no single necessary and sufficient definition for a prototypical con-
cept. For another, family resemblances imply overlapping of the subsets of a
category. To take up the formulation used in the quotation under (ii) above,
if there is no definition adequately describing A, B, C, D, and E, each of the
subsets AB, BC, CD, and DE can be defined separately, but obviously, the
'meanings' that are so distinguished overlap. Consequently, meanings exhib-
iting a greater degree of overlapping (in the example: the senses correspond-
ing with BC and CD) will have more structural weight than meanings that
cover peripheral members of the category only. In short, the clustering of
meanings that is typical of family resemblances implies that not every mean-
ing is structurally equally important (and a similar observation can be made
with regard to the components into which those meanings may be analyzed).
The systematic links between the characteristics mentioned at the beginning
are schematically summarized in Table 1.
As a third remark, it should be noted that the four characteristics are of-
ten thought to be co-extensive, in spite of incidental but clear warnings such
as Rosch and Mervis's remark that a family resemblance structure need not
be the only source of prototypicality (1975: 599). Admittedly, it is easy to
consider them to be equivalent; already in the quotations given above, partial
reasons for their mutual interdependence can be found. More systematically,
the following links between the four characteristics might be responsible for
the idea that prototypicality necessarily entails the joint presence of all four.
Table 1. Characteristics of prototypicality
NON-EQUALITY
differences in structural weight
EXTENSIONAL degrees of representativity
INTENSIONAL clusters of overlapping senses
NON-RIGIDITY
flexibility and vagueness
absence of clear boundaries
absence of classical definition
First, linking the first to the second characteristic is the argument men-
tioned above: if there is no single definition adequately describing the exten-
sion of an item as a whole, different subsets may be defined, but since the
members of a category can usually be grouped together along different di-
mensions, these subsets are likely to overlap, i.e., to form clusters of related
mearungs.
Second, linking the second to the third characteristic is the idea that
members of a category that are found in an area of overlapping between two
senses carry more structural weight than instances that are covered by only
one meaning. Representative members of a category (i.e., instances with a
high degree of representativity) are to be found in maximally overlapping
areas of the extension of a category. (In the example, A and E are less typi-
cal members that B, C, and D, which each belong to two different subsets.)
Third, linking the third to the fourth characteristic is the idea that differ-
ences in degree of membership may diminish to a point where it becomes
unclear whether something still belongs to the category or not. Categories
have referentially blurred edges because of the dubious categorial status of
items with extremely low membership degrees.
And fourth, linking the fourth to the first characteristic is the idea that the
flexibility that is inherent in the absence of clear boundaries prevents the
formulation of an essence that is common to all the members of the category.
Because peripheral members may not be identical with central cases but may
only share some characteristics with them, it is difficult to define a set of
attributes that is common to all members of a category and that is sufficient
to distinguish that category from all others.
These circular links between the four characteristics are, however, mis-
leading. A closer look at some (familiar and less familiar) examples of pro-
totypicality reveals that they need not co-occur.
Bllill
The concept bird (one of Rosch's original examples of prototypicality)
shows that natural categories may have clear-cut boundaries. At least with
regard to our own, real world, the denotation of bird is determinate; edu-
cated speakers of English know very well where birds end and non-birds
begin. They know, for instance, that a bat is not a bird but that a penguin is.
Of course, the principled indeterminacy described by Waismann (1952) as
'open texture' remains: when confronted with an SF creature (a post-World
War III mutant) that looks like a bird but talks like a man, we would not be
sure whether it should be called a bird or not. A boundary problem that is
typical for a prototypical organization of the lexicon would then arise. As it
functions now, however, in present-day English, bird is denotationally
clearly bounded, the archaeopteryx notwithstanding. 14 As has been remarked
elsewhere (Lakoff 1987), the existence of prototypicality effects in clearly
bounded concepts such as bird implies that a strict distinction has to be
made between degree of membership and degree of representativity. Mem-
bership in the category bird is discrete; something is or is not a bird. But
some birds may be birdier than others: the swallow does remain a more typi-
cal bird than the ostrich.
RED
Color terms such as red constituted the starting-point for prototypical re-
search; drawing on the views developed in Berlin and Kay (1969), Rosch's
earliest work is an experimental demonstration of the fact that the borderline
between different colors is fuzzy (there is no single line in the spectrum
where red stops and orange begins), and of the fact that each color term is
psychologically represented by focal colors (some hues are experienced as
better reds than others) (Heider 1972, Heider and Olivier 1972). These pro-
totypical characteristics on the extensionallevel are not matched on the defi-
nitionallevel. If red can be analytically defined at all (i.e., if it does not sim-
ply receive an ostensive definition consisting of an enumeration of hues with
their degree of focality), its definition might be 'having a color that is more
like that of blood than like that of an unclouded sky, that of grass, that of the
sun, that of ... (etc., listing a typical exemplar for each of the other main
colors)'. Such a definition (cp. Wierzbicka 1985: 342) does not correspond
with either the first or the second characteristic mentioned above.
ODD N1J};ffiER
Armstrong, Gleitman and Gleitman (1983) have shown experimentally that
even a mathematical concept such as odd number exhibits psychological
representativity effects. This might seem remarkable, since odd number is a
classical concept in all other respects: it receives a clear definition, does not
exhibit a family resemblance structure or a radial set of clustered meanings,
does not have blurred edges. However, Lakoff (1982) has made clear that
degrees of representativity among odd numbers are not surprising if the ex-
periential nature of concepts is taken into account. For instance, because the
even or uneven character of a large number can be detennined easily by
looking at the final number, it is no wonder that uneven numbers below 10
carry more psychological weight: they are procedurally of primary impor-
tance.
VERS
As I have tried to show elsewhere (1988a), the first characteristic mentioned
above is not sufficient to distinguish prototypical from classical categories,
since, within the classical approach, the absence of a single definition char-
acterized by necessity-cum-sufficiency might simply be an indication of
polysemy. This means that it has to be shown on independent grounds that
the allegedly prototypical concepts are not polysemous, or rather, it means
that prototypical lexical concepts will be polysemous according to a defini-
tional analysis in tenns of necessary and sufficient conditions (the classical
definition of polysemy), but univocal according to certain other criteria.
These criteria may be found, for instance, in native speakers' intuitions
about the lexical items involved, intuitions that may be revealed by tests
such as Quine's (1960) or Zwicky and Sadock's (1975). In this sense, the
first characteristic has to be restated: prototypical categories will exhibit
intuitive univocality coupled with analytical (definitional) polysemy, and not
just the absence of a necessary-and-sufficient definition.
Once this revision of the first characteristic is accepted, it can be demon-
strated that the first and the second criterion need not co-occur. Lexical
items that show clustered overlapping of senses may either confonn or not
confonn to the revised first characteristic. An example of the first situation
is the literal meaning of bird, an example of the second situation the Dutch
adjective vers, which corresponds roughly with Englishfresh (except for the
fact that the Dutch word does not carry the meaning 'cool'). Details of the
comparison between both categories may be found in the paper mentioned
above; by way of summary, Figures 1 and 2 represent the definitional analy-
sis of both items. The distinction in intuitive status between vers and bird
can be demonstrated by means of the Quinean test (roughly, a lexical item is
ambiguous if it can be simultaneously predicated and negated of something
............................................................................................................6 7 ,
kiwi
y o_s_tr_i_ch_-t_:hiCkeJ
3---+-------------+-----+-----'
penguin
5---t--------------'
L--------------4------'
1 being able to fly
4 having wings
7 having a beak or bill
2 having feathers
5 not domesticated
3 being S-shaped
6 being born from eggs
Figure J. A definitional analysis of bird
in a particular context). Thus, taking an example based on the corresponding
ambiguity in the English counterpart of vers, it would be quite normal to
state that the news meant in the sentence there was no fresh news from the
jightini
5
is fresh in one sense ('recent, new') but not in another ('in optimal
condition'): it makes sense to say that the news is at the same time fresh and
not fresh. By contrast, it would be intuitively paradoxical to state that a
penguin is at the same time a bird and not a bird (disregarding figurative
extensions of the semantic range of bird). Nevertheless, the definitional
analyses in Figures 1 and 2 make clear that both concepts exhibit prototypi-
cal clustering. In both cases, too, the structural position of the instances just
discussed (news, penguin) is not in the central area with maximal overlap-
ping. In short, then, the revised version of the first characteristic need not
coincide with the second characteristic.
The insight derived from a closer look at the four examples just described
may be summarized as in Table 2. It is now easy to see to what extent 'pro-
totypicality' is itself a prototypical notion. There is no single set of attributes
that is common to all of the examples discussed here. Rather, they exhibit a
family resemblance structure based on partial similarities. In this sense, the
set of prototypical concepts characterized by clustering of senses overlaps
e.g. infonnation
1
2
e.g. fruit and
other foodstuffs
'---- 1
e.g. air
1 new, novel, recent
2 in an optimal condition, pure, untainted
Figure 2. A definitional analysis of vers
with the subset characterized by fuzzy boundaries (because of vers), and so
on. At the same time, some concepts are more typically prototypical than
others. (Bird and vers are more prototypical than red.) Notice, in particular,
that the category fruit makes a good candidate for prototypical prototypical-
ity, in the sense that it seems to combine all four characteristics. It shares the
prototypical characteristics of bird, but in addition, things such as coconuts
and, perhaps, tomatoes, seem to point out that the denotational boundary of
fruit is less clear-cut than that of bird.
However, although the examples considered above do not have a set of
attributes in common, they do share a single feature, viz. degrees of mem-
bership representativity. It is highly dubious, though, whether this feature
alone suffices to distinguish prototypical concepts from classical concepts. If
the possibility of a single necessary-and-sufficient definition is one of the
features par excellence with which the classical conception has been identi-
fied, it might justifiably be claimed that degrees of representativity are en-
tirely compatible with the classical conception of categorization. It is, in
fact, in that sense that Annstrong, Gleitman and Gleitman (1983) deal with a
category such as odd number. The experiments used by Rosch to measure
degrees of representativity are not, they claim, indicative of prototypicality
since they occur with classical, rigidly definable concepts such as odd num-
ber. To say the least, representativity effects are only a peripheral prototypi-
cal attribute according to Table 2 (cp. Lakoff 1986). But at the same time,
the debate over the status of odd number shows that the concept 'prototypi-
cal concept' has no clear boundaries: it is not immediately clear whether a
concept such as odd number should be included in the set of prototypical
concepts or not.
Table 2. The prototypicality of 'prototypicality'
P:::
~
<Zl
o:l
Cl
P::: Cl
Cl ::2
P:::
~ ~
Cl ~
iQ
>
P::: 0 z
absence of classical definition +
clusters of overlapping senses
+ +
degrees of representativity + + + +
absence of clear boundaries + +
Of course, contrary to the situation in everyday speech, such a boundary
conflict should not be maintained in scientific speech. A discipline such as
linguistics should try to define its concepts as clearly as possible, and the
purpose of this section is precisely to show that what has intuitively been
classified together as instances of prototypical categories consists of distinct
phenomena that have to be kept theoretically apart. In line with prototype
theory itself, however, such an attempt at clear definition should not imply
an attempt to define the 'true nature' or the 'very essence' of prototypicality.
Determining an 'only true kind' of prototypicality is infinitely less important
than seeing what the phenomena are and how they are related to each other
by contrast or similarity.
Still, there might seem to be one way in which decent sense could be
made of the question what the true meaning of prototypicality would be. To
begin with, let us note that the prototypical character attributed to the con-
cept of prototypicality also shows up in the fact that the notion 'prototype' is
an extremely flexible one. This can be illustrated in two ways. First, the
lexical item prototypical is spontaneously used to name a number of phe-
nomena that are linked by metonymy, next to the phenomena linked by simi-
larity that are brought together in Table 2. The lexical item does not only
characterize structural features of concepts, and the concepts exhibiting
those features themselves, but sometimes even particular (viz., highly repre-
sentative) instances of the categories in question (the robin as a prototypical
bird). Second, context may stress one feature of prototypical organization
rather than another (cp. the priming effects in Rosch 1975). The general
purpose of one's investigations may lead one to devote more attention to one
aspect of the prototypical cluster than to another. To name a few examples:
degrees of representativity are important for language development studies
(if it is taken into account that most concepts in early language development
are acquired via their exemplars), while clustered overlapping of senses will
come to the fore in linguistic or lexicographical studies into the structure of
polysemy. And a cognitive interest into the epistemological principles under-
lying natural language will attach more weight to the decoupling of intuitive
univocality and analytical, definitional polysemy.16
In this respect, the question with regard to the true nature of prototypical-
ity might be transformed into the question what might be the most interesting
(or perhaps even the most important) perspective for studying and defining
prototypicality. But here again, the 'ultimate essence fallacy' exposed by
prototype theory itself lurks round the corner: there will be different prefer-
ences for one perspective rather than another, but there will be no single
ultimately and eternally most important conception of prototypicality.
In short, the foregoing analysis corroborates Wierzbicka's remark that
there are 'many senses' to the notion prototype, and that 'the notion proto-
type has been used in recent literature as a catch-all notion' (1985: 343).
However, a more systematic analysis than Wierzbicka's reveals that this
very multiplicity of usage also supports Cognitive Semantics, in the sense
that it shows that the same categorization principles may guide common
sense and scientific thinking. This is, then, a further indication of the
metatheoretical relevance of a cognitive conception of linguistic categoriza-
tion, which I have explored at length elsewhere (1985b). At the same time, it
has become clear that one of the major tasks for the further development of
prototype theory is the closer investigation of the prototypically clustered
characteristics of prototypicality. A major reference in this respect is La-
koff's attempt (1987: Chapter 4-8) to determine which different kinds of
conceptual models may lie at the basis of prototypicality effects.
3. Definitional problems, second Series: 'Prototype theory' as a
prototypical notion
Whereas the previous section made clear that prototypicality as used in lin-
guistic semantics is a prototypically structured concept, it should now be
noted that the prototype-theoretical movement as well is a prototypically
structured approach to semantics. There are, in other words, central as well
as more peripheral examples of prototypical theories. In particular, there
exist a number of theories that combine aspects of the classical approach to
semantic structure with aspects of the prototypical conception. In this sec-
tion, two approaches will be considered that are to some extent semi-
classical as well as semi-prototypical; each of both embodies a strategy for
reinstating particular aspects of the classical view against the background of
an overall cognitive point of view.
To begin with, some of the clarity and neatness of the classical approach
may be recovered by concentrating on the prototypical centre of a category.
If the non-classical indeterminacy of lexical concepts stems primarily from
the flexible extendibility of concepts, discreteness may be reinstalled by
avoiding the problems of clustered polysemy, i.e., by restricting the defini-
tional analysis to the prototypical centre of the category. This approach is
vigorously carried through by Wierzbicka (1985), who explicitly defends the
discreteness of semantics by introspectively considering only the clear, sali-
ent centre of lexical categories. In a discussion of Labov's experimental
investigation into the non-classical characteristics of everyday concepts
(1973), she notes: 17
To state the meaning of a word, it is not sufficient to study its applicability
to things; what one must do above all is to study the structure of the con-
cept which underlies and explains that applicability. In the case of words
describing natural kinds or kinds of human artefacts, to understand the
structure of the concept means to describe fully and accurately the idea (not
just the visual image) of a typical representative of the kind: the prototype.
And to describe it fully and accurately we have to discover the internal
logic of the concept. This is best done not through interviews, not through
laboratory experiments, and not through reports of casual, superficial im-
pressions or intuitions ... but through methodical introspection and think-
ing (1985: 19).
It should be noted immediately that Wierzbicka's reinstatement of dis-
creteness does not imply that her definitions do in fact always consist of
necessary-and-sufficient conditions, and she acknowledges as much (1985:
60). In this respect, Wierzbicka's approach is only partly a departure from
the hard core of prototype-theoretical studies: the absence of necessary-and-
sufficient conditions for the definition of certain core concepts is accepted,
but the avoidance of the clustered polysemy problem 'tidies up' the semantic
description and reinstates some of the classical neatness. Neither does
Wierzbicka's approach imply that lexical items are always univocal; in her
dictionary of English speech act verbs (1987a), several items receive multi-
ple definitions. Each of the definitions does, however, constitute a highly
salient meaning, and again, by disregarding peripheral kinds of usage, the
clustered or radial structure of the polysemy of lexical items does not enter
the picture. The question to be asked, then, is whether Wierzbicka' s restric-
tion of the description to the salient meanings of a category is useful and
adequate from a cognitive point of view.
From a methodological point of view, the periphery of natural, non-
uniquely definable categories is as interesting as their salient centre(s), be-
cause it is precisely the relationship between both that typically characterizes
natural categories. Cognitive Linguistics is not only interested in what con-
stitutes the centre of a category, but also in how this centre can be extended
towards peripheral cases, and how far this extension can go. The mecha-
nisms for incorporating marginal cases into a category at the same time re-
strict the flexibility of that concept; it is only by studying peripheral cases,
for instance, that an answer may be found with regard to the question how
dissimilar things can be before they are no longer recognized as basically the
same. If, in other words, flexible polysemization is indeed one of the major
characteristics of natural language categories, a deliberate restriction of the
description to the salient meanings of a category is methodologically less
propitious, as it may lead to a neglect of this basic feature. 18
A second strategy for salvaging aspects of the classical approach is to in-
voke sociolinguistic mechanisms such as Putnam's 'division of linguistic
labor' (1975). According to Putnam, ordinary language users possess no
more than 'stereotypical' knowledge about natural kinds, that is to say, they
are aware of a number of salient characteristics, such as the fact that water
is a transparent, thirst-quenching, tasteless liquid. The technical definition of
water as H
2
0, on the other hand, is to be located primarily with scientific
experts. It is the experts' knowledge that ultimately determines how natural
kind terms are to be used. On the one hand, a 'division of linguistic labor'
ensures that there are societal experts who know that water is H
2
0, that
there is a distinction between elms and beech, how to recognize gold from
pyrites, and so on. On the other hand, laymen attune their own linguistic
usage to that of the expert scientists, technicians, etc.. The members of the
non-specialized group are not required to have expert knowledge, but if they
wish to be considered full-fledged members of the linguistic community, they
are supposed to know the 'stereotype' connected with a category. A stereo-
type is, thus, a socially determined minimum set of data with regard to the
Prospects and problems ofproto(vpe theory 21
extension of a category. Given the similarity between Putnam' s stereotypes
and the prototypes of Cognitive Linguistics (both consist roughly of the most
salient information connected with a category), the division of linguistic
labor might be used to rescue the classical view of concepts. 19 Expert defini-
tions being classical (they specify an essentialist 'hidden structure' for natu-
ral kinds), the stereotypical concepts of everyday language users might now
be seen as hardly more than a sloppy derivative of those classically defined
expert categories. 'True' (expert) definitions would be classical, and stereo-
typical/prototypical concepts might be dismissed as sociolinguistically sec-
ondary phenomena.
It should be remarked immediately that such a reinstatement of the clas-
sical view is not as obvious for other words than the natural kind terms for
which Putnam' s theory is in fact intended (what is the expert definition of
the preposition for?). Moreover, as a sociolinguistic theory about the social
factors that determine how lexical items may be used, the 'division of lin-
guistic labor' theory is incomplete to say the least. The primacy of expert
definitions would seem to imply that natural language follows the develop-
ments and discoveries of science in a strict fashion. In actual fact, however,
natural language categorization is not only determined by the state of affairs
in the sciences, but also by the communicative and cognitive requirements of
the linguistic community in its own right. One of Putnam's own examples
may serve as an illustration. Although science has discovered that jade re-
fers to two kinds of materials (one with the 'hidden structure' of a silicate of
calcium and magnesium, the other being a silicate of sodium and alumin-
ium), ordinary usage continues to refer to both substances indiscriminately
as jade. That is to say, categorization in everyday language is not entirely
dependent upon scientific research, but seems to be determined at least in
part by independent criteria: if the classificatory exigencies of everyday
communicative interaction do not call for a distinction between the two kinds
ofjade, the scientific splitting of the category is largely ignored. This implies
that an investigation into everyday language categorization as an independ-
ent cognitive system is justified. More generally, if Putnam' s view is seen as
a theory about the sociolinguistic structure of semantic norms, his hierarchi-
cal model (with experts at one end and laymen at the other) is only one
among a number of alternatives, some of which (such as the one described
by Bartsch 1985) link up closely with a prototypical conception of categorial
structure. At the same time, however, it should be admitted that the relation-
ship between classical scientific categorization and prototypical common-
sense categorization may be explored in more depth than is yet the case.
20
To summarize: the confusion associated with the notion of prototypicality
is further increased by the fact that more straightforwardly prototypical ap-
proaches are surrounded by hybrid theories that contain particular strategies
for combining classical discreteness with typically prototypical phenomena.
We have discussed two such approaches (one in which the strategy in ques-
tion is methodological, and another one in which it is sociolinguistic), but
this does not mean that these are the only ones that might be mentioned.
21
The two approaches mentioned here are, however, particularly revealing, as
they link up with two important currents in the history of Western thought.
The first one simplifyingly boils down to the view that the mind is neat (if
you look hard enough into it), but that the world is fuzzy: the non-
discreteness that Cognitive Linguistics is concerned with arises from the fact
that we have to apply clear-cut mental categories to an external reality that
is so to say less well organized. The conception that the world of mental
entities is somehow better organized than the outside world is obviously an
idealistic one (though it does not constitute the only possible kind of ideal-
ism); Wierzbicka herself stresses the Platonist character of her approach. On
the other hand, Putnam' s view that science is neat whereas everyday lan-
guage is fuzzy, links up with the empiricist objectivism of the Ideal Lan-
guage branch of analytical philosophy: the objective structure of reality is
best described by the language of science, and everyday language is at best a
weak derivative of scientific categorization, at worst a conceptual muddle
teeming with philosophical pseudo-problems. As the previous discussion
suggests that hard-core Cognitive Linguistics steers clear of both the idealist
and the objectivist option, we have here one more indication
22
for the neces-
sity of a further investigation into the epistemological, philosophical back-
ground of the prototypical conception of categorial structure.
Notes
1. The discussion in section 2 will make clear that the term p r o t o ~ y p e theory
should be used with care, since the theoretical uniformity that it suggests
tends to obliterate the actual distinctions between the diverse forms of proto-
typicality discussed in the literature. The term is used here as a convenient
reference mark only, to indicate a number of related theoretical conceptions
of categorial structure that share an insistence on any or more of the various
kinds of prototypicality effects discussed in section 2.
2. Though not exclusively: see Rosch (1988: 386).
Prospects and problems o/prototype theory 23
3. Notice that this claim applies just as well to the axiomatic, postulate-based
form of description that developed as the major representational alternative
for Katzian componential analysis. The notion of criteriality is just as much
part and parcel of the classical versions of the axiomatic alternative as it is of
Katzian feature analysis.
4. See, among others, Haiman (1980a) and Geeraerts (1985b).
5. The distinction between semantic and encyclopedic concepts against which
Cognitive Semantics reacts is often misconstrued. In particular, in the state-
ment that there is no principled distinction between semantic and encyclope-
dic information, the words semantic and encyclopedic are not used (as im-
plied by Lehmann 1988b) in the senses 'as may be found in dictionaries' and
'as may be found in encyclopedias', respectively. Rather, the rejected distinc-
tion refers to an alleged distinction within an individual language user's con-
ceptual memory; it involves the presupposition that there is an independent
level of semantic information that belongs to the language and that is distinct
from the individual's world knowledge. The kind of information that is typi-
cally found in encyclopedias involves scientific information ofthe kind 'ovu-
lation triggered by copulation' for the item cat (the example is Lehmann's);
but while the distinction between scientific and laymen's knowledge is pri-
marily a social one, this kind of 'encyclopedic' information is only relevant
for the psychological perspective of Cognitive Semantics if the individual
lexicon to be described is that of someone with a certain amount of scientific
knowledge of cats (or if, through sociolinguistic idealization, the average
language user's lexicon may be supposed to contain that piece of scientific
information).
6. There are, of course, exceptions such as Dowty (1979) to confirm the rule.
The historical sketch of the advent of prototype theory given here is treated
more thoroughly in Geeraerts (1988a).
7. As the semantic interests of the former audience of Generative Semantics
were so to say no longer envisaged by the leading theories of the day, it does
not come as a total surprise, from this point of view, to find George Lakoff,
one of the leading Generative Semanticists, again as one of the leading cog-
nitivists.
8. These antecedents are not the only ones that might be mentioned. I have
elsewhere (1988c) drawn the attention to the similarities between the pre-
structuralist, historical tradition of semantic research and present-day Cogni-
tive Semantics, but there are other (admittedly non-mainstream) traditions of
semantic research with which Cognitive Semantics is methodologically re-
lated: think, e.g., of the anthropological research of Malinowski, Firth, and
the London School in general. Even a structuralist such as Reichling has held
views about the structure of polysemy that come close to the point of view of
prototype theory: his influential work on the word as the fundamental unit of
linguistics (1935) contains an analysis of the Dutch word spel that is awk-
wardly similar to Wittgenstein's remarks about the German equivalent Spiel.
The point to be stressed is this: as a theory about the (radial, clustered, dy-
namically flexible) structure of polysemy, prototype theory is to a consider-
able extent a rediscovery of views that were paramount during the prestruc-
turalist era of the development of lexical semantics, and that lingered on
below the surface in the structuralist and transfonnationalist periods.
9. Because of their large scope, the functionalist approach of Seiler (1986) and
the naturalist approach of Dressier (1985) are particularly interesting for the
use of prototypicality with regard to various aspects of the fonnal organiza-
tion of language.
10. A bibliography of work in Cognitive Linguistics is to be found in Dirven
(1988). It is worth mentioning that Cognitive Linguistics is currently in a
stage of organization: a first international conference of Cognitive linguis-
tics was held in Duisburg in March 1989, and a new journal entitled Cogni-
tive Linguistics, published by Mouton, is scheduled to start appearing in the
beginning of 1990.
11. Next to the link with psycholinguistics, there is a connection with Artificial
Intelligence research, through the correspondences between the notion of
prototypicality and that of frame; see Fillmore (1977a). It needs to be
stressed, though, that the link is relatively weak; specifically, the correspon-
dence just mentioned is to a certain extent counterbalanced by Lakoff's criti-
cism (1987) of the objectivist assumptions of mainstream Artificial Intelli-
gence research (but then again, one of Lakoff's current research projects
involves a connectionist approach to the fonnal modeling of Cognitive Se-
mantic notions such as metaphorical image schemata). In general, sorting out
the relationship between Cognitive Semantics and Artificial Intelligence-
oriented Cognitive Science will be one of the major tasks for the further de-
velopment of Cognitive Semantics.
12. See Lakoff (1987: Chapter 6) for the notion of a radial set and compare
Givon (1986) for a comparison between the views of Wittgenstein and those
of prototype theory. The stress Givon places on the distinctions between both
is slightly exaggerated, as it tends to obscure their mutual rejection of the so-
called classical theory. See also the next footnote.
13. The 'so-called' is added to stress, first, that the views of Aristotle also con-
tain features that correspond rather with a cognitive than with a 'classical'
approach, and second (more generally), that the philosophical position of
prototype theory is in need of further elucidation. The present situation is
rather muddled: while the classical Roschian position is to characterize pro-
totype theory as non-Aristotelian and Wittgensteinian, Givon (1986) has ar-
gued that prototype theory is non-Wittgensteinian (see the previous note), but
whereas Givon also describes prototype theory as non-Platonic, Wierzbicka
(to whom we shall come back in section 3) precisely presents an explicitly
Platonic version of prototype theory. More generally, the philosophical posi-
tion of prototype theory has so far been discussed mainly against the back-
ground of classical philosophy (Aristotle and Plato), and against the back-
ground of contemporary analytical philosophy (see Lakoff 1987). This means
that a large part of the history of Western philosophy passes unmentioned;
this is to be regretted, as the post-Cartesian period in the history of philoso-
phy is concerned with epistemological questions that are of immediate inter-
est to Cognitive Semantics. In particular, if it can be accepted that one of the
major epistemological aspects of a prototypical conception of categorial
structure resides in the fact that categories are interpretive schemata that are
used flexibly and dynamically in our encounters with reality, a major phi-
losophical reference point for prototype theory will lie with those philosophi-
cal theories that recognize the constitutive role of existing knowledge with
regard to new experiences. As I have argued elsewhere (1985b), the Husser-
lian phenomenological movement (as represented, specifically, by Maurice
Merleau-Ponty) provides a good starting-point for a further confrontation
with philosophy.
14. The archaeopteryx is probably regarded as a species separate from either bird
or reptile.
15. The example is taken from the Longman Dictionary of Contemporary Eng-
lish.
16. Considered from this point of view, Lakofrs radial sets as such are not par-
ticularly unclassical: structured polysemy as such is entirely compatible with
the classical view. Kleiber (1988) offers an insightful discussion of the theo-
retical consequences of the growing importance of the structure of polysemy
in prototype-theoretical research.
17. For a more extended discussion of Wierzbicka's views, see Geeraerts
(1988c).
18. Notice that the restriction to the prototypical centre of categories correlates
with Wierzbicka's Platonic, introspective methodology: it seems probable
that the applications of a category that can be accessed introspectively are
only the more salient ones; peripheral cases probably do not always pass the
threshold of conscious attention. What is interesting from a cognitive point of
view, however, is the way people spontaneously categorize and classify
things, not the way in which they introspectively reflect upon their own con-
ceptualizations. Any attempt to describe the peripheral instances of a cate-
gory together with its prototypical centre can therefore not be restricted to an
introspective methodology.
19. This is not say that Putnam actually intended his stereotypical theory as such
an attempted rescue: his problems lay with the notion of reference rather than
with those of polysemy and categorial structure. My remarks about Putnam
are an investigation into some of the possible consequences of the notion of
division of linguistic labor, not an attempt to give an account of Putnam's
view in its original setting. Further, it has to be mentioned that some of Put-
nam's later philosophical views open up entirely different perspectives for a
confrontation with Cognitive Semantics; in particular, see Lakoff (1987) on
Putnam and anti-objectivism.
20. An interesting contribution to such an exploration is found in Lakoff (1987:
Chapter 12), where it is claimed that scientific categories are far from being
as classical as is usually assumed.
21. Again, see Lakoff (1987: Chapter 9) for some more examples; they are situ-
ated within formal psycholexicology rather than within linguistics.
22. Next, that is, to the remarks made in footnote 13.
Chapter 2
Where does prototypicality come from?
Originally published in Brygida Rudzka-Ostyn (ed.), 1988, Topics in Cognitive
Linguistics 207-229. Amsterdam/Philadelphia: John Benjamins.
The prototype-theoretical case study reprinted in the present chapter is not just
included because it provides a concrete illustration of the theoretical model dis-
cussed in Chapter 1. The paper makes two points of a more general nature. First,
on the theoretical level, the paper argues for a functional explanation of proto-
typicality effects, i.e. an explanation in tenns of general requirements on the or-
ganization of human cognition: It is cognitively advantageous to maximize the
conceptual richness of each category through the incorporation of closely related
nuances into a single concept because this makes the conceptual system more
economic.
Second, on the methodological level, the paper introduces an onomasiological
perspective in lexical semantics: it is only by comparing the near-synonyms ver-
nielen and vernietigen that we get a correct insight into the factors that might
explain the presence of prototypicality effects. Over the years, I have become more
and more convinced that such an onomasiological perspective is essential in Cog-
nitive Linguistics: categorization primarily involve onomasiological choices, i.e.
the question 'Why does someone use the category x rather than the category y for
talking about phenomenon zT is cognitively more realistic than the question
'Does z belong in the semantic range of application of xT Starting with the mono-
graph The Structure of Lexical Variation (Geeraerts, Grondelaers, and Bakema
1994), the lines that I have tried to set out with our Leuven research group have
increasingly been geared towards the analysis of onomasiological variation (see
Geeraerts 2002b, Grondelaers and Geeraerts 2003 for a general statement and an
overview, and compare Chapter 11).
1. Hypotheses about the sources of prototypicality
Prototype theory is as it were part of the prototypical core of the cognitive
paradigm in semantics, particularly in lexical semantics. I think it is safe to
say that it is by now quite obvious that gradience and salience are among the
linguistically relevant aspects of semantic structure. One need only recall the
early experimental work by Rosch (1973) and Labov (1973) to appreciate
the importance of gradualness and vagueness for the adequate description of
word meaning. But what about explanatory adequacy? Can we move beyond
the descriptive level and explain why prototypicality exists at all? There are
at least four different hypotheses that have been proposed to explain proto-
typical phenomena. Each of these hypotheses has been formulated (or at
least hinted at) by Eleanor Rosch herself; this is an indication that the hy-
potheses might well be complementary rather than mutually contradictory. I
will call these four hypotheses the physiological, the referential, the statisti-
cal, and the psychological one. Let us have a look at them.
The physiological hypothesis says that prototypicality is the result of the
physiological structure of the perceptual apparatus (Rosch 1973). This hy-
pothesis has been formulated with regard to the prototypicality effects in the
domain of color terms (the first major field in which prototypicality phenom-
ena have been observed). Particular colors are thought to be focal because
the human eye is more sensitive to certain light frequencies than to others.
The scope of the physiological explanation is probably fairly limited; it may
only be applicable to concepts immediately referring to perceptual phenom-
ena, or at least to bodily experiences that have a distinct physiological basis.
Since this is most likely not the majority of cases, additional hypotheses will
have to be invoked to explain the prototypical structure of concepts that have
no immediate physiological basis.
The referential hypothesis states that prototypicality results from the fact
that some instances of a category share more attributes with other instances
of the category than certain peripheral members of the category (or share
attributes with more other instances than these peripheral cases). The pe-
ripheral applications of a category share attributes with relatively few other
cases, or share only a relatively small number of attributes with other, more
central members of the category. This is the family resemblance model of
prototypicality (Rosch and Mervis 1975); in psychological terms, it states
that the prototypical instances of a category maximize cue validity. I have
dubbed this view 'referential' because it considers prototypicality to be an
automatic consequence of the structure of the range of application of a con-
cept. Once you know what objects, events etc. a concept can refer to, you
can compute differences in salience by comparing the number of shared
attributes among those things. One might even say that prototypicality is a
secondary phenomenon: it is a side-effect of the mutual attribute relations
among the instances in the referential range of application of the concept.
Statistical explanations of prototypicality state that the most frequently
experienced member of a category is the prototype. At least, this is the sim-
Where does prototypicality come from? 29
pIe fonn of the frequency model. It can also be combined with the family
resemblance model; the weight of an attribute within a concept is then not
only detennined by its role within the family of applications constituting the
category, but also by the relative frequency with which it is experienced
(Rosch 1975).
The psychological hypothesis is a functional one. It states that it is cog-
nitively advantageous to maximize the conceptual richness of each category
through the incorporation of closely related nuances into a single concept
because this makes the conceptual system more economic. Because of the
maximal conceptual density of each category, the most infonnation can be
provided with the least cognitive effort (Rosch 1977).
In what follows, I would like to show that the functional explanation of
prototypicality is more general than the other ones because it can explain
cases of prototypicality that are counter-examples to the other modes. I will
elaborate the psychological hypothesis by indicating some more functional
sources of prototypicality; I will try to make clear that prototypicality is the
outcome of some deep-seated principles of cognitive functioning.
2. A case study in synonymics
Dutch has a pair of synonyms vernielen and vernietigen, which both roughly
mean 'to destroy'. Though they exhibit some degree of phonetic similarity,
their origin is quite diverse. Vernielen is the older fonn. It is already to be
found in Middle Dutch, and it is fonned by means of the common verb-
fonning prefix ver- and the adjective niel, only a few examples of which
survive, but which probably meant 'down to the ground'. Etymologically,
then, vernielen means 'to throw down to the ground, to tear down'.
Vernietigen, on the other hand, makes its first appearance in the sixteenth
century; it is fonned by means of the same prefix ver- and the adjective
nietig, which is itself a derivation from the negation particle niet (English
not) and the suffix -ig (which corresponds with English -y). Vernietigen
gradually replaces a third fonn verniefen, which is a straightforward deriva-
tion from niet with ver-, and which is extinct by the end of the seventeenth
century. Vernietigen literally means 'to annihilate, to bring to naught'.
The best way to study both words is to turn to the Woordenboek der
Nederlandsche Taal (hence WNT), the major dictionary of Dutch that cov-
ers the period from 1500 up to 1920 and that, by the way, is still uncom-
pleted after about a century of editorial work. This dictionary is being com-
piled on the basis of a huge corpus of quotations; there are as yet no equally
representative corpora for contemporary Dutch, so that it is rather more
difficult to get an adequate picture of twentieth-century usage than it is to
study the semantic history of the vocabulary of Dutch. For the purpose of
this paper, this is not very important: it suffices to pick out one synchronic
period and to see how both words relate to each other in that period. For a
number of practical reasons (among others the amount of available mate-
rial), I will concentrate on the nineteenth century, stretching the temporal
borders of that period with approximately one decade at each end. In this
way, a period from 120 to 130 years can be considered, ranging from
roughly 1790 to 1910. In the light of the history of culture, this seems quite
justified; we more or less envisage the cultural period from the French revo-
lution up to the First World War: the nineteenth century in the broadest
sense.
To get a good picture of the development ofvernietigen and vernielen, it
would be necessary to present and discuss the entire articles that I have
compiled for the WNT, and the complete set of quotations on which they arc
based. For obvious lack of space, I will only give illustrative quotations;
bibliographical references are as in the WNT. The set of quotations shows
that vernielen and vernietigen can be used indiscriminately with the same
range of application. Each numbered pair of quotations gives examples of
one particular kind of usage. These examples should be studied from two
points of view. On the one hand, the question has to be asked whether
vernielen and vernietigen exhibit any syntagmatic differences, i.e. differ-
ences in their collocational properties. On the other hand, the question arises
whether they are paradigmatically different, i.e. whether they exhibit purely
conceptual differences. (More details on the analysis of both verbs can be
found in Geeraerts 1985e.)
(1) Dat huis was ... evennmin als de naburige tegen de verwoestende veeten
dier tijd bestand. Reeds onder den zoon en opvolger des stichters werd
het ... tot den grand toe vernield (Veegens, Hist. Stud. 2, 282, 1869).
[Like the neighboring one, this house was not able to stand up against
the destructive quarrels of the age. Already under the son of the foun-
der, it was demolished down to the ground.]
Alleen zijn de vroegere kruisvensters door vensterramen van nieuweren
trant vervangen en hebben de vrijheidsmannen van 1795 ... het wapen
des stichters in den voorgevel met ruwe hand vernietigd (Veegens, Hist.
Stud. 1, 125, 1864).
[Only, the earlier cross-windows have been replaced by windows in a
newer style, and in 1795, the freedom fighters demolished the founder's
arms in the facade with their rough hands.]
(2) Er gaat dan stroom op den daarvoor gevormden zijweg over, waarbij
genoeg warmte ontwikkeld wordt om de draadwindingen in zeer korten
tijd te vernielen (Van Cappelle, Electr. 214, 1908).
[Electric current is then transferred to the diversion that has been con-
strued to that end, in which case enough warmth is produced to destroy
the coils of the wire in a very short time.]
Zonder deze voorzorg zou het draadje door de enorme hitte van den
gloeidraad vernietigd worden (Van Cappelle, Electr. 295, 1908).
[Without this precaution, the wire would be destroyed by the enormous
heat of the filament.]
(3) Hoeveel het wild vernielt wordt door een Engelschman zeer goed uit-
eengezet bij gelegenheid van een aanval op de beschenning die het wild
aldaar geniet (Volksvlijt 1872, 175).
[How much is destroyed by game is aptly expressed by a certain Eng-
lishman on the occasion of an attack on the protection these animals
enjoy in his country.]
Bij het vernietigen van de onkruiden door het bewerken dient op hunne
voortplanting en ontwikkeling te worden gelet (Reinders, Land. I, 309,
1892).
[During the destruction of weeds by cultivating the land, one should
bear in mind their reproduction and development.]
(4) Wel wat hamer! Wordt door zulke sentimentele zotternyen niet al de
inwendige kracht vernield? (Wolff en Deken, Blank. 3, 220, 1789).
[By golly! Does not such sentiment al foolishness destroy all our inner
strength?]
Stel mij niet zoo hoog, zei ze onthutst, ik zou daaraan niet beantwoor-
den; ik zou uw ideaal vernietigen (Vosmaer, Amaz. 175, 1880).
[Do not put me on a pedestal, she said disconcertedly, I would not live
up to that; I would destroy your ideal.]
(5) De bergstroom in zijn grammen loop Verscheurt zijn zoom, verdrinkt
de dalen: Alzoo vernielt Gij 's menschen hoop! (Ten Kate, Job 53,
1865).
[The mountain stream in its angry course rends its banks, drowns the
valleys: thus, Thou destroys the hope of Man!]
Zy is dan, van kindsbeen af, opgevoed om mynheer Daniel's echtgenote
te worden, en nu is die hoop van een geheelleven vernietigd! (Con-
science, Kwael d. T. 2, 65, 1859)
[She has been raised from childhood to become Master Daniel's wife,
and now this hope of a lifetime has been annihilated!]
Dit toeval vernietigde ons geheele plan (Haafner, Ceilon 103, 1810).
[This coincidence annihilated our entire plan.]
(6) Mij gedenkt ook nog dat Nicolaas Gaal ... mij placht te verhalen ... dat
de oude man om deze ontstolen eer zich zoo ontstelde en vergramde,
dat het ook scheen ofhij dezen diefwel had willen vernielen (Fruin,
Geschr. 1,1974, 1888).
[I still remember that Nicolas Gaal was fond oftelling me that the old
man used to get so angry and upset about this stolen honor, that it
seemed that he would have liked to kill that thief.]
Intusschen heeft de Godin de Natuur besloten nu voor altijd de Drij-
vende Eilanden en al hun inwoners te vernietigen (Quack, Soc. 1,246,
1875).
[Meanwhile, the goddess Nature has decided to destroy the inhabitants
of the Floating Islands once and for all.]
(7) De beroerte, die haar zwakke levenskrachten in een half uur tijds ver-
nielde, had reeds in het eerste oogenblik hare spraak verlamd (Beets,
C.O. 206, 1840).
[The stroke that destroyed her weak life force in half an hour. had from
the first moment paralyzed her speech.]
Hy moet rusten. Zulke driften vernietigen het sterkste gestel (Wolff en
Deken, Leev. 1,290, 1784).
[He has to take a rest. Such passions undennine the strongest constitu-
tion.]
(8) De vrouwen, Lus, zijn zonen, al de anderen bleven stom, vernield van
ontsteltenis, op hun stoelen genageld (Buysse, NeefPerseyn 4545,
1893).
[The women, Lus, his sons, all the others remained silent, destroyed by
disconcertedness, nailed to their chairs.]
Toen antwoordde zij langzaam met een doffe stem, als vernietigd door
haar eigene woorden: 'la, indien het nog mogelijk is' (Buysse, Mea
Culpa 68, 1896).
[Then she answered slowly, in a dull voice, as if struck down by her
own words: 'Yes, if it is still possible'.]
(9) De uitslag van den stryd was ditmael hem niet gunstig: geheel zyn leger
werd vernield of uiteen geslagen (Conscience, Gesch. v. Belgie 110,
1845).
[This time, the result of the battle was not favorable to him: his entire
army was destroyed or dispersed.]
Where does prototypicality comefrom? 33
Het gansch leger der Turken was vernietigd! (Conscience, Gesch. v.
Belgie 352, 1845).
[The entire army of the Turks was destroyed! 1
Syntagmatically, we not only see that both words can be used by the
same author in the same context without noticeable differences, as in (1),
(2), (8) and (9), but also that the range of application of each word can be
divided into three identical major groups, which can moreover be subdivided
along parallel lines. There is a set of applications in which the words are
used with regard to concrete, material objects (1, 2, 3); a set in which they
are used with regard to abstract objects (4, 5), and a set in which they are
used with regard to persons (6, 7, 8, 9). Within the first set, frequently oc-
curring applications relate to buildings (1), other human artifacts (2), and
natural objects, in particular plants and crops (3). With regard to the ab-
stract applications, we can distinguish between the annihilation of the exis-
tence of certain abstract objects as such (4), and applications in which the
realization or fulfillment of certain abstract notions that contain an aspect of
expectation or intention with regard to the future is prevented (5). With re-
gard to persons, (6) expresses their death as such; (7) and (8) indicate how
someone's bodily or mental health, respectively, are undermined. (9) ex-
presses how armies are beaten; this application is half-way between the ab-
stract group (the armies cease to exist as functional entities), and the per-
sonal group (individual soldiers are killed). The existence of analogous
subdivisions within each of the major groups shows that the syntagmatic
equivalence of vernielen and vernietigen is not a coincidence, but that it is
an essential part of their relationship.
Furthermore, the examples also show that there is a paradigmatic, strictly
conceptual equivalence between both: they do not only have the same collo-
cational properties, but they also seem to express the same concepts in the
same contexts. (The distinction between syntagmatic and paradigmatic
meaning is used here for purposes of analysis only; it does not imply any
particular view with regard to the theoretical relation between both aspects
of lexical meaning and particularly with regard to the question whether se-
lectional restrictions are always an automatic consequence of a concept's
paradigmatic characteristics.)
As a preliminary step, notice that the concept 'to destroy' does not only
appear as the notion 'to annihilate the existence of someone or something, to
cause someone or something to disappear out of existence', but that it also
exhibits the weaker nuance 'to undermine someone or something with regard
to some aspect of his existence' (without a complete destruction or a com-
plete removal out of existence being implied). The distinction can easily be
discovered within the personal group of applications. In (6), a person is
killed, taken out of existence, while in (8) (and most likely also in the second
quotation from (7), someone's existence is undermined from one point of
view or another, but not entirely annihilated. Likewise, we can see that
within the abstract group, (4) signifies the suppression of the existence of
some abstract things as such, whereas in (5), plans, hopes, and expectations
are undermined with regard to their realization and fulfillment: the plan as
such is not removed (at least not to begin with), but it is reduced to ineffec-
tiveness and futility. In short, both vernielen and vernietigen express the
notions of complete destruction and partial damage, that is to say, the com-
plete removal out of existence of something or someone, and the less drastic
undermining in some respect, of the existence of people or objects. Accord-
ing to the syntagmatic context, these notions receive further specifications.
For instance, with regard to persons, complete destruction means killing, but
with regard to concrete things, destruction signifies material demolition, and
so on. (For the sake of completeness it should be added that the equivalence
of vernielen and vernietigen is less straightforward in present-day Dutch
than it is in nineteenth-century Dutch. Some of the quotations discussed here
are now felt to be rather awkward; in particular, it would be difficult to use
vernielen with regard to persons. )
On the basis of the foregoing observations, one might be tempted to con-
clude that the semantic structure of vernielen and vernietigen in nineteenth-
century Dutch is completely identical: both syntagmatically and paradig-
matically, they have the same range of application. However, a number of
facts testify that both words have different prototypical structures, i.e., that
they have different conceptual centers. There are two sets of facts to be con-
sidered: corpus-based facts relating to the way in which both words are used
in our corpus of quotations, and introspective facts relating to the way in
which the words are perceived by the speakers of the language. In general,
consideration of these facts will lead to the conclusion that the abstract ap-
plications are central within the structure of vernietigen, and that the mate-
rial applications are central in the cases ofvernielen. As such, each verb has
a different semantic structure in spite of the fact that the elements of these
structures appear to be the same.
3. Usage as evidence for prototypicality
Five observations support the prototypical hypothesis. In the first place, the
abstract group of applications is quantitatively more prominent within the
structure of vernietigen than the material set of applications, while the re-
verse is true of vernielen, in which the material group is the most frequently
occurring one. In both cases, the major group is represented by approxi-
mately three times the quotations of the less central group.
In the second place, the differences in centrality show up in the fact that
the prominent applications exhibit specifications and particular nuances that
they do not have when they are peripheral within the structure of the lexical
item. Thus, the material group of vernielen contains a metonymical exten-
sion of the application with regard to plants and crops, towards an applica-
tion in which the fields and gardens where these plants and crops grow ap-
pear as the direct object of the verb. Likewise, the application with regard to
buildings receives a figurative extension towards an application with regard
to an allegorical 'wall' that separates two people. These extensions are
probably not impossible within the concrete set of applications of vernieti-
gen, but the fact that they do not appear there is statistically interesting: it
indicates that the concrete application is more productive in the case of
vernielen than in the case ofvernietigen. Conversely, the abstract group has
nuances and additional specifications in the case ofvernietigen that are lack-
ing in the same group with vernielen, although it is quite easy to imagine
that they would in fact occur there. For example, vernietigen has a fairly
large set of applications in which social movements, institutions, activities
and so on are abolished, one quotation in which it is said that railway trans-
port destroys distances (obviously, distances do not disappear as such, they
are only functionally overcome), and one quotation in which a philosopher is
said to destroy the soul (again, the soul is not destroyed as such, but the idea
that the soul exists is metonymically abolished by the philosopher in ques-
tion). None of these extensions of the abstract use of the concept 'to destroy'
can be found in the case of vernielen, which is indicative of the fact that the
abstract use is less prominent in the latter verb than in the semantic structure
of vernietigen.
In the third place, the salience of the material kind of usage can be de-
rived indirectly from the nominalizations of both verbs. Both vernietiging
and vernieling have the verbal sense 'the fact, the act or the process of de-
stroying or being destroyed', but only vernieling exhibits the metonymical
extension towards the concept expressing the result of that process or that
36 and salience
act, i.e., the concrete damage that issues from it. (In the latter case, the word
is typically used in the plural: vernielingen more or less equals the notion
'damage'.)
In the fourth place, the internal structure of the set of personal applica-
tions reflects the differences in prototypical structure between both verbs. To
begin with, notice that the personal group contains concrete as well as ab-
stract applications; to kill someone is clearly more concrete than to under-
mine someone's psychological well-being or his social position. If we then
have a look at the mutual relationship between the abstract and the concrete
subgroups of the application with regard to persons, we find that the abstract
subgroup is proportionally dominant in the case of vernietigen, whereas the
reverse is true in the case of vernielen. Also, we find that extensions of the
concrete subgroup of the personal application with regard to other living
beings than people or with regard to personifications, are not as strongly
present in the case of vernietigen than in the case of vernielen. (It should be
added that these observations have to be considered with more care than the
previously mentioned points, since there is a general tendency throughout the
centuries covered by the WNT-material, to re move the personal application
from the structure of vernielen. There are relatively less personal applica-
tions in the structure of nineteenth-century vernielen than in the structure of
either nineteenth-century vernietigen or sixteenth-century vernielen; as has
al ready been mentioned, it is even more difficult to use vernielen with re-
gard to persons in present-day Dutch. In any case, the nineteenth-century
material does seem to show that the material subgroup of the personal appli-
cation of vernielen is more resistant to the tendency in question than the
abstract subgroup, as can be predicted from our centrality hypothesis.)
Finally, the importance of prototypicality can be derived from the fact
that different nuances play a central role within the core of each concept,
whereas those nuances are not particularly important within the corres-
ponding group in the other concept. Thus, the destruction of buildings and
other human constructions is prominent within the material use ofvernielen,
but is only rarely present within the material group of vernietigen. Within
the structure vernielen as a whole, demolishing buildings is the single most
frequently represented kind of usage, but within the structure of vernietigen,
it is merely one among many equally important nuances of the material set of
applications. In the same way, the central, abstract group within the struc-
ture of vernietigen is itself centered round applications relating to the disso-
lution, the cancellation, the annulment of agreements, commitments, en-
gagements, obligations, permissions, rights, and so on, and of the laws,
orders, contracts etc. in which they are contained and through which they
come into existence. Whereas vernielen only rarely exhibits this kind of
usage, it is the most frequently occurring sense within the abstract group of
vernietigen as well as within that word as a whole.
In general, these facts of linguistic usage clearly favor the hypothesis that
the abstract applications of the concept 'to destroy' are prototypical within
the structure of vernietigen, whereas the concrete applications are prominent
in the case of vernielen. Taking into account that each central group is itself
concentrated round a dominant kind of usage, it seems plausible to say that
the latter is the prototypical sense for each of the verbs in question. It should
furthermore be noted that these prototypical phenomena seem to be con-
nected with the etymology of the words. On the one hand, the abstract proto-
type ofvernietigen may well be connected with the abstract character of the
words niet 'not', and nietig 'null and void, insignificant', on which it is
based. Moreover, the common phrase nietig verklaren 'to declare something
to be null and void, dissolve, annul something' corresponds pretty closely
with the central notion within the abstract group ofvernietigen. On the other
hand, the centrality of the application with regard to buildings in the struc-
ture ofvernielen seems to correspond with the etymological meaning 'to tear
down, to throw to the ground' that we reconstructed above as the original
meaning of the verb.
4. Introspective evidence of prototypicality
Before we can deal with the introspective evidence in favor of the prototypi-
cality hypothesis, two preliminary questions have to be answered. In the first
place, how trustworthy is the introspective methodology? The paradoxical
fact of the matter is that it is exactly the unreliability of introspection that
makes it interesting for our purposes. If introspection were able to yield a
completely adequate picture of the facts of linguistic usage (which is doubt-
ful), it would simply reduplicate the results reached in the previous para-
graphs on the basis of a direct examination of linguistic usage. But given the
presupposition that introspection yields only a partial insight into the seman-
tic structure of the words that are investigated, we can also presuppose that
it will be exactly the prototypical kinds of usage of those words, that reach
the introspective consciousness of the language user. We can use the results
of the introspective method as support for the prototypical hypothesis if we
presuppose that prototypical kinds of usage (precisely because they are more
salient than other applications) will more easily pass the threshold of con-
scious attention. Given this presupposition, the introspective judgments of
native speakers may shed light on the question which kinds of usage are
predominant within a certain concept.
In the second place, how can the introspective method be used with re-
gard to historical material? There are no nineteenth-century speakers of
Dutch around to be asked what they think is the meaning of particular
words, so how are we going to get introspective judgments at all? The fact is
that we do have information on how the nineteenth-century speakers of
Dutch perceived the near-synonyms that we are investigating, viz. in the
form of synonym dictionaries. Synonym dictionaries (at least the older ones)
are notoriously unreliable as descriptions of actual patterns of usage; most
of the time, the compilers of synonym dictionaries rationalize away the ac-
tual identity of words by imposing distinctions that cannot be discovered in
the actual facts of usage. However, these rationalizations need not always
have proceeded out of the blue: it seems quite plausible that they were
guided by the introspective judgments of the compilers. So, if we like to
know something of the introspective insights of the nineteenth-century
speakers of Dutch, we can have a look at the synonym dictionaries of that
time to see whether the distinctions they make between vernielen and
vernietigen (however inadequate as a picture of the complete set of possible
kinds of usage) do indeed reflect the differences in prototypical structure of
both words.
And indeed, the nineteenth-century synonym dictionaries of Dutch do dis-
tinguish between vernielen and vernietigen along lines that fit into our hy-
pothesis. On the one hand, there are those that draw the line syntagmat-
ically, such as Weiland and Landre (1825), who state that vernielen can
only be used with regard to 'lighamelijke dingen' (material things), whereas
vernietigen is more widely used, in particular also with regard to 'menscheli-
jke instellingen' (human institutions). De Beer (1897) expresses an analo-
gous point of view. On the other hand, there are those that describe the dis-
tinction along paradigmatic lines, so that there would be an actual notional
difference between the verbs in question, rather than merely a distinction in
selectional restrictions. Whereas vernietigen is defined as 'to bring to
naught, to annihilate' , vernielen is defined as 'to damage, to smash to
pieces, to tear down'. In this sense, vernietigen implies a complete annihila-
tion whereas there may be some pieces left of the original object in the case
ofvernielen. It is easy to see that this paradigmatic point of view, which can
be found among others in Pluim (1894), is connected with the previous, syn-
tagmatic one: it is precisely because vernielen relates to material things that
the notion of remaining debris comes to the fore. Likewise, a complete anni-
hilation (in which the original objects disappear completely) is less likely in
the material world of concrete objects, so that the restriction of vernietigen
to abstract objects will tend to be related to the notion of complete annihila-
tion. This is in fact done by Weiland and Landre (1825), though not all pro-
ponents of the paradigmatic distinction adhere to the syntagmatic distinction.
For instance, De Flines (1810) mentions that verniefigen can in fact be used
with regard to material objects, but that there is a difference with vernielen
in the degree of damage achieved.
By and large, these views faithfully reflect the insight into the prototypes
of vernielen and vernietigen that we have gained by considering the actual
facts of linguistic usage. Syntagmatically, it is recognized that the material
context is more important for vernielen, whereas abstract objects are pre-
dominant in the case ofvernietigen. Paradigmatically, this is reflected by the
fact that vernielen carries overtones of material destruction and damage
(think of the relationship between the prototypical usage of vernielen with
regard to buildings, and the definitions of that word that bring to the fore the
act of smashing and demolishing things), whereas verniefigen calls forth the
idea of complete annihilation (as it were, wiping something off the face of
the earth). As such, the stubborn efforts of the compilers of synonym dic-
tionaries to find semantic differences among near synonyms seem to be not
entirely gratuitous. To the extent that they try to capture the characteristics
of the most salient kinds of usage of both lexical concepts, they strengthen
our own hypothesis about the differences in prototypical structure among the
verbs.
5. The functional explanation of prototypicality
There are a number of interesting conclusions to be derived from the above
analysis of the near-synonyms vernielen and vernietigen. First, pro-
totypicality is an interesting new point of view in the study of synonyms. It is
traditionally well-known in lexical semantics that there are relatively few
true synonyms in natural languages, and the ways in which near-synonyms
differ can be very diverse. Our discussion of vernielen and vernietigen
shows that there is one more factor to be added to the list of differentiating
factors: near-synonyms may be distinct with regard to the prototypical struc-
ture imposed on an otherwise identical range of application. Once again, the
importance of prototype theory for the traditional concerns of lexical seman-
tics becomes apparent (cp. Geeraerts 1983a, 1983c, 1984a, 1985a); proto-
type theory opens up new perspectives in the study of synonyms.
Secondly, there are some indications that introspective judgments in lexi-
cal semantics relate to the prototypically salient instances of concepts rather
than to the full range of actual usage possibilities. If this can be confirmed
by additional comparisons between introspective perceptions of lexical
meanings and actual usage patterns, more will be known about the value of
both methodologies (introspective and corpus-based) in lexical semantics.
Also, if we maintain the classical view of modem linguistics that it is one of
the goals of linguistic theory to account for the introspective judgments of
native speakers, and if these judgments appear to be influenced by prototypi-
cal phenomena, yet one more reason presents itself for incorporating proto-
type theory into lexical semantics.
Thirdly, the fact that vernielen and vernietigen have the same conceptual
and collocational range of application, and yet differ with regard to the core
and the periphery of their categorial structure, indicates that there are at
least some cases of prototypicality that cannot be explained by means of the
referential model. Vernielen and vernietigen refer to the same set of acts and
processes; as such, the differences in their prototypical structure cannot be
the automatic consequence of their referential range, as is implied by the
family resemblance hypothesis. In addition, the physiological and the statis-
tical explanation will not be of much avail either. There is no particular or-
gan or mechanism for the perception of processes of destruction, and even if
there were, we would still need two different physiological structures to ex-
plain the distinction between both verbs, which is beyond all intuitive plausi-
bility. The statistical explanation is inapplicable for the same reason as the
referential hypothesis: since the range of application of both verbs is the
same, the frequency of occurrence of the processes referred to is the same
for both verbs. That is to say, the frequency with which the demolishing of
buildings occurs in reality, relative to the frequency with which, say, agree-
ments are cancelled in reality, has exactly the same effect on both verbs,
since these refer to the same objective reality. Because they denote the same
things in reality, the structure of reality (either with regard to the frequency
of occurrence of its elements, or with regard to the mutual resemblances
among those elements) cannot be invoked to explain the distinction in seman-
tic structure between vernielen and vernietigen.
In short, we can reject all materialistic explanations of the prototypical-
ity effects observed in the verbs under consideration. Indeed, the physiologi-
Where does prototypicality come from.? 41
cal, the referential, and the statistical hypotheses have this in common: that
they try to explain prototypicality on the basis of materialistic data, either
the material structure of the human perceptual apparatus, or the material
characteristics (statistical or otherwise) of the referential range of the con-
cepts involved. Given that we have to reject these materialistic hypotheses,
we can provisionally choose, by elimination, for the psychological, func-
tional explanation of prototypicality. To support this choice, I would like to
make clear that the functional hypothesis has some additional advantages,
besides the fact that it avoids the problem of the materialistic hypotheses.
First, however, three remarks have to be made.
To begin with, it might be claimed that a statistical explanation of the
prototypicality effects in vernielen and vernietigen can indeed be given, if
we take into account, e.g., that the material sense occurs much more fre-
quently with vernielen than with vernietigen, or that the abstract specifica-
tion of the notion 'to destroy' is statistically much more prominent in the
latter verb than in the fornler. However, the frequencies that are mentioned
here are linguistic frequencies, not referential frequencies, i.e., they are fre-
quencies of occurrence of words, not of the things those words refer to. Be-
cause the frequency at stake here is linguistic rather than referential, it can
hardly be invoked to explain prototypicality; as an aspect of linguistic usage,
it is one of the things we have to explain, not one of the things that are them-
selves part of the explanation. We can use linguistic frequencies to determine
what instances of a concept are prototypical (that is what we did in section
3), but explaining prototypicality on the basis of linguistic frequency is put-
ting the cart before the horse. Some kinds of usage are not prototypical be-
cause they are more frequent; they are more frequent because they are proto-
typical. The apple is not a prototypical fruit because we talk more about
apples than about mangoes, but because we experience apples more often
than we encounter mangoes (and this fact, in turn, may be the reason why
we talk more about apples). Frequency of linguistic occurrence may be a
heuristic tool in the pinpointing of prototypes, but it is not the source of pro-
totypicality as meant in the statistical hypothesis.
The second remark has to do with the fact that criticism with regard to
the referential, family resemblance model of prototypicality has already been
formulated elsewhere. This has been the case in the work of Pulman (1983)
and - in more stringent fashion - in the well-known article by Armstrong,
Gleitman, and Gleitman (1983). They argue that gradience can be observed
in concepts with rigid boundaries (their examples relate to natural numbers),
so that family resemblances cannot be invoked to explain the differences in
42 and salience
salience among numbers. There are two reasons, however, why their argu-
mentation is less relevant than they assume. First of all, they more or less
equate prototype theory and the family resemblance model of the sources of
prototypicality, whereas it is quite clear that the family resemblance model is
merely one of a number of hypotheses concerning the sources of prototypi-
cality: ruling out one hypothesis does not mean that one can ignore the oth-
ers. And also, I do not think they are successful in presenting a counter-
example to the family resemblance model. Even if a concept has rigidly de-
fining characteristics, family resemblances may exist among the non-
defining characteristics of the instances of that category. Since Cognitive
Semantics is basically encyclopedist in its approach, these non-defining,
'encyclopedic' attributes should be incorporated into the computation of
degrees of shared attributes. As Lakoff (1982) has shown, such encyclope-
dic, experiential factors do indeed occur with regard to numbers, and they
can be used to explain the prototypicality ratings found experimentally.
My third remark is this: my criticism of the materialistic hypotheses
should not be overgeneralized. The fact that they do not work in the case of
vernielen and vernietigen clearly does not imply that they do not work in
any case, but merely makes clear that next to the physiological, the referen-
tial, and the statistical model, there will have to be at least one other source
of prototypicality.
Let us now come back to the functional model of prototypicality and try
to elaborate it. Remember that the psychological hypothesis involves re-
quirements that the cognitive system is to comply with if it is to function
efficiently: prototypicality exists because it is cognitively advantageous. As
we have seen, Rosch has specified this functional advantage in terms of the
economical effect of informational prototypical categories enable
one to reach the most information with the least cognitive effort. This func-
tional line of reasoning can be supplemented with some additional (and per-
haps even more fundamental) functional reasons for having prototypical
categories. We can base the discussion on one of the fundamental insights of
cognitive psychology, viz. that cognition should combine structural stability
with flexible adaptability. On the one hand, cognition should have a tendency
towards structural stability: the categorial system can only work efficiently if
it can maintain its overall organization for some time, if it does not change
fundamentally any time new information has to be incorporated. At the same
time, however, it should be flexible enough to be easily adaptable to chang-
ing circumstances. To prevent it from becoming chaotic, it should have a
built-in tendency towards structural stability, but this stability should not
Where come from? 43
become rigidity, lest the system stops being able to adapt itself to new and
unforeseen circumstances.
This necessity of flexibility is one of the aspects of lexical semantics that
was recognized by the prestructuralist tradition of historical semantics, but
that has been more or less lost in the meantime, as a result of the struc-
turalist attention for fixed synchronic structures. Be that as it may, it will be
clear that prototypically organized categories are particularly well suited to
fulfill the double demand for flexible adaptability and structural stability. On
the one hand, the fact that slightly deviant nuances can be developed within a
particular category indicates that categories have the dynamic ability to cope
with changing conditions and changing expressive needs. On the other hand,
the same fact (that marginally deviant concepts can be incorporated into
existing categories as peripheral instances of the latter) proves that these
categories have a tendency to maintain themselves as holistic entities, thus
maintaining the overall structure of the categorial system. Prototypical cate-
gories maintain themselves by adapting themselves to changing cir-
cumstances and new expressive needs; at the same time, they function as
expectational patterns with regard to reality: new facts are interpreted in
terms of information that is already at the disposal of the individual. The
flexibility of the cognitive system does not only show up in the fact that it
ran adapt itself to new experiences, but this flexibility is supplemented with
the fact that existing categories have a formative influence with regard to
experience; new experiences are fitted into the expectational patterns pro-
vided by the existing categorial system.
Along these lines, prototypicality appears to be the outcome of some fun-
damental, deep-seated principles of cognitive functioning. The form of the
conceptual system appears to be determined by a set of basic functional
requirements, and prototypically structured concepts admirably meet these
requirements. If this is correct, the same basic principles should also have a
role to play in other cognitive disciplines. That is to say, if prototypicality is
an emanation of some basic characteristics of all cognition, we should be
able to find analogies of the prototypical idea in other fields of cognitive sci-
ence, next to lexical semantics. I have tried to prove at length elsewhere
(1985a) that this is in fact the case: the importance of interpretative sche-
mata mediating between experience and existing knowledge is an idea that
can be traced in a number of cognitive disciplines. It is very much apparent
in Artificial Intelligence (Minsky's frame notion); it can be found in cogni-
tive psychology, particularly in the work of Bruner, and to some extent in
that of Piaget; it can be related to some of the views of the early, Husserlian
phenomenological movement in philosophical epistemology; and it has some
important similarities with the paradigmatic conception of scientific enquiry
inaugurated by Thomas Kuhn.
These are exciting parallels because they suggest that the functional, psy-
chological hypothesis concerning the sources of prototypicality can at the
same time be the basis for a truly integrated cognitive science in which the
insights of linguistics, Artificial Intelligence, cognitive psychology, philoso-
phical epistemology, and the philosophy of science can be brought together
under a common denominator. In this respect, the functional model of proto-
typicality, even if it does not rule out the possible importance of the physio-
logical, the referential, or the statistical explanation, does seem to be more
general than the latter, not just because it is based on fundamental principles
of cognition, but also because similar views have been put forward in other
branches of cognitive science.
6. Onomasiological and semasiological aspects of Cognitive Semantics
Unfortunately, the optimistic perspective of the previous paragraph does not
solve everything. To round oft the discussion, I would like to show that a
complete explanation of all questions to be raised with regard to vernielen
and vernietigen is far from available. The picture we have reconstructed so
far looks like this: apparently, the linguistic conmmnity at some point in its
development finds it convenient to have two distinct categories for the con-
cepts of material and abstract destruction. Thus, a pair of etymologically
distinct words becomes available, originally vernielen and vernieten, later
on vernielen and vernietigen; their compound character ensures that one of
them signifies material destruction, the other abstract annihilation. Gradu-
ally, the flexibility that is inherent in all human categorization extends these
concepts beyond their etymological usage; as a result, they have the same
range of application in the nineteenth century.
But now consider the original situation in which these flexible extensions
have hardly begun taking place. Is it then not irrational to use vernielen to
express abstract annihilation, when you al ready have vernietigen or
vernielen to do so? The question can be put in terms of global and local
efficiency. As we have argued, the global efficiency of the conceptual system
commands its flexible, prototypical organization. But there is, in the case of
vernielen and vernietigen, also a local efficiency principle that says that it is
uneconomic to have two terms expressing the same things. We are then
Where does proto(vpicality come from? 45
forced to ask: why does not the local efficiency principle stop the application
of the global principle? Why is not the prototypical extension of vernielen
towards abstract forms of destruction checked or prevented by the considera-
tion that you already have a lexical category expressing abstract destruction?
There is yet another way of formulating the problem: prototype-based flexi-
bility is necessary because of the expressive needs of the speaker: he may
want to express concepts for which no specific term is available. But why
then would he use these flexible mechanisms of semantic extension if such a
specific term is in deed available?
One kind of answer might simply be that the global principle is stronger
than the local principle; the global principle simply supersedes the local prin-
ciple to the extent that local inefficiencies are created. We are then saying
that the global principle is so general that its strength overrules the local
principle, and that it applies even where it is not strictly necessary. Still, this
does not tell us why the local principle is weaker than the global principle.
Also, it is rather awkward to explain a mechanism that is unfunctionally
overproductive on the basis of functional considerations. Couldn't we there-
fore find a more rational explanation of the flexible extensions?
The way out, as far as I can see, is to take into account other kinds of ex-
pressivity than the purely conceptual one. Using vernielen to express a con-
cept that is commonly expressed by vernietigen may be conceptually super-
fluous, but that does not mean that doing so may not serve particular
expressive purposes. On the level of the linguistic form, for example, it may
be quite functional to use another word than the usual one. The varieties of
such a formally expressive synonymy are well-known in traditional lexical
semantics; near-synonynls may exhibit connotational and emotional differ-
ences (as in euphemisms), stylistic differences (as in popular words versus
poetic terms), or sociolinguistic differences (as in learned words versus
common words). Perhaps we can even say that speakers have an urge for
stylistic variation as such, even if the fornlal variants do not carry specific
overtones; variation may well be governed by a straightforward desire to
avoid monotony, to create new ways of expressing oneself, to experiment
with unexpected innovations as a way of stressing one's own individuality.
Moreover , it may well be that the importance of metaphor in natural lan-
guage is determined precisely by its stylistic expressivity; metaphorical ex-
pressions would then be created primarily to add expressive weight to the
message one wants to convey. (See Rudzka-Ostyn 1988.)
It is quite plausible, then, that factors such as these have governed the ex-
tension of vernielen and vernietigen beyond their original meanings and into
46 and salience
each other's etymological range of application. For instance, using vernielen
to express a process of abstract cancellation may have been stylistically
particularly expressive, because the process of material destruction normally
denoted by vernielen carried overtones of physical violence that were less
marked in the case of vernietigen. The extended use of vernielen would then
have been a case of metaphorical hyperbole. It is, however, very difficult to
pinpoint exactly which form of expressivity is the relevant one with regard to
the two verbs that we are concerned with here; our historical material for the
earliest (Middle Dutch) history of vernielen and vernieten, for instance, is
very hard to interpret with regard to such questions. Still, some clear cases
may in fact be found. For instance, the first quotation of (5) clearly carries
more overtones of violence, force, and intensity than the second quotation in
that pair of examples. (This is mainly made apparent by the presence of a
simile, marked by alzoo.) This suggests that the verbs highlight slightly dif-
ferent aspects of the situation described, or rather , represent the situation
from different points of view (determined by the prototypical core of each
verb). (In Langacker' s terminology, the distinction between the two verbs,
when used with regard to the same process, might then be characterized as a
figure/ground-distinction: vernielen takes the violent process as figure, and
vernietigen the destructive result.) This is not an altogether implausible hy-
pothesis, but it is unfortunately hard to confirm for the simple reason that
the historical texts used here do not give us enough clues to discern such
subtle differences in stylistic or emotional overtones.
On the whole, then, what can we conclude from our discussion of this
additional problem? On the one hand, it inspires caution with regard to our
attempts to explain prototypical phenomena: the linguistic materials at our
disposal do not always allow completely satisfactory answers with regard to
the questions at stake to be formulated. On the other hand (and this is, I
think, the more important conclusion), the discussion suggests that prototype
formation may be influenced by other factors than purely conceptual ones.
Stretching the meaning of a lexical item may be motivated by the desire to
use another form than the one that is usual to express the idea in question;
stylistic, sociolinguistic, connotational expressivity rather than purely con-
ceptual needs may determine the flexible use of a category. In such a case,
the conceptual coherence of the prototypically structured category (i.e., the
fact that the new, peripheral kinds of usage have to be accessible from the
prototypical core) constitutes a limit to the desire for formal variation: you
can use a particular lexical item to express an idea that is usually signified
by another word, but only on the condition that the idea in question is part of
Where does come from? 47
the prototypical potentialities of that lexical item. Basically, you stretch an
item's meaning to express something conceptually new, but you can also
stretch it to express something conceptually old in a formally new way. This
is a very important suggestion, because it implies a warning against a ten-
dency that is a natural characteristic of Cognitive Semantics: the tendency,
in fact, to look for purely cognitive or conceptual explanations of the facts
one encounters. Taking the cognitive, experiential, encyclopedic nature of
linguistic signs seriously should not imply looking only for strictly concep-
tual explanations. Language is not just content: it is also fornl, and its fonllal
side has an expressivity of its own, which does seem to create lexical con-
figurations that can hardly be explained if we only take into account the
conceptual expressivity of language.
In the traditional terms of lexical semantics, this means that the expla-
nation of prototypicality should not restrict itself to the semasiological per-
spective (in which each category is considered on its own), but that the ono-
masiological point of view (in which it is studied how several items may
express similar or identical concepts) should be taken into account as well.
(conceptual expressivity is basically a factor connected with the semasiolog-
ical explanation of prototypicality, whereas the onomasiological influences
on prototype formation seem to refer to other kinds of expressivity, as was
suggested by our study of vernielen and vernietigen. The incorporation of
the onomasiological approach does not mean that Cognitive Semantics
moves away from the functional perspective advocated in the previous sec-
tion; non-conceptual expressivity is just as much a functional principle as
purely conceptual expressivity and cognitive efficiency. Rather, the incor-
poration of onomasiology implies that Cognitive Semantics moves much
closer to the rich tradition of lexical semantics, in which onomasiological
mechanisms and configurations have been thoroughly studied (see Geeraerts
1986a: Chapter 1). Such a link with traditional approaches can only streng-
then the linguistic attractiveness of Cognitive Semantics.
To summarize: I have tried to argue, on the basis of a case study involv-
ing the Dutch near-synonyms vernielen and vernietigen, that the functional
point of view is the most encompassing, most promising one for studying
prototype formation, though it should not be restricted to purely conceptual
expressivity and efficiency, but should also take into account the kinds of
functional mechanisms that have traditionally been studied by the onomasi-
ological approach to lexical semantics. The fact, however, that pro-
totypicality may come from a number of diverse sources, also implies that an
adequate explanation of conceptual structures will not be easy.
Chapter 3
The semantic structure of Dutch over
Originally published in Leuvense Bijdragen. Leuven Contributions in Linguistics
and Philology 1992, 81: 205-230.
The preposition over plays a role in Cognitive Semantics that is somewhat compa-
rable to that of bachelor in Katzian semantics: from Brugman (1981, 1988) over
Vandeloise (1990), Cuyckens (1991), Deane (1992), and Dewell (1994), to Evans
and Tyler (2001), Tyler and Evans (2003), it has been a rallying-point for com-
paring competing forms of semantic analysis. The present chapter reproduces my
own go at the subject. Although 1 would now try to firmly base the analysis on
corpus materials rather than on introspectively derived sample sentences, I believe
that the basic point of the paper still holds: the semantic structure of lexical items
like over has to be seen as a multidimensional structure, in which covariation of
semantic shifts along different dimensions constitutes the essential backbone of a
prototype-based network. Specifically, the semantic structure of over involves at
least two dimensions: on the one hand, the spatially relational dimension in the
strict sense (with regard to which three distinct spatial configurations have to be
distinguished), and on the other, the motional dimension, with regard to which we
have to distinguish between cases of real motion, fictive motion, and zero motion.
A third dimension to be added involves the existence of actual contact between the
prepositional object and the other entity involved in the spatial relation.
This type of analysis contrasts with the more common type of radial set analysis in
Cognitive Linguistics, which is much more atomistic: a classical radial set analy-
sis focuses on individual readings and the links that connect pairs of such read-
ings, somewhat neglecting the underlying structure of dimensions that link vari-
ous readings among each other. Such a focus on the underlying multidimen-
sionality may help to avoid the pitfalls of the polysemy/univocality debate. There
are (as argued in Chapter 5) convincing methodological reasons for being careful
with the question: the criteria for answering it are neither stable nor consistent.
But a practical, descriptive exercise such as the analysis of over may inspire the
same conclusion. If you think of semantic structure primarily in terms of individ-
ual readings, it is natural to ask the question 'How many meanings does a word
have?' (cp. Taylor 1992), i.e. is this a monosemous or a polysemous, or a hyper-
polysemous word? If, however, one recognizes that covariation of different dimen-
sions is a crucial characteristic of the structure in question, the individual readings
(and the number they have) become less important from a practical and theoretical
point of view.
The semantic structure ofDutch over 49
1. Scope and purpose ofthe exercise
The English spatial preposition over occupies an important position within
Cognitive Semantics. On the one hand, Brugman's analysis (1981), which
also appears with slight modifications in Lakoff (1987), is one of the earliest
examples of a radial set analysis of lexical semantic structures. On the other
hand, Vandeloise's reaction against the Brugman/Lakoff-analysis (1990)
raises a number of fundamental questions concerning the nature of polysemy
and the description of polysemy within a Cognitive framework. In the fol-
lowing pages, a similar methodological point concerning lexical-semantic
research in Cognitive Semantics will be dealt with on the basis of an analysis
of over. However, by presenting an alternative for the treatment in Cuyckens
(1991), the paper will focus on the Dutch item over rather than on its Eng-
lish counterpart. Also, the theoretical focus will not be on the possible uni-
vocality of over (as discussed by Vandeloise), but on the importance of a
detailed definitional analysis oflexical items.
Two points will be discussed with regard to which Cuyckens's analysis
would seem to be subject to improvement: a moderate lack of comprehen-
siveness, and an insufficiently specific treatment of the links between the
various senses of over. Both of these characteristics seem to follow from a
tendency (not altogether surprising in the framework of a prototype-
theoretical approach) to concentrate on the most salient instances of use of a
category: on the one hand, there will be less attention for the non-typical
cases; on the other, the salient cases will tend to be seen as self-contained
entities. In contrast with such a synthetic conception of the salient instances
of a category, an analytical alternative will be presented that not only tries to
determine the conceptual relationship between the various instances of a
category as precisely as possible (specifying the various attributes that link
the instances to each other), but that also tries to make sense of less obvious
cases.
There are a number of restrictions on the following treatment of over that
should be mentioned explicitly. Only the prepositional uses of over in the
strictest sense will be taken into account; predicative and adverbial uses (de
pUn is over, de pUn gaat over), or the use of over in compounds (over-
blijven, overdenken, overgaan etc.) are beyond the scope of the present
treatment. The discussion will be limited to the spatial senses of the preposi-
tion; figurative and temporal uses (een disGussie over iets, een vergadering
over drie dagen) will not be envisaged. Further, a lexical field approach
(which would involve comparing over with related prepositions, specifically
with boven) will not be systematically adopted (although there will be inci-
dental remarks about synonymy and near-synonymy). Likewise, there will be
no contrastive analysis (with the BrugmanlLakoff-analysis of English over,
or with German iiber, which combines the readings of over and boven).
Since these are all legitimate topics for investigation that should really be
taken up in a more extensive treatment of the preposition, it will be clear that
the restrictions are a matter of convenience rather than principle. Like
Cuyckens's analysis, the discussion focuses on the core of the semantic
structure of over, which would be the basic step in a more detailed descrip-
tion at any rate.
There is, however, one restriction on the present treatment of over that
does involve matters of principle rather than convenience. In fact, the paper
will remain agnostic about the question whether the various kinds of usage
that will be described, constitute different meanings or not (in the theoretical
sense in which 'meaning' is distinguished from 'vagueness'). How, in fact,
to determine whether a particular instance in which a lexical item is used
constitutes a different meaning of that item? As argued in Geeraerts (l993a),
not only are there various kinds of semantic identity tests that might be in-
voked (and that have been invoked) to settle the question, but also, these
various tests appear to yield mutually contradictory results. To say the least,
linguistic semantics has not come up with a universally accepted, unequivo-
cal operational test for distinguishing polysemy from mere referential vague-
ness. Against the background of this methodological indetenninacy, the pre-
sent paper will not endeavor to establish the meanings of over (if any such
set of clearly delimited semantic entities exists at all), but will rather pursue
a more modest goal; it will try to describe the structure of the referential
range of application of over in some detail. Which portions of the structure
constitute a different meaning in the theoretical sense is a question that is
beyond the intended scope of this paper - but conversely, it is a question that
can probably only be answered properly on the basis of a referential analysis
of the type to be demonstrated here.
Finally, there is a restriction on the originality of the following analysis.
Although it will try to go one step further than Cuyckens' s analysis by pre-
senting an alternative for the overall structure of the item, it also draws
heavily on his work when individual applications are at stake (just like
Cuyckens himself has drawn on the work of Brugman and Lakoft). The
distinctions and correspondences will not be mentioned separately each time
they appear, but a global comparison with Cuyckens' s proposals will be
made in paragraph 6.
2. The basic referential structure
The absence of a generally applicable operational criterion for meaning dis-
crimination does not imply that the existing approaches could not be used as
a starting-point for the kind of structural analysis to be illustrated here. For
instance, lexical polysemies could at least in some context show up as a
sentential ambiguity. To take over eruse' s example (1982), the sentence (1),
when uttered at the meeting of a political committee, may mean either that
delegate Balder has taken another seat in the conference room, or that he has
changed his opinion on the topic at hand. This ambiguity, then, testifies to
the polysemy ofposition.
(1) Bafder has shifted his position!
(2) Wafdemar fletst over de fijn
'Waldemar rides his bicycle over the line'
Now consider (2).2 Sentence (2) may refer to at least three different situa-
tions. First, Waldemar may be riding across the line (as when he moves
from one lane to the other). Second, he may be riding on and along the line,
staying roughly within its boundaries as he follows it. (He might be doing
this as a game, or as a kind of test to enable the police to ascertain whether
he is riding under the influence of alcohol). And third, Waldemar may be
riding at the other side of the line (seen through the eyes of an implicit ob-
server). For instance, imagine a group of children riding their bikes at a
schoolyard, but being forbidden to cross a particular line. Sentence (2) could
then be used to point out that Waldemar ignores the prohibition. Notice that
the implicit observer turns this third reading into an example of subjectifica-
tion in the sense of Langacker (1990b): the spatial relation that is expressed
by over holds between the bicycle (or at least, between the act of riding the
bicycle) and the line, but it is construed with an implicit reference to the
subjective presence of an observer.
In Figures la, 1b, and lc, the triple ambiguity of (2) is graphically repre-
sented. Accordingly, over can be assigned three distinct readings, as defined
in [i]-[iii]. With regard to example (2), the definitions specify the spatial
characteristics of the activity referred to by the verb fietsen: it is an activity
that extends from one point to another (either within the spatial area defined
by the prepositional object x, or by crossing that area), or an activity that is
situated in a region construed as being to the other side ofx.
[i] Extending from a point or region at one side ofx to a point or region at
the other side of x, while crossing x
[ii] Extending from one point or region within the space defined by x, to
another point or region within that space (without going outside of it)
[iii] Situated at the other side of x, relative to the position of an observer
1a
~ , .. _--
1c
Figure 1. The basic semantic configurations of over
1b
Now, it is crucial for the semantic structure of over that the three spatial
relations as just defined can also occur in other syntagmatic contexts than
the one represented by (2).
First, the prepositional phrase can indicate the spatial extension of a fic-
tive rather than a real motion. In (3), for instance, there is no real motion of
the line; still, the line extends from a point on one side of the prepositional
object x to a point to the other side of that object (which motivates the use of
over), and further, this extension is construed, through the use of lopen, as a
process of fictive motion.
3
Second, if over can be used to express stative
spatial relations, it should be possible to combine it with stative verbs that
do not express motion. An example is (4), where it is said that the blanket
extends from one side of the fence to the other; it thus covers the fence, if
one likes. Example (5) presents another syntagmatic context in which the
'zero motion' option is realized; if the prepositional phrase modifies a noun
rather than a verb (and if the noun does not refer to a process or an action),
the interpretation is automatically stative.
(3) Er liep een rode streep over het woord verantwoort
'A red line ran across the word verantwoort' (which contains a spelling
mistake)
(4) De deken hangt over de schutting
'The blanket hangs over the fence'
(5) De deken over de schutting is van Godelinde
'The blanket across the fence belongs to Godelinde'
(6) Het vliegtuig vliegt over de stad
'The airplane flies over the town'
It will be clear by now that the semantic structure of over involves at least
two dimensions: on the one hand, the spatially relational dimension in the
strict sense (with regard to which three distinct spatial configurations have to
be distinguished), and on the other, the 'motional' dimension, with regard to
which we have to distinguish between cases of real motion, fictive motion,
and zero motion. A third dimension to be added involves the existence of
actual contact between the prepositional object and the other entity involved
in the spatial relation. As (6) exemplifies, the reading defined in [i] can also
occur when there is no physical contact between trajector and landmark. (In
fact, (6) could be shown to be ambiguous along the three readings [i]-[iii]).
On the basis of these three dimensions, the referential range of application of
over can be charted systematically. As Table I and the examples (7)-(22)
show, almost all possible combinations of all values on the three dimensions
actually occur.
(7) De haf roft over de fijn
'The ball rolls over the line' (to the other side of it)
(8) Het vliegtuig vliegt over de stad
'The plane flies over the town' (to the next one)
(9) Er liep een rode streep over het woord verantwoort
'A red line ran across the word verantwoort'
(10) De spoorlijn loopt over de rivier
'The railway line runs across the river'
(11) De deken hangt over de schutting
'The blanket hangs over the fence'
(12) Er hangen donkere wofken over de stad
'Dark clouds hang over the town'
(13) Engefram kuiert over de markt
'Engelram strolls over the market-place'
(14) O([ert rijdt over de fijn
'Olfert rides over (on and along) the line'
Table I. Combinations of values in the semantic structure of over
MOTION FICTIVE MOTION ZERO MOTION
+contact -contact +contact -contact +contact -contact
[i1
(7) (8) (9) (10) (11) (12)
[ii1 (13)(14) (15) (16) (17) (18) (19)
[iii1 (20) (21) (22)
(15) Het vliegtuig cirkelt over de stad
'The airplane circles above the town'
(16) Condensatiestrepen lopen kriskras over de blauwe lucht
'Condensation trails run haphazardly over the blue sky'
(17) Ermingards blik dwaalde over het blad
'Ermingard's eye wandered over the sheet of paper'
(18) Er figt een armoedig kleedje over de vloer
'A shabby rug lies on the floor'
(19) Een dunne streep mist hangt over de rivier
'A small trail offog hangs over (above and along) the river'
(20) Over de rivier exerceerde de vijandelijke mifitie
'Across the river, the enemy militia was exercising'
(21) Over de rivier loopt een rij bomen fangs de beek
'Across the river, a row of trees runs along the brook'
(22) Dordrecht figt over de Moerdijkbrug
'Dordrecht lies across the Moerdijk bridge'
(23) Over de brug hangt een ba//on
'Across the bridge, a balloon hangs in the air'
Three refinements are necessary to complete this picture of the basic range
of application of over. First, it will be noted that the distinction between <-
contact> and <+contact> cases is basically irrelevant for the examples of
[iii], because the position of the trajector x to the other side of the landmark
(the prepositional object) automatically implies that there is no contact be-
tween them. It is, of course, possible to interpret the distinction between (22)
and (23) as involving a <-contact> versus <+contact> situation, but because
the contact in question does not involve the bridge (which is the landmark for
the relation expressed by over) but rather the ground in general, the distinc-
tion does not affect the structure of over. In this sense, then, the gaps in the
structure represented in Table 1 are systematic, not incidental ones.
As a second remark, it should be established that the distinction between
[i], [ii], and [iii] is not just one between, respectively, a 'directional' sense of
over, a 'path' reading, and a 'locational' interpretation. Judging only on the
basis of the leftmost column of Table 1, it would seem possible to say that
over specifies either the direction of a particular movement, the path along
which a movement takes place, or the place where something is located.
However, it should be clear from Table 1 as a whole that the <zero motion>
readings of [i] and [ii] are just as locational as [iii] as a whole.
And third, let us notice that there is a particular complication with resul-
tative readings of the verbs, when a location referred to by over is presented
as the result of the movement referred to by the verb. The problem can best
be introduced in connection with transitive causative constructions. In such
constructions, the moving object is not the subject of the verb, but an entity
on which the verbs exerts a direct or indirect force. To begin with, consider
cases of the causative construction with what we have called the <real mo-
tion> case of [i], as in (24)-(25). In these cases, the distinction between [i]
and [iii] fades away, given that there are various possible paraphrases for
the causative construction: while [i] characterizes the directional path of the
movement of the affected entity, [iii] simultaneously characterizes the resul-
tative final state of that movement - and both ways of reading the causative
construction seem equally appropriate. Thus, (24) could be paraphrased as
'Lidewij causes the ball to move over the fence' (i.e. from one side of the
fence to the other, which brings into play the directional reading [i] of over),
but also as 'Nimrod causes the ball to reach a position over the fence' (i.e. at
the other side of the fence, which activates the locational reading [iii]). If we
call the latter interpretation the resultative one, the examples (24)-(25) would
seem to suggest that the locational resultative case is epiphenomenal with
regard to the directional reading of over.
(24) Lidewij gooit de bat over de schutting
'Lidewij throws the ball over the fence'
(25) Fiobert duwt de fiets over de markt
'Fiobert pushes the bicycle over the market place'
(26) Evermoed liet de deken over de schutting vallen
'Evennoed dropped the blanket over the fence' (in the sense that it
hangs over the fence)
(27) De deken va!t over de schutting
'The blanket falls over the fence' (in the sense that it falls on top of it
and subsequently hangs over it)
56 and salience
In (26), however, the movement of the affected entity cannot be described
in terms of any of the <real motion> cases charted in Table 1, although the
resultant state is a simple case of the <zero motion> application of [i]. Con-
sequently, the resultative cases cannot always be reduced to a by-product of
a directional reading, and the <zero motion> label in Table 1 should be read
so that it includes resultative constructions like in (26). At the same time, we
can now also see from (27) that such resultative constructions are not re-
stricted to transitive causatives. The 'resultative' construction undoubtedly
deserves to receive a closer look than the one it has been subjected to here; in
the context of the present paper, however, it will play only a minor role.
3. Sample syntagmatic restrictions
The survey of the range of application of over given in the previous para-
graph has to be further elaborated in two different ways. On the one hand, it
has to be supplemented with some peripheral cases of semantic relations
(which will be the subject matter of paragraph 4). On the other, it has to be
refined by an investigation into the syntagmatic restrictions that the various
applications of over are subject to.
In fact, not any kind of trajector can be associated with any kind of
landmark by means of over. Notice, to begin with, that the <real motion>
case of [i] can be used without restrictions. Or at least, objects, surfaces, and
lines can move over each other in all possible trajector/landmark combina-
tions, as shown in Table 2 and examples (28)-(45). (Within each cell of the
figure, the first example is a <+contact> one, the second a <-contact> onel
(28) Berthilde jiefst over de heuvel naar huis
'Berthilde bicycles home over the hill'
(29) Fredegonde jietst over het plein naar het begin van de Agnietenstraat
'Fredegonde bicycles over the square to the beginning of the Agnieten-
straat'
(30) Dietbaldjietst over de witte lijn
'Dietbald crosses the white line with his bike'
(31) De trein stoomde over de heuvel
'The train steamed over the hill'
(32) De trein stoomde over de vlakte
'The train crossed the plain'
(33) De trein stoomde over de grens
'The train crossed the border'
Table 2. Trajector-landmark combinations in the structure of over
LM OBJECT LM SURFACE LM LINE
TR OBJECT (28)(37) (29)(38) (30)(39)
TRSURFACE (34)(40) (35)(41) (36)(42)
TR LINE (31)(43) (32)(44) (33)(45)
(34) Een Romeins legioen marcheerde over de heuvel
,A Roman legion marched over the hill'
(35) Een Romeins legioen marcheerde over het plein
(36) Een Romeins legioen marcheerde over de grens
(37) Het vliegtuig vloog over het monument
'The plane flew over the monument'
(38) Het vliegtuig vloog over de vlakte naar de stad in het gebergte
'The plane flew over the plain towards the mountain town'
(39) Het vliegtuig vloog over de rivier
'The plane flew over the river'
(40) De wolk dreef over het monument
'The cloud drifted over the monument'
(41) De wolk dreef over de vlakte
(42) De wolk dreef over de rivier
(43) Een rij vliegtuigen vloog over het monument
'A row of airplanes flew over the monument'
(44) Een rz/ vliegtuigen vloog over de vlakte
(45) Een rij vliegtuigen vloog over de rivier
(46) Het vliegtuig cirkelt over de stad
'The airplane circles over the town'
In passing, it could be noted that the classification of an entity as either ob-
ject, line, or surface can be a matter of perspectival construal rather than
matter-of-fact objectivity. To give just one example, a town could be seen as
a point-like object, but it can also be conceived of as a surface. In a sentence
like (46), for instance, assigning [ii] to over automatically entails thinking of
the town as a surface: because the movement of the plane has to remain
within the (vertically projected) boundaries of the town, thinking of the town
as a two-dimensional surface is the natural thing to do.
Now, without systematically exploring the restrictions on all the possi-
bilities mentioned in Table 1, let us consider two major examples of limita-
tions on the combination of trajector and landmark. Importantly, the restric-
58 PrototypicaliZv and salience
tions will turn out to be an automatic consequence of the spatial relations
defined by [i]-[iii]. (Later on, we shall come across an example of restric-
tions that are truly ad hoc.)
To begin with, let us consider [ii]. Because the spatial relation referred to
by over has to remain within the boundaries of the landmark, the preposi-
tional object has to have a line- or surface-like extension: it should be possi-
ble to conceive of it as a line or a surface. This shows up quite clearly when
considering three-dimensional objects. Whereas (47) and (48) are straight-
forward cases of lines and surfaces, (49) is an example of a three-
dimensional object whose third dimension (height) does not prevent it from
having a bounded surface that allows for an over-relation. In (50), however,
the three-dimensional object's vertical dimension is so important that no
surface for aimless wandering can be imagined. As (51) shows, reading [i]
of over does not impose this restriction. This does not mean, however, that
an object such as schutting could not occur at all with reading [ii] of over.
Example (52) shows that the relevant feature is not the objective horizontal
or vertical orientation of the surface, but rather the functional characteristic
of 'treadability': what is not a surface for free movement for human beings
is a natural path for insects. Furthermore, (53) shows that (51) is actually
ambiguous between [i] and [ii]5 Examples (51) and (53) are possible with
[ii] if a situation is envisaged in which someone crawls along the top of the
fence; in that case, the relevant character of the landmark 'fence' is not that
of a surface, but that of a line. That is to say, any interpretation of (51) and
(53) on the basis of[i] would have to link up with (47) rather than (48).
(47) Trudofietste over het pad
'Tmdo rode his bicycle along the path'
(48) Sulfrida kuierde over de markt
'Sulfrida strolled over the market place'
(49) Kunigonde dwaalde over de berg
'Kunigonde wandered over the mountain'
(50) *Dankmar dwaalde over de schutting
'Dankmar wandered over the fence'
(51) Gradolfklom over de schutting
'Gradolf climbed over the fence'
(52) De spin kroop maar door over de schutting
'The spider kept on crawling over the fence'
(53) Odilde kroop maar door over de schuffing
'Odilde kept on crawling over (on and along) the fence'
(54) De trein glijdt over de monorail
'The train glides over the monorail'
(55) Een kolonne soldaten marcheerde over de heuvel
'A line of soldiers marched over the hill'
(56) Een rij pukkels over de wang van Adelheid is voor haar een bron van
veelzorgen
'A row of pimples on Adelheid' s cheek is a source of much concern to
her'
(57) Een rij bomen loopt schuin over het veld
'A row of trees mns diagonally over the field'
On the other hand, because the trajector of an over-relationship according
to [ii] has to remain within the boundaries of the landmark, it will typically
be smaller than the landmark - small enough, at least, to move about on the
surface or along the line constituted by the landmark. Specifically in the case
of line-like landmarks, however, this does not imply that the trajector has to
be smaller than the landmark in all directions. Taking into account that real
lines are never merely one-dimensional, but always have a certain width as
their second dimension, the trajector does not have to be narrower (but it
does have to be shorter). As example (54) shows, the moving object may be
wider than the actual line.
Two additional remarks have to be made at this point. First, the distinc-
tion between line-like and surface-like landmarks in combination with [ii] is
not unimportant, because it correlates with a restriction on the kinds of
movement that are possible with reading [ii]. Line-like landmarks, in fact,
automatically impose an orientation on that movement. Because the move-
ment goes from one point on the line to the other, and because the line re-
stricts the sideward movement of the trajector, the movement follows the
line. Automatically, the line defines the path of the movement. Surface-like
landmarks do not define such a path: the movement from one place to an-
other on the surface may be random (as in (49), where there is a suggestion
of aimless wandering or loss of orientation), but it may also be a straight
line. Because, in the latter case, the straight line will typically lead from one
side of the surface to the other side, there is a really close relationship with
the 'crossing' reading [i]: going over a surface according to [ii] is then the
same thing as being in the middle of the process of going over that surface
according to [i].
As a second remark, let us note that an example like (55) shows that a
line-like trajector may be discontinuous, or, if one wishes, virtual. If we
jump to the <fictive motion> and <zero motion> applications of [i] for a
moment, further examples may be found of such virtual/discontinuous tra-
jectors, as in (56) and (57). Needless to say, examples such as these again
60 Proto(ypicali(y and salience
underline the importance of perspectival construal in natural language: spa-
tially disconnected entities are conceived of as a single entity.
Let us now turn to a second major example of syntagmatic restrictions.
The restrictions that hold for the <zero motion> reading of [i] are to some
extent the reverse of those just described. With regard to [ii], we noted that
the landmark was necessarily line-like or surface-like, and that the trajector
had to be smaller (in the relevant dimension) than the landmark. In the case
that we are concerned with here, the trajector has to be construed as line-like
or surface-like, and the landmark has to be smaller (in the relevant dimen-
sion) than the trajector. On the one hand, the trajector of the <zero motion>
cases of [i] has to have a line-like or surface-like extension of its own. Be-
cause the spatial extension that is expressed by the prepositional phrase does
not involve the path of a movement (of an entity of any dimensionality), but
rather attaches to the entity itself, the trajector cannot be construed as point-
like. On the other hand, because the trajector has to extend from one side of
the landmark to the other, it has to be larger than the latter.
In the case of line-like trajectors, this means that the landmark is smaller
than the trajector in the direction in which the trajector extends. (As the ex-
amples (58) and (59) illustrate, the landmark may very well be larger than
the trajector in the direction orthogonal to the trajector's.6)
(58) Er ligt een kabel over de weg
'There's a cable lying over the road'
(59) Er hangt een deken over de schutting
'There's a blanket lying over the fence'
(60) Een deken ligt over het bed
'A blanket is lying across the bed (covers the bed)'
(61) Heriswitha legt een vel papier over de vlek
'Heriswitha puts a sheet of paper over the blot'
(62) Het rode blad ligt over het gele blad
'The red sheet lies over the yellow one'
(63) Ricfridus heefl sproeten over heel z 'n gezicht
'Ricfridus has freckles all over his face'
(64) Er ligt een hoop zand over het emmertje
'A heap of sand covers the little bucket'
(65) De kaasstolp Iigt over de Camembert
'The cheese cover lies over (covers) the Camembert'
In the case of surface-like trajectors, the trajector may extend from one
side of the landmark to the other in various directions, but it should do so in
at least one direction. At this point, we come across a phenomenon of the
same kind as the one we encountered when we discussed line-like landmarks
in our first example of syntagmatic restrictions. Just like <real motion> ap-
plications of [ii] with line-like trajectors yield the interpretation 'following
along x', <zero motion> applications of [i] with surface-like trajectors may
result in the interpretation 'covering x', as in (60) and (61).7 Another special
case of surface-like trajectors and landmarks is when they are both the same
size. When you have a yellow sheet of paper and a red one of precisely the
same size, (62) is possible when one sheet lies precisely on top of the other.
This is a special case because it is situated at the borderline between [i] and
[ii] (both taken in the <zero motion> interpretation, of course): according to
[i], the top page should be larger, whereas it should be smaller than the bot-
tom page according to [ii]. Equal size is so to speak the upper limit of [ii]
and the lower limit of [i]. However, because it would be natural to say that
the top sheet covers the bottom one, and because the 'covering' interpreta-
tion is a usual one with [i], there would seem to be at least one indication
that special cases of the 'equal size' kind are intuitively conceived of along
the lines of [i] rather than [ii]. Note also that the equal size cases also occur
with 'virtual' examples that are analogous to (56) and (57). In an example
like (63), the virtual surface defined by the freckles is implied to have the
same size as the face; characteristically, the freckles cover the face.
To round off, let us have a look at (64) and (65). Because the trajectors
in these cases are neither clear examples of lines or of surfaces, it might
seem that they contradict the earlier statement that the trajectors of the <zero
motion> interpretation of [i] are construed as line-like or surface-like. One
way out of the problem would be to claim that the relevant 'surface' in these
cases is the vertical projection of the maximal expansion of the trajectors;
and another would be to point out that surface-like trajectors may involve
curved and bended surface. However, rather than trying to salvage the letter
of the original statement of the restriction, it would be more in line with the
spirit of the original statement to accept that it has to be slightly modified:
the basic point is to see that point-like trajectors are excluded (because they
have a minimal extension), but this does not rule out object-like trajectors
that are construed as masses or spheres rather than points.
4. Some less systematic cases
Three cases have to be mentioned that fall outside the range of applications
charted in Table 1. (Each time, the restrictions on the additional cases will
62 and salience
turn out to be ad hoc rather than an automatic consequence of the spatial
relations expressed by over, as in the previous paragraph.)
The most common case is illustrated by (66)-(69); the meaning expressed
by over can be defined as in [iv]. This kind of usage can most easily be seen
as a transformation of those that fall under [iii]; specifically, the preposi-
tional object is now the starting-point of a virtual path rather than the path
itself. In contrast with [iii], the path is necessarily an open area such as a
street or a square (or even a dinner table) rather than just any obstacle that
can be crossed: examples such as (66) are ruled out in a situation in which
the museum is situated at the foot of a hill and it is your intention to make
clear that Gosbert lives over the hill. This is to say that this kind of usage is
subject to specific restrictions that do not follow automatically from the
spatial relation that is being expressed. The reading of over at stake here is
not just a metonymic transformation of [iii], but it is subject to additional
restrictions that are absent in the case of [iii]. The restrictions can probably
best be understood as involving two people that can see each other in the
face (which explains the open space), or entities such as buildings, that are
anthropomorphically modeled as looking out towards each other. In fact, the
English word facing that can be used to translate over in these cases, ex-
presses the same anthropomorphic metaphor.
(66) Gosbert woont over het museum
'Gosbert lives across the museum'
(67) Aan ta/el zat JvJeinol( over JvJombert
At the table. Meinolf was sitting facing Mombert
(68) Het museum bevindt zich over de rechtbank
'The museum is situated across the magistrates' court'
(69) In de erehaag stond Edmar over Notker
'In the double row of honor, Edmar stood facing Notker'
[iv1 Situated at the end of a virtual path that extends from x in a straight
line across an open space
A second additional case to be mentioned is illustrated by (70)-(72). The
meaning of over in these examples can be defined as in [v]. Clearly, this type
of usage links up with the <zero motion> <-contact> cases of [ii]. As in the
latter, an entity is situated spatially within the boundaries (through vertical
projection) of x, but above x. The distinction between the <zero motion> <-
contact> cases of [ii] and [v] is that the former presupposes that the trajector
has a certain extension (as of a line or a surface), whereas the latter specifi-
cally allows for point-like trajectors (that are so to speak without extension).
This kind of usage, then, is a relaxation (in fact, a generalization) of the
reading described under [iii]. As shown by (73) and (74), the relaxation is
subject to the ad hoc restriction that only <no contact> cases are allowed. It
should be remarked that the presence or absence of [v] is particularly sensi-
tive to relatively minor situational differences and differences of perspectival
construal. If, for instance, (75) applies to a situation in which the dangling
movement of the spider covers a sizeable area, (75) would not be interpreted
along the lines of (72), but along those of(15). And whether (76) is a case of
[v] or [ii] would seem to depend on the size of the kettle with regard to the
fire (or on the extent to which these are taken into account by the observer).
Methodologically speaking, the ease with which possible cases of [v] shift
towards one of the other readings testifies to the peripheral character of [v].
(70) De lamp hangt over de tafel
'The lamp hangs over the table'
(71) Het ongeluk gebeurde over de zee
'The accident happened over the sea'
(72) Het spinnetje hangt sti! over de vensterbank
'Motionlessly, the little spider hangs over the window sill'
(73) *Het spinnetje figt sti! over de vensterbank
'Motionlessly, the little spider lies over (on) the window sill'
(74) *Het boek rust over de tafel
'The book rests over (on) the table'
(75) Het spinnetje bungelt over de vensterbank
'The little spider dangles over the window sill'
(76) De ketel hangt over het vuur
'The kettle hangs over the fire'
[v1 Situated above x, within the spatial boundaries of x
A third and final special case is illustrated by examples (77)-(80). In all
of these, the trajector has a circular structure, while the landmark is like a
pole or stick. While (80), which features a horizontal landmark, falls under
the <zero motion> <+contact> application of [i] in roughly the same way
that (11) does, the other cases are problematic because the landmark is in
vertical position. A straightforward allocation to the <zero motion>
<+contact> application of [i] is out of the question, because the trajector
does not cross or cover the landmark: it does extend from one side of the
landmark to the other, but it does so sideways, by going round the landmark
rather than by crossing and covering it. According to this line of thought, the
examples (77)-(79) illustrate a separate reading [vi], which can be consid-
ered a relaxation of the <zero motion> cases of [i]; specifically, the relaxa-
tion involves discontinuous trajectors (roughly, with a hole in them) that
64 Prototypicali(v and salience
extend from one side of the landmark to the other by going round it horizon-
tally rather than by going over it vertically. It is important to notice, how-
ever, that the syntagmatic restrictions on this type of usage still reflect some-
thing of the original meaning [i] from which [vi] appears to be derived. In
fact, as the ungrammaticality of (81) illustrates, the landmark cannot be so
tall that the trajector could not have reached its position by being pushed
downward from above; a trajector that has reached its position of sideways
encapsulation by a roundabout motion cannot be said to go over the land-
mark. Apparently, only the same kinds of covering movement that lead to the
configurations illustrated by cases like (60), (61), and (65) can lead to the
spatial configuration captured by over in its reading [vi]. And because (60),
(61), and (65) are unproblematic examples of [i], the link between [i] and
[vi] shows up in the syntagmatic restrictions on over in reading [vi]. At the
same time, it will also be clear that these restrictions cannot be predicted
from the spatial configuration [vi] itself. They can be derived from the se-
mantic history of the reading [vi] within the structure of over, but from the
point of view of the configuration [vi] as such, the restriction is an arbitrary
one.
(77) Er hangt een krans over het standbeeld
'A wreath hangs over the statue'
(78) De ring ligt over de kegel
'The rings lies over the cone'
(79) Wigbert schuift de rol toiletpapier over de pleeboy
'Wigbert slides the toilet roll over the toilet roll stand'
(80) De zwembandjes hangen over een lange haak
'The lifebelts are hanging on a long peg'
(81) *Een geellint was over de boom gebonden
, A yellow ribbon was tied over the tree'
[vi] Circularly encapsulating x, extending from one side of x to the other by
going round x
To summarize, we have found three special cases that can each be con-
sidered a relaxation or generalization of one of the three basic readings dis-
cussed in section 2: [iv] is linked to [iii], [v] to [ii], and [vi] to [i]. On the one
hand, then, [iv]-[vi] correlate with and testify to the systematicity of the ba-
sic range of application of over as discussed in section 2. On the other hand,
the less systematic features of [iv]-[vi] show up in the fact that the syntag-
matic restrictions that they are subject to CaImot be derived automatically
from the semantic content that they express (in contrast with the syntagmatic
restrictions on the major readings that were treated in section 3).
5. Prototypicality effects
The analysis so far might seem to imply that all applications of over have
the same status as far as structural weight is concerned. This is hardly the
case, though: there are unmistakable prototypicality effects at work. Intui-
tively at least, there can be no doubt that (for instance) the cases discussed in
the previous section are less preponderant than [i] or [ii]. Given that we do
not have psycholinguistic experiments at our disposal to corroborate this
intuition, what criteria do we have to establish differences in structural
weight among the various instances of over?
In the first place, the semantic relations between [i],[ii], and [iii] estab-
lish the central role of [i], at least if it can be accepted that that reading is
central from which the others can be most economically derived. In particu-
lar (restricting, for the sake of the argument, the discussion to cases of real
motion), [i] specifies a path across an obstacle while [ii] refers to a part of
that path, and while [iii] refers to the endpoint of the path. Both [ii] and [iii]
are therefore relatively straightforward metonymical extensions from [i].
Taking into account, however, that metonymy often works both ways, is
there any reason why, for instance, [ii] instead of [i] could not be the logical
centre along this axis of the structure of over? Notice that the transition from
[i] to either [ii] or [iii] can be taken in one step: there is a straightforward
type of metonymy (basically of the synecdochal kind) to justify the develop-
ment. Going from [ii] or [iii] to [i], however, is less straightforward. For
instance, while [ii] merely specifies that a certain movement is situated
within the boundaries of the landmark, there is no compelling reason why
that movement should be intended at crossing the landmark; that is to say,
because movement across x always implies movement within the boundaries
of x, but not the other way round, deriving [ii] from [i] is more plausible
than the alternative. Similarly, while crossing x normally implies that one
reaches the other side of x, being located at the other side of x does not im-
ply that one has reached that position by crossing x.
In the second place, the syntagmatic restrictions on the trajec-
tor/landmark combinations discussed in section 3 point towards the <real
motion> application of [i] as the unmarked case within the structure of over.
This is the only case, it will be remembered, that can be used with trajectors
and landmarks of all types, and that therefore has the widest applicability.
The restrictions on the other applications inevitably mark them off as special
cases. At this point, another particular kind of restriction (one we have not
discussed before) has to be pointed out in connection with the <fictive mo-
66 and salience
tion> and <zero motion> cases. Examples like (82)-(85) show that a particu-
lar situation that does not involve motion can more easily be expressed as a
fictive motion case than as a zero motion case, i.e., using a verb that is
originally a motion verb enhances the grammaticality of the sentences.
?
(82) Er bevindt zich een streep over het bord
'There is a line over the blackboard'
(83) Er loopt een streep over het bord
;A line runs over the blackboard'
(84) Er zijn condensatiestrepen over de lucht
'There are condensation trails over the sky'
(85) Er lopen condensatiestrepen over de lucht
'Condensation trails run over the sky'
The phenomenon would have to be investigated more systematically; for
instance, the <zero motion> applications of [ii] seem to be more susceptible
to the restriction than those of [i]. Even so, the restriction as such seems to
support the claim that <zero motion> cases are somehow more marked
within the structure of over than <real motion> cases.
In the third place, the existence of alternative expressions for the same
referential situation may point to the peripheral status of a particular kind of
usage of over, specifically when the alternative expression is intuitively more
natural than over in that particular context. The underlying idea is that the
basic, central instances of over will also be those that are less like to have
strong competitors for the expression of a particular concept. This is, in fact,
a particular version of the idea that form and meaning have parallel charac-
teristics: the whole idea boils down to the hypothesis that semasiological
centrality (of instances with regard to a category) will be reflected by ono-
masiological dominance (of a form with regard to a referent). Now, if we
take up again some of the examples from Table 1 (and some additional ones
discussed in section 4), we can try to deternline whether over has strong
competitors in those particular contexts. In examples (86)-(100), the alterna-
tive expressions are printed in italics at the end of each sentence. Brackets
round the alternative prepositions indicate that they are less likely than over
in that context. A plus-sign indicates that they are more natural. In general,
the list supports the view arising from the previous criteria, viz. that the
upper left corner of Table 1 represents the central set of instances of over.
(86) De bal rolt over de lijn: (naar de andere kant van)
(87) Het vliegtuig vliegt over de stad: (boven)
(88) De deken hangt over de schutting: (van de ene kant naar de andere)
(89) Er hangen donkere wolken over de stad: boven
(90) Engelram kuiert over de markt: (op)
(91) Het vliegtuig cirkelt over de stad: boven
(92) Er figt een armoedig kleedje over de vloer: op
(93) Een dunne streep mist hangt over de rivier: boven
(94) Over de rivier exerceerde de vijandelijke mifitie:
+aan de andere kant van
(95) Over de rivier loopt een rij bomen langs de beek:
+aan de andere kant van
(96) Dordrecht figt over de Moerdijkbrug: +aan de andere kant van
(97) Over de brug hangt een ballon: +aan de andere kant van
(98) Gosbert woont over het museum: Hegenover
(99) De lamp hangt over de tafel: +boven
(lOO) De ring figt over de kegel: rand
To summarize, we find that there are several indications for a prototypi-
cal structure of over pointing in the same direction. It should be mentioned,
though, that neither of the criteria used above can be considered a foolproof
test of prototypicality; their parallelism, however, does seem to warrant the
conclusion we have drawn from them. It is obvious, at any rate, that addi-
tional research (of a basically psycholinguistic nature) could be invoked to
complement the analysis given here.
6. A comparison with Cuyckens
Having charted the range of application of over, let us try to compare our
own treatment with the one presented by Cuyckens (1991). In Table 3, the
various senses of over distinguished by Cuyckens (1991: 282) are charted
within the same framework as the one presented in section 2 (supplemented
with the three additional readings described in section 4). Figure 2 is a copy
of Cuyckens's own overview of the polysemy of over. In this classification,
over} represents the 'above and across' reading, assuming a process of real
motion. Over2 refers to the related cases where the movement is conceptual-
ized as being situated within the (vertically projected) spatial boundaries of
the landmark. As can be seen in Figure 2, over} and over2 only refer to sur-
face-like landmarks; over3, however, takes care of point-like landmarks in
the case of a movement 'above and across'. Next to theses three readings
referring to <-contact> cases, the possibility of contact is recognized in
over4, avers and over6. Judging from the graphic representations, 4 and 6
refer to cases where a surface-like or object-like landmark (respectively) is
crossed. Overs, on the other hand, considers cases where the movement stays
68 and salience
OVER1D OVER 9
."
."
I--
cb
1
OVER 3
"
"

... :

. :
0
V 7\
. .
t--
f--
.v
V

OVER 1
OVER 2
11
I 1
-E)-
r--
8
OVER 4 OVERS
OVER 6
OVER 7

....
... .)

OVCR8

x- '.V

lOVER12
OVCR13
fI]
@ff]
. : : '.' ...
-
' .. : '." .
. . ' ...
Figure 2. The analysis of over in Cuyckens (1991)
within the confines of the landmark, but specifically envisages the trajector
to pursue a 'fairly rectilinear path' (1991: 280). The addition is not without
importance: over]3 refers to those cases where the movement within a sur-
face has no dominant direction, but rather takes a random form. In the
analysis presented by Cuyckens, over13 ranges with the 'covering' -readings
over11 and over12 rather than with over5. Over11 and over12 themselves in-
volve surfaces (not necessarily of equal size) of which one covers the other.
Overl2 specifically takes into account what we have called the 'discontinu-
ous' cases in section 3. Because the graphic representation suggests that the
trajector of over11 is at least the same size as the landmark, it should be
noted that the examples given by Cuyckens also provide for trajectors that
are smaller than the landmark. These have been included as over11' in Table
3. Other additions are over] and over2', which refer to the <fictive motion>
counterparts of over] and over2. They are mentioned in the text, but are not
counted as separate readings by Cuyckens. The remaining over's represent
fairly straightforward cases. Over? is the 'at the other side of, relative to an
observer' reading (in all possible combinations); over8 is the 'across from,
facing' reading; and overQ and over]O correspond with the 'above' cases that
require no particular extension of the trajector. (The distinction between 9
and 10 depends on whether the landmark is a surface or a point.)
Table 3. Mapping Cuyckens' analysis onto the multidimensional structure of over
MOTION FICTlVE MOTION ZERO MOTION
+contact -contact +contact -contact +contact -contact
[i]
4.6 1,3 1'
11,12 ----> [vi]
[ii] 5,13 2 2' 11' ----> [v] 9,10
[iii] 7 7 7 ----> [iv] 8
In comparison with the analysis presented by Cuyckens, there are a num-
ber of characteristics that are specific to the one conducted in the previous
pages. First and foremost, the present analysis paints a more detailed picture
of the conceptual relations between the various applications of over. As can
be seen in Table 3, the over's that Cuyckens identifies consist of bundles of
co-occurring features (such as a particular type of spatial relation coupled
with a particular value on the 'motion/contact' dimension). But while the
analysis presented here specifies and isolates those features, Cuyckens treats
the bundles that constitute the 'senses' in his description as largely unana-
70 and salience
lyzable wholes. As can be seen in Figure 2, he does identify a number of
larger groupings of related over's, but these groupings do not constitute a
systematic exploration of the semantic relations between the various possi-
bilities of use of over. Specifically, the fact that the relevant features deter-
mine highly overlapping sets of uses is not prominent in his analysis. (The
groupings overlap in the sense that all subsets within the range of application
of over that are defined on the basis of the 'motion' dimension cross-classify
with the subsets defined on the basis of the 'contact' dimension, and with
those defined on the basis of the spatial relations referred to by [i], [ii], and
[iii].)
Conversely, the analysis presented here precisely tries to describe the se-
mantic structure of over as a cluster of subsets; rather than focusing on the
individual readings, it focuses on the structure of the possible groupings of
those individual readings. Thus, it not only systematically explores the rela-
tions between the various subsets into which the range of application of over
can be divided, but it also yields an insight into the question what various
kinds of features enter into the conceptual structure of over. Notice, in fact,
that there are three major kinds of features involved: spatial relations as
defined by [i], [ii], and [iii]; characteristics that pertain to the action, state,
or event involving that spatial relation; and characteristics of the trajector
and the landmark (and their combination). As a fourth type of a structurally
relevant feature, the process of semantic generalization that lies at the basis
of [iv], [v], and [vi] could be mentioned. At the same time, it has also be-
come clear that there is much more systematicity in the structure of over
than the picture presented by Cuyckens (see Figure 2) suggests: it will be
sufficient to compare Table 3 and Figure 2 to appreciate the fact.
The difference in focus between the analysis presented by Cuyckens and
the one presented here can be captured (and summarized) terminologically
by calling his outlook a synthetic one (because the emphasis lies on the indi-
vidual applications of over as a synthesis of co-occurring features), whereas
ours is analytic. Now, the analytic focus leads to two further differences
with the synthetic one. First, whereas Cuyckens is rather hesitant to answer
the question whether over has a prototype, an exploration of (among other
things) the syntagmatic restrictions that the various readings of over are
subject to in terms of trajector/landmark combinations unmistakenly sup-
ports the suggestion that over has a prototype structure (see section 5). Sec-
ond, a systematic exploration of the range of application of over leads to
greater comprehensiveness than Cuyckens has achieved. Apart from the
obvious empty slots in Table 3, it is unclear why some features that are con-
sidered criterial by Cuyckens in one part of the entire range, are not given
the same treatment elsewhere. For instance, it is unclear why the distinction
between <+contact> and <-contact> is considered criterial for distinguishing
between over4 and overj, but does not lead to a similar distinction in the case
of point-like trajectors (why doesn't Figure 2 contain a <+contact> counter-
part of over]?).
7. Methodological concluding remarks
The distinction between the 'synthetic' approach and the 'analytic' approach
has to be reinterpreted on a more general methodological level. Prototype
theory has (rightly) stressed the important role that the individual members
of a category may play in the structure of that category. However, there is a
danger that this kind of emphasis may lead to the tendency to restrict the
analysis to the salient members of a category, thus underestimating the im-
portance of a definitional analysis that not only tries to indicate the semantic
relationships between the salient instances of a category, but that also tries
to make sense of less central cases. The salient instances of a category take
the form of bundles of co-occurring characteristics, but the foregoing analy-
sis of over suggests that the true semantic structure of a category can only
be revealed by systematically unraveling the features occurring in those bun-
dles, their kinds, and their mutual relations. The tendency to concentrate on
the salient instances of a category carries with it the danger of neglecting
such an analytical approach.
At this point, we naturally come back to a question that was raised in
section I: how to determine what the readings of over are? Against the back-
ground of the analysis summarized in Figures 2 and 4, the same question can
be restated: what should we consider to be the 'meanings' of over - higher
level groupings such as [i], [ii], and [iii], or lower level applications such as
those charted by Cuyckens? In order to make the structure of the problem
more apparent, I may refer to a distinction that I introduced in Geeraerts
(l987a). On the one hand, the question how to determine what the meanings
of a lexical item are may be approached from a psychological angle, taking
into account phenomena such as intuitive salience effects. On the other hand,
the question may be tackled in a purely definitional, analytical vein; accord-
ing to the traditional requirements, separate meanings will then be maximally
general subsets of the extension of an item that can still be defined by means
of necessary and sufficient characteristics
8
Now, the intuitive approach
seems to point specifically to the lower level applications of over as distinct
meanings: the psychologically salient aspects of over (for instance, the se-
mantic values that can easily be brought to mind by the native speaker) are
synthetic entities, bundles of characteristics like the various over's discussed
by Cuyckens. Conversely, an analytical, definitional approach automatically
shifts the focus towards the more general groupings of the basic applications
(as represented by [i], [ii], and [iii], and the like).
Is there any way, then, to decide which approach (the intuitive or the
definitional one) is most appropriate? I would say there is not. For one thing,
as argued in Geeraerts (l993a), the existing criteria for distinguishing true
meanings from mere cases of referential vagueness do not yield a clear-cut,
unambiguous picture of what 'meanings' are. Rather, the question itself
what the meanings of over are might well be suspect: it presupposes that
meanings are like entities, stored as clearly separated objects in a word's
contents, whereas the semantic flexibility that has been reintroduced into
lexical semantics by prototype-theoretical studies could very well (when
taken to its extreme) lead to the conclusion that 'meanings' are contextually
flexible rather than stable and unaffected by context. Moreover, even if we
would reach an agreement about what the concept meaning precisely covers,
this would be basically a terminological decision. It would not detract from
the fact that (at least from a cognitive perspective) both the psychological
phenomena and the definitional analysis are indispensable aspects of any
truly adequate study of the semantic structure oflexical items: up to a point,
the lexical-semantic structure that Cognitive Semantics would like to de-
scribe, resides precisely in the interplay between both types of data. In this
sense, my argument in favor of the 'analytic' approach to over does not
imply that the 'synthetic' approach can be neglected; rather, the whole point
is that both approaches have to be taken into account.
To round off, let us note that the analytical type of prepositional descrip-
tion as represented by Figures 2 and 4 can be seen as an application and
concretization of the ideas expressed by Leys (1988). Leys, in fact, has sug-
gested that the basic conceptual structure of spatial prepositional relations
takes the form [X S/M] R [Y], in which X is the trajector, Y the landmark,
R the spatial relation as such, and S/M the stative or motional situation that
determines the 'aspect' of the prepositional construction. It will be clear that
what I have tried to do is precisely to specify the various values that R can
take in the case of over by means of definitions like [i]-[iii], and to identify
the various aspectual values of the construction by means of the distinction
between <motion>, <fictive motion>, and <zero motion> cases. Whether the
attempt is as yet entirely successful is another matter, but I hope it has be-
come sufficiently clear that taking the approach advocated by Leys seri-
ously, enables us to make at least some progress in comparison with alterna-
tive treatments of over.
Notes
1. Actually, eruse's example features Charles rather than Balder. However,
less commonplace proper names will be used throughout this article to draw
the attention to the fact that Odo Leys, to whom this paper and this issue of
Leuvense Bijdragen are dedicated, has made some major contributions to the
study of proper names apart from his work on the central themes of this col-
lection (and his manifold other research interests).
2. Translations and paraphrases of the Dutch sentences are given in square
brackets. Occasionally, round brackets within the square brackets contain ex-
planatory glosses.
3. A classification of the various kinds of fictive motion that could possibly be
distinguished is beyond the scope of the present article.
4. Given the system according to which the examples are construed, English
translations are not always given.
5. In fact, [iii] is possible as well, but it is irrelevant for the present discussion.
6. The label 'line-like' should not be taken too narrowly in this description. The
blanket in (57) could probably also be conceived of as a surface-like trajector;
in that case, the requirement that the landmark should be smaller than a line-
like trajector in the direction in which the latter extends would have to be
modified to provide for oblong surface-like trajectors.
7. It may be noted, however. that a 'covering' interpretation is not necessarily
unavoidable in the case of multidimensional over-ness: when er ligt een
kleedje over de tarel [a napkin lies over the table] is used to describe a situa-
tion in which the napkin is placed like a lozenge, leaving the four corners of
the table free, it does not exactly cover the entire surface of the table, al-
though it does lie over it in more than one dimension. (Of course, another
way of dealing with this situation would be to accept a relaxed interpretation
of 'covering'. so as to include almost complete coverage.)
8. For more details, see the (1987) article, and the (1991) paper. in which the
distinction between both types of polysemy criterion is further elaborated.
Chapter 4
Salience phenomena in the lexicon. A typology
Originally published in Liliana Albertazzi (ed.), 2000, Meaning and Cognition
125-136. Amsterdam/Philadelphia: John Benjamins.
As argued in Chapter 1, the essence of prototype theory lies in the fact that it
highlights the importance of flexibility (absence of clear demarcational bounda-
ries) and salience (differences of structural weight) in the semantic structure of
linguistic categories. But whereas prototypicality was initially described for the
semasiological structure of lexical items, flexibility and salience may be observed
in many other circumstances. In particular, as suggested in Chapter 2, it is crucial
for cognitive studies to switch from a semasiological to an onomasiological per-
spective, i.e. from an analysis conducted on the basis of words to one conducted
on the basis of concepts.
The present chapter, then, explores the concept of onomasiological salience, and
puts it in the context of a systematic analysis of salience phenomena in the lexi-
con. The article draws on (and partially repeats) a number of case studies included
elsewhere in this volume, viz. the over study of Chapter 3 and the beer names
study of Chapter 10. Also, like Chapter 6, it includes some materials taken from
The Structure of Lexical Variation (Geeraerts, Grondelaers, and Bakema 1994).
The detailed empirical investigation of lexical variation (both semasiological and
onomasiological) carried through in that monograph constituted the first step in
the development of an onomasiological line of research within our Leuven re-
search group (again, see Grondelaers and Geeraerts 2003 for an overview, and
compare Chapter 11).
1. Salience as a lexicological variable
When one considers the question of the impact of Cognitive Semantics (in
the sense of Langacker 1990a, Lakoff 1987, or Taylor 1989) on the field of
lexicological research, it is probably correct to say that systematic attention
to salience phenomena is one of the major innovations of the cognitive para-
digm. As an inquiry into language as a system of categories, Cognitive Se-
mantics has received major impetus from the prototype-theoretical insights
into the structure of linguistic categories developed in psycholinguistics and
anthropological linguistics. And prototype theory is basically concerned with
Salience phenomena in the lexicon 75
salience phenomena: it shows how the various semantic applications that
exist within the boundaries of one particular category need not have the
same structural weight within that category.
The cognitive semantic attention to salience, however, has not yet been as
systematic as it might have been. There are forms of salience phenomena
that have not yet been investigated as thoroughly and completely as proto-
type-based salience. The purpose of this paper, therefore, is to provide a
tentative overview of the various forms of salience effects that may be use-
fully distinguished. A typology of salience phenomena will be presented,
based on a distinction between the following basic types: perspectival, se-
masiological, onomasiological, and structural salience.
Although terminological hair-splitting and taxonomic overabundance are
real dangers for an endeavor such as this one, there is one overriding reason
why the present exercise may be useful: salience is the place where structure
and use meet. Consider lexical semantics and prototype theory. Traditional
semantic analysis is concerned with describing the various meanings of lexi-
cal items and the structural relations among those meanings. Katz and Fo-
dor's bachelor example, for instance, describes the readings that 'bachelor'
may possibly yield in actual text, and at the same time describes the concep-
tual relations among those readings, in the form of a hierarchical structure
based on shared semantic features (Katz and Fodor 1963). Such analysis is
basically a structured list of possibilities: it specifies the space of possible
readings that language users may choose from in actual communication. The
study of those communication processes, the actual choices made by the
language users, is not envisaged. Prototype theory, on the other hand, does
incorporate aspects of use in the structural description of the lexical catego-
ries: the fact that some readings are more salient than others in the structure
of the category reflects the fact that they are more readily chosen when using
that category. Admittedly, frequency of occurrence is not the only factor
determining prototypicality, but next to conceptual coherence, it is certainly
an important one.
It should be clear that the combination of structure and use in the study
of salience phenomena is essentially also a combination of semantics and
pragmatics: semantics as the study of structure, and pragmatics as the study
of use, combine when salience is seen as the structural reflection ofprag-
matic phenomena. This interlocking combination of the semantic and the
pragmatic perspectives implies a slightly different conception of the notion
of 'structure' than is customary: whereas linguistic structures would tradi-
tionally be seen merely as ordered sets of possibilities, adding pragmatics-
based salience implies introducing probabilities rather than just possibilities.
2. Perspectival salience
The first step to take involves a definition of the concept of perspectival
salience - a type that more or less stands apart from the other cases to be
treated, and which will only be mentioned briefly here. Perspectival salience
is represented in Cognitive Semantics by such terminological pairs as profile
and base, or more generally, jigure and ground (for definitions and applica-
tions of these concepts, see Langacker 1990a or Talmy 1978, among many
other publications). To mention just the simplest of examples, the lexical
item hand evokes a particular body part, but it primarily does so against a
particular conceptual background, namely, that of the arm. The arm is pre-
supposed in the concept of the hand, but it is the hand itself that is high-
lighted against the backdrop of the arm. Crucially, this is a matter of per-
spectivization: a particular chunk of reality (the hand and the arm) is
perspectivized in a particular way.
The fundamental distinction between this perspectival type of lexical sali-
ence and the ones to be mentioned later, resides in the fact that the other
types defined below invariably involve differences of salience among lexico-
logical alternatives (where the alternatives may be meanings, syntagmatic
contexts, lexical forms, lexical categories, or distinctive dimensions). In the
perspectival type of salience, on the other hand, the salience phenomena
involve just one of these lexicological alternatives (and specifically, one
particular lexical meaning); within such a single meaning, particular aspects
of the extra-linguistic situation referred to receive more weight than others.
Perspectival salience, in short, involves differences of perspectival attention
within one particular construal of reality, whereas the other type (which we
will call variational salience) involves preference relations among different
construals of reality. Perspectival salience is defined between the different
aspects of the extra-linguistic situation (in the broadest possible sense of
'situation') invoked by a linguistic expression: 'hand' invokes the entire arm
as the background of the conceptualization, but it is only the terminal part of
the arm that is saliently conceptualized by 'hand'. Variational salience, on
the other hand, is defined between different linguistic expressions or proper-
ties of those expressions: the semantic property of 'hand' that it may mean
'the terminal part of a human arm' is more salient than the property that it
may also mean 'manual worker' (as in 'hired hand') - which is then the less
prototypical reading.
Confusion should be avoided in particular when the linguistic expressions
to be compared from a variational perspective each embody a different per-
spectival construal of the same chunk of reality. Choosing to describe a
commercial transaction by either the verb 'to buy' or the verb 'to sell' in-
volves an onomasiological choice. If one of both terms is the more obvious
alternative, there is a difference of variational salience between both verbs:
we would then say, for instance, that 'to buy' is a more salient way of talk-
ing about a commercial transaction than 'to sell'. On the other hand, each of
the two verbs separately attaches perspectival salience to different aspects of
the commercial transaction scene. In the case of 'to buy', the person buying
is more salient than the person selling, and the reverse is true in the case of
'to sell'. In examples such as these, comparing the differences in perspecti-
val salience of two linguistic expressions is not the same thing as comparing
the onomasiological salience of the linguistic expressions themselves, that is,
the likelihood that one rather than the other will be used in situations where
both are applicable.
3. Semasiological, onomasiological, and structural relations
As a further step in the construction of the classification, a distinction must
be drawn between semasiological salience and onomasiological salience.
Now, the conceptual distinction between semasiology and onomasiology may
require some background information, because it involves a terminological
pair that has - unfortunately -not made its way into the standard set of lin-
guistic terms to be found in introductory courses. It is a terminological dis-
tinction that was introduced at the beginning of this century, at the time of
the birth of structural methods of semantic analysis. The tern1S were coined
by the Swiss romanist Karl Vossler in 1919. Terminologically speaking, the
distinction is still much alive in those branches of (continental) European
lexicology that preserve the heritage of structuralist semantics (such as the
Coseriu school in Germany). In the Anglo-Saxon world, however, the termi-
nology is hardly known. Basically, the distinction involves two ways of per-
spectivizing the structuralist link between and This ter-
minological pair itself, to be sure, is not without having problems of its own,
but for present purposes it may be sufficient to equate with a word
form, and with the concepts associated with that word form.
A semasiological perspective takes the word as its starting-point and in-
vestigates how several concepts (meanings, if one likes) are associated with
that word. This is, of course, the regular type of semantic analysis in which
the meanings of a word are listed and the semantic relationship between
them indicated. Onomasiological research, on the other hand, reverses the
perspective, and takes the concepts expressed as its starting-point; it investi-
gates, basically, which various words may express a given concept, and
what the structural links between those words are. In actual historical prac-
tice, the onomasiological perspective has been realized primarily in the form
of lexical field research. Bruno Quadri's bibliographical overview of onoma-
siological research, published in 1952, for instance, reads as an overview of
various types of lexical field analysis (Quadri 1952). Kurt Baldinger (again
a Swiss romanist) has succinctly described the difference in the following
way: 'Semasiology [... ] considers the isolated word and the way its meanings
are manifested, while Onomasiology looks at the designations of a particular
concept, that is, at a multiplicity of expressions which form a whole.' (Bald-
inger 1980: 278).
An even more succinct way of putting things would be to say that sema-
siology involves processes of meaning, whereas onomasiology involves
naming. Baldinger's formulation, however, also makes clear that there is a
certain ambiguity in the definition of onomasiology. On the one hand, when
onomasiology is conceived of in tern1S of 'designation' (or naming), it in-
volves the relationship between semantic entities and the lexical items that
name them. If, on the other hand, onomasiology is conceived of in terms of
'a multiplicity of expressions which form a whole', it instead involves the
mutual relations between related lexical items. In this conception, onomasi-
ological research will tend to coincide with lexical field research as a purely
structural endeavor - that is, as an approach in which the actual choices
from among the set of alternatives present within the lexical field are barely
envisaged, but in which most if not all of the attention is directed towards an
analysis of the relations between those expressions. In actual practice, in
fact, so-called onomasiological research has been basically restricted to lexi-
cal field research of the kind just described.
It may be useful, therefore, to impose a terminological distinction be-
tween both kinds of onomasiological research, because it will help us to
distinguish between the purely structural and the pragmaticized conception
of onomasiology: if the term onomasiology is reserved for the pragmatic
approach in which the relationship with actual referents is explicitly envis-
aged, the other approach may then be called structural.
c c
~ - i i
R w w
R
Figure 1. Semiotic relations between the major groups of salience phenomena in
the lexicon
The differences between the three perspectives defined in this way, may
be presented schematically by starting from two semiotic triangles as in Fig-
ure 1. If we maintain a distinction between Word (or lexical item, W in the
figure), concept (C in the figure), and referent (R in the figure), then semasi-
ology investigates how a particular lexical item may express various con-
cepts, or alternatively (putting things extensionally rather than intensionally),
how it may designate various types of referents. This is expressed by the
arrows in the left hand triangle. Onomasiology in the restricted sense defined
a moment ago describes how referents or concepts may be designated by
various expressions; this is indicated in the right hand triangle. A structural
analysis, finally, studies the semantic relationship between expressions. In
the figure, this is indicated by the arrow in between both triangles. The ar-
row actually connects the lines going from C to W rather than just both W's
or both C's alone, because the structural analysis involves expressions as
combinations of forms and meanings, rather than just word forms or seman-
tic concepts alone. (The double directionality of the arrow between the trian-
gles, therefore, does not indicate a one-to-one correspondence; it merely
represents the fact that the structural analysis is a reciprocal one: it looks, so
to speak, at one item from the perspective of the other, and vice versa.)
Given the distinction between semasiology and onomasiology, and the
double interpretation of onomasiology, concepts of semasiological salience,
onomasiological salience, and structural salience may now be defined.
4. Semasiological salience
Semasiological salience is a relationship among the various semantic possi-
bilities of a given lexical item. A lexeme/morpheme may express various
semantic values, as when in expresses containment ('the apple in the bowl ')
or inclusion/surrounding ('the village in the desert'). Some of the values
80 and salience
expressed by the lexical element may be more central than others, for in-
stance because they occur more frequently within the range of application of
the lexical element, or because they are a semantic centre in a network or
family resemblance structure from which the other values are extensions.
For example, containment seems to be more central in the structure of in
than inclusion (see Vandeloise 1994). Practically speaking, then, semasi-
ological salience coincides with prototypicality. As prototypicality effects
constitute the best known type of lexicological salience, it is not necessary to
recapitulate the literature here (see Geeraerts 1989c for an introduction, and
Taylor 1989 for an extensive overview.) Two things require clarification,
though, in preparation for the following discussion of onomasiological sali-
ence.
First, prototypicality may be defined both as a relationship between a
lexical item and its meanings, and as a relationship between a lexical item in
one of its meanings and the referents corresponding with that meaning (com-
pare Kleiber 1990). In the former case, it is pointed out for instance that the
'biological species' sense of 'bird' is more common than the slang sense in
which 'bird' refers to a girl. In the latter case, it may be pointed out that
robins and sparrows are more central members of the 'biological species'
sense than, for instance, ostriches and penguins. This distinction between the
two possible definitions of prototypicality hinges crucially on the distinction
between polysemy and vagueness: of all the multiple applications of a lexical
item, some constitute true meaning differences (and hence involve
polysemy), while others should merely be considered referential specifica-
tions (and hence involve vagueness rather than polysemy). It has recently
been suggested (Taylor 1992, Geeraerts 1993a) that the distinction between
polysemy and vagueness may well be less stable than is traditionally as-
sumed; if this is correct, the distinction between two different levels of proto-
typicality (the semantic one and the referential one) marks the endpoints of a
continuum, rather than a clear-cut dichotomy.
Second, it may be useful to distinguish between the paradigmatic and the
syntagmatic aspects of lexical meanings. To begin with, let us note that each
of the values expressed by a relational lexical element may occur in a spe-
cific syntagmatic frame. For relational spatial concepts, Cognitive Semantics
describes such a frame in terms of a 'trajector' and a 'landmark' (see Lan-
gacker 1990a). For instance, the landmark of a containment relation has to
be a recipient with, say, 'containing potential', whereas the trajector has to
be something 'containable' in relation to the landmark. The containment
relation itself is a paradigmatic value, whereas the containing potential of the
landmark is a syntagmatic phenomenon. These restrictions on syntagmatic
contexts resemble the selection restrictions of old, except that they now ex-
plicitly involve (groups of) referents and not just words.
For a more extended example, we may turn to a word that more or less
functions as a classic example in Cognitive Semantics, namely the preposi-
tion 'over' (see Lakoff 1987). In Geeraerts (l992a) it is argued that there are
basically three spatial relations to be distinguished in the semantic structure
of the Dutch preposition over. They are defined a, b, and c below and illus-
trated by the examples (1 )-(3) (lrn stands for 'landmark', tr for 'trajector').
The range of application of over is not restricted to these cases, amongst
other things because the spatial relations may show up in situations involv-
ing real motion, in situations involving fictive motion, and in examples in-
volving stative relations.
(a) Extending from a point or region at one side of Im to a point or region
at the other side of Im, while crossing hn
(b) Extending from one point or region within the space defined by Im, to
another point or region within that space (without going outside of it)
(c) Situated at the other side ofhn, relative to the position of an observer
(1) De bal rolt over de lijn
'The ball rolls over the line (to the other side of it)'
(2) De agent kuiert over het marktplein
'The policeman strolls over the market-place'
(3) Over de rivier deed de atleet zijn oefeningen
'Across the river, the athlete was exercising'
The relevance of a separate analysis of the syntagmatic restrictions on
trajector and landmark becomes clear, for instance, when one compares the
selectional restrictions holding for these three basic relations. Without going
into all the details, let us briefly note that there are hardly any restrictions on
the trajector and landmark of semantic value (a): both trajector and land-
mark may be one-dimensional (points), two-dimensional (lines), or three-
dimensional (objects), in any combination. In the case of (b), however, both
trajector and landmark are subject to restrictions.
On the one hand, because the spatial relation referred to by over must
remain within the boundaries of the landmark, the prepositional object must
have a line- or surface-like extension: it should be possible to conceive of it
as a line or a surface. This shows up quite clearly when considering three-
dimensional objects. Whereas (4) and (5) are straightforward cases of lines
and surfaces, (6) is an example of a three-dimensional object whose third
dimension (height) does not prevent it from having a bounded surface that
allows for an over-relation. In (7), however, the three-dimensional object's
vertical dimension is so important that no surface for aimless or random
wandering can be imagined. As (8) shows, reading (a) of over does not im-
pose this restriction. This does not mean, however, that an object such as
schutting 'fence' could not occur at all with reading (b) of over. Example
(9) shows that the relevant feature is not the objective horizontal or vertical
orientation of the surface, but rather the functional characteristic of 'tread-
ability': something that could hardly be a surface for free movement for
human beings may be a natural path for insects.
On the other hand, because the trajector of an over-relationship according
to (b) must remain within the boundaries of the landmark, it will typically be
smaller than the landmark - small enough, at least, to move about on the
surface or along the line constituted by the landmark. Specifically in the case
of line-like landmarks, however, this does not imply that the trajector has to
be smaller than the landmark in all directions. Taking into account that real
lines are never merely one-dimensional, but always have a certain width as
their second dimension, the trajector does not have to be narrower (but it
does have to be shorter). As example (l0) shows, the moving object may be
wider than the actual line. (The train is wider than the rail.)
(4) Janjietste over het pad
'John rode his bicycle along the path'
(5) Jeanne kuierde over de markt
'Jane strolled over the market place'
(6) Mark dwaalde over de berg
'Mark wandered over the mountain'
(7) *Jvfarie dwaalde over de schutting
'Mary wandered over the fence'
(8) Karel klom over de schutting
'Carl climbed over the fence'
(9) De spin kroop maar door over de schutting
'The spider kept on crawling over the fence'
(10) De trein glijdt over de monorail
'The train glides over the monorail'
Now, the syntagmatic frame of each semantic value may have typical and
less typical instantiations. For instance, 'the apple is in the bowl' would
probably refer to a typical containment relation, whereas 'the jelly pie is in
the suitcase' is an untypical one (even though the containment relation itself
is the same as in the typical case). Such typicality relations illustrate the
concept of syntagmatic salience. In the over example, those instances of
reading (b) that more readily invoke a line-like or surface-like landmark will
be syntagmatically more salient than the others. In the sample sentences, for
instance, (4) and (5) are more typical than (6) or (9). Along the same lines,
examples in which the trajector is not entirely contained within the bounda-
ries of the landmark - as in sentence (10) - are syntagmatically less canoni-
cal.
5. Onomasiological salience
Turning now to onomasiological salience, there is a possible misconception
to be discarded first. From a very general point of view, salience on the
onomasiological level involves a preference for one expression over another
as the name for a particular referent or type of referent. The existence of
such alternative names would therefore seem to coincide with the kind of
formal variation that is customarily studied in sociolinguistics and dialectol-
ogy. Consider Table 1, which specifies a number oflexical patterns from the
clothing terms project described in Geeraerts, Grondelaers, and Bakema
(1994). In each of the cases in Table 1, there is a different naming pattern in
Belgian Dutch and in Netherlandic Dutch; the clothing types themselves,
however, are the same. The choice between legging and caler;on as a desig-
nation for leggings, for instance, therefore acquires the same value as the
distinction between, for instance, different pronunciations of the same pho-
nological form in different geographical regions or social strata of a linguis-
tic community.
This type of onomasiological variation may be called 'formal' or 'synon-
ymy-based', but it is useful to point out that these terms are not ideal in all
respects. Terms like legging and caler;on are denotationally synonymous,
but they are not absolute synonyms, because they have a different geo-
graphical distribution. The two terms have the same meaning in a narrow,
denotational sense of the word, but they clearly have a different sociolinguis-
tic value (in a broad sense of 'sociolinguistic'). In a similar way, the term
'formal' variation (to which I shall adhere in the rest of the paper) might
suggest that the variation in question is trivial or irrelevant, in the sense that
it could be taken to involve mere 'notional variants'. Again, however, the
alternation between legging and caler;on is not a question of free variation,
but signals a specific stratification of the linguistic community.
84 Proto(vpicali(y and salience
Table J. Differences of naming patterns between Netherlandic and Belgian Dutch
ITEMS
calec;on
legging
leggings
blouson
jack
jeans
spi}kerbroek
BELGIAN SOURCES
40 (38%)
26 (24.7%)
39 (37.3%)
13 (46.4%)
15 (53.6%)
64 (97%)
2 (3%)
NETHERLANDIC SOURCES
91 (100%)
85 (100%)
38 (70.4%)
16 (29.6%)
The misconception, then, consists of restricting onomasiological variation
to this type of 'formal' variation. Next to the synonymy-based variation
meant here, there are, obviously, cases where the choice of an alternative
name carries a semantic load. Consider the following distinction between
Belgian Dutch and Netherlandic Dutch.
(11) Hi} kwam langs de deur naar binnen
'He came in through the door' (Belgian Dutch)
(12) Hi} kwam door de deur naar binnen
'He came in through the door' (Netherlandic Dutch)
In Belgian Dutch, the usual preposition is langs 'via', where the semantic
focus is on the path as such. In Netherlandic Dutch, the usual preposition is
door 'through', which adds the notion of a passage that is penetrated and
crossed. Although the preference for one or the other construction is conven-
tional, it is not conceptually neutral: there is a different perspective involved
in the choice of one over the other. This implies that next to the formal sali-
ence of synonyms that is at stake in the leggings/cale90n example, we have
to distinguish a form of onomasiological variation that is semantic or con-
ceptual in kind: it involves the choice of one semantic category (and not just
one lexical item) rather than the other for talking about a particular real-
world situation (such as coming through the door). In contrast with the 'for-
mal' kind of onomasiological variation, we will call this type the' categorial'
type of onomasiological variation, because it involves the choice of a differ-
ent category.
To be sure, the existence of such a preferred categorial choice is not an
entirely novel idea, as it is part of the foundation of the basic level model of
taxonomical structure formulated by Berlin and colleagues (see Berlin
1978). There are, however, various points with regard to which the basic
level model of onomasiological salience may be refined. At least the follow-
ing two points may be considered.
To begin with, the basic level model attaches onomasiological salience to
hierarchical levels in the taxonomical tree, rather than to individual words.
However, it may be argued (Geeraerts, Grondelaers, and Bakema 1994) that
the notion of onomasiological salience should not be restricted to levels, but
should be extended - generalized, if one wishes - to individual categories. It
may not be necessary to repeat all the arguments here, but Table 2 gives a
sample of relevant observations.
Table 2. Differences in onomasiological salience among co-hyponyms
ITEM
broek
short-shorts
bermuda
legging-Ieggings-caler;on
jeans-jeansbroek-spijkerbroek
t-shirt
blouse-bloeze-bloes
overhemd
topje
shirt
hemd
overhemdblouse
ONOMASIOLOGICAL SALIENCE
46.47
45.61
50.88
45.50
81.66
70.61
61.52
31.45
29.62
29.06
22.31
12.74
The data in the figure are based on the following operational definition of
onomasiological salience: the onomasiological salience of a lexical category
is the frequency of the lexical element naming the category divided by the
cumulative token frequency in the database of the semantic values expressed
by that lexical element. The rationale behind this definition is as follows. For
each semantic value (concept, or type of referents) expressed by a lexical
element. there may exist alternative terms (co-hyponyms, hyponyms, hy-
peronyms). A lexical category is onomasiologically highly salient if it is a
likely choice for the semantic values it expresses, that is, if it is stronger than
the alternatives. Thus, given a corpus of language use, the onomasiological
salience of an item like 'skirt' can be calculated by counting how many times
skirts (i.e. the potential referents of 'skirt') are named in the corpus, and then
checking how many times these are actually referred to with the lexeme
'skirt', rather than alternative ones. Such alternative terms could be hypo-
nyms like 'miniskirt' or 'wrap-around skirt', or hyperonyms like 'garment'
(although the choice of the latter term seems quite unlikely in the case of
'skirt'). The definition of onomasiological salience also takes cases of syn-
onymy into account, but this aspect of the matter need not be spelled out
here.
The point about Table 2, then, is simply this: the onomasiological sali-
ence of different categories on the same taxonomicallevel may differ consid-
erably. For instance, in the upper part of the figure, 'short', 'bermuda', 'leg-
ging', and 'jeans' are co-hyponyms, as they all fall under the hyperonymous
category broek 'two-legged garment covering the lower part of the body,
pants'. However, the onomasiological salience of the different categories
differs considerably: that of the jeans-category, for instance, doubles that of
the legging-category. This means that a potential member of the category
jeans is twice as likely to be designated by an expression that names the
category jeans than a member of the legging-category would be likely to be
designated by an expression that names the category legging.
The second point about the conceptual kind of onomasiological salience
involves the recognition that the referential starting-point for the determina-
tion of onomasiological salience values is subject to variation. The salience
figures given so far represent the likelihood that one specific category, such
as legging or t-shirt will be chosen as the name for a particular referent, in
those cases where that referent is a potential member of the category. The
calculation, to be precise, is made for the category as a whole, that is, for the
overall set of referents of the category. However, the calculation may also
take its starting-point in a particular subset of the entire range of application
of a lexical category. The question asked so far boils down to questions of
the type: what is the probability for the set of leggings as a whole that its
members will be named with an expression that precisely identifies the cate-
gory 'legging'? The modified question now reads: given a specific subset of
the set of leggings as a whole (for instance, leggings that do not reach down
to the ankles but are not longer than the calves), what is the likelihood that it
will be named with the item legging or one of its synonyms?
Obviously, the answer to the question will be determined by the overall
salience of the category legging as calculated a moment ago, but it may also
depend on the structural position of the subset under investigation within the
semasiological structure of legging. The 'local' onomasiological salience
introduced here may be called the 'cue validity' of a particular subset of the
range of application of an item: it indicates how good a cue that subset is for
triggering the name of the category. The suggestion, then, is clearly that this
onomasiological cue validity and the semasiological properties of the subset
exhibit a positive correlation. Put more simply: an expression will be used
more often for naming a particular referent when that referent is a mem-
ber of the prototypical core of that expression's range of application.
When, for instance, a particular referent belongs to the core of item x but
to the periphery of y, it is to be expected that x will be a more likely
name for that referent than y. (At the same time, of course, the choice for
x will be a function of the global salience of the category represented by x,
but that is a factor we have already dealt with).
A brief and rather impressionistic example may suffice to illustrate the
idea (for more detailed, quantitatively elaborated examples, see again Geer-
aerts, Grondelaers, and Bakema 1994). The kinds of usage that are structur-
ally peripheral in the semasiological structure of Dutch over are more read-
ily replaced by alternative ternlS than the central cases. For instance, in the
examples illustrating reading (c) of over, the expression aan de andere kant
van 'to the other side of is more natural. In the same way, there is an exten-
sion of (c), defined and illustrated below sub (d), for which the alternative
expression tegenover 'across, facing' is by far the more natural one.
(d) Situated at the end of a virtual path that extends from lm in a straight
line across an open space
(13) ]v/artha woont over het museum
'Martha lives across the museum'
(14) Aan tarel zat Marianne over Ariane
.At the table, Marianne sat facing Ariane'
(15) Het museum bevindt zich over de rechtbank
'The museum is situated across the magistrates' court'
(16) In de erehaag stand Samuel over David
'In the double row of honor, Samuel stood facing David'
The notion of local onomasiological salience may be subject to two dif-
ferent interpretations, involving the same distinction between polysemy and
vagueness that was discussed above in cOlmection with the concept of proto-
typicality. The referential subset from the range of application of a lexical
item that is taken as the starting-point for a calculation of local onomasi-
ological salience mayor may not correspond to a meaning, in the traditional
sense, of that item. There is no reason to restrict the investigation of onoma-
siological salience to subsets that actually correspond with 'meanings'. This
is not only because (as mentioned before) the distinction between meanings
(underlying polysemy) and mere referential subsets (representing vagueness)
88 PrototypicaliZv and salience
may well be unstable, but also because prototype-theoretical research has
established beyond doubt that even referents that do not constitute a 'mean-
ing' in the strict sense (for instance, central members of the category), may
play an important role in the semantic structure of a lexical item. The mean-
ings whose local onomasiological salience is calculated need not necessarily
constitute denotational meanings: any lexically expressed semantic concept
(whether it be of a denotational, emotive, stylistic, or discursive kind) can be
subjected to the approach.
6. Structural salience
We may now turn to the third major perspective distinguished above. Re-
member that a purely structural conception of onomasiological lexical rela-
tions involves the mutual relationship between lexical categories. Salience
effects along this perspective, then, will have something to do with the
weight of the distinctive dimensions that distinguish various categories from
each other. A semantic feature or a semantic dimension will be structurally
salient when it often occurs in the structure of the lexicon; the structure it-
self, in fact, coincides with the totality of distinctive relations in the lexicon.
For an example, let us consider the lexical field of names for beer current
in Belgium (see Geeraerts 1999a). The methodology can be summarized as
follows: the more a particular feature is expressed in the names of beers, the
more it is cognitively salient. When, for instance, the presence of fruit fla-
vors never surfaces in the names of the beers that have such a flavor, it is
unlikely that fruit flavor is a preponderant feature of beers. Of course, we
have to be more precise about what it means for a distinctive feature or a
distinctive dimension 'to be expressed in the names of beers'. In most cases,
there is no problem: when you find a reference to kriek in a name, and when
you know that kriek is the name for a type of cherry, there is no difficulty in
concluding that the addition of cherry juice is mentioned in the name of the
beer. In other cases, the reference is more indirect. A striking example is the
following: in the French speaking part of Belgium, a peach-flavored beer is
produced with the name La pecheresse, which translates as 'the female sin-
ner, the sinning woman'. When you realize, however, that peche is the
French for 'peach', it will be clear that the name contains a pun with an
indirect reference to the presence of peach flavor.
In general, a feature is expressed by a name (or part of a name) when that
name (or the relevant part of it) only occurs in connection with that feature.
Some more examples may further illustrate the point. The fact that a par-
ticular beer is low on alcohol may be expressed directly by expressions like
alcoholarm 'low on alcohol' or alcoholvrij 'without alcohol'. Indirectly,
however, the item talelbier 'table beer' also signals the light character of
beer, to the extent that it only occurs within the group of very light beers.
Similarly, lambiek and geuze can be taken as expressions of the 'spontane-
ous fermentation' character of beers, to the extent that they only occur in
names for beers with that characteristic.
Table 3. Salience characteristics of sample referential features in the field of Bel-
gian beer names
REFERENTIAL FEATURE REFERENTIAL FREQUENCY
wheat (instead of rye) 139
very light low alcohol 110
seasonal beers 86
added herbs 59
raspberry flavor 40
cherry flavor 19
trappist beer 16
CHARACTERISTIC NAMES
91 (65.5%)
25 (22.7%)
86 (100%)
4 (6.8%)
38 (95%)
18 (94.7%)
16 (100%)
Following these guidelines, we can show that not all features of beers are
equally important. There are clear differences in the extent to which specific
characteristics are expressed in the names. Table 3 presents a sample to
demonstrate the kind of differences in salience that occur. The table should
be read as follows. The first column of figures indicates the number of beer
brands in the data set that has the feature mentioned to the left. The right-
most column indicates how many of those beers actually carry a name that
refers directly or indirectly to the feature in question. The relationship among
the figures mentioned to the right gives an indication of which features carry
distinctive weight. Trappist origin, for instance, seems to be highly valued,
while the addition of herbs is not an especially individuating feature.
In the spatial realm, the featural salience that is at stake here would in-
volve, for instance, the distinctive weight of dimensions like shape, or
movement, or geometry, or the role of the observer across items (or even
across languages). (See Taylor 1988 for an overview of potentially relevant
dimensions).
90 and salience
7. A typology of lexicological salience
In the following overview, each of the seven major cases of lexicological
salience as presented before is defined. Also, a shorter name is suggested for
each of the various types; of course, these names are just suggestions that
should be treated with care as long as there is no uniform lexicological ter-
minology.
1. Perspectival salience: highlighting
= the differences of perspectival attention attached to different parts of the
overall chunk of extralinguistic reality evoked by a particular concept.
2. Paradigmatic semasiological salience:
= the preponderant structural weight of specific senses or members within
the semasiological range of application of a lexical category
3. Syntagmatic semasiological salience: canonicality
= the dominant occurrence of specific selection-restrictional frames within
the syntagmatic range of a paradigmatic reading of a lexical category
4. Formal onomasiological salience: sociolinguistic prevalence
= the preference for a lexical item, in comparison with its synonyms, in a
specific language variety or a specific pragmatic context
5. Global categorial onomasiological salience: entrenchment
= the preference for a specific lexical category as a designation for its
range of application, taken as a whole
6. Local categorial onomasiological salience: cue validity
= the preference for a specific lexical category as a designation for a subset
of its range of application
7. Structural salience: distinctiveness
= the preponderant recurrence of a semantic dimension or a semantic fea-
ture as a distinctive dimension or feature within the semantic structure of
the lexicon
By way of summary, the seven types of lexicological salience are brought
together in their mutual taxonomical relations in Figure 2.
At this point, we may try to chart the seven types against the background
of the semiotic triangle used in Section 3. However, the inadequacy of the
semiotic triangle for dealing with more complex linguistic relations is well-
known; see Hofinann (1993) for a brief history of the matter, and compare
Geeraerts (1983b).
lexicological
salience
perspectival
salience
semasiological
salience
~
variational
salience
onomasiological
salience
~
structural
salience
paradigmatic syntagmatic formal categorial
~
global local
Figure 2. Taxonomical relations among the various types of lexicological salience
An attempt (but certainly no more than attempt) to graphically represent
the six variational types of salience may be based on Figure 3. In Figure 3,
three levels of analysis are distinguished: the level of words W, the concep-
tual level of meanings C of those words, and the level of the referents R
corresponding to those meanings. The word W2 is polysemous; Wl and W2
are synonymous with regard to meaning Cl. The referential range of words
may overlap in yet another case besides synonymy: R4 may be conceptual-
ized either as C3 or as C4 (as when a short wrap-around skirt might be
called either a 'miniskirt' or a 'wrap-around skirt'). The referential range of
CS is represented in a special way to highlight the syntagmatic contexts S in
which the referents of CS occur. The horizontal arrows represent the fea-
tures that distinguish between the concepts C. Salience effects may now be
represented by using bolder lines; the differences between onomasiological
and semasiological salience are expressed by having bottom-up and top-
down arrows respectively.
In Figures 7 to 12, the six types of variational salience are represented
graphically against the background of Figure 3. (The parts of the figure that
are irrelevant are represented by dotted lines.) Figure 4 indicates that C2
(rather than Cl or C3) is the prototypical meaning of W2, and that R3
(rather than R2 or R4) is the prototypical type of referent of C2. Figure S
shows that R-Sl is the canonical syntagmatic context for CS. Figure 6
represents a sociostylistic context in which W2 is the prevalent synonym for
Cl. Figure 7 indicates that W4 is a more highly entrenched category than
W1, in the sense that W4 is more likely to be used (in those cases where it
can in principle be used) than Wl is likely to be used (in those cases where it
is a relevant category). Figure 7 does not contain cases in which Wl and W4
overlap, but obviously a comparison of entrenchment values could just as
well be applied to such a situation. Figure 8 expresses that R5 has a higher
cue validity with regard to C4 than with regard to C3, that is to say, W3 is
more likely to be used to categorize R5 than W2. Figure 9 indicates that the
dimensions or features that distinguish C3 from C4 are structurally more
important than the others (for instance, because they recur elsewhere in the
lexicon).
W2 Wl
I
Cl
W3 W4
I I
_ - - - - _ ~ C2 . 4 - - _ ~ C3 __ C4 _ C5
V/\
Rl R2 R3 R4 R5 R-Sl R-S2
Figure 3. Sample network for representing salience phenomena
Wl W2 W3 W4
Cl
C2 C3
C4 C5
Figure 4. Schematic representation of paradigmatic semasiological salience
C2 C3
C5
1\
Wl
Cl
Rl R2
W2
R3 R4 R5
W3
C4
R-Sl
W4
R-S2
Figure 5. Schematic representation of syntagmatic semasiological salience
Wl W2
1________
Cl
C2 C3
W3
C4
W4
C5
Figure 6. Schematic representation of formal onomasiological salience
Wl W2 W3 W4
I.
i
Cl
................
C2
. ....
C3 C4 C5
1
...
1\
Figure 7. Schematic representation of global categorial onomasiological salience
Wl W2 W3 W4
. ~
t
Cl C2
, ...........
C3 C4 C5
\/
Figure 8. Schematic representation of local categorial onomasiological salience
Wl W2 W3 W4
Cl
_----__ C2 .4--__ C3 ++C4 __ C5
Figure 9. Schematic representation of structural salience
8. Salience and snow
The classification presented in the previous pages is undoubtedly only a first
step towards an exhaustive and systematic treatment of lexicological sali-
ence. This first step was taken for the purpose of clarification, but it may be
interesting to round off by adopting a slightly different perspective: could the
classification throw new light on other linguistic phenomena?
As one example, let us consider the question of conceptual universals. If
semantic universals are likely to be statistical universals rather than absolute
ones, the notion of semantic salience will be crucial in establishing the statis-
tical universality of a particular phenomenon: semantic phenomena that are
cognitively basic are likely to be salient across genetically unrelated lan-
guages. Questions of cultural specificity (the logical counterpart of questions
of universality) likewise involve salience phenomena. Consider, to address
one perennial member of the lexicological stock, the alleged existence of
various words for snow in Eskimo. Let us suppose for a moment that there
are no empirical problems with Boas's statement (1911) that the Eskimo
have four words for snow, whereas English and Aztec have only one (but see
Martin 1986). Boas mentions the following four:
aput 'snow on the ground'
gana 'falling snow'
piqsirpoq 'scurrying snow'
qimuqsuq 'snowstorm, blizzard'.
It has been repeatedly pointed out that this does not imply that people
camlOt notice differences in snow merely because they speak English. Clark
and Clark (1977: 555), for instance, draw attention to the fact that English-
speaking expert skiers may have specific words for snow such as 'powder',
'corn', and 'ice'. More generally, there is always the possibility of ad hoc
compounds and word groups: 'snow glue', 'drizzle snow', 'icicle flakes', or
the translations of the Eskimo words given above ('falling snow' etc.). The
possibilities of expression of a language are not restricted to the vocabulary,
and neither are the possibilities of conceptualization of the speakers of the
language restricted to what is included in the mental lexicon. This attenuated
point of view, however, does not imply that the cultural differences between
the Eskimo and the Aztec disappear altogether. The cognitive salience of a
concept like 'falling snow' may well be higher for the Eskimo than for the
Aztec: it is plausible that they will more easily and more readily conceptual-
ize a particular white substance as 'falling snow' rather than just 'snow'. In
the terms of the classification presented in this paper, this expectation can be
translated into the hypothesis that the entrenchment (the global categorial
onomasiological salience) of 'falling snow' will be higher for the Eskimo
than for the Aztec.
The relationship between salience and universals may be specified from
yet another point of view. In her work on semantic universals, Wierzbicka
(1992) demonstrates that such abstract concepts as emotions and moral no-
tions may be of a culturally specific n a t u r e ~ at the same time, she insists on
the possibility of expressing such concepts in any given language by means
of a language-independent metalanguage: no matter how culture-specific a
term might be, it can be translated in the 'alphabet of human thoughts' that
she has been researching for decades. The foregoing pages, then, suggest one
way of looking for such an 'alphabet of human thought' that might usefully
supplement Wierzbicka' s analytic approach: if these concepts really have the
universal status attributed to them, they should in one form or another have a
high degree of lexicological salience in a large number of natural languages.
96 and salience
The universality question also invokes the problem of pre-linguistic forms
of salience, namely the objective correlates of the linguistic phenomena. To
what extent is linguistic categorization determined by the structure of the
world itself, or at least by the structure of our perceptual apparatus? Ifnam-
ing is often grounded in an act of perception, how do the independent proper-
ties of such perceptual acts influence the linguistic phenomena? The typol-
ogy presented here has steered clear of these perceptual forms of salience,
but it certainly does not mean to exclude the possibility of a pre-linguistic
level of salience. Rather, a cognitive approach to the study of language will
inevitably have to consider the possible impact of such a pre-linguistic type
of salience. Tackling that question, however, is beyond the scope of this
article: before one starts looking for explanations of the linguistic phenom-
ena, one has to be clear about those phenomena themselves.
Section 2
Polysemy
Chapter 5
Vagueness's puzzles, polysemy's vagaries
Originally published in Cognitive Linguistics 1993,4: 223-272.
Polysemy is the basic phenomenon in lexical meaning structures, and arguably the
basic contribution of Cognitive Linguistics to lexicology is the renewed attention
for the structure of polysemy that it generated. But how exactly should polysemy
be defined? What does it mean for a reading of a word to be a different meaning?
In this paper, I argue that the distinction between vagueness and polysemy (which
is crucial to traditional conceptions oflexical-semantic structure) is unstable: what
appear to be distinct meanings from one point of view turn out to be instances of
vagueness from another. In the present chapter, an argumentation for this thesis is
developed in three steps. First, three basic kinds of polysemy criteria are distin-
guished: a logical, a linguistic, and a definitional test. Second, it is shown that
these three kinds of tests may yield mutually contradictory results. Third, it is
argued that such inconsistencies also occur within each criterion, in the sense that
there are contextual variations in the polysemy results that they yield. As a conse-
quence, Cognitive Linguistics will have to start from a highly flexible conception
of meaning.
In the period after the publication of the original article, polysemy research has
moved in various directions relevant to the issues addressed in this chapter. First,
Cognitive Linguists have explored ways for incorporating the flexibility and in-
stability of meaning into the models of semantic structure that they work with.
Levels of schematicity in the Langackerian sense (i.e. the idea that the meaning of
a linguistic unit can be described as a schematic network in which more specific
and more abstract levels of description co-occur) play a crucial role in such at-
tempts: see Tuggy (1993), Dunbar (2001), Taylor (2003), Zlatev (2003). Second,
attempts have been made to link polysemy research in Cognitive Linguistics with
psycholinguistic studies of polysemy (Sandra and Rice 1995), but this has -
somewhat surprisingly, given the very cognitive orientation of Cognitive Linguis-
tics - not developed into a dominant trend; see the discussion between Croft
(1998), Sandra (1998), and Tuggy (1999). Third, the interest in polysemy issues
has spread outside the field of Cognitive Linguistics: see Ravin and Leacock
(2000), Peeters (2000) and Nerlich, Todd, Hennan and Clarke (2003) for samples.
In particular, a major new trend in polysemy research has arisen in the domain of
computational linguistics, in the form of so-called Word Sense Disambiguation
algorithms: computational and statistical analyses of large corpora that take the
100 Polysemy
co-occurrence behavior of lexical items as indications of their semantic range (see
Stevenson 2001, Agirre and Edmonds 2006 for overviews).
Together with detailed corpus-based studies as will be illustrated in the next chap-
ter, psycholinguistic and computational studies of the type just mentioned will, I
think, be of crucial importance for the further development of polysemy research
in Cognitive Linguistics: the theoretical issues have been greatly clarified, but
there is still plenty of room for methodological progress.
1. Prototypicality and polysemy
In my paper 'On necessary and sufficient conditions' (1988a), I pointed out
that there is an intimate connection between the proper evaluation of the
novelty of the prototype-oriented movement in lexical semantics, and the
long-standing issue how to determine whether a word has one or more
senses. In fact, while the revolutionary nature of prototype theory is often
said to reside in the fact that it rejects the classical view that lexical catego-
ries can be defined by means of a sufficient set of necessary conditions for
category membership, the possibility that some lexical items cannot receive a
definition in terms of necessary-and-sufficient attributes is not exactly absent
from the classical tradition. Rather, it is traditionally taken as an indication
of the polysemy of the item in question. Therefore, if prototype theory is to
maintain its revolutionary claims, it will have to redefine them, or at least
make clear that the absence of necessary and sufficient definitions that it
reveals does not simply signal good old-fashioned polysemy.
In his recent monograph La semantique du prototype (1990), Georges
Kleiber has taken up this suggestion, incorporating it into a discussion of the
various trends that exist within linguistic prototype theory. However, as I
already briefly noted in my review of Kleiber's thought-provoking book
(1992), his treatment of the consequences of polysemy for prototypical theo-
rizing seems to assume that the concept of polysemy is methodologically
unproblematic; at least, his book does not deal explicitly with the question
how to identifY the polysemous readings of a lexical item. The present paper
is devoted precisely to that issue. It will compare various criteria for distin-
guishing between vagueness and polysemy, and explore the consequences of
this comparison both for Kleiber's views and for prototype-oriented lexical
semantics at large. In particular, it will point out two sets of problems for
the usual polysemy criteria: first, the various types of criterion will appear to
be mutually inconsistent, and second, any single criterion yields internally
inconsistent results. That is to say, I will try to show that the distinction
Vagueness's puzzles, polysemy's vagaries 101
between vagueness and polysemy is not stable, in the sense that what ap-
pears to be a distinct meaning in one context is reduced to a mere case of
vagueness in another.
The paper is organized as follows. First, the importance of the polysemy
issue for prototype-oriented theories is discussed in more detail. Second, a
classification of types of criteria for distinguishing between vagueness and
polysemy is presented. As a third step, the problems with the criteria will be
presented, first by comparing the criteria among each other, and second by
discussing each criterion separately. In contradistinction with existing treat-
ments (such as Zwicky and Sadock 1975 and Cruse 1982), the emphasis will
not just lie on the evaluation of each criterion in isolation, but also on their
mutual incompatibilities; another point of contrast with existing work is that
the distinction between vagueness and polysemy will not simply be assumed,
but will itself turn out to be somewhat doubtful. After a complementary
discussion of some additional problems, the consequences of these findings
will be explored; it will be argued that the instability of the distinction be-
tween vagueness and polysemy not only suggests a reorientation of our con-
ception of lexical meaning, but also raises important questions about the
objectivity of the methodology of lexical semantics. Although the paper will
throughout assume some familiarity with the concepts of linguistic prototype
theory (such as they may be found in Taylor 1989), the paper will proceed at
a relatively slow pace. The reduced speed is, to be sure, not a sign of leisure,
but rather reflects the difficult and often uncharted character of the territory
that we will be venturing into.
2. Intercategorial and intracategorial prototypicality
In the course of the development of prototype studies in linguistics, various
definitions of prototypicality have been suggested. Apart from the absence of
necessary and sufficient definitions, prototypicality has been said to involve
(among others) categorization on the basis of the salient members of a cate-
gory, vague category boundaries, and a radial set relationship among the
various applications of a category. (Compare Lakoff 1987 and Geeraerts
1989c for attempts to chart the different kinds of prototypicality.) It is now
generally accepted that these characteristics need not coincide; they rather
represent prototypicality effects that a single category may exhibit in various
degrees, and that may arise from various sources. Precisely because the
various prototypicality effects need not be co-extensive, some of them might
102 Polysemy
be typical of single meanings (the separate senses of a lexical item), whereas
others could be typical of the meaning of lexical items as a whole (i.e. of
structured sets of senses). Specifically, the difficulty noted in the introduc-
tory paragraph may be resolved by making clear that the absence of defini-
tions characterized by necessity-cum-sufficiency is in fact a true novelty
when it applies to the single senses of a lexical item (whereas it would only
be a traditional indication of polysemy when it applies to the lexical item as
whole). For instance, if it turns out to be impossible to define bird by means
of a single sufficient set of necessary attributes, this is hardly surprising if
the biological sense of bird is taken together with a metaphorical reading as
in A silver bird took the happy couple from Chicago to Santa Cruz. How-
ever, if the biological sense taken by itself (given that it is a single, separate
sense of bird) cannot be defined in the traditional way, this is a result con-
tradicting traditional beliefs about the definability of word senses.
Kleiber (1990) has pushed this kind of reasoning one step further by
bringing other prototypicality effects into the picture. If we distinguish be-
tween the lexical level (consisting of lexical items as forms, i.e. of signifi-
ants), the conceptual level (consisting of cognitive categories that are at the
same time the senses oflexical items), and the referential level (consisting of
real- or possible-world entities), various types of prototypicality effects can
be contrasted. On the one hand (considering the relation between the lexical
and the conceptual level), there is prototypicality in the radial sets or family
resemblance sense, defined as a relationship among the various senses of a
lexical item (each of those senses being a category on the conceptual level).
On the other hand (considering the relation between the conceptual and the
referential level), there is prototypicality in the original sense of Rosch's
psycholinguistic research, involving typicality ratings and degrees of mem-
bership (and also, in the sense of an absence of classical definitions). Here,
the internal structure of one of the distinct meanings of a lexical item is in-
volved (like the biological sense of bird, in contrast with a metaphorical
reading as when an airplane is called a silver bird). Prototypicality is, then,
either a model for the polysemy of lexical items (involving the relationship
between the lexical and the conceptual level), or a model of categorization
(involving the relationship between the conceptual and the referential level).
In the latter case, the prototypicality effects are intracategorial; in the for-
mer, they are intercategorial. I
This model of prototypicality (which is as attractive as it is simple) po-
tentially considerably diminishes the importance of linguistic prototype the-
ory. On the one hand, it could be argued that the 'categorization' relation-
ship between the conceptual and the referential level is the proper domain of
psychology rather than linguistics; linguistics is primarily concerned (in the
structuralist line of thought) with the relationship between signifis and
signifiants, rather than with the relationship between language and the world.
On the other hand, the importance of the 'polysemy' relationship between the
lexical and the conceptual level is only a relative novelty in the history of
linguistics; both the distinction between central and peripheral meanings, and
the idea that those meanings are related by motivated links (like metaphor
and metonymy) are far from new in the history of lexical semantics (see
Geeraerts 1988a).
However, the second part of this interpretation is hardly disconcerting.
Even if prototype theory as a model of the polysemy of lexical items is a
rediscovery of older concerns rather than a complete novelty, it is important
as a rediscovery. In the article that I just referred to, I have argued that pro-
totype theory links up primarily with prestructuralist historical semantics.
Granting, then, that the structuralist approach in lexical semantics has (in its
mainstream development) downplayed the importance of polysemy, the re-
vivification of the polysemic concerns of prestructuralist semantics is an
important step towards a more adequate conception of lexical-semantic
structure. Also, there can be no doubt that prototype theory is more than a
rediscovery of older ideas; the importance of typical individual members of a
category for the structure of that category is, for instance, hardly to be found
in the prestructuralist conceptions of lexical structure.
The other point is potentially more damaging. If the' categorization' rela-
tionship as defined above does not belong to the proper domain of linguis-
tics, the claim of Cognitive Linguistics that it is a full-fledged theory of
categorization will be undermined, because it would have to leave that aspect
of cognitive structure in which the relationship between the conceptual level
and the referential level is studied, to psychology. That is to say, either Cog-
nitive Linguistics would indeed be studying the 'categorization' relationship
together with the 'polysemy' relationship (but then it would be trespassing
on psychological territory), or it would forsake the study of 'categorization'
(but then its cognitive importance would largely evaporate, because the rela-
tionship between the referential level and the conceptual level is the basic
one in the formation of cognitive categories).
This argumentation rests, among other things, on the debatable assump-
tion that existing disciplinary demarcations should be maintained. The fact
that linguistics and psychology now exist as separate disciplines should not
preclude the formation of an interdisciplinary cognitive science in which
I04 P o ~ y s e m y
intercategorial and intracategorial prototypicality effects are studied to-
gether. But then again, independent arguments should be adduced for ex-
plaining why they should be studied together. It may well be interdisciplinar-
ily legitimate in principle to study them together, but what is the positive
motivation for actually doing so? The most drastic reply possible would be
to challenge the distinction between intercategorial and intracategorial
polysemy itself. Methodologically speaking, the model that we have derived
from Kleiber's views requires a coherent way of establishing polysemy, i.e.
of making a distinction between referential differences that do and those that
do not correspond with conceptual differences. If, however, what is merely a
conceptually neutralized referential distinction at one point turns into a case
of polysemy at another, the categorial structure is not stable, and it would be
natural (nay, unavoidable) to study intracategorial prototypicality together
with intercategorial prototypicality. In fact, the very distinction between both
types would have to be rejected.
What is at stake, in other words, is not just the classification of proto-
typicality effects into intercategorial and intracategorial ones, but also the
unity of prototype theory, that is to say, the legitimacy and the relevance of
an interdisciplinary study of cognitive structures, in which the 'linguistic'
study of polysemy and the 'psychological' study of categorization are ines-
capably intertwined. Proponents of a prototype-oriented perspective in lexi-
cal semantics therefore have every reason to have a closer look at the con-
cept of polysemy.
3. Criteria for polysemy - A classification
The distinction between vagueness and polysemy involves the question
whether a particular piece of semantic information is part of the underlying
semantic structure of the item, or is the result of a contextual (and hence
pragmatic) specification. For instance, neighbor is not polysemous between
the readings 'male dweller next door' and 'female dweller next door', in the
sense that the utterance my neighbor is a civil servant will not be recognized
as requiring disambiguation in the way that she is a plain girl ('ugly' or
'unsophisticated'?) does. The semantic information that is associated with
the item neighbor in the lexicon does not, in other words, contain a specifi-
cation regarding sex; neighbor is vague (or general, or unspecified) as to the
dimension of sex.
Vagueness's puzzles, pozvsemy's vagaries 105
The existence of altemative terms (see Norrick 1981: Ill) for the phe-
nomenon that we are concemed with is related to the fact that the term
vagueness has various applications that do not coincide with the kind of
vagueness meant here. Specifically, it is important to distinguish the cate-
gorical lack of specification illustrated above from the referential indetermi-
nacy (or 'vagueness') that may characterize the individual members of a
category or the category as a whole. The former, individual type of referen-
tial indeterminacy is illustrated by knee: it is impossible to indicate precisely
where the knee ends and the rest of the leg begins. The class type is illus-
trated by any color term: it is impossible to draw a line within the spectrum
between those hues that are a member of the category red and those that are
not. In order to bring some structure into this set of interpretations of the
notion 'vagueness', it should be noted that they involve indeterminacy on
three distinct, hierarchically related levels. The knee example is situated on
the level of individual entities; the question where a particular entity (such as
a knee) ends is specified as the question whether a potential part of the entity
(say, a specific area of the leg) is part of the entity or not. The red example
is situated on the level of groups of entities, i.e. conceptual categories; the
question conceming the boundaries of a particular category is specified as
the question whether a particular entity (such as a particular hue) is part of
the category or not. The neighbor example of vagueness, finally, is situated
on the level of polysemous lexical items, considered as groups of related
conceptual categories; the question where exactly the boundaries of a spe-
cific polysemous item lie is specified as the question whether a potential
conceptual category (such 'male person living next door') is indeed a mem-
ber of the set of meanings of a particular item (such as neighbor). While we
concentrate here on the third kind of vagueness, it is mainly the second kind
that has received a lot of attention in the recent literature on the subject of
vagueness.
2
,3
The difficulties created by the terminological divergences are enhanced
by the existence of various kinds of tests for distinguishing between vague-
ness and polysemy. A definitive treatment of this subject does not yet exist;
despite its importance for theories of semantic structure, it has received rela-
tively little systematic and continuing attention. Without discussing all spe-
cific tests that have been suggested, three types of criteria will be distin-
guished here. (The most comprehensive treatment of tests for lexical
polysemy that I am aware of is to be found in eruse 1982, which will fur-
nish a basic reference point for the following discussion
4
.
5
Other relevant
works will be mentioned in passing in the course of the following pages.)
106 Polysemy
First, from the truth-theoretical point of view taken by Quine (1960:
129), a lexical item is polysemous if it can simultaneously be true and false
of the same referent. Considering the readings 'harbor' and 'fortified sweet
wine from Portugal' of port, the polysemy of that item is established by
sentences such as Sandeman is a port (in a bottle), but not a port (with
ships). Or, in the words of Quine: 'An ambiguous term such as light may be
at once clearly true of various objects (such as dark feathers) and clearly
false of them' (1960: 129).
Second, linguistic tests involve semantic restrictions on sentences that
contain two related occurrences of the item under consideration (one of
which may be implicit or deep-structural); if the granmlatical relationship
between both occurrences requires their semantic identity, the readings that
can be attributed to the resulting sentence may be an indication for the
polysemy of the item. For instance, the identity test described by Zwicky and
Sadock (1975) applies to constructions that were assumed in the Chomskyan
'standard theory' to involve transfomlations such as conjunction reduction
and so-reduction, which require the semantic identity of the items involved in
the reduction. (It may be noted that constructions such as these are nowa-
days no longer discussed in transformational terms. However, as the current
term 'identity-of-sense anaphora' indicates, the idea that there are semantic
restrictions on the construction remains intact. Note also that the type of test
mentioned here did not originate with Zwicky and Sadock; in particular, see
Lakoff 1970.) Thus, at midnight the ship passed the port, and so did the
bartender is awkward if the two lexical meanings of port are at stake; disre-
garding puns, it can only mean that the ship and the bartender alike passed
the harbor (or, perhaps, that both moved a particular kind of wine from one
place to another). A 'crossed' reading in which the first occurrence of port
refers to the harbor, and the second to wine, is normally excluded. Con-
versely, the fact that the notions 'vintage sweet wine from Portugal' and
'blended sweet wine from Portugal' can be crossed in Vintage Naval is a
port, and so is blended Sandeman indicates that port is vague rather than
polysemous with regard to the distinction between blended and vintage
wmes.
Third, the definitional criterion (as infomlally stated by Aristotle in the
Posterior Analytics II.xiii) says that an item has more than one lexical
meaning if there is no minimally specific definition covering the extension of
the item as a whole, and that it has no more lexical meanings than there are
maximally general definitions necessary to describe its extension. Definitions
of lexical items should be maximally general in the sense that they should
Vagueness's puzzles, p o ~ v s e m y 's vagaries 107
cover as large a subset of the extension of an item as possible. Thus, sepa-
rate definitions for 'blended sweet fortified wine from Portugal' and 'vintage
sweet fortified wine from Portugal' could not be considered definitions of
lexical meanings, because they can be brought together under the definition
'sweet fortified wine from Portugal'. On the other hand, definitions should
be minimally specific in the sense that they should be sufficient to distin-
guish the item from other non-synonymous items. A maximally general defi-
nition covering both port 'harbor' and port 'kind of wine' under the defini-
tion 'thing, entity' is excluded because it does not capture the specificity of
port as distinct from other items.
4. Criteria for polysemy - Remarks and refinements
There are three important (sets of) remarks to be made with regard to this
classification of types of criteria for distinguishing between polysemy and
vagueness. First, a number of tests have to be briefly mentioned that have
not been included in the classification because their insufficiency is generally
accepted. Second, the basic types enumerated above may be supplemented
with variants (a possibility that will turn out to be of particular importance
for the logical test). And third, there are some remarks about the theoretical
background of the tests (and in particular, the definitional test) that are
worthwhile mentioning.
4.1. Omissions
For the sake of completeness, let it be noted that a number of tests whose
insufficiency or impracticability has been sufficiently established elsewhere
have not been included in the classification presented above. In particular,
these involve the 'indirect' criteria distinguished by Cruse (1982), which
refer to the relations that an item may contract with other items in the lan-
guage (or in other languages, such as when the existence of different transla-
tions for a lexical item is taken as proof for the polysemy of that item). (The
criteria discussed and rejected by Herringer 1981: 117-118 can be included
in the same class.) As an illustration, consider the view according to which
the existence of distinct antonyms for specific readings of a lexical item
serves as proof of its polysemous nature. In the case of light, the reading
'not having much weight' contrasts with heavy, and the reading 'having a
bright color' contrasts with dark. Although, in other words, the antonymy
108 Polysemy
test seems to work for light, there are various counter-examples. On the one
hand, there are items with identical antonyms for distinct readings, as in the
case of old, which has new as its opposite both in the reading 'former' and
in the reading 'advanced in age'. On the other hand, there are items with
distinct antonyms for what are merely distinct referential specifications
rather than distinct meanings. For instance, when the Dutch word vers
'fresh' is applied to bread, it contracts a relationship of anton;my with oud
'old' but also with oudbakken 'stale'; when vers applies to meat, the anto-
nyms would be oud and adellijk 'high, gamy'. Intuitively and definitionally,
however, the two applications of vers constitute the single reading 'recent,
and as such, optimal for consumption'. Examples such as these lead eruse
to the evaluation that the 'indirect' criteria are less satisfactory than the 'di-
rect' ones, which focus directly on an item's behavior in various contexts
6
4.2. Variants
A second major point concerning the classification of polysemy criteria in-
volves the presence, within each type, of other tests than the ones mentioned
above. The three types of criteria are meant precisely as types: classes of
tests that have something crucial in conunon. If, then, it is relevant to group
polysemy criteria into the three classes identified above, we should be able to
place specific tests that have not yet been discussed into one of the three
classes. Because this point is specifically important for the first criterion, let
us now have a look at the distinct varieties of the truth-functional approach,
trying to bring some order into the scattered suggestions made by other au-
thors. The starting-point for distinguishing between subtypes of the truth-
functional test is the recognition that Quine' s proposal rests on two features:
first, polysemy is signaled by truth-conditional differences between the two
alleged meanings, and second, these truth-conditional differences are re-
vealed by applying the two readings to a single situation. Truth-functional
criteria for polysemy that do not strictly coincide with the Quinean test as
defined above, can now be introduced by taking these two features as a
guide-line; basically, they contain variations on those features as involving
the questions 'What kind of truth-conditional differences are we looking
for?', and 'How do we diagnose them?', respectively.
With regard to the first feature, Quine himself already noted that next to
the oppositeness of truth value implied by the 'p and not p' criterion, a third
truth value 'nonsensical' or 'irrelevant' may have to be taken into account.
Questions and chairs can both be hard, but it is difficult to find a situation in
Vagueness's puzzles, 's vagaries 109
which both readings of hard are relevant (but in which one would be appli-
cable and the other not). Although Quine seems to suggest that such a situa-
tion could be treated as a case of vagueness rather than polysemy, it is much
more in line with the essence of the truth-functional approach to meaning
discrimination, to assume (with Norrick 1981: 113) that a single sentence
such as This book is sad may signal the polysemy of sad. The fact that the
reading 'evoking sadness' of sad may render the sentence true or false ac-
cording to the circumstances, whereas the reading 'experiencing sadness'
yields a nonsensical interpretation, implies that there is a crucial distinction
in the truth-functional properties of This book is sad under both interpreta-
tions. Given that the distinction depends on a distinction in the meanings we
attribute to sad, this should be as good evidence for the polysemy of sad as a
straightforward 'p and not p' case. The truth-functional differences involved
in the test should, in other words, be taken broadly so as to include the dif-
ference between 'either true or false' and 'nonsensical'.
Because the reading 'this book is experiencing sadness' turns This book
is sad into a semantically ill-formed construction, it is difficult to have a
single well-formed sentence of the 'p and not p' type in which the relevant
interpretation of sad and the nonsensical interpretation are combined. This
means that (as far as the second feature of the basic Quinean test is con-
cerned) the differences in truth-functional properties that we are looking for
will not be signaled by the spontaneous acceptability of a sentence of the 'p
and not p' type, but will involve an active consideration of the properties of
potentially distinct readings. The 'p and not p' test does not initially require
a lot of conscious thinking about different meanings: if it is spontaneously
clear that a feather can be light and not light, the polysemy of light is proba-
bly firmly established, psychologically speaking; it springs to mind so easily
that the non-contradictory nature of the 'p and not p' construction is auto-
matically recognized. In this sense, the 'p and not p' test has the same attrac-
tiveness that grammaticality judgments have in syntax: the judgment is un-
mediated. In the case of This book is sad, however, there is a lot of
conscious mediation and active introspection involved, mainly because one
has to actively conceptualize the different potential readings of the item and
evaluate their consequences. Once this more active and
more mediated approach is accepted (and it will have to be accepted if it is
correct that the 'p and not p' frame cannot be applied to cases involving a
nonsensical reading), varieties of this active search for truth-conditional
differences may be explored. I shall mention two, but I do not want to claim
that these are the only possibilities. We shall see later (in section 6.2) that
110 Polysemy
the variants of the logical criterion presented here run into trouble, but for
the moment, we shall merely present them.
To begin with, if a non-contradictory use of the 'p and not p' frame indi-
cates the presence of polysemy, so does a non-redundant use of the 'p and p'
frame. If p has two meanings, it should be possible to assert 'p and p' with-
out raising feelings of redundancy or awkwardness. That is to say, while the
main 'p and not p' version of the logical test considers cases in which a sen-
tence can be used in the same situation with opposite truth values, the 'p and
p' variant considers cases in which p is used in the same situation in two
readings with the same truth value. The obvious question then is how we can
know whether p indeed has two distinct interpretations. Apart from the pos-
sibility of asserting 'p and p' without redundancy or awkwardness (a possi-
bility that is seldom noted in the literature), there is the introspective psycho-
logical phenomenon that in the process of interpreting the sentences in
question, our attention seems to be switching continuously between one read-
ing and the other. Take the sentence Charles has changed his position, in
which Charles is an official delegate at a political summit meeting (the ex-
ample is Cruse's ; cp. Cruse 1982: 70). The readings involved can be para-
phrased as 'Charles has moved to another place in the conference hall' and
'Charles has changed his opinion concerning the topic discussed at the con-
ference'. Now, even in those cases where both readings are true, it is diffi-
cult to mentally conceive of them together, i.e. to fuse them into a single
reading. Rather, at anyone time, we seem to be able to have only one of the
readings at the same time in mind. In the words of Herringer (1981: 95):
Wichtig ist also dieses 'entweder/oder', und das, was man das Kippen zwi-
schen den Bedeutungen genannt hat: Es gibt keine fliessenden Dbergange
zwischen den Bedeutungen, sondern einen Aspektwechsel, der sich pl6tz-
hch und oft iiberraschend einsteUt.
[What is important, then, is this either/or situation, and what has been
called the tilting between meanings: There is no fluent transition between
the meanings, but an aspectual change that imposes itself suddenly and by
surprise.]
Herringer (1981) also discusses a second variant of the basic logical crite-
rion: the truth-conditional differences between sentences can be revealed by
taking into account the implications of potentially distinct meanings: if these
are indeed distinct, they should have different implications. For instance, the
French sentence in (1) contains the polysemous word louer, which may
mean either 'to rent' or 'to praise'. In one reading, (1) entails (2), but in
Vagueness's puzzles, vagaries III
another, it entails (3). However, if we combine these facts into the single
implication in (4), and then compare that sentence with (5), where an unsus-
pected case of vagueness is involved, it becomes clear that the format of the
entailment as such does not always distinguish between vagueness and
polysemy: the necessity to express the entailments of a sentence in disjunc-
tive form does not sufficiently distinguish between both phenomena.
(1) 11 a loue son appartement.
(2) 11 s 'est oblige de payer son appartement.
'He has taken on the obligation to pay the apartment.'
(3) 11 a dit du bien de I 'appartement.
'He has spoken well of the apartment.'
(4) S'il a loue I 'appartement, il s 'est oblige de payer ou il en a dit du bien
'If he has loue the apartment, either he has taken on the obligation to
pay for it, or he has spoken well of it. '
(5) Si c 'est un enfant, c 'est un garr;on ou unefille.
'If it is a child, it is a boy or a girl. '
(6) lfit is not a child, it is not a boy and it is not a girl.
(7) If they did not reach the bank in time, either they did not reach the
money-bank in time, or they did not reach the river-bank in time.
This problem can be avoided, however, when negation is added to the test
(cp. Kempson and Cormack 1981, Cruse 1982). When a vague predicate is
negated, the specifications of the predicate are both negated as well. In (6),
for instance, the distinction between male children and female children (an
example of vagueness with regard to child) lies at the basis of a conjunction
of negations. On the other hand, in the case of polysemy, the negation need
not extend to both readings at the same time. Thus, in (7), the negated am-
biguous readings occur as a disjunction. Notice, in this respect, that the
disjunction is necessary if the 'p and not p' frame is to be applicable in the
case of polysemy: if PI and P2 are the readings involved, and if the negated
form 'not p' would automatically entail both 'not PI' and 'not P2', an inter-
pretation of 'p and not p' as 'PI and not P2' or 'P2 and not PI' would be ex-
cluded.
4.3. Specifications
To round off the discussion of the classification of polysemy criteria, some-
thing more has to be said about the third criterion, if only because it is ab-
sent in the works we have been referring to. The criterion is of particular
importance because it so to speak looks into the lexical items to determine
112 Polysemy
how the various meanings of the item are constituted. Whereas the other two
criteria are at best diagnostic tools that signal the presence of polysemous
readings, the third criterion is an analytical one that examines the internal
constitution of the meanings of the item. In order to be better able to appre-
ciate the impact of this definitional approach, let us have a closer look at
Aristotle's description of it.
We should look at what are similar and undifferentiated, and seek, first,
what they all have that is the same: next, we should do this again for other
things which are of the same genus as the first set and of the same species
as one another but of a different species from those. And when we have
grasped what all these have that is the same, and similarly for the others,
then we must again inquire if what we have grasped have anything that is
the same - until we come to a single account; for this will be the definition
of the object. And if we come not to one but two or more accounts. it is
clear that what we are seeking is not a single thing but several.
I mean, e.g., if we were to seek what pride is we should inquire, in the case
of some proud men we know, what one thing they all have as such. E.g. if
Alcibiades is proud, and Achilles and Ajax, what one thing do they all
have? Intolerance of insults; for one made war, one waxed wroth, and the
other killed himself. Again in the case of others, e.g. Lysander and Socra-
tes. WelL if here it is being indifferent to good and bad fortune, I take these
two things and inquire what both indifference to fortune and not brooking
dishonor have that is the same. And if there is nothing, then there will be
two sorts of pride (Barnes 1984: 161). [An alternative, more literal reading
for the Greek megalopsychia is 'greatness of the soul' rather than 'pride'.]
Granting that Aristotle does not seem to make a distinction between the
definition of the word pride and the definition of the thing 'pride' (granting,
in other and more traditional words, that his approach to definition is a real-
ist rather than a nominalist one), the quotation helps to see the close connec-
tion between the definitional criterion of polysemy and the classical concep-
tion of definitions as sufficient sets of necessary criteria for class
membership. The whole criterion could, in fact, be rendered in simplified
form as the statement that an item is polysemous (or, alternatively, that what
was thought to be a single category in fact consists of several categories) if
there is no single set of necessary and sufficient attributes to describe the
extension of the item or the alleged category. The full criterion does not,
however, coincide entirely with this simplified version, because the latter
only specifies the lower limit of an item's polysemy: if there is no single set
of necessary and sufficient attributes that is minimally specific, the item has
at least two meanings. But the definitional criterion of polysemy also sets an
Vagueness's puzzles, polvsel1ly 's vagaries 113
upper limit to the polysemy of an item, by specifying that an item has no
more meanings than the number of maximally general definitions that are
needed to cover its entire range of application. If, for instance, the two defi-
nitions 'indifference to fortune' and 'intolerance of dishonor' are sufficient
to cover all examples of proud men that you know of, mega/opsychia does
not represent more than two meanings. Notice that this 'upper limit' specifi-
cation of the simple 'lower limit' rule that an item is polysemous if it cannot
be defined by means of a single definition, is an entirely natural one: does
not the very fact that we seem to be looking for single definitions first sug-
gest that the crucial thing about definition is generality? In popular parlance,
the definitional criterion might then read as follows: 'if you cannot achieve
maximal generality, settle for the next degree of generality that you can at-
tain'.
After Aristotle, the definitional criterion for polysemy became a standard
part of treatises on logic as the requirement that definitions should be neither
too broad nor too narrow (see, for instance, Copi 1972: 138). The transition
is not a straightforward one, however, to the extent that a requirement on
definitions is not necessarily a criterion for polysemy: imposing restrictions
on definitions would seem to imply that the definer already knows for certain
what the definiendum is, i.e. what the lexical meanings are that are to be
defined. However, if the definiendum is taken to be a lexical item rather than
an independently discerned lexical meaning, the requirement actually turns
into a criterion for the polysemy of the item. In that case, the requirement
that a definition should not be too narrow embodies the principle that maxi-
mal generality should be aspired at, whereas the requirement that a definition
should not be too broad reflects the necessity of minimal specificity as de-
fined above. In actual practice, the fonnulations of the requirement that may
be found in logical treatises waver between an interpretation of the target of
the definition as a word and an interpretation as a word meaning. For in-
stance, Copi's example ofa definition that is too narrow (1972: 138) is said
to involve 'the word shoe', but practically, it seems that only the meaning
'outer foot -covering for a person' is at stake, and not (for instance) the
meanings 'horseshoe' or 'thing that is like a shoe in shape or use (such as
wheel-drags, sockets, ferrules etc.)'. It would seem, in other words, that the
crucial distinction between both interpretations of the standard requirement
on definitions is not always made.
7
It should be emphasized that the definitional polysemy test is particularly
important in comparison with the other two: it is the only one that embodies
a hypothesis about the principles of categorization that human beings em-
114 Polysemy
ploy. It suggests, in fact, that one of the basic characteristics of naturallan-
guage categorization is a tendency towards generality (or, in other words,
towards maximal abstractions). Thus, while the other two classes of
polysemy criteria merely contain diagnostic tools for establishing whether
interpretations of lexical items constitute distinct meanings, the definitional
test gives us an explanatory surplus, because it is based on an assumption
about why particular senses are distinguished. It is all the more regrettable,
then, that the definitional test has hardly received any explicit attention in the
recent linguistic literature on the subject of lexical polysemy.
5. Problems with the polysemy tests - First series
Given the existence of three major types of tests for lexical polysemy, we
can now try to determine their mutual relationship. We will concentrate pri-
marily on the tests that were earlier presented as the basic representative of
each type, rather than bringing all possible variants of each type (as dis-
cussed in the previous paragraph) into the comparison.
5.1. Pairwise divergences among the criteria
Just a few lexical items may suffice to show that the three tests, taken pair-
wise, may yield different results. First, consider an autohyponymous word
such as dog (i.e. a word one of whose senses is a superordinate category
with regard to one of the other senses). In the reading 'Canis familiaris', dog
is a hyperonym of dog in the reading 'male Canis familiaris'; in the hypero-
nymous reading, dog contrasts with (for instance) cat, whereas the proper
antonym is bitch in the hyponymous reading. But are these readings really
distinct meanings? Applying the tests, it will be easily verified that dog is not
polysemous according to the definitional criterion, but does exhibit polysemy
according to the logical criterion. Taking Lady (from Disney's Lady and the
Tramp) as a prime example of a female dog, it is possible to assert Lady is a
dog alright. but she is not a dog. On the other hand, according to the defini-
tional criterion, the hyponymous 'male Canis familiaris' reading is always a
proper subset of the superordinate reading, so that no definition of that sub-
set could ever claim the maximal generality that is required. Now, what
about to the linguistic test? According to the linguistic test, a 'crossed' read-
ing of sentences containing identity of sense anaphors (or similar construc-
tions) should come out grammatically awkward. That is to say, interpreta-
tions of such a sentence in which the full occurrence of the item under con-
sideration receives meaning x, and the anaphorical element receives meaning
y, are ungrammatical. In the case of dog, however, it could be doubted
whether such an ungrammaticality would ever surface as positive evidence
in favor of the polysemy of dog, given that the anaphorical sentence would
always automatically be interpreted as a grammatical construction involving
just a single one of the two readings of dog. For instance (assuming that
Fido is a male dog), Fido is a dog and so is Lady is spontaneously under-
stood as the correct assertion 'Lady and Fido are both members of the spe-
cies Canis familiaris', or conversely, as the false assertion that both animals
are male members of that species. In neither case is the sentence spontane-
ously felt to be odd in the sense in which At midnight the ship passed the
port, and so did the bartender is. Of course, it could be argued that the very
fact that Fido is a dog and so is Lady can be either true or false (depending
on the reading we attach to dog) in the same situation, testifies to the
polysemy of dog. This is correct, but the invocation of truth-functional dif-
ferences indicates that we are then dealing with a variant of the logical test,
rather than with the linguistic grammaticality test that we are concerned
with.
Let us accept, therefore, that the linguistic test does not testify to the
polysemy of dog. We will then have established a distinction between the
definitional test and the linguistic test on the one hand, and the logical test on
the other. Next, let us consider a case in which the linguistic and the defini-
tional test are at odds. Examples are easy to find (in particular, see Nunberg
1979 and Deane 1988). In the sentence The newspaper has decided to re-
duce its size, the item newspaper refers first to the management or the board
of directors of the paper, and then to the paper as a material object. The
perfect grammaticality of the sentence does not allow us to spot any
polysemy on linguistic grounds. Definitionally, on the other hand, it does not
seem possible to find a description that fits the 'board of directors' usage
and the 'actual. material publication' usage into a single well-defined cate-
gory. This implies that there is a definitional polysemy in the concept news-
paper that is obscured in the sentence The newspaper has decided to reduce
its size. And this in turn implies that we now have examples establishing for
each of the three pairs of tests that can be construed out of the basic set of
three criteria, that the members of those pairs may be at odds with each
other: while the dog example dealt with the definitional/logical and the lin-
guistic/logical pairs, the newspaper example takes care of the defini-
tionalllinguistic couple.
116 Polysemy
To complete the analysis of the newspaper example, let us notice that
applying the logical test also yields polysemy. Speaking of the overpowering
presence of a certain editor-in-chief, we could easily utter something like
That man is the newspaper, meaning that he is the driving force of the whole
business. Because the utterance would be nonsensical if newspaper were
interpreted in the 'material publication' sense, the variant of the logical test
that we discussed earlier in connection with This book is sad reveals the
polysemy of newspaper. It can now also be appreciated that our earlier diffi-
culties with the application of the linguistic test to dog are of relative impor-
tance. If we were to reject our earlier views on the matter and accept that, by
some kind of mental aerobatics, a crossed reading of Fido is a dog and so is
Lady would not only be possible but would also be recognized as odd,
8
the
dog example could no longer be invoked to establish that the linguistic test
and the logical test are not co-extensive. Even then, however, we could rely
on the newspaper example to bring home the point.
5.2. Three-way divergence in a single item
For further illustration, let us now consider a category that is a typical ex-
ample of prototypicality. The category fruit is, in fact, among the categories
originally studied by Rosch (see, in particular, Rosch 1975 and Rosch and
Mervis 1975 for experimental results). An immediate difficulty withfruit is
the vagueness of membership in the category (a problem that is less outspo-
ken, if not absent, in the case of newspaper and dog): is a coconut or an
olive a fruit? Notice, first, that we are not concemed with the technical, bio-
logical sense offruit, but with our folk model of fruit as a certain category
of edible things. Technically, any seed-containing part of a plant is the fruit
of that plant; as such, nuts in general are fruit. In ordinary language, on the
other hand, nuts and fruit are basically distinct categories (regardless of the
possible boundary status of the coconut): nuts are dry and hard, while fruits
are soft, sweet, and juicy; also, the situations in which nuts and fruits are
eaten are typically different. Second, remember that category membership is
not the same thing as typicality: a penguin is undoubtedly an uncharacteris-
tic kind of bird, but it is a bird nonetheless; as to the olive, the question is
not just whether it is a typical fruit, but rather whether it is a fruit at all.
This indeterminacy has an inm1ediate bearing on the applicability of the
various polysemy tests.
To begin with, the borderline status of olives shows up in the following
exchange: Daddy, is an olive a fruit? - Well, it is and it is not (that is to
say, it has some of the characteristics of the things that we typically call
fruits, but it also differs from the clear cases on a number of important at-
tributes). The 'logical' polysemy that shows up here is similar to the use of
hedges: 'up to a point', or perhaps even 'strictly speaking', an olive is a
fruit, but seen from another angle, it is not. Applying the linguistic polysemy
test, on the other hand, does not seem to yield indications of polysemy. De-
pending on one's conception of the category fruit, the sentence An orange is
frUit and so is an olive would either seem to be simply true or false, but not
ungramrnatical (as the linguistic test would have it if there is polysemy).
Further, a definitional analysis is initially hampered by the uncertainty sur-
rounding the extensional boundaries of the category. If there is consensus
that olives are not fruit, we should not include the olive in an analysis of
frUit. Conversely, if an olive is considered to be a fruit (however peripheral
and uncharacteristic), it will have to be included. (We might then expect the
definitional analysis to reveal the same polysemy that is illustrated by
Daddy, is an olive afruit? - Well, it is and it is not.)
To circumvent the problem with olives and their likes, let us restrict the
definitional analysis to clear cases of fruit, i.e. cases for which doubts about
membership do not play a role. Even for these clear cases, it can be shown
that ji'uit cannot be defined by means of a single minimally specific set of
necessary attributes. Because such absences are often assumed but seldom
demonstrated in the literature, it may be useful to be a bit more specific on
this point than is usual. A starting-point for a consideration offruit can be
found in Wierzbicka's (characteristically lengthy) definition of the category
(1985: 299-300). In order to show that this is not a classical, necessary-and-
sufficient definition, it has to be established, on the one hand, that not all
attributes of fruit as mentioned by Wierzbicka are general (even within the
set constituted by the examples of fruit that are high on Rosch's typicality
ratings), and on the other hand, that the remaining set of general attributes is
not minimally specific, i.e. does not suffice to distinguish fruit from, for
instance, vegetables.
The following characteristics mentioned by Wierzbicka are not general. (I
repeat Wierzbicka's formulations, though not in the order in which she pre-
sents them.)
(a) They have a skin harder than the parts inside.
(b) They have some small parts inside, separate from the other parts, not
good to eat. These parts put into the ground could grow into new things of
the same kind growing out of the ground.
118 Polysemy
(c) They are good to eat without being cooked, without having anything
done to them, without any other things, and people can eat them for pleas-
ure.
(d) Eating them uncooked makes one feel good.
(e) Before they are good to eat they can be sour.
(f) They have a lot ofjuice.
(g) Their juice is good to drink.
(h) They are also good to eat dried.
Characteristic (a) is contradicted by the strawberry, which has no skin
worthy of that name. Strawberries likewise do not have the seeds mentioned
in (b); bananas are another case in point. Attributes (c) and (d) indicate that
fruit can be eaten (with pleasant results) without further preparation, but this
does not seem to hold for the lemon, whose sour taste generally requires
sugaring. (Even if this counter-example were not accepted, adding (c) and
(d) to the list of attributes that are general for fruit would not solve the prob-
lem that that list does not suffice to distinguish fruits from some vegetables
and nuts.) Attributes (e) and (t) are not valid for the banana: first, an unripe
banana is bitter rather than sour, and second, there is no juice in a banana.
Because the generality of (g) depends on the generality of (t), it may likewise
be discarded. Finally, as far as (h) is concerned, it is difficult to imagine a
dried melon as being good to eat.
Next, there is a set of characteristics whose non-generality seems to be
accepted (or at least, implied) by Wierzbicka herself.
(i) Wanting to imagine such things, people would imagine them as grow-
ing on trees.
(j) They can be small enough for a person to be able to put easily more than
one thing of this kind into the mouth and eat them all at the same time, or
too big for a person to be expected to eat a whole one, bit by bit, at one
time, but wanting to imagine such things, people would imagine them as
too big for a person to put a whole one easily into the mouth and eat it, and
not too big for a person to be expected to eat a whole one, bit by bit, at one
time, holding it in one hand.
(k) After they have become good to eat they are sweet, or slightly sweet, or
sour but good to eat with something sweet.
(I) Wanting to imagine such things after they have become good to eat,
people would imagine things which are slightly sweet.
(m) Things on which such things can grow can also grow in some places
where people don't cause them to grow, but wanting to imagine such
things, people would imagine them as growing on things growing out of
the ground in places where people cause them to grow.
Vagueness's puzzles, p o ~ y s e m y 's vagaries 119
While (k) is a disjunctively defined attribute (i.e., is a superficial combina-
tion of two characteristics that are each not general when taken separately),
the other features are introduced by the formula wanting to imagine such
things, people would imagine them as; this would seem to indicate that the
attribute is merely typically associated with the concept, rather than being
general. For instance, the sweetness mentioned in (I) does not hold for lem-
ons, and berries do not grow on trees, in contradistinction with the feature
involved in (i)9
The set of general characteristics that is left over after the elimination of
the previous sets contains the following features.
(n) They grow as parts of certain things growing out of the ground.
(0) They don't grow in the ground.
(p) They become good to eat after they have grown long enough on the
things growing out of the ground.
(q) Before they are good to eat they are green or greenish outside.
(r) People cause things of this kind to grow in many places because they
want to have those things for people to eat.
(s) They are good to eat cooked with sugar, or cooked as part of some
things which have sugar in them.
Is this set minimally specific? Up to characteristic (r), the set applies not
only to fruit, but also to nuts, herbs, and large collections of vegetables
(though not to the ones that grow in the ground, like carrots), so that the
crucially distinctive attribute would be (s). However, if one takes into con-
sideration the use of almonds and other nuts in certain types of pastry, the
use of herbs (such as tansy) in pancakes, and the habit of cooking rhubarb
with sugar, it soon becomes clear that there are counter-examples with re-
gard to (s) in each of the three categories (nuts, herbs, and vegetables). All in
all, most of the attributes mentioned by Wierzbicka are not general, whereas
those that are, taken together, apparently do not suffice to exclude non-
fruits.
Given, then, that we cannot define the uncontroversial core applications
of fruit in a classical, necessary-and-sufficient fashion, let us have another
look at the other two polysemy tests. The linguistic test still does not yield
indications of polysemy: An orange is a fruit and so is a banana is straight-
forwardly grammatical. Furthermore, the logical test does not seem to signal
polysemy either: a sentence like A strawberry is a fruit and not a fruit is
simply false, in spite of the fact that the strawberry does not belong to all the
maximal subsets that we can distinguish in the structure offruit at the same
time. That is to say, if each of those maximal subsets is a distinct meaning
120 Polysemy
according to the definitional approach, it should be possible to reveal that
polysemy by applying the logical criterion; but apparently, such an interpre-
tation is impossible for A strawberry is a fruit and not a fruit. However,
consider the following exchange: 1<; a penguin a bird? - Well, it is and it
isn't (that is to say, strictly speaking it is a bird, but it is not a typical bird).
Taking into account that there is no doubt about the membership status of
penguins in the category of birds, the 'p and not p' construction here in-
volves differences in typicality: it sets off the prototypical application from
peripheral cases. We now see that we could also have exchanges like the
following: Is a mango a fruit? - Well, it is and it isn't (strictly speaking it
is, but it does not have the typical characteristics of typical fruits like or-
anges and apples). If it is correct that the logical test does reveal a polysemy,
and if this polysemy precisely involves the distinction between the typical
and the non-typical cases of a category, it is also understandable that the 'p
and not p' frame does not reveal any polysemy when applied to a highly
typical case such as strawberry (even though it is perhaps not the single
most typical case).
But although the logical criterion does, in this case, yield polysemy, it
nevertheless still does not lead to precisely the same kind of polysemy as the
definitional test: if at all, the polysemy test can at most distinguish between
core and periphery, but it cannot single out any and each of the definitionally
maximal subsets that constitute the senses of fruit according to the defini-
tional approach. In other words, the fruit example presents us with a case in
which the application of all three criteria for polysemy yields different re-
sults. What are distinct meanings from one point of view (that is, according
to one test) are reduced to cases of vagueness from another point of view.
6. Problems with the polysemy tests - Second Series
At this point, two possibilities for further developments open up. In the first
place, the recognition that the various criteria lead to different answers to the
question what the meanings of a word are, may lead to the question what
exactly those criteria are criteria of. If they result in contradictory statements
as to what the meanings of a word are, not all of them can actually be crite-
ria for meaning discrimination; in fact, only one could be the true test of
polysemy. Pursuing this line of analysis, then, would mean eliminating the
wrong criteria until the only true one is left over. Such an approach would
still presuppose that words have a stable set of meanings, though. Hence, the
Vagueness's puzzles, polysemy 's vagaries 121
second road to follow would be to undennine this assumption even more
than we have done already. More specifically, the belief in the stability of
meanings will turn out to be ultimately spurious if the vagueness of the
boundary between vagueness and polysemy can also be established when we
restrict the analysis to just one of the three basic types of polysemy test (and
not just when we compare the three among each other). Because this is a
more radical and more fundamental departure from our traditional concep-
tions than the view that there is only one true kind of polysemy test, it is
worthwhile pursuing this line of research first. The hypothesis, then, will be
this: for each of the polysemy criteria, examples can be found in which what
is a distinct meaning in one context, is reduced to a case of vagueness in
another context (and vice versa).
6. I. The linguistic criterion
Contextual influences on the linguistic test have been (implicitly or explic-
itly) noted by several authors. In fact, the recognition occurs relatively early
in the literature on the subject. When Lakoff (1970) introduced the and so-
construction as a criterion for polysemy, he argued that hit is ambiguous
between an intentional and an unintentional reading, because John hit the
wall and so did Fred would constitute an anomalous utterance in situations
in which John hit the wall intentionally but Fred only did so by accident (or
the other way round). Catlin and Catlin (1972), however, noted that the sen-
tence could easily be uttered in a context involving imitation. A situation in
which John hits his head against the wall after stumbling over his vacuum
cleaner and is then comically imitated by Fred, might very well be described
by the sentence in question. Nunberg (1979) further drew the attention to
sentences such as The newspaper has decided to change its size, which
features intuitively distinct senses of newspaper ('management, board of
directors' and 'material publication'). Of course, the point would be to show
that the distinction between both senses is not just a matter of intuition, but
that the linguistic test itself also supports the polysemy of the item under
consideration. This can be done by concentrating on the asymmetries pointed
out by Deane (1988). On the one hand, there are cases like the well-known
The ham sandwich is sitting at table 5 and (with anaphora) The ham sand-
wich
i
tasted lousy, so hei demanded the check. These cases show neutraliza-
tion of the distinction between the literal interpretation of ham sandwich and
the customer metonymically connected with it. On the other hand, The ham
sandwich
i
demanded the check because it
i
tasted lousy is ruled out.
122 Polysemy
While the examples presented so far involve anaphoric references, similar
cases can be found involving coordination. For instance, Norrick (1981:
115) contrasted the (decidedly odd) sentence Judy's dissertation is thought
provoking and yellowed with age with the (perfectly natural) construction
Judy's dissertation is still thought provoking though yellowed with age. If
the coordination generally requires that dissertation be used in the same
sense with regard to both elements of the coordinated predicate, the sen-
tences show that the distinction between the dissertation as a material prod-
uct and its contents mayor may not play a role. Cruse (1982), in his turn,
noted that none of the following series of sentences containing coordination
produces feelings of oddity:
(8) John likes blondes and racehorses.
(9) John likes racehorses andfast cars.
(10) John likes cars and elegant clothes.
(11) John likes elegant clothes and expensive after-shave.
(12) John likes expensive after-shave and vintage port.
(13) John likes vintage port and marshmallows.
Coordinating the first item in the series with the last, however, does produce
an awkward sentence. So, while the awkwardness of John likes blondes and
marshmallows would normally be taken as evidence for the polysemy of
like, the pairings mentioned above suggest that there is a continuum of
meaning rather than a dichotomy. Cruse concludes 'that readings which are
close together can be coordinated without zeugma, but if they are sufficiently
far apart, they are incompatible. If this picture is correct, it does not make
sense to ask how many senses of like there are: there is just a seamless fabric
of meaning-potential' (1982: 79).10
The previous examples sufficiently establish the context-dependent nature
of the linguistic test. These contextual effects are fairly strong, in the sense
that some contexts appear to be able to bridge the gap between widely diver-
gent meanings. Deane (1988: 345) points out that explicit comparison is
such a context; it can even combine actual homonyms. Examples (14) and
(15) reproduce his examples. They can be contrasted with cases such as
(16), in which the same meanings cannot be combined.
(14) The arm ofa giant resembles that ofan ocean.
(15) Financial banks resemble those that you find by rivers: they control.
respectively, the flow ofmoney and o[water.
(16) *The giant put his arm into that o[the ocean.
(17) When Von Mi1nchhausen said a gigantic arm appeared before him
when he came out ofthe jungle in Xanadu, he actually did mean the
arm ofa giant, not that ofan ocean.
(18) The bank John had trouble finding again was not the one where he had
cashed the cheque, but the one where he had moored his boat.
(19) Daddy, what exact!.v do you call a bank: the place where we moor the
boat or the place where I bring my savings? - Well, son, the place
where we moor the boat is a bank, but so is the place where you bring
your savings.
(20) Daddy, what precisely is a race: a group ofpeople with same ~ v p e of
body, or a competition in which you try to be the fastest.? - Well, dar-
ling, competition in speed is a race, and so is a group ofpeople with
the same type ofbody.
Taking Deane's observations one step further, let us note that there are
other, comparison-like contexts that have similar effects. First, there are
contexts of explicit contrast. For instance, in my earlier description of the
difficulty of combining an intentional with a non-intentional reading, I used a
sentence that did precisely that: I talked about situations in which John hit
the wall intentionally but Fred only did so by accident. The case can be gen-
eralized, as illustrated by (17) and (18). A second type of context that exhib-
its the same strong neutralizing effect involves definition and explanation.
Consider the exchanges in (19) and (20). Although it would be correct to
remark that this is at least to some extent a case of metalinguistic use, this is
hardly consequential for our interpretation of the possibility of identity-of-
sense anaphora: the metalinguistic explanation or definitional comment is not
realized as a metalinguistic statement about the words bank and race, but
rather takes the foml of a categorial statement about category membership.ll
At this point, two important remarks suggest themselves. First, we now
have a basis for an evaluation of the position taken by Bosch (1979). His
views are important, because they are the only ones (to my knowledge, at
least) that explicitly take into account the contextual focusing effects that we
have been considering here. Bosch' s suggestion boils down to replacing the
dichotomy between vagueness and polysemy by a threefold distinction, based
on the presence or absence of contexts that bridge the gap between two in-
terpretations of a predicate. Concentrating on do so ellipsis, Bosch proposes
to talk of polysemy when there is no focusing context that could bring to-
gether the two distinct interpretations, of vagueness when there is no context
in which they can be kept apart, and of contextually determined meanings
when there is at least one context in which the contextual focus allows for a
combination of both readings. But as the foregoing suggests that there is
124 Polysemy
always such a context (viz. when we are focusing on comparison, contrast,
or explanation), the tripartite distinction is undermined. And even if the neu-
tralizing effects of these contexts were not universal, they seem to be so
strong that stable polysemy in the sense outlined by Bosch would turn out to
be a relatively marginal case in the lexicon as a whole.
Second, the kind of examples that Nunberg and Deane concentrate on
raises the question whether the linguistic test (in its identity-of-sense anaph-
ora form) is a test of polysemy at all. Is it correct to talk about such ana-
phoric constructions as involving 'identity of sense', when they bring to-
gether such blatantly different things as newspaper directors and paper
copies of the newspaper, or even a river bank and a financial bank? Using
them as evidence in a semantic analysis means treating them as categorial
statements - but are they? Although Deane (1988) continues to treat these
constructions as examples of polysemy, his own explanatory analysis sug-
gests an alternative. On the basis of relevance-theoretic concepts and an
explicit theory of attention, he arrives at a theory that predicts which ana-
phors will be acceptable. What appears to be at stake, however, is less sense
identity than what might be called 'referential accessibility'. Crucial for the
acceptability of constructions like The newspaper decided to reduce its size
is neither the semantic (categorial) identity of the newspaper and it, nor their
coreferentiality, but the question whether the referent of it can be accessed
on the basis of the interpretation of the newspaper (and given certain restric-
tions on the flow of attention). This reinterpretation of 'identity of sense'
anaphora as 'accessibility of reference' anaphora is an example of the re-
analysis of the semantic tests that was suggested in the introductory passage
of this paragraph. As already indicated, we will not pursue this line of think-
ing systematically, but the example given here may help to clarify what is
involved in such a reinterpretation process.
6.2. The logical criterion
There are two types of internal inconsistencies within the logical approach to
polysemy. On the one hand, readings that are indistinguishable in most con-
texts may be separated in specific circumstances. On the other hand, read-
ings that are mostly kept apart may be combined into a single sense in spe-
cific contexts.
We have already seen an example of the first type. Whereas the concept
'bird' is usually used univocally when the biological category is meant (in
the sense in which a sparrow is a bird and not a bird is unintelligible), it is
Vagueness's puzzles, vagaries 125
possible to come across exchanges such as (21), in which non-prototypical
exemplars are implicitly contrasted with typical exemplars. Furthermore,
hedges and definitional specifications provide contexts in which 'p and not
p' constructions can be fitted easily: see the examples (22), (23) and (24).
(21) Is a penguin a bird.? - Well, it's a bird and not a bird.
(22) Is a whale ajish? - Irell, speaking it isn't, but loosely
speaking you might say it is.
(23) Is a kiwi a bird? - Well, speaking it is, but it certainly isn't a
bird par excellence.
(24) 1 bought this really expensive jade necklace last year, and it is cer-
tainly worth its price for all its and srylish appearance. Still,
chemically speaking, is this jade? - Well, it is jade i{you mean
nephrite, but not ifyou mean jadeite.
The final example may require some comment. The traditional precious
stone category jade may refer to two chemically different substances: neph-
rite (silicate of calcium and magnesium), and jadeite (silicate of sodium and
aluminium). In outward appearance (hard and green) and function (used as
ornament or for implements), both substances are similar; as such, ordinary
parlance maintains the traditional unitary, non-polysemous character of the
category. However, in specific circumstances such as the one evoked in the
example, the chemical differences between both substances may become
important the conceptual distinction that is then profiled obviously influ-
ences the possibility of the 'p and not p' construction.
For the second type of internal instability of the logical criterion, it is
natural to turn to cases in which two conceptually distinct readings occur
together in one context. First, let us have a closer look at the second major
variant of the test that we introduced in section 4.2. It can be shown, in fact,
that it runs into difficulties precisely in cases where the two readings whose
relationship we are trying to establish, occur together. Following a sugges-
tion made by Kempson and Cormack (1981), we noted in 4.2 that the nega-
tion of a vague predicate p leads to conjunctive implications of the 'not PI
and not p/ kind, whereas the negation of a polysemous predicate entails a
disjunctive implication of the 'not PI or not P2' kind. It is easy to see that the
latter disjunction is not exclusive; the negative sentence They did not reach
the bank in time would still be true if they neither reached the financial bank
nor the river bank. To take this fact into account, the implication frame for
polysemous predicates could be rendered as 'not PI or not P2 or not (PI and
P2)'. The contextual transformation of polysemy into vagueness then takes
place in those situations in which the single alternatives that feature in the
126 Polysemy
disjunction ('not PI but possibly P2' on the one hand, and 'not P2 but possibly
PI' on the other) are discarded in favor of the third alternative ('neither PI
nor P2')' This will be the case, for instance, when They did not reach the
bank in time is uttered in a situation in which the financial bank to be
reached is situated on the bank of a river. Because in this particular context,
the implication would be of the 'not Pl and not P2' type, bank would have to
be considered vague rather than polysemous in the context, given at least
that an implicational frame of the 'not PI and not P2' type is taken to be an
indication of vagueness.
Further, let us consider the other variant of the logical test that we intro-
duced in 4.2. Remember that we assume that there is polysemy if the 'p and
p' construction is not considered awkward or redundant. However, there are
indications that this criterion needs further refinement. Let us have a closer
look at the Dutch word vers, which we already briefly encountered in section
4.1 (for a full analysis of vel'S, see Geeraerts 1989a). Consider the following
(quoted) examples.
(25) Wie eet er de beste kost, wie koopt er verse vis en wild?
'Who can eat the finest food, who can buy fresh fish and garneT
(26) Nu meende zij, dat zij aan alles zou kunnen wennen, behalve aan gemis
van beweging en verse lucht.
'Now she thought that she would be able to get used to everything, ex-
cept lack of exercise and fresh air. '
(27) Voor een goede verpakking dienen de verse sigaren tot zolang in voch-
tige, weke toestand te blijven.
'In order to be packed properly, the newly produced cigars have to re-
main soft and moist until packing. '
Example (25) illustrates the most frequent reading of vers, the one that is
invariably the first to be found in dictionaries and to be remembered when
informants are asked what vers means. It is used with regard to foodstuffs,
and indicates that the latter are recently produced (caught, bought, harvested
etc.), and are therefore in optimal condition for human consumption; one
typically speaks of fruit, meat, bread etc. as being vel's. Quotations (26) and
(27) show that the two notions that are combined in (25) can occur sepa-
rately and independently. In (27), the notion of recency occurs without the
notion of optimality, and in (26), the reverse is the case. Intuitively speaking,
the central applications of vers constitute a distinct meaning: there is no
spontaneous recognition of an ambiguity when one mentions fresh food, to
the extent that it would be absurd to ask' Vel'S in what sense 7' when there is
talk of vers voedsel. Is it possible, then, to apply the logical test in such a
way that the non-polysemy of the central applications is recognized?
In all the circumstances in which vers could be used in the central mean-
ing, the maximal readings whose overlap actually constitutes the central
application apply. So either the 'p and p' construction can be used without
awkwardness, and then the singularity of the central applications is not rec-
ognized by the logical test, or the 'p and p' frame cannot be used, and then
we have a case of contextual neutralization of the distinction between the
'recency' application and the 'optimality' application (a distinction that can
be firmly established when the logical test is applied in other contexts). Now,
it would seem that the first alternative applies. By uttering, for instance, Dat
verse brood dat ik net van de bakker heb meegebracht smaakt lekker vers
('the fresh bread that I just brought home from the baker's tastes deliciously
fresh') one can enforce a construction that combines the meaning 'newly
produced' in the first occurrence ofvers, with the meaning 'optimal for con-
sumption' in the second occurrence. However, in such a case, we could sal-
vage the logical test by imposing a restriction on it: regardless of the possi-
bility of uttering 'p and p' without awkwardness or redundancy, the
simultaneous applicability in a particular context of two interpretations of p
does not constitute a case of polysemy if the two interpretations are more
closely connected than merely by juxtaposition. What we would like to rec-
ognize is the 'fusion' of the two interpretations into a single one, and such a
fusion is signaled precisely by the fact that there is more to the combination
of those interpretations than merely their simultaneous presence. In the case
of vers, for instance, the two separate readings are fused in the central appli-
cations, because there is a causal connection between them: fresh fish is not
just recently caught and (by coincidence) optimal for consumption, it is op-
timal in taste and nutritional value precisely because it is recent. If this is
accepted, the logical approach appears to be instable: in its 'p and not p'
form, it establishes the distinction between the readings 'optimal' and 're-
cent' of vers, but when we look at the simultaneous applicability of these
interpretations, the restriction just formulated leads to the conclusion that
there are contexts in which the distinction between both interpretations is
overruled.
It may be noted that we have now arrived a point where the logical test
leads to difficulties no matter how we look at it. Suppose, on the one hand,
that we simply stick to the 'p and not p' form of the test (ignoring the 'p and
p' version), or that we accept the test in its 'p and p' format, but without the
revision just suggested. In these cases, the intuitive univocality of the central
128 Polysemy
cases ofvers is not recognized. On the other hand, a revision of the 'p and p'
test leads to the conclusion that the central application of vers is a separate
meaning, and not just an ambiguous combination of the other two. As such,
the distinction between the 'optimal' reading and the 'recent' meaning is
neutralized in the contexts constituted by the central applications of vers. So
either we repeat the observation from paragraph 5 (viz., that the results of
the logical test may contradict those of other polysemy criteria), or we have
to accept the contextual instability of the test. Now, while such a neutralized
reading might seem to be less than damaging if it is permanently available as
a third major meaning ofvers, it is unfortunately the case that the neutraliza-
tion can be entirely contextual, in the sense that it is incidental. As an exam-
ple of such a contextual neutralization that is like the vers case in all re-
spects except for being purely incidental, let us have a look at the following
passage from P.c. Hooft's play Granida. (Hooft is one of the major seven-
teenth-century Dutch authors.) The shepherd Daifilo talks to his master
about his love for the princess Granida.
(28) Mijn heer, den laetsten nacht passeerd'jck bijder straten
Onder de venster van mijn vrouw, - (... ):
Mijn Vrouwe stondt voor 't glas, en soo sij mij sach gaen,
Sondt sij haer voester at, om mij te roepen aen.
Mijn Vrouw vernederde'haer, en quam mij selve vraeghen,
Wat oorsaack inder nacht mij daer ontrent mochtjaeghen.
[My lord, last night I passed through the street
Under the window of my lady:
My Mistress was at the pane, and when she saw me going there,
Sent her servant down to call me.
My Mistress humbled herself, and came to ask me herself
What reason might have driven me there at night.]
The word we are concerned with is vernederen. Literally, it means 'to
bring down' (one may recognize the English nether in the Dutch neder), but
already in the seventeenth century, the figurative meaning 'to humiliate, to
humble' is the primary one. In the quotation, vernederen is used reflexively;
both the literal and the figurative reading are relevant. On the one hand,
Granida comes down from the upper storey to meet Daifilo, but on the other
hand, she humbles herself precisely by taking this step: as a lady, it is below
her status to go out to meet someone of humble standing like Daifilo. Be-
cause she abandons her figuratively superior position by leaving her literally
superior position, the two readings of (zich) vernederen are fused, and we
have a context in which basically separate senses are combined into a single
Vagueness's puzzles, p o ~ y s e m y 's vagaries 129
reading, in the same way in which the two basic meaning components of vers
are fused in the central meaning of that item. And as in the case ofvers, the
fusion involves a causal connection between the fused readings: Granida
humbles herself precisely because she comes down to meet Daifilo.
Although most studies on polysemy would probably ignore cases such as
this one as deliberate literary puns, the remarkable character of the simulta-
neous applicability of both readings of vernederen need not be based on the
semantics of the construction, but may simply be a consequence of the nov-
elty of this kind of use. Admitting that there is indeed something punning in
the passage quoted above (on this point of literary style, the authors of the
Dutch Golden Age are no different from their English counterpart), I do not
think that we are obliged to attribute the pun to the polysemy of the con-
struction. If a punning effect may simply be the result of unexpectedness, the
remarkable character of vernederen may be a straightforward result of the
unusual character of the fusion. In contradistinction with the vers case, the
fused combination is not well-entrenched in the mental lexicon - but that
need not mean that it is less a distinct meaning, at least not at this point in
the discussion. (In section 7.1, we will explore the consequences of restrict-
ing the epithet 'meaning' to readings that are pennanently stored in the men-
tal lexicon.)
To sununarize, let me repeat the steps in the argumentation. First, the
logical test is not only applicable according to a 'p and not p frame', but
should also take into account cases in which two readings of a predicate p
are simultaneously applicable. Second, in the latter case, polysemy does not
occur when the two interpretations of p fuse coherently into a single one, as
in the case of vers. Third, the incidental character of the fusion that we find
in the case of vernederen underlines the unstable character of the distinction
between vagueness and polysemy. More generally, vernederen is an example
of contexts that may join readings that are usually separated, in contrast
with the jade example and similar ones, \vhich distinguish what is usually
considered identical.
6.3. The definitional criterion
Let us now turn towards the contextual variability of the definitional test.
Though it might superficially seem that this polysemy criterion is entirely
objective and decontextualized, this is hardly the case: because the criterion
requires that all the definitions that we take into account in order to establish
the polysemy of an item are themselves unambiguous, the test may yield
130 Polysemy
different results according to whether we are able to accept a definition as
monosemic or not. The requirement in question again refers to a traditional
one in the literature on definitions: 'if the definiens is itself ambiguous, the
definition obviously fails to perform its function of explaining the definien-
dum' (Copi 1972: 138). If the definition of what is allegedly a single sense
of a lexical item is itself polysemous, the definition may either leave in the
dark which of the various polysemous alternatives actually applies to the
definiendum, or it may cover up the actual polysemy of the latter. The first
difficulty occurs, for instance, when dark is defined as 'not light': because
not light is ambiguous between 'dark' and 'heavy', the definition of dark
does not prevent the meaning 'heavy' to be attributed to that word. It is the
second difficulty, however, that is of interest to us here. Suppose you are
describing a language with a word darvy that may be used both with the
meaning 'dark' and with the meaning 'heavy'. A definition of darvy as 'not
light' might then be pragmatically useful in a dictionary, because it would
make economical use of the polysemy of the English expression not light to
describe the range of application of darvy. In the context of the definitional
test of polysemy, however, defining darvy as 'not light' would be mislead-
ing: instead of establishing the monosemy of darvy, the superficially unitary
character of the definition would merely hide the actual polysemy (or even
homonymy) of darvy.
A real example of this kind of hidden polysemy may be found in Wierz-
bicka (1990). It will be recalled that Fillmore (1982) showed that bachelor
(an old favorite of lexical semanticists) can only be defined as 'unmarried
male' if the definition is made relative to a normal model of the world in
which there is a social institution of marriage as a union of two people of
opposite sexes, in which people have to reach a certain age before they can
get married, and so on. Prototypical fuzziness in the application of bachelor
then arises when this background model is applied to cases where the back-
ground conditions fail. Is a member of a homosexual couple (or a male ado-
lescent of 13 or 14) a bachelor? In the sense that he is unmarried, perhaps,
but certainly not in the prototypical sense of the word. Wierzbicka then ar-
gues that it is not necessary to make a distinction between the actual defini-
tion of bachelor and the related set of background assumptions (the Ideal-
ized Cognitive Model in the sense of Lakoff 1987) to explain the fuzziness
of the word: a simple monosemous definition of bachelor as 'an unmarried
man thought of as someone who could marry' (1990: 349) suffices, because
the additional phrase 'thought of etc.' takes care of the fuzziness. Notice,
however, that this is a prime example of hidden polysemy, if only because it
Vagueness's puzzles, p o ~ y s e m y ' s vagaries 131
includes the verb can. If one utters Yes, John can swim, various things could
be meant that clearly need not all be true at the same time. It might mean
(among other things) that John is objectively speaking not in a situation that
impedes his swimming (he is not lying on a hospital bed with two broken
legs), that he knows how to swim (he has acquired the skill), that he has the
permanent right to swim (he is a member of the sports club that owns the
pool), or that he has the momentaneous permission to swim (the bath atten-
dant has made it clear that the pool is no longer overcrowded and that John
may now dive in). Now, if these are distinct meanings of can, the definition
of bachelor as 'an unmarried man thought of as someone who could marry'
copies the polysemy into bachelor, and hides it at the same time. Also, the
distinction between the various ways in which a person could marry is not
irrelevant for the internal structure of bachelor, in the sense that judgments
about bachelorship are related to it. On the one hand, consider Tarzan, who
cannot marry because the objective conditions for marrying (such as the
presence of a judge, parson, priest or whatever) are not met. On the other
hand, consider a young man who does not get his parents' permission to
marry his beloved. Although both cases involve a person who cannot marry
in one of the various senses of can, the epithet bachelor can be applied much
more easily to the latter case than to the former. If the example can be gen-
eralized, this means that the 'permission' sense of can is less important for
the structure of bachelor than the 'objective possibility' sense.
12
The crucial point to make is this: if the definitional test is applied un-
thinkingly, a decision on the polysemy of the lexical item bachelor will con-
textually depend on our ability to recognize the polysemy of can - and such
a recognition is initially an intuitive, individual matter: Wierzbicka, for one,
seems not to have noticed it. But couldn't the recognition of such ambigui-
ties be made less intuitive? If we are using a definitional criterion for deter-
mining the polysemy of a lexical item but have to determine independently
whether the definitions that we consider are not themselves polysemous, two
basic possibilities are available. First, as in our example, we may resort to
other types of polysemy tests: for instance, when we establish that Yes, John
can swim is polysemous because it could be both, in a given situation, true
and false, we are relying on the logical test. But this would mean that the
reliability of the definitional test of polysemy could be reduced to the reli-
ability of the other criteria. Now, not only do we have independent evidence
for the contextual instability of these other tests, but also, at this point in our
argumentation we are specifically concerned with evaluating the definitional
test in itself and by itself.
132 PoZvsemy
Second, then, an iterative application of the definitional test would not
really help either, since this would merely set us off on a series of endless
repetitions. Or at least, at some point an intuitive recognition of the univo-
cality of the words we are using in our definitions would have to intervene to
stop the endless regression. But precisely because we are relying on intuitive
judgments of ambiguity rather than on tests and operational criteria, the
possibility of contextual and individual oversight is not excluded. Moreover,
if decisions about polysemy ultimately depend on an intuitive recognition of
univocality, why could we not simply apply that intuition directly to the
lexical item whose polysemy we are trying to establish? The whole point
about considering operational tests of polysemy is to get away from the im-
mediacy of simple intuitive judgments about polysemy. So either we accept
that the ultimate validity of the definitional test is based on intuition (but
then we have not got much of a test left), or we invoke the other tests (but
then the singularity of the definitional test disappears, and all the problems
connected with the other tests loom as large as ever).13
7. Additional perspectives - The representational escape hatch
Because the extreme contextuality of meaning that is implied by the internal
and the external inconsistencies of the three polysemy tests implies a rather
drastic departure from traditional conceptions of semantic structure, it is
important to proceed carefully and investigate any remaining possibility of
salvaging a stable notion of lexical meaning. Taking for granted, then, that
the tests for polysemy that we have discussed so far are sensitive to contex-
tual effects, an entirely different approach suggests itself. If the 'meanings'
we have been investigating are contextual surface phenomena, what model of
semantic memory do we have to assume to explain the occurrence of those
meanings? If the model is to be flexible, it would have to contain rules to
derive flexible interpretations, together with elements that the rules operate
on. What surfaces in a particular utterance is then nothing else but either one
of the permanently stored elements, or an interpretation derived from apply-
ing the rules for meaning extension to one of those stored elements. In such a
model, the surface-structure interpretations could be distinguished from the
deep-structural meanings,14 and the polysemy problem would boil down to
the question how to delimit the proper domain of the storage component
from that of the rule component. For instance, if literal senses are stored,
figurative ones could be derived by rule and need not be stored. In this
Vagueness's puzzles, 's vagaries 133
framework, we might try to circumvent the problematic question 'What is a
distinct surface interpretation?' by merely asking 'What is a distinct mean-
ing on the deep level of storage?'. The answer to the question is, in fact, easy
in principle: an interpretation represents a distinct meaning if it cannot be
derived by rule.
Before trying to evaluate this approach, let us note, first, that it can be
seen as belonging in the class of representational criteria for meaning dis-
crimination: what ultimately decides on the semantic status of an alleged
meaning depends on our way of representing lexical-semantic phenomena.
Instead of consisting of simple definitions (as in the type of representational
criterion that we discussed before), the representation now takes a 'distrib-
uted' form, part of the representational burden resting on the storage compo-
nent of the mental lexicon, the other part being taken care of by the rule
component. Second, it may be noted that this representational approach is
not uncommon in Cognitive Linguistic debates. For instance, Sweetser
(1987) is an attempt to reduce the polysemy of lie as described in Coleman
and Kay (1981) to a set of contextually derived specifications of a single
Idealized Cognitive Model for lie. And Vandeloise (1990) confronts the
analysis of over that is included in Lakoff (1987) (and that is itself based on
Brugman 1981); against the multiplication of the polysemy of over that he
spots in Lakoff's description, Vandeloise tries to give a monosemic descrip-
tion of prepositions. What we will be concerned with here is not the detail of
these debates, but the general strategy that consists of, first, defining the
polysemy issue as a question involving penllanently stored meanings rather
than surface structure interpretations, and second, assuming that what can
be derived procedurally from stored information need not itself be stored. It
is important to stress that the following objections against this view do not
involve the distinction between stored and derived information as such, but
rather the idea that the strategy as just defined somehow solves the polysemy
Issue.
The two components of the strategy can be evaluated separately. With
regard to the first component, one reason for being careful with the represen-
tational approach is the fact that it seems to be based on a false analogy with
phonology. In the same way that allophones can be contextually derived
from underlying phonemes, contextually different interpretations of a lexical
item are thought to be derived from underlying, stored meanings. Further-
more, there seems to be a functional similarity between allophony and se-
mantic vagueness: the distinction between allophones (either in the form of
free variation or in the form of contextual determination) is not a functional
134 Polysemy
one, and similarly, the contextual distinction between a male and a female
interpretation of neighbor is not semantically functional, in the sense, for
instance, that no truth-conditional differences are considered to correlate
with it. Not surprisingly, then, the contextually determined readings are
sometimes called allosemes (among others, see Deane 1988). The analogy
breaks down, however, when the functional status of the allo-entities is
taken into closer consideration. Take the well-known case of conjunctions. In
Horn (1985), it is argued that sentences such as 5J'he 's not happy, she's sad
versus She's not happy, she's ecstatic do not feature a distinct meaning of
not, because the differences of interpretation can be derived from the con-
text. Prominent among Horn's reasons for merely calling not 'pragmatically
ambiguous' rather than polysemous is the fact that its actual interpretation
can be derived from pragmatic principles. Now, there is a clear truth-
conditional distinction between the two sentences that corresponds precisely
with the 'pragmatic ambiguity' of not: as the addition of the sentences' tails
indicates, different truth conditions hold for she's not happy according to
whether not is taken in its ordinary reading or in its metalinguistic reading.
We may conclude, in other words, that not all allosemes are functionally
neutral; as such, alloseme is a somewhat misleading term.
To be sure, this is not an ultimately compelling objection against the rep-
resentational conception of polysemy. It does not show that it is impossible
to proceed in the representational way, but rather that it may be less impor-
tant to do so than we might think on the basis of the phonological analogy. If
the allo-entities are not structurally functional, they should not receive most
of the emphasis in linguistic descriptions. But if, in the lexical-semantic do-
main, we conceive of contextual interpretations on the model of allophones,
we run the risk of underestimating their functional importance. A similar risk
is inherent in the representational reinterpretation of the tenninological pair
vagueness/polysemy: because vagueness is considered the non-functional
phenomenon in comparison with meaning, redefining meaning in terms of
storage may make us lose sight of the fact that some of the non-stored read-
ings are just as functionally important as the stored ones. In other words, the
question 'What is a distinct meaning?' is then no longer more linguistically
relevant than the question 'What is a distinct interpretation?'. Or, to put it in
yet a different way, if we think of meanings as the semantically functional
phenomena, it would be wrong to equate meanings and stored readings. The
storage question is not an unimportant one, but it is not necessarily a ques-
tion about polysemy, if polysemy is defined in functional terms. Even re-
gardless of the terminological matter whether meaning should be defined in
such a functional way or not, an investigation into the functions of utterance
interpretations cannot be avoided. IS So, if the difficulties we have encoun-
tered so far are interpreted as difficulties concerning the question 'What is a
different surface interpretation?' (rather than 'What is a different stored
meaning?'), they remain unresolved. 16
Now, with regard to the second component of the representational strat-
egy, let us note that the representational approach as defined above is an
example of the 'rule/list fallacy' (Langacker 1987: 29). If the impossibility
of deriving an interpretation by rule is the criterion for distinguishing
polysemy, the crucial assumption is that the set of stored meanings and the
set of derived interpretations do not overlap. For what would be the status of
a reading that is both stored and derivable? It would contradictorily be both
a meaning and not a meaning. So, if the rule/list distinction can be shown to
be untenable in this lexical-semantic context, our problems with the question
'What is a distinct meaning?' remain unresolved. Consider, in this respect,
the fact that lexical items may change their prototypical centre in the course
of their diachronic development. The process typically involves a shift from
the periphery to the centre of one of the applications that is initially an ex-
tension of the original central sense. (Detailed examples of such processes
may be found in, among others, Winters 1989, Melis 1990, Geeraerts
1990b). At the same time, this shift typically involves frequency effects:
even though dominant frequency of occurrence is neither a sufficient nor a
necessary criterion for the centrality of an item's reading, rising frequencies
tend to correlate with growing centrality (cp. Geeraerts 1990b). If a particu-
lar peripheral meaning gets to be used more often, and if, at the same time,
the originally central meaning gradually lapses into disuse, a conceptual
reorganization takes place resulting in a shift of that application towards the
centre of the category. But if the frequency with which an application occurs
influences its prototypical status, this implies that there is some mechanism
in our conceptual memory for keeping a count (however roughly) of the
application's frequency of occurrence - and this would be the case regard-
less of the application's rule-governed nature. If changes in the frequency
with which we use an application (or hear it being used) may result in a
process by which the application becomes more central in our semantic
memory, we are somehow aware of that reading as an individual element in
our mental lexicon: the account we apparently keep of the reading's fre-
quency of occurrence presupposes a representation of that reading - even if
it is entirely derivable from a given stored meaning, a particular context, and
a specific procedure for semantic extension. In short, there are reasons to
136 Polysemy
doubt that a representational conception of univocality, founded on a di-
chotomous conception of the rule/list distinction, will yield satisfactory re-
sults.
17
8. Consequences
The consequences of the argumentation in the foregoing pages can be traced
on two different levels. First, we may have to change our (pre-theoretical or
theoretical) conception about what meaning is. More specifically, we will
have to determine the impact of the instability of the distinction between
vagueness and polysemy on Kleiber's model of prototypicality effects. Sec-
ond, there may be an influence on our model of what semantics is, i.e. on our
conception of the methodology of lexical-semantic research. And here as
well, an interesting consequence for prototype theory may be envisaged.
8.1. Consequences on the pre-theoretical and the theoretical level
If it can indeed be accepted that the distinction between vagueness and
polysemy is not well-defined, the primary conclusion to be drawn relates to
the problem that initially engendered our discussion of polysemy. Because
the instability of the distinction between vagueness and polysemy precludes a
strict dichotomy between intercategorial and intracategorial semantic multi-
plicity, Kleiber's conception of prototype theory need not worry us too much
for the time being.
It will be appreciated, however, that this conclusion is considerably
hedged in. On the one hand, the fact that we have not found a stable criterion
for a distinction between vagueness and polysemy does not mean that such a
criterion is in principle impossible to find. If we have not found a stable
operational test yet, we have perhaps not looked hard enough. Of course, our
pre-theoretical notion of what a distinct meaning is, is not entirely clear, and
in that sense, any search for polysemy criteria will be hampered by the fact
that we do not know precisely what it is we want a criterion for. But this is
not an uncommon situation in linguistics, and so further research into (new
and old) polysemy criteria retains its importance. Needless to say, research
into polysemy criteria also implies an investigation into the question what the
criteria we originally considered to be polysemy criteria are actually criteria
of, if not criteria of distinctness of meaning. On the other hand, our rejection
of Kleiber's attempt to bring some order in prototype theory does not mean
that all is well in prototype country. Here too, further research into the rela-
tionship between the various sorts of prototypicality effects and their sources
remains necessary. The main point we can claim on the basis of the forego-
ing pages, is that Kleiber's attempt to bring order into the prototypical mat-
ter seems to meet with fundamental difficulties - not that prototypicality is
in no need of further investigation. This is all the more so because our own
findings suggest modifications in our conception of lexical structure. After
all, the idea that meanings (as distinct from mere referential specifications)
do not exist, is rather disconcertingly at odds with what we traditionally
believe. So let us try to determine what other consequences this reversal
(conditional though it may be) leads to.
One major thing to be changed, I would claim, is our metaphorical, pre-
theoretical image of what meanings are. Come to think of it, the very word-
ing of the previous sentence reveals something about the way we talk about
meaning: the plural meanings suggests that meanings are entities or entity-
like things, presumably stored somewhere in the big storage room that we
conceive the mental lexicon to be, to be carried by word forms to an inter-
preter who unpacks and understands them. This metaphorical image has, of
course, been identified and labeled before: it is both that of the conduit
metaphor of verbal communication, first described by Reddy (1979), and of
the lexemist conception of the mental lexicon, more recently analyzed by
Hoffinann (1989). The common characteristic of both metaphorical com-
plexes is that of reification: meanings are things, prepackaged chunks of
information that are contained in and carried about by word bags. But what
kind of image could replace this reificational picture? The tremendous flexi-
bility that we observe in lexical semantics suggests a procedural (or perhaps
'processual') rather than a reified conception of meaning; instead of mean-
ings as things, meaning as a process of sense creation would seem to become
our primary focus of attention. The image I would like to propose to make
this conception more graspable, is that of a floodlight: words are searchlights
that highlight, upon each application, a particular subfield of their domain of
application. If this domain of application is seen, in model-theoretical fash-
ion, as a set, each time the word is used a particular subset is selected. But
while our traditional view of the distinction between vagueness and
polysemy entails that the number of subsets that can be lit is fixed and re-
stricted, we now have evidence that the verbal searchlight has much more
freedom. The freedom is not absolute, surely, and there will be preferential
subsets for each word; even so, the distinction between what can and what
cannot be lit up at the same time is not stable. IS
138 Polysemy
Although we should obviously try to move beyond a mere metaphorical
description of our new conception of lexical- semantic structure, the actual
development of a model of categorial structure that takes into account the
shifting nature of meaning lies beyond the reach of this paper. It may never-
theless be noted that, within the broad range of prototype-oriented studies,
the 'vantage point theory' presented by Robert MacLaury (1991) is perhaps
the one that is most congenial to this enterprise (most explicitly so because it
provides for perspectival shifts of the prototypical centre of a category).
8.2. Consequences on the methodological level
Accepting that the distinction between vagueness and polysemy is subject to
contextual variation highlights the interpretative nature of semantics: mean-
ings are interpretations that arise out of a particular context. But if meaning
is radically context-dependent, what is the methodological status of meaning
- given that it is us, the interpreters, who choose or provide the contextual
perspectives that lead to a particular interpretation? Are meanings phenom-
ena that are objectively out there and that can be discovered, or are they
subjectively construed by the interpreter? Are meanings found or are they
made?
The philosophical problem tumbling head over heels into the discussion
here is that of the objectivity of lexical semantics - not just the objectivity of
the object of semantic research, but the objectivity of its method. Cognitive
Linguistics generally follows Lakoffs (1987) characterization of the enter-
prise as an anti-objectivist one, but it is seldom realized that this characteri-
zation can be taken in two different ways. Usually, the anti-objectivism is
taken to be a characterization of the object of the enquiry. But even if lin-
guistic meaning is conceived of as a subjective, experiential, psychological
phenomenon, does this also mean that the method of linguistic semantics is
non-objective? After all, even if the meanings that we investigate are subjec-
tive entities, they might still be there objectively: we could find the meanings
as they are, and we would simply have to extract them from the messages in
which they are transmitted to us. However, this picture smacks of the reified
conduit metaphor and lexemist model that we rejected earlier. What, then, if
we do not find objective meanings ready-made, but construe them subjec-
tively? First, this would be naturally in line with the rejection of the reifica-
tional lexemist/conduit model: while we primarily rejected the model as a
picture of how meanings are put into words, we can now extend this rejec-
tion to the way in which meanings are extracted from words. And second,
extending the anti-objectivist, anti-reiticational stance to the receiver's end
of the communicative process makes good sense of the difficulties we en-
counter when trying to fix the meaning of utterances like the ones we have
been discussing throughout the paper. What we started looking out for was a
unique meaning, i.e. the meaning of the utterance, the authentic and original
one that the author put in the message (or perhaps: highlighted) and that is
objectively there for us to be unpacked (or alternatively: observed). But,
disconcertingly, we have serious difficulties in pinning down the unique
meaning of an utterance, not least because the various perspectives that we
adopt yield different answers to our semantic question. However, if we
abandon the vestiges of objectivism in our methodological self-conception,
the presupposition that there is a unique meaning itself can be rejected.
Rather than a single unique meaning, there would only be the interpretations
that we impose on the material - and our interpretative activities need not
yield a unique result.
It is not the intention of this paper to settle this question, but just to fur-
ther illustrate its importance, two points are worth mentioning. First, the
interpretative nature of the meanings we attribute to utterances has a bearing
on the status of the prototypical structure that we attribute to lexical catego-
ries. In those cases where we do not rely on 'hard' techniques such as statis-
tical analysis to determine the core meaning(s) of an item, but rather use
traditional linguistic methods of data gathering and description, the question
often arises 'How exactly do you determine what the prototypical centre of
an item is?'. Now, if the prototypical core meaning of an item is no more
objectively traceable than the meaning of a word in an individual utterance,
could it not be the case that we choose that meaning as prototypical that
gives us the best starting-point for interpreting the various applications in
which we encounter that item? Such a meaning should not only be applicable
in a straightforward way to the majority of the applications in which we
encounter the word, but should also be suited to arrive at an interpretation of
novel applications by applying regular mechanisms of semantic extension to
the prototype. In that case, the prototype is that meaning that best enables us
to make sense (literally, as an interpretative process) of the various ways in
which words are used. Or, in other words, the prototype would have to be
thought of as an interpretative perspective that helps us to interpret the uses
of a word. (Our analysis could then make claims to objectivity to the extent
that it mimics the interpretative structure other language users impose on the
category.)
140 Polysemy
Second, lurking round the corner here are three major philosophical
strands in the development of modem thinking. First, there is the herme-
neutical tradition as it stretches from Schleiermacher over Dilthey to
Gadamer, Habemlas, and Ricoeur (for an English introduction to the tradi-
tion, see a.o. Bleicher 1980 or Ricoeur 1981: 43-62). A crucial aspect of the
hermeneutical tradition (specifically as embodied in the Diltheyan distinction
between Geisteswissenschaft and Naturwissenschaft) is the methodological
characterization of the human sciences (explicitly including linguistics) as
being based on processes of interpretation rather than on the objectivist
methodology of the natural sciences. In the methodological self-conception
of linguistics, this idea has hardly been more than an undercurrent in the
major part of the twentieth century; see, however, Itkonen (1978, 1983) and
Anttila (1985) for a defense of a hermeneutical conception of linguistics.
Second, there is pragmatism, specifically in the form that our cognitive ac-
tivities are guided by pragmatic interests. For if our interpretations depend
on differing perspectives, the choice of a particular perspective may be
guided by various interests. Third, the whole set of problems touched upon
here extends towards the philosophy of science: the falsificationalistl para-
digmatic theories of science (in the line of Kuhn 1970 and Lakatos 1970)
precisely emphasize the idea that all scientific thinking starts from perspec-
tivist assumptions. (See Geeraerts 1985a for a discussion of the relationship
between prototype theory and this line of thinking in the methodology of
science.)
However briefly evoked, these philosophical connections highlight the
fundamental issue raised by the polysemy problem. If the difficulties we
have with the determination of lexical meanings are indeed indicative of the
ultimately non-objectivist, perspective-bound, hermeneutically interpretative
nature of linguistic semantics, a methodological reexamination of the enter-
prise imposes itself. If our understanding is perspectival, what perspectives
do we actually use, and what perspectives should or could we use? And what
pragmatic interests could lexical semantics serve? It is beyond the scope of
this article to answer these questions, but their far-reaching character should
be sufficient to establish the importance of a systematic exploration into the
methodological ramifications of the polysemy issue.
9. An inconclusive overview
The preceding pages have taken us on a tour through the mysterious world
of lexical meaning. An examination of three basic classes of criteria for dis-
tinguishing between polysemy and vagueness has made clear, first, that those
three types of criteria are in mutual conflict (in the sense that they need not
lead to the same conclusion in the same circumstances), and second, that
each of them taken separately need not lead to a stable distinction between
polysemy and vagueness (in the sense that what is a distinct meaning accord-
ing to one of the tests in one context, may be reduced to a case of vagueness
according to the same test in another context). Rather than leading to the
assumption that only one of the three tests is the true criterion for distinct-
ness of lexical meaning, these observations give rise to the suspicion that the
distinction between vagueness and polysemy is indeed unstable. If so, lexical
meanings are not to be thought of as prepackaged chunks of information, but
as moving searchlights that may variously highlight subdomains of the range
of application of the lexical item in question. Furthermore, the recognition
that the distinction between vagueness and polysemy is not unproblematic is
not just important for lexical semantics in a very general way, but has spe-
cific consequences for prototype-theoretical approaches to the study of word
meaning. In particular, the suggestion (as put forward by Kleiber 1990), that
various types of prototypicality effects should be kept apart on the basis of
their intracategorial or intercategorial status, can now be seen to rely uncriti-
cally on the stability of the distinction between vagueness and polysemy.19
In general, in spite of its vital importance for any theory of word mean-
ing, lexical semantics does not seem to have arrived at a consistent opera-
tional definition of polysemy yet. Within the non-objectivist framework of
Cognitive Semantics, this observation leads to the suggestion that the meth-
odological uncertainty surrounding the concept of polysemy is a side-effect
of the interpretative nature of linguistic semantics. However, rather than
trying to argue for this thesis as such, the paper has merely tried to evoke the
methodological set of problems in the context of which it can arise as a hy-
pothesis. By describing the operationally problematic nature of polysemy, it
has tried to make clear that the methodological status of lexical semantics is
in urgent need of further analysis.
142 Polysemy
Notes
1. This formulation presupposes that category as used in the terms intercate-
gorial and intracategorial is defined as 'lexical meaning, sense of a lexical
item'. This is not terminologically unquestionable, because in actual practice,
category may sometimes be used to refer to lexical items (whether polyse-
mous or not) just as well as to the individual senses of such items; in the lat-
ter case, one might speak of 'semantic categories', in the fonner of 'lexical
categories'. If such lexical categories are polysemous, they constitute a cohe-
sive cluster of semantic categories. Although the il1ter-/intracategorial ter-
minology equates category with 'semantic category', the distinction between
both types of prototypicality that is at stake here, could also be expressed by
distinguishing prototypicality within 'semantic categories' from prototypical-
ity within 'lexical categories'.
2. This is not just the case for prototype theory's investigation into degrees of
membership. In the context of a logical approach to meaning, the vagueness
of predicates like large is discussed in connection with the Sorites paradox:
removing one grain from a large heap of sand does not imply that the heap
stops being large, but after an indefinite number of steps, the process can no
longer be iterated. How, in other words, should the meaning of large be de-
scribed to accommodate this fact and avoid paradoxes? See Ballmer and
Pinkal (1983) for a sample of this line of research.
3. It may be noted that there is yet a fourth type of vagueness to be mentioned,
viz. the interpretative indeterminacy that occurs when there is a contextually
unresolved polysemy in an utterance. The types of vagueness mentioned in
the text involve indeterminate relations between lexical items and their refer-
ents (the relation between language and the world, if you wish). But indeter-
minacy may also include the relationship between the observer (say, the lin-
guist) and the language, in those cases where it is difficult to determine
which reading of an item is realized in a particular context. If, in a particular
context, it is difficult to make out whether a particular predicate p exhibits
reading PI or pz, this could either be because the distinction between PI and
P2 is semantically neutralized in that context, or because the context does not
provide enough information to choose between both. Whereas the first case
represents the kind of vagueness that we will be concerned with in the rest of
the paper (the speaker or writer intended to be neutral with regard to the dis-
tinction between PI and P2), the second case represents unresolved ambiguity
(the addressor of the message intended to communicate either PI or P2, but,
for lack of sufficient clues, the addressee cannot decide which is the case).
An examination of the relationship between both types (which is interesting
primarily for the methodological issues raised in section 8.2) is beyond the
scope of the present text. It might be added at this point that the tenninologi-
cal difficulties that are mentioned with regard to the term vagueness also oc-
cur with regard to the term ambiguity: while it sometimes used as a synonym
of polysemy, it may also occur with different readings (for instance, as a spe-
cific term for the type of unresolved polysemy discussed in this footnote.)
4. Because of the comprehensiveness of Cruse's (1982) treatment, it may be
useful to give a brief indication of the relationship between the tests distin-
guished by him and the major types distinguished here. While he does not
devote attention to the definitional approach, Cruse's tests Band D corre-
spond with our logical and linguistic test respectively. His criterion A will be
incorporated here as a specific subtype of the logical test (viz.. involving ne-
gation; see section 4.2). As for Cruse's criterion C, I would argue that it as
well is a variant of the logical test to the extent that it involves the possibility
of spontaneously recognizing (and asserting), in the same situation, two read-
ings with different truth-conditional properties for a given sentence. Specifi-
cally, this same situation here takes the form of a sentence conjoined to the
one featuring the predicate whose polysemy we are trying to establish. Con-
sider the anomaly of The children wore red shirts, and the boys, blue ones in
comparison with The dogs wore red collars, and the bitches, blue ones. The
latter sentence is true-or-false in the reading in which dogs means 'male
Canis familiaris', but anomalous in the reading in which dog means 'Canis
familiaris' in general; the anomaly derives from the fact that a general read-
ing for dog entails that bitches wear red collars, which is in contradiction
with the second conjunct, specifying that bitches wear blue collars. (This is,
of course, merely an encyclopedic contradiction, not a logical one.) So, in the
same situation (the one in which bitches wear blue collars), the sentence The
dogs ~ j . ' o r e red collars can be spontaneously recognized as either true or false,
or anomalous. As explained in 4.2, this is one of the realizations of the logi-
cal test. In the case of The children wore red shirts, on the other hand, no two
interpretations with distinct truth-conditional properties can be spontane-
ously recognized in the situation in which the boys wear blue.
5. The classification of polysemy criteria presented in Cruse (1986: Chapter 3)
differs on two fundamental points from the one in the article of 1982. First,
the direct criteria are divided into three classes. One of these corresponds
with our 'linguistic' class, while a second one includes our class of 'logical'
and related approaches. The third criterion is a novel one, to the extent that it
invokes a distinction between semantic specifications that are derivable from
context, and those that are context-independent. We will come back to this
criterion in paragraph 7: in particular, see note 16 below. Second, and more
importantly, Cruse' s argumentation in the book of 1986 seems to be partly
convergent with the argumentation in the present article. to the extent that
his text contains some observations questioning the idea of a clear dichotomy
144 Polysemy
between polysemy and vagueness; more specifically, he describes his own
view as a 'compromise position' (1986: 81).
6. Note that the vers case has more drastic consequences than the old case. It
could be argued, in fact, that we should not consider relations such as an-
tonymy separately, but that the entire set of relations that an item contracts
with other items should be taken into account. Thus, although old may have
identical antonyms in the two readings involved, the relationship of synon-
ymy might be invoked to disambiguate between both readings, among other
things because the synonym former only applies in one of the two. Even if
this global approach would enable us to circumvent the class of problems ex-
emplified by old (which would, to be sure, have to established separately), it
would not help against the type of difficulties illustrated by vers: taking into
account the entire set of lexical relations that an item contracts might avoid
relational underspecification of meanings (i.e. the situation that a relational
approach is unable to spot semantic distinctions), but it could not help
against relational overspecifications (i.e. seeing more meanings than is feasi-
ble). On the contrary, it would probably strengthen the danger of overspecifi-
cation. In Cruse (1986: 57), thin is used as an example of the vers-type of
problem.
7. If a longitudinal analysis confirms that this confusion was indeed a regular
feature of the history of logic, part of the explanation might be that the inter-
est in the phenomenon of ambiguity - at least judging from Rosier (1988) -
has traditionally always gone more to 'amphiboly', i.e. the rhetorical use and
abuse of sentential ambiguity, than to the operational definition of lexical
polysemy.
8. A good candidate is the following sentence from Cruse (1986: 64): ? Dogs
become pregnant at J2 months, but mature later than bitches.
9. But could it not be argued that attributes such as those involved in (1) and (i)
are general precisely as typical characteristics? This is (if I read her cor-
rectly) the position defended by Wierzbicka (1990), where she argues for the
inclusion of 'mental characteristics' in definitions. The reasoning can, I
think, be rendered in the following way: although not all fmits grow on trees
(and although, therefore, growing on trees is not a general characteristic of
fmit) , it is a general characteristic of fruit that people usually think of it as
growing on trees. Now clearly, including prototypical characteristics into
definitions is not what we would be arguing against from a cognitive point of
view. Still, how exactly should we constme Wierzbicka's argument? On the
one hand, 'usually growing on trees' as such is not a general characteristic of
fruit: if 'usually growing on trees' is a general characteristic of fruit, any fmit
should usually grow on trees - but of course, one cannot say that a strawberry
usually grows on trees. In other words, if 'usually growing on trees' is con-
stmed as part of our knowledge about fruits, it clearly is not a general attrib-
Vagueness's puzzles, poZvsemy 's vagaries 145
ute. On the other hand. suppose we construe 'usually growing on trees' more
radically in the mental sense intended by Wierzbicka: it is then part of our
knowledge of fruit that (other) people tend to think of fruit as growing on
trees. But again, this attribute (which involves not features of x, but knowl-
edge of features of x) can only be general if, for any thing that belongs to the
category fruit, other people tend to think of it as growing on trees - and we
probably would not claim that other people tend to think of raspberries as
growing on trees. In short. attributes that take into account prototypicality
phenomena by incorporating operators like usually do not restore generality.
10. It might perhaps be remarked that Crose's argumentation is less compelling
than he suggests. If the dichotomy involves (roughly defined) a 'gustatory'
and a 'non-gustatory' sense of like, it could be argued that there is no sen-
tence in the set that actually combines both readings (as would be required to
establish the contextual neutralization of the polysemy). The only candidate
would be (12), but a crossed interpretation of this sentence is not the first one
that comes to mind; rather, it would seem that John non-gustatorily likes
both items as status symbols or life style indicators. However, in Cruse (1986:
72), examples of sense-spectra are given for which the specific difficulties
with like do not occur.
11. Moreover, examples may be found in which metalinguistic overtones are
entirely absent; see David Tuggy's paint example (Tuggy 1993).
12. Furthermore, observe that Wierzbicka' s definition misses the point that apart
from ability, willingness to marry also plays a role in judgments about bach-
elorship: an unmarried male person living together with a woman on a per-
manent basis is not (if at all) a good example of a bachelor, although he
could marry in all relevant respects.
13. This discussion does not imply that there could be no other grounds for estab-
lishing the contextual variability of the definitional approach. McCawley
(1981: 7), for instance, notes that one person may more easily come up with a
general definition than another one: until someone pointed out to him that
the two interpretations of queen ('female ruler' and 'wife of male ruler')
could be covered under the general definition 'enthroned female', he had
been unable to define the word non-enumeratively. In general, the failure to
come up with a single definition may be the result of insufficient ingenuity
on the part of the definer, and ingenuity is obviously a characteristic that can
change from one person to another. Yet another possibility for establishing
the contextual variability of the definitional approach is based on the poten-
tial co-existence of two distinct definitional analyses of the same item. Sup-
pose, for instance, that Aristotle's cases of megalopsychia are analyzed in
such a way that the two groups for which a general definition can be given
are constituted by Alcibiades, Achilles, Ajax, and Lysander on the one hand
(megalopsychia being 'military ambition in the service of one's country'),
146 Polysemy
and Socrates on the other (megalopsychia being 'the unwillingness to waver
from the quest for truth under the pressure of others'). It will be obvious that
what is a distinct meaning in the context of this analysis does not have a
separate semantic status in the context of Aristotle's definitional analysis.
14. The distinction is not new in lexical semantics; see, for instance, Schmidt's
(1963) distinction between lexikalische and aktuelle Bedeutung.
15. There is yet another reason for not forgetting the contextual interpretations.
If the stored meanings are by that very fact unobserved entities, and if, con-
versely, the things that we can actually observe (however indeterminatedly)
are surface structure interpretations, getting an idea of what the surface struc-
ture meanings are is epistemologically prior to determining the background
entities from which they are derived. If representations plus rules are postu-
lated to explain the occurrence of particular surface phenomena. these phe-
nomena have to be determined first before one can start postulating explana-
tory entities.
16. At this point, we may have a look at the third criterion for polysemy intro-
duced by Cruse (1986); see note 5 above. In essence, the criterion relies on a
distinction between interpretations that are totally conditioned by context and
ones that are context-independent. For instance, the reading 'female sover-
eign. queen' of monarch in The Ruritarian monarch is expecting her second
baby does not represent a case of polysemy, because the sex of the monarch
can be derived from the contextual information that she is having a baby.
The argumentation can be reconstructed as follows. 'On the level of surface
interpretations, monarch may be interpreted as either 'male sovereign' or
'female sovereign'. These are distinct surface interpretations, because there is
at least one context in which they are not compatible, i.e. in which the logical
criterion for semantic distinctness applies. (In fact, they are incompatible in
any context). These distinct surface interpretations do not represent distinct
meanings. because the question which interpretation applies always depends
on contextual information (such as the presence of predicates that can only
refer to women)'. Now, consider the question what are the constraints on the
kind of contextual information that is envisaged in this argumentation. There
seems to be no reason to restrict it to purely linguistic contextual information,
if only because any relevant piece of extralinguistic. encyclopedic or situ-
ational information could be turned into a piece of linguistic context by ex-
pressing it in words. But then, the same argumentation that is used to defend
the claim that monarch is vague rather than polysemous with regard to the
distinction between the readings 'male sovereign' and 'female sovereign' can
also be applied to cases of genuine polysemy: 'On the level of surface inter-
pretations, position may be interpreted as either 'occupied place' or 'mental
attitude'. These are distinct surface interpretations. because there is at least
one context in which they are not compatible, i.e. in which the logical crite-
rion for semantic distinctness applies. These distinct surface interpretations
do not represent distinct meanings, because the question which interpretation
applies always depends on contextual information (such as the referents in-
volved)'. Therefore, if the reconstruction of the argumentation and the gen-
eralization of 'context' are justified, there are reasons for doubting whether
the criterion is sufficient yet to distinguish vagueness from polysemy.
17. Another failing escape hatch that should be mentioned could be dubbed 'the
ploy from infinity'. In fact, could it not be argued that the distinction between
vagueness and polysemy is necessary in principle, lest lexical semantics be-
come impossible? The traditional line here is to state that the distinction be-
tween polysemy and vagueness is necessary to avoid endless polysemy. If
there is no stop to referential distinctions turning up as distinct meanings, a
word would have an infinity of meanings. If, for instance, neighbor is polyse-
mous between the readings 'male neighbor' and 'female neighbor' merely
because there is a referential difference between the members of both subsets
of the extension of neighbor, what is there to stop us from claiming that
'neighbor who owns a Volvo' and 'neighbor who owns a Chevrolet', or
'neighbor who is 1.73 m tall' and 'neighbor who is 1.74 m tall' are distinct
meanings? After all, the referential distinctions involved define subsets just
like the distinction between 'male' and 'female'. Furthermore, the fact that
such distinctions could be multiplied ad infinitum would yield an endless
polysemy. So, it might be concluded, even if we do not know yet how to de-
termine the distinction between referential differences that do and those that
do not turn up as separate senses, we are at least sure that such a distinction
has to exist. However, the argument from infinity assumes that any referen-
tial difference would eo ipso turn out to be the basis of a semantic distinction:
only in that case could there be infinity. What we have discovered, on the
other hand, is the instability of the distinction between polysemy and vague-
ness, not its complete absence. The distinction remains valid in the sense that
referential differences do not automatically surface as distinct meanings:
there is always a preliminary process of categorization involved that picks
out particular referential attributes as the basis of categorial similarities or
contrasts. The attributes that are being picked out may not be entirely stable
(in the sense that a distinction that is disregarded in one context is promoted
to semantic distinctiveness in another), but there is no automatic mechanism
that turns referential differences into semantic oppositions.
18. This is not the first time a 'searchlight' or 'highlighting' metaphor has been
used in lexical semantics. The searchlight image has been used by Weisger-
ber (1954) to specify the way in which various items illuminate portions of
lexical fields. The concept of highlighting is used by Crose (1986: 53) to in-
dicate the contextual modulation of an item's meaning, in the way in which
the car needs servicing and the car n e e d ~ washing focus on different parts of
148 Polysemy
the car. It will be clear in what way our own terminology differs from Crose's
and Weisgerber's. While the latter uses the term as an onomasiological con-
cept, ours is a semasiological one. Crose's way of using the term, on the
other hand, is referential rather than semantic: what is highlighted is less a
part of the total set of semantic possibilities of an item than an aspect of the
referent of that item in one of its senses. (This referential conception is not
entirely separated from the semantic conception, however. In particular,
Crose's approach seems to rely on a distinction between contextual modula-
tion and contextual selection of senses that may not be tenable in the light of
our present argumentation. The central tendency of the present paper leads to
the suggestion that what are variations of one sense at one point, may turn
out to be distinct senses at another. For instance, while Crose's examples re-
fer to a situation in which both aspects of cars ('carriage' and 'engine') are
simultaneously applicable but in which one of them is backgrounded, there
are contexts in which both aspects are incompatible: an abandoned and de-
molished car with the engine removed would seem to be a car in the sense of
the car needs washing but not in the sense of the car needs servicing. That is
to say, if Crose's conception of highlighting presupposes that the back-
grounded and the highlighted reading are compatible in the relevant context,
our notion of highlighting does not carry that presupposition; moreover, in-
terpretations that stand in a relationship of compatibility-plus-focusing in one
context. may turn out to be incompatible in another.)
19. Admittedly, by focusing on the problematic, unstable cases we may have
exaggerated the instability. It could still be the case that in actual practice,
the problematic cases constitute merely a peripheral phenomenon. Even so,
they constitute a constitute a problem in principle for semantics, to the extent
that they show that the definition of lexical meaning is not as clear as might
be hoped.
Chapter 6
Classical definability and the monosemic bias
Originally published in Rivista di Linguistica 1994.6: 189-207.
Lexical semantics cannot at the same time retain a monosemic bias and the ideal
of classical definability: there exist cases where sticking to the intuitively plausible
idea that a particular lexical item is univocal leads to the conclusion that it cannot
be classically defined. In the present chapter (which is based on section 3.2 of
Geeraerts, Grondelaers, and Bakema 1994), this claim is supported by means of a
corpus-based case-study involving denotational data based on non-elicited actual
usage. Within this methodological framework, a procedure for determining classi-
cal definability will be demonstrated that operationalizes the requirement that
definitions be both general and distinctive. While the previous chapter, then,
looked at problems of definability from a methodological point of view. with an
eye to the criteria that may be used for establishing either polysemy or univocality,
the present paper takes a more practical approach, illustrating definitional prob-
lems with detailed examples.
Crucially, the paper argues for a rigorous procedure in answering definability
questions. If we assume that monosemy can be established by providing a classical
definition (a definition in terms of necessary and sufficient features that is both
general and distinctive with regard to the item to be defined), there are two re-
quirements that discussions of definability should strictly adhere to. First, the
generality of classical definitions implies that no disjunctive features are included
in the analytical definitions, and second, the distinctiveness of classical definitions
implies that they should make the right predictions about the lexical relations
among the items involved. As the examples in the present paper show. complying
with these demands requires a careful. step by step procedure that stands in sharp
contrast with what semanticians usually do when they posit univocality.
The present version of the text is considerably expanded with regard to the origi-
nally published article. The discussion of legging, as an example of classical de-
finability. was missing from the version in the journal.
1. Polysemy and definability
Definability plays an important role in current discussions about lexical
categories. if only because it is intimately related to the problem of
polysemy. Lexical-semantic research is to a large extent guided by a mono-
150 Polysemy
semic bias: the idea that postulating a single meaning for a lexical item is, on
the whole, preferable over postulating multiple meanings. Authors who have
recently defended this position with varying degrees of explicitness include
Charles Ruhl (1989b), Anna Wierzbicka (1985, 1991), and Claude Van-
deloise (1990). As Chris Sinha has pointed out (p.c.), the monosemic bias is
related to Grice's 'Modified Occam's Razor': the principle that senses are
not to be multiplied beyond necessity (Grice 1989: 47). The monosemic bias
is not universal, however. In my own work on lexical semantics, for in-
stance, I have had a tendency to stress the presence of polysemy (for in-
stance, see Geeraerts 1989a, 1990a) and it has been argued (Landheer 1991)
that I have so to speak exhibited an undue polysemic bias.
Now, the monosemy/polysemy issue is complicated by various factors,
two of which stand out. First, there may be various criteria for deciding
about the presence of monosemy. In Geeraerts (1993a), an overview of some
of the most common criteria is given (and it is shown that they are mutually
inconsistent). Second, there is variation as to the level of semantic analysis
with regard to which the monosemy question is defined. Most linguists, in
fact, would seem to be willing to accept a two-stage model in which a dis-
tinction is maintained between, roughly, permanently stored 'deep stmcture
meanings', and 'surface stmcture meanings' that can be derived from them
on the basis of cognitive and pragmatic principles of semantic extension
interacting with situationally and textually contextual factors. If the
monosemy issue is defined with regard to the stored readings, different solu-
tions will be arrived at in comparison with a situation in which it involves
surface stmcture interpretations. In the former case, for instance, one could
say that something is not a different reading when it can be derived by mle
from a stored reading (cp. Kempson 1977). In the latter case, on the other
hand, one of the possible criteria for semantic difference rests on the possi-
bility of capturing all surface readings of an item under a single definition.
Without trying to solve the problem of monosemy as a whole, the present
paper will try to shed some light on the latter test of polysemy, viz. the defi-
nitional test ofpolysemy that says that two of more instances of a word do
not constitute a different reading when they can be defined together as a
single meaning. In particular, it will be demonstrated, first, how the criterion
can be applied to usage-based data about the referential range of application
of lexical items, and second, that applying the criterion involves more com-
parative computation than is usually assumed. Third, this demonstration will
support the claim that lexical semantics cannot at the same time preserve
the monosemic bias and the ideal of classical definability: cases will be
Classical definability 151
presented where sticking to the intuitively plausible idea that a particular
lexical item is univocal leads to the conclusion that it cannot be classically
defined.
This claim will be illustrated on the basis of a corpus consisting of 9000
Dutch clothing terms, taken from fashion magazines and general magazines
published in 1991. A full description of the project is presented in Geeraerts,
Grondelaers, and Bakema (1994); the present paper corresponds largely with
section 3.2. of the monograph. In the database, each garment is present in
the form of a componential description. The descriptions are based on pho-
tographs (or, occasionally, drawings) that appear in the magazines; only
words referring to such pictures have been included in the database. The
componential system of description differs somewhat according to the major
subtypes of clothing. All types of trousers, for instance, are described by
means of the following set of features.
First dimension: LENGTH
1 The garment does not reach further down than the groins
2 The garment reaches down to the thighs
3 The garment reaches down to the knees
4 The garment reaches down to the calves
5 The garment reaches down to the ankles
Second dimension: WIDTH AND CUT
1 The garment is tight-fitting
2 The garment has a straight cut, neither tight-fitting nor wide
3 The garment has a loose, wide cut
4 The garment is tight-fitting round the hips but has widening legs
5 The garment is loose-fitting round the hips but has straight or tight-
fitting legs
6 The garment has a regular straight cut as far down as the knees, but has
widening legs below the knees
7 The garment has a loose cut from the hip to the knee, but has tight-fitting
legs below the knees
Third dimension: END OF LEGS
1 The ends of the legs exhibit no special features
2 The ends of the legs fit tightly by means of an elastic band
3 The ends of the legs fit tightly by means of tied laces
4 The ends of the legs fit tightly by means of buttons
5 The ends of the legs have an elastic band that fits under the feet
Fourth dimension: 1\1ATERIAL
1 The trousers are made of cotton, linen or a similar smooth cloth
152 P o ~ y s e m y
2 The trousers are made of wool or a similar coarse or fluffY material
3 The trousers are made of denim
4 The trousers are made of corduroy
5 The trousers are made of silk or a similar smooth and shining material
6 The trousers are made of smooth, shiny leather or an imitation of it
7 The trousers are made of rough, mat leather or an imitation of it
8 The trousers are made of stretch
Fifth dimension: DETAILS
I Seams and/or pockets are strengthened by metal buttons
2 The waist part of the garment has folds, pleats or creases
3 The garment has a very low crotch
4 The garment is made up of several visible strips of cloth
5 The legs have sharply pressed creases
6 The garment has an elastic band in the waist
A specific garment may then be described, for instance, by the featural con-
figuration H51431, where H indicates that the garment belongs to the group
of trousers, and where 51431 indicates that it is a pair of trousers with value
5 on the first dimension, value 1 on the second, 4 on the third, and so on.
It should be clear from this presentation that the approach followed here
is a referential, denotational one: lexical meanings are studied not primarily
through introspection, but on the basis of their ranges of application as they
appear in actual usage. The referential, usage-based investigation followed
here is not accepted unconditionally by all lexical researchers. In actual
practice, lexical studies with a Cognitive Semantic orientation exist both in
the form of introspective analyses, and in the form of corpus-based re-
search. Schmid (1993: 272) even considers the corpus-based approach in
work such as that of Rudzka-Ostyn (1988, 1989), Schulze (1988, 1991),
and Dirven (1985, 1990) to be typical of the European branch of the
Cognitive Linguistic movement, in contrast with the more introspectively
conducted studies of American researchers of a Cognitive Linguistic per-
suasion. From a more theoretical point of view, explicit attention for the
way words are actually used would seem to follow straightforwardly from
Langacker's characterization of Cognitive Linguistics as a usage-based
model that rejects the Chomskyan neglect of linguistic performance
(1987: 46). However, Wierzbicka (1985) has coupled a prototype-
oriented form of lexical research with an explicit defense of the introspec-
tive method. What advantages, then, are there to a referential, usage-
based approach that avoids relying exclusively on introspection? Next to
the advantages of a usage-based investigation that are enumerated by Ruhl
(1989: 13-16), let us note that the introspective method may succeed
rather well in pinning down the prototypical core of the items under in-
vestigation, but is hardly able to capture the peripheral uses to which the
core meanings appear to give rise in actual usage. This is a point to
which we will return in section 5.
2. Criteria for classical definability
The definitional criterion only works properly when restrictions on defini-
tions exist: there has to be a way of assessing what is a good definition be-
fore definability can be invoked as a measure of polysemy. Classical defin-
ability implies that a definition can be found that characterizes all the
members of the category to be defined, and only those. The definition has to
be general, in the sense that it applies to all the members of the category,
and it has to be distinctive, in the sense that it adequately distinguishes the
category from all others. For instance, let us assume that we are trying to
define the category 'bird' (as a biological species). We will then have to list
the attributes that all birds have in common, if there are any; further, we will
have to make out whether this list of attributes (or any subset of it) suffices
to distinguish birds from mammals, reptiles, and fishes, to say the least. As
illustrated in Geeraerts (l987a), the attributes that one would be inclined to
mention as general characteristics of birds, often do not have the required
commonality. On the other hand, the attributes that do seem to be general
among birds do not suffice to distinguish birds from other species; even
when the features in question are taken together, the duck-billed platypus is
a counter-example to the alleged definition.
It may be useful to point out here that there are various other ways of
terminologically indicating the classical nature of definitions. One is to say
that classical definitions define all and only the members of the category,
while another is to say that they uniquely define the category. More impor-
tantly, however, it has to be noted that applying the definition meets with
particular problems in the case of our material. Before turning to actual
examples, let us consider each of the two requirements in more methodologi-
cal detail, with specific reference to the kind of material incorporated in the
database described above.
The first part of the joint requirement of generality and distinctiveness
would seem to be easy to check: the componential description of the refer-
ents of each lexical item allows us to check whether there are any attributes
154 PoZVsemy
that these members have in common. There is an important reason, however,
for rejecting such a straightforward and mechanical procedure. The descrip-
tive features that define the various configurations in the referential range of
a lexical item cannot be taken at face value, but have to be interpreted (as
an automatic consequence of which, the issue of classical definability cannot
be settled mechanically). There are basically two forms of interpretation to
be taken into account. For ease of reference, they will be called the quantita-
tive and the qualitative one. The quantitative interpretation involves numeri-
cal dimensions, i.e. dimensions whose values constitute a graded continuum.
The crucial point here is to see that it is not the individual value of a specific
referent with regard to that dimension that is definitionally important, but
rather the range of values with which the dimension occurs. If, for instance,
a dimension like WIDTH receives the values [2], [3], and [4] in the semasi-
ological range of application of an item, we should not say that the item has
no common feature on the dimension WIDTH, but we should rather say that
the width of the referents of the item in question ranges from value [2] to [4].
Although the presence of the values [2], [3], and [4] would superficially
suggest that the referents of the item do not have common characteristics as
far as their width is concerned, they do upon closer inspection: all of them
fall within the range defined by the interval [2]-[4]. On the other hand, a
qualitative reinterpretation of the superficially given values involves hidden
variables. In particular, whereas all the dimensions in the database are visual
ones, there may be covert dimensions of a fimctional nature. For instance, if
the MATERlAL dimension of an item features the values [silk] and [cotton],
there is again, superficially speaking, no common characteristic. If, however,
both silk and cotton are used as light materials serving the purpose of keep-
ing the person cool in warm weather, the common functional feature [light
and cool] reduces the original variation on the MATERlAL dimension to
epiphenomenal status.
The distinctiveness criterion for classical definability should be handled
with equal care. To begin with, notice that the distinctiveness requirement
crucially involves negative evidence. If a definition is to hold for all and only
the members of a particular category, the definition should not apply to any
specific thing that does not belong to the category. The distinctiveness of the
definition is contradicted, in other words, if we can find a referent that falls
within the scope of the definition but that falls outside the scope of the cate-
gory. This does not mean, to be sure, that the items falling within the scope
of the definition could never occur as members of other categories than the
one to be defined. For instance, let us define the attributes plusquint and
deciminus of natural numbers. A natural number is plusquint if it is larger
than five; it is deciminus if it is smaller than ten. Both definitions are classi-
cal: they are as mathematically precise as you can get. At the same time,
both categories naturally overlap: the natural numbers 6, 7, 8, and 9 fall
within both categories. This means that the number 7 may sometimes be
called a plusquint and sometimes a deciminus. Suppose further that we have
actually encountered both ways of speaking in our corpus of mathematical
language, and that we are trying to define the word deciminus. We have
noted that all deciminuses share the property of being smaller than ten, and
therefore propose to define deciminus accordingly; we also notice, however,
that 7 is sometimes called a plusquint. When confronted with such a
plusquint instance of 7, could we then say (repeating the sentence introduced
above) that 7 falls within the scope of the definition of deciminus, but that it
falls outside the scope of the category (because it is not then called a
plusquint)? Of course not: falling outside the scope of the category means
never occurring within it. The number 7 is not a counter-example to the
proposed definition of deciminus because it is occasionally called a
plusquint, but it would be a counter-example if it were never called a deci-
minus. In the same way, the duck-billed platypus is only a counter-example
to the classical definability of bird because it is never categorized as a bird.
At the same time, the fact that not all deciminuses can be called plusquint
implies that the definition of the latter category should not be so broad as to
include the entire range of application of deciminus; it must include those
deciminuses that are sometimes called plusquint, but it must exclude those
that are never so called.
As a practical consequence of this observation, we will have to check any
alleged classical definition of a lexical item against the words with which the
item referentially overlaps. In particular, the definition should not be overly
general, in the sense that the entire overlapping category (rather than just the
intersecting part) is drawn into the category to be defined. The relationship
between two items that share referents may, however, take other forms than
the kind of overlapping (partial co-referentiality, semi-synonymy) that is
illustrated by the plusquintldeciminus case. Systematically, there are three
other relations to be envisaged. When the items are synonymous, no problem
arises when the definition of the definiendum covers the entire range of ap-
plication of the second item. Similarly, when the definiendum is a hypero-
nym of the second item, the definition may (in fact, must) cover all the refer-
ents of the second item. But when the definiendum is a hyponym of the other
word, a definition that exceeds the referential boundaries of the hyponymous
156 Polysemy
item will have to be rejected. To summarize, the distinctiveness criterion
does not apply to the synonyms and the hyponyms of the definiendum; in the
case of overlapping and hyperonymous categories, it should only be applied
to the overlapping and hyperonymous categories as a whole, not to those
subsets of the latter that they share with the definiendum.
3. Legging: a case of classical definability
Now that we have a better idea of how the classical definability of lexical
items can be established, actual examples can be considered. In the following
pages, the lexical item legging will be considered in detail as an illustration
of the methodology outlined above. The other lexical items that will be ana-
lyzed further on in this section will be treated more succinctly; rather than
concentrating on the analytical procedure itself, we will then concentrate on
the results of the analysis.
Table 1. The semasiological range of legging
H3118.v
H4118.v
H41186v
H4211.v
[3]
[36]
[3]
[ 1]
H5118.v
H51186v
H5128.v
H5154.v
[58]
[2]
[1]
[1]
H5211.v
H52115v
H52116v
[3]
[ 1]
[1]
The semasiological information that can be extracted from the database
takes the form of a list of referential descriptions of the form illustrated in
Table 1. The table lists all the referential configurations with which the item
legging occurs, together with their respective frequencies, which are men-
tioned between square brackets. The numbers refer to the five dimensions
illustrated in the previous section. The feature v specifies that the item is
worn by a woman. And a dot instead of a bumber indicates that the relevant
feature is not relevant; in this case, it means that the dimension Detail that
was introduced above is not applicable to the item in question.
Finding out whether legging can be classically defined would now seem
to follow a straightforward procedure: first, it will have to be established
whether there are any characteristics that are common to all referents of
legging, and second, it will have to be investigated whether the resulting
definition is sufficient to distinguish legging from all other categories that
are neither hyponyms nor synonyms of legging. On the basis of this proce-
dure, legging would definitively turn out not to be classically definable. A
glance at the table suffices to appreciate that the only truly general charac-
teristics of all the listed instances of legging are the fact that they all involve
trouser-like gannents, as represented by the feature [H], and the fact that
they are worn by women, as represented by the feature [v]. At the same time,
the database contains various trousers worn by women that are never called
legging; specifically, wider and shorter types of trousers for women fall
outside the category.
But obviously, we have not yet subjected legging to the quantitative and
qualitative reinterpretation process that we described above. Let us now try
to establish whether we can salvage the classical definability of legging by
using a more refined approach. A quantitative reinterpretation is important
for the dimensions WIDTH and LENGTH. The width of the referents of leg-
ging varies between the values [1] and [2], which is to say that leggings are
either tight or narrow. The dimension LENGTH has a range between values
[3] and [5]; the referents of legging reach down at least to the knee, but they
may also cover the entire leg down to the ankle. The impact of a qualitative
reinterpretation can be appreciated when we have a look at the dimension
MATERlAL. The fact that the predominant value on this dimension, viz. [8],
refers to stretchy fabrics suggests that there is a causal connection with the
dimension WIDTH: leggings are mostly tight or narrow precisely because they
are made of elastic material. Could it be the case, then, that the feature [elas-
tic] allows for a reduction of the variation on the dimension MATERlAL? A
renewed consultation of the original pictures on which the database records
were based, reveals that this is indeed the case. On the one hand, the records
that contain the value [1] on the dimension MATERlAL appear to be made of
a finely woven tricot that is at least moderately elastic. On the other hand,
the single record that features the value [4] for MATERIAL involves a
stretchy, very tight-fitting corduroy. In other words, although not all leggings
have the same degree of elasticity, they do share a certain amount of
stretchiness. The resulting picture of the common characteristics of all in-
stances of legging may be summarized as follows:
a legging is a two-legged outer garment for women covering the lower part
of the body from the waist down, ranging in length from the knee to the
ankle, and made from elastic materials such that the width of the legs
ranges from tight to narrow.
The next step involves checking whether this set of common features is suf-
ficiently distinctive to act as a definition of legging. In order to get an idea of
the lexical items that have to be included in an analysis of the distinctiveness
158 Polysemy
of the definition of legging that was given above, Table 2 lists the onomasi-
ological alternatives with which legging-configurations occur in the data-
base. For each of the various configurations that are situated within the se-
masiological range of legging, Table 2 specifies the other lexical items
referring to that configuration, together with their frequency.
Table 2. The onomasiological alternatives for legging
H3118.v
H4118.v
H41186v
H4211.v
H5118.v
H51186v
H5128.v
H5154.v
H5211.v
H52115v
H52116v
broekje J, kniebroekje J, legging 3, piratenbroek J, wiel-
rennersbroek J
broek 12, broekje I, calec;on J, denimbroek J, jeans J,
kuitbroek J, legging 36, leggings 9, piratenbroek 3,
stretchbroek J, tricotbroek J
broek 2, calec;on J, legging 3, leggings 2
broek J7, calec;on J, kuitbroek J, legging J, pantalon J,
streepbroek J
broek J2, broekje J, legging 58, leggings 23, skibroek 3,
stretchbroek J, stretchleggings J, tricotbroek 5
broek 6, calec;on 2, legging 2, leggings 7, pantalon J
broek J, joggingbroek J, legging J, skibroek 2
legging J
bloemenbroek J, broek 74, calec;on J, jeans JJ,
legging 3, pantalon 2
broek 24, legging J, pantalon Uricotbroek J
legging J, broek 3
The status of the onomasiological alternatives in terms of their lexical re-
lationship with regard to legging is specified in Table 3. In accordance with
the methodology set out above, we need not worry about a possible lack of
distinctiveness of the definition of legging with regard to hyponyms like
kabellegging ('cabled !egging', legging decorated with a cable pattern) and
stretchlegging ('!egging in stretchy fabric'), or synonymous expressions like
leggings and cale90n. (In passing, it may be noted that the latter term is
geographically restricted to the Belgian sources, in contrast with the maga-
zines published in The Netherlands.) The hyperonym broek does not present
a real danger either, because the definition of legging specifies an actual
subset of the range of application of broek (which may, in particular, refer
to two-legged gannents that are much shorter and much wider than the refer-
ents of legging).
Table 3. The relations between legging and its onomasiological alternatives
Synonyms
Hyperonym
Hyponyms
Semi-synonyms
leggings, caler;on
broek
kabelleggings, stretchleggings
kuitbroek, kniebroekje, broekje
jeans, denimbroek, stretchbroek, tricotbroek
streepbroek, bloemenbroek
piratenbroek, wielrennersbroek,
skibroek, joggingbroek, pantalon
As suggested in Table 3, various subsets may be distinguished within the
set of semi-synonyms. A first set comprises kuitbroek 'calf-long pair of
trousers', kniebroekje 'knee-long pair of trousers', and broekje 'short pair of
trousers, shorts', which each delimit a specific area within the range of ap-
plication of broek on the basis of the length of the gannent. The second set
consists of jeans, denimbroek, stretchbroek, and tricotbroek, which each
refer to trousers made of a particular fabric. In the third set we find items
that refer to a particular type of motif or decoration: streepbroek indicates
the presence of stripes, and bloemenbroek signals the presence of flowers.
The fourth subset contains piratenbroek 'pirate's trousers', wielrenners-
broek 'cyclist's trousers', skibroek 'ski pants', joggingbroek 'jogging
pants', and pantalon 'pair of trousers'. There is an interesting distinction
between the fourth subset and the previous three, in the sense that the items
in the latter may be adequately defined as a hyponym of broek by referring
to a single dimension (length, material, and motif respectively). These di-
mensions detennine the morphological structure of the words, in the sense
that the first member of the compound refers to the specific value on the
relevant dimension that is crucial for the item. For the items in the fourth
subset, however, various dimensions have to be mentioned at the same time
in order to specify their proper position.
In the cases involved in the first three sets, establishing the referential
overlap with the semasiological range of legging is a relatively straightfor-
ward matter. The items are actual semi-synonyms to the extent that they may
refer to pieces of clothing that are not legging-like on any dimension that is
irrelevant for the item in question. For instance, referents of kuitbroek that
160 Polysemy
are not tight-fitting enough to fall within the definition of legging are not so
called either. Kuitbroek is characterized on the basis of the dimension
LENGTH, and its specific value on this dimension happens to fall within the
range of lengths that is definitional for legging. Because of its 'one-
dimensional' nature, however, it is not definitionally specified with regard to
other dimensions that are subject to restrictions in the case of legging (such
as, in the example, WIDTH), and it may therefore refer to pieces of clothing
that are definitely too wide for leggings. We have to make methodological
allowances, though, for the fact that the number of records we have for the
various items does not always suffice to establish their overlapping status
beyond all doubt. That is to say, we may not have enough examples of an
item like bloemenbroek to establish whether it does indeed occur with refer-
ents that are not leggings. In these cases, we have nevertheless listed the item
as a semi-synonym on the basis of the assumption that the morphological
structure of the item is a good indication of its semantics - on the basis of
the assumption, for instance, that trousers with a flower motif may be called
bloemenbroek regardless of their other characteristics.
For the items of the fourth subset identified above, such an abductive un-
derpinning of the classification on the joint basis of intuition and morpho-
logical structure is less obvious. Although most of the referents of skibroek
are relatively tight-fitting, the widest ski pants are definitely wider than the
widest referent of legging; skibroek, in other words, has a wider range on
the dimension WIDTH than legging. The same criterion also distinguishes
joggingbroek from legging. For wielrennersbroek, on the other hand, the
most important distinctive dimension is LENGTH: whereas legging ranges
from dimensional value [3] to [5], the referents of wielrennersbroek are
never longer than the knee. Similarly, piratenbroek refers to garments that
may be as long as the knees or the calves, but never as long as the ankles.
4. Vest etc.: absence of classical definability
The discussion of legging shows that a careful analysis of semasiological
ranges and lexical relations is necessary to determine the classical defin-
ability of lexical items. The legging example produces positive results, in the
sense that the item in question appears to be classically definable. The next
step will be to discuss a number of cases where the definitional analysis
yields negative results, in the sense that the items in question cannot be de-
fined on a classical basis. The discussion will be based on a subset of the
field of clothing terminology. In particular, we will consider pieces of cloth-
ing that cover the upper part of the body, that can be entirely opened at the
front, and that are never worn as the first layer of clothing above the under-
wear. The items with the highest frequencies in this subset are jack, colbert,
blazer, jasje, and vest
NL
. The distinction between vest
NL
and vests is neces-
sary because there is a marked difference between the ways in which the
item vest is used in the Belgian and the Netherlandic sources; we will come
back to this point below.
Table 4 gives an overview of the ranges of application of the items. The
table is based on individual analyses of the items along the principles dem-
onstrated in connection with legging. That is to say, the dimensions used in
the table do not necessarily reproduce the information structure of the data-
base in a straightforward maJmer, but may be the result of a reinterpretation
of the stored data or even a reconsideration of the original pictures. Dimen-
sions that are not distinctive within the subset have been left out. For in-
stance, all the types of clothing included in the subset are worn by men and
women alike; accordingly, the dimension SEX has not been retained in the
table. A plus sign means that a particular dimensional value occurs within
the range of application of the item; a minus sign indicates that it never oc-
curs. Thus, plus signs on all values of a particular dimension mean that both
values may occur. For instance, the referents ofjasje may either occur with
a type of fastening that can be fastened up to the neck, or with a type of
fastening that stops on the chest somewhat lower than the neck; by contrast,
jack is never used as a name for garments that cannot be fastened entirely.
The only dimensions in the overview for which the relevant values might
have to be restated in terms of ranges are LENGTH and FASTENING, since
both involve measures of length; for instance, the referents of jasje have a
fastening whose length ranges from up to the chest to up to the neck. The
other dimensions consist of discontinuous values.
A first thing to note is that the referential ranges included in Table 4 sug-
gest the existence of certain hyponymy relations. It appears, for instance,
that all dimensional values that occur in the range of jack also occur in the
range of jasje; at the same time, the latter item exhibits a number of dimen-
sional values that are absent in the case of jack. In this particular case, the
suggestion that jack is a hyponym of jasje (because the referential range of
the latter word includes that of the former) is supported by the intuition that
jasje is a cover-term for the entire set of items included in 4. However, the
overview in the table is not really a good way of settling the hyponymy rela-
tions among the items, because the referential ranges are being considered in
162 Polysemy
tenus of separate dimensions rather than dimensions in combination. Con-
sider a fictitious case in which an item A is represented by the referential
types [ac] and [bd], and an item B by the types [ad] and [bc]. In both cases,
the first dimension ranges over the values [a] and [b], and the second dimen-
sion over the values [cl and [d]. Judging on the basis of an overview of di-
mensional ranges, then, A and B would be synonymous, since they have the
same dimensional ranges. Judging on the basis of the dimensional values as
they occur in combination, however, it becomes clear that there is neither a
relationship of synonymy nor hyponymy between both items. It is therefore
necessary to establish hyponymy relations on another basis than Table 4 as
such,
Table .;, The semasiological ranges ofjack, blazer, colbert, vest
Nv
jasje,
jack colbert blazer vestNL jasje
LENGTH
shorter than the waist
+ + +
as long as the waist + + + +
lower than the waist + + + + +
CUT
blousing
+ +
wide and straight
+ + +
narrow and straight
+ + + +
waisted
+ + + +
MATERIAL
woven fabrics
+ + + + +
knitted
+ + +
leather
+ +
FASTENING
up to the neck
+ + +
lower than neck
+ + + +
A corpus-based approach for the recognition of hyponymous relations
may be established as follows. If A is a hyponym of B, B may occur as an
alternative name for all referents of A. Of course, B need not be as frequent
as A for the referential set in question, because A may be more entrenched
than B. Also, it may be expected for statistical reasons that the less common
referential types of A may not occur in the corpus with B as an alternative
denomination; in actual practice, it may be sufficient to establish that B oc-
curs as an alternative for the most common referents of A.
Table 5. Vest
NL
andjasje as onomasiological alternatives for blazer.
blazer: frequency of vest jasje
configurations configuration as alternative as alternative
B1l22v 10 +
B1222v 10 +
B2111m 2 +
B2111v 1
+
B2112v 17
+
B2121m 5 +
B2121v 14
+
B2122v 48 +
B2131v 3 +
B2132v 8 +
B2212v 12 +
B2221m 3 +
B2222v 22 +
B2232v 5 +
C2212v 1 + +
C1312v 1 +
C2311v I + +
C2312v 7 + +
C2322v 1 + +
C2332v 1 + +
C2412v 1 +
C3212v 1 +
C3312v 14 + +
C3332v 1 + +
C3412v 4
+
C3432v 1 + +
In Tables 5 and 6, such overviews of onomasiological alternatives are
given for blazer and colbert. Given the statistical margin that was just men-
tioned, it can be deduced from the table that colbert is a hyponym of blazer
and jasje, and that blazer is a hyponym of ja5je. At the same time, of
course, it should be established that there are cases of B that are not named
by means of A (lest a situation of synonymy rather than hyponymy obtains).
This type of information, however, can be safely derived from overviews like
164 Polysemy
the one in Table 4: for instance, the plus sign on the 'knitted' value of the
dimension MATERIAL for blazer as opposed to the minus sign for colbert
indicates that knitted referents of blazer never occur with the name colbert,
for the simple reason that the referential range of colbert does not include
knitted garments of any kind.
Table 6. Blazer, vestl\JL andjasje as alternatives for colbert.
colbert: frequency of blazer vest jasje
configurations configuration as alternative as alternative as alternative
B1121m 1
B1122v 7
+ +
B2111m 2
+ +
B2121m 52
+ +
B2121v 4
+ +
B2122v 17
+ +
B2131m 5
+
B2211m 1
B2212m 1
B2221m 13
+ +
B2222m 1
+
B2222v 4
+ +
C3212m 1
C3311v 3
+
C3312v 1
+ + +
Repeating the procedure illustrated in 5 and 6 for the items jack and vest
leads to the hyponymy relations that are charted in Figure 1. (The label '+lh'
indicates that there is a relationship of hyponymy between the items in ques-
tion, given that a lexical test of hyponymy as illustrated in 5 and 6 is used.
The label '-lh' signals the absence of hyponymy according to the lexical
criterion.) The definitional question regarding the five items can now be
made more precise: can the items be classically defined on the basis of the
overview in 4 without obscuring the lexical relations summarized in 7? This
involves reviewing all possible classical definitions of the items and checking
whether they respect the relations in 7.
Note, however, that only two of the four dimensions included in Table 4
can be used to establish classical definability. Whereas the values on the
LENGTH and the FASTENlNG dimension can be expressed in terms of ranges,
any reference to the other two dimensions automatically involves disjunctive
values; therefore, including these dimensions in the definition inevitably
turns the definition into a non-classical one. Thus, it will be sufficient to
investigate only the potential definitions that refer to the dimensions LENGTH
and FASTENING.
A further restriction follows from the specific position of jasje. Both
from the point of view of the featural ranges given in 4 and from the point of
view of the lexical relations specified in 7, jasje appears to act as a cover-
term with regard to the other items: it is a hyperonym of the other four, and
the row of plus-signs accompanying jasje in 4 indicates that it indeed in-
cludes all definitional possibilities that are relevant for the other items. In
this sense, the search for classical, distinctive definitions may be confined to
the other four items. In all of the potential definitions mentioned below, then,
jasje 'jacket' will be used as a cover-term; roughly, it may be defined as
referring to garments that cover the upper part of the body, that can be en-
tirely opened at the front, and that are never worn as the first layer of cloth-
ing above the underwear.
hyponym jack co/bert blazer vestNL
hyperonym
colbert -Ih
blazer -Ih
vestNL -Ih
jasje +Ih
Figure 1. Hyponymy relations betweenjack, blazer, colbert, vestNL,jasje.
The definitions that are to be taken into account may be systematically
grouped together in three sets: definitions that refer only to the dimension
LENGTH, definitions that refer only to the dimension FASTENING, and defini-
tions that involve both dimensions at the same time. For ease of reference,
we will call jack] the definition that involves LENGTH, jack2 the definition
that involves FASTENING, etc. All in all, the following twelve definitions have
to be considered.
166 Polysemy
Jack]
Jasje whose length ranges from the region of the hip to the region of the
upper part of the legs
Jack2
Jasje that can always be fastened up to the neck
Jack]
Jasje whose length ranges from the region of the hip to the region of the
upper part of the legs, and that can always be fastened up to the neck
Colbert]
Jasje that reaches down to the region of the upper part of the legs
Colbert2
Jasje that can only be fastened as far as the chest (but not up to the neck)
Colbert]
Jasje that reaches down to the region of the upper part of the legs and that
can only be fastened as far as the chest (but not up to the neck)
Blazer]
Ja.sje whose length ranges from the waist to the region of the upper part of
the legs
Blazer2
Jasje that can only be fastened as far as the chest (but not up to the neck)
Blazer]
Jasje whose length ranges from the waist to the region of the upper part of
the legs and that can only be fastened as far as the chest (but not up to the
neck)
Vest]
the legs
Vest2
Jasje with a fastening whose length ranges from the up to the chest to up to
the neck
Vest]
the legs and that has a fastening whose length ranges from up to the chest
to up to the neck
These twelve definitions may occur in 81 (=3
4
) combinations. That is to
say, the general definability question boils down to 81 questions of the type:
ifjack is defined asjackj, calbert as calbert], blazer as blazer2, and vest as
vest], do the lexical relations that follow from these definitions then conform
to the actual relations that are summarized in figure I? Or, in other words, is
there at least one combination of the twelve classical definitions that makes
the right predictions about the attested lexical relations? Rather than consid-
ering all 81 possibilities separately, it can be shown in the following way
that the question has to be answered in the negative.
First, consider all combinations of two elements from among the set of
definitions that refer only to LENGTH. The co-occurrence of jack] and col-
bert] is to be excluded, because this would counterfactually imply that col-
bert is a hyponym of jack (as the range of LENGTH for jack as defined in-
cludes the range as defined for colbert). The co-occurrence of jack] and
blazer] is to be excluded because it would imply that jack is a hyponym of
blazer. The co-occurrence of jack] and vest] is to be excluded because it
would imply that jack is a hyponym of vest. The co-occurrence of colbert]
and vest] is to be excluded because it would imply that colbert is a hyponym
of vest. And the co-occurrence of blazer] and vest] is to be excluded be-
cause it would imply that blazer and vest are synonymous.
Second, consider all combinations of two elements from among the set of
definitions that refer only to FASTENING. The co-occurrence ofjack2 and
vest2 has to be excluded because it implies that jack is a hyponym of vest.
The co-occurrence of colbert2 and blazer2 has to be excluded because it
implies that colbert is a synonym of blazer. The co-occurrence of colbert2
and vest2 has to be excluded because it implies that colbert is a hyponym of
vest. And the co-occurrence of blazer2 and vest2 has to be excluded because
it implies that blazer is a hyponym of vest.
Third, consider all combinations of two elements from among the set of
definitions that refer to both LENGTH and FASTENING. The co-occurrence of
jack] and vest3 has to be excluded because it implies that jack is a hyponym
of vest. The co-occurrence of colbert] and vest] has to be excluded because
it implies that colbert is a hyponym of vest. And the co-occurrence of
blazer] and vest] has to be excluded because it implies that blazer is a hy-
ponym of vest.
The alternatives that remain at this point are summarized in Table 7. It is
now immediately obvious that there can be no combination of four classical
definitions that respects the existing restrictions, if only because all possible
definitions of vest
NL
have already been ruled out. We may conclude, then,
that there is no set of classical definitions for jack, colbert, blazer, and vest
that sufficiently distinguishes the items among each other and that respects
the lexical relations that appear to exist among them. Whether this is the
dominant situation in the lexicon is difficult to say on the basis of our mate-
rial; after all, we have only been able to examine a few lexical categories.
One general conclusion, at least, is that indeed not all lexical categories can
168 Polysemy
be classically defined. In addition, it is worthwhile to point out that the de-
finability issue seems to be strongly influenced by the specific subfield of the
field of clothing terminology that is being considered. The 'skirts' -subfields,
for instance, contains classically definable categories, whereas the subfield
consisting of shirts, blouses, t-shirts, and their likes is as unclassical as the
subfield analyzed in the previous pages.
Table 7. Combinations of classical definitions for jack, colbert, blazer, vest.
5. Methodological discussion
blazer
2
3
3
vest
What we have tried to illustrate in the foregoing pages (apart from the fun-
damental fact that non-classical definability is a real phenomenon) is the
importance of a rigorous procedure in answering definability questions.
Apart from the importance of distinguishing between polysemy and univo-
cality, there are two requirements that discussions of definability should
strictly adhere to. First, the generality of classical definitions implies that no
disjunctive features are included in the analytical definitions, and second, the
distinctiveness of classical definitions implies that they should make the right
predictions about the lexical relations among the items involved. As both the
legging example and the jasje-subfield show, complying with these demands
requires a careful, step by step procedure that stands in sharp contrast with
two other approaches to the relationship between polysemy and definitional-
ity. In fact, while the approach demonstrated here applies the classical crite-
rion of dejinability to denotational data based on non-elicited actual us-
age, the two most outspoken recent defenders of monosemy appear to drop
either the one or the other of these two features from their methodology
On the one hand, Ruhl (l989b) accepts the importance of usage-based
data, but gives up on the necessity of actually defining the presumed single
meaning of an item. If all the instances of use ofa lexical item are claimed to
exhibit the same meaning, the least one should do to make this statement
falsifiable is to define the semantic value in question. Ruhl, however, explic-
itly denies the possibility of doing this. At the end of his analysis of the Eng-
lish verb bear, for instance, he writes:
So what does bear mean? It should be clear by now that this question can-
not be answered in words: there is no single word or phrase that can com-
prehensively capture exactly what bear contributes. I hope, by trying to
show the unity of bear's contexts, to have revealed a unified meaning; but
such a conclusion is inferential (1989: 63).
Clearly, such an approach gives up on precision in favor of impressionism.
For general methodological reasons of comparability and falsifiability, such
a strategy has to be rejected in favor of the attempt to achieve descriptively
adequate definitions. If it is not even possible to definitionally identify the
allegedly unitary meaning of a lexeme, how can its unitary status be tested at
all? This is not, to be sure, a mere rhetorical question: see Geeraerts (1993a)
for an overview of other approaches to polysemy apart from the definitional
one. Ruhl, however, dismisses the definitional criterion without being ex-
plicit about an alternative that is methodologically sound in the sense of
being precise enough to allow for falsifiablity.
On the other hand, Wierzbicka' s approach retains the importance of pre-
cise definition, but rejects the denotational, usage-based approach. Such an
explicitly introspective strategy runs the risk of being imprecise with regard
to the actual range of application of an item. As an illustration, consider
Wierzbicka's definition of dress (1985: 382):
A KIND OF THING MADE BY PEOPLE FOR WOMEN AND GIRLS TO WEAR. IMAGIN-
ING THINGS OF THIS KIND PEOPLE COULD SAY THESE THINGS ABOUT THEM:
they are made to be worn on the body, below the head, to cover most of the
body
so that all the parts of a woman's body which people think should not nor-
mally be seen are covered with that one thing
and to protect most of the body with undesirable contact with the environ-
ment
and to cause the woman wearing it to look good
they are made in such a way that when they are on the body the lower half
surrounds the lower half of the woman's body from all sides
so that the legs are not separated from one another
and so that the genital area of the woman's body seems to be hidden
and so that women wearing things of this kind look different from men
170 Polysemy
things of this kind are thought of as something suitable for women to wear
in most kinds of places and in most kinds of circumstances.
When we have a look at the actual garments that occur in our database as
instances ofjurk (the Dutch equivalent of dress), we find cases in our mate-
rial that do not conform to the description. If, for instance, 'covering most of
the body' is interpreted as 'covering more than 50% of the body', then a
number of very short summer dresses with open backs and low necklines do
not display the feature in question. And if 'the parts of a woman's body
which people think should not normally be seen' include the upper part of
the thighs, then dresses with long side slits contradict the image. Further-
more, some dresses have such wide armholes and such a plunging decolle-
tage that they could not normally be worn without exposure of the breasts
(unless they are worn with an additional t-shirt or blouse underneath).
The comparison shows, in other words, that the description proposed by
Wierzbicka may well be adequate for the majority of cases in the range of
dress, but does not really cover all possible instances. Admittedly, such a
comparison is risky for at least two reasons. First, we start from the assump-
tion that English dress and Dutch jurk are equivalent as far as their referen-
tial range of application is concerned. As long as we do not have a similar
corpus-based analysis of dress as the one we have made for jurk, the com-
parison will have to remain a conditional one. Second (and more impor-
tantly), it is not even certain that Wierzbicka actually intends the definition
to apply to all the cases in the extension of dress. By introducing the phrase
'imagining things of this kind people could say these things about them', the
perspective is shifted from the objective features of the things that are being
called dresses to the subjective image that people say they have about
dresses when they are asked for it. In a sense, Wierzbicka defines dress by
referring to what people think dresses are. And if what people think dresses
are only involves the central cases of the category' dress', then, of course, it
makes no sense to complain that the description of this mental image does
not apply to non-prototypical dresses: it never intended to do so anyway. On
this reading of Wierzbicka's view, its reference to subjective images could
be construed as implying a conscious restriction of the description to the
prototypical core of the category. And because introspection probably does
work efficiently for retrieving such prototypical images, the introspective
method may be salvaged.
It is not quite clear, however, whether this interpretation ofWierzbicka's
position is a valid one. On the one hand, she argues that 'a valid definition
must be empirically adequate, that is, it must be phrased in such a way that
it covers the entire range of use of a given word, expression, or construction'
(1989: 738). On the other hand, if a definition such as that of dress is to be
applicable to all things that may be called by that name, people should be
able to assert all the characteristics mentioned in the definition any time they
see a dress. But surely, when imagining a less prototypical kind of dress than
the kind whose features are included in the definition, people will not imag-
ine it as a prototypical case. What people could say about dresses changes
when peripheral members of the category are at stake: default dresses, for
instance, may well cover most of the body, but that feature may be sup-
pressed when a fancy type of sununer dress is involved.
Even if, however, we accept that Wierzbickian definitions of the kind il-
lustrated above are explicitly restricted to the prototypical core of the cate-
gories, a counterargument may be advanced in favor of a method based on
observing actual usage. Note that it remains a matter to be settled empiri-
cally whether the lexical knowledge that people have in their in minds is
indeed restricted to a mental image of the core of the category in question. It
is not a priori given that the idea of a category that people may introspec-
tively retrieve from memory is an adequate reflection of the extent of that
person's actual knowledge of the category. On the contrary, if it is part of
his knowledge to produce or accept an application of dress to non-
prototypical cases, then he 'knows' more about the category than would be
included in his introspectively retrieved idea of the category. That knowl-
edge, to be sure, is not necessarily conscious knowledge; it is less 'knowl-
edge that (lexical item x may refer to entities with such and such characteris-
tics)' but rather 'knowledge how (lexical item x may be successfully used)'.
In order, then, to get a better grasp on the lexical 'knowledge how', usage-
based investigations of the type illustrated in this article are vital, precisely if
it is suspected that conscious knowledge may only partially cover the full
extent of a person's 'knowledge how' .
In short, the approach demonstrated above is inspired by the desire to
steer clear both of the danger of theoretical imprecision that is implicit in
Ruhl's strategy, and of the danger of empirical incompleteness that is im-
plicit in Wierzbicka's strategy. But if the approach followed here is indeed
to be preferred for general methodological reasons, what are the findings that
it leads to?
172 P o ~ y s e m y
6. Conclusions
If the methodological framework defended in section 5 is accepted, the inves-
tigation carried out in section 4 shows that not all lexical items can be clas-
sically defined in terms of necessary, jointly sufficient features. If classical
definability is considered a criterion for monosemy, this conclusion further
implies that not all lexical items are monosemous. At the same time, how-
ever, the investigation suggests that classical definability is not necessarily a
psychologically acceptable criterion for monosemy. The items whose defini-
tionally polysemous nature was determined in section 4 are not ones that are
intuitively recognized as polysemous. In this sense, the investigation sup-
ports and further illustrates the point made in Geeraerts (l987a) and
(l993a): the different criteria for polysemy that are quite plausible taken by
themselves may be mutually contradictory; specifically, vest etc. are defini-
tionally polysemous, but not intuitively. This implies, in other words, that
lexical semanticians cannot at the same time maintain the monosemic bias
and the ideal of classical definability. If, on the basis of the monosemic
bias, it is accepted that vest's intuitive univocality is a theoretically viable
reflection of an actual monosemy, the ideal that an individual meaning can
always be classically defined has to be relaxed (because there is no classical
definition for vest in the usage under investigation). Conversely, if one sticks
to the idea that classical definability is an integral part of what it is to be a
distinct meaning, the monosemic bias will have to be relaxed to the point
where even an intuitively non-suspect case of univocality like vest is ac-
cepted to be polysemous.
Section 3
Constructions and idioms
Chapter 7
The semantic structure of the indirect object in
Dutch
Originally published in Willy Van Langendonck and William Van Belle (eds.),
1998, The Dative 11. Theoretical Approaches 185-210. Amsterdam/Philadelphia:
John Benjamins.
To a considerable extent, Cognitive Linguistics is characterized by a strategy that
extrapolates models developed in the study of the lexicon to other areas of linguis-
tics. (Structuralism, by contrast. is characterized by a strategy that extrapolates
models of phonological analysis, like the use of distinctive features and distribu-
tional analysis, to the rest of linguistic structure.) Specifically in the domain of
semantics, productive models of meaning description like prototype theory ma-
tured in the field of lexical semantics and were then applied to grammatical phe-
nomena. This may be clearly seen in the area of Construction Grammar as repre-
sented by Goldberg (1995), Croft (2001), Fried and Ostman (2004), Ostman and
Fried (2005). Constructions are so to speak grammatical patterns treated as lexical
items: more or less schematic entities that should be treated as symbolic units in
their own right, with properties that cannot just be compositionally derived. But if
that is indeed the case, competing models for semantic description such as may be
found in lexical semantics are likely to play a role in Construction Grammar as
well.
In the present paper, I argue that the multidimensional form of prototype-based
analysis that is illustrated in Chapter 3 is relevant for the description of construc-
tions just as well as for the description of lexical items. Such a multidimensional
model, in which co-variation of semantic shifts along different dimensions consti-
tutes the grid of a prototype-based network, contrasts with a simpler type of radial
network analysis (as represented by Goldberg's analysis of the dative construc-
tion) in which the crucial elements of the network are points (meanings, readings,
senses) rather than dimensions.
Methodologically speaking, the paper is based on introspectively derived exam-
ples rather than corpus-based materials. and that is clearly not the way in which I
would proceed at the moment. Current work by Colleman (2005) shows however
that the semantic structure for the Dutch dative construction suggested here may
be fruitfully applied to corpus data.
176 Constructions and idioms
1. Aims and background ofthe exercise
The purpose of the present article may be defined from two points of view: a
more narrow descriptive one, and a broader theoretical one. Narrowly de-
fmed, the paper intends to show that the semantic structure of the indirect
object construction in Dutch can be fruitfully described,
a) by starting from a spatial prototype (the indirect object as the recipient of
a spatial transfer)
b) by deriving further uses from this basis by taking into account four spe-
cific conceptual mechanisms of semantic change, viz. generalization, meta-
phor, metonymy, and subjectification
c) by systematically exploring the various relations of these mechanisms, i.e.
by describing the field of application of the indirect object as a multidimen-
sional space.
Broadly defined, the paper will try to bring home three general theoretical
points about the semantic structure of syntactic constructions. The first point
implies that such constructions can be semantically described by means of
basically the same descriptive mechanisms that apply to lexical categories.
The second point involves the view that the semantic structure of categories
is best described as a multidimensional space. The third point consists of the
recognition that the semantic description of syntactic constructions has be
intimately linked to their syntactic structure. All three points require some
additional discussion.
The first point is a straightforward consequence of one of the major
methodological strategies employed by Cognitive Linguistics, viz. to model
grammatical research on the basis of lexical research: the study of lexical
categories is taken as a methodological point of departure for the study of
categorization in the grammar at large, and grammatical construction types
are considered categories just like lexical items. Given that linguistic catego-
rization is the major focus of attention for Cognitive Linguistics, studying
the lexicon first is, in fact, a plausible step to take: the categorizing function
of the lexicon has received more attention in the linguistic tradition (and is
also, perhaps, easier to investigate) than that of grammatical constructs. The
intention to study the grammar of the language along the same lines as its
lexicon obviously leads to a specific kind of grammatical theory, viz. one in
which the grammar, like the lexicon, is conceived of as an inventory of sym-
bolic elements (i.e. meaningful units). At this point in time, this idea has
received its clearest and most elaborate form of expression in Langacker's
The indirect object in Dutch 177
Cognitive Grammar (1987, 1990a, 1991), and in Fillmore's Construction
Grammar (Fillmore, Kay and O'Connor 1988, Goldberg 1994).
A major difficulty for such a research strategy involves the formal identi-
fication of the grammatical constructions. In the domain of lexical research,
it is usually quite clear what separate entities have to be distinguished: in
spite of (stubborn but relatively marginal) problems surrounding the exact
demarcation of homonymy, the fact that lexical items have a constant phono-
logical form by and large suffices to identify what the basic units of the
analysis should be. In the domain of grammatical research, however, the
formal side of the categories/constructions cannot receive such a straight-
forward definition. For one thing, the grammatical categories do not neces-
sarily have a constant phonological form: word classes, for instance, may
well exhibit a type of formal constancy in terms of their distributional and
morphological properties, but these characteristics are already abstract
properties that do not correspond in a one-to-one fashion with the presence
of an identifiable and constant phonological form. For an other, we tend to
think of the formal side of grammatical constructions as being subject to
variations and transformations (in a non-technical sense), but it is not imme-
diately obvious which changes and alternations are allowed, and which ones
rather undermine the formal identity of the construction.
The purpose of the present article is definitely not to resolve this ques-
tion, but it should be pointed out from the start that the problem has immedi-
ate consequences for the task at hand. Traditionally, various construction
types have been subsumed under the label 'indirect object': ditransitive con-
structions featuring both a direct and an indirect object, some intransitive
constructions, and some prepositional constructions. Against the background
of the methodological question that we just raised, it is not clear whether it is
entirely justified to lump together these various constructions: does the indi-
rect object exist at all in its traditional definition, incorporating <V NP NP>,
some cases of <V NP>, and constructions of the type <V NP aan NP> or
<V NP voor NP>? The present article will not try to answer the question,
but rather circumvent the problem by focusing on the ditransitive construc-
tions (which will belong to the hard core of the construction in any event).
There will be some remarks about intransitive indirect objects (that is, indi-
rect objects in a V NP frame rather than a V NP NP frame), but none about
the prepositional indirect object. As for the demarcation of the single NP
indirect object constructions in contrast with single NP direct object con-
structions, the paper relies on the groundwork laid by Van Belle and Van
Langendonck (1992).
The second theoretical point mentioned above involves the question what
kind of semantic structure best describes the semasiological range of gram-
matical constructions (or, as the case may be, lexical items). Although the
representational formats used in Cognitive Semantics basically have the
same representational power (see Geeraerts 1995a), it is nevertheless the
case that some types of representation inadvertently favor a particular type
of analysis. Radial set representations of the type promoted by Lakoff
(1987), for instance, would seem to suggest a two-dimensional analysis of
semantic structures: starting from a prototypical centre, new readings are
derived directly or indirectly from the prototype in what ultimately consti-
tutes a two-dimensional plane of interconnected senses. As against this im-
pression of two-dimensionality, it has been argued that a more adequate
picture of the semantic structure of lexical items has to take into account the
multidimensional nature of semantic extensions: the structure of polysemous
categories is characterized by covariation on multiple dimensions. A case in
point is the reanalysis suggested in Geeraerts (1992a) of the radial set repre-
sentation of the Dutch preposition over proposed by Cuyckens (1991); a
related argument for a multidimensional approach is to be found in Deane
(1993).
The point to make, to be precise, is not one about representational for-
mats: it is not impossible, for instance, to adapt a radial set representation in
such a way that the multidimensional character of the semantic structure is
revealed (see, for instance, the way in which Brugman 1989 enriches a basic
radial set representation with a type of componential analysis that reveals the
multiple links between the readings in the radial set). Rather, the point is an
empirical one: the description of semasiological structures (of lexical items
and grammatical constructions alike) requires specific attention for the mul-
tidimensional nature of the semantic links among the readings of the item or
the construction. In the case that interests us here, a multidimensional analy-
sis of the indirect object construction will be contrasted with two related
analyses that do not pay as much attention to this multidimensional charac-
ter, viz. Goldberg's radial set analysis of the English ditransitive construc-
tion (Goldberg 1992), and Rudzka-Ostyn' s hierarchical network analysis of
the Polish dative (Rudzka-Ostyn 1996).
The third theoretical point mentioned above might seem to be uncontro-
versial: it is a rather obvious idea that the semantic description of grammati-
cal constructions had better be systematically linked to their formal struc-
ture. More concretely, a construction-based definition will have to take a
relational form: if the indirect object constituent is explicitly defined as part
of the constructional frame <V NP NP>, then it should receive a definition
that includes a reference (however indirectly) to the other constituents in the
construction. Given the overall plausibility of this idea, the present paper
will highlight (by way of example rather than through elaborate theoretical
discussion) the connection between this point and the previous one: it will be
shown that the relational character of the definition of the indirect object
constituent plays a crucial role in the multidimensional nature of its semantic
structure.
2. The overall structure of the analysis
The basic goal of the analysis is to show that the indirect object construction
is semantically cohesive: the various instantiations of the category are linked
to each other by means of semantically plausible links, just like the senses of
ordinary lexical items are linked to each other on through semantic associa-
tions. The links that will be envisaged here largely belong to the traditional
repertoire of semasiology as established by the prestructuralist tradition of
diachronic semantics: we will make use of processes of generalization, of
metonymy, and of metaphor. In addition, we will invoke the notion of sub-
jectification, which is a more recent addition to the inventory of mechanisms
of semantic change.
The starting-point of the analysis consists of those cases in which the in-
direct object refers to the recipient of a benefactive material transfer, as in
hi} gal haar een roos 'he gave her a rose'. The prototypical indirect object,
in other words, is defined in terms of a frame featuring a process of transfer,
something being transferred, and a specific functional character of the trans-
fer. The indirect object itself, of course, is the person to whom the transfer is
directed and who benefits from the transfer. The choice of this particular
construction as the prototype of the category is not an arbitrary one: it links
up with traditional localist conceptions of case relations, in which spatial
aspects of meaning are considered to be basic. Although this lies beyond the
scope of the present paper, it should be noted that independent evidence for
the identification of the prototype might in fact be looked for: a frequency
analysis of corpus data, or data of a psychological type involving the ease
with which informants produce particular examples of the indirect object
construction, might very well confirm the central position of the 'spatial'
kind of use selected here. In the context of the present paper, however, the
choice for this particular prototypical center is a hypothesis that can be con-
finned only indirectly. It will turn out to be a suitable hypothesis, in fact, if
it indeed provides a good starting-point for deriving a plausibly coherent
analysis of the indirect object construction.
It is important for the overall structure of the analysis in the following
pages, that the prototype as defined here contains three major components: if
the prototypical indirect object constituent is the recipient of a benefactive
material transfer, semantic changes may take their starting-point in the 're-
cipient' part of the definition, in the 'transfer' part of the definition, and in
the 'material' or 'benefactive' nature of the transfer. The observation that
the prototype may be analyzed into a number of conceptual components has
interesting repercussions in the light of the third theoretical point mentioned
in the previous section, but discussion of that point will be saved for the final
section of the paper. At this point, it suffices to see that mechanisms of ex-
tension such as metaphor and generalization may attach to different compo-
nents of the prototype. And at the same time, of course, each component
may be subject to different types of semantic shifts.
3. Extension 1: Generalization
Starting from the indirect object as recipient of a material transfer as proto-
type of the construction, the first two dimensions leading to extensions from
the prototype involve processes of metaphorization and generalization. Table
1 charts a number of these extensions, in such a way that their mutual com-
binations become apparent. Generalization features along the vertical dimen-
sion in the table, and metaphorization is charted along the horizontal dimen-
sion in the table.
(1) Ik geefhem een boek
'I give him a book'
(2) Haar werd een hart ingeplant
'Her was a heart implanted, she received a new heart'
(3) De wind blies hem de regen in het gezicht
'The wind blew the rain in his face'
(4) Ik geefhet huis een nieuw laagje verf
'I give the house a new coat of paint'
(5) Ik schenk u mijn koninkrijk
'I donate my kingdom to you'
(6) Ze hebben je verkeerde ideeen ingeprent
'They have impressed- wrong ideas upon you'
(7) 20 'n geluk overkomt je nooit meer
'Such a happiness will never befall you again'
(8) De jury heeft 'De ontdekking van de hemel' de eerste prijs toegekend
'The jury has awarded the first prize to De ontdekking van de hemel'
(9) Ik zeg u dat hij komt
'I tell you that he comes'
Table 1. Generalization and metaphorization of the Beneficiary role
RECIPIENT OF RECIPIENT OF
transfer into the MATERIAL ABSTRACT
sphere of TRANSFER TRANSFER
CONTROL (1) (5)
USE (2) (6)
EXPERIENCE (3) (7)
AFFECTEDNESS
(4) (8)
RECIPIENT OF
COMMUNICATIVE
TRANSFER
(9)
Generalization as a traditional mechanism of semantic extension in lexi-
cal semantics involves all fonns of semantic shift in which the original read-
ing refers to a proper subset of the extension of the new reading. Classical
examples are the development from French arriver 'to arrive at the river"s
edge, to land' to the less specific meaning 'to arrive in general', or the devel-
opment of English moon 'earth"s satellite' to the more encompassing reading
'any planetary satellite. Alternative tenns featuring in the lexical semantic
literature for the concept of generalization include 'extension', 'weakening',
'broadening', and 'schematization' (in the Langackerian sense).
The process of generalization identified in Table 1 basically involves a
weakening of the Beneficiary-role of the recipient of the prototypical trans-
fer. In Table 2, tentative definitions are given of the four positions that are
distinguished along the generalization dimension in Table 1: for each of the
four labels distinguished along the vertical dimension of Table 1, a compo-
nential definition is presented that reveals their mutual relations. In accor-
dance with the definition of generalization, each secondary generalized read-
ing includes the original one as a special case. All cases of 'transfer into the
sphere of control', for instance, imply 'transfer into the sphere of experi-
ence". The overall structure, then, of the meanings affected by the process of
generalization is one of nested subsets.
The broadest category distinguished along the generalization dimension
merely implies that the indirect object constituent is affected by the transfer
referred to by the construction as a whole: the transfer involves a change (in
the broadest possible sense) in the indirect object. At the most generalized
end of the scale, the indirect object merely passively undergoes the effect of
the transfer, without in any way being able to control or manipulate the
transferred entity. At the other end of the scale, the indirect object constitu-
ent is a beneficiary in the strong sense: a human being who is free to use the
received entity according to his own intentions. In between these two extreme
positions, a continuum may be discerned going from 'control' over 'use' and
'experience' to 'affectedness'. Precisely because the generalization dimen-
sion is conceived of as a continuum, borderline cases are not to be excluded.
For instance, because the 'use'-category distinguished in the figure is proba-
bly not a very prominent one as far as frequency of occurrence is concerned,
it might already be considered a more or less marginal borderline case be-
tween the more prominent poles involving active control and manipulation
on the one hand, and mere sentient experience on the other hand. In general,
it is not necessarily the case that the intermediary positions distinguished in
Table 2 are the only ones to be retained on a closer analysis.
Table 2. Distinguishing features for control, use, experience, affectedness
the transfer the transfer the transferred the benefici-
involves a involves a entity is used ary has ma-
change of sentient, mostly but not ma- nipulative
attributes and human being nipulated by freedom in
characteristics who is con- the beneficiary his use of the
in the indirect scious of the transferred
object con- change entity
stituent
CONTROL
+ + + +
USE
+ + +
EXPERIENCE
+ +
AFFECTEDNESS
+
4. Extension 2: Metaphorization
The metaphorization dimension charted in Table I features two relatively
straightforward patterns of metaphorical change. On the one hand, there is a
shift from cases involving material transferred entities to cases involving
abstract transferred entities. The link with the non-metaphorical readings
shows up in the fact that the different positions on the generalization dimen-
sion that we distinguished earlier, also appear to be relevant for the meta-
phorized readings. The various abstract entities that are being transferred,
have effects on the recipient that fall into the same categories as when mate-
rial objects are being transferred.
On the other hand, there are metaphorical extensions into the domain of
communicative actions; broadly speaking, these transfers seem to rely on the
conduit metaphor (Reddy 1979): messages are conceived of as entities that
are being transferred from one person to the other. Logically, the recipient of
a material transfer may thus be metaphorized into the recipient of a commu-
nicative transfer.
One question that immediately imposes itself against the background of
these metaphorical extensions is the following: why is the distribution of the
instances of 'communicative transfer' over the 'control' etc. series so defec-
tive? An immediate answer to the question involves the recognition that a
concept such as 'control' (as defined along the generalization dimension)
simply does not apply to the objects of communicative transfers: the recipi-
ent of the message is more passive with regard to the message received than,
for instance, the recipient of a present with regard to the present. Similarly,
communicative transfers presuppose human (or at least animate) recipients,
which excludes the 'affectedness' category as defined along the generaliza-
tion dimension (because it specifically involved non-human recipients).
Going one step further, however, it may be suggested that the asymmetri-
cal distribution of the examples in Table I derives from a shift of perspec-
tive with regard to the prototype. Each of both metaphorical dimensions
would then capitalize on different aspects of the prototype, and the different
distributional patterns of both metaphors would be a reflection of that shift
of perspective. Pursuing this line of thought, let us first identify the two per-
spectives involved. If the prototype case of the indirect object construction
involves the indirect object as the recipient of a material transfer, the focus
may lie either on the functional aspects of the transfer (the transfer as a
process with specific effects for the recipient), or on the spatial aspects of
the transfer (the transfer as a material change of place of an object). The
simultaneous presence of both perspectives is reflected by the fact that the
indirect object constituent in the prototypical case may be either identified as
a Beneficiary (according to the functional perspective) or as a Recipient
(when the focus is shifted somewhat to the spatial point of view). According
to the functional perspective, the generalization continuum described in Ta-
ble 2 is of prime importance: the various positions on the scale identify dif-
ferent forms of functional effects that the indirect object constituent may be
subject to. Conversely, distinctions between 'sphere of control', 'sphere of
experience', 'sphere of affectedness' etc. recede to the background when the
more purely orientational, spatial perspective is taken. Given, then, that the
conduit metaphor takes its starting-point in the orientational conception of
the literal prototype case rather than in the functional conception, it is not
surprising to find that the distinction between the different positions on the
generalization dimension turns out to be largely irrelevant for the metaphori-
cal shifts into the field of communicative transfers.
It may be noted that comparative evidence for the multiple perspectiviza-
bility of the prototype case might be found in languages where the dative
case (the formally marked indirect object) expresses not just the notion 're-
cipient of transfer', but also the concept 'goal, direction of movement', in a
clear-cut spatial, orientational sense. Such an extension of the category's
range is a generalization of the spatial aspect of the transfer prototype,
rather than of its functional aspects: the orientation of the movement is en-
tirely detached from the notion of transfer.
Incidentally, if we consider the dative in languages like Latin a morpho-
logically marked indirect object, and if we further take into account that this
morphological dative in Indo-European seems to be a development of an
original locative case, it becomes clear that the diachronic relationship be-
tween the goal-reading and the recipient-reading may well be the converse of
the synchronic relationship sketched above: historically speaking, the spatial
aspects may have been primary with regard to the functional aspects. Such
prototype shifts are not a theoretical anomaly, to be sure (compare Geeraerts
1997a).
5. Extension 3: Metonymy
Two kinds of metonymy may play a role in the structure of the indirect ob-
ject. First, we find indirect identifications of the indirect object entity: in a
pattern like hij schrijft {het pa/eis, het stadhuis, het Instituut, de Wetstraat}
een brief 'he writes {the palace, the town hall, the Institute, Downing Street
10} a letter', the human recipient of the communicative transfer is indicated
by means of a reference to something (a building or dwelling-place) that is
characteristically associated with the recipient in question. This is a form of
metonymy that is quite wide-spread, and that is in no way specific for the
indirect object construction. It will not be envisaged any further in what
follows.
Second, and more importantly, we find metonymical ramifications of the
transfer aspect of the core indirect object meaning ('recipient of functional
transfer'): instead of naming the recipient in an actual process of transfer,
the indirect object constituent identifies entities or persons involved in ac-
tions or states of affairs that are metonymically associated with processes of
transfer. Two important pattems may be discemed. On the one hand, there
are states of affairs following acts of transfer: resultant states, effects, effec-
tuated changes. If one says, for instance, dit cadeau is mU een kostbaar
bezit 'this present is a precious possession to me', one describes oneself as
the beneficiary of an act of transfer, but focusing on the situation after the
actual transfer: the offering of the present itself is implied, but it is not ex-
plicitly mentioned. In such cases, the indirect object may be broadly defined
as 'the entity affected by an (explicitly mentioned) state resulting from an
(unmentioned) functional transfer (such as would be named in a core exam-
ple of the indirect object construction)'.
On the other hand, we find references to acts and states of affairs preced-
ing acts of transfer: conditions, preparations, preliminary actions, intentions.
If, for instance, one promises something to someone (ik beloof hem het
koninkrijk, 'I promise him the kingdom'), the transfer has not yet taken
place, but one describes someone as the beneficiary of a preliminary action
(the promise) that will normally lead to the actual transfer. In these cases,
the indirect object may be roughly defined as 'the entity affected by an (ex-
plicitly mentioned) action that is preparatory with regard to an
(unmentioned) functional transfer of which the entity is the recipient'.
Table 3 features examples of these two metonymical pattems ('resultant
state' metonymies and 'preceding conditions' metonymies) for most of the
types of use that we distinguished previously. In order not to clutter the fig-
ure too much, however, some simplifications have been applied: along the
generalization dimension, the 'use' category has been left out; likewise, no
instances of communicative transfers are included.
(10) Hij schenkt haar ko/fie in
'He serves her coffee'
Ik gun hem het boek
'1 grant him the book'
(11) Ik be/aa/hem het koninkrijk
'1 promise him the kingdom'
(12) Er stond hem een regenbui te wachten
'A shower of rain was awaiting him'
(13) Zij gunnen hem dat genoegen niet
'They do not grant him that pleasure'
(14) De burgemeester wenste het nieuwe museum veel bezoekers toe
'The mayor wished the new museum many visitors'
(15) Ik wens deze nieuwe theorie niets dan mislukte experimenten toe
'1 wish this new theory nothing but failing experiments'
(16) Dit cadeau is mij een kostbaar bezit
'This present is a precious possession to me'
(17) Het rijk behoort haar toe
'The empire belongs to her'
(18) De hagel was hem onaangenaam
'The hail was unpleasant to him'
(19) Het valt mij moeilijk dat toe te geven
'Tt is for me difficult to admit that'
(20) De nieuwe kleur staat het huis goed
'The new colour becomes the house'
(21) De aanpassingen waren de theorie tot voordeel
'The changes were to the advantage of the theory'
Table 3. Metonymical extensions of the indirect object construction
CONTROL material
abstract
EXPERIENCE material
abstract
AFFECTEDNESS material
abstract
PRECEDENT ACTIONS
METONYMY
(10)
(11)
(12)
(13)
(14)
(15)
RESULTANT STATE
METONYMY
(16)
(17)
(18)
(19)
(20)
(21)
It may be noted that the 'resultant state' cases typically involve the in-
transitive indirect object. This is probably an automatic, semantically driven
consequence of the shift of perspective involved in the 'resultant state' me-
tonymies: when the attention is shifted to the outcome of the transfer, the
initiator of the initial transfer recedes to the background of the attention.
A second point to note involves the fact that on top of the cases charted in
Table 3, an additional type of metonymical extensions has to be distin-
guished. Let us notice, to begin with, that metonymical extensions need not
imply that the original meaning is always relevant in those contexts where
the new metonymical reading is applied. On the one hand, we find cases like
he drank a glass: what he drinks is a 'quantity of water in a glass', but there
is always at the same time a glass 'recipient of specific form and substance'
around. On the other hand, we get cases like barrel: a barrel is a 'specific
quantity of oil contained in a barrel', even if it is, in the circumstances, con-
tained in a tanker or a pipeline. In the same way, to fall signifies 'to be killed
by a weapon' even if the person does not fall down ('the parachutist who got
caught in the tree fell under enemy fire'). Although this is not traditionally
done, developments such as those affecting barrel and to fall could be seen
as a generalization based upon an initial metonymy. For instance, barrel
would undergo a metonymic shift from the initial 'recipient' reading to the
reading 'amount of oil in such a recipient', and then further by generaliza-
tion to 'same amount of oil, even if not actually in the recipient'.
Now, regardless of the exact kind of analysis that one may wish to apply
to such cases, it is crucial not note that the kind of extension exemplified by
barrel and to fall also obtains in the indirect object construction. When, in
fact, the metonymical 'resultant state' cases contained in Table 3 occur
without reference to processes of transfer, 'pure' cases of 'possessor', 'ex-
periencer', and 'affected entity' readings arise. For instance, the prototypi-
cally active recipient with powers of control and manipulation that features
in the central cases of benefactive transfer, turns metonymically into an ac-
tive, controlling entity in a stative relationship resulting from such a transfer;
by a further process of generalization, it becomes an active, controlling en-
tity in a stative relationship (regardless of whether an initial transfer took
place or not).
Table 4 contains a number of examples of the shift intended here. The
examples are restricted to cases that link up with the 'resultant state' me-
tonymies, but it should be noted that similar examples may be found with
'precedent conditions' metonymies. For instance, Ik gun hem zijn overwin-
ning 'I grant him his victory, I do not begrudge his victory' may be said by
someone who has no power at all to bring about (or to hamper) the victory.
(22) Van al zijn lichaamsdelen was zijn linkerhand hem het dierbaarst
'Of all the parts of his body, his left hand was most dearest to him'
(23) Pannekoeken zijnmij het lekkerst
'Pancakes are the most tasteful to me'
(24) Jungles zijn tropische landen eigen
'Jungles are typical for tropical countries'
(25) Je vermogen om zeer snel te lezen zal je goed van pas komen
'Your capacity for reading very fast will be an advantage to you'
(26) AlIe vriendschap leek hem nutteloos
'All friendship seemed useless to him'
(27) Zwaarmoedigheid is alle romantische muziek eigen
'Melancholy is common to all romantic music'
Table 4. Secondary metonylllical extensions
CONTROL
EXPERIENCE
AFFECTEDNESS
MATERIAL
(22)
(23)
(24)
ABSTRACT
(25)
(26)
(27)
6. Extension 4: Subjectification
On the basis of the 'experiencer' and 'affected entity' readings in the stative
relations described at the end of the previous paragraph, a further extension
leads to the so-called ethical dative, as in the idiomatic expression het is me
wat 'that's something [to me, as far as I am concerned]!': the infonnation
contained in the utterance is part of the sphere of experience of the indirect
object (which is identical with the speaking subject), who is concerned, in-
volved, interested with regard to the objective situation or event described by
the utterance. Characteristically, the verb phrase need not refer to a process
of transfer, which is an indication of the fact that the ethical dative is indeed
a further development from the 'stative' cases described at the end of the
previous paragraph.
Against the background of the framework sketched so far, the ethical da-
tive may be explained as a shift from the referential domain of the utterance
to the discursive context. The interested party is not part of the referential
domain in which the action takes place, but of the pragmatic domain in
which the utterance is made. This shift is an example of the subjectification
processes as described by Langacker (1990b) or Traugott (1989). (There are
some differences between Langacker's and Traugott's conception of subjec-
tification, but the differences need not detain our attention here). In general,
subjectification involves a change in the conceptual construal of a particular
relationship: from an objective construal (in which the relationship is con-
strued without reference to the speech situation), a change is made to a sub-
jective construal, in which the speaking subject (or another element from the
speech situation) features as one of the relata in the relationship in question.
In our case, the relationship could be, for instance, 'affectedness'. In the
more common cases of the indirect object construction, the affected entity is
part of an objective referential situation that is conceived of as being inde-
pendent of the speech situation. In the case of the ethical dative, on the other
hand, the affected entity is the speaking subject, as a basic element of the
speech situation.
7. Extension 5: Perspectival switching
A final extension to be noted involves cases in which the transfers that fea-
ture in the prototypical cases change polarity, so to speak, and are trans-
formed into cases of blocked transfers or deprivations. Rather than receiving
something, the indirect object entity is deprived of something: something is
taken away from him that he already possessed, or he is denied something
that he wishes for or that he expects to receive. The nuances that this reading
occurs with are largely similar to the ones that were identified before. With-
out trying to be exhaustive about the possibilities, Table 5 lists examples
involving actual transfers (where 'negative transfer' equals deprivation),
together with a few examples involving stative relations; in each case, both
material and abstract instantiations are included.
(28) Hij nam haar het speeltje a(
'He took the toy from her'
(29) Een wolk ontnam ons het laatste straaltje zon
,A cloud took the last ray of sunshine away from us'
(30) Dit bericht ontnam haar de laatste kleur die haar gelaat nog sierde
'This message took the last shade of colour away that adorned her face'
(31) Het ontbreekt hem aan het nodige geld om dat huis te kopen
'He lacks the necessary money to buy that house'
(32) Ze ontnamen hem aUe rechten
'All his rights were taken away from him'
(33) Het bericht ontnam Hemline haar gevoel van rust
'That message took Hermione's quiet feelings away'
(34) De nieuwe snelweg ontneemt het landschap zijn laatste aantrekke-
lijkheid
'The new highway deprives this landscape from its last point of attrac-
tiveness'
(35) Het ontgaat haar dat het hoogtepunt van het generativisme achter de
rug is
'She fails to see that the heyday of generativism is over'
Table 5. Perspectival switches in the indirect object construction
negative transfer from the sphere of control
negative transfer from the sphere of experience
negative transfer from the sphere of affectedness
stative relation of affectedness
MATERIAL
(28)
(29)
(30)
(31)
ABSTRACT
(32)
(33)
(34)
(35)
The explanation of the shift of perspective (or, if one wishes, the shift of
polarity) that is illustrated by these examples is less straightforward than in
the previous cases. Probably the most plausible analysis is to see the shift as
the result of a generalization process affecting the 'transfer' component of
the prototype case. Earlier on, we saw that the transfer component was sub-
ject to a metonymical shift in cases like Ik gun hem het back 'I grant him the
book'. In the cases that concern us here, the transfer component is subject to
a different type of change: it is generalized by dropping the 'benefactive'
aspect that was associated with the prototype. The transfers envisaged here,
in fact, do not add something to the sphere of control (etc.) of the indirect
object constituent, but rather take something away: instead of being trans-
ferred in the direction of the indirect object constituent, the transferred entity
moves away from the indirect object. Obviously, the indirect object can then
no longer be characterized as a 'recipient', but had better be characterized as
'affected entity' or 'person concerned'. In the cases intended here, in other
words, the generalization of the transfer component correlates with a shift in
the 'recipient' component.
Incidentally, this discussion shows that the label 'benefactive' that was
used initially to characterize the prototype case may be slightly misleading,
in the sense that it combines (in typically prototypical fashion) two different
features that co-occur in the prototype case, but that are nevertheless sepa-
rable. A benefactive transfer is typically one in which something good is
given to someone: a positive for the indirect object constituent combines with
a directionality oriented towards the indirect object. At the other extreme, the
cases of deprivation illustrated in Table 5 involve transfers in which some-
thing good is taken away from someone. Notice, however, that giving some-
thing bad to someone (hi) gaf hem een klap 'he hit him') combines negative
functionality with an orientation towards the indirect object. In the same
way, taking something bad away from someone (haar antwoord ontnam
hem zi)n somberheid 'her answer took away his melancholy mood') com-
bines a positive, beneficial effect with an orientation of the transfer away
from the indirect object constituent. The occurrence of all four combinations
of directionality and effect shows, in short, that the generalizing extension of
the 'benefactive transfer' aspect of the prototype actually covers two differ-
ent processes. On the one hand, there is a perspectival change involving the
directionality of the transfer. On the other, the functional aspect of the proto-
typical transfer may similarly (but independently) undergo a perspectival
switch.
Needless to say, this discussion as a whole once more establishes the im-
portance of distinguishing between the functional ('benefactive') and the
spatial (directional, orientational) aspects of the prototype case - a point that
was already stressed in our discussion of the metaphorical extensions based
on the conduit metaphor.
8. The resulting picture
If the prototype case of the indirect object construction is defined as 'the
active recipient (with controlling power) of a benefactive transfer of material
entities', various semantic extensions appear to start from each of the as-
pects of the prototype. An overview of the various extensions is presented in
Figure 1. (For reasons of graphical economy, the extension towards the ethi-
cal dative is not included in the figure. Neither have the nuances of the per-
spectival switch discussed at the end of the previous section been explicitly
represented. )
Crucially, the points in the network sketched in Figure 1 are not separate
'senses' of the indirect object construction, but conceptual components that
co-occur in various combinations: Figure 1 only indicates the dimensions
along which the indirect object construction may vary, but it does not chart
all the individual readings that arise from the combination of the different
positions that occur along the variational dimensions. In this sense, the rep-
resentational fonnat of Figure 1 is fundamentally different from the 'radial
sets' representations that are common in cognitive linguistic research: while
the nodes in radial networks a la Lakoff (1987) constitute individual read-
ings of the category in question, the representation of Figure 1 gives an
overview of the conceptual components that may enter into such individual
readings. For instance, a sentence like Zo 'n ge/uk overkomt je nooit meer
(see Table 1) combines a shift from active recipient to passive experiencer
along the generalization dimension (represented by the line going from the
prototype to the lower right hand corner of Figure 1) with a shift from mate-
rial to abstract transfer along the metaphorization dimension (the axis drawn
vertically downward in Figure I).
affected entity
instead of
acti ve recipient
expenencer
instead of
active recipient
ill
ACTIVE RECIPlENT
OF MATERlAL ENTITY
OF BENEFICIAL TRANSFER
deprivation and
blocked transfer
instead of beneficial
transfer
communicative
transfer instead of
benefactive func-
tional transfer
preconditions
and resultant
states instead
of transfers
abstract enti-
ties instead of
material
entities
stative relations
instead of
transfers
ill GENERALIZATION
~ METAPHOR
Cl) METONYMY
Figure I. The multidimensional structure of the indirect object construction
Conversely, it is not necessarily the case that any point in the network
that results from the combination of the dimensions in Figure 1, should be
considered a different 'sense' (in the technical reading of that word). Already
in the field of lexical semantics, distinguishing polysemy from vagueness is
not always an easy task (compare Geeraerts 1993a). In syntax, then, the
difficulties are even greater, because the criteria to apply are even less clear
than in the field of lexical studies. Broadly labeled categories like Benefici-
ary or Affected Entity (Patiens), that occupy specific regions of the overall
network, could certainly be considered candidates, but then again, according
to what methodology should these categories be distinguished and demar-
cated? The road taken in the present paper has not been to try to answer the
polysemy question, but rather to sketch the network of different uses with
regard to which such a polysemy would have to be defined.
9. Further steps to be taken
The analysis of the Dutch indirect object construction presented here does
not pretend to be the final word on the subject, but merely tries to make a
number of points that may prove stimulating for further research. In princi-
ple, there are two possible areas for a further elaboration of the model pre-
sented here. In the first place, the analysis as such will have to be refined.
Specifically, by confronting the analyses given here with corpus data illus-
trating actual language use, it may be checked whether any additional se-
mantic nuances have to be added to the picture drawn in the previous pages.
Also, corpus data will yield frequency data that may confirm the prototype-
theoretical structure assumed here. Is it the case, for instance, that the shades
of meaning that are considered peripheral according to the present analysis,
do indeed occur less frequently?
In the second place, the model may be elaborated beyond the boundaries
of the analysis given here. For one thing, the question arises whether the
model has any typological and comparative validity: can the same structure
be used to describe developments in other languages? Note that this question
has two aspects to it: what languages share a similar construction with the
same prototype as the one identified for the Dutch indirect object (see, for
instance, Rudzka-Ostyn's identification of the prototype of the Polish dative:
Rudzka-Ostyn 1996)? And - given such a prototypical similarity - can the
extensions from the prototype that occur in the different languages be mod-
eled along lines that are similar to the ones set out for Dutch? Another area
for elaboration concerns the whole field of onomasiological research. The
present analysis is a semasiological one: it describes the semantic structure
of a single construction. At least for a number of the senses within the sema-
siological range of application of the indirect object construction, alternative
modes of expression exist. First among these are the 'prepositional indirect
objects', i.e. constructions of the type <V NP aan NP> or <V NP voor NP>.
Assuming that these may indeed be considered different constructions (see
the discussion in the first section of the article), the question arises whether
the exact relationship among these onomasiological alternatives might be
deternlined. In particular, would it be possible to analyze the choice between
the ditransitive construction and the prepositional indirect object along the
same quantitative lines set out in Geeraerts, Grondelaers, and Bakema
(1994)? In that monograph, it is shown that the choice among lexical alter-
natives is determined simultaneously by semasiological and onomasiological
salience effects (which may be calculated, given a sufficiently large corpus).
Now, if the basic research strategy of Cognitive Linguistics (viz. of model-
ing the semantic analysis of grammatical constructions on the example pro-
vided by lexicological analyses) is carried through systematically, an exten-
sion of the analysis towards the onomasiological domain will have to be
envisaged sooner or later.
10. Conclusion and discussion
The major conclusion to be drawn from the present analysis involves the
recognition that it is indeed possible to systematically and coherently de-
scribe the semantic structure of the (ditransitive) indirect object construction
in Dutch on a 'lexical' basis, that is to say, by using models for semantic
description developed in the lexicological domain. In the present analysis,
this takes the form of starting from a prototypical meaning and deriving the
other readings of the construction on the basis of well-known mechanisms of
semantic extension such as metaphor and metonymy. As it stands, this gen-
eral conclusion is not entirely original: the same idea lies at the basis of other
recent analyses of datives and ditransitives, such as Goldberg (1992) and
Rudzka-Ostyn (1996). On specific points of the analysis as well, there are
many parallels with existing studies in a functional or cognitive linguistic
framework. (These parallels have not been systematically indicated here, but
see Rudzka-Ostyn 1996 for relevant references.) In what sense, then, does
our analysis differ from these other proposals - in particular, from Goldberg
(1992) and Rudzka-Ostyn (1996)? Restricting the comparison to these two
articles, and focusing on fundamental matters rather than questions of detail,
three points should be mentioned.
First, instead of restricting the analysis to (or at least focusing it on)
metaphorical links, as in Goldberg's (1992) radial set analysis of English
ditransitives, other types of semantic extensions have to be considered. All
the classical and less classical mechanisms of semantic extension (such me-
tonymy, generalization, and subjectification next to metaphor) have to be
envisaged in the analysis. The 'lexical' structure of the semantics of the
indirect object construction cannot be adequately described by starting from
an impoverished conception of semantic structure. This is an approach that
the present analysis shares with Rudzka-Ostyn's (1996).
Second, in contrast with Rudzka-Ostyn's proposal, the present analysis
attaches no special structural position to extensions based on processes of
generalization (which appear in Rudzka-Ostyn's proposal in the form of
'schematizations'). Just like Goldberg's use of a radial set representation
seems to favor focusing on metaphorical links (given that the radial set
model is predominantly used in the Lakovian tradition of metaphor-based
analyses), Rudzka-Ostyn's adoption of a Langackerian schematic network
representation seems to have predisposed her analysis towards overstressing
the role of generalizations ('schematic' readings). Generalization, however,
is but one of the many mechanisms in which the semantic coherence of a
category may be grounded, and there does not seem to be any specific reason
for making it more important than others. In particular, a category need not
receive a single schematic reading to remain coherent. The obligatory search
for such schematic readings easily leads to forced conclusions, and Rudzka-
Ostyn's analysis is a case in point. Her identification of the topmost sche-
matic reading of the Polish dative as a 'landmark' is, in fact, too schematic:
it insufficiently distinguishes the dative from other constructions that also
function as landmarks. The whole point, of course, mirrors the discussion in
lexical semantics about the possibility of finding single definitions for lexical
categories. Such definitions have to be maximally general and minimally
specific: maximally general in the sense of applying to all the cases falling
within the range of the category, and minimally specific in the sense of dis-
tinguishing the category from other, non-synonymous categories. A defini-
tion of the dative as a 'landmark', then, does not conform to the requirement
of minimal specificity. On a more fundamental level, the point may be for-
mulated as follows: given that lexical semantic research has established be-
yond doubt that monosemic, 'schematic' definitions of lexical categories
need not always be possible (see Geeraerts, Grondelaers, and Bakema 1994),
one has to beware of a forced search for such definitions in the case of gram-
matical constructions as well.
Third (and foremost), instead of linking extended individual meanings di-
rectly to the prototype or to each other, a multidimensional structure has to
indicate how non-prototypical readings are linked among each other by the
co-occurrence of semantic shifts along various dimensions. Both Goldberg's
radial set analysis and Rudzka-Ostyn's schematic network analysis inter-
connect individual readings of the construction in question, but insufficiently
succeed in indicating the multidimensional structure underlying the intercon-
nections between the individual readings.
To appreciate the importance of these underlying dimensions, we may
have another look at Figure 1. The representation used there visualizes an
observation that was already brought to the fore in the second section of the
article: the dimensions along which the semantic extensions take place, relate
to specific components of the prototypical reading. For instance, assuming
that the prototype case may indeed be defined as the recipient of a benefac-
tive material transfer, one dimension of change pertains to the 'recipient'
component, another to the 'transfer' component, and so on. We should now
go one step further, however, and recognize that the major components in the
definition link up with specific components of the syntactic frame in which
the indirect object construction occurs. Focusing once more on the ditransi-
tive <V NP NP>-frame, it will be readily appreciated that the 'transfer'
component of the definition links up with the verbal part of the formal con-
struction, and the 'material' aspect of the definition with the direct object
noun phrase (there is a material transfer when the direct object noun phrase
refers to a material object). The 'recipient' itself, of course, is the second
noun phrase, the indirect object as such. The definition that we have used, in
other words, is a relational one that explicitly takes into account the syn-
tagmatic dimension of the indirect object constituent. One could also say,
alternatively, that the description is aframe-based one, in which the indirect
object is defined as part of a larger conceptual structure. (Incidentally, while
it might be suggested that this type of analysis characteristically distin-
guishes the description of verbal lexical items and grammatical constructions
from that of nominal ones, this is not entirely adequate. Any relational con-
cept, in fact, including relational nouns, will be most adequately described
by incorporating the relational aspects into the definition.) What does not
explicitly appear in Goldberg's and Rudzka-Ostyn's approaches, then, is
certainly not the idea that the indirect object or the dative have to be de-
scribed as (part of) a grammatical construction, but rather the insight that it
is precisely this relational description that underlies the multidimensional
nature of the semantic structure of the indirect object or dative.
To summarize, what we have done, and what largely constitutes the
specificity of the present approach, is a) to define the indirect object con-
stituent as part of a construction: the components of the definition system-
atically evoke the other constituents that feature in the grammatical con-
struction; b) to reveal the multidimensional semantic structure that
correlates with this relational definition: the polysemy of the indirect object
is determined by covariation of semantic extensions affecting the various
components of the relational, construction-based definition; c) to stress the
lexical nature of the semantic organization of the indirect object construc-
tion: an adequate picture of the changes affecting the components of the
definition can only be acquired by envisaging all the mechanisms of semantic
extension that have traditionally been identified in lexical semantics.
Chapter 8
The interaction of metaphor and metonymy in
composite expressions
Originally published in Rene Dirven and Ralf Parings (eds.), 2002, Metaphor and
Metonymy in Comparison and Contrast 435-465. BerlinlNew York: Mouton de
Gruyter.
Idioms and compounds have similar semantic characteristics: both types of ex-
pressions are semantically composite, in the sense that their meaning is (at least
in principle) composed of elementary building blocks, i.e. the constituent parts of
the expressions. Idioms have received considerable attention from Cognitive Lin-
guistics (see e.g. Gibbs 1986, 1994, 1995, Fillmore, Kay and O'Connor 1988,
Dobrovol'skij 1995, Feyaerts 1999), compounds much less so (see e.g. Ryder
1994, Rudzka-Ostyn 2003). The mechanisms of semantic compositionality as well
have received only limited attention, with the exception of work in the line of
mental space and blending theory, as in Sweetser (1999) or (Coulson 2001). The
present paper takes a closer look at the semantics of composite expressions, with-
out however attempting to provide a systematic comparison with the blending
analysis: drawing on Geeraerts and Bakema (1993) and Geeraerts (l995b), it
describes the interaction between the syntagmatic and the paradigmatic axes in
the meaning of idioms and compounds, and then charts the various ways in which
metaphor and metonymy can interact along these axes. The crucial idea is this:
when we study compositional processes in the framework of a semantically ori-
ented grammar like Cognitive Linguistics, we should not only describe the way in
which the constituent elements work together to produce a composite meaning,
but we should also take into account the semantic shifts that each of the elements
(and their combination) may undergo in its own right.
1. Metaphor and metonymy in compounds and idioms
The semantic architecture of idioms and compounds is identical: in both
types of expressions, meaning is compositional to the extent that it is built
up out of the constituent parts of the expression; at the at same time, this
compositional meaning is usually but a first semantic step, to the extent that
processes of meaning extension produce a figurative reading from the com-
positional one. This paper will explore the various ways in which metaphor
The interaction ofmetaphor and metonymy 199
and metonymy may interact in expressions with this type of semantic archi-
tecture. The first part of the paper describes the architectural characteristics,
i.e. the paradigmatic and the syntagmatic axes of composite expressions and
the way in which they interact. This part of the paper reproduces the major
part of Geeraerts (1995b). The phenomena in question will be introduced
with regard to idioms in particular, but the extrapolation to the semantics of
compounds is a straightforward one. The second part of the paper charts the
various ways in which metaphor and metonymy combine and interact within
the structure defined by the syntagmatic and paradigmatic axes. In particu-
lar, it will be shown that metaphor and metonymy may occur either in a
consecutive or in a parallel sequence. This part of the paper is loosely based
on Geeraerts and Bakema (1993). All the examples illustrating the argument
are taken from Dutch.
2. The prismatic architecture of composite expressions
2.1. Isomorphism and motivation
The paradigmatic and the syntagmatic dimension
1
of idioms are both two-
fold, in the sense that both can be considered with regard to the original,
literal meaning of the idiom, and with regard to the derived, figurative mean-
ing. The paradigmatic dimension of idioms primarily involves the relation-
ship between the original meaning of the idiomatic expression as a whole and
its derived meaning. Secondarily, it involves the relationship between the
original, literal meaning of the constituent parts of the idiomatic expression,
and the interpretation that those parts receive within the derived reading of
the expression as a whole. The syntagmatic dimension of idiomatic expres-
sions involves the relationship between the interpretation of the constituent
parts of the expression on the one hand, and the interpretation of the expres-
sion as a whole on the other, but clearly, this syntagmatic dimension can be
envisaged both with regard to the original meaning and with regard to the
derived meaning. Systematically, then, the semantic relations in idioms might
be charted by means of a prismatic structure as in Figure 1. (The figure is
intended to serve purposes of clarification and reference only; it is obviously
not a formal representation as envisaged in formal grammar. To keep mat-
ters simple, the expression is assumed to contain only two lexical items.)
1Expression as a whole in its literal reading
2 First constituent item in its literal reading
3 Second constituent item in its literal reading
4 Expression as a whole in its derived, idiomatic reading
5 First constituent item in its derived reading
6 Second constituent item in its derived reading
Figure 1. The prismatic structure of idioms
In addition, the notion of compositionality may either receive a dynamic
or a static interpretation. Within the dynamic interpretation, compositionality
is thought of as a syntagmatic derivational process in the course of which the
meaning of a compound expression is computed on the basis of the meanings
of the constituent parts of the expression'" In contrast with this dynamic,
bottom-up conception, a static interpretation can be envisaged that merely
notes that a one-to-one correspondence between the parts of the semantic
value of the expression as a whole and the meanings of the constituent parts
of the expression can be detected, regardless of the question whether this
correspondence has come about through a process of bottom-up derivation
or through a top-dO\vn interpretative process. Examples of such interpreta-
tive processes will be given later on; at this point, it suffices to see that a
'non-directional' conception of compositionality is not excluded. Now, be-
cause the term compositionality, through its processual connotation,
strongly calls up the idea of bottom-up derivation, a different term might be
helpful for the neutral, non-directional interpretation. I propose to use the
term isomorphism here: what is ultimately at stake, is a one-to-one corre-
spondence between the formal structure of the expression and the structure
of its semantic interpretation, in the sense that there exists a systematic cor-
relation between the parts of the semantic value of the expression as a whole
and the constituent parts of that expression.
A third preparatory step can be made by refining the notion of derivation.
Although statements to the effect that idiomatic meanings are not composi-
tional, that they are specialized, and that they cannot be derived straightfor-
wardly would seem to be interchangeable, it should now be clear that we
have to distinguish in principle between the syntagmatic underivability of the
meaning of an expression (which is, of course, its non-compositionality), and
its paradigmatic non-derivability. The latter involves the transparency of the
semantic extension that leads from the original meaning of an expression to
its transferred reading. It is illustrated by the interpretative difficulty that we
noted with regard to (3): along the top line of the prism in Figure 1, the tran-
sition from the literal meaning of iemand iets op de mouw spelden to its
idiomatic meaning is opaque. The latter cannot be derived on the basis of the
former, because the motivating image is lost. I propose to call this type of
derivability motivation (in contrast with dynamic compositionality as a syn-
tagmatic kind of derivability).
To round off the preparations, let us note that isomorphism and motiva-
tion as defined here
3
share a common characteristic: both involve the trans-
parency of some of the links indicated in Figure 1. More specifically, iso-
morphism coincides with syntagmatic transparency, whereas motivation can
be defined as paradigmatic transparency.
2.2. Combinations of isomorphism and motivation
We can now arrive at a basic classification of the specialized nature of idi-
oms when we consider the various combinations that isomorphism and moti-
vation can occur in. Isomorphism and motivation can, of course, each be
considered at two points in the relational structure sketched in Figure 1. On
the one hand, motivation may refer both to the paradigmatic top line of the
prism and to the two paradigmatic bottom lines. On the other hand, syntag-
matic isomorphism may involve the front triangle or the back triangle of the
figure. Because the entity that we are primarily interested in is the idiomatic
meaning of the expression as a whole (the top backside corner of the figure),
we will for now only consider motivation and isomorphism as they directly
relate to that part of the structure, i.e., we shall consider isomorphism within
the figurative plane, and motivation on the global level. Extensions towards
other parts of the structure will follow in the next sections. Consider, then,
the following expressions.
(1) De koe bij de horens vatten 'To take the cow by the horns> to take the
bull by the horns, to grasp the nettle'
Aan de weg timmeren 'To practice carpentry at the roadside, to work in
public> to attract attention by one's activities, to be in the limelight'
(2) Met spek schieten 'To shoot with bacon> to tell a tall story, to boast'
Een wit voetje bij iemand hebben 'To have a white small foot with
someone> to be in someone's good books, enjoy someone's favours'
(3) JI/fet de handen in het haar zitten 'To sit with one's hands in one's hair
> to be at one's wit's end, to be in trouble'
Dat heeft niet veel om het lijTThat does not have much around the
body> there's nothing to it it does not mean very much'
(4) De kat de bel aanbinden 'To tie the bell to the cat> to bell the cat, to
take the lead in a dangerous activity'
Als puntje bij paaltje komt 'When point reaches pole> when it comes
to the crunch, when all is said and done, when you get down to the
nitty-gritty,
Systematically, the idiomatic readings in (1) are both isomorphic and mo-
tivated. Those in (2) are isomorphic but not motivated, those in (3) moti-
vated but not isomorphic, and those in (4) neither isomorphic nor motivated.
The isomorphic nature of de kae bi} de harens vatten follows from the fact
that a consistent one-to-one mapping can be defined between the elements of
the global meaning and the meanings of the constituent parts of the expres-
sion. If we paraphrase the idiomatic meaning as 'to tackle a problem or a
difficulty at the central, most dangerous or difficult point', it becomes clear
that the cow maps onto the problem in its entirety, while the horns represent
the most tricky part of it; taking hold of the horns further symbolizes tack-
ling the core of the problematic situation. Similarly, met spek schieten is
isomorphic because the tall tales that are told can be seen to correspond with
spek, while the telling of the tales corresponds with schieten. Conversely, it
is difficult to identify those aspects within a situation of being at one's wit's
end that could map isomorphically onto the various aspects of the situation
described by met de handen in het haar zitten: what would be the hands,
and what would be the hair, for instance? As far as motivation is concerned,
however, it can be readily appreciated that met de handen in het haar zitten
is a metonymic expression for a situation of being in trouble; taking one's
head between one's hands and pondering the situation is precisely what one
does in the circumstances. In the same vein, it is easy to see that the literal
situation described by de koe bi) de horens vatten is a metaphorical image
for tackling a problem at its most difficult spot. But it is unclear why shoot-
ing with bacon should come to indicate boasting, or why a point reaching a
pole should have anything to do with things becoming serious.
For further clarification of this basic classification, four remarks have to
be made. In the first place, the isomorphic relations identified above should
not be confused with the question whether the isomorphically mapped read-
ings of the constituent elements of the idiomatic expressions are themselves
motivated. For instance, while the lexical item koe in de koe bi) de horens
vatten maps onto the 'problem' part of the global figurative reading 'tackle a
problem by its most difficult aspect', there is no independent motivation for
extending the semantic range of koe towards the meaning 'problem'; a se-
mantic shift from 'cow' to 'problem' is not a conventional aspect of the
meaning of koe, nor is there a readily conceivable independent metaphor that
leads from 'cow' to 'problem'. In other words, the bottom paradigmatic lines
in Figure 1 are not present in the case of de koe bi} de horens vatten. 4 It
should now also be clearer why I suggested to take into account isomor-
phism as a non-directional concept of compositionality. Although de koe bi}
de horens vatten exhibits isomorphism, the idiomatic meaning 'tackle the
problem by its most difficult aspect' could never be arrived at by means of a
bottom-up compositional process, because the building blocks for that proc-
ess (for instance, an interpretation 'problem' for koe) cannot be reached
independently. For lack of an independent paradigmatic motivation at the
bottom of the prism, the input for a possible compositional process can only
be retrieved when the output of the process (the global figurative meaning of
the idiom as a whole) is already available.
In the second place, motivation and isomorphism may be partial. Given a
paraphrase 'to give the orders' of de lakens uitdelen (mentioned under (5)
below), it is isomorphically possible to map lakens onto 'orders' and uit-
delen onto 'give'. At the same time, it is possible to imagine a situation in
which the person responsible for distributing the sheets is generally in
charge; as such, the idiomatic meaning is motivated. But the motivational
link is weak: distributing sheets is not the kind of situation that is typically
associated with being in charge (or at least, not any more: apparently, the
image derives from the dominant position of the lady of the house, whose
control over housekeeping is symbolized by her control over the Iinen-
cupboard). In de kogel is door de kerk, an event (such as the making of a
decision) whose coming about has been delayed or hindered by some kind of
obstacle, has finally occurred. The general picture is fairly clear: a material
obstacle (the church) obstructs and slows down the movement of the bullet,
in the same way that various difficulties obstruct and slow down the materi-
alization of the long-awaited event. But why a church and a bullet? Again,
the transparency of the motivating image is only partial.
(5) De lakens uitdelen 'To hand out the sheets> to run the show, be the
boss, play the first fiddle'
De kogel is door de kerk 'The bullet has gone through the church> the
dice has been cast, a decision has been reached, things have taken a fi-
nal turn'
(6) Met spek schieten 'To shoot with bacon> to tell a tall story, to boast'
Abraham gezien hebben 'To have seen Abraham > to be over fifty'
Als puntje bij paaltje komt 'When point reaches pole> when it comes
to the crunch, when all is said and done. when you get down to the
nitty-gritty'
(7) VU de bol gaan 'To go out of one's head> to blow one's top, to go out
of one's mind with excitement'
Het hoofd verliezen 'To loose one's head'
Niet goed bij z'n hoofd zijn 'Not to be well in the head> to be soft in
the head'
Het hoofd loopt mij om 'My head is going round'
Z'n hoofd ergens bijhouden 'To keep one's head to something> to re-
main attentive, to keep one's mind on something'
Buiten zichze!fzijn 'To be beside oneself
Vit z 'n vel springen 'To jump out of one's skin> to be beside oneself
(with rage)'
Vitbarsten 'To burst out, explode'
Exploderen/ontplojJen 'To explode'
In de wolken zijn 'To be in the clouds> to be overjoyed'
In de zevende hemel zijn 'To be in the seventh heaven> to be on cloud
nine'
In de put zitten 'To sit in the pit> to be downhearted, be in the dumps,
feel down'
Door een dal gaan 'To go through a valley> to go through an abyss, to
suffer a depression'
Erbovenop zijn 'To be on top of it> to have overcome one's troubles'
In the third place, loss or weakening of motivation often results from cul-
tural changes. More often than not, the background image that motivates the
figurative shift is an aspect of the material or the immaterial culture of a
language conununity - and when the culture changes, the imagistic motiva-
tion may lose its force. A clear example is met spek schieten (repeated under
(6. Apparently, enemy ships were shot at with bacon (and similar fat sub-
stances) to facilitate setting them afire; the interpretation 'to boast' can then
be reached through the intermediary of an interpretation 'to subject someone
to verbal aggression, to overpower someone verbally'. In this case, the rele-
vant knowledge belongs to the material culture of old-time warfare at sea. In
Abraham gezien hebben, the relevant piece of knowledge belongs to the
realm of the inunaterial rather than the material culture. Because a meaning
'to be no longer young or inexperienced' can be derived to the extent that the
idiom is interpreted as a hyperbolic expression with the reading 'to have seen
someone from a long time ago', the motivation is only partial: why is the
borderline set at 50? In fact, it takes a good knowledge of the Bible to recog-
nize the background of the expression, which derives from the gospel of John
8: 57. It should also be clear by now, that the motivated nature of an expres-
sion is subject to considerable individual variation (depending, among other
things, on individual differences in one's familiarity with the historically
motivating context). This is not to say, however, that a vast encyclopedic
knowledge will always suffice to recover the motivation behind an expres-
sion. The latter may indeed be near to irretrievable, which is typically the
case when professional etymologists disagree on the origin of an expression.
In als puntje bij paaltje komt, for instance, some think of a transformation
of an older expression als putje bij paaltje komt 'when the pit comes to the
pole, when it comes to putting the pole into the pit', while others think of a
variation on de puntjes op de i zetten 'to dot the i > to be meticulous about
the details' .
In the fourth place, the motivating image need not be specific for the ex-
pression in question; moreover, the motivating image may be complex. At
this point, we can link up with the generalized metaphor research in the line
of Lakoff and Johnson (1980). Without going too deep into the matter, con-
sider uit de bol gaan in (7). The expression seems to be motivated by a
combination of at least three images that are each generalized ones, in the
sense that they provide a general motivation for various specific expressions.
First, THE HEAD IS THE LOCUS OF ONE'S SELF-CONTROL underlies het hoofd
verliezen, niet goed bij z 'n hoofd zijn, het hoofd loopt mij om, z'n hoofd
ergens bijhouden. Second, LOSING ONE'S SELF-CONTROL IS LEAVING THE
BODY is to be found in buiten zichze(f zijn, uit z'n vel springen, uitbarsten,
exploderen, and ontplojfen. And third, UP IS POSITIVE / DOWN IS NEGATIVE
is present in in de wolken zijn, in de zevende hemel zijn, in de put zitten,
door een dal gaan, erbovenop zijn.
2.3. Local motivation and absence ofliteral isomorphism
In the previous section, only part of the original representation in Figure 1
was envisaged. Enlarging our perspective to Figure I as a whole, there are
two additional phenomena to be dealt with. On the one hand, the global mo-
tivation of an expression can be contrasted with the 'local' motivation of
each of the elements in the expression, i.e., motivation along the top para-
digmatic line of the prism has to be supplemented with motivation along the
bottom line. On the other hand, isomorphism at the figurative back of the
prism can be contrasted with isomorphism at the literal front of the prismatic
representation.
PareIs voor de zwijnen in (8) below is an example of an idiom in which
Figure 1 is fully realized. The global image is motivated (it is easy to appre-
ciate what it means to throw valuable things at the feet of unworthy beings),
and the figurative meaning is isomorphic (parels maps onto the valuable
things that feature in the idiomatic meaning, and zwijnen maps onto the
lowly beings that they are surrendered to). At the same time (and this a ma-
jor contrast with the de koe bij de horens vatten example that we discussed
earlier), both the transition from pare! to 'valuable thing' and from zwijn to
'unworthy person' is motivated on the basis of an evaluative metaphor. Up
to a certain point, the metaphorical transition may even be said to be lexical-
ized (in the sense of being conventional): zwijn is a regular term of invective
in the same way that pig is, and een pareltje is a term of praise in the same
way that a gem, a jewel, a pearl are. (This is not to say, however, that the
metaphorically motivated readings of parel and zwijn in the idiom coincide
exactly with the lexicalized metaphorical meanings. For instance, as a term
of abuse, zwijn normally implies that the person in question lives an immoral
life, whereas no such implication need be present in parels voor de zwijnen.
The important point to see is that the lexicalized existence of zwijn in a read-
ing like 'unworthy person, specifically because of his immoral behavior'
strengthens the motivated nature of the related reading that the word receives
in the context of the idiom.)
(8) Parels voor de zwijnen gooien 'To cast pearls before swine'
(9) lemand de loe!a(steken 'To take the wind out of someone's sails> to
get the better of someone, to deprive someone of an advantage'
Tegen heug en meug 'Against heug and meug > against one's will, re-
luctantly'
T/an hot noch haar weten 'To know neither hot nor haar > to be totally
ignorant'
(l0) lets aan de kaak stellen 'To put something at the jaw> to expose, de-
nounce something'
lets op touw zetten 'To put something on the rope> to organize, plan,
start, launch something'
Given this analysis of pareIs voor de zwijnen as a fully motivated and
fully isomorphic idiom, two additional remarks can be made. To begin with,
it will be appreciated that the figurative reading of fully motivated and fully
isomorphic idioms can be arrived at along two interpretative routes: either
the global literal meaning is derived first and then transferred into the figura-
tive realm, or the shift from the literal meaning of the individual words to
their transferred meaning is effectuated first, to be followed by their compo-
sitional combination into the idiomatic meaning of the expression as a whole.
In terms of Figure 1, the interpretation can go from the bottom to the top
first, and then to the back, or it can go to the back along the bottom first, and
then move up. Or, in still other words, either the syntagmatic dimension is
gone through first, or the paradigmatic one takes precedence (starting, each
time, at the lower front of the prism). Which of both interpretative paths is
psychologically real (or merely preponderant) is another matter, but it is
important to see that both are at least in principle possible.
Further, the 'local', lexical motivation may be partial, i.e., it need not in-
volve all the items constituting the expression. When iemand de loef afste-
ken is interpreted as 'to deprive someone of an advantage (like an initial
superior position)', loef maps onto 'the advantage', while afsteken can be
associated with the notion of deprivation. In the latter case, the association is
motivated: the metaphorical transfer from the literal meaning 'to cut off of
afsteken to 'to deprive' is a transparent one. In contrast, no such motivating
link is possible in the case of loef, because the latter does not have a literal
meaning for the average speaker of Dutch. Notice also that there are grada-
tions in the degree of motivation,S Conventionalized shifts of meaning like
the ones we mentioned in the case of pareI and zwijn are stronger than the
shift from 'to cut off to 'to deprive' in the case of the latter exten-
sion is possible and perhaps even plausible, but it is not a conventional one.
Together with iemand de loefafsteken, the other expressions under (9) il-
lustrate the second major extension of the basic classification presented in
section 3. Because loe/. heug, meug, hot, and haar are cranberry morphs,
they illustrate the case in which there is no isomorphism on the literal level: a
global literal meaning cannot be computed because one or more of the lexi-
cal building blocks lacks a meaning of its own. (Incidentally, the earlier dis-
cussion of de loef afsteken has made clear that the absence of isomorphism
on the literal level can co-occur with isomorphism on the figurative level.) In
addition, it may be useful to mention the existence of 'hidden cranberry
morphs' such as kaak and touw in (l0). In the context that motivates the
idiom, the words have their older meanings 'pillory' and 'loom' respectively;
at present, these meanings are no longer common usage. The homonyms
kaak 'jaw' and touw 'rope', on the other hand, are high-frequency words.
The expressions aan de kaak stellen and op touw zetten, then, can receive an
interpretation incorporating the readings 'jaw' and 'rope' (as suggested in
(l0, but this is obviously not the kind of literal reading that could motivate
(on the level of the expression as a whole) the figurative interpretation of the
idiom. To the extent that kaak 'pillory' and touw 'loom' survive only in the
expressions aan de kaak stellen and op touw zetten, they are like ordinary
cranberry morphs; to the extent, however, that they formally coincide with
the homonyms kaak 'jaw' and touw 'rope', they can be called 'hidden cran-
berry morphs' .
It appears, in short, that the associative links presented in Figure 1 may
be present in various combinations. A full-fledged investigation into the
semantics of idioms, then, will have to include an overview of the various
ways in which the model sketched here may be partially realized. (See Geer-
aerts and Bakema 1993 for an example of what such an overview may look
like.)
2.4. Reinterpretation processes
The examples of isomorphism on the figurative level that were mentioned in
section 2.2 yield secondary, non-original interpretations for the items in-
volved in the process. For instance, the isomorphic association between
lakens and 'orders' in de lakens uitdelen yields a contextually determined
interpretation for lakens, but because the transition from laken 'sheet' to
laken 'order' is neither conventional nor motivated (in the sense that the
latter reading is a plausible and transparent semantic extension of the for-
mer), the contextually isomorphic interpretation of laken as 'order' is not
likely to acquire much structural weight in the lexicon. The process of con-
textual reinterpretation within the idiom is not always, however, without
structural importance.
6
(11) Spek'Verkoper 'Person who sells bacon> person who boasts'
Van heinde en verre 'From far and wide, from everywhere'
Met zijn talenten woekeren 'To make the most of one's talents> to ex-
ploit one's gifts'
In fact, the reinterpretation process can be shown to be real when the item
in question (in its secondary reading) comes to be used in isolation from its
original idiomatic context. This has happened with spek in met spek
schieten, which has led to the compound spekverkoper, as in (11) below.
The formation of the compound noun can only be explained if it is accepted
that spek in its idiomatically contextual reading 'boasting speech, tall tale'
has been isolated from the original idiom. S'pekverkoper as such cannot be
directly explained as an original metaphor of its O\\Il1: there is no way in
which selling bacon can be associated with boasting except through the in-
tem1ediary of met spek schieten. This type of 'semantic back-fommtion' is
also quite visible in the case of cranberry morphs. When people are asked
for an interpretation of heinde in van heinde en verre, it appears that a ma-
jority understands the item as a synonym of near-synonym ofverre (more or
less like the relationship between far and wide in the English counterpart of
the expression). Etymologically speaking, however, heinde and verre are
antonyms rather than synonyms; heinde is related to hand and basically
means 'what is near, what is at hand, what can be found in the immediate
neighborhood'. When the etymological relationship with hand and the se-
mantic relationship with nearness is lost, however, the overall meaning 'from
everywhere' of the idiomatic expression enables heinde to be reinterpreted as
a synony111 of verre. Met zijn talenten woekeren is an even clearer case.
Whereas talent in its original biblical context referred to a particular coin,
the reading 'personal ability, capacity, aptitude' that it received in the figu-
rative interpretation of the expression is now the major one; it occurs freely
in separation from the original expression.
From a very general perspective, reinterpretation processes such as these?
indicate that a search for isomorphism (defined as syl1tagmatic transparency)
is an active force in the mind of the language user. Apparently, isomorphism
on the figurative level of the idioms is not just real when it is given on the
basis of the literal meanings of the constituent elements of the expression,
but it is also real in the sense of being sought for when it is not given. If this
can be accepted, it also means that interpretation processes are not always
bottom-up, but that they can also be top-down: the overall meaning of met
zijn talenten woekeren determines the specific meaning of talent that has
become the item's major meaning.
(12) Vi/en naar Athene dragen 'Carry owls to Athens/Athena > do some-
thing irrelevant, useless. superfluous'
De kat de bel aanbinden 'Tie the bell to the cat> bring something out
into the open, make something public, ring a bell about something'
Similarly, evidence can be found that the search for paradigmatic motiva-
tion too is a real one. Of course, the search for motivation is quite outspoken
when etymological researchers disagree on the actual motivation of an ex-
pression, as in uilen naar Athene dragen, which is usually interpreted in
terms of the owl as the symbol of the goddess Athena and the city Athens,
but which some see as being motivated by the simple fact that there were
many owls in Athens. But etymologists are professional searchers for moti-
vation, so their interpretative creativity does not tell us too much about the
spontaneous occurrence of such interpretative activities when ordinary lan-
guage users use idioms. More important in this respect are, first, psycholin-
guistic investigations of the type reported on in Gibbs (1990) and later work,
which show that motivating images for idioms are psychologically real. And
second, evidence for actual reinterpretations along the paradigmatic axis
may be found, i.e. evidence for new meanings coming about through the
search for motivation. For instance, the older idiomatic meaning of de kat de
bel aanbinden (viz. 'to bell the cat, to take the lead in a dangerous activity')
refers to the old fable of the cat and the rats. Nowadays, however, it seems
to be shifting towards the interpretation 'to bring something into the open, to
make something public, to ring a bell about something': on the one hand, the
older association with taking the responsibility in a dangerous action (in
favor of other people) disappears into the background; on the other hand, the
notion of drawing the public attention to something (in particular, something
scandalous or negative) is foregrounded. Given that de kat de bel aanbinden
is largely unmotivated for most speakers, the association between the bell
referred to in the expression and the notion of making something public (of
making it heard, that is) enhances the motivated character of the idiom. The
search for greater motivation leads to a shift in the interpretation.
2.5. Summing up
What I have tried to indicate in the previous pages can be summarized in
three points. First, an adequate description of the various forms of semantic
specialization that occur in composite expressions requires that a number of
distinctions are taken into account: the distinction between the syntagmatic
and the paradigmatic aspects of meaning (which can both be found on the
The interaction o.f metaphor and metonymy 211
level of the literal meaning and on that of the figurative meaning of the id-
iom), and the distinction between bottom-up and top-down semantic proc-
esses. Specifically, the concepts of isomorphism (defined as syntagmatic
transparency) and motivation (defined as paradigmatic transparency) are of
primary importance for describing the semantics of composite expressions.
Second, semantic specialization is a matter of degree. More specifically,
the classificatory framework defined on the basis of the conceptual distinc-
tions just mentioned allows for a ranking of the degree of specialization in-
volved. Least specialized are fully isomorphic and fully motivated cases like
parels voor de zwijnen gooien. Somewhat less specialized are cases like de
koe bij de horens grijpen and met de handen in het haar zitfen, which may
not be derivable on a word-per-word basis, but which are entirely transpar-
ent along the upper side of the prismatic structure. Still further down the
line, we find cases like met spek schieten and de kat de bel aanbinden, that
lack the global motivation of the previous examples. Most specialized, fi-
nally, are cases where the literal meaning of the expression cannot even be
recovered, such as iemand de loeI afsteken and other idioms containing
cranberry morphs. In each of these cases, matters are further nuanced by the
existence of degrees of motivation.
And third, semantic interpretation is not just a question of bottom-up
compositionality or literal-to-figurative transfer. The reinterpretation proc-
esses that can be observed point to the existence of top-down and figurative-
to-literal interpretations. It is not just the case that literal meanings determine
figurative ones; figurative meanings also determine literal ones. And it is not
just the case that the meaning of the parts determines the meaning of the
whole; the meaning of the whole also determines the meaning of the parts.
3, Interactions between metaphor and metonymy in composite
expressions
3.1. The metaphor/metonymy continuum
If there is a continuum between metonymy and metaphor, this implies that
there are in-between cases between expressions that are fully metonymical
and expressions that are fully metaphorical. Composite expressions as well
can be fully metaphorical or fully metonymical, when the motivational links
that are present within the semantic architecture of the expression are only
metonymical or only metaphorical. Pare!s voor de zwijnen gooien, as dis-
cussed above, is fully metaphorical: the top level shift from 'to throw pearls
at swine' to 'to present unworthy people with valuable things' is a meta-
phorical one, and so are the bottom level shifts from pare! 'pearl' to 'valu-
able thing',8 and from zwijn 'pig' to 'unworthy person'. Conversely, com-
pounds of the bahuvrihi type, like roodhuid 'redskin', are well-known cases
of metonymical compounds: the link between the initial, compositional
meaning ('red skin') and the derived reading ('Indian, seen as one with a red
skin') is metonymical link of the possessed/possessor type. In the roodhuid
case, to be sure, motivational links at the bottom level fail, because the de-
rived reading 'redskin' cannot be considered isomorphic.
So, given that composite expressions can be either metaphorical or meto-
nymical, how can we chart the in-between cases? How do metaphor and
metonymy occur in mutual combination in compounds and idioms? I will
argue that there are three basic cases to be distinguished: cases in which
metaphor and metonymy occur consecutively, cases in which they occur in
parallel, and cases in which they occur interchangeably. In the following
sections, each of these cases is presented separately. (More, and more intri-
cate, examples of the interaction between metaphor and metonymy in ex-
pressions may be found in Gevaert 1994 and Feyaerts 1997.)
3.2. Consecutive interaction of metaphor and metonymy
A consecutive interaction between metaphor and metonymy occurs when one
of the motivational links in the semantics of the composite expression in-
volves a sequence of two semantic extensions. A first example is presented
in Figure 2, which contains an analysis of the compound schapenkop. Liter-
ally, the word means 'sheep's head' (and the word could actually be used in
this sense, in contrast with some of the other compounds that we will analyze
presently, in which the literal reading is not conventionalized). The derived
reading of schapenkop is 'dumb person', and this reading seems to involve
two steps: first, 'sheep's head' is metaphorically extended towards the read-
ing 'a (human) head like that of a sheep, a stupid head', and second, a meto-
nymical step leads to 'a person with a head like that of a sheep, a stupid
person'. (The representation
9
in Figure 2 can be completed on the bottom
level of the prism, but that is a step that will be taken in section 3.3.)
METONYMY
--8---
METAPHOR
1 Sheep's head
2 Sheep
3 Head
4 (Human) head like that of a
sheep
5 Stupid person
Figure 2. The prismatic structure of schapenkop
A similar consecutive combination occurs in an idiomatic expression
such as groen achter de oren zien. Literally, the reading is 'to be green be-
hind the ears', which is then metaphorically interpreted as 'to be young, (as
if people are like fruit that have a green color in the first stage of their exis-
tence, before they reach maturity). As with so many other expressions indi-
cating young age, the expression next receives a further extension to the
reading 'inexperienced, naive'.
We may also note that the consecutive sequence need not always involve
an alternation of metaphor and metonymy. In an example like hanglip, for
instance, two consecutive metonymical steps may be identified. The literal
reading is composed of the noun lip 'lip', and the verbal stem hang 'to
hang'; the literal reading can therefore be paraphrased as 'hanging lip'. A
first metonymical extension (involving the metonymical relationship between
a specific feature and the bearer of that feature) leads to 'a person with a
hanging or protruding (lower) lip'. A second metonymical extension (involv-
ing the metonymical relationship between a typical effect and the usual cause
of that effect) leads to 'an unhappy, sulky, pouting person' .
3.3. Parallel presence of metaphor and metonymy
A parallel presence between metaphor and metonymy occurs when there is a
difference in type among the different motivational links that occur in the
semantics of a composite expression. If, for instance, the motivational link at
the top level of the prismatic structure is metaphorical and one of the links at
the lower level is a metonymical one, metaphor and metonymy work in par-
allel to produce the derived reading of the expression, or at least, both a
metaphorical and a metonymical path could be reconstructed to arrive at the
derived reading.
1 Sheep's head
4 Sheep-like
7 Stupid
2 Sheep
5 Head
8 Person
3 Head
6 (Human) head like a sheep
9 Stupid person
Figure 3. The prismatic structure of schapenkop completed
As an initial step, let us complete the picture for schapenkop. At the bot-
tom level of the prismatic structure, the intermediate reading 'a (human)
head like that of a sheep, a stupid head' can be considered isomorphic: the
kop-part does not undergo a semantic change, and the 'sheep-like' reading is
related by metaphorical similarity to the original reading of schaap 'sheep'.
The ultimate reading 'stupid person' maintains the isomorphism: there is a
metaphorical link from 'sheep-like' to 'stupid', and there is a conventional
part/whole-metonymy linking 'head' to 'person'. (For instance, de koppen
The interaction a/metaphor and metonymy 215
tellen 'to count the heads' is a conventional expression for 'counting the
individuals in a group'.)
The example shows that the motivating links at the bottom level of the
prismatic structure need not be the same as the ones at the top level: the
identity link is absent at the upper level. In some cases, then, we get an alter-
nation between metaphorical and metonymical links. A case in point is the
expression in de stront zitten, which may be analyzed as in Figure 4.
1To sit in the shit
4 To be in great trouble
2 To sit (in)
5 To be situated (in)
3 Shit
6 Trouble, unpleasantness
Figure -I. The prismatic structure of in de stront zitten
The top level shift is a metonymical one of the cause/effect-type: if you
are literally surrounded by excrements, you are typically in an unpleasant,
troublesome situation. At the bottom level, however, the motivational links
are of a metaphorical type. Zitten 'to sit' is a conventional metaphor for 'to
be characterized by, to experience': dat zit goed 'that sits well' means as
much as 'that is okay', and in moeilijkheden zitten 'to sit in difficulties'
equals 'to have, to experience difficulties'. Stront is likewise a conventional
expression for anything extremely nasty.
As a slightly more complicated example, let us consider droogkloot 'bor-
ing person, bore', which can be analyzed as in Figure 5. The compositional
literal reading 'dry testicle' is the basis for a roodhuid-type extension, yield-
ing the possessive compound 'person with dry testicles'. This reading, how-
ever, is itself the input for a further metaphorical extension, leading to the
'boring person' sense. At the same time, the derived reading is isomorphic:
the aspect 'boring' correlates with the 'adjective droog, and the aspect 'man,
person' correlates with the noun kloot. Such an isomorphic analysis is fur-
ther supported by the motivational link between the initial and the derived
readings of the constituent parts. Droog is in fact conventionalized in the
reading 'boring, dull, dreary' (een droge klaas is 'a tedious fellow', where
klaas is originally a proper name), and kloot is a conventional derogatory
expression for 'man'.
1 Dry testicle
4 Person with dry testicles
7 Unworthy man
2 Dry
5 Boring person
3 Testicle
6 Lifeless, dull
Figure 5. The prismatic structure of droogkloot
3.4. Interchangeability of metaphor/metonymy analyses
The prismatic model
10
implies that the meaning of composite expressions
may sometimes be construed in two different ways: from bottom to top and
then from front to back at top level (which is undoubtedly the standard
pathway) or conversely from front to back at bottom level, and then from
bottom to top. The 'non-uniqueness of semantic solutions' (to borrow the
words introduced by Nunberg 1979 in a slightly different context) goes even
further, if we can show that different prismatic analyses may - with equal or
near-equal plausibility - be construed for one and the same expression. To
the extent that such alternatives involve different configurations of meta-
phorical and metonymical motivational links, we shall say that we have in-
terchangeable metaphor/metonymy.
Compounds pointing in this direction are the following. Badmuts literally
means 'swimming cap' but is also used jocularly for a bald person. The shift
can be analyzed in two ways. Either 'swimming cap' leads metonymically to
'a person with a swimming cap' and from there by metaphorical similarity to
'a person who looks as ifhe was wearing a swimming cap, a bald person'.
Or 'swimming cap' is directly metaphorized as 'a head that looks as if it is
covered by a swimming cap, a bald head' and from there metonymically to
'a bald-headed person'. The reconstruction of the semantic process can go
either way, and there is no principled way to favor one analysis over the
other.
An example with an idiom rather than a compound is over de rooie gaan
'to go over the red one', in which 'the red one' is a red mark on a gauge
indicating the point of maximal pressure. The derived reading 'to explode
with anger' may be arrived at in either of the following ways. To go beyond
the point of maximal pressure is the metonymical cause of a material, literal
explosion (as when a boiler explodes), and the literal explosion can then be
metaphorically used to conceptualize an emotional outburst. Conversely, the
notion of crossing the point of maximal pressure may be directly meta-
phorized into the psychological domain (' go beyond the point of maximal
emotional strain'), and this event may then causally (i.e. metonymically) lead
to an outburst.
It should be clear that the alternatives need not always involve different
sequences of metaphors and metonymies. It may also be the case, for in-
stance, that two metonymies occur in alternative orders. Zultkop provides us
with an example. Literally, it means 'head filled with or made from brawn';
the derived reading is again 'stupid person'. The consecutive steps could be
from the literal reading to 'stupid head' via a metonymicallink (the presence
of brawn rather than brain is the cause of the idiocy), and from there to 'stu-
pid person' via another metonymical link of the part/whole-type. Or the se-
quence might involve an initial part/whole-metonymy producing 'a person
with a head full of brawn' and hence to 'stupid person' through the interme-
diary of the effect/cause-metonymy.
4. Metaphtonymy and prismatic semantics
Within the context of recent metonymy studies, there is an obvious link be-
tween the phenomena described above and the notion of metaphtonymy in-
troduced by Louis Goossens in 1990. The two types of metaphtonymy (i.e.
interaction between metaphor and metonymy) distinguished by Goossens
resemble the two basic interaction types identified above: what Goossens
calls metaphor from metonymy refers to a sequential operation of the two
mechanisms that can be linked to the consecutive type described in section
3.2, and what Goossens calls metonymy within metaphor/ metaphor within
metonymy can be associated with the simultaneous, parallel type of interac-
tion described in section 3.3 of the present article. But how far exactly does
the correspondence go?
Goossens introduces metaphor from metonymy by referring to the
polysemy of giggle. The verb initially means 'to laugh in a nervous way',
but this meaning can be used metonymically in a context like 'Oh dear', she
giggled, 'I'd quite forgotten', in which giggle comes to mean 'say while
giggling'. A further extension towards 'to say as if giggling' then constitutes
the 'metaphor from metonymy' reading. Whereas the consecutive operation
of a metonymical and a metaphorical shift links up with the cases discussed
in section 3.2, it will also be clear that the approach in the present paper has
a wider scope than Goossens's. We have identified not just successions of
metonymies followed by metaphors, but we have illustrated a larger variety
of sequences: metaphors followed by metonymies, metonymies followed by
metonymies, etc. From a broader point of view, it is important to realize that
neither our 'consecutive interaction of metaphor and metonymy' nor
Goossens's 'metaphor from metonymy' can be considered real innovations
in the context of lexical semantics. The recognition that mechanisms of se-
mantic extension such as metaphor and metonymy may operate in succession
(and in fact, in series with multiple steps) is a natural and time-honored one
in diachronic semantics (cp. Geeraerts 1997a). What is being added to that
idea in the prismatic model described above, is precisely the importance of a
second dimension for an adequate description of composite expressions.
As to Goossens's 'metonymy within metaphor', it involves cases like
catch someone's ear 'ensure someone's attention'. Such examples (which
invariably involve idiomatic expressions rather than single lexemes) receive
a straightforward interpretation in the context of the model sketched in the
present paper, as can be gathered from the analysis in Figure 6.
1 Take hold of someone's organ of hearing
2 Take hold of, capture
3 Organ of hearing
4 Force to listen, obtain the attention
5 Attain, obtain
6 Attention
Figure 6. The prismatic structure of to catch someone's ear
The literal meaning 'take hold of someone's organ of hearing' is meto-
nymically extended to 'to obtain someone's attention'. Goossens basically
sees a metaphorical shift at the level of the expression as a whole, but at the
same time allows for a metonymic interpretation. In the context of the pris-
matic model, the metonymic interpretation would seem to be more plausible:
materially taking hold of someone's ear is metonymically conceptualized as
a cause (or at least, a contributing factor) for getting someone's attention. At
the same time, there is indeed a metaphorical aspect to the expression, but it
involves the development of catch at the bottom level of the two-dimensional
structure: the verb undergoes a metaphorical shift from a material to an im-
material reading. To be sure, non-uniqueness surfaces again, to the extent
that it could also be said that obtaining something is the result of taking hold
of it (and of course, the relationship between action and result, or cause and
effect, is a metonymical one). Ear, finally, is metonymically linked to the
notion of attention: the hearing organ is one of the media for channeling a
person's attention.
In the light of this analysis, the advantage of the prismatic model can be
defined as follows: it draws the attention to the fact that the more specific
semantic development is not restricted to one of the constituent parts (as
might be suggested by Goossens's examples) but actually has to be deter-
mined for all of them.
At least for composite expressions, then, the prismatic model appears to
have a wider scope than the notions introduced by Goossens, specifically
because it allows for other sequences than just metaphor from metonymy.
More importantly, the model combines the intuitions behind 'metaphor from
metonymy' and 'metonymy within metaphor' / 'metaphor within metonymy'
by bringing the two relevant aspects of the development of composite ex-
pressions together: on the one hand, the semantic development of the expres-
sion as a whole, on the other, the role of the constituent parts of the expres-
sion and their independent development. The model thus allows for a
uniform and more detailed description of the semantics of composite expres-
sIOns.
Notes
1. In the Saussurean tradition, syntagmatic relations involve associations be-
tween linguistic expressions that exist in presentia, whereas paradigmatic re-
lations involve associations that exist in absentia. In an expression like
dames en heren, for instance, the association between dames and heren is re-
alized in the expression dames en heren itself. The semantic association be-
tween dames 'ladies' and vrouwen 'women', on the other hand, exists even if
it does not show up explicitly in the expression being used as such. Paradig-
matic relations may be of various sorts; they do not just include semantic as-
sociations of the type just mentioned, but also morphological relations be-
tween a lexical base and the derivates or compounds in which it features.
Among the semantic paradigmatic associations, metaphor traditionally fea-
tures prominently (see among others Jakobson 1971: 74); note that in this
case, the association exists not beween to different words, but between two
readings of the same word. The paradigmatic relations that will be envisaged
in this paper are precisely of the kind illustrated by metaphor: semantic asso-
ciations between different readings of one linguistic expression. Apart from
metaphor, the relevant associations involve semantic relations like meton-
ymy, generalization, and specialization.
2. The bottom-up interpretation is present in quotations like the following:
'Whatever linguistic meaning is, there must be some sort of compositional
account of the interpretation of complex expressions as composed from the
interpretations of their parts and thus ultimately from the interpretations of
the (finitely many) simple expressions contained in them and of the syntactic
structures in which they occur' (Chierchia and McConnell-Ginet 1990: 6).
3. In most work in the tradition of Cognitive Semantics, the concept of motiva-
tion is used in a slightly broader way than the way in which it is defined
here. In Lakoff (1987) and related work, for instance, 'motivation' involves
the principles that explain (or make plausible) why a particular linguistic ex-
pression means what it does. The concept is explicitly introduced as an alter-
native to the more traditional notion of predictability: even if meanings are
not entirely predictable, they may be motivated by existing tendencies and
schemata. The distinction that is drawn here between 'motivation' and 'iso-
morphism' tries to be more specific about the general concept of motivation
by distinguishing between its syntagmatic and its paradigmatic form. The
concept of 'isomorphism', on the other hand, links up with existing work
within the Cognitive tradition relating to the iconicity of grammar (see e.g.
Haiman 1980b). Isomorphism as used here is a form of iconicity to the extent
that features of meaning (in particular, its complex nature) are reflected by
features of the linguistic form (viz., its composite nature).
4. The question might be asked how we can put koe as interpreted in the figura-
tive context into correspondence at all with koe as interpreted in the literal
plane. On the one hand, we would still be willing to say that koe 'problem'
corresponds with koe 'cow'. But on the other hand, there is no associative
semantic link from 'cow' to 'problem'. So can we say at all that koe maps
onto 'problem'? What is there to stop us from mapping bi} de horens vatten
onto the 'problem' part of the idiomatic reading of the expression? In this
particular instance, of course, the link between vatten 'to seize (literally)' and
the figurative reading 'to tackle' is not unmotivated, and hence, byelimina-
tion. koe is easily mapped onto 'problem'. But even if this paradigmatic link
between both interpretations of vatten were to be just as non-transparant as
that between 'cow' and 'problem', the syntactic structure of the expression
(as interpreted literally) would favor a figurative interpretation of koe as a
noun, and one of vatten as a verb. This would seem to lead to the conclusion
that there is always some paradigmatic link at the bottom of the prism be-
tween the literal readings of the constituent items and their figurative inter-
pretation: at the very least, the literal reading would motivate the figurative
reading because the latter is consistent with the word class of the former. In
principle, such a weak form of motivation can be accounted for by accepting
degrees of motivation: it will be made clear further on in the text that this is a
useful step to take in any case. Empirically speaking, however, it remains to
be seen whether figurative readings are always consistent with the word class
of the constituent elements of the literal expression. (Notice that at least in
the realm of morphology, reinterpretative processes may violate the initial
syntagmatic structure: the isomorphically metanalytic reanalysis of ham-
burger as ham + burger violates the initial formal structure.)
5. The representation in Figure 1 might be adapted to take into account degrees
of motivation by drawing thicker or thinner lines, or by attaching a numeric
weight to them.
6. The reinterpretation processes mentioned in this paragraph constitute one
kind of proof of the cognitive reality of the semantic structure embodied in
the prismatic model of Figure 1. Other kinds of support for the validity of the
model will have to be explored in further research. Two main alternatives
have to be envisaged. First, psychological investigations (involving on-line
processing tasks, or on the basis of questionnaires) may be invoked to estab-
lish the psychological reality of a specific analysis. Second, synchronic lin-
guistic phenomena (rather than the diachronic reinterpretation processes
mentioned here) may point to the structural importance of the model. Con-
sider, for instance, the possibility of incorporating anaphoric elements in the
expression. As a working hypothesis, it would seem that only those idioms
that are isomorphic on the figurative level allow for the introduction of ana-
phoric demonstrative pronouns referring to a previous instantiation of the
figuratively interpreted concepts. In this way, it is quite plausible to have a
sequence like the following: 'Then came the problem of formatting the text
according to the style sheet. To take this bull by the horns appeared to be
much more difficult than applying the revisions required by the editors'. Be-
cause bull maps onto 'the problem' that is to be tackled, this may be intro-
duced to refer to a previous identification of that problem. In the case of met
de handen in het haar zitten 'to sit with one's hands in one's hair> to be at
one's wit's end', however, the absence of a clear interpretation for handen in
the idiomatic context makes sequences like the following: 'Toen moesten er
camera-ready kopieen van de figuren gemaakt worden. Met deze handen in
het haar te zitten bleek veel erger dan het schrijven van het oorspronkelijke
artikel geweest was (Then came the problem of producing camera-ready figu-
res. To sit with these hands in one's hair appeared to be much more taxing
than writing the original paper had been)' rather implausible. This is, to be
sure, just an example of the type of phenomena to be studied, but it illustrates
how additional evidence for the linguistic reality of the prismatic model may
be sought.
7. An example of a reinterpretation of compounds is provided by the element
scharrel-, the verbal stem of scharrelen 'rummage about, scratch, scrape'.
From the compound scharrelkip 'free-ranging chicken' it is extrapolated to
compounds like scharrelei 'an egg ofa free-ranging chicken': scharrel- is re-
interpreted as 'produced by biological farming'.
8. There might be some discussion with regard to this specific case: it could
perhaps also be considered a generalization. Examples of competing and in-
terchangeable motivations will be discussed explicitly further on in the arti-
cle.
9. In Geeraerts and Bakema (1993), the prismatic representation of compounds
looks slightly more complicated than in the representation used here. For in-
stance, the compositional reading 'sheep's head' would be analyzed as a spe-
cialization of a more general reading 'head having something to do with
sheep'. The motivation behind this approach is the underdetennination of
compounds. Schapenkop could in fact mean many things: a sheep's head, but
also 'a head with a talent for or a specific interest in sheep' (just like a
studiekop is 'a bright head, a head with a talent for of a specific interest in
study'). The construction of nominal compounds in Dutch does not formally
differentiate between the possessive reading and the alternatives; by contrast,
the syntactical construction of idiomatic expressions is much more specific as
to the semantic role of the constituent parts. This underdetermination of a
compound XY can be expressed by merely defining the initial compositional
reading as 'a Y that has something or other to do with an X'. The next step is
then invariably a semantic specialization yielding specific readings like 'a
sheep's head' or 'a head with a talent for or a specific interest in sheep'. In
the present article, this complication at the front end of the prismatic dia-
grams for compounds has not been included, because it does not add very
much to the line of thought that is relevant in this text.
10. It may be useful to point out that the 'prismatic model' is a model precisely
because it involves a certain degree of abstraction: the semantics of the com-
posite expressions is rendered in a schematic way (in the sense, for instance,
that the number of constituent elements is systematically reduced to two).
Section 4
Meaning and culture
Chapter 9
Looking back at anger. Cultural traditions and
metaphorical patterns
Originally published as Dirk Geeraerts and Stefan Grondelaers, 1995, in John
Taylor and Robert E. MacLaury (eds.), Language and the Construal o[the World
153-180. BerlinJNew York: Mouton de Gruyter.
This paper addresses the question of the influence of the Medieval doctrine of the
four humors and the four temperaments on our contemporary vocabulary. Specifi-
cally, given the psychological part of the humoral theory, is there any way in
which the influence of the theory can be felt in the way emotions are being talked
about today? Concentrating on the concept of anger, the paper shows that there is.
By taking a closer historical look at Kovecses' analysis of expressions for the
emotional concept 'anger' (1989), it is argued that his a-historical method ob-
scures the possible role of cultural traditions as a source of emotion concepts.
In the context of the development of Cognitive Linguistics, the paper helped to
reveal the possible tension between a universalist and a culture-specific interpreta-
tion of the 'experientialist' nature of Cognitive Linguistics. Cognitive Linguistics,
by its very cognitive nature, has a tendency to look at language from a psychologi-
cal point of view, i.e., at language as (part of) the organization of knowledge in
the individual mind. But if cognition is embodied, and if our bodies are basically
the same. then underlying cognitive patterns will be largely the same for all. In
Kovecses' analysis of anger metaphors in terms of the ANGER IS HEAT pattern, the
universalist assumption is that the pattern is physiologically motivated.
However, if the experientialist nature of language and cognition includes the cul-
tural and historical 'situatedness' of human experience, cultural and historical
factors are likely to influence our cognitive patterns. That is precisely what the
present paper tries to show.
In more recent publications, Kovecses (2000. 2002) accepts the point that cultural
determination may play a role in shaping cognitive patterns, and investigating the
universality (or the cultural specificity) of metaphorical patterns has now become
a standard topic in metaphor oriented cognitive linguistic research. Such studies
focusing on anger include Taylor and Mbense (1998), Yu (1998), Mikolajczuk
(1998), Forceville (1999), Barcelona (2001), Gevaert (2001), Harkins and Wierz-
bicka (2001), Soriano (2003), and Maalej (2004).
More generally, there is a growing interest in Cognitive Linguistics in the social
aspects of meaning and cognition. A number of researchers (Palmer 1996, Sinha
and Jensen de L6pez 2000, Harder 2003. Itkonen 2003. Tomasello 2003. Geer-
228 Aleaning and culture
aerts 2005 and others: see Kristiansen and DilVen, forthcoming) emphasize that
the experientialist nature of Cognitive Linguistics does not only refer to material
factors (taking a notion like 'embodiment' in a physical and physiological sense)
but that the cultural environment and the socially interactive nature of language
should be recognized as primary elements of a cognitive approach. Further, laying
the foundations for a true cognitive sociolinguistics, variational phenomena are
being studied empirically in work such as Kristiansen on phonetic variation
(2003), Berthele (2004) on differences in syntactic construal between dialects, and
Grondelaers (Grondelaers 2000) on grammatical phenomena whose distribution is
determined by a combination of internal (structural or semantic) and external
(contextual or sociolinguistic) factors. (See also the work mentioned in the intro-
duction to Chapter 11.)
1. The mysteries of masturbation
In the course of 1989, the Belgian Department of Education started a school
campaign against truancy. The major slogan of the campaign read Van spij-
belen word je doof 'Playing truant makes you deaf - a jocular (but probably
inefficient) reference to the old belief that excessive masturbation could
cause deafness. It is not likely that this belief itself is still very much alive in
our post-sexual revolution, sex education era, but the very fact that knowl-
edge of it was assumed in the campaign seems to indicate that it is still
around, and that it was being handed down from educators to pupils not too
long ago. But what was the origin of that belief (which, incidentally, came in
a number of variants, in the sense that next to deafness, blindness and dete-
rioration of the spinal marrow were cited as the sinful results of promiscuous
self-indulgence)? Was it just a conspiratorial invention of priests and par-
ents, intended to keep personal frustration and public morality up? Or was
there an actual basis for it? Let us turn to a specialist for an answer.
In 1772, the honorable doctor Tissot (member of learned societies in
London, Basel, Bern, and Rotterdam) published a lengthy treatise entitled
L 'Onanisme. Dissertation sur les Maladies produites par la Masturbation
(Grasset, Lausanne), in which all is revealed about this 'crime obscene'.
After an extensive treatment of the detrimental influence of masturbation, he
asks the question 'Col11l11ent une trop grande emission de semence produit-
elle tous les maux que je viens de decrire?' [How does an excessive emission
of sperm produce all the evils that I have just described?]. On page 69, he
begins his answer with a reference to the father of medicine, Hippocrates of
Looking back at anger 229
Kos (approx. 460-377 BC). (The relevant passages are in the Hippocratic
treatise known as De Genitura.)
Hippocrate a cru qu'elle [la semence] se separoit de tout le corps, mais sur-
tout de la tete. La semence de l'homme vient, dit-il, de toutes les humeurs
de son corps, elle en est la partie la plus importante.... Il Ya des veines &
des nerfs qui de toutes les parties du corps vont se rendre aux parties geni-
tales; quand celles-ci se trouvent remplies & echauffees, elles eprouvent un
prurit, qui se communiquant dant tout le corps, y porte une impression de
chaleur & de plaisir; les humeurs entrent dans une espece de fermentation,
qui en separe ce qu'il y a de plus precieux & de plus balsamique, & cette
partie, ainsi separee du reste, est portee par la moele de l'epine aux organes
genitaux.
[Hippocrates thought that semen secreted itself from the entire body, but
specifically from the head. A man's semen, he says, comes from all the
humors of his body, of which it is the most important component. ... There
are veins and nerves that go towards the genital organs from all over the
body, and when these organs are filled up and wanned up, they experience
an urge that communicates itself through the entire body, producing an
impression of warmth and pleasure. The humors then enter into some sort
of fermentation that separates out the most precious and balsamic sub-
stance they contain, and this part, when it is separated from the rest, is car-
ried to the genital organs by the spinal marrow.]
If, in other words, semen is produced by a process of fermentation and
distillation of the 'humors', it is plausible that an overproduction of sperm
weakens the body, given the vital importance of those 'humors'. Specifi-
cally, the role of the head and the spinal cord in this process explains why, in
particular, the functions of the head (such as seeing and hearing), and the
spinal marrow may suffer the detrimental effects of excessive sperm produc-
tion. In this sense, it all falls it into place - but it only does so against the
background of the doctrine of the four humors. In Classical and Medieval
physiology, the human body was thought to contain four kinds of fluids, or
humors, which regulated the body's functioning and whose disproportionate
presence could cause illness. Admittedly, conflicting views were expressed
within the humoral framework about the origins of sperm, and the humoral
doctrine as a whole was no longer valid in its original Hippocratic form by
the end of the eighteenth century. But, although Tissot hardly follows Hip-
pocrates in detail, he quotes him approvingly, and like most of his contempo-
raries does retain the basic idea that the production of semen (and the harm-
230 Meaning and culture
ful effects of overproduction) involves the extraction of the seminal sub-
stance from the vital bodily humors.
Against the background of this historical link, extending from classical
antiquity well into the modem era, the lingering belief in the negative effects
of masturbation appears to be a recently deceased (or at least moribund)
renmant of what was once solid science. And to be sure, it is not the only
relic. In the Dutch-speaking part of Belgium, for instance, the concept 'cold,
inflammation of the mucous membrane of nose and throat' is expressed in a
majority of the dialects by the word valling. While valling is morphologi-
cally complex (being a nominalization of the verb vallen 'to fall '), it is not
semantically transparent to the majority of speakers. Historically speaking,
however, its formal complexity makes perfect sense in the framework of the
theory of humors: when the nose runs or sputum is expectorated, what 'falls'
is nothing else than phlegm, one of the four humors. Given that the head was
considered to be the major locus of phlegm (and taking into account that
phlegm was described as a cold humor, in contrast with, for instance, blood
as one of the warm humors), a valling or a cold is nothing but a precipita-
tion of the brain's fluid. And of course, in non-historical parlance, the Eng-
lish word phlegm now refers precisely to the thick semi-fluid secretion of the
mucous membranes of the respiratory passages. Similarly, we speak of ca-
tarrh (derived from the Greek katarrheo 'to flow down'), and - in the case
of another disease attributed to an excess of phlegm - of rheumatism, in
which the Greek verb rheo 'to flow' can be discerned (cp. Siegell968: 323).
Faced with examples such as these, we would like to address the question
of the influence of the humoral doctrine on our contemporary vocabulary
more systematically. What other relics of the old beliefs can we find? Spe-
cifically, given the psychological part of the humoral theory, is there any
way in which the influence of the theory can still be felt in the way we talk
about emotions? Concentrating on the concept of anger, we will try to show
that there is. By taking a closer historical look at Kovecses' analysis of emo-
tional expressions in terms of generalized metaphors (1989), we shall argue
that his a-historical method obscures the possible role of cultural traditions
as a source of emotion concepts. Our purpose, in other words, is factual to
the extent that we will try to establish the importance of the old humoral
theory for our contemporary emotional vocabulary, critical to the extent that
we will try to qualifY Kovecses' analysis, and methodological to the extent
that we will stress the methodological importance of a diachronic perspective
for linguistic studies with a cultural orientation. Before dealing with the
specific linguistic part of the investigation, we will first give some more
information on the theory of humors and its historical importance.
2. The history of the humors
In this section, we will briefly (and simplifyingly) present the humoral doc-
trine, and sketch its historical development. On various aspects of the history
of medicine at large and the humoral theory in particular, more information
may be found in among others Lindeboom (1985), Godderis (1988), Beek
(1969), Irwin (1947), Siegel (1968), Major (1954), Schafer (1966), and
Diepgen (1955). Klibansky, Panofsky and Saxl (1964) deserve to be men-
tioned separately for their detailed history of the humoral doctrine up to the
seventeenth century (with special emphasis on the concept of melancholy).
The foundations of the humoral doctrine were laid by Hippocrates of
Kos. Three aspects of his approach should be mentioned: the physiological,
the psychological, and the medical. Physiologically, the four humoral fluids
regulate the vital processes within the human body; the secretion of the hu-
mors underlies the dynamical operation of our anatomy. Psychologically, on
the other hand, they define four prototypical temperaments, i.e. a person's
character is thought to be determined by the preponderance of one of the
four vital fluids in his body. Thus, the choleric temperament (given to anger
and irascibility) is determined by a preponderance of the yellow bile, while
the melancholic, gloomy and fearful, suffers from a constitutional excess of
black bile. The phlegmatic personality is typically placid and unmoved,
while the sanguine temperament (defined in correlation with blood, the
fourth humor) is passionate, optimistic, and brave. The singular combination
of physiological and psychological concepts that characterizes the theory of
humors also shows up in the fact that a disequilibrium of the fluids does not
only characterize constitutional temperaments, but also causes temporary
diseases - which are then typically described in bodily, biological terms as
well as in psychic terms. For instance, an overproduction of yellow bile may
be signaled by the patient's vomiting bile, but also by his dreaming of fire.
In the same line, an excess of blood shows up in the redness of the skin and
swollen veins, but also in carelessness and a certain degree of recalcitrance.
In this sense, the humoral theory is a medical doctrine: it identifies diseases
and their symptoms, and defines a therapy. Obviously, the basic therapeutic
rule will be to restore the balance of the humors, given that a disturbance of
their well-balanced proportion is the basic cause of the pathological situa-
tion. The long-lasting popularity of blood-letting, for instance (a standard
medical practice that continued well into the nineteenth century) has its his-
torical origins in the theory ofhumors.
The connection between yellow bile and fire that was mentioned a mo-
ment ago is not accidental. It is part of a systematic correlation between the
human, anatomical microcosm and the macrocosm, thought to be built up
from four basic elements. Thus, yellow bile, black bile, phlegm, and blood
corresponded with fire, earth, water, and air respectively. In the Aristotelian
elaboration of the Hippocratic doctrine, a 'componential analysis' was added
to these correlating sets of microcosmical and macrocosmical basic ele-
ments. They were defined, in fact, as combinations of four basic features:
cold, warm, wet, and dry. (Needless to say, these four features are them-
selves related along two dimensions.) Blood was thought to be warm and
wet, phlegm cold and wet, yellow bile warm and dry, and black bile cold and
dry.
The classical humoral doctrine received the form in which it was to
dominate the Middle Ages in the work of Galen (129-199). His incorpora-
tion of the humoral approach into an encompassing theory of the human
digestive system is of particular interest. Galen distinguishes between three
successive 'digestions'. In the first digestive process, food is transformed
into chyle in the stomach; the residue of this first digestion is faeces. In the
second step, the humoral fluids enter the picture. For instance, by the trans-
formation of chyle in the spleen, black bile is produced, while the liver re-
fines chyle into blood; the residue of the second digestion is urine. The third
step takes the blood and carries it through the body, sustaining the growth of
the body; the residue of this third digestion is perspiration. But while the
substance that ensures the growth and maintenance of the body is known as
the nutrimental spirit, there are also two other spirits to be taken into ac-
count in this third step. In a continuing and cumulative refining process, the
heart produces the vital spirit (which regulates the temperature of the body
and controls the passions), and the brain produces the animal spirit (Aris-
totle's pneuma psychicon, which commands the movement of the body, but
also feeling and the workings of the mind).
Further, Galen's digestive anatomy leads to a dietary pharmacology. All
plants (and foodstuffs in general) could be characterized by one of four de-
grees of warmth, cold, wetness, and dryness. Given that diseases are caused
by an excess of one of the four humors, and given that these are themselves
characterized by the four features just mentioned, the basic therapeutic rule
is to put the patient on a diet that will ensure a decrease of the superfluous
humor. For instance, because yellow bile is hot and dry, patients suffering
from choleric diseases should avoid plants such as garlic and ginger, which
are both warm in the fourth degree and dry in the fourth degree. Rather, they
should resort to plants such as opium (wet in the fourth degree) and henbane
(cold in the fourth degree).
Table 1. A system of humoral correspondences
PHLEGM BLACK BILE YELLOW BILE BLOOD
CHARACTERISTIC cold & moist cold & dry warm & dry warm & moist
ELEMENT water earth fire air
TEMPERAMENT phlegmatic melancholic choleric sanguine
ORGAN brain/bladder spleen liver/stomach heart
COLOR white black yellow red
TASTE salty sour bitter sweet
SEASON winter autumn summer spring
WIND North West South East
PLANET moon Saturn Mars Jupiter
ANIMAL turtle sparrow lion goat
In the course of the Middle Ages, the Galenic framework was further de-
veloped into a large-scale system of signs and symbols. In a typically medie-
val analogical way of thinking, widely divergent phenomena (ranging from
the ages of man to astrological notions such as the system of the planets and
the signs of the zodiac) were fitted into the fourfold schema presented by the
medical theory. In Table I, an overview is given ofa number of those corre-
lations. It should be mentioned, however, that the system was not entirely
without unclarities (which is not surprising for a system that was to a large
extent devised independently of empirical observation). For instance, while
there was general agreement on the core of the system, authors would differ
as to the more peripheral elements (such as the question which planet corre-
lates with which humor; in particular, the associated animals are highly un-
stable across authors). Also, the system so to speak contained its own
sources of confusion. There is, for instance, a marked ambiguity in the use
of the concept 'blood', which was not only considered to be one of the four
basic fluids, but which was also thought to transport the other humors, and
which could hence also be used to refer to the mixture of the humors that
was carried through the body (cp. Schiifer 1966: 4). And while the basic
color associated with yellow bile is obviously yellow, it was believed that
yellow bile turned red when heated (and black when it was entirely burned
up), so that the color 'red' could receive multiple interpretations within the
system.
The humoral edifice began to be undermined as soon as the Renaissance
introduced renewed empirical medical investigations. Harvey's discovery of
the circulation of the blood, for instance, was in direct contradiction with the
traditional position of the blood in the Galenic 'digestive' system. However,
the disappearance of the theory from the medical scene was only very grad-
ual, and it took approximately another three centuries before the last vestiges
of the humoral framework were finally removed. The standard view of the
historians of medicine is, in fact, that only in the middle of the nineteenth
century (and more particularly, with the publication of Rudolf Virchow's
Die Cellularpathologie of 1858) did the humoral pathological conception
receive its final blow. This 'final' character only holds, of course, for the
official medical science: we have already seen in the introductory section
about masturbation that traces of the old doctrine continued to exist much
longer in popular belief. Along the same lines, it could probably be shown
that the contemporary revival of herbalist medicine at the fringe of official
medicine has direct links with the Galenic dietary pharmacology.
3. Anger in art
As we have seen, the humoral doctrine had developed into a full-fledged
semiotic system in the course of the Middle Ages: an ordered set of signs for
medical and psychological interpretation. As a first indication of the fact that
this semiotic system was not confined to the field of medicine, let us see how
it influenced the artistic production of the Renaissance. We shall give two
examples, one from the pictorial arts, and one from the dramatic arts.
Cesare Ripa's Iconologia of 1593 was undoubtedly one of the major ref-
erence works for the seventeenth-century graphic artist. It contained a the-
matic inventory of the emblematic subject-matter of art, that is to say, of the
topics, motifs, and symbols that could be used in paintings, drawings, en-
gravings, and the like. There is a separate section in Ripa devoted to the four
temperaments, with a detailed enumeration of the iconography associated
with each of the four types. This is how Ripa introduces the choleric tem-
perament.
Vn giouane magro di color gialliccio, & con sguardo fiero, che essendo
quasi nudo tenghi con la destra mano vna spada nuda, stando con pron-
tezza di voler combattere. Da vn lato (cioe per terra) sar. .. vno scudo in
mezo del qual sia dipinta vna gran fiamma di fuoco, & dall'altro lato vn fe-
roce leone.
[A thin young man in yellow, with a ferocious face, almost naked, holding
a drawn weapon in his right hand, ready to fight. From one side, a shield
will be placed on the ground. with a flame of fire painted in the middle,
and from the other side, a ferocious lion.]
Each of these characteristics is then further explained and elaborated in
the course of Ripa' s expose, which is interspersed with references to and
quotations from authorities such as Oalen, Ovid, Seneca, and Avicenna. In
general, the attributes mentioned in Ripa' s description can be easily related
to the characteristics mentioned in Table 1. Basically, the irascibility of the
choleric person is symbolized by depicting him as a battle-prone warrior.
Note that each of the details subtly contributes to the meaning of the whole;
in particular, the fact that the young man is naked, and the fact he is not
carrying his shield but that it is merely lying on the ground, indicate the im-
pulsiveness of his hot-tempered nature: in his fits of rage, he does not even
think about his own protection. If this impulsiveness is the negative side of
his personality, the braveness symbolized by the lion is its positive side.
Further elements that can be traced easily are the fire, and the yellow calor
(corresponding, of course, with the yellow bile that is the physiological basis
of this type). Less clear perhaps is the leanness of the young man's body, but
this is an expression of the consuming character of the dry heat that is typi-
cal of the choleric physiology.
It is worthwhile noticing that Ripa's description contains only the basic
iconography of the four temperaments. It suffices to have a look at Kli-
bansky, Panofsky and Saxl's (1964) magisterial monograph on Diirer's
well-known wood-cut Melancolia I to get an idea of the intricacies and sub-
tleties that arise when the humoral iconology is used and transformed by a
truly creative artist.
But the influence of the humoral semiotic system was not confined to the
graphic arts. For instance, it has been described by various authors (Camp-
bell 1930, Cruttwell 1951, Draper 1965, Schafer 1966, Pope 1985, Kail
1986) how the psychology of Shakespeare's dramatic characters unmistak-
enly refers to the theory of humors. Just a few quotations from The Taming
of the Shrew suffice to demonstrate this.
(1) Were I not a little pot and soon hot [IV: 1:5]
(2) Is she so hot a shrew [IV: 1: 17]
(3) I tell thee, Kate, 't was burnt and dried away,
and I expressly am forbid to touch it,
for it engenders choler, planteth anger;
and better it were that both of us did fast,
since, of ourselves, ourselves are choleric [IV: 1: 156]
(4) Gm. What say you to a neat's foot?
Kath. 'Tis passing good. I prithee let me have it.
Gm. I fear it is too choleric a meat.
How say you to a fat tripe finely broil' d?
Kath. I like it well. Good Gmmio, fetch it me.
Gm. I cannot tell. I fear 'tis choleric.
What say you to a piece of beef and mustard?
Kath. A dish that I do love to feed upon.
Gm. Ay, but the mustard is too hot a little [IV:3:25]
The conceptualization of anger in these quotations conforms to the model
furnished by the theory ofhumors: anger is caused by choler (3), the produc-
tion of which may be stimulated by certain kinds of food (3,4); while a chol-
eric temperament is a permanent personality trait (3), the main attribute of
the choleric personality is hotness (1,2). (This is not to say, by the way, that
Shakespeare's use of the humoral doctrine is unoriginal: see Pope 1985 on
the vivid and original way in which he handles the humoral concepts.) The
fact that passages such as the ones quoted above can be multiplied from the
work of Webster, Marlowe, or Jonson, leads Schafer (1966) to the conclu-
sion that the humoral conception of physiology and psychology is something
of a true fashion in Elizabethan drama. He attributes this to the fact that it is
only in the middle of the sixteenth century that the doctrine became known to
a wider audience that that of learned men who could read the medical au-
thorities in their Latin and Greek originals. It is only, in other words, after
the invention of printing that works such as Thomas Elyot's Castel of
Helthe (1539), Andrew Boorde's A Breuyary of Helth (c. 1542) and A
Compendyous Regyment or A Dyetary of Helth (c. 1542), or Thomas Vi-
cary's A Profitable Treatise o.lthe Anatomie ofMans Body (1548) could be
widely distributed, and that they could contribute to the spreading of the
humoral doctrine to the community at large. (In Chapman 1979: 277, the
wide distribution of almanacs is mentioned as a specific factor contributing
to its popularity.) But if this dissemination of the doctrine of humors from
the realm of learned knowledge to that of popular belief implies that it is
technically a piece of gesunkenes Kulturgut, the question arises how far it
actually sank. In particular. how deep did it become entrenched in the lan-
guage itself?
4. The lexical legacy
We have already seen, in the case ofvalling, catarrh, rheumatism, and even
cold, that single lexical items that are current today may be traced back to
the humoral doctrine. These items are not isolated cases. In Table 2, we have
systematically brought together a number of items and expressions in three
European languages (English, French, and Dutch) that can be considered a
part of the legacy of the theory ofhumors.
Table 2. Lexical relics of the humoral doctrine
ENGLISH FRENCH DurCH
PHLEGM phlegmatic avoir un jlegme valling
calm, cool, apa- imperturbable (dialect) cold, infec-
thetic, undisturbed to be imperturbable tion of the nose
BLACK BILE spleen melancolie zwartgallig
organ filtering the sadness, sorrow, sad, depressed (lit.
blood; sadness moroseness black-bilious)
YELLOW BILE bilious colere Z 'n gal spuwen
angry, irascible, anger, bad temper, to vent (lit. to spit
hot-tempered wrath, rage out) one's gall
BLOOD full-blooded avoir du sang dans warmbloedig
vigorous, hearty, les veines passionate (lit.
sensual to have spirit, luck warm-blooded)
It will be noticed that the items exhibit various kinds of etymological or
semantic relationships with regard to the older medical vocabulary. To begin
with, there are items like melancolie, colere, andjlegme that refer directly to
the original Latin denominations of the four basic fluids or types of person-
ality. Next, there are items such as bilious and zwartgallig, which are based
on a synonym (bile) or a translation in the vernacular (gal) of the technical
term for the humor in question. Finally, there are items that have a more
indirect relationship with the humors, in the sense of being metonymically
related with them. Thus, spleen and valling are not formed on the basis of
the name of one of the basic fluids, but rather refer, respectively, to the or-
gan typically associated with black bile (and hence to the associated tem-
perament), and to a physiological effect thought to involve phlegm. Not in-
cluded in the figure but equally revealing are items such as French humeur
'temperament; mood', that involve the generic term for the four fluids. Ifwe
zoom in on one of the cells of Table 2, still further examples may be found.
According to Roget's Thesaurus, the items listed under (5) all refer to anger
or related concepts (the glosses are our own).
(5) choler 'anger'
gall 'anger'
rouse one's choler 'to elicit anger'
stir one's bile 'to elicit anger'
galling, 'vexing, causing anger'
choleric 'irascible'
liverish 'irascible'
splenetic 'irascible'
hot-blooded 'irascible'
.fiery 'irascible'
hot-headed 'irascible'
Although we will return to methodological problems more systematically
further on in the paper, a brief methodological remark may be useful at this
point. The basis for quoting a particular item as evidence for the influence of
the humoral doctrine is the degree of etymological or semantic motivation
that may be attributed to the item in question when it is interpreted in that
historical light. Because items such as gall, liverish, choler(ic), and to stir
one 's bile would simply remain etymological puzzles if the historical medi-
cal background were not taken into account, a humoral interpretation has
explanatory value for them. But not all of the cases mentioned under (5) are
equally clear. Two kinds of more or less problematic cases can be distin-
guished.
In the first place, there are items whose global motivation in terms of the
theory of humors is plausible, but whose local motivation within the theory
is not entirely clear. Take the case of hot-blooded: if the typical fluid asso-
ciated with anger is yellow bile, how come this expression contains a refer-
ence to the warming up of blood rather than bile? However, we have already
seen that blood had a highly specific position in the whole doctrine: it is not
only a humor in itself, but also carries the other humors through the body. If,
then, blood can also refer to the mixture of the four humors as it circulates
through the body, it is not surprising that the warming up that causes anger
may be metonymically said to involve the entire mixture.
A similar but less easily explainable case of a possible local lack of moti-
vation is splenetic: although this expression supports our general point that
the influence of the humoral theory on our contemporary emotional vocabu-
lary can be demonstrated, there seems to be a contradiction in the fact that
items referring to the spleen may be either related to the melancholic tem-
perament (see Table 2) or to the choleric temperament, as in (5). Given that
the link between the spleen and melancholy is the orthodox one in the
framework of the humoral approach, there are at least two ways in which the
association between the spleen and anger could be explained. First, the asso-
ciation between the spleen and anger could be an effect of the confusion
about aspects of the humoral doctrine that we have already drawn the atten-
tion to. Second, the association could be motivated by specific subtleties and
refinements of the theory that we have not yet dealt with. In fact, Klibansky,
Panofsky and Saxl (1964: 88) draw the attention to a passage in Avicenna
where a distinction is made between the natural, primary form of melan-
choly, caused by an overproduction of black bile, and a secondary form of
melancholy caused by a combustion of one of the other humors; thus, there
is a specifically 'choleric' form of melancholy, which typically expresses
itself as a state of frenzy (compare Starobinsky 1962, Jackson 1986 for the
history of melancholy). At stake here is the notion of 'adust melancholy',
which was thought to lead to more aggressive behavior and less fearfulness
and sorrow than the natural melancholy that was engendered in a straight-
forward manner by an overabundance of black bile. (On the distinction be-
tween natural melancholy, adust melancholy, and choler - and on the confu-
sion it leads to among scholars - see Soufas' s 1990 argumentation that Don
Quixote is an adust melancholic rather than the choleric type he has been
made out to be in earlier humoral interpretations of Cervantes's work.)
Our intention here is not to choose between these alternatives, but to
make the methodological point that settling the question requires a detailed
diachronic analysis of the development of the humoral theory and of its in-
fluence on our emotional vocabulary. If it is in general clear that the histori-
cal motivation behind the meaning 'irascible' of splenetic has to be sought in
the older physiological-psychological conceptions of the theory ofhumors, a
closer historical look at the development of that theory would yield valuable
information about the specific history of splenetic. If, on the one hand, an
explanation is sought in the less central aspects of the humoral theory, the
historical analysis should be able to show how, for instance, the dissemina-
tion of Avicenna's view that was mentioned above led to the lexical associa-
tion between the spleen and anger. If, on the other hand, that association is
the result of a confusion, it is probably a later development, caused by impu-
rities in the dissemination of the theory from its learned origins to the com-
mon people (or, perhaps, by the fact that the theory became less transparent
when it gradually lost its scientific and medical respectability).
If, then, the cases that are characterized by an apparent local lack of mo-
tivation merely establish the need for more detailed historical research, the
items that raise global motivational questions are potentially more damaging
for the humoral hypothesis. Consider an example like fiery: the (metaphori-
cal) reference to heat could be attributed to the lingering influence of the
humoral doctrine, but it could also be motivated on entirely different
grounds. Suppose, in fact, that increased body heat is a physiological effect
of being in a state of anger. and that anger is metonymically conceptualized
in terms of its physiological effects. Rather than an historical motivation as a
relic of a now abandoned medical theory, an expression such as fiery would
then have an a-historical physiological motivation. At this point, we can
include another set of expressions for the concept 'anger' into the discussion.
In an analysis that has been published in several places (Kovecses 1986,
Lakoff and Kovecses 1987, Lakoff 1987, Kovecses 1989), conventionalized
phrases such as those in (6) have been subsumed by Kovecses and Lakoff
under the general metaphor ANGER IS HEAT, which is further specified into
ANGER IS THE HEAT OF A FLUID IN A CONTAINER when the heat applies to
fluids, and into ANGER IS FIRE when the heat is applied to solids. (We will
base our discussion on Kovecses 1989; there are only minimal differences in
any case between the four published versions of the analysis).
(6) I had reached the boiling point. She was seething with rage. He lost his
cool. You make my blood boil. He wasfoaming at the mouth. He'sjust
letting ofsteam. Don't get hot under the collar. B i / ~ v ' s a hothead. They
were having a heated argument. When I found out, I almost burst a
blood vessel. He got red with anger. She was scarlet with rage. I was
fuming. When I told him, he just exploded. Smoke was pouring out of
his ears. He was breathing fire. Those are inflammatory remarks. That
kindled my ire. He was consumed by his anger.
At a still lower level of analysis, these and many similar expressions are
grouped together under labels such as when the intensity of anger increases,
the fluid rises (his pent-up anger welled up inside him), intense anger pro-
duces steam (I was fUming), and when anger becomes too intense, the person
explodes (when I told him, he just exploded). Next to the basic general
metaphor ANGER IS HEAT, less elaborate metaphorical patterns such as AN-
GER IS lNSANITY, ANGER IS AN OPPONENT, ANGER IS A DANGEROUS ANI-
MAL, and CAUSING ANGER IS TRESPASSlNG are identified.
It will be obvious that the general metaphor ANGER IS THE HEAT OF A
FLUID IN A CONTAlNER neatly fits into the humoral views: the body is the
container of the four cardinal fluids, and anger involves the heating up of
specific fluids (either yellow bile as the direct source of ire, or blood as the
mixture of the four humors). However, given the alternative explanation of
the general metaphor in terms of a physiological metonymy, is there any way
in which we can say that the humoral hypothesis provides a better explana-
tion of the motivation behind this particular subset of our emotional vocabu-
lary? Apart from the general methodological point that a humoral explana-
tion achieves greater generality by being able to combine an explanation of
the cases under (5) with an explanation of those under (6), we have to con-
sider two specific reasons for preferring it over a purely physiological expla-
nation.
First, it seems better able to motivate the reference to fluids in the expres-
sions. Kovecses explains these references in the following terms: 'The fluid
version [of the basic metaphor] is much more highly elaborated. The reason
for this, we surmise, is that in our overall conceptual system we have the
general metaphor the body is a container for the emotions' (1989: 53). The
latter is illustrated by expressions such as he was filled with anger and she
could not contain her joy. However, it is not clear how this metaphor com-
bines with the basic ANGER IS HEAT metaphor to yield the application to
fluids (as Kovecses claims it does): the fact that the body is a container for
the emotions does not predispose the interpretation towards a conception of
the emotions as fluids; after all, the contained emotions could just as well be
solids or gases as far as the container metaphor is concerned. We will pres-
ently have more to say about the 'solids' version of the basic metaphor, but
it can already be remarked here that in Kovecses's view of the matter, one
would not expect the fluid version to be more elaborate than the solid ver-
sion, because the container metaphor that is invoked as an explanation does
not seem to favor the one over the other.
Second, the humoral interpretation may help us to make sense of cases
that are beyond the reach of a physiological explanation. In general, one is
tempted to argue that a physiological interpretation entails that like physio-
logical effects lead to like patterns of lexicalization. In this sense, it would be
a counterargument for the physiological approach that an emotion such as
shame, which is no less characterized by redness in the face (flushing) and a
subjective impression of increased body temperature than anger, is not lexi-
calized by the same set of expressions as anger. It would make no sense, for
instance, to say that one's blood boils with shame, or that someone is fuming
with shame. However, Kovecses has rightly pointed out that there need not
be a simple correlation between physiological effects and linguistic patterns,
and that motivation does not equal prediction (1989: 85). The physiological
effects of anger motivate our anger vocabulary, but because they do not
predict the linguistic situation, emotions with similar physiological effects
may be differently conceptualized. This element of caution does not, how-
ever, work as easily in the other direction: similar physiological effects need
not have similar lexical reflections, yet similar patterns of lexicalization had
better correlate with similar physiological effects if the physiological expla-
nation is to have any generality. To take up an example, there exists a rather
hackneyed set of expressions to the effect that love is a fire: you can let the
flame of your love die out, you can have a steadily burning devotion for
someone, and you can feel warm towards that person. On the one hand, this
accords well with the humoral belief that love is one of the 'hot' emotions.
On the other hand, it is physiologically unlikely that persons in love have a
permanently raised skin temperature (we, at least, are not aware of physio-
logical research to that effect). Granting, in other words, that the methodo-
logical key element is motivation rather than prediction, it does seem to be
the case that taking into account the historical humoral background may lead
to better motivational success.
(A related point that will only be mentioned in passing here concerns the
subjective experiential prominence of the physiological effects thought to
underlie our emotional vocabulary. Kovecses refers to the experimental re-
sults of Ekman, Levenson and Friesen (1983) to prove that anger indeed
correlates with higher skin temperature, whereas fear correlates with a de-
crease. However, independent evidence is needed to show that these objective
increases and decreases correlate with subjective experiences of warm and
cold. This question is relevant because the changes that were measured were
rather small: an increase of O. 15 degrees in the case of anger, and a decrease
of 0.01 degrees in the case of fear. Are these changes noticed at all by the
individuals concerned? Are they sufficient to cause the subjective experi-
ences that could influence our vocabulary?)
On the other hand, let us now play the devil's advocate. A possible objec-
tion against the hypothesis that the ANGER IS HEAT metaphor is a legacy of
the humoral theory could be based on those cases in which the basic meta-
phor is applied to solids rather than to body fluids: there is no reference to
solids, after all, in the original medical doctrine. Notice, however, that most
of the expressions cited by Kovecses as evidence for the 'solids' interpreta-
tion, as listed under (7), refer to fire as the source of combustion rather than
to a solid substance that is being warmed up; this is aptly rendered by
Kovecses's labeling of this metaphorical subpattern as ANGER IS FIRE. By
contrasting heat as applied to solids with heat as applied to fluids,
Kovecses's formulations suggest that in the former case, the solids have the
same function within the metaphorical image as the fluids in the latter case.
But while the fluids are the object of the process of warming up, there is no
reference to solids as things that are being warmed up in most of the expres-
sions in (7). On the contrary, we mainly find references to fire as the source
of the process of warming up. Therefore, while there would be an incom-
patibility in the images referring to fluids and to solids as the object of the
heating process, there is merely a complementarity between the expressions
referring to fluids as the object of the heating up and the expressions refer-
ring to fire as the source of the combustion process. And of course, these
complementary images dovetail with the hypothesis that the expressions
historically have humoral origins.
(7) Those are inflammatory remarks. She was doing a slow burn. He was
breathingfire. Your insincere apology just addedfuel to the fire. After
the argument, Dave was smolderingfor days. That kindled my ire. Boy,
am 1 burned up! He was consumed by his anger.
But what about the last two expressions in (7)? In the metaphorical im-
age, there is an unmistakable reference to the person's body as a solid sub-
stance being consumed. But either in the presupposition that the fire of anger
naturally takes its fuel from the body, or in the presupposition that it may
detrimentally spread to the whole body and consume substances that are not
its natural source of fuel, there is again no contradiction with the humoral
conception of anger. On the contrary, we have already seen in Ripa's de-
scription of the choleric that a consumptive burning up of the body, resulting
in leanness and thinness, is part and parcel of the original views. We see no
reason, in short, to argue that the 'solids' interpretation of the ANGER IS
HEAT metaphor endangers a humoral interpretation of the motivation behind
that metaphor.
5. Attenuating the analysis
The foregoing does not imply, to be sure, that the humoral interpretation of
our emotional vocabulary is without problems. In order to avoid misunder-
standing, we would now like to specify a number of views that we explicitly
do not intend to propagate. First and foremost, we do not think that our
analysis could not be further corroborated. Specifically, because we claim
that a sizeable portion of our contemporary anger vocabulary is part of the
lexical legacy of the theory of the four humors, we are convinced that a his-
torical analysis of the development of our emotional vocabulary is necessary
to supplement the foregoing remarks. It would have to be shown, in this
respect, that the conceptual model of anger that we attribute to the humoral
doctrine has indeed entered the language under the influence of the popular
dissemination of the latter, and further, that there is a continuous tradition
from that period to ours. Because the (a-historical) physiological model does
not impose such restrictions on the historical development of the language
(assuming at least that the physiological correlates of our emotions are his-
torically stable), the ultimate test for the humoral hypothesis consists of a
diachronic lexicological analysis. The purpose of this paper, to be sure, is
not to carry out this diachronic analysis, but merely to show how it follows
in an obvious manner from the humoral hypothesis.
Further, we do not want to create the impression that the whole of our
emotional vocabulary can be motivated in humoral terms. We are well aware
that various patterns of conceptualization can be discerned in our emotional
the fundamental importance of Kovecses's research is precisely
that it takes a major step towards the identification of those patterns. As a
consequence, we do not claim that the humoral doctrine has had the same
amount of impact on every possible emotion concept. Our basic claim so far
has merely been that such an influence cannot be disregarded if a proper
insight is to be obtained into the motivation behind our contemporary emo-
tional vocabulary. Precisely how far the humoral influence goes is another
matter, and one that can only be solved by further research.
Specifically, we do not wish to imply that physiological factors are un-
important for the structure of our emotional vocabulary, nor that they could
not interact with the historical humoral influences. Such an interaction could
take various forms, one of which is that the physiological factors have a
marked influence on the reinterpretation process that expressions with a
humoral origin undergo in the course of time. The fact that a number of
contemporary emotional expressions has its historical origin in the theory of
humors does not imply, to be sure, that the theory synchronically determines
the interpretation of those expressions: though our vocabulary for the con-
cept of anger may still bear the imprint of ancient medical theories, we no
longer believe in the theory as such (notwithstanding isolated relics like the
masturbation beliefs mentioned above). This implies that the expressions
have gone through a process of reinterpretation. Pope (1985: 179) correctly
identifies this reinterpretation process as one in which expressions that were
once taken literally acquire a figurative interpretation:
Though it [the humoral doctrine] now may be dead in our minds it is far
from dead on our tongues. We have been taking each other's temperatures
for over a hundred years and finding them steady at around 98.4 OF, but we
still use and understand the language of humoral psychology. The only dif-
ference is that when we describe somebody as having hot blood or a cold
heart or a dry wit we realize that we are talking metaphorically, whereas in
the past we would have believed ourselves to have been talking about
physical qualities.
We would suggest, then, that the physiological factors that Kovecses
concentrates on could be a crucial factor in this reinterpretation process. As
the original literal motivation gradually disappears, the elements of our emo-
tional vocabulary could receive a new interpretation as figurative expres-
sions of the physiological effects of particular emotions. Such a physiologi-
cal reinterpretation would not be automatic, however: in some cases, the
expressions could simply loose all transparency (following the valling
model), while in others, the new figurative meaning could be purely meta-
phorical rather than metonymical along the 'physiological effects' line. For
instance, taking for granted that the origins offiery in the sense 'irascible'
are humoral, and also taking for granted that the expression has not become
totally opaque in the way in which valling has, its contemporary reinter-
preted meaning could be based on the physiological metonymy that anger
causes body heat (as Kovecses would suggest), but it could also be the case
that fiery is synchronically interpreted on the basis of a metaphorical image:
the propensity of the irascible person to burst out abruptly could be com-
pared with the fire's tendency to flare up suddenly.
Methodologically, what is required here is an investigation of the way in
which the expressions in our contemporary emotional vocabulary are actu-
ally interpreted: what kind of interpretation (if any) do people associate with
them? Again, we wish to emphasize that a closer scrutiny of the reinterpreta-
tion process also naturally includes a historical analysis - if only because the
reinterpretation process is a historical phenomenon. By following the histori-
246 Nfeaning and culture
cal development of our emotional vocabulary on a step by step basis, infor-
mation about the reinterpretation process can be obtained. In particular, it
can be hypothesized that those lexical items that are not easily reinterpreted
will sooner disappear from the language than others. On the basis of this
assumption, the relevance of more specific hypotheses can be determined. Is
it correct, for instance, that expressions that are easily reinterpreted along
the physiological lines set out by K6vecses, are more resistant to a process
of lexical loss? The obsolescence of a number of the expressions mentioned
under (5) could be an illustration of the same reinterpretation process: if the
expressions under (5) are synchronically less lively than those under (6), this
may very well signal a historical shift from a humoral to a physiological
interpretative framework. It is beyond the scope of this paper to deal with
the question in detail, but we hope to have made clear that historical ques-
tions such as these follow logically from a consideration of the possible hu-
moral origins of our contemporary emotional vocabulary.
Finally, it should be pointed out that we have not tried to answer the
question where the humoral theory itself comes from, conceptually speaking.
Obviously, it is based on anatomical observations concerning the bodily
fluids, but is there any reason why, within the theory, the concept of anger
should be specifically linked to the yellow bile, and to fire? Within a physio-
logical conception of our emotion vocabulary, it seems attractive to postulate
that the humoral theory itself draws on a pre-theoretical physiological ex-
perience of the emotions. At the time of its conception, the humoral theory
would then be a literalization of a pre-existing, physiologically motivated
metaphorical understanding: the conceptualization of anger as fire, for in-
stance, would then primarily be a physiological metaphor that is later turned
into a literal statement in the framework of the medical theory of humors.
This is a position that is implicit in K6vecses (1995): accepting the possible
influence of culture-specific influences on the emotion vocabulary available
in a specific language, he argues that there exist cross-culturally uniform
factors of a physiological nature that constrain and stabilize the cultural
conceptions.
K6vecses argues that such a conception contradicts the suggestions made
in the present paper, but this is a conclusion that we explicitly have to op-
pose: as should have become clear from the foregoing remarks, we do not
claim that only cultural factors are important, and that physiological factors
could not play a role in the development of our emotion vocabulary. It is
important, in this respect, to distinguish between the methodological and the
substantive part of the present article. The article purports to do two things:
to argue for the importance of culture-specific historical research when pre-
sent-day emotion vocabularies are being considered, and to put forward a
particular hypothesis (the humoral one) within such a perspective. Kovecses
(1995) accepts the methodological point to the extent that he explicitly rec-
ognizes the potential influence of cultural factors on the development of our
emotion vocabulary, but he exaggerates the weight we would like to attach
to the humoral theory. In particular, we claim that including the humoral
doctrine into the picture is important for accounting for our present emotion
vocabulary, but we have made no statement about the origins of the doctrine
itself. In general, we would therefore like to leave open the possibility that
Kovecses is right when he suggests that the humoral doctrine is a culture-
specific rationalization of a universal physiologically-based metaphorical
understanding of the emotions. However, we would like to stress that there is
once again a methodological point to be made: if the physiological conceptu-
alization of anger (as typically embodied in physiological metaphors) pre-
cedes the humoral theory, the only way to establish this is by doing historical
research. If the suggestion implicit in Kovecses (1995) is correct, the pre-
Hippocratic conceptualization of anger in Classical Greek should be based
on physiological metaphors. It is beyond the scope of this article to test the
hypothesis, but it is methodologically important to see that it is an empirical
hypothesis that can be tested through historical research. Here again, our
conclusions on the methodological level are more important than those on the
substantive level: regardless of whether the origins of the humoral theory are
indeed physiologically metaphorical or not, the very question about the ori-
gins of the humoral doctrine calls for historical research.
To summarize, the present paper is to a large extent hypothesis-forming:
we claim that it is necessary to take into account the historical background
of our emotion concepts to get a clear picture of the present-day situation,
and in particular, that it is necessary to include the humoral doctrine into the
investigation. Also, this investigation naturally entails a longitudinal histori-
cal analysis of the development of our emotional vocabulary. But if we hope
to have established the necessity and the attractiveness of such a research
programme, we certainly do not pretend that we have already carried it out.
6. Methodological musings
In the previous sections, we have presented an alternative for Kovecses's
analysis of the general metaphor ANGER IS THE HEAT OF A FLUID IN A CON-
TAlNER. Instead of a straightforward physiological interpretation, we suggest
that it has undergone the influence of the humoral doctrine, but that the
original set of humoral expressions has been subjected to a process of rein-
terpretation and obsolescence. We are now in a position to make some fur-
ther methodological remarks. Two related topics will be discussed: the use of
conventionalized language, and the relationship between folk models and
scientific knowledge.
Kovecses explicitly takes the conventionalized way in which a particular
culture talks about the emotions as an indication of the way in which that
culture conceptualizes the emotions; the conventionalized language under
scrutiny includes idioms, cliche's, sayings, proverbs, collocations, and set
expressions in general (1989: 43). Now, while Kovecses states with some
emphasis that 'each and every expression related to a concept has to be ex-
amined if we wish to uncover the minute details of the concept' (1989: 44),
the question arises why expressions such as those that are mentioned under
(5) are not included in the observational basis of his treatment of the concept
'anger'. Why has Kovecses picked out for consideration the particular set of
expressions that he actually concentrates on?
It could be hypothesized that Kovecses has explicitly restricted his analy-
sis to those expressions that are the most transparent ones for a contempo-
rary audience, i.e. those expressions whose metaphorical nature is still a live
one, or, more generally, those expressions that are most readily considered to
be motivated by today's speaker's of English. The question of motivation
can be illustrated by comparing an expression such as valling with an item
such as to make one's blood boil. Although they have a common historical
motivation in terms of the theory of humors, valling is entirely fossilized and
opaque for the contemporary language user, whereas to make one's blood
boil could possibly receive a motivation along the physiological lines set out
in the previous section. The problem that is at stake here is the same as the
one mentioned there; also, it has been identified several times in connection
with the 'generalized metaphors' approach of Lakoff and Johnson (1980), on
which Kovecses' s identification of metaphorical patterns is based: if general-
ized metaphors are cited as evidence for our contemporary way of conceptu-
alizing the world, it does not suffice to identify the metaphor, but it has to be
shown on independent grounds that the metaphors are not just dead ones (see
among others Traugott 1985, Geeraerts 1981). Those 'independent grounds'
could be the researcher's intuition, but also, for instance, psycholinguistic
experiments in the line of Gibbs (1990). So, what would have to be shown
before expressions such as those in (6) are cited as evidence for our present-
day way of conceptualizing emotions is not just that they are not dead meta-
phors (in the sense in which valling is opaque and fossilized), but also that
the motivation that they actually receive is in tenns of the physiological ef-
fects of anger. (On related questions, see also Ortony 1988.)
Now, it would be unfair to claim that Kovecses ignores the question
whether the conventionalized language we use to talk about the emotions
actually reflects our current beliefs. He discusses the problem by making a
comparison with our astronomical model of the world: expressions such as
the sun came up and the sun went down cannot be used as evidence for a
geocentric folk model, basically because fonnalized education has pro-
foundly influenced our way of thinking. This is, says Kovecses, in marked
contrast with the situation on the emotional field.
As a result of certain scientific discoveries, our educational system has
spent several centuries on changing our geocentric view of the relationship
between the earth and the sun. The consequence is that, despite our lan-
guage use, anyone with at least some elementary education would refuse
the geocentric view as his or her folk model of the earth/sun relationship.
Nothing like this has been the case with the emotions. No such large-scale
attempts have been made to change our thinking about them (. .. ). As a re-
sult, we pretty much believe what we say about them. (... ) It seems then
that, as far as the emotions go, we still live by and think in terms of a geo-
centric emotional universe (1989: 45-46).
There are various things to be said about this view. To begin with, it does
not invalidate the methodological problem identified above. Because
Kovecses reaffinns the necessity to take into account all expressions - and
to take them at face value - the fact remains that he does not follow his own
methodological dictum. But if he would have followed it, the unmistakable
presence of items derived from the theory of humors might have led him to
the conclusion that we indeed still (partially) adhere to a 'geocentric' view of
emotions (i.e. to a prescientific, medieval theory), but that the theory is a
humoral one rather than the physiological one he suggests. So we are faced
with a dilemma: either Kovecses is right in affinning that our present-day
views about the emotions have not been influenced by scientific discoveries
(but then his method of taking expressions at face value would lead to the
conclusion that we still have a humoral conception of anger), or he would
have to reconsider his statements about the influence of scientific theories on
our emotional vocabulary, (together, in fact, with his belief that emotional
expressions can be taken at face value, i.e. without considering the possibil-
ity of reinterpretations).
250 lvfeaning and culture
The fact, on the other hand, that we clearly no longer take the humoral
expressions literally (the fact, that is, that we no longer believe the theory)
can only be attributed to the same kind of dissemination of scientific theories
that led to the downfall of the geocentric view in astronomy. Because we
have learned about the new anatomical and physiological discoveries, we
have abandoned our earlier folk models (by and large, i.e., not counting relic
beliefs like those about masturbation that we started off with). And even this
older folk model itself was not a pure folk model. As we have seen, it was a
piece of high, Latinate culture that was gradually incorporated into the
common culture through the intermediary of popularizing publications.
There is no reason, in short, to believe that our emotional vocabulary is free
of scientific influences, and there is no ground for a methodological exploita-
tion of such a conception of the specificity of our emotional vocabulary.
7. Culture and cognition
To sum up, we have tried to establish the following points. First, the medie-
val physiological-psychological theory of the four humors and the four tem-
peraments has left its traces on our emotional vocabulary. Second, the AN-
GER IS THE HEAT OF A FLUID IN A CONTAINER metaphor identified by
K6vecses (1989) can be seen as one of those traces. It is then not motivated
directly by the physiological effects of anger, as K6vecses suggests, but it is
part of the historical (but reinterpreted) legacy of the humoral theory. Third,
K6vecses's neglect of the historical background of our emotional vocabulary
prevents him from appreciating the possible impact of the humoral theory;
once this possible impact is taken into account, inconsistencies in K6vecses's
methodology become apparent. Fourth, a further corroboration of the his-
torical-humoral hypothesis requires a longitudinal scrutiny of the historical
development of our emotional vocabulary.
Because they are the most wide-ranging, we consider the methodological
consequences of our investigation to be of primary importance. The basic
point as we see it is this: an adequate analysis of the motivation behind cul-
tural phenomena in general and language in particular has to take into ac-
count the diachronic dimension. Cultural models, i.e. the more or less coher-
ent sets of concepts that cultures use to structure experience and make sense
of the world are not reinvented afresh with every new period in the culture's
development. Rather, it is by definition part of their cultural nature that they
have a historical dimension. They can only fulfill their role of shaping a
conununity's life if they have a historical pennanence, that is, if they can be
transmitted from generation to generation, assuring continuity over and
above an individual's and an individual generation's activities (though not,
to be sure, unaffected by them). If cognitive models are cultural models, they
are also cultural institutions, and as such, they carry their history along with
them: their institutional nature implies their historical continuity. It is only
by investigating their historical origins and their gradual transfonnation that
their contemporary fonn can be properly understood.
Now, while one of the major steps forward taken by Cognitive Semantics
has been to put the study of meaning back into its cultural and experiential
context, it would seem that the natural consequence of including the dia-
chronic dimension into the investigation has perhaps not yet been fully ap-
preciated. There is an instructive parallel to be drawn here between the cog-
nitive-semantic study of single lexical concepts (as in prototype theory) and
the cognitive-semantic research into supra-lexical structures such as the
cultural models of emotion that K6vecses concentrates on. In the case of
purely lexical research, the emphasis on the mechanisms of semantic flexi-
bility that underlie the structure of polysemy (such as metaphor and meton-
ymy) naturally entails a renewed interest in diachronic semantics (see Geer-
aerts 1988a): to a large extent, the synchronic polysemy of lexical items is a
reflection of their diachronic development. The point that we are trying to
bring home here is that an awareness of the synchronic reflection of dia-
chronic patterns is just as natural and just as important in the case of su-
pralexical cognitive structures as in the case of lexical concepts.
If cultures are only cultures because they have a tradition, and if, there-
fore, cognitive models are only cultural models if they have a chronological
continuity and a historical pennanence, an awareness of the history of ideas
is methodologically indispensable for Cognitive Semantics.
Chapter 10
Beer and semantics
Originally published in Leon De Stadler and Christoph Eyrich (eds.), 1999, Issues
in Cognitive Linguistics 35-55. Berlin/New York: Mouton de Gruyter.
This paper is a revised version of the opening address presented on July 18, 1993
at the 3rd International Cognitive Linguistics Conference. Both the tone and the
topic of the paper bear witness to the original occasion: the topic introduced the
opening reception of the conference (which featured a representative sample of
choice Belgian beers), and the tone is accordingly somewhat lighter than linguis-
tic papers tend to be. Even so, the present analysis is of interest beyond the origi-
nal context, for two reasons.
First, it illustrates some of the salience phenomena that were described in Chapter
4 of the present volume. Structural salience was there defined in tenns of the
weight of the distinctive dimensions that differentiate between lexical items: a
semantic feature or dimension is more structurally salient to the extent that it
occurs more often in the structure of the lexicon, i.e. to the extent that it is more
often expressed or evoked. In the present paper, this is illustrated both with regard
to the purely descriptive and with regard to the metaphorically evocative brand
names for Belgian beers. The analysis of the non-metaphorical brand names
shows how some distinctive characteristics of specific beer types are more readily
expressed than others. Beers that are brewed in trappist monasteries, for instance,
are invariably identified as such, but the addition of herbs for flavoring is only
seldom a motif in naming. More importantly, the analysis of the metaphorically
evocative names reveals that beer is associated with a culturally specific cluster of
conceptual metaphors in the Lakovian sense: it combines pleasure, prestige, local
identity and tradition - and a sense of adventure and transgression.
Second, the paper adds a specific perspective to the study of proper names in
Cognitive Linguistics. Proper names have not yet been studied intensively by
cognitive linguists, but to the extent that they have, stress has fallen on the theo-
retical question what kind of meaning, if at all, is involved in the use of proper
names (Van Langendonck 1999). In addition, studies like Jakel (1998, 1999) have
looked at the metonymical mechanisms at work in proper names. The present
paper adds the perspective of lexical salience and that of metaphor analysis as
ways of uncovering underlying patterns in name-giving. More work on beer
branding may be found in Wagner (2002, 2003), who studied the image of beer in
German advertisements.
Beer and semantics 253
1. Belgium as beer's own country
If Cognitive Linguistics is about the relationship between language and cul-
ture, Cognitive Linguistics conferences should be concerned with learning
about cultures as much as they are concerned with learning about language
and languages. That is why it seemed appropriate to start off the 3rd Inter-
national Cognitive Linguistics Conference with an investigation into a cen-
tral aspect of Belgian culture, namely beer. In approaching a culture, in fact,
it seems only fair to tackle those areas of behavior first that are experienced
as being important by the members of the culture themselves. And it seems
that there is indeed sufficient evidence for the claim that beer is part of the
pride and joy of the inhabitants of that minuscule patch of land along the
coast of the North Sea.
First, some factual data may demonstrate the importance of beer in Bel-
gium. To begin with, the consumption of beer per head of the population is
relatively high. Data provided by the lnterbrew breweries in 1987 list the
following per capita beer consumption in twelve European countries: Swe-
den 5lliters, Norway 51, Germany 145, Czechoslovakia 130, Belgium 121,
Austria 118, England Ill, The Netherlands 85, Ireland 75, Spain 65, Portu-
gal 40, Italy 25. The figures demonstrate the existence of a so-called Euro-
pean beer-belt, situated geographically between the spirit-belt in the north of
Europe, and the wine-belt in the south.
The figures, of course, do not show that Belgium is beer country number
one as far as per capita consumption is concerned: it comes third after Ger-
many and Czechoslovakia. The key concept regarding Belgium's status as a
beer country, however, is not quantity, but quality. And the quality shows up
specifically in the enormous variety and the abundant diversity of the types
of beer produced in Belgium. In the data set that was used for this investiga-
tion (and which is taken from Peter Crombecq's annual overview of Belgian
beers for the year 1992) no less than 1454 different brands are distinguished.
In Crombecq' s own classification of these beers according to taste and alco-
holic strength, no less than 304 distinct types of taste appear. A comparison
with Holland, Belgium's nearest neighbor, strengthens the impression of
diversity. Against the 1454 different brands to be found in Belgium, there
are only 354 Dutch ones. And while Crombecq distinguishes 304 different
Belgian tastes, he only tastes 94 Dutch ones. In short, there is sufficient
evidence for the variety of Belgian beers.
At least as important as these factual data is the recognition that Belgians
identify themselves with the country's production of high quality beers. If
254 A:feaning and culture
you ask people from Belgium what their country is good at, they will proba-
bly mention beer and chocolates. The recognition that beer, and the quality
of Belgian beer, is an explicit topic in Belgian culture, may be further illus-
trated in two ways. One is the following slogan from a beer advertisement:
Het bier van het land van het bier, i.e. 'the beer from the country of beer',
or, more freely, 'the beer from beer's own country'. Belgium as beer's own
country - this is apparently a live notion. The second illustration consists of
the number of coffee table books that are published in Belgium on the topic
of beer (although the name coffee table books is obviously somewhat para-
doxical in this respect). Here is a list of some more or less recent ones; if
nothing else, it at least shows that there is a live interest in beer, and that
beer is considered a quality product worthy of luxurious publications of the
type that is usually restricted to wine and art.
M. Jackson [1977]. Spectrum bieratlas (translation of The world gUide to
beer). Utrecht!Antwerpen.
W. Patroons [1984]. Alies over Belgisch bier. Antwerpen/Weesp.
1. Tulfer [1986]. Belgische biergids. Antwerpen.
M. Jackson [1991]. De grote Belgische bieren. Een volledige gids en een
hulde aan een unieke cultuur (translation of The great beers ofBelgium).
Anwerpen.
G. van Lierde [1992]. Bier in Belgie. Gids voor bieren en brouwerijen.
Roeselare.
2. Methodological considerations
Given the importance of beer in Belgian culture, the cognitively interesting
question is this: how is beer conceptualized in Belgium? What are the cogni-
tively salient concepts associated with it? And more specifically, starting
from a Cognitive Linguistic perspective: what does language reveal about
the conceptualization of beer? The two crucial aspects of this question that I
would like to focus on are the following. First, a general methodological
question: how can salience effects in lexical fields be measured? Second, a
specific question about beer: what are the specific values associated with
beer? Before I try to be more specific about these questions, I have to clarify
a few points.
To begin with, there are a number of restrictions on what I will be doing
next that have to be mentioned explicitly. The vocabulary that I will be talk-
ing about consists basically of brand names, i.e. the names of types of beer.
This means that various parts of the lexical beer field will pass unnoticed.
For instance, I will not have the opportunity to mention the names of pubs,
inns and cafes (in spite of such gems as The ship o.ffools, The last judg-
ment, The kingdom of heaven, The ultimate hallucination). The restriction
also implies that the vocabulary for talking about beer will not enter the
picture. I will have nothing to say, that is, about predicates like earthy,
fruity, refreshing, complex, seductive, and others that may be used to de-
scribe beers. (Surprisingly perhaps, this vocabulary is far less extended than
the one that exists for wine and that has been analyzed so elegantly by
Lehrer 1983.)
Another consequence of the restriction to brand names is that the every-
day names of the beers are not considered systematically in what follows.
The brand names, if you like, are full names, constituting a truly individual
identification of the beers. The everyday names, on the other hand, are like
Christian names: shorter but at the same time more general forms that are
communicatively sufficient in most circumstances. In the technical terms of
lexical semantics, the brand names are certainly not basic level terms: they
are used much less frequently than the everyday names, and they are taxo-
nomically much more specific. In most cases, the generic basic level de-
nominations are based on overall types of beer. Whereas the full names are
clearly proper names, the everyday names are common nouns. When you ask
for a geuze in a cafe, it will mostly be of no interest to you whether you are
served an Eylenbosch gueuze Iambic, a Drie Fonteinen geuze, a Mort
Subite geuze or any of the other brands of geuze. But by a typical prototypi-
cality effect, the everyday name is not necessarily originally a common
name. When you ask for one of those typical white beers made on the basis
of wheat you would probably ask for a Hoegaarden, which is the most
typical representative of this class - and you probably would not object if
you were then served a Dentergems witbier, which is another example of the
same class.
It could be argued, no doubt, that the basic level terms are somewhat
more interesting for lexical analyses than the brand names, precisely because
they are cognitively more salient. Note, however, that systematic research
into the actual use of the everyday names would require extensive obser-
vation in pubs and cafes, and I am afraid that the investigation would then
be diverted very soon onto scientifically unfruitful although extremely pleas-
ant side-roads. Moreover, precisely because the brand names are more spe-
cific than the everyday names, they may reveal more of the conceptualization
of beer. Again, it could be remarked that the conceptualization they reflect is
not the consumer's but the brewer's, but even so, the brewers are likely, for
obvious commercial reasons, to choose names that have a broad appeal to
the consumers. Even though we study names that are not actually given by
the consumer, the names can be expected to fit the consumers' image of
what beer should be.
From a methodological point of view, the field of beer terms is interesting
for two reasons - two complications, in fact. The first methodological point
to note involves the fact that the names found in the data set subclassify into
two fundamentally distinct subsets, which each require their own type of
analysis.
The majority of the brand names conforms to a general pattern: they con-
sist of an identification of the firm producing the beer, and a specification of
the type of beer involved. Here are some examples: Louwaege pils, Liefmans
frambozenbier, Belle Vue geuze. There is considerable variation in this pat-
tern, not least because neither the first nor the second element is absolutely
necessary. In some cases, a single reference to the origin of the beer suffices.
This is the case, for instance, with some of the beers produced in monaster-
ies (I will come back to these further on). A name like Orval identifies the
monastery together with the single kind of beer produced there: there is only
one Orval kind of beer, and it is a trappist beer. Conversely, only the type of
beer may be specified, without reference to the brewer, but then, the generic
name will be subclassified by the addition of various specifications as in the
following: Dentergems witbier, Aarschotse bruine, where a placename indi-
cates that this is the type of white beer or brown beer or whatever as typi-
cally brewed in Dentergem, or Aarschot, or wherever. The subclassification
may also occur in combination with an identification of the brewer, as in
Liefmans geuze gefilterd. Liefmans geuze ongefilterd, where it is indicated
that the geuze beer produced by the firm Lindemans comes in two varieties:
a filtered one and an unfiltered one. In cases like the Orval-type, where the
element identifying the producer at the same time suggests the kind of beer
involved, the subclassifying element occurs without an explicit mention of
the generic type of beer. In the following set, the various additions identify
the varying degrees of alcoholic strength of the beers. The name West-
vleteren is again a name for a monastic brewery; it identifies the beer in
question as a trappist beer: Westvleteren dubbel 4, Westvleteren special 6,
Westvleteren extra 8, Westvleteren abt 12.
All the names mentioned so far stand in dramatic contrast with a type of
name illustrated by the following examples: Duvel, Luc!!er, Judas, Ver-
boden Vrucht ('Forbidden fruit'). The basic distinction between this type
(which fonns a sizeable minority of the total set) and the previous one re-
sides in the fact that the type 1 names are literal names, whereas the type 2
names are figurative, or at least associative names. A name like Lie/mans
geuze gefilterd is made up of various infonnational elements that identify
aspects of the beer: the fact that it is of the geuze type, the fact that it is
produced by the Liefinans brewery, the fact that it has gone through a filter-
ing process. Each of these infonnational elements is identified directly in the
name of the beer, i.e. by means of words whose literal meaning it is to ex-
press the infonnational elements in question. The literal meaning of the type
2 names, on the other hand, does not as such identify the relevant infonna-
tional element. The name Duvel literally means 'devil', but the infonnation
that is relevant for the beer name is only indirectly connected with this literal
meaning: it is a concept that has something to do with the devil, and that for
one reason or another is also appropriate to talk about this kind of beer. Not
surprisingly, the identification of the relevant infonnational element is not
always easy with type 2 names. The beer may have something devilish, but
what exactly is that? Does it lead you to destruction? Is it a sin to drink it?
Does it overpower you in the way the devil might? One could say, perhaps,
that the type 2 names attribute somewhat vague qualities to the beers,
whereas the type 1 names describe rather clearly defined features.
In summary, the type 2 names are associative or, if one wishes, evocative
names, because they function indirectly, and because they involve less
clearly defined characteristics than the type 1 names. The consequence of the
distinction is that the type 2 names have to be analyzed on a different basis
than the type 1 names. In the case of the type 1 names, one can base the
analysis rather straightforwardly on the literal readings of the tenns. In the
case of type 2 names, one has to take into account the relationship between
the original literal meaning and the evocatively associated meaning. This
type of analysis, of course, is well-known within Cognitive Linguistics: it is
analogous to what we all do when we analyze metaphorical patterns in
Lakovian style, i.e. to identify source and target domains and pinpoint the
precise motivational link that connects them. In the vocabulary subset that
interests us here, the associative names are not all metaphorical, but the
analytical procedure used for metaphors can be generalized.
The second complication that I hinted at before involves the fact that one
of the things that we are crucially interested in are salience effects in lexical
.fields. Given a set of categories that belong together in a conceptual field
(like types of beer), what we would like know is which categories are con-
ceptually more important than others. This is not a question that is ade-
quately answered in traditional lexical field theory, but I think we should
recognize that even Cognitive Linguistics, with its outspoken interest in sali-
ence effects, has not yet developed a proper methodology for treating this
kind of issue. The salience effects Cognitive Linguistics has been primarily
interested in are salience effects within separate categories, an interest that is
epitomized by prototype-theoretical research (cp. Geeraerts 1989c). Apart
perhaps from Berlin's basic level model (see Berlin 1976, 1978), not much
systematic attention has been devoted to salience effects that involve not the
various applications of a single category, but rather the relationship between
various lexical categories as altemative names for the same kind of referent.
There is, to be sure, a nuance to be added to this picture. As far as the
associated values as expressed by Type 2 names are concemed, we do have
a clear-cut idea of how to measure salience: an associated value is more
salient to the extent that it recurs throughout the set of evocative names.
This is basically the measure of cognitive salience behind the search for
generalized metaphors introduced by Lakoff and Johnson (1980). Note,
however, that it involves only Type 2 names, so that it remains important to
try to find a measure of salience that specifically applies to Type I names.
3. The origins of the diversity
To see what is involved in tackling the methodological question raised a
moment ago, we should first have a closer look at the structure of the refer-
ential field itself, i.e. at the variety that I hinted at before. In order to get a
better grasp of it, it is important to know something about how beer is made.
Basically, you put barley into water and let it ferment. However, an elabo-
rate beer culture could not have developed if this process had not been sub-
ject to refinement. Schematically, what actually happens can be divided into
three steps. First, the barley is let to soak and germinate, and the budding
barley is then dried or sometimes roasted. The product of this first step is
known as malt, which reveals, by the way, that whisky is the distilled coun-
terpart of beer. In a second step, the malt is crushed and cooked, so that the
natural sugars that it contains become free. During the cooking stage, addi-
tional substances for flavoring are added, such as hop or extra sugar. The
product of this second step is called wort. When yeast is added to the wort, it
can then ferment, and at the end of this third stage, you get beer.
At each of the three stages, altematives with regard to the standard proc-
ess exist. Instead of starting off with barley, for instance, you can begin with
wheat, or rye, or a mixture. In the second stage, various other flavoring sub-
stances can be added. Typical for a number of Belgian beers is the addition
of fruit flavors like cherry and raspberry, or certain herbs like ginger, lico-
rice, coriander, or mint. And in the third stage, different forms of fermenta-
tion exist. This is an important point that needs some elaboration. Basically,
there are three forms of fermentation: high fermentation, low fermentation,
and spontaneous fermentation. In the first case, the fermentation process is
relatively short and intense, producing strong ales. In the case of low fermen-
tation, the process is a slow one, the ripening process takes a long time, and
the resulting beer is less heavy than in the case of high fermentation. These
are the beers of the regular pilsener or lager type. Spontaneous fermentation,
on the other hand, is much less common than high or low fermentation, and
is really extremely typical for one portion of the Belgian beers. In the sur-
roundings of Brussels, the art of producing beers by spontaneous fermenta-
tion has been refined in ways not known anywhere else in the world, and
some of the most typical names associated with Belgian beers have to do
with spontaneous fermentation. Specifically, I first have to mention lambiek,
which is the product of 100% spontaneous fermentation. As lambiek as such
is rather sour, you will not find very often in its pure form. Usually, sweet
flavors are added in the form of fruit extracts, which yields names like kriek
lambiek 'cherry Iambic'. Another additional process that Iambic may go
through is a second fermentation, which yields the well-known geuze, which
is sweeter than Iambic, and which has a unique, champaign-like sparkling
character.
'_____ __
germination,
drying/roasting
addition of hop,
addition of sugar etc.,
cooking
addition of yeast,
fermentation
Figure 1. Schematic representation of beer production
The variety does not stop there, however. At least two additional points
have to be mentioned. For one thing, some Belgian beers are brewed only on
a special occasion. Traditionally, this could be the harvesting season, or
Christmas, or the annual festival of the local patron saint. Nowadays, these
gelegenheidsbieren or 'occasional beers' also include beers made to com-
memorate specific historical events, like the anniversary of the foundation of
one or another city, or the birth of a historical figure.
Another additional factor besides the temporal one is a geographical one,
or at least, it involves the place where the beer is being brewed. This is par-
ticularly important for abbeys and monasteries producing beer. In fact, refin-
ing the art of brewing is one salient aspect of the civilizing influence that
monasteries and abbeys have exerted in the Low Countries. Traditionally, all
monasteries brewed their own beer, like many farmers once did. Today,
there are still five monasteries left where the monks brew their own beer. As
they all belong to the Trappist order, their beer is commonly known as trap-
pist. The monasteries in question are those in Orval, Chimay, Rochefort,
Westvleteren and Westmalle. These beers are all of the strong, high fermen-
tation type, but there are many differences among their individual flavors. It
is useful, by the way, to distinguish the trappist beers from the larger cate-
gory of abdijbieren or 'abbey beers'. These are beers of basically the same
type as the trappist beers, but they are being produced by commercial brew-
eries who have taken over the original monasterial breweries, or who have
merely paid for the license to use the abbey's name. Examples are Leffe,
Grimbergen, Maredsous, Affligem, Corsendonck, and Tongerlo.
On the basis of what I have explained so far, we can now define, on a ge-
neric level, some of the most common types of Belgian beers.
pils: a light blond beer of the pilsener or lager type, produced through
low fermentation (common brands: Stella, Jupiler)
witbier: 'white beer', a light beer on the basis of wheat rather than bar-
ley, with a blond, somewhat cloudy appearance (common brands: Hoe-
gaarden, Dentergems)
geuze: a sparkling, spontaneously fennented beer with a reddish color
and a sour-sweet taste (common brands: Belle-Vue, Mort Subite)
kriek: a moderately sweet beer obtained by adding cherry juice to geuze
or lambiek (common brands: Belle-Vue, Mort Subite)
trappist: strong, mostly dark brown beer of the high fermentation type,
produced in Trappist monasteries (common brands: Orval, Chimay,
Rochefort, Westmalle). (Sometimes, the name trappist will also be used
for the comparable commercial abbey beers).
4. Salience effects for type 1 names
Although it would seem that the list at the end of the previous section gives
us a nice and neat set of categories to start the onomasiological investigation
with, things are less clear than they seem. Once we start looking at more
specific categories, there appears to be a lot of overlap and cross-
classification among the various groupings of beers. Consider the following
example. As we have seen, we can distinguish between geuze, lambiek, and
kriek in the category of spontaneously fermented beers. Among the beers
with high fermentation, there are the whitebeers, based on wheat, but also a
number of barley beers, such as the popular, very strong beer called Duvel
('devil'). Geuze and lambiek have in common with the whitebeers that wheat
is used in the production process. Duvel and geuze, on the other hand, have
in common that their fermentation process continues when they are already
bottled. And further, there are kriek beers that are not based on geuze or
lambiek. like the Felix kriek oudbruin. The resulting picture is one of multi-
ple overlapping, as shown in Table 1.
Table 1. Multiple overlapping in the lexical field of beer names
::::
..::,: ..::,:
"-
.s;
<\l <\l
-.
. ~
"-..::,:
<\l
:.c; ..::,::.c; <\l
>,
;>.
~ ~ . ~
~
lE::
<\l -
~
0;:: ::::
Q
;;:
<\l
~ . . : : , : ~
Cl"::': I:lI)
high fermentation
+ + +
spontaneous fermentation + + +
fermentation in bottle + +
wheat based
cherry f1avor
+ +
Each beer type, in other words, can be considered a bundle of features,
each of which can be the basis for the name of the beer (or part of the name
of the beer). But how then can we go about determining the conceptual sali-
ence effects that we are interested in? If each beer type were to belong
uniquely to a neatly separated, taxonomically well-behaved category, we
would just have to investigate those categories. But if the categories them-
selves are extremely varied and exhibit multiple overlapping, what should
one do? I earlier referred to Berlin's basic level model as the only systemati-
cally elaborated model for salience within lexical fields. But note that Ber-
lin's model deals with the salience of levels in a taxonomy, while we are
interested here in salience phenomena among categories on fundamentally
the same level of a taxonomy. So it seems that it may be worthwhile to sug-
gest ways of dealing with intercategorial (rather than intracategorial) sali-
ence effects in a way that goes beyond the basic level model. (The ideas
presented here are developed in more theoretical and empirical detail in
Geeraerts, Grondelaers, and Bakema 1994. The notion 'intercategorial sali-
ence' is briefly introduced in Geeraerts 1993b; it is basically a quantitatively
operationalized version of the concept of entrenchment defined by Langacker
1987: 59-60. A useful presentation of the distinction between intracategorial
and intercategorial entrenchment may also be found in Kleiber 1991.)
The obvious thing to do, I would suggest, is to start from the features
themselves that constitute the bundles representing each individual type of
beer. A practical methodology can then be formulated as follows: the more a
particular feature is expressed in the names of beers, the more it is cogni-
tively salient. When, for instance, the presence of fruit flavors never surfaces
in the names of the beers that have such a flavor, it is unlikely that fruit fla-
vor is a preponderant feature of beers. On a general methodological level, I
would like to suggest the following operational definition of conceptual sali-
ence in lexical fields: a category in a lexical field is more salient to the
extent that its members are more often identified by names that are typical
for the category. Of course, we have to be more precise about what it means
for a feature 'to be expressed in the names of beers'. In most cases, there is
no problem: when you find a reference to kriek in a name, and when you
know that kriek is the name for a type of cherry, there is no difficulty in
concluding that the addition of cherry juice is mentioned in the name of the
beer. In other cases, the reference is more indirect. A striking example is the
following: in the French speaking part of Belgium, a peach-flavored beer is
produced with the name La pecheresse, which translates as 'the female sin-
ner, the sinning woman'. When you realize, however, that peche is the
French for 'peach', it will be clear that the name contains a pun with an
indirect reference to the presence of peach flavor.
In general, a feature is expressed by a name (or part ofa name) when that
name (or the relevant part of it) only occurs in connection with that feature.
Some more examples may further illustrate the point. The fact that a par-
ticular beer is low on alcohol may be expressed directly by expressions like
alcoholarm 'low on alcohol' or alcoholvrij 'without alcohol'. Indirectly,
however, the item tafelbier 'table beer' also signals the light character of
beer, to the extent that it only occurs within the group of very light beers.
Similarly, lambiek and geuze can be taken as expressions of the 'spontane-
ous fermentation' character of beers, to the extent that they only occur in
names for beers with that characteristic. Kriek, however, does not express
the feature of spontaneous fermentation, because (as we have seen in Figure
3) it also occurs in the name of beers of high fermentation.
Following these guidelines, we can show that not all features of beers are
equally important. There are clear differences in the extent to which specific
characteristics are expressed in the names. Table 2 presents a sample to
demonstrate the kind of differences in salience that occur. I should add im-
mediately that it is practically impossible to systematically explore all possi-
ble features, mainly because their presence is often uncertain: even persistent
tasting (to the extent that the researcher can muster it) does not always suf-
fice to establish the objective characteristics of a particular brand. The table
should be read as follows. The first column of figures indicates the number
of beer brands in the data set that has the feature mentioned to the left. The
rightmost column indicates how many of those beers actually carry a name
that refers directly or indirectly to the feature in question.
Table 2. Salience characteristics of sample referential features
REFERENTIAL FEATURE
based on wheat
very light, low alcohol
occasional beers
added herbs
raspberry flavor
cherry flavor
trappist beer
REFERENTIAL FREQUENCY NAME FREQUENCY
139 91 (65,5%)
110 25 (22,7%)
86 86 (100%)
59 4 (6,8%)
40 38 (95%)
19 18 (94,7%)
16 16 (100%)
The relationship among the figures mentioned to the right gives an indica-
tion of which features are considered special enough to merit a name of their
own, or at least, special enough to be referred to in the beer names. Trappist
origin, for instance, seems to be highly valued, while the addition of herbs is
not a terribly individuating feature. It is tempting, of course, to speculate
about the origin of these preferences. It is not unlikely that they reflect his-
torical traditions within the culture, but this is, to be sure, something that
cannot be established on the basis of this scanty set of examples alone - one
would need a longitudinal diachronic analysis to get any certainty on this
point.
On the other hand, the examples at least allow for the falsification of one
hypothesis. It might be suggested that the salience of a particular feature is
inversely correlated with its referential frequency: an infrequent feature like
the presence of cherry flavor is lexically expressed to a high degree, pre-
cisely because it is a marked, unexpected feature. This phenomenon proba-
bly does play a role in the lexical field as a whole: no one would expect a
reference to water in the name of beers, precisely because it is obvious to
anyone with the least encyclopedic knowledge of the domain that all beers
contain water. However, the figures in the table clearly establish that such an
inverse correlation with referential frequency could at most explain part of
the data. There seem to be at least some real preferential phenomena at play.
5. Salience effects for type 2 names
I would now like to tum to the analysis of what I have earlier called the as-
sociative values of the beer denominations. In Figure 5, five salient associa-
tive motifs are identified. They are recurrent ones not just because they show
up in various individual names, but also because they generalize over more
than one source domain. It would lead us too far to discuss all the names
separately. but here is a sample analysis of some illustrative cases.
A first recurrent pattem represents beer as something prestigious. In its
most direct form, the high status of beer is evoked by borrowing names from
the lexical field of wine (the prestigious drink par excellence), as in Vlaamse
Bourgogne 'Flemish burgundy'. The same value may be evoked more indi-
rectly by referring to important people, like kings (Grand Monarque 'Great
monarch') or members of the aristocracy (Ridder 'Knight'). A link with the
religious motif that occurs a number of times in the set is constituted by
names referring to hierarchically important people in religious organizations
(like Monseigneur 'Monsignor' and Kapittel abt 'Chapter abbof). Another
set of prestigious figures involves historically important people, who may be
political leaders like emperor Charles V (Charles Quint) or culturally
prominent people like the painters Breugel, Rubens, and Rembrandt. In most
cases, these figures have a direct historical link with the Low Countries, but
names like Napoleon and Da Vinei show that this is not an absolute precon-
dition. Next prestigious people, expensive materials and other valuables may
be evoked, as in Salir; a cross-references to one of the other subcategories
already mentioned appears in a name like Gouden Carolus. which is origi-
nally the name of the gold coin minted by emperor Charles V.
A second major motif represents beer as an adventure at the edge of the
accepted norms. Up to a certain point, the notion of adventure (danger,
physically extreme conditions, violence) and transgressive behavior (flouting
existing norms) may be separated, but because they so often occur together,
they are here treated as one category. The norms that are violated may be
basically religious or social. The first case yields names referring to sin
(Verboden Vrucht, 'Forbidden fruit'), witchcraft (Sorcieres d 'Ellezelles
'Witches of Ellezelles'), and the devil (Lucifer, Satan, Duvel 'Devil'). The
second case yields names referring to pirates (Piraat, Boucanier), smugglers
(Smokkelaar), highwaymen (Bokkereyer) and related unorthodox profes-
sions. Also, names like Schavuit, Deugniet, and Stouterik 'rascal' indicate
moderated mischievousness; they are typically the kind of name to be used
hypocoristically to a child. Brigand and Kerelsbier are particularly interest-
ing cases. For instance, in the Belgian context, Brigand refers to the peas-
ants and farmers who rebelled against the French occupation at the end of
the eighteenth century (the Boerenkrijg 'peasants' war'). Looking back at the
historical figures mentioned above, it now becomes clear that they too often
(as in the case of Ambiorix, Artevelde, Breydel and De Coninck) represent a
struggle for independence against a foreign oppressor. (It will be suggested
below that an irreverent or even rebellious attitude with regard to established
authority may be a general characteristic of Belgian culture.) Names like
Stoeren Bonk 'sturdy fellow, son-of-a-gun' evoke virility and force. Against
the background of what was just said, it seems plausible that these names do
not however evoke machismo pure and simple: rather, the image is at least in
part one of strong and independent men who stand up for their rights. A final
major source domain in this group points to the physical effects of over-
abundant consumption of beer, as in Delirium Tremens. The idea probably
is that you have to be a strong person (of the type evoked in the previous
group) to take such strong drinks. Note, in addition, that multiple associa-
tions may again be active: Houten Kop 'wooden head' may not only mean
'hangover', but also stubbornness, which is again related with the spirit of
independence.
Names like Flierefluiter 'Iayabout, idler', Lekkerbek 'gourmet', and Pal-
lieter (the name of a famous literary figure who has become symbolic for an
epicurean attitude) are typical for a third major motif, in which beer is her-
alded as a source of pleasure and as part of an easy-going way of life. Char-
acteristically, the religious motif appears once again when beer is so to
speak presented as a foretaste of the pleasures that may await us in heaven.
This is the case in among others Engeltjesbier 'little angels' beer'.
The religious source domain also contributes to the large group of names
in which beer is represented as a link with the traditional culture of the past.
Religion, to be sure, is such an aspect of the tradition, in the same way in
which the guilds of old (as in Gildenbier 'beer of the guilds' are. References
to old methods of producing beer (Traditions des Moines 'The monks' tradi-
tions') serve a similar function. In A la ferme 'On the farm', Verte cam-
pagne 'Green countryside', and Bosbier 'Beer of the woods' rural if not
pastoral associations are activated: to the extent that beer is part of the leg-
acy of the past, it is clearly the pre-industrial era that is at stake. Historical
references, to be sure, are a recurrent phenomenon in the set presented here.
Apart from the historical figures mentioned before and the references to local
history mentioned below, there are straightforward references to the past in
names like Vieux Temps 'Old times'. In Druide 'Druid', l'iguanodon 'The
iguanodon', and Blonde du Menhir 'The blonde one of the menhir', the his-
torical reference takes a remarkable prehistoric, pre-Christian shape: the
historical reference is so to speak pushed to its extreme, but at the same
time, the notions of primitivity and force that seem to accompany these ref-
erences echo some of the vigorous, vitalistic values connected with the sec-
ond motif discussed above.
The final major motif represents beer as an aspect of local identity, by re-
ferring to people and things associated with a particular region. The relevant
region can change in surface, to be sure. It may be Belgium as a whole (as
with historical figures like the painters Rubens and Breugel, or the cartogra-
pher Mercator), or it may be one of the main regions of Belgium (as when
the historical figures Artevelde, Breydel, and De Coninck evoke the past of
Flanders in particular, and when Elckerlyc and Pallieter refer to figures
from Flemish literature); very often, it is a specific town or village (as in the
many names that mention a particular placename, or the place's patron
saint). The latter case is also represented by those names that make reference
to aspects of local folklore, like famous local figures (Ketje) , mysterious
beings that appear in legends and traditional stories (Antigoon, Cnudde), or
salient landmarks (Manneken Pis). Linguistically interesting, to round off,
are those names where a local identity is evoked by dialect forms (which are
often at the same time archaic). The relevant linguistic features include in-
flection (Stoeren Bonk would be Stoere Bonk in standard Dutch), pronuncia-
tion (Duvel instead of dUivel), or spelling (as the typically archaic sch in
Oude Tongerschen Bruinen). These features may, of course, be combined
with each other and with values discussed before. The result may be quite
surprising: in particular, notice how the orthographical form Lowie Kators
for the name of the French king Louis Quatorze (cp. the source domain of
royalty and aristocracy) evokes a Dutch, dialect-like pronunciation, and as
such ironically combines the lofty position of the foreign king with the down-
to-earth status of the local dialect.
Table 3. Recurrent associative values in beer names
Beer is a prestigious entity
Wine tenninology
Royalty and aristoc-
racy; VIPs
Historical figures
Wealth and riches
Vlaamse Bourgogne, Cuvee Chateau des Flandres,
Caves, Rose de Cambrinus
Ada Royal, Grand Monarque, King's Ale, Regal, Rid-
der, Royal pils, Marquise
Monseigneur, Moeder Overste, Kapittel abt
Ambiorix, Artevelde, Breydel en de Coninckbier, Her-
tog Jan pils, Charles Quint
Breugelbier, Rubensbier, Rembrandt's bier, Vondel,
Mercator
Lowie Kators, Napoleon, Da Vinci
Fortuyn, Gold Pils, Gouden Carolus, Parel Pils, Silver
Pils, Safir, Toison d'or
Beer is an adventure at the edge ofthe accepted norms
Transgression of
religious norms
Transgression of
social norms
Virility and force
Strong bodily effects
Duivels bier, Duvel, HopduveL Judas, Lucifer, La Pe-
cheresse, Satan, Verboden Vrucht, Sorcieres
d'Ellezelles
Boucanier, Piraat, Smokkelaar, Schavuit, Stouterik,
Averechtsen, Bokkereyer, Deugniet, Bakelandt, Bri-
gand, Kerelsbier, Voyoll
Het geheim van Jan de Sterke, Kastaar, Bink, Brugse
Straffe Hendrik, Stoeren Bonk, De Soldaat
Delirium Tremens, Houten Kop, Katerke, Killer Pils,
Mort Subite
Beer is a source ofpleasure
Carpe diem Bacchus, Cupido, Drongens plezierke, Flierefluiter,
Joyeuse blonde, Lekkerbek, Pallieter, Plaisir, Relax,
Cote d'azur speciale
Heavenly pleasure Engeltjesbier, Bons voeux, Sanctus, Onze vader z'n
bier + various names of saints
Beer constitutes a link with the traditional culture ofthe pre-industrial era
Traditional methods
of production, old
trades
Rural origins
History and prehis-
tory
Religion
Brouwmeester, Gildenbier, Keurbier, Traditions des
Moines, Oerbier, Retro oude methode, Natuurbier,
Saison 1900
A la fenne, Bosbier, Boskeun, Verte campagne, Fenne
de Grand Pre, Pays vert, Peerdevisscher, Oelens boerke
Antiek, Vieux Temps, Druide, Blonde du Menhir,
l'Iguanodon; and see below ('Local history') and before
('Historical figures')
as before: Monseigneur, Moeder Overste, Kapittel abt
Engeldesbier, Bons voeux, Sanctus, Onze vader z'n
bier
+ various names of saints
Beer is an aspect oflocal identity
Dialect forms and
archaic forms
Local history
Local folkore
Duvel, Lowie Kators. Stoeren Bonk, Peerdevisscher,
Drongens plezierke, Oude Tongerschen Bruinen
Ambiorix, Artevelde, Bakelandt. Breydel- en de
Coninckbier, Kerelsbier, Hertog Jan pils
Breugelbier, Rubensbier. Rembrandt's bier, Vondel,
Mercator
Gouden Carolus, Toison d'or
Elckerlyc, Pallieter, Prutske bier, Hercule Poirot
Antigoon, Cnudde kriek, Het geheim van Jan de Ster-
ke, Ketje, Rosse Lei, Rooie Rietje Stekselbier, Tineke
van Heule, Manneken Pis, Geitenbier
There are three additional remarks to be made with regard to this over-
view of associated values. First, it is not always easy to get the etymological
origin of the names right, or rather, the actual etymological origin is not
necessarily the one that is relevant for the value that seems to be evoked now
by the name. For instance, Mort Subite as 'sudden death'. To the extent that
it is interpretable at all, it could only be in tenns of the physiological impact
of strong beers. Originally, however, the name was that of a particular cafe
plus brewery in Brussels; the name of the cafe itself apparently referred to a
particular type of backganunon popular in the pub. Some methodological
caution, in other words, is necessary when dealing with this kind of material.
As a second remark, it will be noted that the list contains both French and
Dutch names. In principle, it might have been possible to analyze both sets
of names separately, on the assumption that we might then get an insight in
the differences between the culture of the Dutch speaking part of Belgium
(Flanders) and its French speaking part (the Walloon country). I will, how-
ever, proceed on the assumption that, as far as beer goes, Belgian culture is
fairly uniform, This is supported by the recognition that there is no unambi-
guous link between the language of the brad name and the geographical posi-
tion of the brewery. Some French names, for instance, derive from breweries
in Flemish territory. Moreover, it appears from sociological surveys (most
recently, Kerkhofs, Dobbelaere, Voye, and Bawin-Legros 1992) that the
opinions and behavior that distinguish Flemings from Walloons are by far
outweighed by what they have in conunon in contrast with the Dutch, Ger-
mans, or French (to name only the inunediate neighbors).
A third remark is simply that a specific name may be motivated by vari-
ous evocative values at the same time, as has become apparent at various
points in the preceding discussion. A name like Ambiorix, for instance, re-
fers to the leader of one of the Celtic tribes that rebelled against Julius Cae-
sar's invasion of this part of Gaul. In this sense, the name Ambiorix com-
bines the motifs of local identity, of prestigious leadership, of the far
historical past, and of resistance to an imposed authority.
The fourth remark is the crucial one. If beer is as central to the Belgian
self-conception as I claimed earlier, do these aspects of Belgium's concep-
tion of beer reveal something of Belgium's self-conception? Methodologi-
cally, answering such a question requires going beyond linguistics and coop-
erating with sociologists, historians, and anthropologists. That is more than
could be done in the space of this paper, but let us note that linguistics is not
entirely unequipped to deal with the question. At least one thing we can do
as linguists is to compare the Belgian situation with that in other countries.
Taking The Netherlands as a point of comparison, some differences do ap-
pear that support the idea that the pattern that emerges for Belgium is a cul-
turally specific one. For instance, while the prestige motif is as outspoken in
The Netherlands as it in Belgium (which may be due to a universal conuner-
cial strategy of appraisal), the link with the pleasure motif is much weaker,
and references to local identity (at a level below that of the country as a
whole) are rather scarce. So are references to religion. Most tellingly, per-
haps, references to anti-authoritarian transgressive behavior are almost en-
tirely absent.
These differences can be meaningfully interpreted against the background
of the different cultural and political history of both countries. (These differ-
ences can only be summarily evoked here. For more background informa-
tion, see Verdoodt 1978, where a sketch of the cultural and political history
of the Low Countries provides the backdrop for a sketch of the divergent
linguistic evolution of Belgian Dutch and Netherlandic Dutch.) First, the
protestant, calvinistic religion that was dominant in Holland from the six-
teenth century on did not favor the enjoyment of earthly goods, while the
Catholic religious atmosphere in Belgium was much less austere. This may
explain why the link with the religious motif and with the pleasure motif is
almost absent in the brand names found in The Netherlands. Second, Hol-
land became a nation, a unified national state in the seventeenth century,
while the Belgian provinces remained under foreign (Spanish, Austrian,
French) influence until the nineteenth century. This correlates with the fact
that Dutch people identify more readily with the national state, whereas the
inhabitants of Belgium traditionally have a genuinely irreverent attitude with
regard to the political authorities at large. Apparently, then, this difference in
mentality also shows up in the different patterns of associated values for
beer names in both countries.
6. Concluding remarks
Possible conclusions to be drawn from this field trip through lexical beer
country include the following. First, a systematic investigation into salience
phenomena in lexical fields should be high on the list of methodological pri-
orities of Cognitive Linguistics, if it is to reach its aim of doing cultural
analysis by linguistic means. Second, in analyzing such salience phenomena,
a distinction has to be maintained between the salience of 'literal' referential
features and the salience of associated values. In the foregoing pages, this
distinction showed up in the distinction between 'Type l' and 'Type 2'
names. Third, proper names (and in particular, commercial brand names) are
amenable to a cognitive analysis in terms of recurrent associative and meta-
phorical values. Fourth, the wealth and diversity of Belgian beers is reflected
in the variety of beer names; moreover, there is at least some evidence that
the semantic patterns that emerge from the linguistic analysis correspond
with real culturally specific aspects of behavior and mentality. And fifth, the
lexicology of beer names can be fun - but whether it is more fun than the
tasting of beer itself is for the reader to explore.
Chapter 11
Cultural models of linguistic standardization
Originally published in R e l l ( ~ DilVen, Roslyn Frank and Martin Piitz (eds.), 2003,
Cognitive Models in Language and Thought. Ideology, Metaphors and Aleanings
25-68. Berlin/New York: Mouton de Gruyter.
What are the models that shape our thinking about language as a social phenome-
non? What are the paradigms that we use to think about language, not primarily
in terms of linguistic structure, but in terms of linguistic variation: models about
the way in which language varieties are distributed over a language community
and about the way in which such distribution should be evaluated? In this paper, I
argue that two basic models may be identified: a rationalist and a romantic one. I
chart the ways in which they interact, describe how they are transformed in the
course of time into a nationalist and a postmodern model, and explore how the
models can be used in the analysis of actual linguistic variation.
The paper constitutes a link between the type of semantic studies illustrated in the
previous chapters, and the variationist, sociolinguistically oriented studies that
grew out of the shift towards an onomasiological form of lexicology as described
in Chapter 4 and elsewhere in this volume. There are various substrands to this
type of research, which is the main focus of our Leuven research group Quantita-
tive Lexicology and Variational Linguistics: sociolectometrical research into the
relationship between language varieties (as illustrated in the final part of the pre-
sent paper, and in studies such as Geeraerts, Grondelaers and Speelman 1999;
Geeraerts 200la, 200lb; Grondelaers, Geeraerts, Speelman and Tummers 2001;
Geeraerts 2002a; Speelman, Grondelaers and Geeraerts 2003; Van Gijsel, Geer-
aerts and Speelman 2004), statistically advanced, corpus based analyses of multi-
factorial sociolinguistically sensitive phenomena (as in Grondelaers, Brysbaert,
Speelman and Geeraerts 2002; Grondelaers, Speelman and Geeraerts 2002; De
Sutter, Speelman, Geeraerts and Grondelaers 2003; Tummers, Speelman and
Geeraerts 2004), and theoretical arguments for incorporating social variation into
the scope of Cognitive Linguistics (Geeraerts 2002b, 2005).
1. Cultural models and language variation
In line with well-known trends in cultural theory (see Burke, Crowley and
Girvin 2000), Cognitive Linguistics has stressed the idea that we think about
social reality in terms of models - 'cultural models' or 'folk theories': from
Cultural models ofstandardization 273
Holland and Quinn (1987) over Lakoff (1996) and Palmer (1996) to Dirven,
Hawkins and Sandikcioglu (2001) and Dirven, Frank and Hie (2001), cogni-
tive linguists have demonstrated how the technical apparatus of Cognitive
Linguistics can be used to analyze how our conception of social reality is
shaped by underlying patterns of thought. But if language is a social and
cultural reality, what are the models that shape our conception of language?
Specifically, what are the models that shape our thinking about language as
a social phenomenon? What are the paradigms that we use to think about
language, not primarily in ternlS of linguistic structure (as in Reddy 1979),
but in terms of linguistic variation: models about the way in which language
varieties are distributed over a language community and about the way in
which such distribution should be evaluated?
In this paper, I will argue that two basic models may be identified: a ra-
tionalist and a romantic one. I will chart the ways in which they interact,
describe how they are transformed in the course of time, and explore how the
models can be used in the analysis of actual linguistic variation.
There are two preliminary remarks that I should make in order to situate
the present paper against a wider background. First, the analysis is a mar-
ginal offshoot of a more central interest in empirical methods for studying
linguistic variation and change. The work that I have been doing over the
last ten years or so with my research group has specifically focused on vari-
ous aspects of lexical variation and change: diachronic semantics (Geeraerts
1997a), the relationship between semantic and lexical variation (Geeraerts,
Grondelaers, and Bakema 1994), and lexical variation within pluricentric
languages such as Dutch (Geeraerts, Grondelaers and Speelman 1999).
Within the latter line of research, we have been particularly concerned with
the development of quantitative techniques for measuring lexical variation
and processes of lexical standardization (see section 4 below).
There are two ways, then, in which the present more or less essayistic
paper links up with the more rigorous descriptive and methodological work
that is my basic field of interest. For one thing, an investigation into linguis-
tic usage needs to be complemented by an investigation into the way in
which the users of the language perceive the actual situation. The cultural
models that I will be talking about define, in a sense, basic language atti-
tudes - and an adequate interpretation of language variation should obvi-
ously take into account language attitudes along with language behavior.
At the same time, both perspectives (the behavioral and the attitudinal)
have links with Cognitive Linguistics. Whereas the attitudinal approach
draws inspiration from the Cognitive Linguistic analysis of cultural models
and folk theories, the descriptive approach is a further development of the
Cognitive Linguistic interest in lexical-semantic variation as represented by
prototype theory. Underlying the publications mentioned above is a logical
line of development from semasiological prototype theory (Geeraerts 1997a)
to a model of lexical variation encompassing onomasiological variation
(Geeraerts, Grondelaers, and Bakema 1994), which then further broadens to
the investigation of 'external', sociolectal and dialectal factors of variation
(Geeraerts, Grondelaers and Speelman 1999).
As a second preliminary remark, is there a difference between a 'cultural
model' and an ideology? It is a common idea in Cognitive Linguistics that
the cultural models underlying reasoning and argumentation are to some
extent idealized entities (see, for instance, the notion of ICM's or 'Idealized
Cognitive Models' as introduced in Lakoff 1987). Actually occurring phe-
nomena and situations usually differ to a smaller or a greater extent from the
models that act as cognitive reference points: the models themselves, then,
are to some extent abstract, general, perhaps even simplistic, precisely be-
cause we use them to make sense of phenomena that are intrinsically more
complicated.
With regard to social phenomena, this means that cultural models may
turn out to be not just idealized entities, but also ideological ones. Cultural
models may be ideologies in two different respects: either when their ideal-
ized character is forgotten (when the difference between the abstract model
and the actual circumstances is neglected), or when they are used in a pre-
scriptive and normative rather than a descriptive way (when they are used as
models of how things should be rather than of how things are). In the latter
case, an ideology is basically a guiding line for social action, a shared sys-
tem of ideas for the interpretation of social reality, regardless of the re-
searcher's evaluation of that perspective. In the former case, an ideology is
always to some extent a cover-up, a semblance, a deliberate misrepresenta-
tion of the actual situation, and a description of such ideologies will of ne-
cessity have to be critical.
The distinction is of course well-known in ideology research, and there is
an extensive linguistic literature probing the relationship between language
and ideology. There are two basic (and to some extent overlapping) ap-
proaches here: on the one hand, all forms of critical discourse analysis, as
represented by Van Dijk (1998), Wodak and Meyer (2001), or Blommaert
and Bulcaen (1997); and on the other, the 'ideologies oflanguage' approach,
as represented by Joseph and Taylor (1990), Woolard, Schieffelin and
Kroskrity (1998), and Schiffman (1996). The former approach critically
analyzes any text with regard to its position in the social power play - with
regard to the way, that is, in which it reproduces or counteracts existing
social relations. The latter approach concentrates on how beliefs about lan-
guage variation and specific linguistic varieties manifest themselves explic-
itly (as in language policies) or implicitly (as in educational practices), and
how they interact with group identity, economic development, social mobil-
ity, political organization.
In the following pages, I will not take a critical approach, but rather start
from a neutral and descriptive conception of linguistic cultural models.
Rather than critically analyzing specific practices and policies as ideological,
I will try to explore the underlying structure and the historical development
of the competing cultural models that lie at the basis of such practices and
policies as well as their critical analysis.
2. The rationalist and the romantic model
In this section, I will present the two basic cultural models that I think need
to be distinguished if we want to get a grip on the logic of standardization
debates: the rationalist one and the romantic one. I will present them in mu-
tual contrast, showing how they are to a large extent each other's counter-
part, and how they are dialectically related. The present section will not
however exhaust the comparison between both models. In the next section,
the comparison will be further expanded, leading to the identification of two
historical transformations of the basic models, in the form of a nationalist
and a postmodem model.
2.1. The rationalist model
So what are the characteristics that are ideally (and perhaps ideologically)
attributed to standard languages? The most conspicuous feature is probably
the generality of standard languages. Standard languages, in contrast with
dialects and other restricted languages, are general in three different ways.
They are geographically general, in the sense that they overarch the more
restricted areas of application of dialects. Further, they are socially general
because they constitute a common language that is not the property of a
single social group but that is available to all. Finally, they are thematically
universal in the sense that they are equipped to deal with any semantic do-
main or any linguistic function. More advanced domains of experience in
particular (like science or high culture) fall outside the range of local dia-
lects.
Because of their generality, standard languages have two additional fea-
tures. First, they are supposed to be a neutral medium, with a mediating
function, in an almost philosophical sense of 'mediation'. Standard lan-
guages, in fact, transcend social differences: they ensure that men and
women from all walks of life and from all corners of the nation can commu-
nicate freely.
In that sense, they are a medium of participation and emancipation. Be-
cause of their neutrality and because of their functional generality, standard
languages are a key to the world of learning and higher culture: functional
domains par excellence for standard language use (or, reversing the perspec-
tive, functional domains that cannot be accessed on the basis of dialect
knowledge alone). Perhaps even more importantly, standard languages are
supposed to contribute to political participation. The possibility of free
communication is a feature of a democratic political organization, in the
sense of the ideal 'herrschaftsfreie Kommunikation' as described by Jurgen
Habermas. If then linguistic standardization contributes to mutual under-
standing and free communication, it is a factor of political emancipation -
just as it is a factor of social emancipation when it contributes to the spread-
ing of culture and education. By contrast, if you believe in the beneficial
effects of standardization, dialects are mere relics of an obscurantist social
and political system that opposes democracy and emancipation.
In a context of postmodern ideological debunking, such a positive con-
ception of standardization is definitely suspect, but it is crucial for my line of
argumentation that at least in the context in which it originated (that of the
eighteenth-century Enlightenment), there was a genuine positive appraisal of
standardization. To illustrate, let us have a look at some excerpts from re-
ports presented to the revolutionary Convention in France. Barere (1794
[1975]) puts matters as follows.
(1) Citoyens, la langue d'un peuple libre doit etre une et la meme pour
tous. (1794 [1975]: 297)
[Citizens, the language of a free people has to be one and the same for
all.]
(2) Les lumieres portees a grands frais aux extremites de la France
s'eteignent en y arrivant, puisque les lois n'y sont pas entendues. (1794
[1975]: 295)
[The lumieres, when they are brought with great difficulty to the remote
corners of France, die out when they arrive there, because the laws are
not understood.]
(3) Laisser les citoyens dans I'ignorance de la langue nationale, c'est trahir
la patrie; c' est laisser le torrent des lumieres empoisonne ou obstrue
dans son cours; c'est meconnaitre les bienfaits de l'imprimerie, car
chaque imprimeur est un instituteur public de langue et de legislation.
(1794 [1975]: 296-297)
[To maintain the citizens in their ignorance of the national language is
to betray the country. It permits the torrent of the lumieres to be poi-
soned or obstructed in its course. It means disavowing the blessings of
the printing press, because all publishers are public teachers of the lan-
guage and the legislation.]
(4) Citoyens, les tyrans coalises on dit: l'ignorance fut toujours notre auxi-
liaire le plus puissant; maintenons l'ignorance; elle fait les fanatiques,
elle multiplie les contre-revolutionnaires; faisons retrograder les Fran-
<;:ais vers la barbarie: servons-nous des peuples mal instruits ou de ceux
qui parlent un idiome different de celui de l'instruction publique. (1794
[1975 ]: 291)
[Citizens, the allied tyrants have said: ignorance has always been our
most powerful helper. It creates fanatics, it breeds counter-
revolutionaries. Let's make sure the French degrade into barbarity: let's
take advantage of the badly educated peoples or of those that speak a
language that is different from that of public education.]
(5) Les habitants des campagnes n'entendent que le bas-breton; c'est avec
cet instrument barbare de leurs pensees superstitieuses que les pretres et
les intrigants les tiennent sous leur empire, dirigent leurs consciences et
empechent les citoyens de connaitre les lois et d'aimer la Republique.
Vos travaux leur sont inconnus, vos efforts pour leur affranchissement
sont ignores. (1794 [1975]: 292-293)
[The inhabitants of the countryside speak only the Breton dialect. It is
with that instrument of their superstitious way of thinking that the
priests and the plotters keep them under their thumb, control their
minds, and prevent the citizens from knowing the laws of the Republic.
Your works are unknown to them, your efforts to bring them liberty are
ignored.]
The characteristics that we have attributed to standard languages (gener-
ality and communicative neutrality, emancipatory and participatory effects,
opposition to obscurantism) can be easily identified in these fragments.
Fragment (1) expresses the generality and uniformity of the standard lan-
guage. Fragments (2) and (3) stress the emancipatory function of knowledge
of the standard: citizens who only know their dialect will not understand the
laws of the Republic (the assumption being, of course, that these have a
liberating effect), nor will they, more generally speaking, be able to profit
from the benefits brought by the printed press. Fragments (4) and (5) associ-
ate dialects more directly with counter-revolutionary obscurantism: it is sug-
gested that priests and 'tyrants' deliberately maintain ignorance by prevent-
ing the common people from acquiring the standard language.
A similar pattern can be found in the following quotes from Gregoire
(1794 [1975]), who actually presents an entire educational project to the
Convention to 'abolish the dialects and generalize the use of the French lan-
guage'. (His notion of 'dialect' actually includes not just the dialects of
French, but also the different languages spoken in the territory of France,
like German in the Alsace region, Flemish in the northern area, or Breton in
Brittany.)
(6) Mais au mains on peut unifonner le langage d'une grande nation, de
maniere que tous les citoyens qui la composent puissent sans obstacle se
communiquer leurs pensees. Cette entreprise, qui ne fut pleinement
executee chez aucun peuple, est digne du peuple f r a n ~ a i s , qui centralise
toutes les branches de I'organisation sociale et qui doit etre jaloux de
consacrer au plut6t, dans une Republique une et indivisible, 1'usage
unique et invariable de la langue et de la liberte. (1794 [1975]: 302)
[But at least one can standardize the language of a great nation, to the
extent that all its citizens can mutually communicate their thoughts un-
hindered. Such an enterprise, which no people has fully achieved as yet,
is worthy of the French nation, which centralizes all aspects of the so-
cial organization and which must endeavor to endorse as soon as possi-
ble, in a Republic that is one and indivisible, the sole and invariable use
of language and freedom.]
(7) 'Il y a dans notre langue, disait un royaliste, une hierarchie de style,
parce que les mats sont classes comme les sujets dans une monarchie'.
Cet aveu est un trait de lumiere pour quiconque reflechit. En appliquant
I'inegalite des styles cl celle des conditions, on peut tirer des consequen-
ces qui prouvent I' importance de man projet dans une dernocratie.
(1794 [1975]: 316)
['There is in our language, a certain royalist said, a hierarchy of styles,
because the words are classified just like the citizens in a monarchy'.
This confession constitutes a ray of insight for any thinking person. If
we apply the inegality of the styles to the inegality of the conditions un-
der which people live, we may come to conclusions that prove the im-
portance of my project (of linguistic standardization through an educa-
tionallanguage policy) in a democracy.]
(8) Tous les membres du souverain sont admissibles atoutes les places; il
est adesirer que tous P4issent successivement les remplir, et retourner a
leurs professions agricoles ou mecaniques. eet etat de choses nous pre-
sente l'alternative si ces places sont occupees par des hommes
incapables de s'enoncer, d'ecrire dans la langue nationale, les droits des
citoyens seront-ils bien garantis par des actes dont la redaction presen-
tera l'impropriete des termes, l'imprecision des idees, en un mot tous
les symptomes de l'ignorance? Si au contraire cette ignorance exclut
des places, bientot renaitra cette aristocratie qui jadis employait le pa-
tois pour montrer son affabilite protectrice aceux qu'on appelait inso-
lemment les petites gens. [... ] Ainsi l'ignorance de la langue compro-
mettrait le bonheur social ou detruirait l'egalite. (1794 [1975]: 303)
[All members of the sovereign people are eligible for all positions. It is
desirable that all may successively fill these positions, and afterwards
return to their agricultural or industrial professions. This state of affairs
yields the following alternative. If the positions are taken up by men in-
capable of expressing themselves or of writing in the national language,
will the rights of the citizens be safeguarded by laws that are character-
ized by improper choice of words, by imprecise ideas, in short by all
symptoms of ignorance? If on the contrary this ignorance prevents peo-
ple from taking up office, then soon enough we will witness the rebirth
of that aristocracy that once used the dialects to demonstrate its affabil-
ity with regard to those that it insolently named 'the small people'. [... ]
Thus, ignorance of the language either compromises social happiness
or destroys egality.]
Fragment (6) points to the communicative generality of the standard lan-
guage: having a unitary language not only symbolizes the unity of the nation,
but it also ensures that the citizens can freely communicate their thoughts.
Fragment (7) symbolically links the absence of standardization to the pre-
revolutionary situation: the existence of hierarchically ordered varieties
within the language mirrors the hierarchical organization of society. Frag-
ment (8) aptly describes the politically emancipatory function of standardi-
zation. The egalitarian ideal implies that any citizen can take part in the
government of the nation; in fact, the ideal would be that all citizens succes-
sively fulfill political functions and then return to their professional envi-
ronment. However, in order to be able to fulfill these functions, a thorough
knowledge of the common language is necessary. People should not be pre-
vented from taking up office by their ignorance of the language. Hence, an
educational effort to ensure standardization is necessary: Gregoire is an ar-
dent defender of the 'Ecole publique' as a standardizing force.
In section 3, I will describe the transformations that the rationalist,
Enlightenment ideal of standardization goes through in the course of the last
two centuries. Even in its transformed shape, however, the positive evalua-
tion of standardization refers to one or another of the features mentioned
here: a neutrally mediating conmmnicative function, and an emancipatory
and participatory effect, both of these supported by an educational system
geared towards the spreading of the standard language.
2.2. The romantic model
The romantic conception of standardization may be easily defined in contrast
with the two dominating features of the rationalist model. First, as against
the emancipatory and participatory goals of the enlightened view, a romantic
view will tend to point out that standard languages are themselves instru-
ments of oppression and exclusion. At this point, of course, the analysis of
standardization takes the form of an ideological criticism: it will argue that
the enlightened ideals are not often realized, and that, in fact, processes of
standardization typically achieve the reverse of what they pretend to aim at.
Although the term is not often used, this type of critical discourse boils down
to a demonstration that linguistic standardization exemplifies what Hork-
heimer and Adomo (1947) called the 'Dialektik der Aufklarung' - the (nega-
tive) dialectic of Enlightenment. Horkheimer and Adomo argue that rational-
ist positions have a tendency to lead to their own dialectical counterpart (in
the sense, for instance, in which a growing technical mastery of man over
nature may lead to the destruction of the natural world).
Now, if we look back at the three types of generality that standard lan-
guages are supposed to characterize, it is easy to see that the actual realiza-
tion of the ideal may tend to contradict the ideal - which is then a case in
point of the 'Dialektik der Aufklarung'.
First, standard languages are supposed to be geographically neutral, but
in actual practice, processes of standardization often have their starting-
point in a specific region that is economically, culturally, and/or politically
dominant. For people in the other, outer provinces, then, the standard lan-
guage is not an impartial medium, but it rather affirms the dominance of the
leading province. Standard French, for instance, is not just an unbiased lan-
guage coming out of the blue; it is the language of the upper and the middle
classes of Paris and the Ile-de-France, and it is associated with the role that
the central province has played since the medieval era.
Second, standard languages are supposed to be functionally general, but
in actual practice, they are typically used in cultural, educational, scientific,
administrative, and political contexts - at least in those circumstances in
which a language community is not entirely standardized. Non-standard
varieties may then naturally acquire additional, contrastive overtones. For
one thing, if the standard language is the language of public life, the non-
standard varieties will be appreciated as the language associated with inti-
macy, familiarity, the personal rather than the public sphere. For another, if
the standard language functions in typically intellectual contexts (education
and science), non-standard varieties will be invested with emotional values.
For speakers of a dialect, the dialect is often the language of the emotions, of
spontaneity, of naturalness, in contrast with the official and educational
language. Ironically, the functional generality of standard languages engen-
ders a functional specialization, separating the public sphere from the per-
sonal, and the emotional sphere from the intellectual.
Third, standard languages are supposed to be socially neutral, but in ac-
tual practice, they are typically the language of an elite. The link between an
economical, cultural, or political elite and the standard language is in fact an
inevitable side-effect of the functional generality of standard languages. If
standard languages are typically used in cultural, educational, scientific,
administrative, and political contexts, then those speakers of the language
that act in these contexts will more easily learn the standard language or
adopt it as their first language than speakers who remain foreign to these
functions. The outsiders may then perceive the greater linguistic proficiency
of the elite as a factor contributing to social exclusion. In Gregoire's view,
knowledge of the standard language contributes to social mobility, but con-
versely, the real social distribution of standard language functions may turn
the standard language into an instrument of discrimination.
We can see, in other words, how the alleged generality of standard lan-
guages actually takes the form of a series of specializations. The process of
standardization takes its starting-point in the language of specific regions,
specific groups of speakers, specific domains and functions, and this largely
inevitable fact may subvert the very ideal that standardization was supposed
to serve. When that happens, the original ideal may be critically unmasked
as an ideological pretence.
Needless to say, this dialectical reversal may also affect the educational
system. If the standard language is recognized as an instrument of oppres-
sion, discrimination, social exclusion, the educational system will likewise be
rejected as contributing to such processes of social exclusion. Rather than
seeing the school as an institution that spreads knowledge of the common
language (and knowledge in general), creating possibilities for social mobil-
ity, it will then be pointed out that the educational system, relying on perhaps
more than contributing to the knowledge of the language, favors those lan-
guage users whose background makes them more familiar with the standard
language, and thus reproduces rather than neutralizes social inequality.
But why call this critical reversal of the appreciation of the standard lan-
guage a 'romantic' model? Why not simply call it a 'realistic' or a 'critical'
or an 'anti-ideological' one? The reason is that this critical stance is often
(though not necessarily always) accompanied by a second feature, that may
be contrasted with the second characteristic of the rationalist model. That is
to say, we have just seen how a critical approach questions the emancipa-
tory, participatory conception of the Enlightenment model. But what about
the second feature? What about the communicative aspects of the rationalist
model?
We get a truly 'romantic' model of language variation when the critical
attitude towards official standards is coupled with a view of language as
expression rather than communication. According to the Enlightenment
perspective, languages are means of communication, and a standard lan-
guage is a superior communicative tool because it is functionally general and
socially neutral.
According to a romantic perspective, languages are primarily expressive
rather than communicative. They express an identity, and they do so because
they embody a particular conception of the world, a world view or 'Weltan-
schauung' in the sense of Herder. The link between this well-known roman-
tic conception of the relationship between language and thought and the
standardization debate will be clear. If languages or language varieties em-
body a specific identity, then a preference for one language or language vari-
ety rather than another implies that the specific identity of a specific group
of people is neglected or denied. Not recognizing the language is not recog-
nizing the language users. If some language varieties are relegated to second
rate status through the existence of a standard variety, then the speakers of
those language varieties are denied a fundamental right: the right to express
themselves in their own language - the only language, in fact, that could do
justice to their individual identity, according to the romantic conception of
the relationship between language and identity.
Cultural models oJstandardization 283
A correlate of this position is the positive evaluation of variety. Whereas
the rationalist approach cherished linguistic unifonnity as the symbolic ex-
pression of a free and open community in which all citizens have equal rights
to speech, the romantic approach values diversity as a recognition of a fun-
damental respect for different identities.
In short, a fully romantic view of language variation and linguistic stan-
dardization opposes the Enlightenment view of language as communication
with a view of language as the expression of an individual identity. It op-
poses the emancipatory and participatory rationalist ideal with a critical
view of standardization as a tool of discrimination and exclusion, and it op-
poses the positive appreciation of education as an instrument for the dis-
semination of linguistic knowledge with a fundamental distrust of schools as
part of a system reproducing social inequality.
In order to illustrate this model, I will not (as I did in the case of the ra-
tionalist model) use an historical example, but I would like to have a brief
look at the current debate about linguistic genocide and the international
position of English. In that interlinguistic fonn of variation, English replaces
the standard language of intralinguistic variation, and minority languages
threatened with disappearance replace the non-standard varieties. All the
objections that a romantic approach would level against a dominating stan-
dard variety, could then be applied against the international domination of
English. Consider, as an example, the following excerpts from an abstract of
Skutnabb-Kangas (2000) (The abstract, by the author herself, may be found
on the author's homepage).
(9) Indigenous peoples and minorities are the main bearers of linguistic
and cultural diversity in the world - over 80% of the world's languages
exist in one country only and the median language has no more than
5,000 speakers. Some of the direct main agents of linguistic (and cul-
tural) genocide today are parts of what we call the consciousness indus-
try: formal educational systems and the mass media. [... ] The book
shows that the education of most minorities and indigenous peoples in
the world is organized in ways which both counteract sound scientific
principles and lead to the disappearance of linguistic and cultural diver-
sity. [... ] Schools are every day committing linguistic genocide. [... ]
They also do it by forcibly moving children from one group (indigenous
or minority) to another group (the dominant group) through linguistic
and cultural forced assimilation in schools. [... ] This inevitably in-
cludes a consideration of power relations. The book shows how the
formal educational systems participate in maintaining and reproducing
unequal power relations, here especially between linguistic minorities
284 Aifeaning and culture
and others, but also more generally, and how the ways of doing this
have changed and are constantly changing, and how control and domi-
nation are resisted and alternatives are constantly created and negoti-
ated, managed and controlled, and recreated. The deficiency-based
models that are used in most minority education invalidate the linguis-
tic and cultural capital of minority children and their parents and com-
munities. They make the resources of dominated groups seem handi-
caps or deficiencies, instead of valued and validated non-material
resources, or they render them invisible and therefore not possible to
convert into material resources and positions of structural power. This
happens just as much in global international relations and the Mcdon-
aldization of the world as it happens in ESL classrooms.
Regardless of whether Skutnabb-Kangas is right or not (see Joseph
forthcoming for a critical reaction), the components of the romantic ap-
proach are conspicuously present in her statement. First, the disappearance
of languages and the disappearance of cultures are equated. The very notion
of 'linguistic genocide' in fact invokes the extermination of an entire people
(or at least culture) together with its language. As opposed to this process of
forced assimilation and disappearance, different cultures have to be accepted
as fundamentally equal, and diversity should be treasured as an end in itself.
Second, the international dissemination of English does not lead to emanci-
pation and participation, but rather serves purposes of international oppres-
sion, notably by multinational companies. And third, the text has explicit
misgivings about the role schools play in this linguistic and cultural power
play.
2.3. The rationalist and the romantic model and the origins of language
In the previous pages, I have not only sketched the rationalist and the roman-
tic model of standardization, but I have also indicated that they exhibit a
specific and narrow relationship. Not only is one the counterpart of the
other, but there is a dialectical relationship between the two, in the sense that
actual processes of standardization seem to be caught in a negative 'dialectic
of Enlightenment' in the course of which the positive rationalist ideals tend
to be subverted, thus giving way to a romantic, critical appreciation of the
standardization process. This dialectical relationship, which is summarized
in Table 1, does not however exhaust the links that exist between the two
basic conceptions.
Table I. The rationalist and the romantic models of standardization
THE RATIONALIST MODEL THE ROMANTIC MODEL
linguistic-philosophical
basis:
conception of standardi-
zation:
conception of language
variation:
language as a medium of
communication
a democratic ideal:
standard language as a
neutral medium of social
participation
language variation as
an impediment to eman-
cipation
language as a medium of
expression
anti-ideological criticism:
standard language as a
medium of social exclu-
sion
language variation as
expressing different iden-
tities
In this section, I will have a brief look at the theoretical linguistic back-
ground of the basic models: is there a specific conception of language that
goes hand in hand with either of the perspectives? I will argue that eight-
eenth-century theories about the origins of language complement the picture
painted so far. It was suggested above that the romantic and the rationalist
approach differ in their theoretical conception of language: the former starts
from an expressive conception of language and the latter takes its starting-
point in a communicative conception. The distinction between both perspec-
tives can get more relief if we have a brief look at some of the theories on the
origins of language that were formulated in the eighteenth century - in the
period, that is, in which we situate the intellectual sources of the rationalist
and the romantic conception of language variation. The expressive concep-
tion of the origins of language may be illustrated on the basis of Herder's
views, while Condillac may serve as an illustration of the conununicative
VIew.
Herder's conception of language as the expression of an inner view ofthe
world shapes the way in which he describes the birth of language: language
arises when man expresses his understanding of the world. In the following
passage, Herder explains how man, in giving names to the animals (in this
case, a sheep) identifies and expresses a specific feature (in this case, the
sheep's bleating).
(10) Weiss, sanft wollicht - seine besonnen sich iibende Seele sucht ein
Merkmal, - das Schaaf blocket! sie hat Merkmal gefunden. Der innere
Sinn wiirket. Dies Bl6cken, das ihr am starksten Eindruck macht, das
sich von alien andern Eigenschaften des Beschauens ul1d Betastens los-
riss, hervorsprang, am tiefsten eindrang, bleibt ihr. Das Schaaf kommt
wieder. Weiss, sanft wollicht - sie sieht, tastet, besinnet sich, sucht
Merkmal - es b16ckt, und nun erkennet sies wieder! 'Ha! du bist das
Blockende!' fuhlt sie innerlich, sie hat es Menschlich erkannt, da sies
deutlich, das ist, mit einem Merkmal, erkennet und nennet (1772
[1978]: 33)
[White, soft, woolly - its cautiously exercizing soul looks for a charac-
teristic feature. The sheep bleats! It has found a feature. The inner sense
is working. This bleating. which impresses it to the utmost, which de-
taches itself from all other visual or tactile features, which springs
forth, which makes an impression, stays with it. The sheep comes back.
White, soft, woolly - the soul sees, feels, reflects - the sheep bleats, and
now it recognizes it! 'Ha! You are the bleating one', it feels within it-
self, it has recognized it in human fashion, because it has recognized
and named the sheep clearly, that is, by means of a characteristic fea-
ture.]
Herder's story may be contrasted with Condillac's view about the origins
of language. Whereas Herder looks for the origins of language in a cogni-
tive, epistemological urge to identify things (and more specifically, to iden-
tify them in a specific way that reflects man's understanding of the things in
question), Condillac points to a communicative situation. In his story about
two children, he describes how a cry for help by one is recognized by the
other, and is gradually conventionalized as a specific sign.
(11) Celui qui souffroit parce qu'il etoit prive d'un objet que ses passions lui
rendoient necessaire, ne s'en tenoit pas apousser des cris: il faisoit des
efforts pout l'obtenir, il agitoit sa tete. ses bras. et toutes les parties de
son corps. L'autre. emu ace spectacle. fixoit les yeux sur le meme ob-
jet; et [... ] il souffroit de voir souffrir ce miserable. Des ce moment il se
sent interesse a le soulager, et il obeit acette impression, autant qu'il
est en son pouvoir. Ainsi, par le seul instinct, ces hommes se deman-
doient et se pretoient de secours [... ]. Cependant les memes circonstan-
ces ne purent se repeter souvent, qu'ils n'accoutumassent enfin aatta-
cher aux cris des passions et aux differentes actions du corps, des
perceptions qui y etoient exprimees d'une maniere si sensibles (1746
[1973]: 194-195).
[A person who suffered from being deprived of an object that his pas-
sions made necessary for him, did not restrict himself to shouting: he
made an effort to obtain it, shook his head and arms and body parts.
The other person. moved by this spectacle, fixed his gaze on the same
object, and suffered from seeing the miserable other person suffer. From
this moment on, he grows interested in coming to his relief, and he
obeys this sentiment to the best of his powers. In this way, by instinct
alone, men mutually ask and give help. [... ] But such events could not
repeat themselves often before people got used to attaching to the pas-
sionate cries and bodily actions the views that they expressed so tangi-
bly.]
The difference between Herder and Condillac correlates with a difference
in their view of the relationship between language and culture. Unlike
Condillac, Herder does not think it possible to explain the origins of human
language from instinctive emotional cries. Understanding needs to be present
before human, conscious language can arise. In Condillac's account under-
standing or 'reflection' takes shape gradually, through the use of signs in a
communicative situation. For Herder, the birth of language presupposes the
existence of a human culture, a culture that is being expressed in the lan-
guage. From Condillac's point of view, human culture develops from in-
stinctive beginnings through the use of signs and the development of lan-
guage. (Compare Sapir 1907 for a more elaborated view.)
If it is indeed correct that the intellectual origins of our cultural models of
linguistic standardization have to be sought in the eighteenth century, quota-
tions (l 0) and (ll) suggest that there was a link with the linguistic theorizing
of that period. This recognition opens the way towards the broader historical
study of the relationship between theories in linguistics and the philosophy of
language on the one hand, and language policies on the other. To what extent
have theories about the origin, nature, structure of language influenced the
way in which people (and governments) have thought about matters of lan-
guage variation and linguistic standardization? The existing literature offers
many points of departure but mostly concentrates on standardization proc-
esses and language policies within a single language or nation. One interest-
ing way to use the cultural models identified above, could be to see whether
they can be used as the basis for a more synthetic view of these historical
links between linguistic theory and standardization.
3. The historical transformation of the models
The present section will have a look at various distinctive moments in the
development of the competing models, in particular charting the transforma-
tions that they go through in the nineteenth and the twentieth centuries. First,
I will argue that the nationalist model of standardization that rose to promi-
nence in the nineteenth century constitutes a specific blend of the rationalist
and the romantic model. Further, I will have a look at the way in which our
contemporary postmodern awareness influences the competition between the
rationalist and the romantic model. It is an interesting question, by the way,
whether the models ever occur in their purest form. Even in the examples
from the French revolutionary period, a link with patriotic nationalism is
present. If this is indeed the case, the models presented in the previous sec-
tion are to be seen as analytic reference points, as 'idealized cognitive mod-
els' in the sense of Cognitive Linguistics.
3. 1. The nineteenth century and the nationalist model
Both the rationalist and the romantic model have a problem with the level at
which they should be situated. If the rationalist model is carried to its ex-
treme, it implies the necessity of a universal, international language. If the
driving force behind standardization is maximizing mutual communication,
then a universal language that transcends all existing language variation is to
be recommended: the neutralization of interlinguistic variation complements
the neutralization of intralinguistic standardization. And of course, the ideal
of a universal, ideal language (Esperanto, Volapilk and the like) is precisely
the historical realization of this consequent interpretation of the rationalist
approach.
In actual practice, however, Esperantist movements and the like remained
marginal. The real level at which standardization processes took place, lay at
a lower level - that of the nation. Starting from the Enlightenment model,
there is a simple logic to this (which can, in fact, be identified in the quota-
tions from Barere and Gregoire that we discussed): if standardization aims at
democratic, political participation, then obviously the nation, as the ideal
form of political organization, becomes the locus of standardization proc-
esses and the educational efforts supporting them. In itself, then, a link be-
tween nationalism and the rationalist view of standardization cannot come as
a surprise. Linguistic standardization is primarily standardization within a
nation, because it is within the nation that the processes of political decision
making take place that linguistic standardization is supposed to contribute
to. A terminological clarification may be useful at this point. 'Nationalism'
is the political ideology in which a state, as a political organization, derives
its political legitimacy from its people, rather than from tradition, divine
right, or the like. A state that lives up to this requirement is a nation. Nation-
alism, in other words, claims that any state should be a nation. The national-
ist relationship between the people and the state may be conceived of in two
different ways: according to a distinction that is customary in the literature
on nationalism, we may make a distinction between civic nationalism and
identity nationalism. On the one hand, civic nationalism is the conception of
nationalism in which the nation derives its legitimacy from the active par-
ticipation of its citizens, through a system of political representation. This is
the liberal, rationalist conception of nationalism. On the other hand, identity
nationalism is the conception of nationalism in which the nation derives its
political legitimacy from the cultural identity of the people. This is the ro-
mantic conception of nationalism.
'Nationalism' also refers to the claim and the efforts of a particular
group to become a nation. Existing states are not necessarily nations accord-
ing to the nationalist view: either because they do not achieve democratic
legitimacy (the liberal point of view), or because they do not recognize the
cultural identity of certain groups (the romantic point of view). Historically
speaking, then, 'nationalist movements' may be either movements trying to
establish a liberal democracy, or movements claiming independence for a
specific group or region. (In contemporary usage, though, the focus tends to
lie more on the latter type.)
The link between nationalism and language that we described above
clearly involves the liberal, rationalist version of nationalism: if the nation
derives its legitimacy from the active participation of its citizens, then
maximizing mutual communication through standardization is an instrument
of participation. But if we turn to identity nationalism, nationalism has a
similar, and maybe even stronger link with a romantic conception of lan-
guage. Whereas the rationalist perspective contains a tendency towards uni-
versality, the romantic perspective has a tendency towards individuality. If
carried to its extreme, the romantic conception of language variation implies
that each person may have his or her own language. Just like the rationalist
perspective tends to maximize communicability, the romantic perspective
tends to maximize individual variation. Again, in actual practice, this is an
extreme position that can hardly be realized as such. Except perhaps in the
romantic admiration for the 'individual voice of the poet' and the like, the
romantic conception deals with the language of groups rather than with the
language of individuals. The identity that is expressed by the language is the
identity of a community, and the community is a nation when it acquires
political autonomy. Hence the well-known romantic link between national-
ism and language: see, among many others, Deprez and Vos (1998). On the
one hand, language correlates with identity according to the romantic model,
290 Aleaning and culture
and on the other, nations may derive their legitimacy from the cultural iden-
tity of the people (which is not to say that all nationalism is linguistic na-
tionalism: as is well known, the sense of identity may come from many other
sources, like religion or ethnicity).
From two different angles, then, nationalism links up with language, and
this recognition may be linked to the distinction between two basic types of
nationalism that is often made in political theory. On the one hand, civic
nationalism is the conception of nationalism in which the nation derives its
legitimacy from the active participation of its citizens, through a system of
political representation. In such a liberal, rationalist conception, the common
language is the medium of participation. On the other hand, identity nation-
alism is the conception of nationalism in which the nation derives its politi-
cal legitimacy from the cultural identity of the people, and language is one of
the factors establishing such identity.
The actual alliance between both forms of reasoning may be briefly illus-
trated by the following quotes from Verlooy (1788). A Dutch-speaking law-
yer in Brussels, which was then under Austrian rule, Verlooy argues against
the growing use of French in public life and in favor of the use of the native
Dutch tongue. In 1789, Verlooy played a role in the 'Brabantse Omwen-
teling', an (ineffective) insurrection against the Austrians. His pamphlet of
1788 may be read as the intellectual basis of his nationalist stance of the
next year. But what is the role attributed to language in Verlooy's national-
ism?
(12) Het is zonder twyffel een goed voor eenigelyk wel ter tael en ter spraek
te zyn, en zyne redens vaerdig en onbelemmert te voeren. Doch hier toe
is een' zekere frankheyd noodig. Maer, gelyk by ons gezien en geplo-
gen is, wanneer wHen wy frank zyn in die vremde tael? (1788 [1979]:
58)
[Without any doubt it is good for any person to be able to speak flu-
ently, and to engage in conversation freely. But to achieve this a certain
candor is necessary. However, as can be observed in our case, when will
we obtain such candor in this foreign language?]
(13) Door ons frans schynen wy van die middelbare geleertheyd en borger-
Iyke wysheyd af geheeI het gemeyn, onze bestgemoedde en weetgirige
borgers, ambachtslieden, akkennans, en onze vrouwen: die 't frans
teenemael niet, of ten minsten zoo verre niet en weten, dat-ze 't met
vennaek of zonder moyelykheid konnen lezen: die daer door als als ge-
doemt schynen tot een' gezogte onwetendheyd. (1788 [1979]: 49)
[By speaking French, we separate from this common knowledge and
this civic wisdom all the common people, our well-humored and in-
quisitive townsmen, craftsmen, farmers, and our women: who do not
know French, or at least not well enough to read it easily and effi-
ciently, and who therefore seem to be condemned to ignorance.]
(14) Voor het vaderlanderschap eener natie is zeer dienstig zoo veel eygen
en bezonder te hebben als mogelyk is (... ) en zelfs hoe meer een' zaek
uytwendigs heeft, gelyk de taeL dragten, toneelen, godsdienst, zeker
plechten: hoe meer zy de gemoederen van 't volk zal aentrekken. [... ]
Waerom werken wy dan om zoo bekwamen band van vaderlander-
schap, de moederlyke tael, te bannen? (1788 [1979]: 59-60)
[For a feeling of national identity within a nation, it is useful to have as
many common and specific features as possible, and these features will
more readily attract the hearts of the people to the extent that they can
be externally observed, like the language, the attire, the theater and the
public entertainments, the religion. Why then do we endeavor to dis-
card our mother tongue, which constitutes such a strong tie of patriot-
ism?]
Quotation (12) emphasizes the individual and emancipatory perspective:
it is important for people to be able to express themselves freely, and this
can only be guaranteed in their mother tongue. In the same vein, quotation
(13) stresses the importance ofa common language for an open communica-
tion within a given society and for the dissemination of knowledge: the fur-
ther use of French would engender an undesirable rupture between the mid-
dle classes and the lower classes. By contrast, quotation (14) stresses the
importance of a common identity for nation-building. Both rationalist and
romantic themes, in other words, may appear in the discourse of proponents
of nationalist movements.
With the development of the national movements in the nineteenth cen-
tury, though, the nationalist emphasis tended to fall more and more on the
romantic notion of national identity. Minorities aspiring towards independ-
ence naively assume or explicitly construct an identity, and nation states may
blatantly enforce a common identity, linguistic or otherwise. These processes
are well known from the nationalism literature (in the line of Hobsbawm,
Anderson, Gellner, Smith). For our present linguistic purposes, the crucial
point is to see that this romantic nationalism reveals the paradoxes of the
romantic cultural model that we identified above. The transition from the
romantic model as described earlier to the nationalist model constitutes so to
speak a 'Dialektik der Romantik' that parallels the 'Dialektik der Aufk-
Hirung', i.e. an almost natural process through which the original romantic
model becomes subverted and contradicts and least some of its own starting-
points.
The paradox of the romantically inspired nationalist model, in fact, is
this. On the one hand, it claims recognition of diversity, equal rights, politi-
cal independence for one (linguistic) group with regard to other groups. On
the other, it has to assume an internal homogeneity within that group, for the
simple reason that within the romantic logic, it is the identity of the group
that legitimatizes the claim for recognition. And so, the identity may have to
be imposed or constructed, and dissident voices within the group may have
to be stifled.
The romantic model, then, is no less prone to contradictory developments
than the rationalist one. In the linguistic debate, the specific form of the ro-
mantically nationalist position is a concern for the purity of the language.
Defending the common language against foreign influences (loan words,
basically) is at the same time a defense of the cultural identity of the people.
In the nationalist subversion of the initial romantic model identities are not
only expressed, but they are also made permanent. Again, the link between
purism and nationalism is well-known, and there is an extended literature on
purism. What I would like to stress, in this respect, is less the phenomenon
as such, but rather how it fits into the overall pattern that defines the para-
doxicallogic of the rationalist and the romantic model of language variation.
This paradoxical logic, to sum up, resides in the following points. First,
although the basic models are opposites, they find a common ground in the
notion of nationalism. Because the rationalist model cannot easily realize its
extreme universalist claims, and because the romantic model cannot easily
realize its radical individualist claims, both models meet on a middle ground
where groups of people claim political identity and independence. Second,
this coalescence of the models does not annihilate the tensions that exist
between them: the history of the past two centuries brims with examples of
conflicts between a more rationalist Staatsnationalismus ( civic nationalism
at the level of the nation-state) and a more romantic Volksnationalismus
(ethnic or cultural identity nationalism). Third, in addition to the tensions
between the models, we have to take into account tensions within each
model: the rationalist model is subject to the danger of a discriminatory
'Dialektik der AufkHirung', and the romantic assumption of internal homo-
geneity may likewise have oppressive side-effects.
3.2. The twentieth century and the postmodern awareness
Living as we do in the aftermath of the nationalist era, we should complete
our overview of the historical transformations of the cultural models of lan-
guage variation by charting what changes are brought to the debate by our
post-nationalist environment. The current situation can best be characterized
by two overlapping developments: globalization and postmodernism.
Globalization is economic and political (to the extent that the growing
importance of international organizations diminishes the older importance of
the nation state). But it is also linguistic: the international spread of English
almost realizes the old rationalist's dream of a universal language.
The postmodern awareness, on the other hand, resides in two features.
First, the so-called 'disappearance of the Great Narratives' signals a weak-
ening of the older patterns of interpretation. There is a great deal of suspi-
cion with regard to the rationalist model of a smooth, emancipatory progress
as well as (and perhaps more dominantly so) with regard to the nationalist
model. Postmodern thinking is the self-consciousness of the late twentieth
century: progress is not automatic, and nationalism is dangerous.
This critical attitude entails a second feature: if the old models are no
longer self-evident, a dehierarchicahzation and informalization occurs. If, for
instance, the original hierarchical ordering of high culture and low culture is
rejected as part of the old models of interpretation, then popular culture may
claim equal rights with regard to high culture.
These two features imply that postmodernism is to a large extent a re-
newed form of the original romantic attitude: it renews the critical, 'counter-
cultural' attitude with regard to the official stories, and it revives the claims
for diversity.
The interesting question from our point of view is whether these changes
lead to a fundamental transformation of the cultural models that are used to
discuss language variation. Basically, there is a growing emphasis on the
international relationship between languages rather than the national rela-
tionship between language varieties. The initial models of standardization
are essentially models of standard languages in comparison with dialects or
other varieties of the same language. In the nationalist era, the debate some-
times involves national languages as opposed to minority languages, but it is
only in our days that the debate concentrates on the international relationship
between different languages, viz. the relationship between English as a world
language in comparison with local, possibly endangered languages.
Now, to the extent that the position of global English is at stake, the old
opposition between rationalist and romantic attitudes receives a new impe-
tus. I have shown above how Skutnabb-Kangas's argumentation about the
treatment of minority languages is largely situated within what I would call a
'romantic' frame. At the same time, it is not difficult to see what form the
basic pattern of a rationalist reply with regard to the position taken by Skut-
nabb-Kangas would probably take.
First, against the identification of language and culture, the rationalist
could point to cases where the same language is unproblematically shared by
different cultures, or conversely, where the same culture unites people with
different languages.
Second, against the allegation that the international dissemination of Eng-
lish is discriminatory, the rationalist might want to stress the actual emanci-
patory effects of a knowledge of English. If English is indeed the key to in-
ternational communication (and if, indeed, acquiring English is possible for
all), then it can only be welcomed that more and more people are able to
participate in that kind of communication.
At the same time, though, the contemporary discussions seem to lead to
the development of a model based on a functional differentiation between the
varieties involved - an 'and/and' -model rather than an 'either/or' -modeL so
to speak. In the discussion about the international situation in particular,
there is a growing recognition that multilingualism is a natural situation.
Interestingly, the shift towards multilingualism as a (so to speak) dialectic
synthesis of the opposite forces may be derived from the rationalist as well
as from the romantic model. In the previous section, we saw that the shift
towards nationalism fitted into the logic of both basic models, if account was
taken of the 'problem of levels'. At this point, we can see in a similar way
that a new focus on multilingualism fits in with both models.
On the one hand, a multilingual solution seems to presuppose some form
of functional distribution: one language is used for a specific set of circum-
stances, and the other for another set of circumstances. Such a diglossic or
polyglossic situational specialization is not incompatible with the original
rationalist model. After all, the rationalist model is motivated by a desire to
assure maximal democratic participation in what are sometimes called 'sec-
ondary domains' of social life: specific, public domains of experience, to
begin with higher education and political life. For the 'primary domains',
beginning with the more private aspects of life, the existence of less uniform,
more local language varieties does not fundamentally contradict the ideologi-
cal basis of the model.
Cultural models o.fstandardization 295
On the other hand, the postmodern twist of the romantic model entails a
new attitude towards the question of personal identity. It is often said, in
fact, that one of the hallmarks of the postmodern mentality is the fragmenta-
tion of identity. People no longer experience a single personal identity, but
they exhibit a number of different, possibly shifting identities, of a profes-
sional, social, ethnic, cultural nature. Different languages may then, follow-
ing the original 'expressive' logic of the romantic attitude, express this
fragmentation (or perhaps rather multiplication) of identities.
Although multilingualism would thus appear to provide a possible syn-
thesis of the initial models, the multilingual solution does not, however, com-
pletely remove the tensions. Just like the nationalist convergence of the mod-
els in the nineteenth century engendered a tension between nation states and
minorities, the multilingual convergence entails tensions about the exact
functional and situational distribution of the language varieties. A simple
case in point is the current reform of higher education in Europe. The impo-
sition of a uniform Bachelor/Master system is intended, among other things,
to stimulate student mobility, and this in turn increases the pressure to intro-
duce English as a language of instruction at least at the Master level. But
many, of course, are reluctant to accept such a functional restriction on the
original national language.
Likewise, internal tensions within each model continue to crop up. For
instance, those welcoming international English as an opportunity for all to
participate equally in a global culture may easily be blind to the fact that
access to English is not equal for all. And the champions of linguistic diver-
sity may readily overlook the fact that the people they purport to defend
often prefer the educational and professional opportunities provided by the
non-native language (as appears to be the case, for instance, in most African
countries that are ex-colonies of Britain).
All in all, then, we may sum up the present situation in a way that largely
parallels the summary at the end of section 3.2. On the one hand, just like
nationalism allowed for a coalescence between the rationalist and the roman-
tic model, multilingualism may constitute a point of convergence for the
post-nationalist manifestations of the models. On the other, just like nation-
alism did not abolish the basic tension between the models nor the internal
tensions within the models, a multilingual model does not cancel out the
tension between rationalistically seeing global English as a communicative
and educational opportunity and romantically seeing it as a threat to diver-
sity and local identity. At the same time, though, we should keep in mind
that the multilingual model is only beginning to emerge, and that the posi-
tions in the current debate have not yet crystallized as much as they have in
the older nationalism debate.
3.3. Summarizing the models and their transformations
We can now identify the pattern that emerges from the discussion in the
previous pages. Referring to the philosophical and cultural climate of the
eighteenth century, we have distinguished between a rationalist and a roman-
tic basic model of linguistic standardization. Starting from a communicative
conception of language, the fonner stresses the emancipatory function of a
common language as an instrument of political and educational participa-
tion. Starting, on the other hand, from an expressive conception of language,
the romantic model stresses how the imposition of a standard language may
discriminate specific cultural identities. There is, then, a tension between the
models to the extent that they are each other's counterpart. That tension is
enhanced by the 'Dialektik der AufkHirung', the mechanism, through which
the implementation of the rationalist ideals may generate its own opposite.
Table 2. Cultural models of standardization and their historical transformations
18TH CENTURY: 19TH CENTURY: LATE 20TH CENTURY:
the basic models nationalism postmodernism
the rationalist the common lan- the nation as the global English as an
position: guage as an instru- basis of a liberal opportunity / multi-
ment of political and democracy lingualism as func-
educational partici- tional specialization
pation
the romantic the standard lan- the nation as a global English as a
position: guage as a discrimi- focus of cultural or threat / multilingual-
nation of specific ethnic identity ism as the expres-
identities sion of fragmented,
postmodern identi-
ties
the tension opposition between conflict between what is the exact
between both the models. en- nation states and shape of the func-
positions: hanced by the ethnic/cultural tional specializa-
'Dialektik der Aufk- groups tion?
Hirung'
In the successive transformations that the models undergo in the nine-
teenth and the twentieth centuries, we have not only identified variants of the
two models as such, but we have also indicated how the tension that exists
between them reappears in different forms. Table 2 charts the various posi-
tions.
The nineteenth century witnesses a partial convergence of the models
round the concept of nationalism. Both basic models do, in fact, have a
problem of levels: how general should the communicative community of the
rationalist model be? And how specific can the identity get that is expressed
according to the romantic model? Rather than going for (respectively) uni-
versalist or individualist extremes, the nationalist interpretation of the mod-
els focuses on an intermediate level. From the rationalist point of view, the
nation is, by definition, the level at which democratic participation should be
guaranteed. From the romantic point of view, the identity to be expressed is
a common identity, the identity of the group that constitutes a nation. These
two forms of nationalism are well known, of course: liberal nationalism
(civic nationalism, Staatsnationalismus) sees the nation as the basis of a
liberal democracy, and romantic nationalism (identity nationalism, Volksna-
tionalismus) sees the nation as a focus of cultural or ethnic identity. The
tensions between both are equally well-known: the liberal nation state need
not coincide with a single cultural or ethnic group, and these groups may
then claim a status as an independent nation. Needless to say, linguistic dif-
ferences may play a crucial role in the definition of the groups involved.
Note, in addition, that a federal political model constitutes a rather frequent
compromise for the tension between claims of nation states and those of
nationalist groups.
The late twentieth century is characterized by a process of political and
economic globalization that has its attitudinal counterpart in a postmodem
view of the world, and that has its linguistic counterpart in the global spread
of English. The debate accordingly shifts towards the position of English vis
a vis local, possibly endangered languages. Although the process has not
perhaps reached its culmination yet, the debate seems to find a new focus in
the concept of multilingualism as a way of reconciling the different posi-
tions. In the same way in which the nationalist focus of the nineteenth cen-
tury followed logically from the initial models (through the problem of lev-
els), the focus on multilingualism can be equally motivated on the basis of
both models. For the rationalist model, multilingualism involves an accept-
able functional specialization of different languages: if language is an in-
strument of communication, different communicative situations may require
different languages. For the romantic model, multilingualism correlates with
the fragmented identity of the postmodern individual: if people may so to
speak have different identities, they may use different languages to express
those identities. However, the shift towards multilingualism does not elimi-
nate the tension: the exact functional specialization of the languages involved
remains a cause for conflict.
4. Applying the models
The rationalist, romantic, nationalist, postmodern models that we have iden-
tified may basically be used in two different ways in linguistics. On the one
hand, they may be used as the basis for an analysis of any discourse con-
cerning problems of standardization. The previous pages have already given
some examples of how the cultural models can be used in the analysis of the
logic of standardization debates. In general, such an analysis identifies the
components of a given discourse as expressing, amending, opposing aspects
of the basic models. We could go one step further by identifying standard
patterns of statements and replies that may be expected in this kind of dis-
cussion, but it is not my intention to do this systematically here.
On the other hand, language attitudes based on one or the other of the
models may influence the actual development of standardization processes.
In the following sections, I will illustrate this second domain of application. I
will do so by having a closer look at the work that I have been doing with my
research team on the internal variation of contemporary Dutch. In section
4.1, I will summarize the main results. (This section could be skipped by
readers less interested in the technicalities of the project.) In section 4.2, I
will discuss the way in which the cultural models of standardization may be
brought to bear on the findings of the project.
4.1. Contemporary Dutch as a pluricentric language
Dutch basically comes in two varieties: Dutch as used in The Netherlands,
and Dutch as used in the Flanders region of Belgium (sometimes referred to
as 'Flemish'). The situation of the standard language in both countries is
somewhat different. In Flanders, the standardization process that started off
(as in most European countries) in the Early Modem Period was slowed
down as a result of Flanders' political separation from The Netherlands
during the Eighty Years' War. Standard Dutch developed in The Nether-
Cultural models ojstandardization 299
lands in the course of the seventeenth century, but as Flanders was politi-
cally separated from The Netherlands, remaining under foreign (Spanish or
Austrian) rule, it did not link up with this process of standardization. Rather,
French was used more and more as the language of government and high
culture, a practice that received an important impulse after the birth of the
Belgian state in 1830. Dutch then survived basically in the form ofa variety
of Flemish dialects.
However, as a result of a social and political struggle for the emancipa-
tion of Flanders and the Flemish-speaking part of the Belgian population,
Dutch again gained ground as a standard language (the language of learning,
government, and high culture) in Flanders. This process started somewhat
hesitantly in the late nineteenth century as a typically romantic movement,
gained momentum during the first half of the twentieth century, and finally
made a major leap after World War 11 and during the booming 1960s. Still,
most linguists agree that the standardization process has not yet reached its
final point, or at least, that the level of standardization has not reached the
same height as in The Netherlands.
The latter observation is the starting-point for our research: can we quan-
tify the relationship between Belgian Dutch and Netherlandic Dutch (and the
internal stratification of both varieties)? Can we calculate how close or how
distant both varieties of Dutch are with regard to each other? More specifi-
cally (given that our research team is primarily interested in lexical varia-
tion), how does one quantify lexical convergence or divergence between two
language varieties? In Geeraerts, Grondelaers and Speelman (1999), a
measure of lexical overlap was developed, based on the notions onomasi-
ological prOfile and unUormity.
Table 3. Onomasiological profiles for shirt
hemd
overhemd
shirt
Belgian Dutch 1990
31 %
69 %
0%
Netherlandic Dutch 1990
17 %
46%
37%
The onomasiological profile of a concept in a particular source is the set
of synonymous names for that concept in that particular source, differenti-
ated by relative frequency. Table 3 contains the onomasiological profiles for
OVERHEMD 'shirt' in the Belgian and the Netherlandic 1990-database.
Uniformity is a measure for the correspondence between two onomasi-
ological profiles. Our computation of uniformity has its starting-point in the
idea that a common language norm triggers unifornl linguistic behavior. In
its most extreme form, lexical uniformity in the naming of a concept obtains
when two language varieties have an identical name for that concept, or
several names with identical frequencies in the two varieties. Much more
frequent than these examples of 'ideal' uniformity, however, are such partial
correspondences as illustrated in Table 3. Let us, for the sake of illustration,
assume that the relative frequencies in Table 3 represent 100 actual naming
instances in each of both profiles, rather than percentages. The partial over-
lap between the profiles in Table 3 is quantified by counting the naming
instances for which there is a counterpart in the other profile. In the ideal
scenario outlined above, each of the 100 naming events in each of both pro-
files has its counterpart in the other profile, yielding a maximal uniformity of
100 %. In Table 3, however, 14 instances of hemd in B90 have no counter-
part in N90, 23 Belgian overhemd's have no Netherlandic counterpart, and
there are no Belgian counterparts for the 37 Netherlandic shirts. On the
grand total of 200 naming events in the two profiles, only 200-(14+23+37)
=126 instances have counterparts in the other profile, which yields a uni-
formity of 126/2=63%. For the sake of quantitative convenience, it should
be noticed that this percentage equates the sum of the smallest relative fre-
quency for each alternative term, i.e. 17+46+0=63%.
If more than one concept is investigated, a uniformity index U is defined
as the average of the uniformity indexes of the separate concepts, whereas
uniformity index U' is defined as a weighted average, in which the relative
frequency of each concept in the investigated samples is taken into account.
In the present context, we will focus exclusively on the weighted uniformity
U', in which high frequency concepts have a more outspoken impact on the
overall uniformity.
The empirical foundation of the research project consisted of 40.000 ob-
servations of language use. We collected the different names (and their fre-
quencies) used to denote thirty concepts, fifteen from the field of clothing
terminology, and fifteen from the field of football (i.e. soccer) terminology.
The resulting database allows us, for instance, to calculate the proportion in
Belgian and Netherlandic sources of the term buitenspel 'offside' and the
loanword offside for the concept OFFSIDE; in the case of the concept JURK
'dress', we can determine whether the lexical choices involve a preference
for either jurk, japon, or kleed. The core of the observed material consisted
of magazine and newspaper material recorded in 1990.
This core was extended in two ways. In the first place, similar material
was collected for 1950 an 1970, which enabled us to carry out a 'real time'-
investigation of lexical convergence or divergence processes. In addition, the
stratification of language use was taken into account. Between standard and
dialect, there are a number of 'strata' on which register differences may co-
occur with an increasing geographical specialization. For an investigation of
the relationship between Belgian and Netherlandic Dutch, these strata - viz.
the regionally colored informal variants of the standard language - are ex-
tremely relevant: it can be expected that the linguistic differences between
Belgium and The Netherlands will increase on this regiolectic level. This
intermediate level between dialect and written standard language was repre-
sented by the clothing terms we collected from labels and price tags in shop
windows in two Belgian (Leuven and Kortrijk) and two Netherlandic towns
(Leiden and Maastricht). The intended audience of this form of communica-
tion is more restricted than the national or binational audience which is the
target of the magazines from which the core material was selected. The fact
that we are dealing with written language in a semi-formal situation, on the
other hand, ensures that we steer clear of the purely dialectal pole of the
stratificational continuum.
Table 4. U' values comparing Belgium and The Netherlands (1950-1970-1990)
and comparing written data from magazines and newspapers with local
shop window data (1990)
B50 / N50:
B70 / N70:
B90 / N90:
B90 / B
sw
90:
N90 / N
sw
90:
69,84
74,59
81,70
45,90
67,75
Given this database, what can we expect to find with regard to the rela-
tionship between the various language varieties? With respect to the status
and the development of Belgian Dutch, two uncontroversial hypotheses can
be found in the linguistic literature. First, there is an expectation of dia-
chronic convergence between Belgian and Netherlandic Dutch. The stan-
dardization process in Flanders is characterized by an explicit normative
orientation towards Netherlandic Dutch: the standardization of Belgian
Dutch took the form of an adoption of the Dutch standard language that
existed already in The Netherlands. In addition, the unfinished character of
the standardization of Belgian Dutch is believed to manifest itself in a larger
synchronic distance between local and national language in Belgium than in
The Netherlands; even to the untrained observer, it is obvious that the differ-
ences between regional and supraregional registers are much larger in Bel-
gium than in The Netherlands.
The diachronic and the synchronic hypothesis may now be made opera-
tional in terms of uniformity values as defined above. Diachronically, con-
vergence and divergence can be quantified as increasing or decreasing uni-
formity. Synchronically, the larger distance between national and local
language we expect in Belgian Dutch, will manifest itself in a smaller uni-
formity between magazine and shop window material in Belgian Dutch than
in Netherlandic Dutch.
Table 4 contains the relevant results. B50 stands for 'Belgian data from
1950', N50 stands for 'data from The Netherlands from 1950'. B
sw
90 refers
to the shop window materials in Belgium, in contrast with B90, which stands
for the data taken from magazines and newspapers. The data in Table 4
unambiguously confirn1 the diachronic as well as the synchronic hypothesis.
Diachronically, the increase in uniformity between Belgian and Netherlandic
Dutch suggests an evident lexical convergence between both varieties:
U'(B50,N50) <
69.84 <
U'(B70,N70)
74.59
<
<
U'(B90,N90)
81.70
Synchronically, the delayed or unfinished standardization of Belgian Dutch
manifests itself in a distinctly lower unifonnity between the Belgian maga-
zine and shop window data than between the Netherlandic magazine and
shop window material:
U'(B90, B
sw
90)
45.90
<
<
U'(N90, N
sw
90)
67.75
4.2. Cultural models and the evolution of Belgian Dutch
The overall situation of the different varieties of Dutch may be summarized
as in Figure 1. The left-hand side of the figure represents Netherlandic
Dutch, whereas the right-hand side represents Belgian Dutch ('Flemish').
The vertical dimension represents the internal stratification of both national
varieties, distinguishing between the register of standard speech, the collo-
quial register, and the local dialects. Distances in the figure symbolize lin-
guistic distances. In the upper register, both national varieties are close to
each other, but the distances increase in the lower registers. Crucially, the
colloquial Belgian Dutch (colloquial Flemish) is much further apart from the
high register in Belgium than the distance that exists between colloquial
speech and polite speech in The Netherlands. So how would Belgian Dutch
evolve?
Basically, there are three possibilities. First, the present situation is
merely a transitory one in a gradual process of convergence and standardiza-
tion. We have seen that there is some evidence for such a converging devel-
opment, at least in the upper stratum of the language, and it might be as-
sumed that this evolution will continue in the lower strata. In that case,
colloquial Flemish is likely to move upwards in the figure, perhaps until it
reaches a position that is similar to that of Netherlandic Dutch. Second, the
process of convergence might be reversed. Typical features of colloquial
Flemish might be incorporated into the standard register used in Belgium,
and this informalization of the standard language may in turn lead to a grow-
ing distance in the highest stratum between Netherlandic Dutch and Belgian
Dutch. And third, the situation might simply be stable.
Netherlandic
5tandard
D.1tch
/
Netherlandic
colloCPJial D.1tch
dialects
Belgian
standard
D.1tch
colloquial
Flemish
\
dialects
Figure 1. The stratificational structure of contemporary Dutch
Now, what I would like to suggest is that the actual choice from among
these three logical possibilities may be determined at least in part by the
extent to which the three models of standardization are prevalent within the
language community.
The first possibility (in which colloquial Flemish moves upwards in the
direction of the upper stratum) would receive the strongest impetus from a
rationalist attitude towards processes of standardization. A growing ten-
dency to participate in those activities that are typically associated with
standard language use would go hand in hand with a further dissemination of
the standard language in its present form. Whether this process would imply
a further convergence of the upper register with Netherlandic Dutch is less
clear: apart from the overall universalist aspect of the rationalist stance, i.e.
the tendency to broaden the conmmnicative scope of the language, the func-
tional motivation for such a convergence would seem to be relatively weak.
We should note that a process of continuing convergence could also be
triggered by a nationalist attitude, if the Flemish and the Dutch feel as one
supra-national people, united by a common language across the borderline
between Belgium and The Netherlands. However, although there exists a
minor tradition of people advocating such an 'ethnic' union of Holland and
Flanders, it plays no significant role in popular opinion. The feelings that
exist between the Dutch and the Flemish are feelings of competition and
gentle animosity rather than ethnic, language-based solidarity.
The second possibility (an increasing internal uniformity within Belgian
Dutch, achieved through an informalization of the highest register) links up
with the nationalist model. Flanders has achieved a political emancipation
within Belgium, and this might be reflected in a growing linguistic independ-
ence - in the development, in other words, of an independent standard lan-
guage nornl that is increasingly different from the imported Netherlandic
Dutch that was the initial reference point for the standardization process in
Flanders. If the nationalist model prevails (and if it is oriented towards Flan-
ders alone and not towards an 'ethnic' union of Holland and Flanders), the
speakers of Flemish might develop a growing preference for the indigenous
language forms that they encounter at the colloquial level rather than for the
imported forms in the higher registers.
The third possibility (a status quo) has a romantic and postmodern ring to
it, given that the romantic attitude in its purest form involves anti-
establishment feelings. The romantic appreciation of standardization (if it is
not lifted to the nationalist level) is basically a suspicious one: standardiza-
tion is seen as a threat to diversity in general and to one's own local or indi-
vidual identity in particular. Distancing himself from the official standard,
the romantic has several options: he can plead for diversity (the acceptance
of all varieties as equal) or he can try to impose his own language as the
standard (the nationalist option), but he could also simply maintain the dis-
Cultural models o.fstandardization 305
tance, accepting that there is a language for official occasions with which he
does not identify, and a colloquial language in which he feels at home and
that is sufficiently distant from the official language to be perceived as an
expression of his own 'tme' identity.
Are there any indications at all that Flanders might indeed have a mental-
ity that favors such a stable form of diglossia? Its own popular self-image
would certainly seem to support such a claim. Flanders perceives itself as
being mildly anarchistic, in the sense that it is highly suspicious of official
authorities, and in the sense that it does not take laws and regulations all too
seriously. Tax evasion, for instance, is frequently described as a national
sport. Also, historical reasons are often cited to explain this mentality. As
mentioned above, Flanders has a history of foreign mle, and this is supposed
to have engendered a mentality of critical distancing with regard to all forms
of official authority. Postmodem tendencies towards informalization and
dehierarchicalization would of course strengthen such an attitude. So, could
it be that this mentality supports a romantically diglossic attitude in linguis-
tic matters? The suggestion is certainly fascinating enough to warrant fur-
ther empirically sound research about the mentality of the Flemings and their
cultural history.
To be sure, I am not able to make any predictions about the probable
course of events in Flanders. The evolution will certainly not only be deter-
mined by the attitudinal models mentioned here, and I do not have any atti-
tudinal data yet that determine the strength of one or the other attitude. What
I have tried to show, though, is that it makes sense to think about actual
processes of standardization in terms of the cultural models identified in the
first part of this paper. Cultural models of standardization are real not only
to the extent that they shape public debate, but also to the extent that they lie
at the basis of the language attitudes of the language users.
5. Conclusions
I have tried to do a number of things in this paper. First, I have argued that
our thinking about questions of language variation and linguistic standardi-
zation can most often be reduced to two underlying, opposing cultural mod-
els: a rationalist one and a romantic one.
Second, I have suggested that the models originated in the eighteenth cen-
tury, and then went through a number of transformations in the course of the
following two hundred years. Specifically, the nineteenth century witnessed
the development of a very influential nationalist model that combines ele-
ments of both basic models, and the late twentieth century witnessed a shift
towards questions of globalization and the international position of English.
At each of these stages, the models exhibit a certain tendency towards a
coalescence, but at the same time, tensions remain - tensions between the
models, and tensions within the models.
Finally, I have tried to demonstrate that the models in question cannot
only be used in discourse-analytic fashion to scrutinize argumentations and
discussions, but that they may also be seen as shaping the language attitudes
of linguistic communities. If they have a real presence, they do not only
show up in the way in which people talk about standardization, but also in
the way in which they act.
Each of these points may be further developed. The identification of the
models may be expanded towards a systematic map of standardization dis-
cussions, charting recurrent patterns of statements and replies. The historical
sketch might be developed into a synthetic historical overview of standardi-
zation processes, standardization debates, and their relationship with linguis-
tic theorizing. And the suggested link between models and language attitudes
could lead to empirical attitudinal research. All of these possible develop-
ments, though, should contribute to a common goal: a better understanding
of the underlying logic of standardization debates.
Chapter 12
Caught in a web of irony: Job and his embarassed
God
Originally published in Ellen Van Wolde (ed.), 2003, Job 28. Cognition in Con-
text 37-55. Leiden: Brill.
One of the best indications for the success of Cognitive Linguistics is the way in
which its models of analysis spread to disciplines that lie outside linguistics in the
traditional sense, like literary studies (as represented by Steen 1994, Freeman
1995, 2000, Turner 1996, Brandt 1998, Hamilton 2002, Danaher 2003, Popova
2003, Hiraga 2003) or the study of religious language (see among others Bie-
buyck, Dirven, and Ries 1998, Boeve and Feyaerts 1999, Feyaerts 2003, Taba-
kowska 2000, Jakel 2002). The present paper belongs to the latter group; it de-
rives from a conference held in Amsterdam in 2002, in which biblical scholars
and cognitive linguists convened to discuss the Book of Job. Admittedly, the prin-
ciples of Cognitive Linguistics put at work in the present study are only high-level
ones, not very specific modes and models of analysis. Still, the paper illustrates
nicely how a contextualized reading (both in terms of the internal context of the
text and the cultural context in which it was written) may enhance understanding.
1. Dealing with discrepancies
In an ironical but at the same time esthetically pleasing way, the position of
the reader with regard to the Book of Job is analogous to the position of Job
vis avis God's behavior: for neither of them, it is an easy matter to arrive at
a satisfactory interpretation. But if both God and the text are difficult to
make sense of, it should come as no surprise that the Book of Job has been
subject to postmodern analyses. Postmodern criticism is very much con-
cerned with two related characteristics of texts that feature prominently in
the Book of Job: on the one hand, the absence of a definitive interpretation
(what is, ultimately, the significance of human suffering?), and on the other,
the presence of incongruities and discrepancies (why, for instance, does God
answer Job by basically repeating a number of the things that Job himself
has already said?) In Job scholarship, these two characteristics seem to lie at
the basis of two different strategies of interpretation, one somewhat less
308 lvleaning and culture
extreme than the other. Whereas a more moderate approach shows how the
tensions in the text contribute to meaning rather than subverting it, the more
radical form of postmodern interpretation takes its starting-point in the ab-
sence of a definitive interpretation, and dialectically turns this absence into
the very message of the text: the impossibility of arriving at an ultimate
meaning is the meaning of the text.
A postmodern approach, however, is merely one way of making sense of
the alleged inconsistency and inconclusiveness of the text. More traditional is
the earnest philological habit of attributing difficulties to textual corruptions
and interpolations, and there have of course been many suggestions as to the
philologically heterogeneous character of the Book of Job. I am not in a
position to judge the philological value of such arguments, but I will assume
that it is good hermeneutic practice to apply an interpretative Occam's razor:
if we can find a coherent meaning, that option is to be preferred over postu-
lating different textual sources. The same methodological guideline applies,
to be sure, to radical postmodernism: renouncing a coherent textual interpre-
tation in favor of a meta-textual reading is something to be done only as a
last recourse, when all other possibilities have been rejected. So have they?
What I propose to do in this paper, then, is to steer clear of both the
Scylla of postmodem extremism and the Charybdis of philological facility.
Incoherence, a lack of harmony, clashes between different aspects of the text
may in fact be interpreted in yet a third way: as symptoms of humor. As a
literary genre, humor is the form par excellence to harbor incongruities. So
would a humorous reading be an adequate way of dealing with the interpre-
tative difficulties of the Book of Job? To what extent can it be read as an
ironic text?
The question is not new, and what I will be able to add to the debate is
merely a specific reading of God's speech in chapters 38-41. I will develop
my argument in three steps. First, I will review some of the evidence for the
presence of irony in the Book of Job. Second, I will chart the ironists' posi-
tions with regard to the theophany of chapters 38-41: for those writers who
have taken the ironical aspects of the Book of Job seriously, the nature of
God's speech is as much a matter of debate as it for those writers who have
simply taken the Book of Job seriously. And third, I will have a closer look
at chapters 38-41, arguing that they can indeed be interpreted ironically. The
outcome is a reappraisal of the dramatic figure of God: I will argue that the
God-character is best interpreted as an embarrassed God.
Caught in a web ofirony 309
2. Methodological preliminaries
The methodology that I will use in the following pages is not specifically
linguistic in any technical sense. (It is not even specifically lexical, in spite
of my personal specialization in lexical semantics and lexicology.) There is,
however, a consistent inspiration from Cognitive Linguistics, which it may
be useful to make explicit at this point. (For an introduction into Cognitive
Linguistics, see Taylor 1995, Ungerer and Schmid 1993. Foundational
works are Langacker 1987, 1991 and Lakoff and Johnson 1981, Lakoff
1987.)
As a first step, let us note that my analysis involves three specific the-
matic discussions: an investigation into the concept of wisdom as it appears
in the Book of Job, a pragmatically oriented analysis of Job's speech, and an
attempt to answer the question whether God's reply to Job can be considered
ironical. These topics are situated at different levels of linguistic structure:
the analysis of the notion of wisdom looks at a single category, whereas the
other two involve utterances as a whole.
As I will try to make clear presently, the three topical questions illustrate
a number of views on meaning that are cherished by Cognitive Linguistics.
This does not mean, to be sure, that a similar analysis could not be arrived at
by other means, i.e. starting from a different framework. I have since long
claimed, for instance, that Cognitive Semantics links up in fundamental re-
spects with a traditional philological approach, and (notwithstanding the
recognition that Cognitive Linguistics has developed a battery of analytical
notions and techniques that go far beyond anything to be found in the philol-
ogical tradition), I could very well conceive of the following analysis as
coming out of a more traditionally oriented philological approach.
The relevant views on meaning, then, are the following. Each of them re-
lates specifically to one of the three topical discussions.
I Meaning involves categorization. Linguistic meaning is not a straightfor-
ward reflection of the world, but it is a way of shaping reality, of making
sense of the world. In the case of a concept like 'wisdom', what is to be un-
derstood by that term is not a priori given. It is, as we shall see, a category
that derives its experiential content from its relationship with other concepts.
In the Book of Job, wisdom is not just 'the quality of being wise' or some
such lexicographical definition, but it is closely connected with age and ex-
perience: I will argue that it is crucial for an understanding of the Book of
Job that wisdom comes with age. If we merely think of lexical concepts in
the way in which they tend to be economically defined in dictionaries, such
aspects of a concept might easily be considered of secondary importance. If
one thinks of meanings as categories of experience, however, the entire can-
vas of associations that the concept entertains becomes part of the meaning.
2 Meanings may invoke culturally specific background knowledge and
assumptions. As categories making sense of the world, specific categories
are situated against a wider context of lived experience, which may be either
physiological/biological or cultural. In the pragmatic analysis of Job's
speech, I will suggest that Job seems to be flaunting a culturally specific
pragmatic politeness hierarchy that has its roots in a patriarchal context.
3 Meanings are contextually flexible. The meanings that are realized in
actual speech contexts need not be completely identical with the meanings
that are more or less permanently stored in our linguistic memory. The inter-
play between specific circumstances and the linguistic message may create
new or different readings. On the level of individual categories, this flexibil-
ity is well-known in the form of prototype effects (Geeraerts 1997a). In the
Book of Job, the importance of context becomes specifically clear if, at the
level of utterances rather than individual categories, we look at the way in
which irony comes about. Irony is typically a case in which an expression
receives a different interpretation in one context rather than another - even
extremely so: one reading is the opposite of the other, and moreover, the
contrast between the two readings is part of the effect that the speaker wants
to achieve. When, for instance, Job replies to Sofar and the other friends:
'wisdom will die with you '(12:2), the utterance receives an ironic interpreta-
tion because we know that Job is far from happy with the comfort that his
friends purport to give. In the context, Job's reply acquires a sarcastic read-
ing. The impact of context in the constitution of irony is complicated by the
fact that the knowledge that triggers the ironic reading is not necessarily
shared by all participants (and the reader, to be sure, is one of those partici-
pants).
3. Humor in the Book of Job
Before focusing on the possible irony of God's speeches in chapters 38-41,
let us first establish that a reading of the Book of Job as a humorous text is
not an altogether implausible option. Humor is far from absent in the Old
Testament (Good 1965, Knox 1969, Friedman 2000), and a number ofau-
thors (Whedbee 1977, Robertson 1973, 1977, Hoffmann 1983) have specifi-
cally analyzed the Book of Job with regard to its humorous content. The
most radical position in this respect is that of Robertson and Whedbee, who
argue that the text as a whole belongs to the genre of comedy. I will discuss
the views of Whedbee, Robertson, and Hoffinann more thoroughly in the
next section. At this point, I will merely enumerate some of the humorous
aspects of the text. Many of the points enumerated here may be found in the
works of the authors just mentioned, but I have tried to bring some order into
the matter by grouping the observations in four categories: the conversa-
tional style of the book, the characters, the plot as a whole, and specific de-
tails of the plot.
First, throughout the conversations and speeches that constitute the main
body of the text, there are sarcastic remarks and ironical interchanges among
the main characters. For one thing, Job regularly scorns his friends for their
failure to provide an adequate answer to his misery. (Quotes are from the
Revised Standard Version of 1947.)
12:2 No doubt you are the people, and wisdom will die with you.
13:5 Oh that you would keep silent, and it would be your wisdom!
26:2 How you have helped him who has no power! How you have saved the
ann that has no strength!
3 How you have counseled him who has no wisdom, and plentifully de-
clared sound knowledge!
4 With whose help have you uttered words, and whose spirit has come
forth from you?
For an other, God mocks Job's claim to wisdom (the claim, that is, that he
could explain something to God that he does not know yet):
38:4 Where were you when I laid the foundation of the earth? Tell me, if
you have understanding.
5 Who determined its measurements-surely you know! Or who stretched
the line upon it?
and Job's challenging attitude:
40: 10 Deck yourself with majesty and dignity: clothe yourself with glory
and splendor.
11 Pour forth the overflowings of your anger, and look on every one that is
proud, and abase him.
12 Look on every one that is proud, and bring him low; and tread down the
wicked where they stand.
13 Hide them all in the dust together; bind their faces in the world below.
14 Then will I also acknowledge to you, that your own right hand can give
you victory.
Second, the characters have caricatural features. To begin with, Job's
friends are a caricature of the wise counselor: they are supposed to comfort
Job and provide him with good counsel, but blatantly fail in their task. Next,
Elihu is even more ridiculous: there is a comical contrast between the tension
and the expectation that is created when Job so to speak calls God's bluff,
and the unexpected appearance of Elihu, who is portrayed as a bombastic,
long-winded and self-important young upstart. In contrast with the awesome
God whose appearance is expected, Elihu is a joker who relieves and at the
same time prolongs the tension. Finally, even Job himself, working himself
stubbornly from a state of utter desolation into a frenzy of rebellion in which
he challenges God to a showdown, displays an obsessive single-mindedness
that would suit any of Moliere's comical archetypes.
Third, the overall narrative structure of the text corresponds with that of
a comedy (not in the sense of an amusing story with funny characters and
farcical incidents, but in the sense of a literary work with a happy ending).
The overarching structure moves from a blissful, fairytale-like beginning to
a situation in which the main character lapses into wretchedness and misery.
When the main character has gone through the test (and has changed his
ways accordingly), a happy end follows.
And fourth, at a more detailed level of the plot, the progress of events is
characterized by comical repetitions and comical reversals. The dominant
example of repetition is obviously the conversation between Job and his
friends. Time and again, the attempts to comfort and convince Job fail, but
time and again, the pattern of futile argument and emotional reply is re-
peated. A specifically ironical touch is the way in which the pattern of repe-
titions gradually comes to a halt, with ever shorter interventions of the
friends, and finally the complete absence of the expected third speech by
Sofar. The ineffectiveness of the interventions, in other words, is reflected by
the gradual fading away of the friends' speeches: they literally have no ar-
guments left. An ironic reversal occurs in Job's attitude with regard to God:
after God has only once raised his voice, Job's rebellious challenge abruptly
changes into subservience. Similarly, there is an ironical reversal in Job's
relationship with his friends: while they were supposed to act as intercessors
for Job but failed to do so properly, Job, after his restoration, functions as an
intercessor for them.
Caught in a web a/irony 313
4. The irony of God's speech I
Both microstructurally and macrostructurally, then, humor and irony seem
to play an important role in the construction of the Book of Job. But all in-
terpretations of the book stumble over God's speech - the final chapters of
the poetical part of the text that are so hard to interpret that some critics
have even suggested that they do not belong to the original text. Writers like
Hoffmann, Whedbee, and Robertson, who have devoted specific attention to
the humorous and ironic aspects of the book, are likewise divided as to the
meaning to be attributed to the intervention of God. In this paragraph, I will
try to show that the positions they take correspond to a specific pattern. At
the same time, it will appear that there is a hole in the pattern, and the next
paragraph will suggest how the hole can be filled.
The position of Robertson and Hoffmann may be contrasted with that of
Whedbee: whereas the former argue that God's speech is ironic and evasive,
and therefore does not constitute a proper answer to Job's question, the latter
underlines the non-ironic, relevant aspects of God's answer. Let us take a
closer look at the various positions.
Robertson (1973) interprets God's speech against the larger context of
the book, and specifically in the light of chapter 9, where Job predicts what
God would do in a face-to-face encounter: God would overawe Job with his
intellectual superiority and his overwhelming powers. And when God does
indeed appear, he acts precisely as predicted by Job: Job simply does not
stand a chance, and he acknowledges as much. God then appears as an in-
considerate tyrant rather than a just and merciful ruler: 'So God's rhetoric,
because Job has armed us against it, convinces us that he is a charlatan God,
one who has the power and skill of a god but is a fake at the truly divine task
of governing with justice and love' (1973:464). Job's repentance, in turn,
can only be a non-authentic stance, faking rather than feeling submissive-
ness. The overall meaning of the Book of Job, then, is that of curing man's
fear of fate, his destiny, the unknown by 'ridicule of the object feared'
(1973: 468): 'So we know of him what we know of all tyrants, that while
they may torture us and finally kill us, they cannot destroy our personal
integrity' (1973: 469).
Whedbee (1977), as we have seen, accepts Robertson' s overall premise
that the Book of Job belongs to the genre of comedy, but challenges Robert-
son's reading of God's speech. Robertson, he claims, underplays the signifi-
cance of God's intervention by merely concentrating on the ironic interplay
between Job's prior speeches and God's speeches. God does in fact go fur-
314 lvfeaning and culture
ther than just echoing Job's predictions: he redirects them towards a positive
vision of the universe. Yes, God is all-powerful (as Job predicted), but
whereas Job paints a bleak picture of God's overwhelming strength, God
himself brings, as Whedbee states it, a 'playful, festive note in the portrayal
of creation', stressing the magnificence and the order of the universe rather
than the chaos and oppression that Job highlighted. Job's confession, then, is
a genuine recognition that he saw creation in the wrong light, and the resto-
ration of Job, rather than being ludicrous in all in its implausibility (as
Robertson would have it) is a true happy ending.
Robertson (1977), incidentally, has answered Whedbee, but in a very
half-hearted way, admitting that 'though neither theory is free from defects,
both are about as coherent and comprehensive interpretations of the Book of
Job as we are likely to get'.
Hoffmann's position (1983, 1996), to round off the overview, is an inter-
esting variant of Robertson's. Hoffmann too recognizes God's speech as
ironical, but it is not irony of God with regard to Job, but rather irony of the
author with regard to himself. It is worthwhile to quote Hoffmann (1983: 19)
in full:
The author has ironically presented himself as someone who knows the an-
swer to the questions he has raised. However, in the manner of irony, that
which is written (i.e.: here is the well-ordered answer to the questions
raised heretofore) implies the exact opposite. What is really being said is: I
cannot answer the difficult questions I have presented, and they are still as
problematic as before. What we have here is selfirony, in that throughout
the book the author creates the impression that he knows the answer and
will tell us at the end. ( ... ) Hence we arrive at the self-irony of man: All I.
as a human being, can do is to delude myself that there is a solution to the
problem of divine recompense and mask my delusions one way or another.
The ironic truth is that I must accept my fate, come what may.
The inaptness of God's words with regard to Job's question turns the text as
a whole into an ironic statement. All along, the text seems to work its labori-
ous way towards an answer, but when the answer finally comes, it is way off
the mark.
The pattern that emerges from the positions of these three authors is a
simple one: either God's speech is an appropriate albeit indirect response to
Job (Whedbee), or it is ironic and evasive (Robertson and Hoffmann). But
are these alternatives mutually exclusive? At least in one way they are not:
the view that God's speech is a corruption of the original text in the form of
a later insertion can be equated with the defeatist view that it is neither ap-
Caught in a web a/irony 315
propriate nor ironic, that it is, in other words, uninterpretable. The true hole
in the pattern, then, involves the possibility that God's speech is both appro-
priate and ironic. What would happen if we pursued this line of thought?
Can such an interpretation be maintained?
I will explore the possibilities in the next paragraph, but I first have to
indicate how the approach that I will propose differs from the positions dis-
cussed so far.
On the one hand, I agree with Whedbee that Robertson underestimates
the impact of God's speeches: they do not just mimic Job's earlier speeches,
but they add something to them. I disagree with Whedbee, however, as to
what it is that is being added. Even though there is a element of playfulness
in God's speech, the speech as a whole is not really a straightforward hymn
to creation. If it convinces Job, it has to be on other grounds than just the
festive nature of creation. A closer scrutiny of God's speech is in order, then.
On the other hand, I entirely agree with Hoffmann that we have to take
into account different levels of irony in the text. Whereas Robertson only
envisages the irony of God with regard to Job, we have to allow for forms of
irony that are perceived only by the author and the audience, and not by the
characters. However, I think we can be more precise about the conspiracy
between the author and the audience than Hoffmann is, by taking a closer
look at the narrative framework of the text. If the audience knows more than
the characters, it is through the narrative setting introduced in the prologue.
So how does God's speech relate to the prologue?
5. The appropriateness of God's speech I
There are two points of a pragmatic nature that need to be elucidated first.
To begin with, God's speech is definitely an indirect speech act, and indirect
speech acts can constitute extremely appropriate answers. If you ask me:
'Will you attend Ron Langacker's lecture?', and I reply: 'Have men walked
on the moon?', then my rhetorical question is not a direct answer, but it does
constitute an appropriate response of the' obviously' -type.
Further, appropriate responses to speech acts may take the form of ques-
tioning the felicity conditions of those speech acts. Felicity conditions are
criteria that have to be fulfilled if a speech act is to be properly made. They
may involve the sincerity of the speaker, the authority invested in the
speaker, the ability of the speaker to live up to the consequences of his com-
mitments, and so on. In all cases, satisfaction of the felicity conditions is a
necessary prerequisite for a certain speech act to be pragmatically valid.
This implies that certain acts may be appropriately countered by pointing
out that the felicity conditions of the act do not hold. Imagine a 15-year old
at the box office of a theater showing only X-rated movies, asking whether
the film is shown with or without a break. Ifhe gets the answer: 'Sorry, but
you are under age', then it is clearly not a direct answer to his question, but
it is an appropriate answer, because it makes clear that he was not in a posi-
tion to ask the initial question to begin with. The presuppositions that are
necessary for the question to be a relevant one, are not met, and pointing that
out is a conversationally adequate answer.
My suggestion now is that God's speeches may be read as indirect
speech acts questioning the felicity conditions of Job's question. I will ar-
gue for this position in two steps. First, I will analyze the concept of wisdom
in the Book of Job, because Job's speech act is fundamentally franled as an
investigation into knowledge and truth. If Job is asking for insight but is
asking for it in the wrong way, then we will first have to be clear about the
conception of knowledge and insight that permeates the Book of Job (and
that is insufficiently heeded by Job). Second, I will have to deal with the
bipartite structure of God's speech, because each of those parts addresses a
different aspect of Job's pragmatically inappropriate behavior.
The concept of wisdom (and insight, as both terms are often mentioned
together: see 11:6, 15:8-9, 26:3, 28: 12, 32:7-8, 34:34-35, 38:36, 39: 17) is
basically presented with three features (apart from the fact, of course, that
wisdom is difficult attain).
First, wisdom is a condition for appropriate speech. See e.g. 11 :47,20:3,
26:3-4, 32:7-8, 34:35, 36:4 and other passages, of which God's comment in
38:2-4 is perhaps the most poignant one. In itself, this association between
wisdom and the pragmatic validity of a given discourse is not surprising. In
the dramatic setting of Job's attempt to justify himself, speaking the truth,
expressing wisdom, formulating the right insight is probably the motor con-
cept par excellence of the whole text. We will see presently, however, that
this unsurprising link between possessing wisdom and speaking appropri-
ately is the basis for a specific pragmatic logic.
Second, wisdom correlates with age. Take the second speech by Elifaz,
lines 15:7-10. Elifaz questions Job's insight by pointing out that Job is not of
infinite age.
7 Are you the first man that was born? Or were you brought forth before
the hills?
8 Have you listened in the council of God? And do you limit wisdom to
yourself?
9 What do you know that we do not know? What do you understand that is
not clear to us?
10 Both the gray-haired and the aged are among us, older than your father.
Superior knowledge, in other words, presupposes superior age. Similar lines
are to be found in Sofar's speech in 20:4, in Elihu's speech in 32:7-9, per-
haps also in Job's words in 28: 13, and most certainly in God's remark in
38:4. Job cannot have infinite wisdom because he does not have infinite age.
If we accept that the experience needed for ultimate wisdom exceeds the
bounds of human life, then, of course, wisdom cannot be found in the land of
the living. The patriarchal association of wisdom with experience and age is,
in other words, extrapolated towards God. In the ordinary context of the
patriarchal society, the elders know best. God however is a superhuman
hyper-patriarch who has to know even better, if only because of his superior
age.
Third, it follows from the combination of the first and the second feature
that age is a prerequisite for appropriate speech, or at least, that there is
likely to exist a pragmatic hierarchy of speaking rights based on age and
assumed wisdom: the young do not speak before the old. As a form of a
politeness hierarchy, such a hierarchical restriction on the structure of con-
versation would be a conventional part of a patriarchal society, but in the
context of the Book of Job, it acquires a specific depth through the way in
which the text focuses on the concept of wisdom.
This is particularly apparent in the speech of Elihu. On the one hand,
Elihu presupposes the patriarchal model, expressing his disappointment in
the older people who have spoken before him, but who have not exhibited the
wisdom that might be expected of them: see 32:7-9. On the other hand, Elihu
breaks the hierarchy. According to the patriarchal politeness hierarchy, Elihu
speaks before his time. He is a young person challenging and the elders, and
moreover challenging them on their own domain: that of wisdom. That is
why Elihu can be depicted as a fool, making a lot of wind, inflating himself,
making himself more important, and continually re-affinning his own impor-
tance.
Elihu, in short, challenges the patriarchs without being in a position to do
so. But ironically, this is precisely what Job does with regard to God. The
position of Elihu with regard to Job and his patriarch friends is structurally
analogous to the position of Job with regard to the divine super-patriarch.
The epistemological hierarchy of wisdom and age, coupled with the prag-
matic hierarchy of politeness and conversational turn-taking, leads to a lin-
ear ordering where Elihu is lower on the ladder than Job and where Job is
lower on the ladder than God.
But then, Job calling God to account is not observing the linear ordering
just as much as Elihu is in reprimanding the elders. In challenging God, Job
is making a non-felicitous speech act, because he is not observing the hierar-
chy of politeness and conversational rights, and in claiming that he can teach
God something about his innocence, he is ignoring the hierarchy of wisdom
that is itself (as we have seen) intimately connected with the pragmatic hier-
archy.
God, to be sure, makes all this apparent, putting his mighty finger on
Job's breach of felicity conditions. God makes clear (in a roundabout way,
by pointing out rather than by stating explicitly) that Job is going beyond the
pale. Note that we have just identified two relevant conditions that Job is not
observing: one having to do with a hierarchy of politeness of conversational
rights, and the other having to do with a hierarchy of wisdom and knowl-
edge. That is why we have two speeches of God: in each of the speeches,
God addresses one of Job's pragmatic trespasses. The first speech addresses
the hierarchy of wisdom (i.e. Job's flaunting of the presupposition that supe-
rior wisdom correlates with superior age, and that ultimate wisdom relies on
infinite age). The second speech addresses the hierarchy of power (i.e. Job's
flaunting of principles of conversational politeness).
6. The appropriateness of God's speech 11
But the binary nature of God's speech is one of those aspects of the text that
have troubled the interpreters. It has been claimed that the double speech of
God is a flaw in the text, an interpolation perhaps, and in any case a super-
fluous repetition. Why are there two speeches of God? Why is there a second
speech if Job has already submitted after the first one? Is God just sadisti-
cally rubbing it in, delivering a final blow to an already subdued enemy? I
would argue that there are subtle differences between the two speeches that
indicate that they have a different focus. There are three basic observations
that I would like to advance in support of this idea.
First, God's opening lines, setting the topic of the speeches, unmistakenly
mark a difference. The opening of the first speech takes Job at issue for
wanting to teach God something, as if God were saying: 'So you want to
instruct me? Well okay, be a man, and let's see if you can teach me any-
thing'. The second speech puts it no less explicitly in temlS of power rela-
tions: 'You want to challenge me? Well okay, let's see how far you can get.'
Whereas 38:2 invokes knowledge (and 'words without knowledge') as the
theme of the speech, 40:8 invokes condemnation and justification, i.e. the
notion of authority. Needless to say, the two aspects that we have identified
above involve precisely the way in which Job counteracts or at least ignores
the prevailing models of knowledge and authority.
Second, Job's reaction is fitting with regard to the topic of the speeches.
After the first speech, he basically admits that he has no knowledge to speak
of, and then he shuts up - which is an entirely appropriate reaction in a cog-
nitive context, i.e. in a context focusing on knowledge and wisdom. After the
second speech, by contrast, Job does something rather than say something:
he submits and repents. The reaction is not a cognitive one, but a symbolic
one; it involves making symbolical amends. Whereas Job's first reaction is a
cognitive recognition of ignorance, his second reaction is a symbolical rec-
ognition of impudence, followed by a ritual act of making amends.
And third, the text of the speeches, the argument that each of them con-
tains, is clearly different. In the second speech, we find images of power -
the two monsters - and of power over power. God conjures up the terrible
image of Leviathan and Behemoth, and further shows (in a positively hilari-
ous way) that he controls the monsters: 'Look, I can draw a straw through
the crocodile's nose and lift it up. Can you?' In the first speech, on the other
hand, we get images of the birth of the cosmos and the principles that organ-
ize it. Given the model of wisdom that we have been able to identify, this is
entirely appropriate: God displays his superior knowledge which is grounded
in his superior age, i.e. the fact that he (and not Job; compare 15:7) wit-
nessed and engineered the creation of the world. Once again, God is over-
whelming Job: 'Look, I know all of all this. Do you? And if you do not, how
can you question me?'
In short, then, God seems to be saying something like this: 'Well, Job,
you are asking me for a chance to defend yourself, but actually, you are not
in a position to ask. You challenge me to explain what is going on, but come
to think of it, you do not have enough wisdom to understand what is going
on, and moreover, you are not powerful enough to challenge me. '
There is, however, another step to take. What is the function of the sec-
ond part of the first speech? Chapter 39 does not contain cosmic images but
rather describes a number of animals (though not powerful monsters as in
chapters 40 and 41). One way of reading 39 is just as an addition to 38, as
further proof of God's powers of creation (and hence his unsurpassed
knowledge). I would argue, however, that God is adding a nuance to the first
part of his first speech. To be sure, we cannot really say that God is noticea-
bly subtle in his speeches. He uses extremely powerful images of cosmic and
animal forces, and Job is duly overpowered; predominantly, God is just put-
ting Job down. But in 39, a touch of subtlety does creep in.
Overall, the chapter presents multiple images of carefree and proud ani-
mals. What is the function of this? I would suggest that the key lies right in
the center of the chapter, where God describes the ostrich.
13 The wings of the ostrich wave proudly; but are they the pinions and
plumage of love?
14 For she leaves her eggs to the earth, and lets them be warmed on the
ground,
15 forgetting that a foot may crush them, and that the wild beast may tram-
ple them.
16 She deals cruelly with her young, as if they were not hers; though her
labor be in vain, yet she has no fear;
17 because God has made her forget wisdom, and given her no share in
understanding.
18 When she rouses herself to flee, she laughs at the horse and his rider.
The passage (and particularly 39: 17, which seems to mirror 28:20) adds a
crucial nuance to the conclusion that Job arrives at in 28:28. The ostrich is
presented as carefree (39: 13) but at the same time as careless: it does not
mind leaving its children in the desert (39: 14-15) nor treating them harsWy
(39: 16). Moreover (39: 17), it does not mind if its efforts are useless, and it
thinks higWy of its own strength, lightly attacking humans, i.e. beings that
are higher on the scale of creation (39: 18).
The ostrich, then, is the ironical counter-image of Job himself: Job does
worry, conspicuously so, about his children coming to harm and about the
fruit of his life's work being destroyed. And even if Job stands up against a
higher, divine power, he does not do so laughingly. This ironical context
gives specific depth to line 39: 17, which reads as an explanation of the os-
trich's behavior against the background of the hierarchy of knowledge: the
ostrich is carefree precisely because it has no insight. This could be called
the 'tree of knowledge' argument: because man has eaten from the tree of
knowledge, he is no longer happy and carefree like the animals. By implica-
tion, the very fact that Job does worry establishes that he is not entirely de-
void of insight.
God, in other words, refers to the hierarchy of wisdom to point out to Job
that he got his presuppositions wrong (chapter 38), but at the same time, he
subtly suggests (chapter 39) that the hierarchy of wisdom has to be extended
in another sense as well: starting from Job, the Great Chain of Being has to
be extended upwards from man to God, but it also has to be extended in a
downward direction from man to animal. So, we have to modify our sum-
mary of God's first speech slightly: 'Job, you are challenging me on the
point of wisdom. Well, you're in no position to do so, because my knowl-
edge is infinite and yours is limited. But if it can be of any comfort to you,
remember that your limited knowledge is precisely what distinguishes you
from the animals.' Job is asking questions that he cannot answer, but the
very fact that he can ask them demonstrates that he is closer to God than he
may be inclined to think.
7. The irony of God's speech 11
In an indirect, roundabout speech, replete with images and innuendo's, God
suggests that the felicity conditions for Job's challenging attitude are not
met, with regard to the question of wisdom just as well as with regard to the
question of power and authority. This is a relevant answer, given the hierar-
chical model of knowledge and authority that permeates the text. And be-
cause Job clearly agrees, both in thought and in behavior, God's reply can
only be considered an effective one.
But even if God's discourse is an appropriate and effective one, yielding
an answer that Job can genuinely (though perhaps uncomfortably) live with,
God is still being evasive and dismissive. He is sending Job off none the
wiser, or at least, just a tiny bit wiser. Shouldn't we expect a different an-
swer from God? It is important to make a distinction at this point between
the expectations we may have as a reader and the expectations we may have
as a believer. As believers, identifying with Job, we may either find God's
answer unsatisfactory, or we may accept the idea that God's logic escapes
us. As readers, however, we need not even ask the question, because we
know what the answer is. This is a recognition that is remarkably absent
from most discussions. If Job's question is: 'Why do you torment me so
while I am ilIDocent?', we as readers know the answer, and we have known it
from the very start of the book: God is tormenting Job merely because God
has been persuaded by the satan to test Job. God has been bragging about
Job, and being challenged by the satan to prove himself right, God lets the
satan have his way with Job.
As readers, then, we know very well that God's answer to Job, though
pragmatically relevant and conversationally effective, is not the true answer.
So why doesn't God give the true answer? At least one plausible solution, it
would seem to me, is to recognize that God may well be embarrassed. Even
if the satan in the story were not the devil of later Judaism and Christianity
but rather a member of God's courtly council (see Pagels 1995), God is not
exactly well-behaved by common human standards. He has been playing a
bad joke on Job, and he is not really in a position to confess to it. Job's insis-
tence creates a painful moment for God: he could not really admit that he has
been agonizing Job because he is his most faithful servant, could he?
The suggestion of an embarrassed God, outlandish though it may seem,
does not only follow from the narrative context. There are two other features
of the very final chapter of the book that fit in well with such an interpreta-
tion.
First, how can God declare that Job has spoken right of him when Job
himself admitted that he had no understanding and that he had been talking
nonsense? God's declaration in 42:7 can hardly apply to Job's earlier state-
ments, which were dismissed in 38:2 as lacking understanding. At the same
time, it can hardly apply to Job's newly acquired submissive attitude (and
silence), because it is not different enough from the deferential attitudes ad-
vocated earlier by Job's friends, who are now being punished. Logically
speaking, then, God's statement that Job has spoken right of him would
seem to be paradoxical on any account. However, if we take it as a gesture
of atonement, as a comforting signal that everything is alright now, as a way
of saying that in spite of everything, Job is still God's best friend, then of
course it matches the assumption that God somehow feels indebted to Job.
Second, the restoration of Job carries a hint of overcompensation. The
reversal is so abrupt and the reparation so exaggerated that it can only be a
sign of God's bad conscience. God is trying to make amends, and his feeling
of unease induces him to do so lavishly.
On the whole, then, coming back to our earlier discussion of the views of
Whedbee, Robertson, and Hoffmann, we can conclude that the opposition
between Whedbee on the one hand and Robertson and Hoffmann on the
other is not an exclusive one. God's speech is appropriate, but at the same
time, his behavior appears to be profoundly ironic, in a sense that may be
perceived by the audience and God himself, but hardly by Job. There is an
ironic discrepancy between God's show of power with regard to Job, and his
actual state of embarrassment. God's authority, his wisdom, and his strength
overpower Job, but the reader can recognize them as a disguise of God's
vulnerability. Likewise, there is an ironic discrepancy between the message
Caught in a web o{irony 323
of God's indirect speech act ('you're in no position to ask') and the message
that we assume God is trying to avoid ('I am in no position to answer').
8. The theology of God's embarrassment
Within the dramatic structure of the text, then, God's behavior exhibits an
ironical embarrassment. But what does this mean for our interpretation of
the text as a whole, if we think of it as a religious text? How should we in-
terpret God's embarrassment? I would like to suggest that we have two basic
alternatives.
If we contrast the conunon expectation of an onmipotent God with the
image drawn in the Book of Job - the image, that is, of an all too human,
fumbling, embarrassed God - then the text has an overall ironic meaning.
We could even conjecture that we are dealing here with a form of what we
would now identify as Jewish humor, with the author mocking his own ini-
tial, pre-existing image of God, and his own subordination to a God that he
does not entirely fathom. But as we know next to nothing about the way in
which the original audience may have understood the text (how old, in fact,
is Jewish humor?), it may be cautious to contrast the radically ironic inter-
pretation with a more charitable one: what if the Book of Job is trying to
make a specifically subtle point about the relationship between God and
man?
At this point, I am again drawn to the ostrich passage as a key for an al-
ternative interpretation. On the ladder of Creation, human intelligence has a
specific place. It alone can recognize the wonders of creation and the power
of God, but it can only do so at a tremendous price - the loss of innocence
and carefreeness. God's embarrassment at the enormity of the price, in re-
turn, would seem to ask for much more drastic measures than the mere res-
toration of Job. Perhaps the Job episode, this practical joke that almost got
out of hand, convinced God that he needed to set things straight with man-
kind on a much grander scale. And so Job would be vindicated, as a prefigu-
ration...
It should be clear that the context that could disambiguate between these
alternatives can hardly be found within the text itself. I have so far tried to
formulate an interpretation that is as consistent as possible with what we
know about the internal features of the text and the external context (such as
the patriarchal social structure) in which it originated. But there are natural
limits to this approach, and the choice between the alternative interpretations
suggested here far exceeds the contextualized methodology that I have tried
to follow. But that, I would say, is merely one more irony of this profoundly
ironical text.
Section 5
Lexicography
Chapter 13
The lexicographical treatment of prototypical
polysemy
Originally published in Savas L. Tsohatzidis (ed.), 1990, Meanings and Proto-
types. Studies in Linguistic Categorization 195-210. London: Routledge.
Conceptions of semantic structure should not only be theoretically interesting,
they should also fit the applied perspective of practical lexicography: large-scale
descriptive work of the lexicographical type provides a crucial testing ground for
theoretical conceptions. In this paper, I argue that a prototype-theoretical concep-
tion of semantic structure sheds an interesting light on specific problems of dic-
tionary making (like the linearization problem, involving the difficulty of map-
ping a multidimensional semantic structure onto a two-dimensional ordering of
definitions).
The paper is one of a number of publications in which I deal either with specific
questions about the lexicography of Dutch (1986b, 2000b, Geeraerts and Janssens
1982) or with matters pertaining to theoretical lexicography at large (l985b,
1987b, 1989b, 1997b, 2000a, 2003). There's a biographical background to this
interest in theoretical lexicography: I started my professional career working as a
historical lexicographer on the Woordenboek der Taal, the Dutch
counterpart (eventually completed in 1998) of the historical Oxford English Dic-
tionary. and later, I acted as editor-in-chief of the 14th edition of the Van Dale
Groot woordenboek van de Nederlandse taal, which is the main contemporary
dictionary of Dutch (Den Boon and Geeraerts 2005). My views on the importance
of prototype theory for semantic change (as laid down in my 1997 book on dia-
chronic prototype semantics) would never have come about had it not been for the
broad experience with historical meaning phenomena that I acquired when work-
ing on the Woordenboek der Nederlandsche Taal.
1. Lexicography and prototype semantics
There have hitherto been two main directions in the application of the proto-
typical conception of semantic structure to the problems of lexicography. On
the one hand, Wierzbicka (1985, 1987a, 1987b) has presented a method of
semantic analysis which replaces traditional dictionary definitions by ex-
haustive, in-depth analyses of the prototypes underlying natural language
328 Lexicography
concepts. On the other hand, I myself have argued that the prototypical
model of lexical polysemy is an indispensable aspect of any theory trying to
classify the diverse forms of semantic information that may be incorporated
in lexicographical reference works (Geeraerts 1985b, 1987b).
Both transpositions of prototype theory to the field of applied linguistics
differ basically with regard to the type of lexicographical problem they ad-
dress. In fact, while Wierzbicka is concerned with the microstructural pres-
entation problem, I have tried to (re)formulate the theoretical background of
the microstructural selection problem. The terminology used here is based on
the assumption that the problems of lexicography are concentrated on two
times two domains. On the one hand, dictionaries contain a macrostructure
(a set of lexical explananda) next to a microstructure (explanatory data for
each of the elements of the macrostructure); on the other hand, lexicogra-
phers are faced with problems of selection and representation with regard to
each of those two sets of elements (cp. Geeraerts 1984b). That is to say,
while the macrostructural selection problem basically involves the question
'What lexical items should be incorporated into the dictionary?', the ques-
tion 'What information should be presented with regard to the selected
items?' is raised on the level of microstructural selection. Similarly, while
the macrostructural presentation problem involves, among other things, the
choice between an alphabetical and a thematic organization of the selected
items, questions with regard to the arrangement of the distinct meanings of a
particular word (should they, for instance, be listed chronologically, logi-
cally, or according to frequency?) belong to the field of microstructural pres-
entation. Questions concerning the definitional techniques to be used with
regard to individual meanings likewise belong to the microstructural presen-
tation problem, and that is why Wierzbicka's approach is to be primarily
situated there as well. By contrast, my own attempts to identify and classify
the different types of semantic information (prototypical, stereotypical, tech-
nical) that can be incorporated into dictionaries are situated on the micro-
structural selectional domain.
In this respect, both approaches are basically complementary with regard
to each other. Whereas Wierzbicka's research is concentrated round the
question 'Given that one decides to incorporate the prototypical meaning of
natural language categories into one's dictionary, how should one define
them?', mine addresses the question what other kinds of meanings, next to
the prototypical one, might be selected for lexicographical treatment. This
basic methodological complementarity between the two approaches does not
imply, however, that the points of view developed in each case are necessar-
Lexicographical treatment o/prototypical polysemy 329
ily in accordance with each other. In particular, there are two remarks that I
believe should be made with regard to the practical relevance of Wierz-
bicka's point of view. In each case, the pragmatic purposes of dictionaries
play an important role in the argumentation.
A first and minor remark involves Wierzbicka's answer with regard to
her own basic question. Should prototypical meanings indeed receive the full
definitions of 'gargantuan length' (1985: 37) defended and admirably illus-
trated by her? From a theoretical point of view, she is undoubtedly right:
prototypical concepts are 'encyclopedic' entities that should be described in
all relevant aspects, and that cannot be reduced in the structuralist fashion to
minimal sets of contrastive features. This is a position that I have elsewhere
defended at length myself (1985b), and which I see no reason to take theo-
retical issue with. From the practical, applied-linguistic point of view of
lexicography, however, the question arises whether definitions should always
be as lengthy as Wierzbicka proposes. The point to be made is this: the kind
of information to be incorporated into a dictionary is ultimately determined
by the purposes the dictionary is likely to fulfill for its intended audience.
For certain types of audiences, it may not be necessary to have definitions
specifying the full conceptual detail of prototypical categories. This is par-
ticularly the case if the dictionary user can be expected to be familiar with
the concept in question; the definition should then do no more than enable the
dictionary user to identify the concept in question, that is, to retrieve it from
his own conceptual memory. Adult native speakers of English may, for in-
stance, be assumed to know what houses are; by definition, they are sup-
posed to know everything that is contained in a Wierzbickian analysis of the
concept, because this analysis is intended precisely to describe the native
speaker's conceptual knowledge. A definition of house in a dictionary for
adult native speakers of English may therefore be restricted to a brief gloss
identifying the concept in question Ca building for people to live in'). While
such a minimal rather than a maximal Wierzbickian definition may help the
dictionary user to distinguish between the entries devoted to house as a verb
and house as a noun, the general question nevertheless arises why such fa-
miliar words should then be incorporated into the dictionary at all. One
should not forget, however, that the entry in question will usually also con-
tain information that is more likely to lead to a consultation of the dictionary
to the extent that it is less familiar. (People may, for instance, wonder what
eat someone out of house and home, or bring the house down mean.) This
brings us straight to the second remark to be made with regard to Wierz-
bicka's application of prototype theory to lexicography, but before going
330 Lexicography
into that, it may be useful to fonnulate on a more general level the conclu-
sion that may be drawn from the present remark.
Most important is the recognition that there is a distinction between theo-
retical semantics and practical lexicography; a definitional technique that is
optimally justified from a theoretical point of view need not be so from the
pragmatic point of view of practical dictionary making. In this respect, it is
not my intention to claim that Wierzbickian definitions could never be used
when prototypical concepts are to be defined for adult native speakers; the
only point I am trying to make is that the theoretical preference for such
definitions cannot be transferred automatically and without further (prag-
matic) justification to practical lexicography.
Now, the impression that Wierzbicka tends to underestimate the distinc-
tion between practical lexicography and theoretical semantics is further cor-
roborated by the fact that she explicitly restricts her analyses to the proto-
typical concepts at the centre of natural language categories. Her solution
for the microstructural presentational problem thus seems to be coupled with
a very specific answer with regard to the question of microstructural selec-
tion, viz. that it is the salient senses of a lexical category that should be de-
fmed. It is not entirely clear whether she indeed believes that such an ap-
proach should be generalized to everyday dictionaries, but if so, this point of
view should receive some nuances. As suggested a moment ago, the kind of
things people may want to consult a dictionary for will very often involve
precisely the peripheral, less familiar applications of a category. From a
practical point of view (that is to say, with an eye to the pragmatic purposes
a dictionary has to meet for a particular audience), restricting the micro-
structural selection to prototypes may be useful for certain types of diction-
aries, but is certainly not generally applicable. Dictionaries have to provide
infonnation about the polysemy of words just as well as about their basic
meanmgs.
2. The linearization problem and its background
Once it is accepted (on the microstructural selectionallevel) that dictionaries
may wish to describe cases of prototypical polysemy, the microstructural
presentational problem presents itself: how can such a prototypically
polysemous cluster of meanings be adequately treated in the framework of a
traditional dictionary? In general, the problem is the following: whereas lexi-
cographical microstructures basically consist of a list of neatly separated,
Lexicographical treatment polysemy 331
consecutively numbered lexical meanings, the structure of prototypically
organized concepts is characterized by clustering and maximal overlapping.
How then can the multidimensional structure of prototypical concepts be
mapped onto the linear order of the dictionary? What I would like to show
here, is that there are a number of tricks and techniques that are intuitively
used by lexicographers to solve the linearization problem. This is entirely as
may be expected, or at least, if prototype theory is right, one may expect
lexicographers to be troubled by the linear order of the dictionary. If the
prototypical conception of categorization is correct, lexicographers should
be looking for ways to make the description more faithful with regard to the
perceived semantic structure of the category. That this is indeed the case
may be demonstrated by having a look at a dictionary entry for an indubita-
bly prototypical concept. Before, however, there are some preliminary re-
marks to be made.
In the first place, all dictionaries are not equally deviant with regard to
the canonical linear order. Once again, the pragmatic purposes of lexicogra-
phy appear to be very important. If it is sufficient for a particular dictionary
merely to enumerate the basic meanings of an item, without bothering too
much about their semantic interrelations, no specific attempts to render the
prototypical structure of the concept as adequately as possible should be
expected. The only intrusion of prototypicality will then probably be the fact
that the preponderant meaning will be enumerated first, less frequent ones
being mentioned near the end of the entry. But if a dictionary intends to give
a highly detailed, linguistically adequate description of the vocabulary of a
language, matters will be different. Not only will the prototypical structure
of the category become more apparent to the extent that more semantically
distinct examples of usage are incorporated into the description, but also, the
lexicographer will be more inclined to give an adequate description of proto-
typical polysemy to the extent that the scholarly level of his work rises.
That is to say, first, if a dictionary concentrates on describing only the
more salient senses of a word, the conceptual distance between those senses
will be greater than if it also incorporates less frequent kinds of usage; the
borderline cases in which the clustered structure of the prototypically organ-
ized category shows up most clearly will be precisely the less frequent appli-
cations. In this sense, one may stick roughly to the rule that larger dictionar-
ies are more likely to show the effects of the linearization problem.
Second, the pragmatic intention to render the semantic structure of a
category as adequately as possible is stronger in those dictionaries that are
compiled at least partly with a scholarly purpose. Whereas the ordinary,
332 Lexicography
everyday dictionary serves a primarily pedagogical purpose (transmitting
practical, communicatively useful information about a specific language or a
language variety), a number of dictionaries are explicitly conceived with the
primarily scientific rather than pedagogical purpose of giving a linguistically
adequate description of a language or a language variety. Clear cases are
historical dictionaries such as the Dictionary of the Older Scottish Tongue
and Bosworth and ToIler's An Anglo-Saxon Dictionary, or etymological
dictionaries: these works are first and foremost independent pieces of lin-
guistic research, in which pedagogical considerations only play a minor role.
Obviously, then, the burden of linearization will be more felt in dictionaries
for which linguistic descriptive adequacy is more important than pedagogical
efficiency.
This means that our example should preferably come from a large,
'scholarly' dictionary. Specifically, we shall use the entry vers from the
Woordenboek der Nederlandsche Taal, which is the major historical dic-
tionary of Dutch, ranging from 1500 up to the beginning of the twentieth
century. (Actually, the entry is labeled versch in accordance with the older
spelling of Dutch, but the contemporary spelling shall be used here.) In its
scope, its methods, and its nineteenth century origins, the Woordenboek der
Nederlandsche Taal is the Dutch equivalent of Murray's Oxford English
Dictionary and Grimm's Deutsches Worterbuch. Whereas the latter have
been completed for some time, the Woordenboek der Nederlandsche Taal,
however, has not yet reached completion. This is at least partly due to the
fact that the level of detail at which the dictionary works, is even higher than
in the other two; having progressed up to the letter W, the Woordenboek
now totals 33 volumes with an average of 1500 pages per volume. (For more
information about the Woordenboek, see Heestermans 1979.)
As a second preliminary remark, it should be noted that the linearization
problem has something to do with the traditional form of the dictionary as a
written text, but does not derive from it in any absolute fashion. On the one
hand, the necessity to list meanings in a linear way is imposed by the basic
linear structure of written text; on the other hand, graphic representations
such as those used in this article (see Figure 1) might circumvent the greater
part of the difficulties arising from the linearization problem. The fact, then,
that such graphic representations are never used in dictionaries should
probably be attributed partly to the force of lexicographical traditions, partly
to economical reasons. It should not be forgotten, in fact, that most diction-
aries are written under the pressure of financial constraints; with regard to
his possible audience just like with regard to his financial sources, it is in-
dispensable for the lexicographer to keep the final costs down. As the incor-
poration of figures showing the conceptual relations among the senses of a
word greatly enhances the required printing space, and hence, the cost of the
dictionary, this may indeed be a reason for not using them. In this respect, it
should be mentioned that computerized dictionaries open up new perspec-
tives for dealing with prototypically clustered polysemy. Not only is storage
space hardly a problem any more, but also, the access to the different mean-
ings of an item becomes much more flexible. It is not difficult to imagine, for
instance, how a graphic representation such as Figure I (together with ex-
planatory glosses) would form the basic semantic information in a particular
entry; further information at varying levels of specificity would then be re-
trievable by highlighting specific areas and points in the figure.
But developing blueprints for changes in lexicographical practice is not
the primary intention of this paper. Rather, as we are dealing with an analy-
sis of existing practices, the question now arises whether the linearization
problem is only created by the practical constraints just described, or
whether the canonical linear order also reflects a particular conception of
semantic structure. Is the linear dictionary model of semantic structure a
direct consequence of the classical theory of categorization against which
prototype theory reacts? Lakoff's propagandistic denunciation of the classi-
cal conception, which 'in the course of two thousand years ... has become so
entrenched that we tend not to even notice its presence' (1982: 14) seems to
suggest a positive answer to this question: if the classical theory has indeed
been as universally dominant as Lakoff claims, it is natural to suppose that
lexicographers have been influenced by it.
However, as I have tried to show on a number of occasions, traditional
conceptions of categorization are far from being as monolithically 'classical'
as Lakoff believes. First, the prestructuralist historical-philological tradition
of semantic research in linguistics is characterized by a fundamental meth-
odological similarity with the contemporary cognitive approach (Geeraerts
1988a). Second, Kantian and early phenomenological conceptions of phi-
losophical epistemology likewise correspond in basic respects with the view
of human knowledge implied by the prototypical theory of categorization
(Geeraerts 1985a). And third, clustered overlapping of lexical meanings is
not in principle ruled out by the classical theory. A closer look at the Aristo-
telian sources of the classical conception reveals that what is new about
prototype theory is not the discovery of clustered overlapping as such, nor
the recognition that lexical categories as a whole may not be definable by
means of a single set of necessary and sufficient characteristics, but rather
334 Lexicography
the fact that the overlapping is as a rule maximal and structural instead of
minimal and incidental, and more importantly, the fact that applications that
are intuitively considered to represent a single meaning, may not be analyti-
cally definable by means of a single set of necessary and sufficient attributes
(Geeraerts 1987a).
Given these nuances with regard to Lakoff's simplified view of the tradi-
tion, it is not entirely sure whether the canonical linear form of the dictionary
is indeed a straightforward consequence of the classical theory of categoriza-
tion. On the one hand, the classical expectation of minimal overlapping
among lexical meanings will probably have exerted a certain degree of influ-
ence, but on the other hand, it is equally probable that lexicographers,
through their extensive empirical familiarity with actual semantic structures,
have been long aware of the difficulties surrounding the classical expecta-
tions. This view of things is strengthened by the fact that lexicographical
manuals explicitly recognize these difficulties. It may be sufficient, for in-
stance, to see how Zgusta's influential Manual oflexicography discusses the
existence of peripheral and divergent uses of a word (1971: 47-49) to appre-
ciate the fact that the classical view is not necessarily the lexicographically
canonical one. And obviously, if the influence of theoretical conceptions on
the linear order of dictionaries is indeed restricted, the material constraints
mentioned earlier acquire more weight.
3. The structure of the problem
In order to see how the representational constraints imposed by the linear
order of the dictionary may be circumvented, we shall have to compare the
lexicographical description of our example vers with an independent seman-
tic analysis of that word. The latter may be found in Figure I, which gives a
prototypical analysis of the nineteenth century material for vers that is avail-
able in the corpus of the Woordenboek der Nederlandsche Taal. The re-
striction to the nineteenth century material is motivated by the traditional
attempt to avoid diachronic distortions in what is basically a synchronic
analysis. Because prototype theory undermines the dichotomous distinction
between diachrony and synchrony (cp. Geeraerts 1985d), the selection and
description of a single synchronic period as a static unity is clearly an ideali-
zation, but it is a necessary one for the analysis to start at all. There is a
further idealization to be mentioned: for the sake of simplicity, a small num-
ber of highly peripheral applications of the category has not been included in
the figure. As it stands, the figure sufficiently illustrates the prototypical
character of the category. For reasons of succinctness, a justification of the
analysis based on a detailed discussion and mutual comparison of the avail-
able quotations (which takes about thirty pages) is left out; by the same to-
ken, no illustrative quotations are given. Whereas the Arabic numerals refer
to specific, 'low level' meanings of vers, the Roman numerals indicate lar-
ger, 'higher level' groupings of those basic applications.
,..... i
r-iii
r-- ii
.....-y-
1:3 12 11
x-
xi-
I
14 4 19 18
~ I ~
6
17 5
~ M
3
xii -
~
15 7
20
16 a 10
21
9
Figure J. A prototypical analysis of the nineteenth-century material for vers avail-
able in the corpus of the Woordenboek der Nederlandsche Taal
336 Lexicography
In an optimal condition: not characterized by decay consisting in the
loss of certain attributes that are considered optimal
ii Optimal in a functional sense; optimal for use; optimal for the function
it has to fulfill in a particular context
iii Optimal in a material sense; in an optimal condition with regard to its
physical, material characteristics
iv Optimal for consumption; in the full possession of the useful and/or
pleasant attributes that motivate its consumption
v Recent, new; not having been present previously or for a long time in a
particular context
VI New as such; not having existed for a very long time
vii New by contrast; distinct from members of the same category that are
contextually implied or explicitly mentioned
(I) With regard to foodstuffs: recently produced and hence optimal for
consumption as far as taste and nutritive value are concerned; recently
harvested, caught, prepared etc., and hence not yet subject to decay
(2) With regard to foodstuffs: not preserved: not smoked, dried, salted,
sugared, frozen etc. to prevent decay
(3) With regard to organic materials in general: recently harvested, picked
etc., and hence not yet contaminated by a natural process of organic de-
cay
(4) With regard to artifacts: only recently produced, purchased, taken into
use etc., and hence not yet subject to wear or damage
(5) With regard to things that are inherently characterized by decay and
contamination, viz. with regard to manure: recently produced and
hence most powerful in its effects
(6) With regard to things that are inherently characterized by decay and
contamination, viz. with regard to wounds: recently made and hence
most painful
(7) With regard to organic materials: not yet contaminated by a natural
process of organic decay, in contrast with similar things that are men-
tioned or implied in the context
(8) With regard to people or animals: physically in an optimal condition, in
good shape, in contrast with other people or other animals that are
mentioned or implied in the context
(9) With regard to artifacts: not subject to wear or damage, in contrast with
other things that are mentioned or implied in the context
(10) With regard to foodstuffs: optimal for consumption, in contrast with
other foodstuffs that are mentioned or implied in the context
(11) With regard to foodstuffs: optimal for consumption, regardless of their
inherently or contextually recent character
(12) With regard to people: in an optimal physical condition, well rested,
not tired, regardless of any contrast with other people
(13) With regard to abstract things (such as memory impressions): vivid,
intense, powerful (regardless of recency)
(14) With regard to abstract things (such as memory impressions): having a
recent origin, and hence vivid, intense, powerful
(15) With regard to abstract things (such as courage): intense, powerful, in
contrast with other examples of the same category that are mentioned
or implied in the context
(16) With regard to people: without moral stains, uncorrupted, in contrast
with other people that are mentioned or implied in the context
(17) With regard to abstract things (such as one's conscience): recently
come into existence, and hence still free of moral stains
(18) Recently present or available in a particular context, regardless of any
contrast with other things, and regardless of inherent novelty
(19) Recent as such, inherently new, regardless of any contrast with other
things
(20) Recent as such, inherently new, and hence distinct with regard to other
things that are mentioned or implied in the context
(21) Recently present or available in a particular context, in contrast with
other things that are mentioned or implied in the context, though re-
gardless of inherent newness
In general, the method followed in drawing the figure is based on the two
criteria mentioned at the end of the previous paragraph: on the one hand,
there is the analytical attempt to define subsets of the observed applications
of vers, on the other, there are intuitive observations as to the distinctions
and similarities between those applications. For more details about the rela-
tionship between these two criteria for meaning discrimination, I refer once
again to the previously mentioned article (l987c). Briefly, it may be argued
that the basic discovery of prototype theory is the fact that what is intuitively
recognized as a single meaning need not be so from an analytic, definitional
point of view; for instance, the 'biological species' -reading of bird is intui-
tively recognized as a single lexical meaning of the word, in spite of the fact
that there is no single definition in terms of necessary and sufficient proper-
ties to cover precisely that reading. That is to say, the reading in question is
not ambiguous from an intuitive point of view (which may e.g. be demon-
strated by applying the identity test described by Zwicky and Sadock 1975),
but it is ambiguous from a definitional point of view (given the absence of a
single necessary-and-sufficient definition). It should be noted that the con-
verse relationship between the intuitive and the definitional criterion also
338 Lexicography
exists: autohyponymous terms such as dog are intuitively ambiguous ac-
cording to the identity test, but definitionally univocal according to the ne-
cessity-cum-sufficiency criterion.
Now, once it is accepted that semantic analyses are based on the inter-
play between both criteria, the representational dilemma facing the lexicog-
rapher in the case of vers may be analyzed in more detail. A linear represen-
tation of the semantic structure of vers might, in fact, enumerate either the
specific meanings or the higher level groupings. Both alternatives, however,
raise problems of adequacy. In its simplest form, the problem shows up in
the following way. On the one hand, merely enumerating the higher level
groupings obscures intuitively valid distinctions within those groups. On the
other hand, listing only the lower level specifications ignores the fact that
these may often be definitionally subsumed under a more general reading,
and would thus have to be considered cases of vagueness rather than ambi-
guity according to the analytical criterion. In actual fact, the structure of the
problem may be a bit more intricate, for instance in the sense that intuitive
distinctness need not be restricted to the lower level applications, but may
also apply to larger groupings. For expository purposes, however, we shall
restrict the discussion to the problem in its simplest form. An illustration of
the problem may then be found at the very centre of the category.
To begin with, let us note that vers has roughly the same meaning as the
English word fresh, except for the fact that the meaning 'moderately cool' is
absent from the semantic structure of verso As in English, vers prototypi-
cally applies to foodstuffs; fresh fruit and vegetables are in an optimal con-
dition for consumption because they have not been produced, harvested,
picked etc. for a very long time. As can be seen in Figure 1, this central
meaning is constituted by the overlapping of two concepts that may also
occur independently: on the one hand, the notion 'in an optimal condition',
on the other, the notion 'new, recent'. The intricacy of the semantic structure
of the word derives largely from the fact that each of these two notions may
receive further specifications and nuances which may then, of course, be
combined in the overlapping area. Moreover, additional distinctions may
occur within each of the areas determined in the manner just described. For
instance, the most central meaning 'with regard to food: recently produced
(etc.), and therefore optimal for consumption', which is represented by [1] in
the figure, gives rise to the particular nuance represented by [2]. Here, vers
means something like 'recently produced, and as such, not (yet) smoked,
salted, dried, frozen etc.; that is to say, not having been subjected to a pre-
serving treatment' .
From an intuitive point of view, [l] and [2] are distinct; something may
be vers in the sense of [I] though not in the sense of [2]. The expression vers
spek, for instance, is ambiguous. According to [2], it is a fixed expression
for the notion 'pork, unpreserved meat from pigs', contrasting with gerookt
spek and gezouten spek, 'bacon'. According to [I], however, vers spek
might for instance simply mean 'recently purchased pork or bacon' (regard-
less of preservation). Similarly, it would not be contradictory to say het
verse spek is niet vers meer (literally 'the fresh pork is no longer fresh')
meaning that the pork has gone off.
From an analytical, definitional point of view, however, the distinction
between [1] and [2] is a case of referential vagueness rather than semantic
ambiguity, as both concepts may be subsumed under the notion 'recently
produced'. In fact, as I have tried to show in the 1987c article, the distinc-
tion between vagueness and ambiguity is analytically determined by the rule
of thumb that the senses of a polysemous item should be defined as exten-
sionally broad as possible. In this way, different shades of red do not consti-
tute different meanings of red, as they can be subsumed under a general
definition of the color in question; the distinction between the political read-
ing of red and its interpretation as a color term does, on the other hand, rep-
resent a case of ambiguity rather than vagueness, since both readings cannot
be subsumed under a general definition sufficient to distinguish red from
other adjectives. Applying this criterion (which again may be traced to Aris-
totle) to vers does not only yield the conclusion that the distinction between
[1] and [2] is not a case of ambiguity, but actually reduces the semantic
analysis of the word as a whole to an enumeration of the largest groupings in
Figure 1. Disregarding some of the peripheral applications that are not in-
cluded in the figure (see however Figure 2), vers would then basically only
have the meanings 'in an optimal condition' and 'recent, new', as these are
the only ones that cannot be reduced to each other.
It should by now be clear how the Iinearization problem arises. Linearly
enumerating only low level specifications such as [1] and [2] is analytically
inadequate, as they analytically represent cases of vagueness rather than
truly semantic ambiguity (and it is, of course, ambiguity with which linguis-
tics should be concerned, not vagueness). Linearly enumerating only high
level groupings such as [i] and [v] is intuitively inadequate, as the intuitively
valid distinction between cases such as [1] and [2] is obscured. (In fact, the
whole prototypical area formed by the overlapping of [i] and [v] could never
be listed separately, as it can always be subsumed under either of both.)
Given, then, that a semantic analysis should indeed take into account ana-
340 Lexicography
lytical just as well as intuitive considerations, a simple linear representation
is out of the question. What semanticians and lexicographers should do, is to
describe the basic specifications together with the different higher level
groupings to which they give rise. And this automatically implies a descrip-
tion of the clustered nature of those groupings.
4. Circumventions of the Iinearization problem
The primary step to be taken by the lexicographer will obviously be the im-
position of a hierarchical structure on the semantic material. This is a way
of describing basic specifications together with higher level groupings that is
still rather close to the 'classical' conception of semantic structures, at least
in the sense that the ideal of rigid boundaries between different applications
of a concept may be maintained. As shown in Figure 2 (which is a transla-
tion of part of the structural description of vers in the Woordenboek der
Nederlandsche Taal) , the hierarchical description may reach considerable
depth, containing up to 6 distinct levels. (It should be mentioned that the
partial structure given in Figure 2 contains some of the minor nuances that
were not incorporated into Figure 1.) The point to be made is, of course, that
the neatness and the rigidity of the hierarchy do not adequately represent the
actual semantic structure of prototypical concepts, since the larger groupings
of basic specifications do not take the form of a taxonomy, but are rather
characterized by, first, multiple overlapping, and second, uncertain bounda-
ries. That is to say, the taxonomic model of semantic structure meets the
prototypical model half-way, in that it recognizes the necessity to describe
higher level groupings alongside basic specifications. On the other hand, it
still implies that each of the hierarchical subsets is well-defined, and that
there is no overlapping among them. How then can lexicographers remedy
these defects of a linearly hierarchical model? In particular, how does the
Woordenboek der Nederlandsche Taal cope with vers?
In the first place, the overlapping between the different groups of mean-
ings is explicitly indicated in the dictionary entry for vers by using cross-
references to other parts of the hierarchical structure. A prime example can
be found at the highest level of the structure. As mentioned in the previous
paragraph, the prototypical application of vers with regard to foodstuffs is
constituted by the overlapping of the concepts 'in an optimal condition' and
'new, recent'. This fact appears to be mirrored in the structure of the dic-
Lexicographical treatment ofprototypical polysemy 341
tionary entry by the fact that I is explicitly described as the conjunction of
the concepts represented by II and m.
In the second place, the vers-entry in the Woordenboek der Nederland-
sche Taal uses different means to indicate that each of the groupings in the
hierarchical structure may have unclear boundaries in that it may harbor
slightly deviant nuances. To begin with, it may be explicitly pointed out that
a particular group of senses have a 'core meaning' in common (see, e.g., the
definitions of II and m in Figure 2). Such a definitional technique is a warn-
ing for the dictionary user that the applications that are grouped together
need not be strictly identical with regard to the attributes mentioned in the
definition, but should rather be considered nuances of a central meaning
(which is then given by the superordinate definition). The subordinate senses
in the hierarchical structure are grouped together on the basis of similarity
rather than strict identity with regard to the concept identified by their su-
perordinate definition; the superordinate definition indicates the basis of the
similarity that holds among the subordinate senses, not the basis of their
identity. Within a taxonomical conception of a hierarchical structure, subor-
dinate senses merely add attributes to the superordinate definition; within a
prototypical conception, however, the subordinate specifications need not
possess all the attributes identified by the higher level definition, in the sense
that they may be related by similarity rather than taxonomic identity. This
fact, then, may be lexicographically signaled by using specific definitional
techniques, such as the explicit mentioning of the existence of a core mean-
ing rather than an essentialist definition in the classical sense.
Further, the non-taxonomical grouping of senses is achieved by using
definitions that are themselves combinations of similar but distinct concepts.
A very clear example is to be found at I-A-l in Figure 2. The use of the
and/or-specification, the presence of an optional element in the definition
(the one between brackets), and the incorporation of an open-ended enumera-
tion clearly mark the definition as a disjunctive rather than an essentialist
one. Again, what lies at the basis of the grouping is a similarity that allows
for variation and deviation rather than a taxonomical relationship that im-
plies essentialist identity. And again, non-classical definitional means are
used to convey this fact to the dictionary user.
Finally, there are structural next to definitional means for indicating that
hierarchically grouped senses may be related rather than taxonomically iden-
tical. In fact, their relatedness is not always based on similarity (as may have
been suggested by the foregoing discussion). For instance, as witnessed by
the lowest level in the partial hierarchical structure as presented in Figure 2,
342 Lexicography
VERS
I) In applications in which the
notions 'new' (cp.lI) and 'in
optimal condition' (cp. Ill) occur
in combination
A) In the central application,
with regard to foodstuffs
I) Recently produced and hence,
through its taste and/or nutritive
value, (pre-eminently suited for
consumption); recently har-
vested, caught, killed, perpared
etc., and hence not subject to
decay
a) In general
11) In application with the core
meaning 'in an optimal condi-
tion, not characterized by decay
consisting ofthe loss of certain
optimal characteristics that the
thing in question possesses'
B) In applications that link up
with the specification 'not
having come into existence of
having appeared for a long time
yet' of the basic notion 'new':
because of its recent character
not yet subject to the loss of its
original, optimal characteristics
2) Through specialization of I):
unpreserved, not having been
subject to a preserving treatment;
not smoked, dried, salted, sug-
ared, frozen etc.in order to
prevent contamination and decay
b) In specialized applications
Ill) In application with the core
meaning 'new, recent', either in
the sense 'not having come into
existence or having appeared for
a very long time yet', or in the
sense' other, distinct from what
was mentioned or implied
before'
C) In applications that link up
with the specification 'other,
distinct' ofthe basic notion
'new': distinct from what was
mentioned or implied before, viz.
because of its better condition
1
0
Often with regard to fish (viz.
herring), and then more particu-
larlyas a technical term in the
fishing industry: not salted
I
- Metonymically with regard to
industrial fishing: concerned
with fish that is not meant to be
salted or otherwise preserved,
but that is meant to be brought
directly to the market
2 Verse worst: unsalted and
unsmoked sausage
3 Verse waar: garbage from
slaughtered animals, meat from
animal organs, specifically when
sold as food for dogs or cats
Figure 2. Translation of part of the structural description of vers in the Woorden-
boek der Nederlandsche Taal
Lexicographical treatment polysemy 343
conceptual nuances that are structurally grouped together may be related
through metonymy rather than similarity. Within the group of applications 1-
A-I-b-I
o
, the metonymical nuance is structurally singled out: on the one
hand, its close relationship to the application defined by I-A-I-b-Io is indi-
cated by its subordinate status with regard to the latter, but on the other
hand, the fact that it is a metonymical extension that is not exactly identical
with I-A-I-b-Iois made apparent both by its distinct structural position
within that group, and by its being explicitly labeled as metonymical. In this
sense, the subordinate position is not an indication of a taxonomical relation-
ship, but rather of a close relationship that is yet to be distinguished from
true identity.
There is ample evidence, in short, that lexicographers use hierarchical
structures in a non-taxonomical way; various provisions are made to repre-
sent the indeterminacy and overlapping of the hierarchical groupings. To
round off the discussion, notice that Figure 2 contains an example where
both structural characteristics are interwoven. The definition under III is one
in which a core meaning is explicitly identified, while at the same time, the
two major nuances of that core application are mentioned; as such, it is
made clear that the grouping of meanings under III (the structure of which is
not represented in detail in the figure) is based on similarity rather than strict
identity. But as I is constituted by the overlapping of 11 and Ill, the main
nuances of III should also be identified within I. This is indeed the case: I-B
and I-C contain an explicit cross-reference to the main nuances that are
grouped together under Ill.
5. Conclusions
There are a number of points that may be retained from the foregoing dis-
cussion. Basically, there is no reason to identify the classical conception of
categorization with an allegedly lexicographical model of definitional struc-
tures. An examination of an actual dictionary entry reveals that neither a
classically linear or a semi-classical taxonomic model is followed. Rather, a
hierarchical structure is used to accommodate the tension between analytical
and intuitive distinctness of meaning, i.e., to incorporate both basic specifi-
cations and higher level groupings into the entry. At the same time, the hier-
archical structure is modified in several ways in order to bring out most
clearly the indeterminate and clustered nature of those groupings. It follows
from this observation that prototype semantics is well suited as a theoretical
344 Lexicography
basis for a lexicographical metatheory, since it accurately models the kind of
semantic phenomena that lexicographers have to face up to in their prag-
matically restricted descriptive activities. In this respect (and given the re-
sults mentioned at the beginning of the article), a further rapprochement
between the recent, cognitive developments in linguistic semantics on the one
hand and lexicography on the other may be envisaged. For one thing, lexico-
graphical metatheories had better abandon all hope of building an adequate
theory of lexicography on a linguistic basis that does not conform with their
own empirical experience; in particular, extant attempts to impose a struc-
turalist conception of semantic structure on the methodology of lexicography
should be rejected as misguided: there are now simply much more suitable
semantic theories available. But the rapprochement could not be unidirec-
tional: it would seem that prototype semantics might profit a lot from a
closer acquaintance with the detailed descriptions of prototypical structures
contained in large-scale traditional dictionaries such as the Woordenboek
der Nederlandsche Taal.
Chapter 14
The definitional practice of dictionaries and the
cognitive semantic conception of polysemy
Originally published in Lexicographica 2001,17: 6-21.
This chapter puts the previous one in a wider perspective. First, while the previous
paper focuses on a specific type of dictionary (the large-scale diachronic diction-
ary of the Oxford English Dictionary type), the present takes its examples from
common desk dictionaries. Second, while the previous chapter deals with the
relationship between prototypicality and lexicographical definitions only, the
present refers to other lexicographically relevant aspects of Cognitive Semantics
as well. And third, while the previous chapter basically addresses an audience of
linguists, trying to persuade them of the interest of lexicographical data for se-
mantic theorizing, the present paper takes the reverse perspective, presenting
relevant aspects of Cognitive Semantics to an audience of lexicographers who are
not necessarily acquainted with recent developments in theoretical semantics.
(Section 2 of the paper, then, will be quite familiar for readers who have pro-
gressed up to this point in the present volume.)
In the years between the initial publication of the article and the present publica-
tion, the situation has changed drastically with regard to at least one of the desid-
erata mentioned at the end of the article: in general, the use of corpus data in
Cognitive Linguistics has increased significantly (see Tummers, Heylen, and
Geeraerts 2005 for a critical overview), and more specifically, the use of existing
collocational techniques has been popularised through the 'collostructional' ap-
proach of Gries and Stefanowitsch (among many other publications, see Ste-
fanowitsch and Gries 2003, Stefanowitsch and Gries 2003).
1. Gist of the argument
Cognitive Semantics has had a major impact on lexical studies: more than
any other recent theory, it has in the last fifteen years led to a renewed inter-
est in lexical research. But what about lexicography? What is the relation-
ship between Cognitive Semantics and lexicography? And specifically, in the
context of the present issue of this journal, what is the impact of Cognitive
Semantics on the lexicographical treatment of polysemy? In what follows, I
will try to answer that question - succinctly and perhaps a bit sketchily, but
346 Lexicography
also as fair as possible to both the theoretical, lexicological and the practical,
lexicographical part of the comparison. I emphasize my attempt at fairness,
because I deliberately want to avoid any form of theoretical imperialism. It
is definitely not an assumption of this paper that theoreticians in principle
have the answers, and that lexicographers simply have to follow. I have
pointed out earlier (Geeraerts 1997a: 5) that there should be a relationship of
mutual inspiration between both disciplines, and even though the present
paper looks from lexicological theory to lexicographical practice rather than
the other way round, I still hold the reverse perspective to be equally impor-
tant.
In general, I will argue that a number of existing definitional and descrip-
tive practices in the dictionary that are somewhat suspect from an older
theoretical point of view receive a natural interpretation and legitimacy in the
theoretical framework offered by Cognitive Semantics. More specifically,
there are three aspects of the Cognitive conception of lexical semantic struc-
ture that have to be discussed: the importance of prototypicality effects for
lexical structure, the intractability of polysemy, and the structured nature of
polysemy. I will argue that each of these points inspires a specific conclusion
for lexicographical practice, or at least, that it vindicates existing aspects of
lexicographical practice.
First, the importance ofprototypicality effects for lexical structure blurs
the distinction between semantic information and encyclopedic information.
This does not entail that there is no distinction between dictionaries and en-
cyclopedias as types of reference works, but rather that references to typical
examples and characteristic features are a natural thing to expect in diction-
anes.
Second, the intractability ofpolysemy involves the absence of a coherent
set of criteria for establishing polysemy; a more charitable way of wording
things would be to say that distinctiveness between senses of a lexical item is
to some extent a flexible and context-based phenomenon. Dictionaries, then,
will use various definitional techniques to accommodate the flexibility of
meamng.
Third, the structured nature of polysemy involves, basically, the radial
set structure of polysemy. While lexicography has certainly never denied the
existence of links between the various readings of a lexical item, Cognitive
Semantics has added a number of new insights: the clustered nature of
polysemic structures is now being analyzed in more detail than ever. For
lexicography, this implies a recognition of the linearization problem that
traditional dictionaries face.
Dejinitional practice ofdictionaries 347
Given these three focal points, the paper has a more or less deductive
structure. I will first present the facets of Cognitive Semantics that I want to
focus on. (I will not, however, try to present an overall introduction to the
Cognitive approach: see Taylor 1995, Ungerer and Schmid 1996, Palmer
1996, Violi 1997, Dirven and Verspoor 1998 for introductory volumes. The
discussion in section 2 will be largely superfluous for readers familiar with
Cognitive Semantics.) Next, I will identify the specific expectations with
regard to lexicographical practice that may be deduced from this theoretical
analysis, and then proceed to show that these predicted features are indeed
part and parcel of actual lexicographical practice - in spite of what might be
expected on the basis of other theoretical approaches to semantics. The pa-
per closes with an attempt to place these observations in a wider context the
relationship between lexicography and Cognitive Semantics is not exhausted
by the discussion of polysemy.
2. A primer of Cognitive Semantics
What are the structural characteristics of semasiological structures high-
lighted by Cognitive Semantics? A good starting-point to define some of the
topics central to Cognitive Semantics is provided by the well-known distinc-
tion between the level of senses and the level of referents (in logic-semantic
parlance, between intension and extension).
Consider the word fruit. This is a polysemous word: next to its basic,
everyday reading ('sweet and soft edible part of a plant, containing seeds'),
there are various other readings conventionally associated with the word. In
a technical sense, for instance ('the seed-bearing part of a plant or tree'), the
word also refers to things that lie outside the range of application of the ba-
sic reading, such as acorns and pea pods. In an expression like the fruits of
nature, the meaning is even more general, as the word refers to everything
that grows and that can be eaten by people (including, for instance, grains
and vegetables). Further, there is a range of figurative readings, including
the abstract sense "the result or outcome of an action' (as in the fruits ofhis
labor or his work bore fi'uit), or the somewhat archaic reading 'offspring,
progeny' (as in the biblical expressions thefruit ofthe womb, thefruit o(his
loins)
Each of these readings constitutes a separate sense offruit, but in turn,
each sense may be thought of as a set of things in the outside world. The
basic sense of fruit, for instance, corresponds with a set including apples,
348 Lexicography
oranges, and bananas (and many other types of fruit). If you think offruit in
this central sense as a category, the set consists of the members of the cate-
gory. These members are 'things' only in a broad sense. In the fruit example,
they happen to be material objects, but in the case of verbs, they could be
actions, or situations, or events; in the case of adjectives, they could be
properties; and so on. Also, the 'things' featuring in the set need not exist in
the real world. The set contains all real and imaginary apples and oranges
(etc.) that fruit could possibly name, in the same way in which goblin will
have a set of members associated with it, regardless of whether goblins are
real or not.
Given the distinction between the intensional and the extensional level of
semasiological analysis, we can now describe three structural characteristics
that receive specific attention within a Cognitive Semantic framework.
2.1. Differences of structural weight
Differences in salience involve the fact that not all the elements at one level
of analysis have the same structural weight. On the semantic level, for in-
stance, the everyday reading of fruit occupies a more central position than
the archaic reading 'offspring' or the technical reading. Various indications
may be adduced for this central position. For one thing, the central reading
more readily springs to mind when people think of the category: on being
asked what fruit means, you are more likely to mention the edible parts of
plants than a person's offspring. For another, the 'edible part' reading is
more frequent in actual language use.
In addition, the 'edible part' reading is a good starting-point for describ-
ing the other readings. It would probably be more easy to understand the
expression frUit of the womb (if it is new to you) when you understand the
'edible part' reading than the other way round. The basic reading, in other
words, is the center of semantic cohesion in the category; it holds the cate-
gory together by making the other readings accessible. Three features, in
short (psychological salience, relative frequency of use, interpretative advan-
tageousness), may be mentioned as indications for the central position of a
particular reading.
Centrality effects are not restricted to the level of senses, however, but
may also be invoked at the referential level. When prompted, Europeans will
more readily name apples and oranges as types of fruit than avocados or
pomegranates, and references to apples and oranges are likely to be more
frequent in a European context than references to mangos. (This does not
exclude, to be sure, cultural differences among distinct parts of Europe.)
The terminology used to describe these differences of structural weight is
quite diverse, and the description in the foregoing paragraphs has featured
such (intuitively transparent) terms as salience, typicality, and centrality.
The most technical term however is prototypicality: the central reading of an
item or the central subset within the extensional range of a specific reading is
the prototype. The linguistic literature on prototypes is by now vast. Apart
from the introductory works mentioned above, see Mangasser-Wahl (2000)
for an interesting overview of the development of the approach.
2.2. Demarcation problems
The elements at one particular level of the semasiological analysis need not
necessarily be clearly distinguishable with regard to each other. As an illus-
tration, let us consider the question whether the central sense offruit can be
delimited in a straightforward fashion. Such a delimitation will take the form
of a definition that is general and distinctive: it is general in the sense of
naming characteristics that are common to all fruits, and it is distinctive in
the sense of being sufficient to distinguish the category fruit (in the relevant
sense) from any other category. (If a definition is not distinctive, it is too
general: it will cover cases that do not belong in the category to be defined.)
Now, many of the characteristics that one might be inclined to include in
a definition of the central reading offruit do not have the required generality:
they are not necessarily sweet (lemons), they do not necessarily contain parts
that are immediately recognizable as seeds (bananas), they are not necessar-
ily soft (avocados). There are, to be sure, a number of features that do have
the required generality: all fruits grow above the ground on plants or trees
(rather than in the ground); they have to ripen before you can eat them, and
if you want to prepare them (rather than eat them raw), you would primarily
use sugar, or at least use in them in dishes that have a predominantly sweet
taste. Taken together, however, these features do not suffice to prevent al-
monds (and other nuts), or a vegetable like rhubarb (which is usually cooked
with sugar), from being wrongly included into the category that is to be de-
fined.
We have to conclude, then, that the central sense ofjruit cannot receive a
definition that is both general and distinctive. If we shift the attention to the
referential level, similar effects may be observed: the borderline of categories
350 Lexicography
is not always clearly delineated. For instance, is a coconut or an olive a
fruit?
Definitional difficulties like the one just illustrated are enhanced by the
existence of various kinds of tests for distinguishing between vagueness and
polysemy - and specifically, by the fact that the existing tests may yield
results that are to some extent divergent with regard to each other. To briefly
illustrate the main point, and without discussing all specific tests that have
been suggested, three types of criterion can be distinguished.
First, from the truth-theoretical point of view taken by Quine
(1960,129), a lexical item is polysemous if it can simultaneously be clearly
true and clearly false of the same referent. Considering the readings 'harbor'
and 'fortified sweet wine from Portugal' ofport, the polysemy of that item is
established by sentences such as Sandeman is a port (in a bottle), but not a
port (with ships).
Second, lingUistic tests involve acceptability judgments about sentences
that contain two related occurrences of the item under consideration (one of
which may be implicit or deep-structural); if the grammatical relationship
between both occurrences requires their semantic identity, the resulting sen-
tence may be an indication for the polysemy of the item. For instance, the
identity test described by Zwicky and Sadock 1975 applies to constructions
that were assumed in the Chomskyan 'standard theory' to involve transfor-
mations such as conjunction reduction and so-reduction, which require the
semantic identity of the items involved in the reduction. (It may be noted that
constructions such as these are nowadays no longer discussed in transforma-
tional terms. However, as the current term 'identity-of-sense anaphora' indi-
cates, the idea that there are semantic restrictions on the construction re-
mains intact.) Thus, at midnight the ship passed the port, and so did the
bartender is awkward if the two lexical meanings of port are at stake; disre-
garding puns, it can only mean that the ship and the bartender alike passed
the harbor (or, perhaps, that both moved a particular kind of wine from one
place to another). A 'crossed' reading in which the first occurrence of port
refers to the harbor, and the second to wine, is normally excluded. Con-
versely, the fact that the notions 'vintage sweet wine from Portugal' and
'blended sweet wine from Portugal' can be crossed in Vintage Naval is a
port, and so is blended Sandeman indicates that port is vague rather than
polysemous with regard to the distinction between blended and vintage
wmes.
Third, the dejinitional criterion (as informally stated by Aristotle in the
Posterior Analytics II.xiii) says that an item has more than one lexical
meaning if there is no minimally specific definition covering the extension of
the item as a whole, and that it has no more lexical meanings than there are
maximally general definitions necessary to describe its extension. Definitions
of lexical items should be maximally general in the sense that they should
cover as large a subset of the extension of an item as possible. Thus, sepa-
rate definitions for 'blended sweet fortified wine from Portugal' and 'vintage
sweet fortified wine from Portugal' could not be considered definitions of
lexical meanings, because they can be brought together under the definition
'sweet fortified wine from Portugal'. On the other hand, definitions should
be minimally specific in the sense that they should be sufficient to distin-
guish the item from other non-synonymous items. A maximally general defi-
nition covering both port 'harbor' and port 'kind of wine' under the defini-
tion 'thing, entity' is excluded because it does not capture the specificity of
port as distinct from other things.
Now, the existence of various polysemy tests is non-trivial to the extent
that they need not always (in contrast with the port example) yield the same
results. In the case of autohyponymous words, for instance, the definitional
approach does not reveal an ambiguity, whereas the Quinean criterion does.
In fact, given that dog is autohyponymous between the readings 'Canis fa-
miliaris' and 'male Canis familiaris', the latter definition is not maximal,
because it defines a proper subset of the Canis familiaris' reading; the sen-
tence Lady is a dog, but not a dog, on the other hand, is not ruled out. Such
divergences between polysemy tests occur on a larger scale: see Geeraerts
(1993a), and the further discussion in Tuggy (1993), and specifically Cruse
(2000a). Following up on the lead provided by Geeraerts (1993a), Cruse
(2000a) systematically explores various configurations of divergence and
convergence of the polysemy tests. From the point of view of the present
article, this line of discussion is important because it lends further support to
the recognition that there is not necessarily a unique and optimal solution to
drawing dividing lines around and between the meanings of a lexical item.
2.3. Multidimensional structural relations
The relationship that exists between the various elements at each level of the
analysis is not restricted to the quantifiable phenomena described in section
2.1: the links between those elements may also be described in a more quali-
tative way. On the level of senses, in particular, it appears that the relation-
ship between the meanings of a word may be described in terms of a more or
less limited set of basic conceptual links. The senses ofjruit, for instance, do
352 Lexicography
not exist in isolation, but they are related in various ways to the central sense
and to each other. The technical reading (,seed-containing part') and the
sense illustrated by the fruits ofnature are both related to the central mean-
ing by a process of generalization. The technical reading generalizes over the
biological function of the things covered by the central meaning, whereas the
meaning 'everything that grows and that can be eaten by people' focuses on
the function that those things have for human beings. The figurative uses, on
the other hand, are linked to the other meanings by a metaphorical link, but
notice also that the meaning 'offspring' is still closer to the central sense,
because it remains within the biological domain. The overall picture, in
short, takes the form of a cluster of mutually interrelated readings.
This observation is, of course, a familiar and time-honored one in lexical
semantics: the terminology used to describe the links among senses origi-
nated with diachronic semantics in the late nineteenth century, with the very
birth of lexical semantics as a separate subdiscipline of linguistics. What is
new in Cognitive Semantics, though, is the emphasis on the overall structure
of the related meanings rather than on the individual links: an emphasis on
the multidimensional nature of the overall structure, and an emphasis on the
cohesive role of prototypical centers within such structures. In fact, multidi-
mensional analyses of the semantic structure of lexical items are a common
feature of Cognitive Semantics: they are a crucial feature of the radial set
model of semantic structure that has become popular through the work of
Brugman (198 I) and Lakoff (1987) and many others. Apart from prototype
theory as described above and the conceptual metaphors introduced by La-
koff and Johnson (1980), the radial set model of semantic description may
indeed be the most widely known feature of Cognitive Semantics.
2.4. Summarizing the position of Cognitive Semantics
The semasiological model arising from the previous pages may be summa-
rized in the graphical format of Figure 1. Without in any way being exhaus-
tive, the picture shows how a word like fruit can, on a first level of analysis,
be associated with various senses. On a second level of analysis, each of
those senses is itself associated with a set of referents. These sets are repre-
sented in a form that resembles the representation with Venn-diagrams that
is usual in mathematics. Examples of entities at the referential level are in-
cluded only in the set associated with the sense 'edible part'. This is, of
course, a matter of graphical economy rather than principle. At each level,
specific structural characteristics have to be taken into account. Of the three
basic characteristics mentioned above, two have received a graphical expres-
sion in the picture. The differences of centrality and structural weight among
the elements at each level are indicated by drawing them in different sizes
and by topologically ordering them in a way that reflects the dine from cen-
ter to periphery. The structured nature of the relations between the elements
is indicated by an explicit identification of the relevant links. (Again for
reasons of graphical economy, this is restricted to the level of senses.)
fruit
apple, orange,
pear, banana
pomegranate
seed-bearing part edible part result of
of plant .. of plant an action
GENERALIZATION METAPHOR
Figure I. The semantic structure offruit
We have now reached the point where we can summarize the specific ap-
proach taken by Cognitive Semantics in just a few points.
First, it was suggested above that there is an outspoken homology be-
tween the structure of the referential level and the structure of the semantic
level: the same structural characteristics shape both levels. This suggests
that both levels are less far apart than has traditionally been thought. The
structuralist view of lexicology in particular tends to suggest that only the
354 Lexicography
semantic level (the level of senses) is worthy of linguistic analysis. By con-
trast, the so-called Cognitive Semantic conception that has meanwhile arisen
in opposition to this earlier structuralist view, stresses the fact that the refer-
entiallevel has to be included in the analysis.
Second, while structuralist approaches to semantics tend to be reluctant
to take into account differences of structural weight and demarcational fuzzi-
ness, Cognitive Semantics readily accepts these phenomena as relevant as-
pects of semantic structure.
And third, linking up with prestructuralist semantics, Cognitive Seman-
tics puts a new emphasis on the multidimensional, clustered nature of sema-
siological structures.
3. From theory to practice
Now, what would be the consequences for lexicographical practice? Or
rather, if the Cognitive conception of semantic structure is by and large cor-
rect, what could we expect to find in actual dictionaries? The three charac-
teristics highlighted in the previous paragraph lead to the following hypothe-
ses.
To begin with, if it is correct that the referential level of semantic struc-
ture is part and parcel of a proper semantic description, we may expect dic-
tionaries to include references to that level - in spite of the traditional, strict
distinction between the semantic and the encyclopedic level of description. In
particular, we may expect dictionaries to refer to prototype instances of
categories or to typical (rather than general) features of the members of
those categories.
Further, if it is correct that the description of meaning has to come to
terms with fuzziness, demarcation problems, and non-uniqueness, we expect
dictionary definitions to use definitional methods that take into account these
characteristics. Instead of definitions that rigidly take the form of separately
general and mutually distinctive features, we expect the intrusion of unor-
thodox definitional methods such as enumerations, disjunctions, and the
accumulation of near-synonyms.
And finally, if it is correct that semantic structures predominantly take
the form of a multidimensional radial set structure, we may expect dictionar-
ies to face a linearization problem: how can the multidimensional nature of
the semantic structures be mapped onto the linear order of the dictionary?
In the following subsections, these expectations will be confronted with actual
examples. It will be shown that the expectations are basically correct.
3.1. Pratotypicality effects in lexical structure
Consider the following definitions (of separate meanings or idiomatic ex-
pressions) fram the New Shorter Oxford English Dictionary (cd-ram ver-
sion, 1997).
abiogenesis The production of organic matter or compounds, other than by
the agency of living organisms; esp. the supposed spontaneous generation
of living organisms.
baritone A I The male voice between tenor and bass, ranging typically
from lower A in the bass clef to lower F in the treble clef; a singer having
such a voice; a part written for such a voice.
cup b An ornamental vessel, typically of silver and comprising a bowl with
a stem and base, that is offered as a prize in a competitive event.
defoliate Remove the leaves from; cause the defoliation of, esp. as a mili-
tary tactic.
dwarf A I b Any of a mythical race of diminutive beings, typically skilled
in mining and metalworking and often possessing magical powers, figur-
ing esp. in Scandinavian folklore.
hear! hear! An exclam. calling attention to a speaker's words, e.g. in the
House of Commons, and now usu. expressing enthusiastic assent, occas.
ironical derision.
heart 5 A central part of distinct confonnation or character, e.g. the white
tender centre of a cabbage, lettuce, etc.
honours of war Privileges granted to a capitulating force, e.g. that of
marching out with colours flying.
model 2 a (fig.) A person or thing resembling another. esp. on a smaller
scale.
tea 5 A meal or social gathering at which tea is served. Now esp. (a) a
light afternoon meaL usu. consisting of tea, cakes, sandwiches, etc. (also
more fully afternoon tea, five o'clock tea); (b) (in parts of the UK, and in
Australia and NZ) a main meal in the evening that usually includes a
cooked dish, bread and butter, and tea (also more fully high tea)
356 Lexicography
tee A conical metallic structure, usually hung with bells, surmounting the
pagodas of Myanmar (Burma) and adjacent countries.
thimblerig A sleight-of-hand game or trick usually played with three in-
verted thimbles and a pea, the thimbles being moved about and bystanders
encouraged to place bets or to guess as to which thimble the pea is under.
In each of these definitions, words such as especially, e.g., typically,
usually and often introduce descriptive features that are not general but that
rather identify typical (prototypical, if one likes) characteristics or instances
of the category. Within a structuralist conception of semantics, this would be
inadmissible, because these elements belong to the 'encyclopedic' level
rather than the semantic level. In actual practice, however, this prototype-
oriented definitional technique can hardly be called exceptional in the context
of the dictionary as a whole. The expression esp., for instance, is used no
less than 28335 times in 18274 entries in the dictionary as a whole.
Does this mean, by the way, that the difference between dictionaries and
encyclopedias is a spurious one? The question asks for a brief excursion. An
early discussion of the question between Haiman (1980a) and Frawley
(1981), with a further reply by Haiman (1982), provides a good starting-
point for delimiting the Cognitive point of view (for a more recent discussion
of the theoretical question, see the contributions in Peeters 2000). On the one
hand, the theoretical basis for a distinction between dictionaries and ency-
clopedias cannot be provided by the structuralist approach (as in Lara
1989): it is a crucial aspect of Cognitive Semantics that the distinction be-
tween the two levels of description is not as strict as presupposed by the
structuralist doctrine. On the other hand, there is a practical difference be-
tween dictionaries and encyclopedias that need not be abolished: there is a
difference in scope and content between, say, the Encarta or the New
Shorter Oxford English Dictionary, or between the Encyclopaedia Britan-
nica and the Oxford English Dictionary, and no cognitive linguist would
argue against the distinction.
This distinction basically resides in two features. Macrostructurally, the
encyclopedia focuses on proper names, nouns, and maybe a number of other
elements from open word classes, whereas the dictionary includes all word
classes (typically excluding all or most proper names). Microstructurally,
the encyclopedia focuses on expert information as provided by scientific,
technical, or professional experts, whereas that information is only one of
the types of semantic description that the dictionary may include, together
with the more everyday uses of the words.
Definitional practice ofdictionaries 357
But if Cognitive Semantics accepts this distinction, how can it justify it?
As a theoretical background for the distinction between the type of informa-
tion typically included in encyclopedias and that included in dictionaries, we
need a 'sociosemantic' theory: a theory about the distribution of semantic
knowledge within a linguistic community. Scientific, technical, professional
information is, in fact, primarily information that is produced and certified
by a specific group of people - the experts, who are recognized by the com-
munity as such and on whom the community relies when expert knowledge is
at stake. Although no such 'sociosemantic' theory is as yet available with
any reasonable degree of comprehensiveness, a starting-point is provided by
Putnam's theory of the 'division of linguistic labor' (1975), which explicitly
distinguishes between extensional concepts (the expert's knowledge) and
stereotypes (the basic semantic knowledge that language users are supposed
to possess if they are to count as full-grown members of the linguistic com-
munity). A combination of Putnam' s approach with prototype theory is not
impossible (see Geeraerts 1985b, 1987b): if a prototypically organized con-
cept combines all the various nuances with which a lexical item may be used
within a linguistic community, then extensional and stereotypical concepts
are particular members of the full prototypical set of applications of an item.
Extensional concepts are characterized by their expert nature, whereas
stereotypes represent the minimal amount of semantic knowledge that the
language user is supposed to possess if he is to count as mastering the lan-
guage. Roughly speaking, stereotypes are likely to coincide with the most
common, most central senses within a prototypical cluster: what people are
primarily supposed to know are the central readings of the cluster.
This recognition of a possible theoretical combination of prototype theory
and a theory of the division of linguistic labor yields a theoretical framework
for reference works that naturally provides a place for both the encyclopedia
and the dictionary (see Geeraerts 1985b, 1987b). In fact, three basic types
may be distinguished.
First, technical, professional, scientific expert knowledge is treated in en-
cyclopedias and terminological dictionaries.
Second, the full prototypically organized set of senses of a lexical item,
including nuances and less frequent or more specialized readings, is treated
by large-scale dictionaries, of the size represented by (to name just a few)
the New Oxford Dictionary of English or Merriam Webster's Collegiate
Dictionary, and any dictionary beyond that size.
358 Lexicography
And third, standard desk dictionaries can be related to the notion of
stereotype: they make a selection from the full prototypical set by presenting
only the most central, most frequent senses.
Closing the excursion, we may conclude that a Cognitive Semantic con-
ception of the relationship between semantic and encyclopedic knowledge
does not preclude a theoretical justification for the distinction between dic-
tionaries and encyclopedias as different types of reference works.
3.2. The intractability of polysemy
Definitional demarcation problems show up in the fact that dictionaries ap-
pear to use definitional techniques that are 'unorthodox' from the point of
view of a traditional conception of meaning. Consider the following set of
entries, again from the New Shorter Oxford English Dictionary. (The entries
are rendered in a reduced form: etymologies, quotations, dates, and a number
of labels have been left out.)
primer I
1 A prayer-book or devotional manual for the laity.
2 An elementary textbook (orig. a small prayer-book) used in teaching chil-
dren to read.
b A small introductory book on any subject; fig. something introducing or
providing initial instruction in a particular subject, practice, etc..
c (A child in) an elementary class in a primary school.
3 A size of type. Chiefly & now only in great primer, long primer.
primer 11
I a = priming-wire.
b A cap, cylinder, etc., containing a compound which responds to friction,
electrical impulse, etc., and ignites the charge in a cartridge etc.2 A sub-
stance used as a preparatory coat on previously unpainted wood, metal,
canvas, etc., esp. to prevent the absorption of subsequent layers of paint or
the development of rust.3 A person who primes something.
4 Aeronaut. A small pump in an aircraft for pumping fuel to prime the en-
gine.
5 a Biochem. A molecule that serves as a starting material for a polymeri-
zation.
b Zoo/. &Physio/. A pheromone that acts initially on the endocrine system,
and is thus more general in effect than a releaser.
primer III
I First in order of time or occurrence; early; primitive.
2 First in rank or importance; principal, chief.
In almost half of the fourteen senses or subsenses presented here, we fmd
definitional techniques that would seem to be inadmissible if one assumes
that meanings have to be defined in terms of necessary and sufficient, gen-
eral and distinctive characteristics. To begin with, we find disjunctions in I 1
('A prayer-book or devotional manual for the laity'), in I 2b ('something
introducing or providing initial instruction in a particular subject, practice,
etc. '), in 11 2 (' A substance used as a preparatory coat on previously un-
painted wood, metal, canvas, etc., esp. to prevent the absorption of subse-
quent layers of paint or the development of rust'), in III I ('First in order of
time or occurrence'), and in III 2 ('First in rank or importance'). From a
traditional point of view, disjunctions are barred from definitions, because
they fail to capture the common aspects of the category to be defined.
In a similar way, open-ended enumerations should be avoided: they may
illustrate or partially demarcate a category, but they do not define it, if you
assume a rigid conception of definitions. In the examples, however, quite a
number of open-ended enumerations appear: in I 2b ('something introducing
or providing initial instruction in a particular subject, practice, etc. '), in 11 1
a CA cap, cylinder, etc., containing a compound which responds to friction,
electrical impulse, etc., and ignites the charge in a cartridge etc.'), in 11 2 CA
substance used as a preparatory coat on previously unpainted wood, metal,
canvas, etc. ').
Finally, we may note that the juxtaposition of near-synonyms is yet an-
other way of loosening up the definitions. In the example III 1, the near-
synonyms early and primitive do not have exactly the same meaning (what
is early is not necessarily primitive, and vice versa). At the same time, they
add something to the analytical definition; in particular, the near-synonym
primitive adds a nuance of lack of sophistication that is not explicit in the
definition 'First in order of time or occurrence'.
Lexicographical practice, in short, appears to be in accordance with the
lexicological observation that the distinction between meanings need not be
clear-cut. This fact has not escaped the lexicographers themselves, to be
sure: among others, see Ayto (1983), Stock (1983), Hanks (1994). In the
neighboring field of computational lexicography, similar voices may be
heard: Kilgarriff (1997).
3.3. The structured nature of polysemy
Let us consider the first seven senses of the adjective fresh in the Oxford
English Dictionary, 2nd edition. (In the overview below, the definitions are
360 Lexicography
sometimes rendered only partially. Some meaning nuances have been left
out.)
fresh
I New, recent
1. a. New, novel; not previously known, used, met with, introduced, etc. b.
In weaker sense: Additional, another, other, different, further.
2. Recent; newly made, recently arrived, received, or taken in.
3. Making one's first acquaintance with a position, society, etc.; raw,
inexperienced; unsophisticated, 'green'.
n. Having the signs of newness.
4. Of perishable articles of food, etc.: New, in contradistinction to being
artificially preserved; (of meat) not salted, pickled, or smoked; (of butter)
without salt; (of fruits, etc.) not dried or preserved in sugar or the like
5. Of water: Not salt or bitter; fit for drinking.
6. Untainted, pure; hence, possessed of active properties; invigorating,
refreshing. Said esp. of air
7. Retaining its original qualities; not deteriorated or changed by lapse of
time; not stale, musty, or vapid.
The article exhibits a linear ordering of the meanings, with a higher-
order, taxonomical structure of three levels. Even a cursory inspection of the
defInitions reveals that the hierarchical ordering does not make explicit all
the relations that exist among the different senses.
In the fIrst place, the senses 1-3 within group I are related by similarity,
with sense 1 probably as the prototypical center of the group. Roughly,
sense I can be paraphrased as 'new according to the perspective of a be-
holder'. Sense 2 is 'new as such, newly produced'. Sense 3 may receive the
paraphrase 'new in a specifIc context, new in a given position or function'.
The senses within group II are likewise related by similarity, but 7 seems to
be a more encompassing one than the others: if 7 is paraphrased as 'retain-
ing its originally optimal character', then both the 'pure and strong' reading
of 6 and the 'optimal for consumption, still in possession of all its nutritional
value' reading of 4 are specializations of7. Sense 5 'fIt for drinking', on the
other hand, belongs together with 5. In short, the linear order within group I
and within group II does not have an identical value, or at least, the semantic
relations within each group are more specifIc than can be expressed by a
mere linear ordering.
In the second place, the relationship between group I and group II is a
metonyrnical one: having the features of newness is a causal result of being
new, in whatever sense. However, such a metonyrnical relationship also
appears within group I. The nuance 'raw, inexperienced, unsophisticated'
that appears after the colon in definition 3 is as much a 'sign of newness' in
sense 3 as the meanings 4-7 are signs of newness in the sense defined by 2.
We see, in other words, that the same type of relationship is not always
treated in the same way. This also holds for the relationship of semantic
specialization that links 7 to 4, 5, and 6. Notice, in fact, that reading lb is a
semantic specialization of Ia. The things that are fresh in Ib are not just
novel from the point of view of the beholder, they are novel in comparison
with a set or series of similar things.
All in all then, the semantic structure of the item is a multidimensional
one. A further, more detailed analysis would undoubtedly reveal more di-
mensions, but at this point, it may be sufficient to take into account the three
dimensions that came to the fore in our cursory analysis: the relationship of
similarity between 1, 2, and 3; the relationship of specialization that exists
between 7 and 4,5,6 on the one hand and between la and lb on the other;
and the metonymical relationship between 2 and 7, and between 3 and 3'
(where 3' refers to the reading 'raw, inexperienced, unsophisticated').
The point, to be sure, is not that the linear order in the Oxford English
Dictionary should be condemned as an inadequate rendering of the underly-
ing semantic structure. The point is rather that any traditional form of linear
ordering cannot do full justice to the multidimensional nature of semantic
structures. In an earlier article (Geeraerts 1990a), I called this the lexico-
graphical linearization problem: the fact that lexicographers compiling tra-
ditional dictionaries have to project a multidimensional, clustered semantic
structure onto the linear order of the dictionary. In that article, I presented a
detailed analysis of the word vers (the Dutch counterpart of English fresh)
and its treatment in the Woordenboek der Nederlandsche Taal (the Dutch
counterpart of the Oxford English Dictionary), and analyzed the various
mechanisms (like hierarchical groupings, labels, and cross-references) that
lexicographers may employ to circumvent the problem.
The main point then, as now, was not a practical but a theoretical one: if
the linearization problem is indeed a recurrent problem for practical lexicog-
raphy, then a lexicographical metatheory had better start from a linguistic
theory that explicitly recognizes the underlying semantic multidimensional-
ity.
362 Lexicography
4. The wider picture
The discussion in the previous pages suggests that the conception that Cog-
nitive Semantics has of polysemy and semantic structure is consonant with
the actual practice of dictionaries. I deliberately use the word suggests, be-
cause the few examples taken into consideration here could hardly tell the
whole story. Even so, what Cognitive Semantics seems to offer to lexicogra-
phy is a conception of semantic structure that is perhaps in a number of
respects more realistic than what many other semantic theories (in particu-
lar, theories of a structuralist persuasion) can provide. This recognition does
not, however, exhaust the interaction between Cognitive Semantics and lexi-
cography. There are at least three further points that should be mentioned to
put the present contribution in a wider context.
First, the previous discussion was restricted to the way in which Cogni-
tive Semantics encompasses a theoretical perspective that so to speak vindi-
cates an existing definitional practice. However, Cognitive Semantics may
also suggest ways of dealing with the links between the senses of lexical
items that go beyond common practice. Swanepoel (1992, 1998) and Van
der Meer (2000), for instance, argue for devoting more explicit attention to
the motivational link between core senses and figurative subsenses. Such
motivational links could specifically involve conceptual metaphors in the
Lakovian sense (Van der Meer, Swanepoel), or even image schemata (Swan-
epoel). Interestingly, Van der Meer's suggestion is part of a critical ap-
praisal of the New Oxford Dictionary of English, which is perhaps the first
dictionary to refer explicitly to prototype theory as the basis of its organizing
principles (cp. Hanks 1994). Up to a point, then, Van der Meer's comments
can be read as the suggestion that an even greater influence of the Cognitive
Linnguistics approach could be lexicographically useful.
Second, there are aspects of lexicography that Cognitive Semantics has
touched upon, but that fall outside the scope of the present paper, focusing
as it does on problems of polysemy. In particular, frame theory has proved a
highly stimulating framework for the description of verbal meaning, both
theoretically and lexicographically: see Fillmore and Atkins (1992, 2000) for
the more theoretical side of the approach, and compare the website descrip-
tion of the Berkeley FrameNet project for the lexicographical applications.
And third, to complete the picture, it should also be mentioned that there
are important aspects of current lexicographical practice that Cognitive Se-
mantics has only marginally touched upon, in spite of the fact that the lexi-
cographical approach has proved extremely rewarding for lexical analysis at
large and for the study of polysemy in particular. Specifically, a colloca-
tional approach to polysemy, identifying different meanings through differ-
ences in collocational patterns, is a methodological focus for many current
lexicographical projects (see among others Moon 1998). However, although
the use of corpus materials is ardently advocated by a number of linguists
working in the tradition of Cognitive Semantics (Geeraerts, Grondelaers, and
Bakema 1994, Barlow and Kemmer 2000), collocational methods to get a
grip on polysemy are not (yet) among the standard equipment of cognitive
linguists.
In short, although Cognitive Semantics appears to offer an exciting per-
spective for the further development of lexicography and lexicographical
theory, the real interaction has clearly only started to emerge.
Section 6
Theory and method in lexical semantics
Chapter 15
Cognitive grammar and the history of lexical
semantics
Originally published in Brygida Rudzka-Ostyn (ed.), 1988, Topics in Cognitive
Linguistics 647-677. Amsterdam/Philadelphia: John Benjamins.
Innovative approaches may tend to overstate the novelty of their approach, and in
the early years of Cognitive Linguistics, sweeping statements could be found about
the revolutionary impact of the new framework (as they may perhaps still be in the
rhetoric of converts and novices). An awareness of the history of linguistics is a
welcome antidote against such exaggerations. In the article reprinted here, I argue
that Cognitive Semantics. rather than being a radical break with some two thou-
sands years of thinking about language, is in a number of important respects a
return to the prestructuralist, diachronically oriented fonn of semantics that
reigned supreme before the advent of structuralist semantics in the first half of the
twentieth century.
In the meantime, many other studies have discussed the position of Cognitive
Linguistics in the context of the history of linguistics: apart from the extensive
work done by Nerlich (1990, 1992), examples to be cited include Swiggers (1988),
Lutzeier (1992), lakel (1997), Taylor (1999), Elffers (1999), and Haser (2005). As
a companion piece to the present paper, I published an article entitled 'Cognitive
Semantics and the history of philosophical epistemology' (Geeraerts 1993c) in
which I argued that from a philosophical point of view as well, the exact implica-
tions of the cognitive framework can only be properly understood against the
background of the history of the discipline.
The present article is most profitably read in conjunction with the next chapter,
which adds a number of important points to the historical picture painted here.
L Introduction
De semasiologie of semantiek is dus dM deel der taalwetenschap, dat zich
ten doel stelt de levensgeschiedenissen van alle woorden stelselmatig te
verzamelen, te schiften en te klassificeeren, om daarna te trachten al de
diepere psychologische en sociale oorzaken van die levensgeschiedenissen
te achterhalen niet alleen, maar weer onderling te vergelijken, om zoo tot
de kennis van eenige weinige algemeen geldige semasiologische wetten te
368 Theory and method
komen. Daar nu de woorden alleen leven van ons menschelijk psychisch en
sociaal leven, zullen dit dus psychische en sociaal-psychische wetten moe-
ten zijn (Van Ginneken 1912: 6).
[Semasiology, or semantics. is that part of linguistics that tries to system-
atically collect the histories of all words, to select and classifY them, in or-
der not only to try to discover the deeper psychological and social causes of
those histories, but also to compare these among each other, so that an in-
sight into some few generally valid semasiological laws can be attained.
But because words can only exist and develop through our human psycho-
logical and social life, these will be psychological and socio-psychological
laws.]
The view of semantics expressed in this quotation has an immediate appeal
to proponents of the new school of Cognitive Grammar that is exemplified
by the studies in this volume. Indeed, studying the semantic history of words
to learn something about the 'laws' of the human mind can easily be appre-
ciated as a particular realization of the more general attempt to study lin-
guistic phenomena at large as an emanation and concretization of human
cognition and its working principles. Is Van Ginneken merely an incidental
precursor of Cognitive Grammar, or are his views typical of a broader tradi-
tion of semantic research that can be shown to have some remarkable affini-
ties with Cognitive Grammar?
In this paper, I will try to demonstrate the latter. I will try to point out
some revealing similarities between Cognitive Grammar and the historical-
philological tradition of semantic research to which Van Ginneken belongs
(albeit with the restriction that I will be concerned exclusively with lexical
semantics). Also, I will try to make clear that Cognitive Grammar may profit
from a greater awareness of the historical-philological tradition of semantics,
and I will place the observed similarities against the background of a brief
survey of the history oflexical semantics as a branch of linguistics.
2. Historical-philological semantics
The historical-philological approach can be exemplified by major works
such as Paul (1880), Darmesteter (1887), Breal (1897), Wundt (1900), Ny-
rop (1913), Wellander (1928), Sperber (1923), Carnoy (1927), and Stem
(1931). It can be seen as the first truly linguistic fornl of semantics, i.e. as a
form of semantic research that is embedded in linguistics, as an autonomous
empirical discipline, rather than being a part of speculative philosophy or of
History oflexical semantics 369
rhetoric (as is the case with most of the semantic studies carried out before
the nineteenth century). The historical-philological approach is characterized
by three prominent features. I In the following, these will be illustrated with
quotations from Breal (1897), not because Breal is the first or the most im-
portant exponent of historical semantics, but because his work clearly ex-
presses the major methodological features of the historical-philological ap-
proach
2
It should be mentioned that the three characteristics listed below
need not be simultaneously present in all of the works belonging to the his-
torical-philological era. Individual differences of emphasis notwithstanding,
they do, however, adequately characterize the basic methodological outlook
that is shared by most of the semantic studies in this period.
1. Already on the first page of Breal's Essai de semantique. the historical
orientation of semantics is indicated as an intuitively obvious matter of fact.
Talking about linguistics, he notes:
Si 1'on se borne aux changements des voyelles et des consonnes, on reduit
cette etude aux proportions d'une branche secondaire de I'acoustique et de
la physiologie; si I'on se contente d'enumerer les pertes subies par le meca-
nisme grammatical, on donne I'illusion d'un edifice qui tombe en ruines; si
1'on se retranche dans de vagues theories sur I'origine du langage, on ajou-
te, sans grand profit, un chapitre al'histoire des systemes. 11 y a la, iI me
semble. autre chose a faire... La linguistique parle aI'homme de lui-meme:
elle lui montre comment il a construit, comment il a perfectionne, a travers
des obstacles de toute nature et malgre d'inevitables lenteurs, malgre meme
des reculs momentanes, le plus necessaire instnullent de civilisation (1897:
1-3).
[If one restricts oneself to the study of the changes of vowels and conso-
nants, this discipline is reduced to a secondary branch of acoustics and
physiology; if one merely enumerates the losses suffered by the grammati-
cal mechanism, one creates the illusion of a building tumbling into ruins; if
one hides behind vague theories about the origin of languages, one adds,
without much profit, a chapter to the history of systems. There is, it seems
to me, something else to be done... Linguistics talks to man about himself:
it shows how he has constructed, how he has perfected, through difficulties
of all sorts and in spite of an inevitable inertia, in spite even of temporary
retreats, the most indispensable tool of civilization.]
It is even the case that an adequate understanding of words in their con-
temporary meaning requires a thorough knowledge of their semantic history:
'L'histoire peut seule nous donner aux mots le degre de precision dont nous
avons besoin pour les bien comprendre' (Only history can give to the words
the degree of precision that we reqUIre to understand them adequately)
(1897: 124).
This historical orientation is, of course, a general aspect of nineteenth-
century linguistics. As an autonomous empirical discipline, linguistics comes
into being as a form of historical research, and the birth of historical linguis-
tic semantics in the nineteenth century is merely one more aspect of the
overall diachronic outlook of the first phase in the development of modem
linguistics. It should be noted, however, that the birth of semantics within
that young linguistic science was not just a question of completeness, but
rather of necessity. The study of meaning was not simply taken up out of a
desire to study linguistic change in all of its aspects, but a thorough knowl-
edge of the mechanisms of semantic change appeared to be a prerequisite for
adequate historical investigations into the formal aspects of languages. In-
deed, one of the most important branches of historical linguistics was the
etymological reconstruction of the history of words and phrases. The meth-
odology of comparative reconstruction requires that the word forms from
different languages that are to be compared with each other, be semantically
related. But such a relationship is not always obvious. For instance,
throughout the older Germanic languages, there is a fairly systematic formal
resemblance between words for the concept 'beech', and words for notions
such as 'book' and 'letter'. (Compare, e.g., Old High German buohha
'beech' and buoh 'book', or Old Saxon bOka 'beech' and bOk 'book, writing
tablet'.) Now, in order to justify a reconstruction of these forms as being
related to the same Proto-Germanic root, their semantic relationship has to
be clarified. (In this particular case, an awareness of the frequent metonymi-
cal relationship between names for substances and the name of objects made
of those substances will have to be combined with archaeological evidence
showing that wooden tablets were used for writing purposes.) Considering a
number of lexical forms as cognate requires that their semantic relationship
can be plausibly established, and this in turn requires an overview of the
regular mechanisms of semantic change. As such, diachronic semantics was
not merely taken up as an end in itself, but also as an auxiliary discipline
with regard to etymological studies.
3
It should also be noted that the enormous success of historical semantics
in the nineteenth century and the first decades of the twentieth, does not only
show up in the large number of articles and monographs devoted to it, but
also in the major lexicographical projects undertaken during that period.
Voluminous historical and etymological dictionaries such as the Deutsches
Worterbuch, the Oxford English Dictionary, and the Woordenboek der
Nederlandsche Taal testify that the early interest in the semantic histories of
words has led to an hitherto unsurpassed amount of empirical semantic re-
search.
2. The psychological orientation of historical-philological semantics has
two aspects to it. In the first place, lexical meanings are considered to be
psychological entities, i.e., (a kind ot) thoughts or ideas: '(Le langage) objec-
tive la pensee' (Language makes thought objective.) (Breal 1897: 273). The
mental status of lexical meanings links up directly with the overall function
of thinking, i.e., with the function of cognition as a reflection and reconstruc-
tion of experience. As such, language expresses thought as a storeroom of
encyclopedic knowledge: 'Le langage est une traduction de la realite, une
transposition ou les objets figment deja generalises et classifies par le travail
de la pensee' (Language is a translation of reality, a transposition in which
particular objects only appear through the intermediary of the generalizing
and classificatory efforts of thought.) (1897: 275). Language, then, is not
autonomous; it is embedded in and linked with the total set of cognitive ca-
pacities that enable men to understand the world with ever more refined con-
ceptual tools.
In the second place, historical-philological semantics tries to explain
meaning changes as the result of psychological processes. The general
mechanisms of semantic change that can be derived from the classificatory
study of the history of words constitute patterns of thought of the human
mind. Breal caBs these mechanisms 'les lois intellectueBes du langage', but
he hastens to add that 'law' means something different here than in the natu-
ral sciences: a law of semantic change is not a strict rule without exceptions,
but it represents a tendency of the human cognitive apparatus to function in
a particular way. In a passage that opposes the restriction of linguistics to
the study of the formal aspects of language, he remarks:
Nous ne doutons pas que la linguistique, revenant de ses paradoxes et de
ses partis pris, deviendra plus juste pour le premier moteur des langues,
c'est-a-dire pour nous-memes, pour I'intelligence humaine. Cette myste-
rieuse transfonnation qui fait sortir le franc;ais du latin, comme le persan
du zend et I'anglais de I'anglo-saxon, et qui presente partout sur les faits
essentiels un ensemble frappant de rencontres et d'identites, n'est pas le
simple produit de la decadence des sons et de I'usure des flexions; sous ces
phenomenes ou tout nous parle de ruine, nous sentons I'action d'une pen-
see qui se degage de la fonne a laquelle elle est enchainee, qui travaille a la
modifier, et qui tire souvent avantage de ce qui semble d'abord perte et
destruction. Mens agitat molem (1897: 338-339).
[We do not doubt that linguistics, giving up its paradoxical prejudices, will
treat more fairly the primary forces in languages, i.e. ourselves, human in-
telligence. The mysterious transformation that makes French grow out of
Latin (just as Persian out of Zend, and English out of Anglo-Saxon), and
that everywhere shows a remarkable set of similarities and parallelisms
with regard to its essentials, is not simply the product of the decay of
sounds and the wearing off of endings. Behind these phenomena in which
everything seems to speak of destmction, we feel the active efforts of hu-
man thought liberating itself from the form in which it is constrained, try-
ing to modifY it, and very of ten turning to its advantage what at first sight
appears to be mere loss and decay. Mind moves matter.]
The moving force of the human mind also shows up in the fact that the
fundamental factor that brings the psychological mechanisms of semantic
change into action, consists of the communicative needs of the language
user. Languages change because people try to express their thoughts as ac-
curately and satisfactorily as possible.
Le but, en matiere de langage, c'est d'etre compris. L'enfant, pendant des
mois, exerce sa langue aproferer des voyelles, aarticuler des consonnes:
combien d'avortements, avant de parvenir aprononcer clairement une syl-
label Les innovations grammaticales sont de la meme sorte, avec cette dif-
ference que tout un peuple y collabore. Que de constmctions maladroites,
incorrectes, obscures, avant de trouver celle qui sera non pas l'expression
adequate (il n'en est point), mais du moins suffisante de la pensee (1897:
8).
[The goal, as far as language is concerned. is to be understood. During
months, the child exercises his tongue to produce vowels, to articulate con-
sonants: how many failures, before he can clearly pronounce a syllable. On
the grammatical level, innovations are of the same sort, with this difference
that an entire people is involved. How many clumsy, incorrect, obscure
constmctions, before the one is found that will be, not the perfect expres-
sion of thought (there is none), but at least a sufficient expression of it.]
Note, however, that the expressive, communicative intent that lies at the
basis of linguistic change need not always have the conscious, voluntary
characteristics attributed to it by Breal. Sperber (1923), e.g., stresses the
importance of subconscious emotional factors.
Also, the overall psychological orientation of historical-philological seman-
tics leaves room for authors such as Meillet (1906), who stresses the social
rather than the psychological background of semantic change. Likewise, the
earliest classifications of semantic changes, relying heavily on traditional
rhetorical classifications (as is still the case in, e.g., Darmesteter 1887), were
much less psychologically orientated than later ones.
4
3. Methodologically, historical-philological semantics is a hermeneutical
discipline. In the following quotation, Breal does not simply repeat the point
that semantics is a historical science, but he also has something to say about
the way in which that scientific project is put into practice:
Si I' on admet une difference entre les sciences historiques et les sciences
naturelles, si I'on consictere I'homme comme fournissant la matiere d'un
chapitre apart dans notre etude de I'univers, le langage, qui est l'oeuvre de
I'honune, ne pourra pas rester sur I'autre bord, et la iinguistique, par une
consequence necessaire, fera partie des sciences historiques (1897: 278).
[If one admits that there is a distinction between the historical and the
natural sciences, that is, if one considers man as being the subject matter of
a separate chapter of our study of the universe, language (which is the
product of man), cannot stay on the other side, and linguistics wil1 inevita-
bly be a branch of the historical sciences.]
Clearly, the natural sciences also study historical processes (as in geology or
the study of biological evolution), and that is why the difference between the
natural and the human sciences that is mentioned in the quotation has to be
sought on the methodological level rather than on the level of the object mat-
ter of both approaches. The distinction made by Breal probably refers to the
theories of the German philosopher Wilhelm Dilthey, whose views on the
relationship between the natural and the human sciences (Naturwissenschaft
versus Geisteswissenschaft) were widely popular near the end of the nine-
teenth century.
5
The methodological autonomy of the human sciences with
regard to the natural sciences resides in the fact that they try to understand,
by means of an empathic process of interpretation (Einfuhlung), the cultural
forms of expressions in which men have, throughout history, laid down their
experience of the world. The natural sciences, on the other hand, try to ex-
plain the characteristics of the material world by means of rigid laws. Next
to being historically and culturally orientated, the human sciences in the
Diltheyan sense are hermeneutical par excellence: they try to reconstruct the
original experience that has lain at the basis of particular forms of human
expression that have been transmitted from earlier times to the present day;
they look for the expressive intention behind historical forms of expression.
The connection between the Diltheyan conception of the human sciences
and the kind of linguistic semantics sketched above will be clear: through its
historical approach, through its cultural, experiential orientation (language is
an expression of a particular experience of the world - see the quotation
given above from Breal 1897: 275), and through the importance it attaches
to the expressive intentions of language users as the source of linguistic
change, historical-philological semantics fits nicely into the Diltheyan view
of the human sciences.
This is reflected on the methodological level. Because linguistic seman-
tics is a historical discipline, its primary material consists of texts from dead
languages or from previous stages in the development of a living language.
Its basic methodological act is therefore the interpretation of those texts,
only afterwards can changes between periods (and the mechanisms guiding
them) be recognized, classified, and explained. The primary methodological
step of the historical semantician is that of the historical lexicographer and
the philological scholar: to interpret historical texts against the background
of their original context by trying to recover the original communicative
intention of the author.
3. Cognitive Semantics
With Rosch's prototypical conception of categorial structure as one of its
leading notions (cp. Rosch 1977, Mervis and Rosch 1981), Cognitive Se-
mantics is exemplified by studies such as Fillmore (1977a, 1977b), Lakoff
(1982a, 1982b), Lakoff and Johnson (1980), Langacker (1983), Talmy
(1983, 1985b), Coleman and Kay (1981), Lindner (1981), Brugman (1981,
1984), Hawkins (1984), Sweetser (1984), Vandeloise (1985a), Dirven
(1985), Rudzka-Ostyn (1985), Geeraerts (1983a, 1985a). The characteris-
tics of the cognitive trend in lexical semantics can be summarized by con-
trasting it with a view of categorization that has, implicitly or explicitly,
been fashionable for some decades. This view is sometimes referred to as the
classical or 'Aristotelian' view (cp. Rosch 1975). As far as the history of
lexical semantics is concerned, this is a somewhat misleading terminology.
The purpose of this paper is precisely to show that the 'classical' approach
to lexical structure has not been dominant throughout the whole history of
lexical semantics as a branch of modem linguistics, but is rather typical of
the structuralist period between the older historical-philological era and the
birth of Cognitive Semantics. As far as Aristotle is concerned, the predicate
'Aristotelian' is justified by the Aristotelian (and scholastic) metaphysical
distinction between accidental and essential attributes, with the latter con-
stituting the true definition of a concept. One should not forget, however,
that in his poetical and rhetorical treatises, Aristotle also has a lot of atten-
tion for topics such as metaphor, analogy, and the use of concrete examples
or 'paradigms' in rhetoric, that come close to the interests of cognitive se-
manticists.
The following list of characteristics gives only a rough overview of Cog-
nitive Semantics. I trust that the other papers in this volume will give the
reader an exemplary view of the cognitive approach. I will not try to prove
or defend the theses listed below: I merely present them as a summary of
what seem to be the major tendencies in cognitive semantic research on the
lexical-semantic level, a summary that may serve as an aide-memoire to the
comparison of Cognitive Semantics with the historical-philological tradition
that will be the subject matter of section 4.
It should be mentioned that there is a lot of variation within Cognitive
Semantics: the views listed below need not all be held by the specific studies
that can be situated within the cognitive framework. In this respect, Cogni-
tive Semantics as a theoretical paradigm corresponds with its own views on
categorization. One of its basic tenets is that lexical concepts are not rigidly
defined by a set of essential attributes shared by all the instances of those
concepts. Likewise, a rigid definition of Cognitive Semantics, consisting of
an enumeration of attributes shared by all cognitive semantic studies is hard
to give. Notice that this is in keeping with the paradigmatic conception of
science defended by Kuhn, Lakatos and others, which considers scientific
research programmes to be clusters of interconnected theories. (The method-
ological consequences of this analogy between the theoretical views of Cog-
nitive Semantics and the metatheoretical views of the Kuhnian, paradigmatic
theory of science are explored in Geeraerts 1985a.) What, then, are the cen-
tral theses of Cognitive Semantics?
1. Lexical concepts have vague boundaries, in the sense that they contain
peripheral zones round clear conceptual centers (versus: lexical concepts are
discrete, well-defined entities). - The prototypical view of categorization
implies that categories may have marginal instantiations that do not conform
rigidly to the central cases. This extensional fact is paralleled by the inten-
sional phenomenon that it may not be possible to define a concept by means
of a set of attributes applicable to all the in stances in the extension of that
concept. Intensionally as well as extensionally, then, concepts may be char-
acterized by the existence of a chiaroscuro-zone, either with regard to the
category membership of instantiations of a category, or with regard to the
definitional status of attributes.
2. Lexical concepts are polysemous clusters of overlapping semantic nu-
ances (versus: the various senses of a lexical item can always be strictly
separated from each other). - The vagueness that characterizes conceptual
categories as a whole is mirrored by their internal structure. The blurred
edges of lexical concepts as a whole are paralleled by the possible absence of
clear boundaries among the senses that constitute the category as a whole.
The most common expression of this view is the family resemblance model
of categorial structure (Rosch and Mervis 1975).
3. The distinction between analyticity and syntheticity, and that between
essential and accidental attributes cannot be rigidly maintained (versus: there
is a strict distinction between analytical and synthetic statements, and be-
tween essential or accidental attributes). - This is a definitional consequence
of the previous points. There need not be a set of attributes that is applicable
to all the members of a category (and only those), and that defines the 'es-
sence' of that category (all other attributes being accidental). Since the Aris-
totelian distinction between essential and accidental attributes is paralleled
by the Kantian distinction between analytic and synthetic statements (at least
on the lexical level), the dichotomous distinction between the latter cannot be
maintained either, although there may be a gradient pattern of more or less
inherent attributes.
4. Lexical concepts may be disjunctively defined (versus: lexical concepts
are necessarily defined conjunctively). - This is again a consequence on the
level of definitions of points 1 and 2. In the classical view, a concept is de-
fined by a set of essential attributes all of which can be predicated of the
members of the category in question. Within a prototypically organized cat-
egory, however, categorial membership may be based on sufficient similarity
rather than identity; some of the allegedly necessary ('essential') attributes
of the central exemplars of the category may appear to be optional at the
periphery. As Labov (1973) has remarked, traditional lexicography very
often uses disjunctive rather than conjunctive definitions, a fact that clearly
indicates the practical importance of disjunctive definitions for semantic
description.
5. Attributes within (or exemplars of) a category may have different degrees
of salience (versus: all attributes within the definition, or exemplars in the
extension of a concept have an equal degree of salience). - Attributes not
shared by all the members of a category (or a majority thereof) are less im-
portant definitionally than attributes that do appear in all or most of the ex-
emplars of a category. These frequent attributes carry more weight within
the semantic structure of the category. Analogously, some members of a
category carry more weight as better exemplars of the category, in contrast
with the peripheral, less salient exemplars.
6. Lexical concepts function in a flexible and analogical manner (versus:
lexical concepts function in a rigid, algorithmical fashion). - The fact that
category membership can be defined by similarity rather than identity entails
that conceptual categories can be used in an extremely flexible way; making
the criteria for the use of a category less strict enhances the usage potential
of that category. One aspect of this extremely important characteristic of
lexical concepts is discussed in my other contribution to this volume (Geer-
aerts 1988d); there, I try to make clear that the flexibility inherent in proto-
typical categories as interpretative schemata with regard to experience, may
well be one of the major functional, psychological reasons for having such a
type of categorial structure. Another point to be mentioned is this: the flexi-
bility of prototypical concepts bridges the gap between synchrony and dia-
chrony (a moot point in structuralist conceptions of language) by incorporat-
ing the dynamism that is usually considered to be characteristic for a
diachronic conception of language among the synchronic features of lexical
concepts. This is, of course, a point that is of particular interest with regard
to the comparison of historical-philological semantics and Cognitive Seman-
tics.
7. Lexical concepts have to be studied as a proper part of human cognition
at large (versus: lexical concepts have to be studied as a part of an autono-
mous linguistic structure). - What is at stake here is the fundamentally
structuralist claim that the lexicon of a natural language contains a proper
semantic structure that is independent of the broader cognitive organization
of the human mind, and which it is the task of linguistic semantics to reveal;
this implies that semantics is an autonomous discipline.
6
By contrast, Cogni-
tive Semantics claims that lexical concepts can only be studied adequately
against the background of the human cognitive capacities at large, and in
particular , that there is no specifically linguistic, semantic organization of
knowledge, separate from conceptual memory in the broadest sense. If there
is no autonomous linguistic-semantic memory, lexical semantic research
cannot be done autonomously, but it has to be conducted in close coopera-
tion with other sciences studying the human mind. Of these, psychology is
obviously the most important one, but there are others such as Artificial
Intelligence, neurophysiology, and cultural anthropology.
8. There is no distinction between semantic and encyclopedic knowledge
(versus: the semantic definition of a lexical concept is to be distinguished
from the encyclopedic data that can be connected with the latter). - This
point is a specification of the former. The distinction between semantic and
encyclopedic knowledge (which parallels that between essential/analytic and
accidental/synthetic information) is based on the distinction between an
autonomous linguistic conceptual structure on the level where lexical items
are defined, and conceptual memory in the broadest sense. If the latter dis-
tinction is discarded, the former also falls.
9. Semantic studies cannot ignore the experiential and cultural background
of the language user (versus: semantic phenomena should be studied apart
from user- or culture-specific background data). - This is a further conse-
quence of the non-autonomous approach mentioned in point 7. If language is
one of the basic cognitive tools of man, it should not be studied autono-
mously, but it should be considered in the light of this cognitive function, i.e.
of interpreting, ordering, retaining, and expressing human experience. This
experience may, of course, be individually or culturally specific. This im-
plies, e.g., that attention will be given to cultural differences in the meta-
phorical patteming of experience (see Lakoff and Johnson 1980), and that
semantic universals will be seen as general strategies for coping with experi-
ence rather than as substantive universals.
4. Comparing the cognitive and the historical-philological tradition
In the light of the foregoing sections, it should not be too difficult to recog-
nize the similarities between the cognitive and the historical-philological
approaches, particularly when they are set off against the 'classical' ap-
proach.
I. A first point of primary importance is the psychological orientation of
both approaches. Both consider meanings to be primarily psychological
entities, that is to say, they identify 'meanings' and 'mental concepts', as
against an approach that strictly separates linguistic meanings and (encyclo-
pedic) psychological concepts. At the beginning of his Griechische Bedeu-
tungslehre, Max Hecht sums up the disciplinary position of historical-
philological semantics:
Insofem sie zugunsten der Lexikographie die Bedeutungen in zeitlicher
Folge ordnet und im Interesse der Etymologie die Gesetze der Bedeutungs-
anderung aufstellt, hat sie sprachwissenschaftlichen Weft. Soweit sie aber
diese Gesetze aus der Natur des Geistes herleitet und eine Geschichte der
Vorstellungen gibt - Bedeutungen sind Vorstellungen -, rallt sie auf das
Gebiet der empirischen Psychologie 15 (1888: 5).
[(Semantics) is linguistically valuable to the extent that it chronologically
classifies meanings in the interest of lexicography, and writes down the
laws of semantic change in the interest of etymology. To the extent, how-
ever, that it derives these laws from the nature of the mind and that it
writes a history of ideas - meanings are ideas -, it falls within the realm of
empirical psychology.]
In a vigorously polemical article that is properly to be regarded as the first
theoretical statement of structural semantics (with Trier's 1931 monograph
on semantic fields being the first descriptive achievement), Leo Weisgerber
(1927) blames Hecht for relegating semantics to the field of psychology.
Treating meanings as psychological 'representations' (Vorstellungen) , says
Weisgerber, entails that meanings cannot be studied autonomously, as a
proper part of linguistic structure.? This structuralist desire for an autono-
mous approach to linguistic semantics is then again discarded by Cognitive
Semantics in favor of a whole-heartedly psychological approach in which the
study of linguistic structure is required to be in accordance with what is
known about cognitive processing in general (see Langacker 1983: 1/6-7).
An immediate consequence of the rejection of an autonomous approach is
the fact that neither the historical-philological nor the cognitive tradition
imposes a strict distinction between encyclopedic and purely semantic data
in lexical analyses. One may say that, at the time of historical-philological
semantics, the distinction had not yet gained wide acceptance, whereas it is
no longer generally accepted in contemporary lexical semantics (see Haiman
1980a, Langacker 1983: 11/63).
Another consequence of the psychological approach is the fact that ex-
perimental psychological research will have an influence on linguistic con-
ceptions of categorial structure. Taking the cognitive nature of language
seriously entails taking psychological studies of meaning seriously. It is, in
fact, possible to identifY major psychological influences on the philological
and the cognitive research programmes. The latter's conception of categorial
structure has to a large extent been shaped by Eleanor Rosch' s experimental
work on prototype semantics (cp. supra), whereas the former has been pro-
foundly influenced by Wilhelm Wundt's apperceptionalist psychology
(Wundt 1900). It is true, though, that Wundt's views have not been gener-
ally accepted by all semanticists of the historical-philological persuasion, but
even when they are systematically criticized (e.g., Marty 1908), the tendency
is rather to substitute another psychological conception for Wundt's, than to
abandon the psychological outlook as a whole.
A final consequence of the psychological orientation of the research pro-
grammes that we are trying to compare is their common attention for the
experiential background of semantic phenomena. In historical-philological
semantics, this is embodied in what I have called its hermeneutic character:
it considers lexical meanings as an expression of a particular (individual or
cultural) view of the world. As for Cognitive Semantics, its experiential
nature has probably been brought to the fore most vigorously by George
Lakoff in his attack on the classical, 'objectivist' conception of categoriza-
tion (but see also, e.g., Langacker 1983: 1/6-7). In the following quotation,
the link between the experiential nature of semantics and its psychological,
non-autonomous orientation is clearly made apparent:
After a generation of research in which it was implicitly assumed that lan-
guage could be described on its own terms, it has become more interesting
to ask how much of the structure of language is determined by the fact that
people have bodies with perceptual mechanisms and memory and process-
ing capabilities and limitations, by the fact that people have to try to make
sense of the world using limited resources, and by the fact that people live
in social groups and have to try to communicate with each other (Lakoff
1982b: 155).
Comparing the importance of experiential factors in historical-
philological and in Cognitive Semantics, there may be a shift in emphasis
from meanings as the expression of subjective experience to conceptual
categories as interpretative schemata shaping experience, but the basic
point is the same: language is an integral part of the intellectual, emotional,
and social life of human beings, and this is reflected by the nature of linguis-
tic categories. As such, these cannot be considered apart from the human
endeavors to make sense (literally) of the world.
2. Next to the fundamental correspondence in the methodological orientation
of historical-philological and Cognitive Semantics, it should also be men-
tioned that there is a lot of similarity between both approaches with regard to
their conception of the nature of semantic categories, although this theoreti-
cal similarity is not as complete as the basic methodological one. One point
that is quite clear, is the importance of polysemy for categorial structure.
Both approaches tend to conceive of lexical concepts as polysemous clusters
of senses, and to study them accordingly. In particular, if the main descrip-
History oJlexical semantics 381
tive efforts of historical-philological semanticists are directed at the classifi-
cation of semantic changes, this is because the mechanisms of semantic
change are responsible for the polysemy of lexical items. (One might say:
mechanisms of change such as metaphor and metonymy are the structural
links between the various senses of an item.) It is no surprise, then, that the
cognitivists' interest in polysemy leads to a revival of the research into some
of the earlier semanticists' favorite topics. This is, of course, most clearly
true for the study of metaphor, but the phenomenon is not restricted to it.
(There is, e.g., the rediscovery of sound symbolism by Lakoff 1983; see
Kronasser 1952: 160-168 for an introduction to the older tradition's treat-
ment of it.)
Furthermore, the notions of vagueness and centrality that figure so
prominently in prototypical semantics are far from absent in historical-
philological theories. They may not receive the same emphasis as in present-
day theories, but they are clearly present in a number of works in the histori-
cal-philological vein. They can be found, e.g., in Paul's influential classifica-
tion of semantic changes (1880), and in Erdmann (1910), whose particularly
lucid formulation of the characteristics of lexical concepts justifies an exten-
sive quotation.
Denn welche Theorien iiber Wesen, Bedeutung und Entstehung der Begrif-
fe man auch vertreten mag: vom Standpunkt der Logik wird man immer
fordern miissen, dass sie eine unzweideutige, klare Grenze aufweisen, dass
sie einen bestimmten Inhalt und Umfang haben. Und Begriffe dieser Art
werden durch Worte nicht ohne weiteres bezeichnet. Worte sind vielmehr
im allgemeinen Zeichen fur ziemlich unbestimmte Komplexe von Vorstel-
lungen, die in mehr oder minder loser Weise Zllsammenhangen.... Die
Grenzen der Wortbedeutung sind verwaschen, verschwommen, zerflies-
send. Treffender aber noch wird meines Erachtens der Sachverhalt ge-
kennzeichnet, wenn man iiberhaupt nicht von Grenzlinien des Umfangs
redet, sondern ... von einem Grenzgebiet, das einen Kern einschliesst. Ver-
anschaulicht man sich gewohnlich den Umfang eines logisch vollkomme-
nen Begriffs durch eine scharfe Kreislinie, wie sie annahrend ein gutge-
spitzter Bleistift erzeugt, so kann man sich die Abgrenzung eines Wortum-
fanges durch einen mehr oder minder breiten, in sich zumcklaufenden
Streifen versinnlichen, wie ihn ein in Farbe getauchter Pinsel auf einer
Flache hinterlasst (1910: 4-5).
[But whatever theories about the essential nature, the meaning and the ori-
gin of concepts one may adhere to, from the point of view of logic one will
always have to require that they exhibit unambiguous, clear boundaries,
that their extent and content are clearly specified. But words do not simply
indicate concepts of that kind. Words in general are rather signs for fairly
unspecific complexes of psychological representations that belong together
more or less loosely. '" The boundaries of word meanings are vague, un-
clear, indeterminate. The situation is, I think, even more adequately de-
scribed if one simply does not talk about the borderline of the range (of a
word), but ... if one talks about a border area that includes a central area. If
one usually represents the range of a logically perfect concept by means of
a sharp borderline, such as is approximately drawn by a well-sharpened
pencil, one can represent the boundary of the content of a word by means
of a more or less wide, gradually narrowing band, such as is produced on a
flat surface by a brush that has been dipped into paint.]
The correspondence with the contemporary prototypical view is too obvious
to need much comment.
The basic correspondence established in the previous points does not im-
ply that there are no differences between the two research traditions com-
pared here. The basic difference is the shift from a diachronic to a syn-
chronic approach to semantics, mirrored by the fact that the emphasis shifts
from the mechanisms of semantic change to the internal prototypical struc-
ture of polysemous lexical concepts. This shift is, of course, part of the leg-
acy of structuralism, but one that is not accepted without fundamental al-
terations. In fact, the extent to which Cognitive Semantics differs from
structuralism with regard to the diachrony/synchrony-distinction is more
important than its primary adherence to that distinction. A strictly syn-
chronic approach as favored by structuralism tends to overlook the fact that
languages are in constant flux.
8
By incorporating the dynamic principles that
structuralism can only conceive of diachronically into the synchronically
flexible nature of lexical concepts, cognitive semantics can avoid the struc-
turalist paradox that the true nature of linguistic structure is considered to be
a steady state, whereas such a steady state can hardly bc found in linguistic
reality. The prototypical view of natural categories is so to say the syn-
chronic side of the diachronic coin; what appears as diachronic change from
one point of view, is merely the realization of synchronically available pos-
sibilities from another. In short, flexibility is the intersection of diachrony
and synchrony at the level of lexical structure, and that is a point of view
that may not have been completely absent in historical-philological seman-
tics, but that still deserved to be emphasized as it is now done in Cognitive
Semantics.
In this respect, then, Cognitive Semantics is a fundamental improvement
over the older tradition. But does this relationship also work the other way
around, i.e., can Cognitive Semantics profit from the historical-philological
tradition? The latter constitutes such a huge body of observations, classifica-
tions, and explanations (indeed, its gigantic proportions may well be the very
cause of its present-day neglect), that it is hard to imagine that it contains
nothing of interest for Cognitive Semantics. That a confrontation with his-
torical-philological semantics may be very fruitful is a point I would like to
illustrate on two different levels, first with regard to a very specific topic of
research, and second with regard to a rather more general, and more funda-
mental aspect of semantics.
a. Both Van Ginneken (1912) and Sweetser (1984) study the figurative
transfers of (among many others) expressions connected with auditory ex-
perience. Both do so from an essentially cross-linguistic, etymological point
of view, but their perspective is slightly different. Sweetser concentrates on
words for the act of hearing itself, and notices among other things that verbs
for the concept 'to hear' shift towards the notion 'to listen, to heed', and
further to 'to obey'. She also mentions that derived words coming from
'hear' -roots regularly mean 'tale, report, fame, glory, news'. Van Ginneken
considers attributes for the things we hear , and points out that auditory
characteristics may be transferred, among others, to visual sensations (loud
colours, aloud dress), to tactile sensations (Dutch dovenetel 'dead nettle'
from door, 'deaf), or to emotions (to quieten someone's fear).
A confrontation of Van Ginneken's and Sweetser's results yields a par-
ticular question for further research: do the patterns of figurative transfers in
both sets of expressions correspond with each other, and if they do not, what
cognitive reason can there be for conceptualizing one set differently from the
other? For instance, if 'loud' is an attribute for things that can be heard well,
and if 'to be heard' shifts towards 'to be famous', is it reasonable to expect
words for the notion 'loud' to be applicable to fame or famous things and
persons? Conversely, if attributes for noises are applicable to non-auditory
perceptual sensations, is it also possible to express the notion 'to experience
a non-auditory perceptual stimulus' with words that etymologically mean 'to
hear'? These questions need not, of course, be answered here; the important
thing is to see how they arise as specific research hypotheses from a con-
frontation of parallel studies in Cognitive Semantics and in the prestructural-
ist tradition of historical semantics.
b. One of the major questions in prototypical semantics concerns the internal
coherence of natural categories. What keeps a prototypically organized cate-
gory together, and more particularly, how far can the senses in a polysemous
cluster be apart from each other or from the prototypical centre before they
'drop out' of the category? The question can be refonnulated in tenns of the
relations that hold among those senses: what kind of relations are accept-
able? Are they gradable, and if so, to what extent can they hold? $0 far,
cognitive semantics has answered these questions almost exclusively in
tenns of similarity (obviously a gradable notion); think of a notion such as
'family resemblance', of the statement that prototypically organized con-
cepts are clusters of overlapping senses (the overlapping senses are similar
to the ex tent that they have a number of features in common), and of the
cognitivists' interest in metaphor.
Now, it would not be implausible to equate the relations that can hold
among the diverse senses within a natural category, with the mechanisms of
semantic change identified by diachronic semantics. In fact, if the flexibility
of prototypical categories is indeed a phenomenon of the same type as the
semantic changes of old (as I have argued above), the same type of mecha-
nisms should govern both. The mechanisms that have been studied by his-
torical-philological semantics as types of diachronic change are then incor-
porated into synchronic categories as the mechanisms that underlie their
flexibility and that specify the relations between the original senses and the
new ones that arise from the flexible use of the category. These mechanisms
detennine and restrict the range of possible extensions of a concept, given a
prototypical centre or an existing cluster of applications. As such, they are
interpretative principles, strategies that help one get from existing knowl-
edge to new, contextually functional applications of a concept.
But historical-philological semantics has revealed the existence of much
more mechanisms of change than mere similarity (though the latter is un-
doubtedly one of the major types). The associations principles leading from
existing senses to new ones also comprise relations of contiguity, which
fonn the basis of what is traditionally known as metonymy (including synec-
doche), and relations that might be called mimetic because they presuppose
the existence of another expression that already has the new sense of the
word we are investigating, and with which the investigated word is associ-
ated for one .reason or another. A typical example of this mimetic type of
change is semantic borrowing: the Gothic word daupjan is associated with
the Greek word baptizein through their common base meaning 'to cause to
go down, in particular, to submerge' , and next copies the secondary sense
'to baptize' of the Greek word. Other examples of mimetic change are ellip-
sis (in which case the association is syntactic), the semantic merging of
homonyms (where phonetic associations are involved), and parallel devel-
opments among antonyms (again a case of an initial semantic association).
(See Geeraerts 1986a: 38-41 for a more thorough treatment.)
I believe it will be one of the major tasks for lexical semantics in the next
few years to study the relevance of these traditional associationist mecha-
nisms of diachronic change for the cognitive theory of categorization. Do
they have the same cognitive status as associations through (literal or meta-
phorical) similarity, in providing principles that guide the interpretative ef-
forts of the hearer, enable the speaker to use lexical concepts flexibly, and
define the structural coherence of natural categories? Or do prototypical
categories only allow certain types of relations, so that semantic extensions
based on principles of contiguity or lexical mimesis immediately lead to the
creation of another category, distinct from the original one? Again, the ques-
tion need not (in fact, could not) be answered here. It suffices to see that the
confrontation with the historical-philological tradition opens up a major area
of research for Cognitive Semantics.
In short, then, the affinity between Cognitive Semantics and historical-
philological semantics also seems to entail that the former may profit from a
closer acquaintance with the latter. To deepen our insight into the affinities
between both approaches, we shall now have a closer look at the history of
lexical semantics as a whole.
5. A survey of the history of lexical semantics
Near the end of his book A History o.fSemantics, W.T. Gordon remarks that
semanticists have tended to take a tabula rasa approach to their discipline,
either out of a conscious rejection of previous theories, or because they were
simply unaware of the work of their predecessors (1982: 244). However,
Gordon's remark is more indicative of his own failure to identify major
schools and traditions in the history of semantics, than of the actual absence
of such major trends. At least in lexical semantics (and let us not forget that
up to the incorporation of semantic problems in transformational grammar,
semantics had been primarily lexical), it is possible to point out the existence
of a number of global methodological perspectives that shape the semantic
endeavors of a majority of researchers within a particular period. Table 1
gives an approximation of what such a methodological history of lexical
semantics might look like. The justification for the figure can only be dealt
with summarily here; a more detailed treatment of the major historical trends
in lexical semantics can be found in Geeraerts (1986a).
Next to the historical-philological and the cognitive approach, the figure
identifies two other basic methodological approaches to lexical semantics.
The logical approach to lexical semantics is clearly a minor tradition. When
it is not primarily a formalization of descriptive ideas developed within
structuralist semantics (cp. Dowty 1979), its descriptive contribution to
lexical semantics resides mainly in the analysis of lexical items corres-
ponding with logical operators (such as quantifiers and connectives), and in
the analysis of the syntactic meaning of word classes (the 'types' of Mon-
tague semantics). The proponents of Montagovian semantics (which is the
most sophisticated form of logical semantics) have themselves expressed
serious doubts with regard to the applicability of intensional logic to word
meaning beyond the level of the syntactic meaning of word classes (cp.
Thomason 1974: 48). Regardless of the results achieved by logical seman-
tics in other domains, its contributions to lexical semantics are as yet too
restricted to consider it a major trend in the history of lexical semantics.
Structuralist semantics is undoubtedly the most diverse of the four ap-
proaches mentioned in the figure. Disregarding the distributionalism of,
among others, Dubois (1964) and Apresjan (1966), which was never very
successful, there are three main approaches to be mentioned within struc-
turalist semantics. Two of the alternatives are clearly present in the debate
between the proponents of a decompositional and an axiomatic approach to
the formalization of lexical meaning, such as was held within the framework
of transformational granunar in the course of the 1970s (see, e.g., Katz and
Nagel 1974 versus Fodor, Fodor and Garrett 1975). For reasons to be ex-
plained in the following section, I consider transformational semantics to be
the methodological culmination of structuralist semantics. The axiomatic
and decompositional alternatives within transformational semantics, then,
are related to earlier forms of structural semantics. Componential analysis is
an outcome of the semantic field approach made popular by Trier (1931), in
the sense that the conceptual dimensions from which components are de-
rived, are primarily the distinctive oppositions within a semantic field. The
field origin of componential analysis is less clearly visible in Katz and Fo-
dor's introduction of the method into transformational granunar (1963), but
it is quite obvious in Lounsbury's (1956) and Goodenough's (1956) anthro-
pological studies, in which the componential method for meaning analysis
was applied for the first time, and from which Katz and Fodor seem to have
borrowed the idea. The field-theoretical origin of componential analysis is
equally apparent in the introduction of the method by European linguists
such as Pottier (1963) and Coseriu (1964). It is not quite clear. incidentally,
whether this European branch of componential analysis came into being
independently of Lounsbury, Goodenough, and/or Katz and Fodor.
Table 1. Dominant theories in the history oflexical semantics
MAIN THEORETICAL OBSERVATIONAL EMPIRICAL EXPLANATORY
PERIOD FRAMEWORK PRINCIPLE DOMAIN PRINCIPLE
1870- HISTORICAL- language as the statements about principles of effi-
1930 PHILOLOGICAL self-expression meaning change ciency and expres-
SEMANTICS of an individual sivity
or a community
1930- STRUCTURAL language as an statements about meaning decompo-
1975 SEMANTICS autonomous paradigmatic and sition, or meaning
structure syntagmatic se- postulates
mantic relations
1970- LOGICAL language as statements about algorithmical com-
SEMANTICS reference to the truth condi- positionality
possible worlds tions of proposi-
tions
1975- COGNITIVE language as a statements about natural categoriza-
SEMANTICS cognitive tool the flexibility, tion (prototypical-
internal structure, ity) and cognitive
experiential na- strategies
ture, encyclopedic
character of lexi-
cal concepts
On the other hand, the axiomatic (or 'meaning postulate') approach is
methodologically related to earlier purely structuralist work by Lyons (1963,
1968), in which relations such as hyponymy, synonymy, and antonymy
among unanalyzed lexical items are taken to be the core structure of the
lexicon (rather than analytic relations among the items in semantic fields).
It is extremely important to realize that the decompositional and the
axiomatic approach are compared here, not as formal methods of descrip-
tion, but as methodological perspectives on word meaning. In particular.
they methodologically define different observational bases for lexical seman-
tics: the componential approach focuses on the structural relations among
the items in semantic fields, whereas the axiomatic approach focuses on
structural relations of the type represented by hyponymy and synonymy. As
such, the point to be made here is simply that structural semantics may focus
on different types of semantic structure. Notice in this respect that compo-
nential analysis and meaning postulates as formal methods of description are
not exclusively tied up with the different observational approaches with
which they are primarily connected. In fact, componential analysis can be
used to explain the synonymy, antonymy etc. of lexical items (this is the
Katzian position), whereas meaning postulates may serve as a descriptive
method for indicating distinctive oppositions within semantic fields. Also, as
methods of formalization, the origins of the decompositional and the axio-
matic approach do not lie with structural semantics, but in phonology and
logic, respectively. In short, the methodologically important point is not the
formal method of description, but the difference in observational basis with
which it is usually connected: structuralists may focus on semantic field
structures, or on structural relations such as hyponymy, antonymy, and the
like.
Next to these two types of paradigmatic structures, they may direct their
attention at a third type of structure, viz. the syntagmatic semantic relations
that have become known in transformational grammar as selectional restric-
tions, but that had actually been discussed already by Porzig (1934).9
There will be much more to say about structural semantics in the fol-
lowing section, but first, our brief discussion of Table 1 has to be concluded
with a note on the chronological periods distinguished in it. These indicate
the heyday of each of the traditions mentioned, i.e., the period in which each
of the methodological perspectives was most intensively applied and in
which it constituted (one ot) the major approach(es) to lexical semantics of
that era. This implies that studies within a particular methodological para-
digm may well be found outside the periods mentioned in the figure, at a
time when that paradigm is no longer (one ot) the major one(s) in lexical
semantics. Thus, componential approaches to meaning description can still
be found today (see, e.g., Lipka 1985), but it would be misleading to say
that structuralist componential analysis is the leading, theoretically most
advanced conception of lexical semantics. Likewise, the fact that the classi-
fication takes its starting-point around 1870 does not mean that semantic
studies are completely absent in the previous period, but merely that it is
around that date that semantics be gins to receive frequent and continuous
attention , and that lively discussions on semantic matters begin to take
place. Still, I would like to stress that the classification is intended to cover
only the history of linguistic lexical semantics, i.e., semantic studies that can
be situated within linguistics as a discipline that combines empincal descrip-
tion with theoretical analysis. In this sense, the older (roughly, pre-nineteenth
century) traditions of semantics (speculative philosophy, practicallexicogra-
phy without theoretical interests, and the rhetorical classification of stylistic
tropes) are not envisaged here.
6. Main lines in the history of lexical semantics
The survey of Table 1 is based on a methodological analysis of the various
approaches. The three headings used to distinguish among them (empirical
domain I observational principle I explanatory principle) are well-known
notions in the philosophy of science. They correspond roughly with the ques-
tions 'What kind of phenomena does a theory deal with?', 'From what point
of view are those phenomena described?', and 'What are the basic principles
used in explaining the empirical observations?'. As with any classificatory
perspective, this methodological approach entails certain choices that might
be reversed from another point of view. For instance, Generative Semantics
and Interpretive Semantics are not mentioned separately because they are
distinguished mainly by the position they want a semantic component to take
within a transformational grammar, but otherwise share the basic methodo-
logical assumptions of transformationalist semantics. Even the latter is not
singled out for separate treatment, in spite of its mentalist views, which seem
to contradict the initial anti-psychological position of structuralism. Initially,
however, the mentalist self-definition of transformational semantics hardly
played a major methodological role, among other things because introspec-
tion had always been tacitly accepted as one of the methodological tech-
niques of semantics, even by structuralists. (Distributionalists are the possi-
ble exception, but then, their failure merely shows the difficulty of avoiding
the introspective method.)
Methodologically speaking, transformationalist semantics is a culmina-
tion of, rather than a departure from, the structuralist approach. The Katzian
definition of the methodology of lexical semantics (1972) is typically struc-
turalist in that it assumes the existence of an autonomous level or semantic
structure that is distinct from cognition at large and that underlies the lin-
guistic judgments of native speakers on such matters as implication, ho-
monymy, synonymy and the like. The basic methodological tenet of struc-
turalist semantics, the fundamental point by which it defines itself in contrast
with the older historical-philological tradition is the following (see Weisger-
ber 1927): meaning is not primarily a psychological, but a linguistic phe-
nomenon; hence, it is by studying linguistic phenomena that semantic struc-
tures can be discovered. The whole point is easily illustrated by the notion
'distinctive opposition'. Given a number of semantically related words
within a lexical field, a structural analysis will have to reveal the distinctive
oppositions among the items in question, and stop there, discarding other
attributes as linguistically irrelevant. In a restricted set of items such as ram:
ewe :: stallion : mare, the opposition MALE is distinctive. The fact that
ram and ewe are covered with relatively long hair from which clothes can be
made, whereas stallion and mare are not, is then considered to be linguisti-
cally irrelevant: it is part of our encyclopedic knowledge of the world, but it
is not a part of the structure of language. Structural semantics originates by
making precisely that distinction, viz. between semantic structure (a linguis-
tic phenomenon involving necessary and sufficient conditions for explaining
lexical and semantic relations) and cognition (a psychological phenomenon
involving encyclopedic knowledge). By adhering to the idea that there is such
an autonomously linguistic level of semantic structure, transformational
semantics is basically structuralist.
Next, what makes it a culmination of the structuralist approach is the
range of linguistic facts it indicates as the observational basis of lexical se-
mantics. In fact, it brings together the three major kinds of facts to which
structuralist semanticists had addressed themselves: the semantic field rela-
tions with which structuralist semantics originated and which dominated
structuralist semantics up to the beginning of the 1960s, the lexical relations
highlighted by Lyons (1963), and the syntagmatic aspects of semantic struc-
ture which had once been indicated by Porzig (1934), but which had more or
less
10
been forgotten until they were taken up simultaneously in the com-
ponential theories of Katz and Fodor (1963) and Pottier (1963). It is of
course to these syntagmatic aspects of semantic structure, embodied in the
notion of selection restrictions, that most of the transformationalist attention
was directed, at least in the sense that it was precisely the role of semantics
with regard to syntax that engendered the debate between Generative and
Interpretive Semantics. However, the methodologically basic point is that the
Katzian definition of the methodology of lexical semantics (1972) incorpo-
rates into the observational basis of semantics all three kinds of structural
relations that had been studied separately by pretransformationalist struc-
turalists. It is the linguistic judgments of native speakers with regard to these
relations that reveal the autonomous semantic structure of language; con-
versely, a fonnal description of that autonomous level of semantic structure
explains the behavior of the native speakers.
It is true, though, that Generative Semantics gradually moved towards a
position in which the autonomy of linguistics appeared to be threatened. The
growing importance given to semantic factors within the transfonnational
framework runs parallel with a move away from the original ideal of linguis-
tic autonomy (an ideal that is no less paramount in the Chomskyan concep-
tion of grammar than in the structuralist conception of lexical semantics).
One need only recall, e.g., Lakoff's incorporation of pragmatic considera-
tions in the fonn of transderivational constraints (1970) to appreciate the
extent to which the Generative Semantics movement slid away from the au-
tonomist ideal. To the extent that Generative Semantics threatened to violate
that basic ideal, it is right to say, following Newmeyer (1980), that Genera-
tive Semantics destroyed itself, but only with the proviso that it destroyed
itself as a branch of transformational grammar. It is remarkable, in fact,
that some of the most prominent figures of Cognitive Grammar (in particu-
lar, Lakoff and Langacker) were fonnerly much closer to the Generativist
than to the Interpretivist approach. As an approach within the transfonna-
tional paradigm, Generative Semantics disappeared because it became in-
compatible with some of the basic assumptions of transfonnational gram-
mar, but the primacy it gave to meaning now reappears within a new school
of thought that has abandoned the autonomist creed.
We can see, then, that the transition from transfonnational to Cognitive
Semantics does not occur at random, but is detennined by substantial points
of debate. Let us therefore have a closer look at the developmental lines that
interconnect the traditions identified in Table 1. The transition from the first
to the second stage has already been described: in trying to establish an
autonomously linguistic approach to the study of meaning, structuralist se-
mantics rejected the historical-philological 'confusion' of encyclopedic cog-
nition and semantic structure. And of course, the structuralist primacy of
synchronic analysis also favored abandoning the older, primarily diachronic
approach.
To see how structural semantics gave way to the logical paradigm, it is
once more the incorporation of lexical semantics into transfonnational
grammar that has to be taken into account. In line with the general tendency
of generative grammar, this incorporation led to an increased interest in the
fonnalization of semantic analyses. Justified criticism with regard to Katz's
fonnalizations (see e.g. Bierwisch 1969) induced people like McCawley
(1971) and Lakoff (1972) to take over the more sophisticated fonnalisms of
predicate logic. When it was recognized that logical and semilogical formal-
izations alike have to be referentially interpreted (see e.g. Lewis' 1972 well-
known sarcasm with regard to the uninterpreted 'Markerese' of earlier trans-
formationalist semanticists), the gradually spreading interest in logic eventu-
ally led to the adoption of the intensional-logical format of Montague gram-
mar.
The transition from the structuralist to the cognitive paradigm can be at-
tributed to two factors, one of which has already been mentioned: the Gen-
erative Semanticists' tendency to broaden the empirical scope of generative
grammar contained a movement away from the autonomy aspired at by
Chomskyan linguistics. Next to this, Cognitive Grammar can be considered
as taking seriously the methodological consequences of the mentalism pro-
claimed so vigorously by transformational semantics. Although (as we have
seen) the basic methodological position of transformational semantics is in
line with the autonomist views of structuralism, its mentalist pretensions did
become methodologically relevant at a particular point in its development. In
fact, the mentalist strand in transformational semantics became important in
the debate between the decompositional and the axiomatic approach to
meaning representation when the choice between both approaches came to be
seen as relying on the psychological reality of a set of primitive, non-
decomposable semantic features (the basic vocabulary of componential rep-
resentations). As might be expected, the discussion about the psychological
reality of the set in question moved away from the transformational frame-
work to a more purely psychological approach. That is to say, through the
appeal to experimental psychological evidence, the discussion on meaning
representation was taken over by psychologists pure and simple. (See, e.g.,
the debate between Glass and Holyoak 1975 and Smith, Shoben and Rips
1974.) But psychologists obviously do not bother about the ideal of linguis-
tic autonomy: in their survey of psycholexicology, Smith and Medin (1981)
explicitly suggest that a psychologically adequate representation of linguisti-
cally encoded concepts may have to rely on a complete representation of all
the cognitive capacities of man. In this sense, taking the psychological impli-
cations of the transformationalist mentalism methodologically seriously
carries one away from the autonomist approach of transformational gram-
mar. This is, of course, precisely the position taken by Cognitive Semantics.
If semantics (as grammar in general, according to Chomsky) is indeed a
psychological discipline, linguistic meaning can no longer be studied in iso-
lation from the other cognitive capacities of man, but it will have to be con-
sidered in the light of its basic psychological function, e.g. as one of the
basic cognitive tools of the human mind.
In this sense, then, transformational semantics represents the grandeur et
decadence of structural semantics, On the one hand, it is the methodological
culmination of the latter, but on the other, its predilection for formalization
and its mentalist pretensions embody the springs of its own dissolution. In
particular, by giving up the earlier anti-psychological views of structuralism
in favor of a mentalist self-conception, transformational semantics under-
mines its own autonomistically structuralist methodology. The developmen-
tal lines set out here are summarized in Figure 1; in bringing out the pivotal
position of transformational semantics, the figure clearly refines the classifi-
cation of Table 1, in which transformational semantics is incorporated into
the structural approach without further nuances,
On a more general level, we can now see that the history of lexical se-
mantics exhibits an oscillation between an autonomous and an experiential,
psychological approach. Structuralist semantics reacted against the psycho-
logical perspective of historical-philological semantics, but nowadays, the
pendulum is swinging back (partly out of an attempt to take the mentalist
claims of neostructuralist transformational semantics methodologically seri-
ously). It should furthermore be noted that, with regard to this basic dichot-
omy, logical semantics ranges with structural semantics for two intercon-
nected reasons. First, logical semantics is definitely not a psychological
approach. Its subject matter (the truth-functional relationship between lin-
guistic expressions and the world) is defined regardless of the cognitive sub-
ject: the truth conditions of a proposition exist regardless of the person(s)
entertaining that proposition, and can be described accordingly. Second, by
abstracting away from subjective, psychological factors, structuralism and
logical semantics alike tend to be objectivist approaches. In both cases,
meaning is primarily defined as an objective entity that is independent of the
individual mind: the supra-individual structure of language, or possible-
world truth conditions. As opposed to this. historical-philological semantics
and Cognitive Semantics take into accoullt the subjective, individual Sitz im
Leben of language. The opposition between 'objectivist' and 'subjectivist'
approaches leads us back to the Diltheyan distinction between, respectively,
'natural' and 'human' science (see section 2). Against the background of the
history of lexical semantics, the fundamental methodological affinity be-
tween cognitive and historical-philological semantics seems to reside primar-
ily in the fact that, through their experiential, psychological orientation, both
are subjective, 'human' sciences in the Diltheyan sense, in contrast with the
objectivist tendencies of structural and logical semantics, which seem to take
HISTORICAL-
PHILOLOGICAL
SEMANTICS
i
meaning should be studied
from a synchronic, non-
psychological, autonomous
point of view
~
STRUCTURAL SEMANTICS
paradigmatic and syntagmatic relations
reveal the autonomous semantic structure
of the language
TRANSFORMATIONAL SEMANTICS
-
mentalism fonnalization
-
a psychologically adequate
study of meaning requires a
non-autonomistic approach
1
COGNITIVE
SEMANTICS
a fonnally adequate representa-
tion of meaning requires a
model-theoretic interpretation
1
LOGICAL
SEMANTICS
Figure I. Stages and transitions in the history of lexical semantics
their methodological mode Is primarily from the 'natural' sciences. This
paper is obviously not the place to decide the issue between both basic ap-
proaches; the important thing is merely to see how such a major methodo-
logical dichotomy underlies the similarities between prestructuralist and
Cognitive Semantics that we have described in section 4.
7. Conclusion
In the foregoing pages, the following theses have been developed. Prima,
Lakoff's statement that 'The classical theory of concepts and categories has
been accepted in the West for two thousand years' (l982a: 99) overstates
the novelty of the cognitive approach. Within the history of lexical semantics
as a branch of linguistics, it is rather a partial return to the methodological
position of the prestructuralist, historical-philological tradition of semantic
research. Secunda, the major methodological alternative in the history of
lexical semantics is a Diltheyan dichotomy between semantics as an objec-
tivist, autonomist, and as an experiential, psychological approach. Tertia,
the historical-philological tradition contains a huge amount of empirical
observations, classificatory mechanisms, and explanatory hypotheses that
may be extremely valuable for the further development of Cognitive Seman-
tics.
It is precisely because of the large quantity of studies in the historical-
philological paradigm, that this article has had to be restricted to some main
lines. Nevertheless, I hope that it will have become clear that a closer look at
prestructuralist semantics will be worthwhile. This conclusion should not,
however, obscure the fact that Cognitive Semantics also takes some funda-
mental steps forward in comparison with the older tradition. For instance,
Gestalt psychologists have long been aware of the figure/ground alignment
in perception, but its importance for lexical-semantic analysis has come to
be appreciated only recently through the work of Langacker and Talmy.
Similarly, the prototypical conception of lexical structure opens up hitherto
scarcely explored possibilities for bridging the gap between the synchronic
and the diachronic approach to linguistic analysis. Although Cognitive Se-
mantics shares its fundamental methodological position with historical-
philological semantics, it is much more than a mere return to prestructuralist
semantics.
Notes
1. It may be useful to point out that I use the tenn 'philological' mainly in a
methodological sense; I intend it to be the indication of a methodological ap-
proach to the study of semantic phenomena, and not just of a topic or field of
research. In this sense, the studies grouped under the label 'philological'
need not be concerned with the study of a particular text or a particular liter-
ary tradition; rather, they are by and large supposed to share some important
methodological characteristics with philological studies in the narrow (lit-
erary) sense. In particular, the most typical representatives of this line of se-
mantic research are concerned with the historical interpretation oflinguistic
material as the expression of thought, philology in the narrow sense being
concerned mainly with the historical interpretation of literary works as the
expression of an individual author's views. (Needless to say, this fonnulation
refers to the characteristics of literary philology in its prestructuralist stage;
in our own century, literary hermeneutics has changed just as much as lin-
guistic semantics. The changes are at least partly similar: by moving away
from a roman tic insistence on the individuality of the author to a modern-
istic interest in the literary work as an autonomous 'textual object', the de-
velopment of literary scholarship parallels the linguistic transition from the
historical towards the structural study of meaning.)
2. Neither does my treatment of BnSal imply that his ideas are completely novel
within the history of linguistics at large. As was pointed out to me by Pierre
Swiggers (to whom I am gratefiJl for a number of interesting remarks with
regard to an earlier draft of this paper), it may be revealing to link Breal's
views with eighteenth-century thought. In view of the chronological limits of
this paper, this is a line which will not be pursued here.
3. Different approaches within the etymological study of words might be used to
identify particular schools, trends, and developments within the historical
tradition of lexical research that have not been mentioned separately in this
article. Thus, the W6rter und Sachen-movement, and the dialect-
geographical method introduced by Gillieron deserve explicit treatment in a
complete history of lexical semantics.
4. Still, readers who may have doubts as to the pervasiveness of the psychologi-
cal outlook in diachronic semantics will find Weisgerber's attack on histori-
cal semantics in general and psychologism in particular (1927) extremely in-
structive.
5. At the same time, Breal's insistence on the fact that linguistics is a human
instead of a natural science, is a reaction against the biological naturalism of
Schleicher, according to which language is an organism evolving independ-
ently of man.
6. This formulation explicitly uses the phrase fundamentally structuralist claim
instead of fundamental structuralist claim to account for the fact that the
Bloomfieldian, behaviorist branch of structuralism does not make the claim,
but instead discards the semantic description of linguistic forms from the
domain of linguistics. The point is this: to the extent that structuralists accept
the feasibility of a semantic description of language, they typically posit the
existence of an autonomously linguistic, non-encyclopedic lexical-semantic
structure, but this feasibility is not accepted by all forms of structuralism.
References to structuralist semantics in this paper obviously exclude those
structuralists that do not accept the possibility of a linguistic semantics.
(Moreover, it should be kept in mind that not all semantic studies by linguists
that take an overall structuralist approach need be typically structuralist on
the methodological level: see note la.)
7. There are two additional remarks to be made with regard to Weisgerber's
comments on Hecht. In the first place, nineteenth-century linguists some-
times make a distinction between BegrUf(concept in the most neutral sense)
and Vorstellung (concept as a psychological entity, mental representation).
As such, Hechfs use of the term Vorstellung is itself an indication of his psy-
chological outlook. In the second place. Weisgerber's major objection against
the older tradition of semantics is the following: identifying meaning and
cognition entails that meanings exist and develop independently of language;
meanings (thoughts) have an independent psychological life, and words are
simply attached to them as labels. It may be good to mention that this objec-
tion is ill-founded. Identifying semantic and encyclopedic knowledge need
not imply that language may not help to shape and support thought, nor does
it entail that cognition develops autonomously, without the help of the sym-
bolical tools provided by language. On the contrary, it is precisely the accep-
tance of an intimate connection between language and cognition that makes
it possible to envisage a relationship of reciprocal influence between them.
(For an elaboration of this point, see Geeraerts 1985a, sections 1.4. and 1.6.)
8. To avoid misunderstandings: the 'synchronic' character of the structural
method does not exclude diachronic studies, but conceives of change as a
succession of distinct synchronic stages. As such, the investigation into syn-
chronic structures is taken to be methodologically basic with regard to the
study of the transition between synchronic stages. However, such stages are
practically very hard to identify, since linguistic systems change neither en
bloc nor abruptly.
9. The structuralist views of Reichling (1935), which come close to the proto-
typical view (see Geeraerts 1983d), will not be discussed as they do not be-
long to 'mainstream' structuralism.
10. 'More or less': the Firthians might be mentioned (with some provisos) as an
exception.
Chapter 16
The theoretical and descriptive development of
lexical semantics
Originally published in Leila Behrens and Dietmar Zaefferer (eds.). 2002, The
Lexicon in Focus. Competition and Convergence in Current Lexicology 23-42.
Frankfurt: Peter Lang Verlag.
The picture of the history of lexical semantics painted in Chapter 15 is restricted
in two ways. First, it could only take into account the very beginnings of Cogni-
tive Semantics, focusing in particular on prototype theory. Second, it concentrated
on the relationship between Cognitive Semantics and prestmcturalist semantics.
Although it did give stmcturalist semantics its rightful place in the chronology, it
paid relatively little attention to the offshoots of the stmcturalist and generativist
tradition that exist alongside Cognitive Semantics.
Almost fifteen years after the initial article, I returned to the topic in a brief at-
tempt to remedy these two points. In the present chapter, which was originally
published in a volume featuring examples of the major traditions of lexicological
research, a specific position is reserved for the 'neostmcturalist' and the 'neogen-
erativist' approaches that continue the lines of the tradition that dominated before
Cognitive Semantics entered the scene. Further, the history of lexical semantics is
not just described as a succession of theoretical approaches, but also as a gradual
widening of the descriptive domain of semantics in which all major traditions play
an important role. It is shown that there is an underlying pattern to this develop-
ment of empirical broadening, going from semasiology to onomasiology, and from
purely stmcture-oriented to usage-oriented phenomena.
The paper rounds off with a consideration of the possibilities for future conver-
gence between the different theoretical approaches: an awareness of history is not
just important for recovering past achievements that might otherwise be lost, but
also for seeing more clearly what the future might bring.
1. Introduction
Existing treatments of the history of linguistic semantics, such as Nerlich
(1992) and Gordon (1982), concentrate on the analysis ofthe views of indi-
vidual authors or separate periods rather than on the general lines of devel-
opment that characterize the development of semantics. Counterbalancing
Theoretical and descriptive development 399
such more or less atomistic (but clearly indispensable) analyses of individual
views and separate stages in the development of the discipline, I earlier pub-
lished a number of articles (specifically, 1988a and 1988b) that tried to pre-
sent a rational reconstruction of the overall progression of linguistic lexical
semantics. In these articles, I concentrated particularly on the way in which
the development of lexical semantics was pushed forward by competing
theoretical conceptions of word meaning.
In the present paper, I will have a look at the history of lexical semantics
from a slightly different angle: I will reconsider the classification into differ-
ent theoretical approaches, taking into account the developments that oc-
curred in the course of the 1990s, and I will then consider the way in which
these various theoretical traditions concentrate on different empirical do-
mains. At the same time, I would like to show that these shifts in descriptive
emphasis fit together in an encompassing conception of the empirical domain
of lexical semantics: it can be shown, I suggest, that the various theoretical
approaches that may be distinguished in the history of the discipline gradu-
ally occupy and fill out the potential range of investigation that could ideally
be associated with lexical semantics. A similar view of the empirical devel-
opment of lexical semantics was presented in Geeraerts (l999b). In com-
parison, the picture in the present article is a more comprehensive and more
synthetic one. As I will inevitably have to deal with quite a number of differ-
ent approaches, I will, for the sake of succinctness, assume a basic familiar-
ity with most of them. The various approaches will be briefly presented, but
it is obviously beyond the scope of the present contribution to do so in much
conceptual or bibliographical detail.
2. Theoretical currents in the history of lexical semantics
For starters, I would like to suggest that four major theoretical currents
shape the development of lexical semantics: prestructuralist semantics,
structuralist and neostructuralist semantics, generativist and neogenerativist
semantics, and Cognitive Semantics. (These currents, by the way, involve
linguistic semantics only: psycholinguistic or philosophical approaches to
semantics are not included in the overview.)
2.1. Prestructuralist semantics
Prestructuralist diachronic semantics dominated the scene from roughly
1870 to 1930. This does not mean, to be sure, that there is no form oflexical
semantics before 1870. It is only by the middle of the nineteenth century,
however, that research into word meaning establishes itself as a distinct sub-
discipline of the new science of linguistics. At the level of individual re-
searchers, prestructuralist diachronic semantics is represented by Breal and
many others, like Paul, Darmesteter, Nyrop, and (at the end of the relevant
period) Carnoy and Stem. A systematic treatment of the main figures and
currents may be found in Nerlich (1992).
At the methodological level, the type of semantic research meant here
may be characterized by the following three features. First, in line with the
overall nature of nineteenth-century linguistics, the orientation is a dia-
chronic one: what semantics is interested in, is change of meaning. Second,
change of meaning is narrowed down to change of word meaning: the orien-
tation is predominantly semasiological rather than onomasiological (or
grammatical, for that matter). Third, the conception of meaning is predomi-
nantly psychological, in a double sense. Lexical meanings are considered to
be psychological entities, that is to say, (a kind of) thoughts or ideas. Fur-
ther, meaning changes (the main domain of research for prestructuralist
semantics) are explained as resulting from psychological processes. The
general mechanisms that are supposed to underlie semantic changes and
whose presence can be established through the classificatory study of the
history of words, correspond with patterns of thought of the human mind. A
concept like metonymy is not just a linguistic concept, it is also a cognitive
capacity of the human mind.
2.2. Structuralist and neostructuralist semantics
The origins of structuralist semantics are customarily attributed to Trier
(1931), but while Trier's monograph may indeed be the first major descrip-
tive work in structuralist semantics, the first theoretical and methodological
expose of the new approach (which, needless to say, took its inspiration from
De Saussure) is to be found in Weisgerber (1927). Weisgerber criticizes
prestructuralist historical semantics precisely on the three characteristic
points mentioned above. First, the study of meaning should not be atomistic
but should be concerned with semantic structures. Second, it should be syn-
chronic instead of diachronic, and third, the study of linguistic meaning
should proceed in an autonomously linguistic way. Because the meaning of a
linguistic sign is detennined by its position in the linguistic structures of
which it is a part, linguistic semantics should deal with those structures di-
rectly, regardless of the way in which they may be present in the individual's
mind. Because the subject matter of semantics consists of autonomous lin-
guistic phenomena, the methodology of linguistic semantics should be
autonomous, too.
The realization of this attempt to develop a synchronic, non-psycho-
logical, structural theory of semantics depended on the way in which the
notion of semantic structure was conceived. In actual practice, there are
mainly three distinct definitions of semantic structure that have been em-
ployed by structuralist semanticians. More particularly, three distinct kinds
of structural relations among lexical items have been singled out as the
proper methodological basis of lexical semantics. First, there is the relation-
ship of semantic similarity that lies at the basis of semantic field analysis,
inaugurated by Trier and ultimately leading to componential analysis in the
work of anthropological linguists such as Goodenough (1956) and Louns-
bury (1956), and, apparently independently, in the work of European struc-
turalists such as Pottier (1964). Second, there are unanalyzed lexical rela-
tions such as synonymy, antonymy, and hyponymy; these were for the first
time systematically selected as the methodological basis of structural seman-
tics by Lyons (1963). Third, syntagmatic lexical relations were identified by
Porzig (1934) under the name of 'wesenhafte Bedeutungsbeziehungen'; they
later reappeared as selectional restrictions in the neostructuralist kind of
semantics that was incorporated into generative grammar by Katz and Fodor
(1963).
The contemporary descendants of structuralist semantics (which may be
tenned 'neostructuralist' to distinguish them from the original approaches)
link up with these three distinct interpretations of the notion of lexical struc-
ture. First, lexical field theory in its original fonn is most explicitly further
developed by Coseriu and his followers (Coseriu and Geckeler 1981). The
approach does not, however, seem to generate a lot of new work anymore,
and its own advocates admit to a certain stagnation (Geckeler 1993).
Second, the relational approach has been further developed by Cruse
(1986); in his most recent work (2000b), however, the structuralist origins
of his approach are attenuated by the impact of Cognitive Semantics. A
consequent elaboration of the relational perspective is to be found in the
work of Mel9uk and his collaborators (1988, 1996), whose purpose is to
identify a universally valid set of relations that allows for the description of
all semantic associations between lexical items in the vocabulary of any
given language. The approach is not a purely theoretical one: it involves the
preparation of a number of formalized dictionaries (Melyuk 1984). In addi-
tion to Melyuk's work, a major descriptive achievement based on a relational
conception of lexical structure is the WordNet project (Fellbaum 1998).
WordNet is an online electronic lexical database organized on relational
principles, which now comprises nearly 100,000 concepts.
Third, following the lead provided by Firth (1957), the syntagmatic con-
ception of lexical relations developed into a highly productive tradition of
collocational analysis (where 'collocations' are, roughly defined, sets of
regularly co-occurring words). Many different specific methods of colloca-
tional analysis exist, and there is no single name that can be mentioned as the
leading figure in this form of semantics. In the present context, it may be
sufficient to mention the incorporation of collocational analysis in the
framework developed by Halliday (Halliday and Hasan 1976), together with
the highly successful incorporation of the approach into lexicography
(among many others, see Sinclair 1987, Moon 1998). Because a colloca-
tional analysis naturally involves a description of the facts of lexical co-
occurrence in actual language use, this approach links up directly with cor-
pus linguistics (which, as we shall see, is an important point for the overall
line of the present article).
2.3. Generativist and neogenerativist semantics
All through the second half of the 1960s and the major part of the 1970s, the
neostructuralist, generativist model of lexical-semantic description intro-
duced by Katz and Fodor (1963, later developed by Katz, specifically in
Katz 1972) formed a reference point for studies in lexical semantics. The
appeal of Katzian semantics was at least partly due to its incorporation into
generative grammar; it profited from the superior position that the generative
paradigm occupied in linguistic theorizing in the period in question. Intrinsi-
cally (i.e. as a theory of lexical semantics), Katzian semantics combined a
culmination of the structuralist approach with two characteristics that were
intimately connected with its incorporation into generative grammar. First,
Katz took over the Chomskyan requirement that linguistic analyses be rig-
idly formalized. In particular, componential analysis in the Katzian model
was at the same time a method of descriptive analysis and a formal appara-
tus that seemed indispensable to comply with the requirement of algorithmic
formalization imposed by Chomskyanism.
Second, Katzian semantics took over the mentalist self-conception of
Chomskyanism. By defining the subject matter of semantics as the compe-
tential 'ability to interpret sentences' of the language user, semantics came to
share the promises of explanatory adequacy that constituted so much of the
appeal of generative grammar.
Third, Katzian semantics brought together the three types of semantic re-
lations that could lie at the basis of structuralist semantic theories. In the
first place, paradigmatic similarity relations along the lines of lexical field
theory showed up in Katz and Fodor's adoption of componential analysis. In
the second place, syntagmatic restrictions on the combination of words were
captured in selectional restrictions (such as that the direct object of eat has
to refer to something edible). And in the third place, the paradigmatic lexical
relations highlighted by Lyons (1963) were not discussed in the 1963 article,
but in 1972 (apparently as a result of the publication of Lyons's book), Katz
pointed out explicitly that semantic theory should be concerned with lexical
relations such as synonymy, antonymy, and hyponymy.
In short, Katzian semantics was a singular combination, within the
framework of generative grammar, of a basic structural semantic methodol-
ogy, a mentalist philosophy of language, and a formalized descriptive appa-
ratus. The further development of lexical semantics was characterized by
two tendencies. In each case, semantics moved away from the structuralist
pole of the Katzian synthesis towards one of the other two poles. On the one
hand, the demands of formalization diminished the structuralist influence in
semantics in favor of logical approaches to meaning analysis. On the other
hand, attempts to take the mentalist position of Katzian semantics seriously
led to a straightforward psychological, cognitive orientation in semantic
studies. For reasons to be explained, the former approach can be called
'neogenerativist'; it will be described presently. The latter, cognitive ap-
proach is introduced in the next section.
In the context of an historical overview of lexical semantics, the evolution
towards logical semantics basically means a shift of emphasis from lexical
semantics to sentential semantics. Truth conditions are the primary point of
interest of logical semantics, and truth conditions are properties of proposi-
tions, not of individual words. Thus, Thomason's statement that a semantic
theory need not specify the way in which items such as walk and run differ
in meaning (1974: 48) is typical for the shift in interest from lexical rela-
tional structures to sentential structures. As a result of this shift in interest,
the logical approach to lexical semantics initially remained somewhat of a
minor tradition - within the field of lexical semantics, that is.
In the 1990s, however, Pustejovsky (1995) gave a new impetus to the
Katzian ideal of a formalized semantic representation by basing it on a logi-
cal rather than a featural formalism. At the same time, he goes way beyond
the Katzian approach by emphasizing the necessity of building a lexicon that
is truly generative - in the sense that it does not just consist of an enumera-
tion of word senses, but incorporates the possibility of formally deriving new
readings from stored ones. In this sense, Pustejovsky's approach (which
links up closely with lackendoff's) is a 'neogenerativist' one: it develops the
Katzian ideal of a formal semantic representation by introducing semantic
flexibility and a logical formalism.
2.4. Cognitive Semantics
The Katzian approach is characterized by a specific tension between theo-
retical starting-point and actual methodology. Using psycholinguistic data
could be the ultimate consequence of the mentalist position of Katzian se-
mantics; if semantic descriptions do indeed have psychological reality, it
should be methodologically acceptable to use all kinds of psychological evi-
dence. On the other hand, there are two other aspects of Katz's position that
opposed this methodological extrapolation of the mentalist stance. Both the
generative notion of competence and the structuralist attempt to develop an
autonomous method for linguistic semantics led to reluctance with regard to
the use of psychological data. In the light of Weisgerber's charges against
psychological approaches in semantics, it can established that Katz's men-
talism is not a straightforward methodological position: it characterizes the
object of the investigation as something that is psychological real, but it
apparently does not influence the method used to study that object. That
method remains structural, in the sense that it is based on static relations
between linguistic elements, rather than on actual psychological processes.
However, other linguists than Katz himself actually pursued the non-
autonomous methodological approach that could be derived from the Kat-
zian position. Linking up with existing work on natural language categoriza-
tion and meaning representation as carried out in psycholinguistics and arti-
ficial intelligence, this kind of research opens up the field of lexical
semantics to the more general question 'What does an adequate model of
man's use and knowledge of meaning look like?'. To the extent that re-
searchers in psycholinguistics and artificial intelligence generally do not
estimate that the linguistic capacities of man can be studied in isolation from
his other cognitive capacities, the autonomist methodological ideal of struc-
turalism is abandoned, not just by using psycholinguistic data, but more
generally by incorporating the study of natural language semantics into cog-
nitive science at large. In linguistic semantics, this cognitive reorientation
has led to the school of Cognitive Linguistics of which Lakoff (1987) and
Langacker (1987, 1991) are the leading figures.
Cognitive Semantics has proved itself to be a highly productive paradigm
for linguistic research. As far as lexical semantics is concerned, it is perhaps
the most productive of the current approaches, as will become clear in sec-
tion 3.2. (see Violi 1997 for an introduction and further refinements, and
compare Ungerer and Schmid 1996). Among the major conceptual contribu-
tions of Cognitive Semantics to the study of lexical meaning, the following
should be mentioned.
The prototypical theory of categorial structure developed in psycholin-
guistics by Eleanor Rosch is taken as the basis for models of the internal
structure of natural language categories (see Taylor 1995, Geeraerts
1997a, Mangasser-Wahl 2000). (Although it is impossible to elaborate
the point in the present context, I should add that I see the Wierzbickian
approach to semantics - see Wierzbicka 1996, Goddard 1998 - as one of
the variants of the prototype-theoretical conception.)
Along the lines of Lakoff and Jolmson (1980), a renewed interest in
metaphor has led to a new wave of research into the epistemological role
of cognitive models and their experiential background. Conceptual meta-
phors, as sets of related metaphorical expressions with the same source
and target domain, are defined as metaphorical conceptions that tran-
scend the boundaries of individual lexical items. This type of research
has further developed into new investigations of semantic structuring at
large and metonymy in particular (Panther and Radden 1999).
Fillmore's scenes-and-frames approach to semantics takes its starting-
point in the assumption that the human conceptual apparatus does not
consist of isolated concepts, but is organized into larger, internally struc-
tured wholes. These larger chunks of knowledge, comprising coherent
sets of human beliefs, actions, experiences, or imaginations, are called
'scenes'. 'Frames', on the other hand, are the linguistic means available
to refer to (aspects of) the scene. Frame theory has proved a stimulating
framework for the description of verbal meaning, both theoretically and
lexicographically: see Fillmore (1977a), Fillmore and Atkins (1992,
2000).
The background of the basic level hypothesis (see Berlin 1978) is the
ethnolinguistic recognition that folk classifications of biological domains
usually conform to a general organizational principle, in the sense that
they consist of five or six taxonomical levels, one of which is the 'ge-
neric' level. To the extent that the generic level is the core of any folk
biological category, it is the basic level. The generic level, in other words,
is the most salient one within the taxonomy: within the lexical field de-
fined by the taxonomy, the generic level specifies a set of preponderant
items. The basic level thus embodies a set of naming preferences: given a
particular referent, the most likely name for that referent from among the
alternatives provided by the taxonomy will be the name situated at the
basic level.
2.5. An overview of the developments
Given this bird's eye overview of the traditions that have dominated the
course of development of lexical semantics, Figure 1 suggests a schematic
representation of the relevant lines of development. Each of the boxes in the
figure represents one of the approaches identified above. While the vertical
axis is basically a historical one, the bottom line of the figure charts the ap-
proaches that are currently around.
The lines connecting the boxes indicate the relationships between the ap-
proaches. The double arrow between prestructural historical semantics and
structural semantics indicates that the latter is a reaction to the former. The
single arrow between structuralist semantics and generativist semantics em-
phasizes the methodological continuity between both; as described above,
though, generativist semantics also adds some crucial new features to the
study of the lexicon. A similar relationship links generativist semantics to
neogenerativist semantics: there is a fundamental continuity, but there is also
much that is new in Pustejovsky's approach. The relationship between struc-
turalist and neostructuralist semantics is a somewhat more straightforward
line of development, while the relationship between generativist and Cogni-
tive Semantics is rather one of contrast (hence the double arrow).
Two additional features of the development sketched in the figure should be high-
lighted.
First, the approach inaugurated by Katz appears to have had a pivotal
function in the history of lexical semantics: it sununarizes the structural
approach, but at the same time, the two innovations that it added to a more
traditional structuralist methodology eventually led to forms of semantics
that went far beyond the initial structuralist position. The shift towards neo-
generativist semantics on the one hand and toward Cognitive Semantics on
the other is a consequence of the attempt to take, respectively, the formalized
and the mentalist character of Katzian semantics seriously.
Prestructuralist
Semantics
Structuralist
Semantics
Generativist
Semantics
Ir
Neostructuralist
Semantics
Cognitive
Semantics
Neogenerativist
Semantics
Figure 1. Main lines in the history oflexical semantics
Second, as I have explained at greater length elsewhere (1988a), the cog-
nitive approach is to a large extent a return to the questions and interests of
the prestructuralist tradition of diachronic semantics. For one thing, the cog-
nitive revival of diachronic semantics links up with the gigantic amount of
historical work done in the prestructuralist stage of the development of lexi-
cal semantics. Further, Cognitive Semantics and traditional historical seman-
tics share, by and large, a psychological conception of meaning. And finally,
both approaches start from an encyclopedist conception of meaning, in the
sense that lexical meaning is not considered to be an autonomous phenome-
non, but is rather inextricably bound up with the individual, cultural, social,
historical experience of the language user. On these three points, then, there
is a basic methodological contrast between Prestructural and Cognitive Se-
mantics on the one hand, and structuralist semantics (and its immediate neo-
structuralist descendants) on the other.
Against the background of this fundamental methodological distinction,
the neogenerativist approach shares with Cognitive Semantics an interest in
the flexibility and dynamism of meaning. Pustejovsky's work, as we men-
tioned above, is very much concerned with the mechanisms of polysemy, just
like a lot of Cognitive Semantic work. However, the Pustejovskian approach
is reluctant to adhere to what is, in contrast, one of the basic assumptions of
Cognitive Semantics, viz. the encyclopedic nature of meaning - the idea that
it is hardly useful for the analysis of natural language to adhere to a strict
distinction between the level of linguistic meaning stricto sensu and the level
of world knowledge. By maintaining the distinction, formalized semantics of
the lackendoff/Pustejovsky school remains close to the tradition of structural
semantics. (For a good illustration of the consequences of accepting or re-
jecting the distinction, see the discussion between Taylor 1996 and Jacken-
doff 1996.)
In short, the history of lexical semantics is characterized by a succession
of different theoretical approaches that are related by lines of similarity and
extrapolation as well as mutual opposition. As a basic underlying distinc-
tion, we may single out the contrast between an encyclopedist approach on
the one hand, and a more restrictive approach on the other.
3. The empirical scope of lexical semantics
Given the main traditions of lexical semantics as identified in the previous
paragraph, what has been their contribution to lexical semantics in terms of
the empirical field of research covered? Answering that question requires a
conceptual map of the field of lexical semantics, and such a map, I believe,
should be based on two fundamental distinctions: the distinction between
semasiology and onomasiology, and the distinction between the qualitative
and the quantitative aspects of lexical structures. Once the conceptual map
has been set up, we can chart the contribution that was made by the various
theoretical currents as described above.
3.1. A conceptual map of lexical semantics
Before we can go on to present the distinction between semasiology and
onomasiology, and that between the qualitative and quantitative aspects of
lexical structure, it should first be pointed out that these dimensions do not
in themselves exhaust the possibilities for distinguishing between different
forms of doing lexical semantics. At least two additional distinctive dimen-
sions would have to be included to get a truly comprehensive picture: the
distinction between diachronic semantics and synchronic semantics, and the
distinction between referential and non-referential types of meaning. These
two additional dimensions cross-classify with the three basic ones and with
each other: one can study onomasiological changes, for instance, or changes
of non-referential meaning. The reason for not including them, then, is one of
practice rather than principle: in order to get a first picture of the field, I
think it may be acceptable to start off with those dimensions that relate to the
core area of lexical semantics, viz. the synchronic (rather than diachronic)
structure of referential (rather than non-referential) meanings. Also, the di-
mensions that are not explicitly included in the classification are quantita-
tively less important in actual lexical semantic research: diachronic seman-
tics was a dominant field of investigation only in the first stage of
development of lexical semantics, and non-referential types of meaning (like
the emotive or stylistic value of words) have never occupied a central posi-
tion within the actual research activities of lexical semanticists.
Let us now have a closer look at each of the core dimensions. The dis-
tinction between onomasiology and semasiology is a traditional one in Con-
tinental structural semantics and the Eastern European tradition of lexico-
logical research, but it has hardly found its way to the canonical English
terminology of linguistics. It does not appear, for instance, in most standard
textbooks of linguistics. Essentially, it equals the distinction between mean-
ing and naming: semasiology takes its starting-point in the word as a form,
and charts the meanings that the word can occur with; onomasiology takes
its starting-point in a concept, and investigates by which different expres-
sions the concept can be designated, or named.
Traditionally, the onomasiological perspective leads to an investigation
into lexical structures, i.e. to the study of semantically related expressions
(as in lexical field theory, or the study of the lexicon as a relational network
of words interconnected by links of a hyponymical, antonymical, synony-
mous nature etc.). However, studying the designations of a particular con-
cept also opens the way for a contextualized, pragmatic conception of ono-
masiology, involving the actual choices made for a particular name as a
designation of a particular concept or a particular referent. This shift of
perspective involves a shift from an investigation of structure to an investi-
gation of use. The structural conception deals with sets of related expres-
sions, and basically asks the question: what are the relations among the al-
ternative expressions? The pragmatic conception deals with the actual
choices made from among a set of related expressions, and basically asks the
question: what factors determine the choice for one or the other alternative?
Such a perspective automatically leads to a 'quantitative' approach: some
onomasiological choices will be more frequent than others. Turning the shift
of perspective the other way round, these differences in the probability of
occurrence of certain expressions will be reflected on the structural level: we
can then say, for instance, that certain expressions occupy a dominant or
central role within the set of available expressions. More concretely, what
are the different types of investigation that relate to the qualitative and quan-
titative aspects of onomasiology?
The qualitative onomasiological question takes the following form: what
kinds of (semantic) relations hold between the lexical items in a lexicon (or a
subset of the lexicon)? The outcome, clearly, is an investigation into various
kind of lexical structuring: field relationships, taxonomies, lexical relations
like antonymy and so on.
The quantitative onomasiological question takes the following form: are
some categories cognitively more salient than others, that is, are there any
differences in the probability that one category rather than another will be
chosen for designating things out in the world? Are certain lexical categories
more obvious names than others? This type of quantitative research is fairly
new. The best-known model to date is Berlin and Kay's basic level model,
which involves the claim that a particular taxonomical level constitutes a
preferred, default level for categorization (see above). In Geeraerts, Gronde-
laers, and Bakema (1994), the quantitative approach to onomasiology is
generalized: it is shown that onomasiological preferences do not attach to
taxonomical levels (as suggested by the basic level hypothesis) but to indi-
vidual categories. At the same time, corpus-based methods are introduced
for determining the factors that influence actual lexical and categorial
choices in language use.
The distinction may be further illustrated by extrapolating it to semasi-
ological structures. Qualitative aspects of semasiological structure involve
the following questions: which meanings does a word have, and how are they
semantically related? The outcome is an investigation into polysemy, and the
relationships of metonymy, metaphor etc. that hold between the various
readings of an item. Quantitative aspects of lexical structure, on the other
hand, involve the question whether all the readings of an item carry the same
structural weight. The outcome, obviously, is an investigation into proto-
typicality effects of various kinds. Needless to say, the qualitative perspec-
tive is a much more traditional one in semasiological lexicology than the
quantitative one, which was taken systematically only recently, with the birth
and development of prototype theory.
The overall picture of the field of lexical semantics that emerges from the
foregoing is charted in Table 1.
Table 1. Aconceptual map of lexical semantics
'QUALITY':
investigating structure
(elements and relations)
'QUANTITY':
investigating use and
differential weights
within structures
SEMASIOLOGY
senses and semantic
links among senses
(metaphor, metonymy
etc.)
prototypicality as in-
volving salience phe-
nomena, core versus
periphery
ONOMASIOLOGY
semantic relations
among lexical items
(fields, taxonomies,
networks etc.)
differences in cognitive
salience between catego-
ries, basic level phenom-
ena and entrenchment
3.2. The contribution of the traditions
The next step to take involves the recognition that each of the major areas in
the overview presented in the previous paragraph is typically covered by a
specific theoretical tradition from the history of lexical semantics. Each of
the three major traditions mentioned in the first paragraph focuses specifi-
cally on a particular area of the conceptual map. The prestructuralist tradi-
tion of diachronic semantics, to begin with, deals predominantly with the
'qualitative' aspects of semasiology - with processes like metaphor and
metonymy, that do not just function as mechanisms of semantic extension,
but that constitute, at the same time, the synchronic links between the vari-
ous readings of a lexical item. Structuralist semantics, on the other hand,
focuses on 'qualitative' phenomena of an onomasiological (a structural
onomasiological) kind, such as field relations, taxonomical hierarchies, and
lexical relations like antonymy. Cognitive Semantics, finally, shifts the at-
tention towards the 'quantitative' aspects of lexical structure: all forms of
prototypicality effects in the semasiological realm, and basic levels on the
onomasiological side.
This overview of the different descriptive focus of the various traditions
of lexical semantics also shows how the conceptual map presented in Table
1 got gradually filled out by different theoretical approaches with different
descriptive priorities. There is a historical progression, in fact, from 'qualita-
tive' semasiology to 'qualitative' onomasiology, and hence to an interest in
'quantitative' phenomena on the semasiological as well as on the onomasi-
ological level. The actual historical development from prestructuralist se-
mantics over structuralism and generativism to Cognitive Semantics thus
constitutes a gradual extension of the descriptive scope of lexical semantics.
The overview of the descriptive preferences of the major traditions of lexical
semantics is summarized in Table 2.
Table 2. Dominance of the theoretical traditions within lexical semantics
SEMASIOLOGY ONOMASIOLOGY
'QUALITY':
investigating structure prestructuralist semantics structuralist semantics
(elements and relations)
'QUANTITY':
differential weights Cognitive Semantics Cognitive Semantics
within structures
In order to avoid misunderstanding, it is important to strike a note of
warning: the figure charts dominant domains of activity rather than strictly
confined fields. The various traditions, that is, are certainly not restricted in
their activities to the empirical areas mentioned in the overview. Let us con-
sider some important nuances with regard to each of the major traditions.
First, prestructuralist semantics does not have an exclusively semasi-
ological orientation. Onomasiological considerations may enter the classifi-
cation of types of semantic change wherever necessary. The notion of ana-
logical changes of meaning, in particular, presupposes an onomasiological
perspective. 'Analogical change' is the general name for semantic extensions
that copy the meaning of another, related word. Analogical changes can be
subclassified on the basis of the associative onomasiological link that exists
between the expression that is subject to the semasiological extension, and
the expression that furnishes the model for the extension. In cases of seman-
tic loans, for instance, the associative link between the source expression and
the target has a semantic character. In cases of ellipsis (when a word takes
over meanings from a syntactically complex expression of which it is a
part), the relationship between the two expressions involved in the analogical
change is syntagmatic rather than semantic.
Second, structuralist semantics does not have an exclusively onomasi-
ological orientation. By focusing on distinctiveness as a basic phenomenon
to be considered in lexical semantics, it generally proclaims the primacy of
onomasiological phenomena over semasiological phenomena. The research
strategy of componential analysis, for instance, is basically to derive the
semasiological description of the meaning of individual lexical items from an
analysis of the onomasiological relations that the item in question entertains
with other items in the same lexical field. Semasiological analysis, in short,
is far from absent from the structuralist approach, but depends on an initial
onomasiological analysis.
Third, in the generativist and neogenerativist approaches that grew out of
structuralism, the semasiological orientation is even more outspoken. Katz's
componential analysis of bachelor, for instance, focuses on the distinctive-
ness of the various readings of a lexical item among each other, rather than
on the distinctiveness of the items in a lexical field. And Pustejovsky's inter-
est in regular polysemy and the dynamism of meaning has an obvious sema-
siological orientation.
Fourth, Cognitive Semantics is not restricted to prototype theory and ba-
sic level research. Specifically, there are at least two important contributions
that Cognitive Semantics has made to the structural onomasiological per-
spective (both of which have been mentioned above): on the one hand, the
development of the Fillmorean frame model of semantic analysis, and on the
other hand, the introduction of generalized metaphor research in the line of
Lakoff and Johnson. Both approaches are in fact extensions of the set of
lexical relations studied within the realm of onomasiology. On the one hand,
conceptual metaphors involve onomasiological sets of metaphorically related
expressions. On the other, the typical perspective of Fillmorean frame se-
mantics is of a metonymical kind (cp. Koch 1999). To consider an example,
studying verbs like buy and sell by bringing buyers and sellers and goods
and prices into the picture, equals studying the 'referential syntagmatics' of
a lexical item: the way in which the referent of the item (in this case, a par-
ticular commercial transaction) occurs in reality in temporal and spatial and
functional conjunction with other entities, locations, processes, activities, or
whatever.
Given these nuances with regard to the picture drawn in Table 2, we now
have a fairly accurate idea of the contribution made by the various traditions
of research to the overall domain of lexical semantics. In Table 3, the major
nuances are incorporated into the picture presented by Table 2.
Table 3. The contribution of the traditions to the development of the field
'QUALITY':
investigating struc-
ture (elements and
relations)
'QUANTITY' :
differential weights
within structures
SEMASIOLOGY
prestructuralist
semantics: mechanisms
of semantic change
neogenerativist
semantics: regular
polysemy
Cognitive Semantics:
prototype theory
ONOMASIOLOGY
structuralist semantics:
taxonomies and lexical
relations; lexical fields;
syntagmatic relations and
collocations
conceptual metaphors and
metonymies; frames
basic levels and
entrenchment
4. Competition and convergence
Essential to the topic of the present paper is the recognition that there is
empirical progress in the development of the discipline. The actual historical
development from prestructuralist semantics over structuralism and genera-
tivist semantics to Cognitive Semantics constitutes a gradual extension of the
descriptive scope of lexical semantics. This form of empirical progress con-
stitutes a nuance with regard to the historical picture that was painted be-
fore: while the theoretical development of lexical semantics is characterized
by shifts and oppositions, the empirical development is characterized by at
least some degree of complementarity and accumulation.
In the context of the present volume, it seems legitimate to ask whether
this empirical complementarity may lead to a convergence of the various
theoretical approaches. I would venture the suggestion that such a process of
convergence - if it will take place at all - could find a focal point in a prag-
matic, usage-based perspective to lexical semantics. Three observations lie
at the basis of this suggestion.
First, the currently most productive offshoot of structuralist semantics
(the corpus-based analysis of collocational patterns) involves a shift from an
analysis of structure as such (the initial approach of structuralism) to an
analysis of patterns of actual language use. In structuralist parlance, this
constitutes a shift from the level of langue to the level of parole - or, if one
likes, a pragmatic shift towards a usage-based perspective.
Second, a somewhat similar shift characterizes the interest in regular
polysemy of Pustejovskian neogenerativist semantics. Regular patterns of
polysemy distinguish between stored meanings - meanings that have to be
included in the mental lexicon - and derived meanings - meanings that can
be determined through the operation of rules for regular polysemy and con-
textual interpretation. This constitutes a shift from an interest in semantic
representations as such to an interest in semantic representations as the basis
of semantic interpretations - and the process of interpretation is obviously to
be situated at the level of usage.
And third, although the overview presented above makes clear that the
major conceptual contributions to the 'quantitative' perspective have so far
come from Cognitive Semantics, it is equally clear that this area of research
is far from exhausted. Investigations into the actual onomasiological choices
made by language users and the factors influencing such choices have only
started to emerge, so that a continuing interest from the side of Cognitive
Semantics for this field of investigation is far from unlikely.
In various guises, then, a pragmatic turn characterizes many of the con-
temporary approaches to lexical semantics. The usage-based aspects of lexi-
cal semantics would therefore seem to be a primary starting-point for explor-
ing the possibility of theoretical convergences - or, as the case may be, for
establishing fundamental incompatibilities.
Chapter 17
Idealist and empiricist tendencies in cognitive
semantics
Originally published in Theo Janssen and Gisela Redeker (eds.), 1999, Cognitive
Linguistics: Foundations, Scope, and Methodologv 163-194. BerlinfNew York:
Mouton de Gruyter.
Two metatheoretical threads run through the papers collected in this volume.
First, as an antidote against the danger of reinventing the wheel, publications like
the previous two chapters stress the importance of situating Cognitive Linguistics
against a wider background: the history of linguistics, and the interdisciplinary
field of philosophy, psychology, and cognitive science at large. Second, as an
antidote against idle speculation, it is imperative for Cognitive Linguistics to do
research on a firm empirical basis. In the current line of research of the Quantita-
tive Lexicology and Variational Linguistics research group that I lead in Leuven,
this methodological conviction takes the form of statistically advanced forms of
corpus linguistics. In the final chapter of the present volume, the question is tack-
led from a theoretical rather than a practical point of view: how can one defend
the value of an empirical approach against a more idealistically, introspectively
oriented method? The discussion takes the uncharacteristic but somewhat ironic
form of a platonic dialogue, with a prologue and an epilogue.
Prologue
The methodological situation in present-day Cognitive Linguistics is charac-
terized by the existence of two methodological extremes. On the one hand,
there is the idealistic approach most conspicuously advocated by Anna
Wierzbicka in her numerous publications (among which 1985, 1992, 1996).
On the other hand, there exist various tendencies to objectivize the methods
used in Cognitive Linguistics. Roughly, there are three main tendencies at
this end of the methodological opposition: psycholinguistic research (as in
Sandra and Rice 1995 or Gibbs 1994), neurophysiologial modeling (as in
Regier 1995), and quantitative corpus-based analysis (as in Geeraerts,
Grondelaers, and Bakema 1994). Most actual work in Cognitive Linguistics
is probably situated somewhere in-between both extremes, or rather, the
Idealist and empiricist tendencies 417
practical methodology used is more often introspective rather than data-
driven, but without the outspoken idealistic commitments that Anna Wierz-
bicka explicitly describes. Whence the title of the following dialogue (taken
from Machado de Assis' QUincas Borba): even those cognitive linguists who
have never reflected on the epistemological underpinnings of their practical
work, are implicitly caught up in the debate between the idealistic and the
more empiristically minded approaches. My own position in the debate is
clearly not on the idealist side, but the following dialogue (which suitably
takes a platonic form) tries to establish why that choice is less self-evident
than it may appear once one has made it.
THE CAT, THOUGH IT HAS NOT READ KANT,
MAY WELL BE A METAPHYSICAL ANIMAL
A Philosophical Dialogue
Dramatis Personae:
DUODECIMUS EMPIRICUS, a cognitive linguist specializing in corpus-based
techniques of analysis and experimental methods; author of the book Quanti-
tative Methods in Cognitive Semantics.
ANTIPODA PROLIPHICA, a cognitive linguist of a platonic persuasion; author
of the book Universal Fundaments o(Meaning.
HISTRIO POLYMORPHUS, teacher and friend to Duodecimus and Antipoda, a
perpetual source of knowledge, inspiration, and academic tolerance to both
of them.
The scene is set in the garden of Histrio' s villa, where the three of them meet
regularly to discuss on a theme set by Histrio.
D. Well, master, the lavishness of last week's feast has more than soothed
the hunger of our stomachs, but our minds are as thirsty for knowledge as
ever. What then will be the topic for today's meeting? To what surprising
views will your argumentation lead us, as unexpected as the vistas open-
ing from your garden's paths?
H. Gardening, my young friend, is a matter of patience. Before the bushes
and lawns, flowerbeds and trees acquire their ultimate architectural form,
years of preparatory work are necessary. The wilderness has to be con-
quered, the grounds have to be laid out, and the plants have to be guided
and protected year after year before the global pattern finally emerges.
Allow me, therefore, to introduce you indirectly, with some sort of a de-
tour, to the theme that I have in mind.
A. Most certainly so, master, but do allow us to drink to your health first.
Such a long journey as you promise us should not be undertaken without
invigorating the body first.
D. Right you are, Antipoda: may Histrio's thoughts continue to flower like a
cherry tree in spring!
H. Your are too kind to a man in the autumn of his life. But let us take the
first step - a step that brings us to physics rather than metaphysics. Do
you know the story of the ether?
A. We know ether is a colorless liquid used as a solvent, or as an anesthetic.
Are you referring to the way it was discovered?
H. The word ether actually has yet another meaning, not well known nowa-
days, but extremely important in the physical theories of the nineteenth
century. By the end of the nineteenth century, the majority of physicists
was convinced of the existence of an ether: an odorless, colorless and
tasteless substance that filled the universe and that functioned as the ma-
terial bearer of waves of light. If light is a wave, it indeed requires a me-
dium through which its waves are transported, just like the waves of the
sea move through the material medium that is water.
D. So the ether was supposed to be a medium that filled the gaps between
particles of matter? This sounds like a physical kind of horror vacui!
H. Exactly. But the analogy with a better known medium like water led to
further refinements of the concept. A body like the earth, then, that
moved through the ether would inevitably cause an ether wind: the ether
particles in front of the body would be pressed together, while behind it
an area with lowered pressure, so to speak, would be formed.
D. I can see that, but if this ether is odorless, colorless and tasteless, how
could its presence be detected? As you know, such methodological ques-
tions particularly interest me.
H. Scientists tried to experimentally measure the ether wind: a light wave
moving in the same direction as the earth would travel more slowly than a
light wave that went along with the ether wind, because the former would
have to overcome the resistance of the tightly packed ether particles in
front of the moving earth. In 1881, this hypothesis was put to an experi-
mental test by Michelson and Morley. Unfortunately, it appeared that the
beam of light that was sent out in the same direction as the earth did not
go faster than the one that shone in the opposite direction.
D. Perhaps the experimental equipment used had not been sophisticated
enough?
H. I can see you have some experience with experimental research, dear
Duodecimus: your suggestion is precisely what Michelson and Morley
thought. So the experiment was repeated in 1887 with more accurate in-
struments and with a truly ingenious design (which I will not describe,
not to tax your patience). This new design did not bring the expected re-
sults either, however.
A. I can very well imagine the reason why not, master: don't we all know
that the speed of light is invariable?
H. Certainly. But at the time of the experiments, science was not sufficiently
advanced yet to have realized this: we all know now that the speed of
light is constant, but that assumption as such is neither self-evident nor
empirically obvious. Only in 1905, in fact, did Einstein proclaim the in-
existence of the ether wind together with the invariability of the speed of
light.
D. I am rather surprised that you say that Einstein 'proclaimed' the inexist-
ence of the ether wind. Should we not rather say that he discovered its in-
existence?
H. From our point of view, we can certainly say so - because we no longer
believe in the ether either. But for Michelson and Morley things were not
that obvious: they actually continued to believe in the existence of the
ether a long time after Einstein described his alternative view, and they
tried out ever more refined experiments - without much success, as you
can Imagme.
A. So were they not just being stubborn?
H. Not necessarily. The interesting thing about the story, in fact, is its meth-
odological structure. When an experiment fails, two courses can be fol-
lowed: either you assume that there is something wrong with the observa-
tional apparatus with which the experiment was carried out, or you
assume that the theoretical assumptions of the experiment should be sub-
jected to severe criticism. In the former case there is uncertainty concern-
ing the actual result of the experiment. The experiment is considered un-
decided, and it is assumed that the uncertainty can be ironed out by using
more refined experimental techniques. In the latter case the experiment is
recognized as an actual rejection of an hypothesis. Michelson and Mor-
ley, by their repeated endeavors to make the experimental environment
better fit to measure the existence of an ether and the presence of an ether
wind, followed the first tack. Einstein, on the other, took the opposite
course, by taking Michelson and Morley's experimental failure as a
straightforward rejection of the hypothesis that there existed an ether.
D. I must confess that the first surprise in your story comes earlier than
expected. Are you trying to tell us that falsification of scientific hypothe-
ses is not a straightforward matter? That there are cases where it is diffi-
cult to disprove a hypothesis directly?
H. Well, try to look at it in this way. The existence of alternative ways of
dealing with negative experimental results implies that, at least to some
extent, theoretical presuppositions define what is a theoretically relevant
fact: if one sticks to the existence of an ether wind, then the experiment
simply does not yield an acceptable observation; due to the presumed
problems with the experimental design, no theoretically interpretable fact
at all is observed. If, on the other hand, one starts from the presupposi-
tion that there is no ether wind at all, the experiment produces a clear re-
sult.
D. But surely, one could not go on indefinitely ignoring the failure of the
experiments?
H. I would like to leave open the question how long the first position can be
maintained after repeated failures: the crucial point to see is that observa-
tional results do not exist in isolation, but that 'the facts' depend at least
partly on the intensity with which one cherishes certain theoretical con-
cepts - or conversely, on the degree to which one may be ready to give up
such concepts. Of the utmost importance here is the notion of a 'theory of
interpretation' as the embodiment of the view that there are no bare facts:
facts only become facts through the intervention of background assump-
tions and interpretative principles.
A. When I let my initial surprise subside, I think I see a relationship with
what we, cognitive linguists, know as 'internal realism'. Following ideas
from Hilary Putnam, George Lakoff (1987) has characterized the episte-
mological position of Cognitive Linguistics as a rejection of both hard-
core objectivism and radical idealism: there is no doubt about the exis-
tence of the outside world, but at the same time, the world as we know it
is shaped by the concepts we impose upon it. Well, I need not repeat this
to you, of course. As I see it now, the ideas that you just introduced de-
fine a kind of internal realism on the level of scientific methodology.
Facts do not exist as such, but only within a particular framework?
H. I knew you would be clever enough to see that yourself, Antipoda.
A. But that does not mean I believe in it, yet.
H. Knowing your independence of mind, I had not expected otherwise. Be-
fore we start discussing alternatives, however, I have to say something
about the status of such a position of internal realism within the philoso-
phy of science. There is, in fact, a particular conception of the philosophy
of science that embodies such an internal-realistic view, viz. the paradig-
matic theory of science as represented by Thomas Kuhn, Imre Lakatos,
and Paul Feyerabend. Internal realism is present in Kuhn's conception of
scientific paradigms in the following way. Apparent counterevidence with
regard to a paradigmatic theory is not simply accepted as such, but may
merely be considered an indication of a problem that will be solved when
the theory is developed further. Rather than occasioning an immediate re-
jection of the existing theory, counterevidence (if it is not simply disre-
garded as based on irrelevant experiments) leads to the attempt to explain
away or solve the problems in the framework of the existing paradigm.
Paradigmatic theories are given credit, they are given time to solve puz-
zles that arise.
D. That suggests to me that the structure and the epistemological role of
scientific paradigms (seen as a set of theories clustered round a number
of central assumptions) are similar to those of natural language catego-
ries (seen as clusters of conceptual applications concentrated round a
prototypical center). Since both have a formative influence with regard
to, respectively, new theories and new concepts, prototypical categories
may be seen as a reflection, on the microstructural level of individual
natural language concepts, of the same 'paradigmatic' principles that
govern the development of science on the macrostructural level of scien-
tific theories.
H. I would say so, yes. But notice that the similarity between both positions
should not be thought of in a simplistic way. Cognitive Linguistics char-
acterizes the relationship between natural language (and natural thinking,
if you like) and the world in terms of internal realism. But Cognitive Lin-
guistics itself is a scientific theory that may be characterized by internal
realism.
D. So there are two epistemological levels involved: the level of thinking
about the world (the level of folk knowledge, so to speak), and the level
of scientifically thinking about thinking. And you suggest that there is a
homology between both levels: the relationship between thinking and its
object is the same on both levels, even if the object on the second level is
thinking itself. Where is the problem with that?
H. As you will remember from the times when we discussed the concept of
self-reference, interactions between different epistemological levels may
yield remarkable results. In our case, questions about the objectivity of
semantics arise. If not only our thinking as such, but also our thinking
about thinking is influenced by pre-existing conceptualizations that ftmc-
tion as interpretational schemata, what kind of a scientific enterprise is
semantics? If it does not only deal with the way in which the speakers of
a language construct interpretations for the expressions they encounter,
but if semantics itself imposes interpretations on its subject matter, what
claims to objectivity can it make? It's like the Liar's paradox, is it not? If
you are Cretan and claim that all Cretans are liars, you undermine your
own statement. In the same way, if you claim that all human knowledge
is a non-objectivist construction, you inevitably undermine your own po-
sition, to the extent that you suggest that your own general statement is
something of a construction.
A. I get the impression we are reaching the central topic for today. Makes
me thirsty, actually. Shall I fill your glasses?
H. Please do.
A. Here you are. To be honest, I am less convinced there is a problem than
you are. After all, you should not forget that there is a crucial difference
between the two levels. On the first level, there is an ontological distinc-
tion between the object and the subject of the epistemological relation-
ship: the known object is the outside world, and the knowing subject is a
human person (a mind, if you wish). But on the second level, subject and
object coincide. The object is the way we humans conceptualize the
world, that is to say, the way in which the world is known to us in our
knowing minds. At the same time, it is that very same mind that investi-
gates its own knowledge of the outside world. On the first level, we deal
with consciousness, but on the second level, the discussion is about self-
consciousness. And because subject and object coincide in the case of
self-consciousness, there is a transparent relationship between them.
D. So you believe in the unmitigated powers of introspection, the full epis-
temological presence of the mind to itself, the ultimate transparency of
self-knowledge?!
A. There is no reason to be ironical. You know my methodological starting-
point perfectly well. I am convinced that the best guide to and the ulti-
mate ground in semantic analysis is an examination of one's own con-
sciousness - what one knows of the concepts to be analyzed. 'Chasing the
phantom of 'objectivity' through supposedly scientific methods one loses
the only firm ground there is in semantics: the terra firma of one's own
deep intuitions' (Wierzbicka 1985: 43).
D. That makes me an objectivist phantom-chaser, then. You know perfectly
well that I practice semantics in the opposite way: not by denying the im-
portance of intuition, to be sure, but by supplementing and supporting
any reliance on introspection with corpus analysis or experimentation.
When I make a claim about a word's meaning, for instance, I want to
check that claim against the way in which people actually use the word.
Or rather the other way round: if! try to define a word's meaning, I need
evidence about the way the word is being used first. For instance, a defi-
nition should fit the referential range of application of a word: if you do
not know what objects a word can refer to, how can you define the word?
A. But if you know the word, you know how to use it, do you not? So you
will not have to look any further for evidence about how it is to be used:
just consult your own knowledge.
D. No, that will not do, because there is sufficient evidence that an intro-
spective analysis of how people actually use words does not necessarily
conform to what people actually do. I could give you several examples,
but let us maybe just restrict the discussion to a single case that both of
us have already done research on. At one point (Wierzbicka 1985: 382),
you defined dress in the following way:
A KIND OF THING MADE BY PEOPLE FOR WOMEN AND GIRLS TO WEAR. IMAGIN-
ING THINGS OF THIS KIND PEOPLE COULD SAY THESE THINGS ABOUT THEM:
they are made to be worn on the body, below the head, to cover most of the
body
so that all the parts of a woman's body which people think should not nor-
mally be seen are covered with that one thing
and to protect most of the body from undesirable contact with the en-
vironment
and to cause the woman wearing it to look good
they are made in such a way that when they are on the body the lower half
surrounds the lower half of the woman's body from all sides
so that the legs are not separated from one another
and so that the genital area of the woman's body seems to be hidden
and so that women wearing things of this kind look different from men
things of this kind are thought of as something suitable for women to wear
in most kinds of places and in most kinds of circumstances.
Now, as you know, I did a corpus-based study with my collaborators
exploring the actual referential use of clothing terms in Dutch. When we
have a look at the actual garments that occur in our database as instances
ofjurk (the Dutch equivalent of dress), we find cases in our material that
do not conform to the description. If, for instance, 'covering most of the
body' is interpreted as 'covering more than 50% of the body', then a
number of very short summer dresses with open backs and low necklines
do not display the feature in question. And if 'the parts of a woman's
body which people think should not normally be seen' include the upper
part of the thighs, then dresses with long side slits contradict the image.
Furthermore, some dresses have such wide armholes and such a plunging
decolletage that they could not normally be worn without exposure of the
breasts (unless they are worn with an additional t-shirt or blouse under-
neath). The comparison shows, in other words, that the description you
propose may well be adequate for the majority of cases in the range of
dress, but does far from cover all possible instances.
A. This is irrelevant. What my definition describes is not the objective fea-
tures of the things that might be called dresses, but the conceptual knowl-
edge that people have of dresses. Not an extension, if you like, but rather
an intension. In fact, by introducing the phrase 'imagining things of this
kind people could say these things about them-in the definition, the per-
spective is shifted from the objective features of the things that are being
called dresses to the subjective knowledge that people know they have
about dresses when they examine their own consciousness.
D. So you define dress by referring to what people think dresses are?
A. Yes. Is that not a natural thing to do for a so-called 'cognitive' analysis?
What I try to define is the concept 'dress' that people have, not actual
dresses. Let us maybe look at another example. You once made similar
remarks about my analysis of the category fruit. In my definition of the
category, I included the feature that people, 'wanting to imagine such
things [fruits], would imagine them as growing on trees'. You then com-
mented that this feature does not apply to all fruit: raspberries, for exam-
ple, are fruit, and yet they do not grow on trees. 'In my view, however,
this indisputable fact does not disprove the existence of a conceptual link
between 'fruit' and 'trees' (just as the fact that ostriches don't fly does
not disprove the existence of a conceptual link between 'birds' and 'fly-
ing') .... From the fact that people think of ostriches as birds, and of birds
as flying, it doesn't follow that they think of ostriches as flying' (Wierz-
bicka 1996: 164). And from the fact that people think of fruit as growing
on trees, and of raspberries as fruit, it does not follow that they think of
raspberries as growing on trees.
D. That amounts to saying that a definition should not necessarily cover the
entire range of the category to be defined. But such a position contradicts
what you once said, namely that 'a valid definition must be empirically
adequate, that is, it must be phrased in such a way that it covers the en-
tire range of use of a given word, expression, or construction. It is not
good enough if it only fits the more common or more typical cases'
(Wierzbicka 1989: 738).
A. But again, do not forget that the features in my definition refer to mental
states, so to speak, not to objective features of dresses or fruit or birds.
What I claim is that people think of trees any time they think of fruit.
D. I am afraid I cannot follow you there. Even if we think of the features in
your definition as truly conceptual features, they still do not have the re-
quired generality. For instance, if a definition such as that of dress is to
be applicable to all the things that may be called by that name, people
should be able to assert all the characteristics mentioned in the definition
any time they see a dress. But surely, when imagining a less typical kind
of dress than the kind whose features are included in the definition, peo-
pie will not imagine it as a typical case. Or would they? What people
could say about dresses changes when peripheral members of the cate-
gory are at stake: default dresses, for instance, may well cover most of
the body, but I would say that that feature is suppressed when a fancy
type of summer dress is involved. Or, in the case of fruit, what would be
the semantics of a sentence like A raspberry is a fruit? If you paraphrase
it as 'a raspberry is one of those things we think about (among other
things) as growing on trees', you do imply that raspberries are thought of
as growing on trees, and that is obviously not the case. And similarly for
ostriches and birds: if An ostrich is a bird reads 'an ostrich is one of
those living beings thinking about which we would imagine them as being
able to fly', then the sentence would be false. Which amounts to saying
that the type of definition you use is incapable of accounting for categori-
zation judgments. The definition offruit that you suggest is unable to ex-
plain why a raspberry belongs to the category in question.
A. Do not jump to conclusions, dear Duodecimus. I would say there are two
aspects to your remark. First, how is it we can say that raspberries are
fruit, if our concept offruit does not allow us to decide? Second, is a sen-
tence like Raspberries are fruit contradictory? To tackle the first point
first, the referential range of a category like fruit might be settled in a
conventional way, more or less in the same way in which Hilary Putnam
(1973) describes the semantics of natural kind terms. Whether a particu-
lar thing is to be called tiger is one thing, and the knowledge that people
have of tigers is quite something else. On the one hand, there is a social
convention about the names of things, on the other, there is a social con-
vention about the knowledge (called 'stereotypical' by Putnam) that the
speakers of a language are supposed to have of the category. Crucially, it
is not what you know about jade that allows you to determine whether
something is jade or not. According to Putnam, ordinary language users
possess no more than 'stereotypical' knowledge about natural kinds, that
is to say, they are aware of a number of salient characteristics, such as
the fact that water is a transparent, thirst-quenching, tasteless liquid. The
technical definition of water as H
2
0, on the other hand, is to be located
primarily with scientific experts. It is the experts' knowledge that ulti-
mately determines how natural kind terms are to be used. On the one
hand, a 'division of linguistic labor' ensures that there are societal ex-
perts who know that water is H
2
0, that there is a distinction between
elms and beeches, how to recognize gold from pyrites, and so on. On the
other hand, laymen attune their own linguistic usage to that of the expert
scientists, technicians, etc.. The members of the non-specialized group
are not required to have expert knowledge, but if they wish to be consid-
ered full-fledged members of the linguistic community, they are supposed
to know the 'stereotype' connected with a category. A stereotype is, thus,
a socially determined minimum set of data with regard to the extension of
a category.
D. Well, I am fairly familiar with Putnam's theory, and I do not think ,that
his particular model of the division of linguistic labor is empirically ade-
quate. Also, his model is restricted to the semantics of natural kind terms,
so we should be careful with generalizing it. But I admit that that is less
relevant for what you are trying to say: if you assume that naming is a
conventional act that need not be backed up by conceptual knowledge,
then yes, you need not adapt your definition offruit to allow for raspber-
ries. But what about the other objection? Is a sentence like Raspberries
are fruit contradictory?
A. Let me ask you a question in return. Do you think that the semantic
knowledge that we possess at one particular moment is always sufficient
to categorize the entities, situations, events, processes or whatever that
we encounter in reality?
D. I do not see yet where this is leading to, but no, I do not think so. One of
the crucial aspects of human cognition is the way in which it can adapt
itself to changing circumstances in the outside world. The knowledge that
we have has to be applied in flexible ways to new experiences. Inciden-
tally, that is precisely the reason why prototype-theoretical models of se-
mantic structures exert such an appeal on cognitive linguists. At least to
some extent, the peripheral, slightly deviant cases of a category are pre-
cisely those that result from the application of an existing category to ca-
ses that are similar enough to be included, but that may differ in subtle
ways from the existing concept.
A. So you accept that there is a distinction between stored knowledge and
actualized knowledge? Between the semantic knowledge that is mentally
stored and the meanings that appear in actual utterances?
D. Absolutely. This is, to be sure, intimately connected with the 'cognitive'
nature of Cognitive Linguistics itself. Cognitive Linguistics is cognitive
in the same way that cognitive psychology is: by assuming that our inter-
action with the world is mediated through infonnational structures in the
mind.
A. Now if you accept that, Raspberries are fruit is not contradictory once
you realize that my definition of fruit describes the knowledge that we
have of the concept 'fruit' on the deep-seated level of stored meanings. It
tells you what we know about fruit out of context, so to speak. Con-
versely, Raspberries are fruit is an actualization of that knowledge on
the surface level, where the conceptual interaction between what we know
about raspberries and what we know about fruit may lead to a contextual
attunement of the underlying concept.
D. So you have a semantic deep structure, and a semantic surface structure,
and semantic transfonnations that change deep structure into surface
structure?
A. Your tenninology has unfortunate historical connotations, but okay, let
us accept this way of putting it as a convenient figure of speech. An al-
ternative tenninology could be to distinguish a level of 'deep semantics'
from a level of 'contextual' semantics.
D. Ab yes, but then I do not quite understand why you concentrate your
powers of analysis and description (which are considerable, let me be
quite clear about that) on the deep-structural, deep-semantic level. The
surface level, after all, is the level where most of the semantic action is to
be found; I mean, most of the phenomena that a semanticist could be in-
terested in are surface level phenomena. It follows from your own analy-
sis of Raspberries are fruit, for instance, that truth conditions are to be
investigated on the surface level. And the same holds for polysemy. Stan-
dard tests for polysemy, like the zeugma test, or Quine's p and not p-test,
can only be properly applied to the utterance level.
A. Really? But if you define polysemy in tenns of definability (two mean-
ings are different if you cannot define them as being identical), and if
definition refers to the deep-structural, stored readings, then clearly
polysemy questions can be answered on the basis of an introspective
methodology.
D. Still, it remains a matter to be settled empirically whether the lexical
knowledge that people have in their minds confonns to the definition that
you retrieve introspectively. It is not a priori given that the idea of a cate-
gory that people may introspectively retrieve from memory is an adequate
reflection of the extent of that person's actual knowledge of the category.
On the contrary, if it is part of his knowledge to produce or accept an ap-
plication of dress or fruit to cases that do not conform to the stored con-
cept as introspectively defined, then he 'knows' more about the category
than would be included in his introspectively retrieved idea of the cate-
gory. That knowledge, to be sure, is not necessarily conscious knowl-
edge; it is less 'knowledge that (lexical item x may refer to entities with
such and such characteristics)' but rather 'knowledge how (lexical item x
may be successfully used)'. In order, then, to get a better grasp on the
lexical 'knowledge how', usage-based investigations of the type illus-
trated by my own studies are vital, precisely if it is suspected that con-
scious knowledge may only partially cover the full extent of a person's
'knowledge how'. So I really believe that your position will only remain
defensible if you provide us with an explicit model that describes how
stored meanings are transformed into contextual readings at the utterance
level.
A. I am sorry to disappoint you, but it may take some time before I come up
with such a model. After all, why should I? The crucial phenomenon is
the underlying concept, and the only proper way to investigate it is by in-
trospection. So why should I spend my energy on what is (from my point
of view) a secondary phenomenon?
D. One reason could be that your approach implies that your position be-
comes practically immune to criticism if you do not provide a full model
of what happens when stored conceptual knowledge is contextualized. If
an introspective examination of your own consciousness is the ultimate
basis of semantics, how can your analyses be falsified at all? Is not your
introspective methodology a gigantic immunization strategy? To put it
slightly differently, how can you ensure that what you come up with in
the analysis has anything to do at all with the knowledge of the language
that we share, as speakers of the same language?
A. 'My own experience in lexicographical analysis carried out over many
years in seminars and group discussions has led me to the conclusion that
the problem of objectively identifying the 'shared knowledge' and the
shared notions may in fact be more apparent than real [... ] Certainly if
you ask a group of people what a cup or a skirt is, they will at first come
up with overlapping rather than identical responses. If, however, each of
them is induced to investigate his or her own concepts in depth, investing
in the task a large amount of time and effort, then the differences in their
responses will gradually diminish. If they are given a chance to discuss
their differences the agreement will grow even further and the residual
differences will often disappear. In the process, it will become clear that
the underlying concept is the same for all the participants and that the ini-
tial differences in their responses were due to the fact that they had not
yet explored sufficiently their own consciousness' (Wierzbicka 1985: 42-
43).
D. I doubt whether the results of such a discussion will always be as uni-
form as you pretend, but I can see that, within your framework, intersub-
jectivity would indeed be a way of avoiding excessive .... Master?! Are
you feeling alright? Quickly, pour him something to drink!
H. Thank you, thank you children, but I am perfectly okay. I was actually
just closing my eyes to get a better view of your argumentation. Quite a
nice pattern is emerging from your lively conversation, I must say.
A. Surely not as well-balanced as the pattern of your pleasant garden, I dare
say.
H. Oh yes, there is symmetry in what you have said. If you have not been
carried off entirely by your ardent interchanges, you will remember that
your discussion was triggered by my little story about the ether wind. Re-
call that the essential point about the story was this: ontological assump-
tions, i.e. assumptions about what is, need not be subject to straightfor-
ward falsification. Now, in what both of you have said so far, I discern
two different 'paradigms' in the Kuhnian sense; two different ways of
considering cognitive phenomena that start from more or less opposing
ontological assumptions. Your position, dear Antipoda, is idealistic in a
double sense: an ontological sense and a methodological one. Ontologi-
cally, you believe in the existence of well-defined, deep-seated meanings
that are the true essence of our conceptual knowledge - a platonic realm
of ideas, really, but with a psychological rather than a metaphysical sta-
tus. Methodologically, your position is an idealist one to the extent that
you believe in the transparency of the mind to itself - in the epistemologi-
cal primacy of introspection, in other words.
A. That is a fair picture of my position. What about Duodecimus?
H. I would say that his position is the opposite of yours. He believes in the
epistemological primacy of utterance meaning. Ontologically speaking, I
think he would say that meaning is to be situated primarily at the level of
performance. Meanings exist primarily as realized meanings, that is to
say, as meanings inherent in utterances, in actual events of using lan-
guage. Methodologically, this view correlates with the idea that utter-
ances are the place where meanings are closest to us, where they are so to
speak most palpable.
D. Precisely. Corpus-based studies and experimental psycholinguistic re-
search are important to see what actually happens when people use the
language. It has sometimes been said that Cognitive Linguistics is a us-
age-based model, but I feel cognitive linguists sometimes neglect to draw
the methodological consequences of that recognition - viz. that actual
language use is the methodological basis of Cognitive Linguistics.
A. But since you do believe in the importance of stored meanings (as you
explained earlier), what is the status of such stored meanings?
D. That is an interesting point, and I must confess that I have not yet arrived
at an entirely satisfactory picture of what is going on. Methodologically
speaking, the matter is quite clear: stored semantic knowledge is inferred;
it can hardly be observed directly, but it functions as a hidden entity that
helps one to explain the actual behavior that one finds reflected in the us-
age data. There is nothing extraordinary about that, surely: making refer-
ence to hidden entities as hypothetical constructs is common practice to
quite a number of sciences. Ontologically speaking, however, things are
less straightforward. Traditionally, cognitive scientists have conceived of
such stored meanings in a symbolic fashion, as information encoded in
some form of mental language or 'Mentalese'. But that is a far-reaching
assumption that should be handled with great care. Connectionism in par-
ticular has made it clear that we had better be careful with thinking about
stored knowledge as something encoded in a symbolic fashion. So, to be
on the safe side, I would not want to go any further at this point than
simply say that the stored knowledge is the mental apparatus that enables
us to understand and produce meanings (where 'meanings' are to be
found on the level of usage). As far as I am concerned, it is not even cer-
tain that it is useful to speak of 'meanings' as a specific kind of entity on
this level at all. At least as far as I can see, we have only a very rudimen-
tary view of how to think about mental meaning, which gives me one
more motivation to start from the other end - that is, the level of utter-
ances, performance, usage.
A. I can see clearly now how our views mirror each other. But I am not yet
convinced that the ether wind story is applicable to our situation. Master,
how exactly did you see the link?
H. Let us look back at the conversation the two of you had a moment ago,
and reconstruct it along the lines of my earlier story. Antipoda started
from the initial assumption that, for instance, the word dress had a spe-
cific unitary meaning. You, Duodecimus, formulated an experiment that
could prove Antipoda wrong. The 'experiment' (an empirical investiga-
tion into real language use, actually) did not establish the correctness of
Antipoda's view, just like Michelson and Morley's experiment did not es-
tablish the existence of the ether. But instead of accepting your experi-
ment as a straightforward rejection, Antipoda used different possibilities
to salvage her original position: either to claim that your methodology
was not the most appropriate one (meaning can only be grasped by an in-
trospective methodology), or to make slight theoretical adjustments (the
meaning to be defined is a 'deep-structural', 'deep-semantic' phenome-
non). Michelson and Morley, as I mentioned, took the road of methodo-
logical adjustment, but in the case of the ether just as in the case of An-
tipoda's meanings, some people took the road of theoretical adjustments.
(One such approach involved so-called 'Lorentz Fitzgerald contraction' -
the idea that the effect of the ether wind could not be measured because
the moving objects underwent a contraction.) And the ironical thing, Du-
odecimus, is that you cannot even blame Antipoda for doing so, because
her strategy is a legitimate one, if you are willing to extrapolate your po-
sition of internal realism from the theoretical to the metatheoretical level.
D. Good. You have established that deciding between the two models is not
self-evident. But a decision should be possible, should it not? After all,
the debate about the ether was decided eventually. I mean, ontological as-
sumptions cannot be selected at random: if I start from the ontological
assumption that the earth has two moons, the world itself will prove me
wrong. You have made the point that the extrapolation of the notion of
internal realism to the methodological level creates a broad margin for
theoretical and methodological differences; quite different assumptions
may make it more difficult to assess competing theories in a straightfor-
ward way. But there clearly are limits to that flexibility, are there not? If
you have a theory that produces a model that systematically describes
how surface readings are related to stored semantic knowledge, then
surely the pure scope of such a theory would be a reason for preferring it
over a model that restricts itself to stored meanings only!
A. The point is rather too hypothetical for my taste: you will have to admit
that at this point, neither of the two approaches has developed such an
encompassing model yet. So it is rather irrelevant to compare them on
this point. There is another thing, however, that the two of you seem to
forget. The relationship between both approaches, in fact, is not as sym-
metrical as Histrio's picture suggests. I am convinced, dear Duodecimus,
that you are somewhat of a closet introspectivist.
D. Again a surprise! What do you mean?
A. Let us consider the clothing terms project that you referred to earlier. The
methodological starting-point of the project was to investigate to which
actual referents specific terms referred. Right?
D. Most certainly so. By choosing clothing terms as the field of investiga-
tion, we could get independent access to the referents of the terms in
question, for instance by examining the photographs that the clothing
terms refer to in fashion magazines and the like. Then we described the
various features of the referents of any specific clothing term, and on the
basis of this featural description, described the structure of the semasi-
ological range of application of the different garment names.
A. Now notice that your approach is not completely free of intuitive aspects,
in the sense that your own understanding of the instances of language use
under investigation is entirely ignored or suppressed. More precisely, the
referential approach does not entail that the investigation proceeds in a
purely objective fashion, without any recourse to interpretative activities
on the part of the investigator. The point may be illustrated by consider-
ing the initial selection of the descriptive features included in the compo-
nential system. The choice of those features cannot be dictated automati-
cally by the referents of the words themselves. In principle, an infinite
number of characteristics could possibly be included in the descriptive
framework. In the case of trousers, for instance, it would be possible to
refer to the presence of lining in the legs, to whether the hip pockets have
a flap or not, or to the number of nooses in the waist intended to hold a
belt. The fact that, in actual practice, you decided not to include these
features in the componential system is undoubtedly determined by as-
sumptions about their relevance for the description. Ultimately, I would
say that it is your own pre-existent knowledge of the field of Dutch cloth-
ing tenns, and your own intuitions about what would be pertinent fea-
tures for describing that field, that have played their role in the choice of
a particular componential system.
D. Oh, but I have no wish to deny that any semantic analysis has to rely at
least to some extent on an intuitive understanding of linguistic expres-
sions. But we opt for a referentially enriched corpus (as we prefer to call
it) precisely to guide our intuitions, and to restrict the possibilities of an
uncritical or unmitigated reliance on intuitive judgments.
A. But if you accept that meanings are only realized mentally, and that you
need intuition anyway to grasp them at all, why not just accept introspec-
tion as the basic method of research? After all ...
H. Excuse me for interrupting you, but I feel we are reaching the heart of the
matter here, and I would like you both to see quite clearly what is at stake
here.
D. Master, we are eminently interruptible when you summon us. Pray ex-
plain the stakes to us.
H. It will be a pleasure. But shall we perhaps leave our seats and stroll a-
round the garden for a while? I feel we could use a little exercise.
A. We certainly can. Could we perhaps walk to the sundial down yonder? I
have never had a chance to take a good look at it.
H. Certainly. But I suggest not to follow the path that leads directly to it; we
could take the path through the orchard instead. In fact, just as I intro-
duced the topic of our conversation by making a detour, I would now like
to follow another side-path, to see whether another excursion can help us
to bring the discussion to a close.
D. Fine. Let us go peripatetic then.
H. What Antipoda's last remarks make clear, is the fact that semantics is
basically a hermeneutic enterprise. Lexical description, for instance, does
not simply consist of recording referents, but of trying to detennine what
features of the referents motivate or license the use of a particular item,
in short, of understanding expressions. I would now like to show to you
that the opposition between both of you is not uncommon in the henne-
neutic sciences.
D. 'The henneneutic sciences' - what do you understand by that?
H. From a very general point of view, henneneutics is the 'theory of inter-
pretation', a methodological reflection on the principles underlying inter-
pretative activities concerned with literary products, works of art, juridi-
cal texts, and so on. It is, in fact, from the sources of biblical exegesis
and philological scholarship that henneneutics as a theoretical enterprise
came into being in the course of the nineteenth century. More specifi-
cally, henneneutics is a philosophical tradition, inaugurated by Wilhelm
Dilthey at the end of the previous century, that takes interpretation to be
the basic methodological concept of the human sciences (Geisteswissen-
schaften). Dilthey's explicit aim was to provide the human sciences with
a philosophical understanding of their methodological basis that could be
set off against the positivist self-understanding of the natural sciences.
D. So the henneneutic sciences coincide with the human sciences, because
they use an interpretative methodology?
H. More accurately, Dilthey's views consist of a characterization of the
subject-matter of the human sciences, and of a characterization of their
method. The products of the human mind that constitute the subject-
matter of the human sciences are the expression of an inner experience,
an understanding of life in tenns of the intellect, the will, and the emo-
tions that all human beings possess. The method of the human sciences,
in return, consists of an interpretative attempt to recover the original ex-
perience behind the expressions. In this sense, the logic of the human sci-
ences is fundamentally different from that of the natural sciences:
whereas the latter are based on 'Erklarung' (positivistic explanation in
tenns of the immutable laws of nature), the fonner are based on 'Verste-
hen' (understanding of lived experience through its expressions).
A. This emphasis on experience must imply a natural affinity between Cog-
nitive Linguistics and Diltheyan henneneutics, then.
H. All the more so because Dilthey thinks of that experience in tenns that
will sound familiar to cognitive linguists. I think you will also be happy
to note, for instance, that Dilthey mentions the study of language as the
epitome of the henneneutic enterprise. And because experience is a men-
tal phenomenon, there is a link between henneneutics and psychology.
This is not to say that the experiences that are being expressed are always
and only highly personal phenomena. Rather, human experience is also a
cultural and historical phenomenon, and products of human life such as
systems of laws, rituals, and everyday conventions clearly embody the
historically transmitted experience of cultures rather than just individuals
- but even then, the cultural message would cease to exist if it were not
incorporated into the personal experience of individuals. At the same
time, the henneneutic methodology is just as encyclopedic as human ex-
perience itself. While the experience that is being expressed is an encom-
passing phenomenon in which feelings, thoughts, memories, expectations
etc. come together, the interpretative methodology of the human sciences
will ideally be an interdisciplinary fusion of historical, psychological, an-
thropological, and linguistic research.
D. I understand there is an affinity with Cognitive Linguistics, but we have
had many conversations before in which you revealed that Cognitive Lin-
guistics has many predecessors, both in philosophy and linguistics. What
makes Dilthey special?
H. You are absolutely right. If we restrict the perspective to philosophy,
Dilthey is but one representative of a larger number of theoreticians who
worked round the turn of the century and who discussed epistemological
problems that are quite relevant to us today. Think of all those, like Bren-
tano and Husserl, who explored the concept of intentionality. But Dilthey
is special in the context of our present discussion, because he has explic-
itly drawn the attention to the methodological importance of interpreta-
tion for the human sciences. Moreover, and this is the reason that I men-
tioned his name at all, both his own work and the further development of
philosophical henneneutics foreshadow the debate between the two of
you.
A. Neither of us ever had the intention of being absolutely original, I am
sure.
H. Please do not feel diminished - I would be the last to overestimate his-
torical originality. But I think I had better explain to you where the ten-
sion in Dilthey's work may be situated. If the method of the human sci-
ences is based on interpretation (that is, on recovering the experience
behind the expressions), can this process of recovery be concluded in an
objective fashion? Does it lead to certainty? Can we be sure that our in-
terpretations are correct? Can we truthfully detennine the original experi-
ence and the authentic intention behind the expressions? At this point, a
particular tension within Dilthey's work becomes apparent. On the one
hand, Dilthey's attempt to place the human sciences on as secure and as
respectable a footing as the natural sciences induces him to stress the ob-
jectivity of the human sciences. In his text 'Die Entstehung der Herme-
neutik' (1961, originally 1913), for instance, he reacts against the danger
of 'romantic arbitrariness' in the conception of the hermeneutical enter-
prise. If the hermeneutic act is basically an empathic subjective re-
enactment of an original creative act, objective standards for assessing
the success of an interpretation may be hard to come by. Conversely, if
hermeneutics is modeled on the example provided by biblical (and gener-
ally, philological) textual exegesis, interpretative rules might indeed be
found. At this point, Dilthey is quite outspoken that the latter course is
the one to take. He emphasizes the importance of subjecting the herme-
neutical, interpretative enterprise to critical scrutiny, and stresses that the
task of hermeneutics should be to develop a general theory of interpreta-
tion, a universally applicable interpretative methodology that specifies the
criteria for correct interpretations.
A. You spoke of a tension in his work, so there must be other tendencies at
work as well.
H. In fact, Dilthey's emphasis on the historicity of human experience also
introduces a note of relativism into his views. While nature is governed
by the necessity of natural laws, human life is governed by purposive ac-
tion and conscious effort; the realm par excellence of human life is the
unfolding of history. As such, can there be a general interpretative
method that universally spans the entire range of the history of mankind,
disregarding the historical differences between the objects of its interpre-
tative activities? Can all historical products be interpreted on the same
basis? More importantly still, if man in general is caught in history, can
interpreters free themselves from their own historical background when
confronting products from the past?
A. I can see the tension quite clearly now, but did he find a solution for it in
any way?
H. In Dilthey's own work, the tension between the claim to universality im-
plied by the search for objectivity, and the relativist tendency suggested
by the emphasis on historicity is far from resolved. In the later develop-
ment of hermeneutics, however, the non-universalist, non-objectivist ten-
dency becomes most prominent, most clearly so in the developmental line
leading from Martin Heidegger to Hans-Georg Gadamer on the one hand,
and to Jacques Derrida on the other.
D. If I understand you correctly, you are making the point that the differ-
ences of opinion that Antipoda and I have do not involve a choice be-
tween an idealist and an objectivist conception of linguistic semantics,
but rather between two ways of thinking about the interpretative, or if
you wish 'hermeneutic' aspects of linguistics?
H. Indeed. In your discussion of a few moments ago, Duodecimus tried to
find hard and fast evidence that Antipoda's view is straightforwardly
wrong, in the way in which it would be indubitably wrong to claim that
the earth has two natural satellites. But at least as long as we do not have
two full-fledged models to compare, such a straightforward falsification
may be difficult to attain at this stage in the development of Cognitive
Linguistics. Also, the methodological extrapolation of the concept of in-
ternal realism teaches us that theoretical comparison need not be self-
evident. So, perhaps, it may be useful to think of your conflicting views
from a different angle, viz. as different ways of dealing with the interpre-
tative nature of semantics. Specifically, while Antipoda basically trusts
intuitive interpretation, the position taken by Duodecimus is a prolonga-
tion of the critical approach that I mentioned in connection with Dilthey.
D. I have no objections to thinking about my point of view in this way.
When I insist on the importance of corpus-linguistic (or for that matter,
psycho-experimental or neurophysiological) data as the backdrop for se-
mantic research, I basically try to achieve a semantic analysis that is con-
sonant with as many factors as possible: the way words are actually ap-
plied to things in the world, the way in which language is actually
processed, the material embodiment of our linguistic skills, but also all
contextual information we have about the historical and cultural back-
ground of the speakers of the language - everything, in a sense, in which
language may be said to be grounded. Indeed, grounding may well be the
essential concept in a Cognitive Linguistic theory of interpretation.
A. That would seem to put me in the corner of 'romantic arbitrariness', or
even worse, postmodem license a la Derrida. I really have to protest
against that impression! There are at least two aspects of my approach
that counteract any tendencies towards subjective arbitrariness. First,
there is the importance of intersubjectivity which I drew your attention to
a few minutes ago already.
D. And second?
A. Second, I firmly believe that semantic description should be couched in a
vocabulary of universal concepts - a set of innate indefinables that is
common to all languages of the world and that constitutes the core of
their vocabulary. If definitions are required to be formulated in terms of
such a natural set of primitive semantic elements, that will sufficiently re-
strict the process of interpretation, will it not?
D. Again a very platonic idea, this universal set of innate ideas. But is it
indeed introspectively clear what those innate ideas are? Is it part of our
intuitive knowledge that we can identify the universal set of primitive se-
mantic elements?
A. I do not think so, actually. The set of universal concepts rather emerges
from the definitional analyses themselves. Carefully defining yields a set
of undefinables.
D. How is that? I would say such an idea contradicts the idea that this set of
indefinables has restrictive power. On the one hand, the set of primitives
is supposed to restrict the set of possible definitions. On the other hand,
the definitional analysis itself is supposed to yield the set of innate con-
cepts. Is that not circular: the definitional analysis yields the set of re-
strictions to which it is supposed to conform?!
A. I may have expressed myself inaccurately. The actual criterion for the
definitions is their cross-linguistic transparency, their understandability,
if you like. A superior definition, you will agree, is one that can be under-
stood without prior knowledge of the language to be defined. In actual
practice, this is achieved by couching the definition in terms that are
cross-linguistically unambiguous, such as SEE and HEAR, THIS, I, YOU,
WHERE, GOOD and BAD, SAME and OTHER, or THINK. SO the ultimate re-
striction is not the innateness of the defining terms, but the cross-
linguistic transparency of the definitions.
D. I sometimes doubt whether transparency is exactly what you achieve with
your type of definitions, but anyway, what strikes me as particularly odd
in this conversation as a whole, is the affinity that now becomes apparent
between Cognitive Linguistics and formal semantics.
A. Now it is my turn to be surprised - flabbergasted, rather. A likeness be-
tween Cognitive Linguistics and formal semantics? What in love's name
are we to understand by that?
D. You may remember that Histrio once tried to convince us that formal
semantics was (originally at least) a normative enterprise, a method of
stipulating requirements for how to describe meanings. But perhaps, mas-
ter, you had better tell us about that yourself. Am I overlooking some-
thing?
H. You are absolutely right from a historical point of view, Duodecimus.
Formal semantics, as it was developed by applying formal logic to natu-
ral language, was explicitly intended by at least some of its founding fa-
thers (like the members of the Vienna Circle) as a tool for separating
metaphysical speculation from empirically viable statements. Logic is
concerned with the validity of arguments, and the merits of formalization
in logic consist in restricting possible propositions to ones whose validity
may be tested, in contrast with mere speculation. And so formal seman-
tics, like Antipoda's approach, tries to solve Dilthey's problem (if we
may call it that - I mean the problem of controlling the process of inter-
pretation lest it slide off towards 'romantic arbitrariness') by imposing
restrictions on how possible interpretations might be expressed. The clar-
ity of the logical language is a way of ensuring the epistemological viabil-
ity of statements. What can be legitimately expressed as an interpretation
of utterances in a natural language, is restricted by how it can be ex-
pressed in the language of formal logic.
D. In the same way in which Antipoda wishes to restrict the boundless free-
dom of interpretation by imposing requirements on how the interpreta-
tions are expressed. My approach to finding restrictions, by contrast, is
not representational, but rather empirical: I want suggested interpreta-
tions to be empirically grounded in what we know of the cultural back-
ground, the actual behavior of the language users, the physiology of the
human conceptual apparatus.
A. All that is very fine, but I do not exactly feel like a formal semanticist. I
think you have an obligation to explain the differences between my posi-
tion and the formalists. But master, what are you scribbling there in the
sand? I hope we are not boring you.
H. On the contrary. I was thinking about a way to answer your question.
Perhaps you would care to have a look at my little sketch here. If we
make a distinction between the restrictive principles applied by the vari-
ous approaches on the one hand, and their major aim on the other, the dif-
ference between Antipoda's approach and the formalists' is to a large ex-
tent that the latter are hardly interested in cognitive relevance, in the
psychological sense in which Cognitive Semantics tends to conceive of
that term. Cognitive Semantics is interested in the full range of semantic
phenomena that seem to play a role in actual language use - including
those that cannot be formally described within the framework of logical
semantics. Formalists, by contrast, seem willing to accept the restriction
of the description to the truth-functional properties of language alone,
given that those may well be (they feel) the only ones that can be de-
scribed according to the high demands of formalizability that they wish to
impose on linguistics.
RESTRICTIVE METHODOLOGICAL POSI- MAJOR
PRINCIPLE TION AIM
empirical empiricist Cognitive
restrictions Linguistics
cognitive
idealist Cognitive
relevance
representational
Linguistics
restrictions
formal logical
I
semantics validity
I
Figure I. Methodological relations between the different approaches to semantics
A. You take the words right out of my mouth, master. 'Despite its name,
'formal semantics' [... ] doesn't seek to reveal and describe the meanings
encoded in natural language, or to compare meanings across languages
and cultures. Rather, it sees its goal as that of translating certain care-
fully selected types of sentences into a logical calculus. It is interested not
in meaning (in the sense of conceptual structures encoded in language)
but in the logical properties of sentences such as entailment, contradic-
tion, or logical equivalence' (Wierzbicka 1996: 8).
D. Very illuminating indeed - but I notice the sundial is just on the other
side of this hedge. What a beautiful object! How should one read it?
H. Let me show you. See that shadow here? The number that it falls on
along this circular band indicates the hour of the day: what you see is not
time itself, but a sign of time.
D. And just so, linguistic behavior is not the conceptualization as such, but
only a sign to be interpreted.
A. Hm, if you interpret this shadowy sign correctly, you will notice that the
tables are now likely to be being laid out; the hour of dinner is approach-
ing. Shall we perhaps walk back to the house?
H. Yes, we should: a happy hour is awaiting us.
Epilogue
In the foregoing philosophical dialogue, I have compared an empiricist and
an idealist conception of the methodology of Cognitive Semantics. Against
the background of my previous research, this comparison continues the con-
frontation of Cognitive Semantics with philosophical epistemology that I
began in Geeraerts (1985a), where the possible 'backfiring' of Cognitive
Semantics' adherence to internal realism was first described. Other work
along this line of thought includes Geeraerts (1993c), and the final chapter
of Geeraerts (1997a), which includes a discussion of Diltheyan hermeneu-
tics. At the same time, the dialogue is a methodological reflection on the type
of work presented in Geeraerts, Grondelaers, and Bakema (1994), which
takes a heavily empiricist attitude.
The dialogue proceeded in eight steps. First, the notion of internal realism
was introduced on the basis of an example taken from the natural sciences,
involving the interpretation of Michelson and Morley's experiments. Second,
it was recalled that internal realism characterizes the epistemological con-
ception that Cognitive Semantics has of the relationship between language
and the world. Third, the adherence to internal realism was shown to raise a
methodological problem: can it be extrapolated to the relationship between
language and the linguistic description of language? And if so, what safe-
guards are there against a lapse into radical subjectivism? Empiricist and
idealist approaches in Cognitive Semantics were presented as different an-
swers with regard to these questions. Fourth, the initial discussion between
the proponents of both approaches remained an open-ended one. Although it
was shown that an introspective, idealist methodology does not always yield
an adequate description of the way in which words are actually being used,
there were at least two possibilities for the idealist to escape from the conse-
quences of this discrepancy and to keep the debate open: to dissociate judg-
ments about category membership from conceptual knowledge (in a way
linking up with Putnam's theory of natural kind terms), or by imposing a
radical distinction between semantic structure and surface structure. Fifth,
the open-ended discussion was placed in a wider framework, by interpreting
it against the philosophical background of Wilhelm Dilthey' s notion of her-
meneutics. It was shown that the Diltheyan conception of language fits Cog-
nitive Semantics very well. Sixth, starting from this recognition, it was sug-
gested that the debate involves a choice between two ways of thinking about
the interpretative aspects of linguistic semantics, rather than a straightfor-
ward choice between empiricist objectivism and idealist subjectivism in any
crude form. More specifically, the two competing approaches were seen as
representing alternative ways of avoiding the danger of 'romantic arbitrari-
ness' as formulated by Dilthey. Whereas the 'empiricist' approach tries to
achieve maximal empirical grounding for posited interpretations, the 'ideal-
ist' approach tries to restrict the set of possible interpretations by couching it
in a vocabulary of allegedly universal concepts. Seventh, without enforcing a
[mal decision, I suggested that the grounding approach represents the more
radical safeguard against arbitrariness. Finally, the two approaches within
Cognitive Semantics were compared with formal semantics as yet another
answer to Dilthey's problem. Relying on a vocabulary of universal primi-
tives, the idealist approach within Cognitive Semantics shares with formal
semantics a 'representational' strategy for imposing restrictions on linguistic
descriptions. At the same time, it shares with the empiricist approach in
Cognitive Semantics a desire to describe the full range of semantic phenom-
ena that play a role in natural language use, rather than restricting the do-
main of enquiry to the 'logical', truth-functional properties of language.
What emerges from the discussion, then, is not exactly a straightforward
answer with regard to the initial question - although I do feel that an empiri-
cist approach imposes stronger demands on the semantic enterprise than the
idealist one, and as such, provides a more outspoken answer to Dilthey's
problem. The picture emerging from the discussion does, on the other hand,
provide an illuminating view of the present-day situation in linguistic seman-
tics. Perhaps surprisingly, the major conceptions of linguistic semantics
appear to pattern together as different answers with regard to Dilthey's
problem, that is, the problem of what constitutes a legitimate interpretation
of natural language expressions. I would like to suggest that this is a good
reason for putting the interpretative nature of linguistic semantics in the
forefront of theoretical attention.
References
Agirre, Eneko, and Philip Edmonds (eds.)
2006 Word Sense Disambiguation. Algorithms and Applications.
Berlin: Springer.
Aijmer, Karen
1985 The semantic development of will. In Historical S'emantics -
Historical Word-Formation, Jacek Fisiak (ed.), 11-21. Ber-
lin,lNew York: Mouton de Gruyter.
Anttila, Raimo
1985 Language, cognition, and linguistics. In Linguistics and Phi-
losophy. Essays in Honor of Rulon S. Wells, Adam Makkai and
Alan K. Melby (eds.), 11-22. AmsterdamlPhiladelphia: John
Benjamins.
Apresjan, Juri
1966 Analyse distributionnelle des significations et champs semanti-
ques structures. Langages 1: 44-74.
Armstrong, Sharon L., Lila R. Gleitman, and Henry Gleitman
1983 What some concepts might not be. Cognition 13: 263-308.
Ayto, John
1983 On specifying meaning. In Lexicography. Principles and Prac-
tice, Reinhard R.K. Hartmann (ed.), 89-98. LondonlNew York:
Academic Press.
Baldinger, Kurt
1980 Semantic Theory. Towards a Nfodern Semantics. Oxford: Basil
Blackwell.
Ballmer, Thomas, and Manfred Pinkal (eds.)
1983 Approaching Vagueness. Amsterdam: North Holland Publishing
Company.
Barcelona, Antonio
2001 On the systematic contrastive analysis of conceptual metaphors:
Case studies and proposed methodology. In Applied Cognitive
Linguistics lJ: Language Pedagogy, Martin Piitz, Susanne Nie-
meier, and Rene Dirven (eds.), 117-146. BerlinINew York:
Mouton de Gruyter.
Barere, Bertrand
1975 Rapport du Comite de Salut Public sur les idiomes. In Une poli-
tique de la Langue: La Revolution franr;aise et les Patois, Mi-
chel de Certeau, Dominique Julia, and Jacques Revel (eds.),
291-299. Paris: Editions Gallimard. Original edition 1794.
446 References
Barlow, Michael, and Suzanne Kemmer (eds.)
2000 Usage-based Models ofLanguage. Stanford: CSLI Publications.
Barnes, Jonathan (ed.)
1984 The Complete works ofAristotle. The Revised Oxford Transla-
tion. Princeton: Princeton University Press.
Bartsch, Renate
1985 Sprachnormen: Theorie und Praxis. Ttibingen: Niemeyer Ver-
lag.
Beek, Henri H.
1969 Waanzin in de Middeleeuwen. Haarlem: De Toorts.
Berlin, Brent
1976 The concept of rank in ethnobiological classification: some evi-
dence from Aguarana folk botany. American Ethnologist 3: 381-
400.
1978 Ethnobiological classification. In Cognition and Categorization,
Eleanor E. Rosch and Barbara B. L10yd (eds.), 9-26. Hillsdale,
NJ: Lawrence Erlbaum.
Berlin, Brent, and Paul Kay
1969 Basic Color Terms. Berkeley: University of California Press.
Berlin, Brent. Donald E. Breedlove, and Paul H. Raven
1973 General principles of classification and nomenclature in folk
biology. American Anthropologist 75: 214-242.
Berthele, Raphael
2004 The typology of motion and posture verbs: a variationist account.
In Dialectology meets Typology. Dialect Grammar from a
Cross-Linguistic Perspective, Bernd Kortmann (ed.), 93-126.
BerlinINew York: Mouton de Gruyter.
Biebuyck, Benjamin, Rene Dirven, and John Ries (eds.)
1998 Faith and Fiction. Interdisciplinary Studies on the Interplay
between Metaphor and Religion. Frankfurt am Main: Peter
Lang.
Bierwisch, Manfred
1969 On certain problems of semantic representations. Foundations of
Language 5:153-184.
Bleicher, Josef
1980 Contemporary Hermeneutics. London: Routledge and Kegan
Paul.
Blommaert, Jan, and Chris Bulcaen (eds.)
1997 Political Linguistics. AmsterdamJPhiladelphia: John Benjamins.
References 447
1988
1984
Boas, Franz
1911
Vagueness, ambiguity. and all the rest. In Sprachstruktur, Indi-
viduum, und Gesellschaft. Marc van de Velde and Willy Van-
deweghe (eds.), 9-19. Ttibingen: Max Niemeyer Verlag.
Brandt, Per A.
1998 Cats in space. Baudelaire's 'Les chats' read by Jakobson and
Levi-Strauss. In The Roman lakobson Centennial Symposium
1996, Per A. Brandt (ed.), 339-347. Copenhagen: C. A. Reitzel.
Breal, Michel
1897 Essai de Semantique. Paris: Hachette.
Brugman, Claudia
1981 The story of over. MA thesis, University of California at Berke-
ley.
The very idea. A case study in polysemy and cross-lexical gener-
alizations. In Papers from the Parasession on Lexical Seman-
tics, David Testen, Veena Mishra, and Joseph Drogo (eds.), 21-
38. Chicago: Chicago Linguistic Society.
The Story o.fOver: Po(vsemy, S'emantics and the Structure ofthe
Lexicon. New York: Garland.
Burke. Lucy, Tony Crowley, and Allan Girvin (eds.)
2000 The Routledge Language and Cultural Theory Reader. Lon-
donINew York: Routledge.
Bybee, Joan, and Carol L. Moder
1983 Morphological classes as natural categories. Language 59: 251-
270.
Campbell, Lily B.
1930 Shakespeare's Tragic Heroes: Slaves of Passion. Cambridge:
Cambridge University Press.
Carnoy, Albert
1927 La Science du Mot: Traite de Semantique. Leuven: Universitas.
Catlin, Jane-Carol, and Jack Catlin
1972 Intentionality: a source of ambiguity in English? Linguistic In-
quiry 3: 504-508.
Introduction. In Handbook of American Indian Languages,
Franz Boas, 1-83. Washington DC: Government Printing Of-
fice.
Boeve, Lieven, and Kurt Feyaerts (eds.)
1999 Metaphor and Godtalk. Bern: Peter Lang.
Bosch, Peter
1979
448 References
Chapman, Allan
1979 Astrological Medicine. In Health, Medicine, and Mortality in
the Sixteenth Century, Charles Webster (ed.), 275-300. Cam-
bridge: Cambridge University Press.
Chierchia, Gennaro, and Sally McConnell-Ginet
1990 Meaning and Grammar. An Introduction to Semantics. Cam-
bridge, Mass.: MIT Press.
Clark, Herbert R, and Eve V. Clark
1977 Psychology and Language. An Introduction to Psycholinguistics.
New York: Harcourt Brace Jovanovitch.
Coleman, Linda, and Paul Kay
1981 Prototype semantics: the English word lie. Language 57: 26-44.
Colleman, Timothy
2005 Central and less--central subsenses of the Dutch ditransitive:
Evidence from a collexeme analysis. Paper presented at the In-
ternational Cognitive Linguistics Conference 9, Seoul, Korea.
Condillac, Etienne Bonnot de
1973 Essai sur l'Origines des Sciences humaines. Charles Porset (ed.).
Auvers-sur-Oise: GaliIee. Original edition 1746.
Copi, Irving M.
1972 Introduction to Logic. 4th ed. New York: MacMillan.
Coseriu, Eugenio
1964 Pour une semantique diachronique structurale. Travaux de Lin-
gUistique et de Litterature 2 : 139-186.
Coseriu, Eugenio, and Horst Geckeler
1981 Trends in Structural Semantics. Tiibingen: Gunter Narr Verlag.
Coulson, Seana
2001 Semantic Leaps: Frame-shifting and Conceptual Blending in
Meaning Construction. Cambridge: Cambridge University Press.
Craig, Colette (ed.)
1986 Noun Classes and Categorization. AmsterdamJPhiladelphia:
John Benjamins.
Croft, William
1998 Linguistic evidence and mental representations. Cognitive Lin-
guistics 9: 151-173.
2001 Radical Construction Grammar: S:vntactic Theory from a Typo-
logical Perspective. Oxford: Oxford University Press.
Cruse, D. Alan
1982 On lexical ambiguity. Nottingham Linguistic Circular 11/2: 65-
80.
1986 Lexical S'emantics. Cambridge: Cambridge University Press.
References 449
2000a Aspects of the micro-structure of word meanings. In Polysemy,
Yael Ravin and Claudia Leacock (eds.), 30-51. Oxford: Oxford
University Press.
2000b Meaning in Language. Oxford: Oxford University Press.
Cruttwe11, Patrick
1951 Physiology and psychology in Shakespeare's age. Journal of the
History ofIdeas 12: 75-89.
Cuyckens, Hubert
1991 The Semantics of Spatial Prepositions in Dutch. PhD thesis,
Universitaire Inste11ing Antwerpen.
Danaher, David S.
2003 A cognitive approach to metaphor in prose: Truth and falsehood
in Leo Tolstoy's 'The Death of Ivan Illich'. Poetics Today 24:
439-470.
Darmesteter, Arsene
1887 La Vie des Mots etudifie dans leurs Signijications. Paris: Dela-
grave.
De Beer, Taco
1897 Tijdscluiften. Noord en Zuid 20: 460.
De Flines, Quirijn
1810 Proeve van Hollandsche of Nederduitsche Synonyma. Amster-
dam: Bolt.
De Sutter, Gert, Dirk Speelman, Dirk Geeraerts, and Stefan Grondelaers
2003 Woordvolgordevariatie in tweeledige werkwoordelijke eindgroe-
pen. Naar een statistische evaluatie van zes factoren. In Artike-
len van de Vierde Sociolinguistische Conferentie, Tom Koole,
Jacomine Nortier, and Bert Tahitu (eds.), 111-119. Delft: Ebu-
ron.
Deane, Paul D.
1988 Polysemy and cognition. Lingua 75: 325-361.
1992 Polysemy as the consequence of internal conceptual complexity:
The case of over. Proceedings of the Eastern States Conference
on Linguistics (ESCOL), 9. Ohio State University, CLC Publica-
tions: 32-43.
1993 At, by, to and past: a study in multimodal image theory. In Pro-
ceedings of the Nineteenth Annual Meeting of the Berkeley Lin-
guistics Society, Joshua Guenter, Barbara Kaiser, and Cheryl
Zol1 (eds.). 112-124. Berkeley: : Berkeley Linguistics Society.
Den Boon, Ton, and Dirk Geeraerts
2005 Van Dale Groot Woordenboek van de Nederlandse Taa/. 14th
ed. Utrecht, Antwerpen: Van Dale Lexicografie.
450 References
Deprez, Kas, and Louis Vos (eds.)
1998 Nationalism in Belgium: Shifting Identities. Basingstoke : Mac-
millan.
Dewell, Robert B.
1994 Over again: On the role of image-schemas in semantic analysis.
Diepgen, Paul
1955 Geschichte der Medizin. Berlin, New York: De Gruyter.
Dilthey, Wilhelm
1961 Die Entstehung der Hermeneutik. In Gesammelte Schriften v:
317-331. Leipzig: Teubner. Original edition 1913.
Dirven. Rene
1985 Metaphor and polysemy. Cahiers de l'lnstitut de Linguistique de
Louvain 11: 9-27.
1988 Bibliography of Cognitive Linguistics. Duisburg: Linguistic
Agency of the University of Duisburg.
1990 Metaphorical divergences in the evolution of Dutch and Afri-
kaans. In Diachronic Semantics (Belgian Journal of Linguistics
5), Dirk Geeraerts (ed.), 9-34. Brussels: Editions de l'Universite
de Bruxelles.
2001 The metaphoric in recent cognitive approaches to English
phrasal verbs. Metaphorik.de 1: 39-54.
Dirven, Rene, Bruce Hawkins, and Esra Sandikcioglu (eds.)
2001 Language and Ideology 1. Theoretical Cognitive Approaches.
AmsterdamJPhiladelphia: John Benjamins.
Dirven, Rene, Roslyn Frank, and Cornelia Hie (eds.)
2001 Language and Ideology JJ. Descriptive Cognitive Approaches.
Dirven, Rene, and Marjolijn Verspoor (eds.)
1998 Cognitive Exploration of Language and Linguistics. Amster-
Dobrovol'skij, Dmitri
1995 Schiss und Espenlaub: Idiome der Angst. Folia Linguistica 29:
317-346.
Dowty, David
1979 IVord kfeaning and IvJontague Grammar. Dordrecht: Reidel.
Draper, John W.
1965 The HUl1lors and Shakespeare's Characters. New York: AMS
Press.
Dressier, Wolfgang
1985 Morphophonology. Ann Arbor: Karoma.
1999
Elffers, EIs
1999
References 451
Dubois, Jean
1964 Distribution, ensemble et marque dans le lexique. Cahiers de
Lexicologie 4 : 5-16.
Dunbar, George
2001 Towards a cognitive analysis of polysemy, ambiguity and vague-
ness. Cognitive Linguistics 12: 1-14.
Ekman, Paul, Robert W. Levenson, and Wallace V. Friesen
1983 Autonomic Nervous System Activity Distinguishes among Emo-
tions. Science 221: 1208-1210.
Predecessors of cognitive semantics and speech act theory? In
History of Linguistics 1996. Selected Papers from the Seventh
International Conference on the History of the Language Sci-
ences 11. From Classical to Contemporary LingUistics, David
Cram, Andrew R. Linn and Elke Nowak (eds.), 307-320. Ams-
terdamlPhiladelphia: John Benjamins.
Erdmann, Karl Otto
1910 Die Bedeutung des Wortes. Leipzig: Haessel.
Evans, Vyvian, and Andrea Tyler
2001 Reconsidering prepositional polysemy networks: The case of
over. Language 77: 724-765.
Fellbaum, Christiane (ed.)
1998 WordNet. An Electronic Lexical Database. Cambridge, Mass.:
MIT Press.
Feyaerts, Kurt
1997 Die Bedeutung der Metonymie als konzeptuellen Strukturprin-
zips. Eine kognitiv-semantische Analyse deutscher Dummheits-
ausdrticke. PhD thesis, University ofLeuven.
Metonyl11ic hierarchies: The conceptualization of stupidity in
German idiomatic expressions. In Metonymy in Language and
Thought, Klaus-Uwe Panther and Giinter Radden (eds.), 309-
332.. AmsterdamlPhiladelphia: John Benjal11ins.
Feyaerts, Kurt (ed.)
2003 The Bible through Metaphor and Translation: A Cognitive Se-
mantic Perspective. Bern, Peter Lang.
Fillmore, Charles J.
1975 An alternative to checklist theories of meaning. In Proceedings
of the First Annual Meeting of the Berkeley Linguistics S o c i e ~ y ,
Cathy Cogen, Henry Thol11pson, and James Wright (eds.), 123-
131. Berkeley: Berkeley Linguistics Society.
452 References
1977a Scenes-and-frames semantics. In Linguistic Structures Process-
ing, Antonio Zampolli (ed.), 55-81. Amsterdam: North Holland
Publishing Company.
1977b Topics in lexical semantics. In Issues in Linguistics, Richard
Cole (ed.), 76-138. Bloomington: Indiana University Press.
1982 Towards a descriptive framework for spatial deixis. In Speech,
Place, and Action: Studies in Deixis and Related Topics, Robert
Jarvella and Wolfgang Klein (eds.), 31-59. Chichester: Wiley.
Fillmore, Charles 1., Paul Kay, and Kathryn O'Connor
1988 Regularity and idiomaticity in grammatical construction: The
case of let alone. Language 64: 501-538.
Fillmore, Charles 1., and Beryl T.S. Atkins
1992 Towards a frame-based lexicon: the semantics of risk and its
neighbors. In Frames, Fields and Contrasts, Adrienne Lehrer
and Eve Kittay (eds.), 75-102. Hillsdale, NI: Lawrence ErI-
baum.
2000 Describing polysemy: the case of crawl. In Polysemy. Theoreti-
cal and Computational Approaches, Yael Ravin and Claudia
Leacock (eds.), 91-110. Oxford: Oxford University Press.
Firth, John R.
1957 Papers in Linguistics 193-1-1951. London: Oxford University
Press.
Fodor, Janet, Jerry A. Fodor, and Merrill Garrett
1975 The psychological unreality of semantic representations. Lin-
guistic InqUiry 6: 515-531.
Forceville, Charles
1999 The metaphor COLIN IS A CHILD in Ian McEwan's, Harold Pin-
ter's, and Paul Schrader's 'The Comfort of Strangers'. Metaphor
and S'ymbo/14: 179-198.
Frawley, William
1981 In defense of the dictionary. A response to Haiman. Lingua 55:
53--61.
Freeman, Margaret H.
1995 Metaphor making meaning: Dickinson's conceptual universe.
Journal ofPragmatics 24: 643-666.
2000 Poetry and the scope of metaphor: Toward a cognitive theory of
literature. Afetaphor and Metonymy at the Crossroads: A Cogni-
tive Per.spective, Antonio Barcelona (ed.), 253-281. BerlinlNew
Fried, Mirjam, and Jan-Ola Ostman (eds.)
2004 Construction Grammar in a Cross-Language Perspective. Am-
sterdamlPhiladelphia: John Benjamins.
References 453
Friedman, Hershey
2000 Humor in the Hebrew Bible. Humor. International Journal of
Humor Research 13: 258-285.
Geckeler, Horst
1993 StruktureUe Wortfeldforschung heute. In Studien zur Wortfeld-
theorie. Studies in Lexical Field Theory, Peter R. Lutzeier (ed.),
11-21. Ttibingen: Max Niemeyer Veriag.
Geeraerts, Dirk
1981 Review of George Lakoff and Mark Johnson, Metaphors we Live
by. Quaderni di Sel1lantica 2: 389-396.
1983a Prototype theory and diachronic semantics: a case study. In-
dogermanische Forschungen 88: 1-32.
1983b Review of Kurt Baldinger, Semantic Theory. Lingua 60: 231-
238.
1983c Reclassifying Semantic Change. Quaderni di Semantica 4: 217-
240.
1983d Lexicografie en linguistiek: Reichling gerehabiliteerd. Tijd-
schrijl voor Nederlandse Taal- en Letterkunde 99: 186-206.
1984a Type en prototype. Tijdschrijl voor Tekst- en Taalwetenschap 4:
69-8
1984b Dictionary classification and the foundations of lexicography.
ITL. Review ofApplied LingUistics 63: 37--63.
1985a Paradigm and Paradox. Explorations into a Paradigmatic The-
ory of Meaning and its Epistemological Background. Leuven:
Universitaire Pers.
1985b Les donnees stereotypiques, prototypiques et encyclopediques
dans le dictionnaire. Cahiers de Lexicologie 46: 27-43.
1985c Cognitive restrictions on the structure of semantic change. In
Historical Semantics - Historical Word Formation, Jacek Fisiak
(ed.), 127-153. Beriin/New York: Mouton de Gruyter.
1985d Diachronic extensions of prototype theory. In Meaning and the
lexicon, Geer Hoppenbrouwers, Pieter Seuren and Ton Weijters
(eds.), 354-362. Dordrecht: Foris.
1985e Preponderantievreschillen bij bijna-synoniemen. De Nieuwe
Taalgids 78: 18-27.
1986a Woordbetekenis. Een Overzicht van de Lexicale Semantiek.
Leuven: Acco.
1986b Lexicografische vernieuwing in de vroege achttiende eeuw.
Documentatieblad Achttiende Eeuw 1-14.
1987a On necessary and sufficient conditions. Journal of Semantics 5:
275-291.
454 References
1987b
1988a
1988b
1988c
1988d
1989a
1989b
1989c
1990a
1990b
1992a
1992b
1993a
1993b
1993c
Types of semantic information in dictionaries. In A Spectrum of
Lexicography, Robert Ilson (ed.), 1-10. Amster-
damlPhiladelphia: John Benjamins.
Cognitive Grammar and the History of Lexical Semantics. In
Topics in Cognitive Linguistics, Brygida Rudzka-Ostyn (ed.),
647-677. AmsterdamlPhiladelphia: John Benjamins. [This vol-
ume, Chapter 16.)
Katz revisited: aspects of the history of lexical semantics. In
Understanding the Lexicon, Werner Hiillen and Rainer Schulze
(eds.), 23-35. Tiibingen: Max Niemeyer Verlag.
Review of Anna Wierzbicka, Lexicography and Conceptual
Analysis. Language in Society 17: 449-455.
Where does prototypicality come from? In Topics in Cognitive
Linguistics, Brygida Rudzka-Ostyn (ed.), 207-229. Amster-
damlPhiladelphia: John Benjamins. [This volume, Chapter 2.]
Wat er in een Woord Zit. Facetten van de Lexicale 5,'emantiek.
Leuven: Peeters.
Principles of monolingual lexicography. In Worterbiicher. Ein
internationales Handbuch zur Lexikographie, Franz Josef
Hausmann, Oskar Reichmann, Herbert Ernst Wiegand and La-
dislav Zgusta (eds.), 1: 287-296. BerlinlNew York: De Gruyter.
Prospects and problems of prototype theory. Linguistics 27: 587-
612. [This volume, Chapter I.]
The lexicographical treatment of prototypical polysemy. In
Meanings and Prototypes. Studies in Linguistic Categorization,
Savas L. Tsohatzidis (ed.), 195-210. LondonlNew York:
Routledge. [This volume, Chapter 13.]
Homonymy, iconicity, and prototypicality. Belgian Journal of
Linguistics 5: 49-74.
The semantic structure of Dutch over". Leuvense Bijdragen 81:
205-230. [This volume, Chapter 3.)
Prototypicality and polysemy. On Georges Kleiber, La semanti-
que du prototype. Cognitive Linguistics 3: 219-231.
Vagueness's puzzles, polysemy's vagaries. Cognitive Linguistics
4: 223-72. [This volume, Chapter 5.]
Generalised onomasiological salience. In Perspectives on Lan-
guage and Conceptualization (Belgian Jounal of Linguistics 8),
Jan Nuyts & Eric Pederson (eds.), 43-56. Brussel: Editions de
l'Universite de Bruxelles.
Cognitive semantics and the history of philosophical epistemol-
ogy. In Conceptualizations and Mental Processing in Language,
1995a
1995b
1997a
1997b
1999a
1999b
2000a
2000b
2001a
2001a
2002a
2002b
References 455
Richard Geiger and Brygida Rudzka-Ostyn (eds.), 53-80. Ber-
linINew York: Mouton de Gmyter.
Representational formats in cognitive semantics. Folia Linguis-
tica 39: 21-41.
Specialisation and reinterpretation in idioms. In Idioms. Struc-
tural and Psychological Perspectives, Martin Everaert, Erik-Jan
van der Linden, Andre Schenk, and Rob Schreuder (eds.), 57-
74. Hillsdale, NJ: Lawrence Erlbaum.
Diachronic Prototype Semantics. Oxford: Clarendon Press.
Entfernung, Fortschritt und UnvoIIstiindigkeit. Das historische
VerhaItnis von Lexikologie und Lexikographie. In Linguistische
Theorie und lexikographische Praxis, Klaus-Peter Konerding
and Andrea Lehr (eds.), 7-16. Tiibingen: Max Niemeyer Verlag.
Beer and semantics. In Issues in Cognitive Linguistics, Leon De
Stadler and Chris Eyrich (eds.), 35-55. BerlinINew York: Mou-
ton de Gmyter. [This volume, Chapter 10.]
Hundred years of lexical semantics. In Aetas do jO Encontro
Internacional de Linguistica Cognitiva, Mario Vilela and
Fatima Silva (eds.), 123-154. Porto: Faculdade de Letras.
Adding electronic value. The electronic version of the Grote Van
Dale. In The Ninth Euralex International Congress Proceedings,
Ulrich Heid, Stefan Evert, Egbert Lehmann and Christian
Rohrer (eds.), I: 75-84. Stuttgart: Institut fur MaschineIIe
Sprachverarbeitung.
Honderd jaar Belgisch in Van Dale. Nederlands van Nu 48: 19-
23.
Everyday language in the media. The case of Belgian Dutch
soap series. In Sprache im Alltag. Beitrage zu neuen Perspekti-
ven in der Linguistik Herbert Ernst Wiegand zum 65. Geburtstag
gewidmet. Matthias Kammerer, Klaus-Peter Konerding, Andrea
Lehr, Angelika Storrer, Caja Thimm, and Werner Wolski (eds.),
281-291. BerlinlNew York: WaIter de Gmyter.
On measuring lexical convergence. In Linguagem e C o g n i ~ a o . A
Perspectiva da Linguistica Cognitiva, Augusto Soares da Silva
(ed.), 51-61. Braga: Associac;ao Portuguesa de Linguistica.
De 19de eeuw als lexicale breuklijn. Verslagen en Mededelingen
van de Koninklijke Academie voor Nederlandse Taal- en Let-
terkunde 112: 489-507.
The scope of diachronic onomasiology. In Das Wort. Seine
strukturelle und kulturelle Dimension. Festschrift fur Oskar
Reichmann zum 65. Geburtstag, Vilmos Agel, Andreas Gardt,
456 References
Ulrike Hass-Zumkehr and Thorsten Roelcke (eds.), 29-44.
Tiibingen: Max Niemeyer Verlag.
2003 Meaning and definition. In A Practical Guide to Lexicography,
Piet Van Sterkenburg (ed.), 83-93. All1sterdamIPhiladelphia:
John Benjamins.
2005 Lectal data and empirical variation in Cognitive Linguistics. In
Cognitive Linguistics. Internal Dynamics and Interdisciplinary
Interactions, Francesco Ruiz de Mendoza Ibaiiez and Sandra
Peiia Cervel (eds.), 163-189. BerlinINew York: Mouton de
Gruyter.
Geeraerts. Dirk, and Peter Bakell1a
1993 De prismatische semantiek van idioll1en en composita. Leuvense
Bijdragen 82: 185-226.
Geeraerts, Dirk, Stefan Grondelaers, and Peter Bakema
1994 The Structure ofLexical Variation. Meaning, Naming, and Con-
text. BerlinINew York: Mouton de Gruyter.
Geeraerts, Dirk, Stefan Grondelaers, and Dirk Speelman
1999 Convergentie en Divergentie in de Nederlandse Woordenschat.
Een Onderzoek naar Kleding- en Voetbaltermen. Amsterdam:
Meertens Instituut.
Geeraerts. Dirk. and Guy Janssens
1982 Wegwijs in Woordenboekel1. Een Kritisch Overzicht van de
Lexicograjie van het Nederlands. Assen: Van Gorcum.
Gevaert, Caroline
1994 Prismatische breking bij idiomen. Het prismatische betekenis-
model in de diachrone studie van rei'nterpretatieprocessen bij
idiomen. MA thesis, University of Leuven.
2001 Anger in Old and Middle English: A 'hot' topic? Belgian Es-
says on Language and Literature, 2001: 89-101.
Gibbs, Raymond W.
1986 Skating on thin ice: Literal meaning and understanding idioms
in conversation. Discourse Processes 9: 17-30.
1990 Psycholinguistic studies on the conceptual basis of idiomaticity.
1994 The Poetics of Mind. Figurative Thought, Language, and Un-
derstanding. Cambridge: Cambridge University Press.
1995 Idiomaticity and human cognition. In Idioms: S'tructural and
Psychological Perspectives, Martin Everaert, Erik-Jan Van Der
Linden, Andre Schenk, and Rob Schreuder (eds.), 97-116. Hills-
dale, NJ: Lawrence Erlbaum.
References 457
Gibbs, Raymond W., and Jennifer E. O'Brien
1990 Idioms and mental imagery: The metaphorical motivation for
idiomatic meaning. Cognition 36: 35-68.
Gibbs, Raymond W., Josephine M. Bogdanovich, Jeffrey R. Sykes, and Dale J.
Barr
1997 Metaphor in idiom comprehension. Journal of Memory and
Language 37: 141-154.
Givon, Talmy
1986 Prototypes: between Plato and Wittgenstein. In Noun Classes
and Categorization, Colette Craig (ed.), 78-102. Amster-
Glass, Amold and Keith Holyoak
1975 Altemative conceptions of semantic theory. Cognition 3: 313-
339.
Goddard, Cliff
1998 Semantic Ana(vsis. Oxford: Oxford University Press.
Godderis, Jan
1988 Galenos van Pergamon over Psychische Stoornissen. Leuven:
Acco.
Goldberg, Adele
1992 The inherent semantics of argument structure: the case of the
English ditransitive construction. Cognitive Linguistics 3: 37-
74.
1995 Constructions: A Construction Grammar Approach to Argument
.)'tructure. Chicago, University of Chicago Press.
Good, Edwin M.
1965 Irony in the Old Testament. Philadelphia: The Westminster
Press.
Goodenough, Ward H.
1956 Componential analysis and the study of meaning. Language 32:
195-216.
Goossens, Louis
1990 Metaphtonymy: The interaction of metaphor and metonymy in
expressions for linguistic action. Cognitive Linguistics 1: 323-
340.
Gordon, W. Terrence
1982 A History of Semantics. Amsterdam/Philadelphia: John Benja-
mins.
Gregoire, Henri-Baptiste
1794 [1975] Rapport sur la necessite et les moyens d'aneantir les
patois et d'universaliser l'usage de la langue franc;;aise. In : Mi-
chel de Certeau, Dominique Julia, and Jacques Revel (eds.), Une
458 References
Politique de la Langue: La Revolution franr;aise et les Patois,
300-317. Paris: Editions Gallimard.
Grice, Paul
1989 Studies in the Way of Words, Cambridge, Mass., Harvard Uni-
versity Press.
Gries, Stefan Th.
2003 Multifactorial Analysis in Corpus Linguistics: A Study ofParti-
cle Placement. London: Continuum Press
Grondelaers, Stefan
2000 De distributie van niet-anaforisch er buiten de eerste zinsplaats.
Sociolectische, functionele en psycholinguistische aspecten van
er's status als presentatief signaal. PhD thesis, University of
Leuven.
Grondelaers, Stefan, and Dirk Geeraerts
2003 Towards a pragmatic model of cognitive onomasiology. In Cog-
nitive Approaches to Lexical Semantics, Hubert Cuyckens, Rene
Dirven and John Taylor (eds.), 67-92. BerlinJNew York: Mou-
ton de Gmyter.
Grondelaers, Stefan, Dirk Geeraerts, Dirk Speelman, and Jose Tummers
2001 Lexical standardisation in internet conversations. Comparing
Belgium and The Netherlands. In Proceedings of the First In-
ternational Conference on Language Variation in Europe, Josep
M. Fontana, Louise McNally, M. Teresa Turell, and Enric Vall-
duvi (eds.), 90-100. Barcelona: Universitat Pompeu Fabra.
Grondelaers, Stefan, Dirk Speelman, and Dirk Geeraerts
2002 Regressing on er. Statistical analysis of texts and language
variation. In 6iemes Journees Internationales d'Analyse Statisti-
que des Donnees Textuelles - 6th International Conference on
Textual Data Statistical Analysis, Anne Morin and Pascale Se-
billot (eds.), 335-346. Rennes: Institut National de Recherche en
Infonnatique et en Automatique.
Grondelaers, Stefan, Marc Brysbaert, Dirk Speelman, and Dirk Geeraerts
2002 Er als accessibility marker: on- en offline evidentie voor een
procedurele interpretatie van presentatieve zinnen. Gramma/TTT
9: 1-22.
John Haiman
1980a Dictionaries and encylopedias. Lingua 50: 329-357,1980.
1980b The iconicity of grammar. Language 56:515-54.
1982 Dictionaries and encylopedias again. Discussion. Lingua 56:
353-355.
Halliday, Michael A.K., and Ruqaiya Hasan
1976 Cohesion in English. London: Longman.
References 459
Hamilton, Craig A.
2002 Conceptual integration in Christine de Pizan's 'City of Ladies'.
In Cognitive S'tylistics. Language and Cognition in Text Analy-
sis, Elena Semino and Jonathan Culpeper (eds.), 1-22. Amster-
damlPhiladelphia: John Benjamins.
Hanks, Patrick
1994 Linguistic norms and pragmatic exploitations, or why lexicogra-
phers need prototype theory, and vice versa. In Papers in Com-
putational Lexicography, Ferenc Kiefer, Gabor Kiss, Julia Pajzs
(eds.), 89-113. Budapest: Hungarian Academy of Sciences.
Harder, Peter
2003 The status of linguistic facts. Rethinking the relation between
cognition, social institution and utterance from a functional
point of view. Mind and Language 18: 52-76.
Harkins, Jean, and Anna Wierzbicka (eds.)
2001 Emotions in Cross-linguistic Perspective. Cognitive Linguistics
Research 17. BerlinINew York: Mouton de Gruyter.
Harris, Catherine L.
1989 A connectionist approach to the story of over. In Proceedings of
the Fifteenth Annual Meeting of the Berkeley Linguistics S'oci-
ety, Kira Hall, Michael Meacham, and Richard Shapiro (eds.),
126-137. Berkeley: Berkeley Linguistics Society.
Haser, Verena
2005 Metaphor, Metonymy, and Experientialist Philosophy: Chal-
lenging Cognitive Semantics. BerlinINew York: Mouton de
Gruyter.
Hawkins, Bruce W.
1984 The semantics of English spatial prepositions. PhD thesis, Uni-
versity of California, San Diego.
Hecht, Max
1888 Die griechische Bedeutungslehre, eine Aufgabe der klassischen
Philologie. Leipzig: Teubner.
Heestermans, Hans
1979 Das Woordenboek cler Nederlandsche Taa!. In Praxis der Lexi-
kographie, Helmut Henne (ed.), 68-82. Tiibingen: Max Niemey-
er Verlag.
Heider, Eleanor R.
1972 Universals in color naming and memory. Journal ofExperimen-
tal Psychology 93: 10-20.
Heider, Eleanor R., and Olivier, D.e.
1972 The structure of color space in naming and memory for two
languages. Cognitive Psychology 3: 337-354.
460 References
Herder, Johann Gottfried
1978 Abhandlung iiber den Ursprung der Sprache. Wolfgang Pross
(ed.) . Munchen: Hanser. Original edition 1772.
Herringer, Hans Jurgen
1981 Die Unentscheidbarkeit der Ambiguitiit. In Logos S'emantikos,
Wolf Dietrich and Horst Geckeler (eds.), Ill: 93-126. Madrid:
Editorial Gredos, BerlinINew York: De Gruyter.
Hiraga, Masako K.
2003 How metaphor and iconicity are entwined in poetry: A case in
Haiku. In Jconicity in Language and Literature IlI. From Sign to
Signing, Wolfgang W. G. Muller and alga Fischer 3: 317-335.
AmsterdamIPhiladelphia: John Benjamins.
Hullen, Wemer, and Rainer Schulze (eds.)
1988 Understanding the Lexicon. Meaning, Sense, and World Knowl-
edge in Lexical Semantics. Tubingen: Max Niemeyer Verlag.
Hoffmann, Yael
1983 Irony in the Book of Job. Immanuel 17: 7-21.
1996 A Blemished Perfection: The Book ofJob in Context. Sheffield:
Sheffield Academic Press.
Hoffmann. Robert R.
1989 Some ambiguities in the study of ambiguity. In Resolving se-
mantic ambiguity, David S. Gorfein (ed.), 204-222. New York:
Springer.
Hofmann, Thomas R.
1993 Realms of Meaning. An Introduction to Semantics. London:
Longman.
Holland, Dorothy, and Naomi Quinn (eds.)
1987 Cultural Models in Language and Thought. Cambridge: Cam-
bridge University Press.
Horkheimer, Max. and Theodor W. Adorno
1947 Dialektik der Aujklarung. Amsterdam: Querido.
Horn, Laurence R.
1985 Metalinguistic negation and pragmatic ambiguity. Language 61:
121-174.
Irwin, James R.
1947 Galen on the Temperaments. The Journal of General Psychol-
ogy 36: 45-64.
Itkonen, Esa
1978
1983
Grammatical Theory and Metascience. AmsterdamIPhila-
delphia: John Benjamins.
Causality in LingUistic Theory. London: Croom Helm. Bloom-
ington: Indiana University Press.
References 461
2003 What is Language? A Study in the Philosophy of Linguistics.
Turku: Abo Akademis tryckeri.
Iwata, Seizi
2004 Over-prefixation: A lexical constructional approach. English
Language and Linguistics 8: 239-292.
Jackendoff, Ray
1996 Conceptual semantics and cognitive linguistics. Cognitive Lin-
guistics 7: 93-129.
Jackson, Stanley W.
1986 Melancholia and Depression. From Hippocratic Times to Mod-
ern Times. New Haven: Yale University Press.
1999
2002
Jakel,Olaf
1997 European predecessors of a cognitive theory of metaphor. In
Human Contact through Language and Linguistics, Birgit
Smieja and Meike Tasch (eds.), 69-86. Frankfurt, Peter Lang.
Metonymy in onomastics. In Metonymy in Language and
Thought, Klaus-Uwe Panther and Giinter Radden (eds.), 211-
289. Amsterdam/Philadelphia: John Benjamins.
Hypotheses revisited: The cognitive theory of metaphor applied
to religious texts. Metaphorik.de 2: 20-42.
Jakobson, Roman
1971 Selected Writings 2: Word and Language. The Hague: Mouton.
Joseph, John, and Talbot Taylor (eds.)
1990 Ideologies ofLanguage. London: Routiedge.
Joseph, John
forthc. Linguistic identities and language shift in modern Europe. In
Linguistic Identities, Language Shift and Language Policy in
Europe, Bert Cornillie, Jose Lambert and Pierre Swiggers (eds.).
Leuven/Paris: Peeters.
Kail, Aubrey C.
1986 The Medical Mind of Shakespeare. Balgowla: Williams and
Wilkins.
Katz, Jerrold J.
1972 Semantic Theory. New York: Harper and Row.
Katz, Jerrold J., and Jerry A. Fodor
1963 The structure ofa semantic theory. Language 39, 170-210.
Katz, Jerrold J., and Richard Nagel
1974 Meaning postulates and semantic theory. Foundations of Lan-
guage 11: 311-340.
Kempson, Ruth, and Annabel Connack
1981 Ambiguity and quantification. Linguistics and Philosophy 4:
259-309.
462 References
Kempson, Ruth
1977 Semantic Theory, Cambridge, Cambridge University Press.
Kerkhofs, Jan, Karel Dobbelaere, Liliane Yoye, and Bernadette Bawin-Legros
1992 De Versnelde Ommekeer: de Waarden van r'1amingen, Walen en
Brusselaars in de Jaren '90. Tielt: Lannoo.
Kilgarriff, Adam
1997 I don't believe in word senses. Computers and the Humanities
31: 91-113.
Kleiber, Georges
1988 Prototype, stereotype: un air de famille? DRLA V Revue de Lin-
guistique 38: 1--61.
1990 La Semantique du Prototype. Categories et Sens Lexical. Paris:
Presses Universitaires de France.
1991 Hierarchie lexicale: categorisation verticale et tennes de base.
Semiotiques 1: 35-57.
Klibansky, Raymond, Erwin Panofsky, and Fritz Saxl
1964 Saturn and Melancholy: Studies in the History of Natural Phi-
losophy, Religion, andArt. London: Nelson.
The traditional roots of Jewish humor. In Holy Laughter, Con-
rad Hyers (ed.), 150-165. New York: The Seabury Press.
2000
2002
1989
1995
Frame and contiguity. In Metonymy in Language and Thought,
Klaus-Uwe Panther and Giinther Radden (eds.), 139-167. Ber-
Iin/New York: Mouton de Gruyter.
Kbvecses, Zoltan
1986 Metaphors ofAnger, Pride, and Love. Amsterdam/Philadelphia:
John Benjamins.
Emotion Concepts. New York: Springer.
Anger: its language, conceptualization, and physiology in the
light of cross-cultural evidence. In Language and the Cognitive
Construal of the World, John Taylor and Robert E. MacLaury
(eds.), 181-196. Berlin/New York: Mouton de Gruyter.
A1etaphor and Emotion. Language, Culture and Body in Human
Feeling. Cambridge: Cambridge University Press.
Aletaphor: A Practical Introduction. Oxford: Oxford University
Press.
Kristiansen, Gitte
2003 How to do things with allophones: linguistic stereotypes as cog-
nitive reference points in social cognition. In Cognitive Models
in Language and Thought: Ideologies, Metaphors, and Mean-
Knox, Israel
1969
Koch, Peter
1999
References 463
1983
1987
1996
1986
1970b
1972
1982b
1982a
ings, Rene Dirven. Roslyn Frank, and Martin Piitz (eds.), 69-
120. Berlin/New York: Mouton de Gruyter.
Kristiansen, Gitte, and Rene Dirven (eds.)
forthc. Cognitive S'ociolinguistics. Berlin/New York: Mouton de
Gruyter.
Kronasser, Heinz
1952 Handbuch der Semasiologie. Heidelberg: Winter.
Kuhn, Thomas
1970 The Structure ofScientific Revolutions. Chicago: The University
of Chicago Press.
Labov, William
1973 The boundaries of words and their meanings. In New Ways of
Analysing Variation in English, Charles 1. Bailey and Roger
Shuy (eds.), 340-373. Washington: Georgetown University
Press.
Lakatos, Imre
1970 Falsification and the methodology of scientific research pro-
grammes. In Criticism and the Growth of Knowledge, Imre
Lakatos and Alan Musgrave (eds.), 91-196. Cambridge: Cam-
bridge University Press.
Lakoff, George
1970a A note on vagueness and ambiguity. Linguistic Inquiry 1: 357-
359.
Global rules. Language 46: 627-639.
Linguistics and natural logic. Semantics of Natural Language,
Donald Davidson and Gilbert Hannan (eds.), 545-665.
Dordrecht: Reidel.
Categories and cognitive models. Trier: Linguistic Agency of the
University of Trier. Also as: Berkeley Cognitive Science Report
no. 2 (Berkeley: Institute for Human Learning).
Experiential factors in linguistics. Language, Mind, and Brain,
Thomas W. Simon and Robert 1. Scholes (eds.), 145-156. Hills-
dale, NJ: Lawrence Erlbaum.
Elements of the sound symbolism system. Lecture at the Codoc
symposium on Cognitive Linguistics. Brussels, 26 March 1983.
Classifiers as a reflection of mind. In Noun Classes and Catego-
rization, Colette Craig (ed.), 13-51. Amsterdam/Philadelphia:
John Benjamins.
Women, Fire, and Dangerous Things. What Categories Reveal
about the Mind. Chicago: The University of Chicago Press.
Moral Politics. What Conservatives Know that Liberals Don't.
Chicago: The University of Chicago Press.
464 References
Some remarks on spatial prepositional structure. In Worlds be-
hind Words. Essays in Honour of F.G. Droste on the Occasion
ofhis Sixtieth Birthday, Frans Heyvaert and Frieda Steurs (eds.),
77-84. Leuven: Leuven University Press.
1990a
1987
1990b
1991
Lakoff, George, and Mark Johnson
1980 Metaphors we Live by. Chicago: The University of Chicago
Press.
Lakoff, George, and Zoltan Kovecses
1987 The cognitive model of anger inherent in American English. In
Cultural Models in Language and Thought, Dorothy Holland
and Naomi Quinn (eds.), 195-221. Cambridge: Cambridge Uni-
versity Press.
Landheer. Ronald
1991 Tekstsemantiek en lexicale semantiek. Jaarboek van de Stich-
ting Instituut voor Nederlandse Lexicologie 1990: 53-66.
Langacker Ronald W.
1983 Foundations of Cognitive Grammar. I: Orientation, 11: Semantic
Structure. Trier: Linguistic Agency of the University of Trier.
Foundations of Cognitive Grammar f. Stanford: Stanford Uni-
versity Press.
Concept, Image, and Symbol. The Cognitive Basis of Grammar.
BerlinINew York: Mouton de Gruyter.
Subjectification. Cognitive linguistics 1: 5-38.
Foundations of Cognitive Grammar If. Stanford: Stanford Uni-
versity Press.
Lara, Luis Fernando
1989 Dictionnaire de langue, encyclopedie et dictionnaire encyclope-
dique: le sens de leur distinction. In Worterbucher. Dictionaries.
Dictionnaires, Franz Josef Hausmann, Oskar Reichmann, Her-
bert Ernst Wiegand, Ladislav Zgusta. (eds.), I: 280-287. Berlin,
New York: Waiter de Gruyter.
Lelunann, Winfred P. (ed.)
1988a Prototypes in Language and Cognition. Ann Arbor: Karoma.
Lehmann Winfred P.
1988b Review of R. Langacker, Foundations of Cognitive Grammar.
General linguistics 28: 122-30.
Lehrer, Adrienne
1983 Wine and Conversation. Bloomington: Indiana University Press.
Leys, ado
1988
Lyons, John
1963
1968
References 465
Lewis, David
1972 General semantics. In Semantics of Natural Language, Donald
Davidson and Gilbert Harman (eds.), 169-218. Dordrecht: Rei-
del.
Lindeboom, Gerrit A.
1985 Inleiding tot de Geschiedenis der Geneeskunde. Amsterdam:
Rodopi.
Lindner, Susan 1.
1981 A lexico-semantic analysis of English verb-particle constructions
with up and out. PhD thesis, University of California, San
Diego.
Lipka, Leonhard
1985 Inferential Features in Historical Semantics. Historical .s'eman-
tics - Historical Word-Formalion, Jacek Fisiak (ed.), 339-354.
BerlinlNew York: Mouton de Gruyter.
Lounsbury, Floyd
1956 A semantic analysis of Pawnee kinship usage. Language 32:
158-194.
Lutzeier, Peter R.
1992 Wortfeldtheorie und kognitive Linguistik. Deutsche .s'prache 20:
62-81.
Structural Semantics. Oxford: Blackwell.
Introduction to Theoretical Linguistics. Cambridge: Cambridge
University Press.
Maalej, Zouhair
2004 Figurative language in anger expressions in Tunisian Arabic:
An extended view of embodiment. Metaphor and Symbol 19:
51-75.
MacLaury, Robert E.
1991 Social and cognitive motivations of change: measuring variabil-
ity in color semantics. Language 67: 34-62.
Major, Ralph H.
1954 A History ofMedicine. Springfield, Ill.: Thomas.
Mangasser-Wahl, Martina
2000 Von der zur empirischen ,)'emantik. Frankfurt:
Peter Lang Verlag.
Martin, Laura
1986 Eskimo words for snow. American Anthropologist 88: 418-23.
Marty, Anton
1908 Untersuchungen zur Grundlegung der a/lgemeinen Grammatik
und .s'prachphilosophie. Halle: Niemeyer.
466 References
McCawley, James D.
1971 Where do NP's come from? In Semantics, Danny D. Steinberg
and Leon A. Jakobovits (eds.), 217-231. Cambridge: Cambridge
University Press
1981 Everything that Linguists Have Always Wanted to Know about
Logic - but IVere Ashamed to Ask. Oxford: Blackwell.
Medin, Doug L., and Edward E. Smith
1984 Concepts and concept formation. Annual Review of Psychology
35, 113-138.
Me19uk, Igor
1984 Dictionaire et Combinatoire du Franryais Contempo-
rain. Montreal: Les Presses de L'Universite.
1988 Semantic description of lexical units in an explanatory combina-
torial dictionary: basic principles and heuristic criteria. Interna-
tional Journal ofLexicography 1, 165-188.
1996 Lexical functions: a tool for the description of lexical relations in
a lexicon. In Lexical Functions in Lexicography and Natural
Language Processing, Leo Wanner (ed.), 37-102. Amster-
dam/Philadelphia: John Benjamins.
Melis, Ludo
1990 Pronominal verbs in Old and Modern French, or How prototypes
can be restructured on the basis of permanent meaning effects.
Belgian Journal ofLinguistics 5: 87-108.
Mervis, Carolyn B., and Eleanor Rosch
1981 Categorization of natural objects. Annual Review of Psychology
32: 89-115.
Mikolajczuk, Agnieszka
1998 The metonymic and metaphorical conceptualisation of anger in
Polish. Speaking of Emotions: Conceptualisation and Expres-
sion, Angeliki Athanasiadou and Elzbieta Tabakowska, 153-
190. BerlinlNew York: Mouton de Gruyter
Moon, Rosamund
1998 Fixed Expressions and Idioms in English. A Corpus-based Ap-
proach. Oxford: Oxford University Press.
Nathan, Geoffrey S.
1986 Phonemes as mental categories. In Proceedings of the Twelfth
Annual Meeting of the Berkeley Linguistics Society, Vassiliki
Nikiforidou, Mary VanClay, Mary Niepokuj, Deborah Feder
(eds.), 212-224. Berkeley: Berkeley Linguistics Society.
References 467
Neisser, Ulrich
1987 Concepts and Conceptual Development. Ecological and Intel-
lectual Factors in Categorization. Cambridge: Cambridge Uni-
versity Press.
Nerlich, Brigitte
1990 Change in Language: Whitney, Breal and Wegener. London:
Routledge.
1992 Semantic Theories in Europe 1830-1930. From Etymology to
Contextuality. AmsterdamIPhiladelphia: John Benjamins.
Newmeyer, Frederick J.
1980 Linguistic Theory in America. New York: Academic Press.
Norrick, Neal R.
1981 Semiotic Principles in Semantic Theory. AmsterdamIPhila-
Nunberg, Geoffrey
1979 The non-uniqueness of semantic solutions: polysemy. Linguis-
tics and Philosophy 2: 143-184.
Nyrop, Kristoffer
1913 Grammaire Historique de la Langue Franr;aise IV. Semantique.
Kopenhagen: Gyldendalske Boghandel.
Ortony, Andrew
1988 Are emotion metaphors conceptual or lexical? Cognition and
Emotion 2: 95-103.
bstman, Jan-Ola, and Mirjam Fried (eds.)
2005 Construction Grammars: Cognitive Grounding and Theoretical
Extensions. AmsterdamIPhiladelphia: John Benjamins.
Pagels, Elaine
1996 The Origin ofSatan. Harmondsworth: Penguin.
Palmer, Gary
1996 Toward a Theory of Cultural Linguistics. Austin: University of
Texas Press.
Panther, K1aus-Uwe, and Giinther Radden
1999 Metonymy in Language and Thought. Berlin/New York: Mouton
de Gruyter.
Paul, Hermann
1880 Prinzipien der Sprachgeschichte. Halle: Niemeyer.
Peeters, Bert (ed.)
2000 The Lexicon-Encyclopedia Interface. Oxford: Elsevier Science.
Pluim, Teunis
1894 Keur van Nederlandsche synoniemen. Purmerend: Muusses.
468 References
Pope, Maurice
1985 Shakespeare's Medical Imagination. In :;;hakespeare Survey,
Stanley Wells (ed.), 38: 175-186. Cambridge: Cambridge Uni-
versity Press.
Popova, Yana
2003 'The fool sees with his nose': Metaphoric mappings in the sense
of smell in Patrick Siiskind's Perfume. Language and Literature
12: 136-151.
Porzig, Waiter
1934 Wesenhafte Bedeutungsbeziehungen. Beitrage zur Geschichte
der Deutschen Sprache und Literatur 58: 70-97.
Posner, Michael
1986 Empirical studies of prototypes. Noun classes and categoriza-
tion, Colette Craig (ed.), 53-61. Amsterdam/Philadelphia: John
Benjamins.
Post, Michael
1986 A prototype approach to denominal adjectives. In Linguistics
across Historical and Cleographical Boundaries, Dieter
Kastovsky and Aleksander Szwedek (eds.), 11: 1003-1013. Ber-
linlNew York: Mouton De Gruyter.
Pottier, Bernard
1963 Recherches sur I'analyse semantique en linguistique et en traduc-
tion mecanique. Nancy: Faculte de Lettres et Sciences Humai-
nes.
1964 Vers une semantique moderne. Travaux de Linguistique et de
Litterature 2: 107-137.
Pulman, Steve
1983 Word A1eaning and Belief London: Croom Helm.
Pustejovsky, James
1995 The Generative Lexicon: A Theory of Computational Lexical
semantics. Cambridge, Mass.: MIT Press.
Putnam, Hilary
1975 The meaning of Meaning. In Mind, Language, and Reality.
Philosophical Papers n, Hilary Putnam, 215-271. Cambridge:
Cambridge University Press.
Quadri, Bruno
1952 Aufgaben und Methoden der Onomasiologischen Forschung.
Bern: Francke.
Quine, Willard v.o.
1960 Word and Object. Cambridge, Mass.: MIT Press.
References 469
Ravin, Yael, and Claudia Leacock (eds.)
2000 Polysemy. Theoretical and Computational Approaches. Oxford:
Oxford University Press.
Reddy, Michael
1979 The conduit metaphor - a case of frame conflict in our language
about language'. In Metaphor and thought, Andrew Ortony (ed.),
284-384. Cambridge: Cambridge University Press.
Regier, Terry
1995 A model of the human capacity for categorizing spatial relations.
Reichling. Anton
1935 Het Woord. Zwolle: Tjeenk Willink.
Ricoeur, Paul
1981 Hermeneutics and the Human Sciences. Edited, translated and
introduced by John B. Thompson. Cambridge: Cambridge Uni-
versity Press, Paris: Editions de la Maison des Sciences de
I'Homme.
Robertson, David
1973 The book of Job: A literary study. Soundings 56: 446-469.
1977 The comedy of Job: A response. Semeia 7: 41-44.
Rosch, Eleanor
1973 On the internal structure of perceptual and semantic categories.
In Cognitive Development and the Acquisition of Language,
Timothy E. Moore (ed.), 111-144. New York: Academic Press.
1975 Cognitive representations of semantic categories. Journal of
Experimental Psychology: General 104: 192-233.
1977 Human categorization. In Studies in Cross-cultural Psychology
J, Warren, Neil (ed.), 1-49. New York: Academic Press.
1978 Principles of categorization. In Cognition and Categorization,
Eleanor Rosch and Barbara B. Lloyd (eds.), 27-48. Hillsdale,
NJ: Lawrence Erlbaum.
1988 Coherences and categorization: a historical view. In The Devel-
opment of Language and Language Researchers. Essays in
Honor of Roger Brown, Frank Kessel (ed.). 373-392. Hillsdale,
NI: Lawrence Erlbaum.
Rosch, Eleanor, and Carolyn B. Mervis
1975 Family resemblances: Studies in the internal structure of catego-
ries. Cognitive Psychology 7: 573--605.
Rosch, Eleanor, Carolyn B. Mervis, Wayne D. Gray. David Johnson, and
Penny Boyes-Braem
1976 Basic objects in natural categories. Cognitive 8:
382-439.
470 References
Rosch, Eleanor, Carol Simpson, and Scott R. Miller
1976 Structural bases of typicality effects. Journal of Experimental
Psychology: Human Perception and Peformance 2: 491-502.
Rosier, Irene (ed.)
1988 L'Ambiguite. Cinq Etudes Historiques. Lille: Presses Universi-
taires de Lille.
Ross, John R.
1987 Islands and syntactic prototypes. In Papers from the 23rd An-
nual Jvfeeting of the Chicago Linguistic Society, Barbara Need,
Eric Schiller, and Anna Bosch (eds.), 309-320. Chicago: Chi-
cago Linguistic Society.
Rudzka-Ostyn, Brygida
1985 Metaphoric processes in word formation. The case of prefixed
verbs. The Ubiquity of Metaphor, Wolf Paprotte and Rene
Dirven (eds.), 209-241. AmsterdamJPhiladelphia: John Benja-
mins.
1988 Semantic extensions into the domain of verbal communication.
In Topics in Cognitive Linguistics, Brygida Rudzka-Ostyn (ed.),
507-553. AmsterdamJPhiladelphia: John Benjamins.
1989 Prototypes, schemas, and cross-category correspondences, the
case of ask. Linguistics 27: 613-662.
1996 The Polish dative. In The Dative 1. Descriptive Approaches,
Willy Van Langendonck and William Van Belle (eds.), 341-394.
2003 Word Power: Phrasal Verbs and Compounds. A Cognitive Ap-
proach. BerlinlNew York: Mouton de Gruyter.
Ruhl, Charles
1989 On Monosemy. A Study in LingUistic Semantics. Albany: State
University of New York Press.
Ryder, Mary E.
1994 Ordered Chaos: The Interpretation ofEnglish Noun-Noun Com-
pounds. Berkeley: University of California Press.
Sandra, Dominiek
1998 What linguists can and can't tell you about the human mind: A
reply to Croft. Cognitive Linguistics 9: 361-378.
Sandra, Dominiek, and Sally Rice
1995 Network analyses of prepositional meaning: Mirroring whose
mind - the linguist's or the language user's? Cognitive Lingu-
istics 6: 89-130.
References 471
Sapir, Edward
1907 Herder's 'Urspmng der Sprache'. Modern Philology 5: 109-
142. Reprinted in Historiographia Linguistica 11: 355-388
[1984].
Schafer, Jiirgen
1966 Wort und Begriff Humour in der Elisabethanischen KomOdie.
Miinster: Aschendorff Verlag.
Schiffman, Harold
1996 Linguistic Culture and Language Policy. London: Routledge.
Schmid, Hans-Jorg
1993 Cottage und Co., idea, start vs. begin. Die Kategorisierung als
Grundprinzip einer diJferenzierten Bedeutungsbeschreibung,
Tiibingen: Max Niemeyer Verlag.
Schmidt, Wilhelm
1963 Lexikalische und Aktuelle Bedeutung. Berlin: Akademie-Verlag.
Schulze, Rainer
1988 A short story of down. In Understanding the Lexicon. Meaning,
Sense, and World Knowledge in Lexical Semantics, Werner
Hiillen and Rainer Schulze (eds.), 395-414, Tiibingen: Max
Niemeyer Verlag.
1991 Getting round to (a) round, towards the description and analysis
of a 'spatial' predicate. In Approaches to Prepositions, Gisa
Rauh (ed.), 253-274. Tiibingen: Gunter Nan Verlag.
Seiler, Hansjakob
1986 Apprehension: Language, Object, and Order. Tiibingen: Gunter
Nan Verlag.
Shibatani, Masayoshi, and Sandra A. Thompson (eds.)
1996 Grammatical Constructions: Their Form and Meaning. Oxford:
Oxford University Press.
Siegel, Rudolph E.
1968 Galen's System ofPhysiology and Medicine. Basel: Karger.
Sinclair, John
1987 Looking up. London and Glasgow: Collins.
Sinha, Chris, and Kristina Jensen de Lopez
2000 Language, culture and the embodiment of spatial cognition.
Skutnabb--Kangas, Tove
2000 Linguistic Genocide in Education ~ or lVorldwide Diversity and
Human Rights? Mahwah, NI: Lawrence Erlbaum.
Smith, Edward E., and Doug Medin
1981 Categories and Concepts. Cambridge, Mass.: Harvard Univer-
sity Press.
472 References
Smith, Edward E., Edward J. Shoben, and Lance 1. Rips
1974 Structure and process in semantic memory: a featural model for
semantic decisions. Psychological Review 81. 214-241.
Soriano, Cristina
2003 Some anger metaphors in Spanish and English. A contrastive
review. International Journal ofEnglish Studies 3: 107-122.
Soufas, Teresa Scott
1990 Melancholy and the Secular Mind in Spanish Golden Age Lit-
erature. Columbia: University of Missouri Press.
Speelman, Dirk, Stefan Grondelaers, and Dirk Geeraerts
2003 Profile-based linguistic uniformity as a generic method for com-
paring language varieties. Computers and the Humanities 37:
317-337.
Sperber, Hans
1923 Einfuhrung in die Bedeutungslehre. Bonn: Schroeder.
Starobinsky, Jean
1962 History ofthe Treatment ofMelancholy from the Earliest Times
to 1900. Basel: Geigy.
Steen, Gerard 1.
1994 Understanding Metaphor in Literature: An Empirical study.
London, Longman.
Stefanowitsch, Anatol, and Stefan Th. Gries
2003 Collostructions: Investigating the interaction between words and
constructions. International Journal of Corpus linguistics 8:
209-243.
Stern, Gustav
1931 Meaning and Change ofMeaning. G6teborg: Elanders.
Stevenson, Mark
2001 Word Sense Disambiguation. Combining Knowledge Sources for
5,'ense Resolution. Stanford: CSLI Publications.
Stock, Penelope F.
1983 Polysemy. In LEXeter '83. Proceedings of the Exeter lexicogra-
phy conference, Reinhard RK. Hartmann (ed.), 131-140.
Tilbingen: Max Niemeyer Verlag.
Swanepoel, Piet
1992 Linguistic motivation and its lexicographical application. South
African Journal ofLinguistics 10: 49--60.
1998 Back to basics: prepositions, schema theory, and the explanatory
function of the dictionary. In Euralex '98 Proceedings, Thierry
Fontenelle, Philippe Hiligsmann, Archibald Michiels, Andre
Moulin, Siegfried Theissen (eds.), 655--666. Liege: Universite
de Liege.
References 473
Sweetser, Eve E.
1984 Semantic structure and semantic change: English perception
verbs in an Indo-European Context. Trier: Linguistic Agency of
the University of Trier.
1987 The definition of lie: an examination of folk models underlying
a semantic prototype'. In Cultural models in language and
thought, Dorothy Holland and Naomi Quinn (eds.), 43-66. Cam-
bridge: Cambridge University Press
1999 Compositionality and blending: Semantic composition in a cog-
nitively realistic framework. In Cognitive Linguistics: Founda-
tions, Scope, and Alethodology. Theo Janssen, Gisela Redeker
(eds.), 129-162. BerlinlNew York: Mouton de Gruyter.
Swiggers, Pierre
1988 Grammatical categories and human conceptualization: Aristotle
and the modistae. In Topics in Cognitive Linguistics, Brygida
Rudzka-Ostyn (ed.), 621-646. Amsterdam/ Philadelphia: John
Benjamins.
Tabakowska, Elzbieta
2000 Grammar and the cult of the virgin. A case study of Polish reli-
gious discourse. In Evidence for Linguistic Relativity, Susanne
Niemeier and Rene Dirven (eds.), 223-233. Amster-
Talmy, Leonard
1978 Figure and ground in complex sentences. In Universals of Hu-
man Language 4. Syntax, Joseph H. Greenberg (ed.), 625-49.
Stanford: Stanford University Press.
1983 How language structures space. In Spatial Orientation: Theory,
Research, and Application, Herbert Pick and Linda Acredolo
(eds.), 225-282. New York: Plenum Press.
1985 Force dynamics in language and thought. In Papers .from the
Parasession on Causatives and Agentivity at the Twenty-First
Regionallv/eeting of the Chicago Linguistic Socie(v, William H.
Eilfort, Paul D. Kroeber, and Karen L. Peterson (eds.), 293-337.
Chicago: Chicago Linguistic Society.
Taylor, John
1988
1989
Contrasting prepositional categories: English and Italian. In
Topics in Cognitive Linguistics, Brygida Rudzka-Ostyn (ed.),
299-326. Amsterdam/Philadelphia: John Benjamins Publishing
Company.
Linguistic Categorization. Prototypes in linguistic Theory.
Oxford: Clarendon Press.
474 References
1992
1995
1985
1996
1999
How many meanings does a word have? S'tellenbosch Papers in
Linguistics 25: 133-168.
Linguistic Categorization. in Linguistic Theory. 2nd
ed. Oxford: Oxford University Press.
On running and jogging. Cognitive Linguistics 7: 21-34.
Cognitive semantics and structural semantics. In Historical
Semantics and Cognition, Andreas Blank and Peter Koch (eds.),
2003 Polysemy's paradoxes. Language Sciences 25: 637-655.
Taylor, John, and Thandi G. Mbense
1998 Red dogs and rotten mealies: How Zulus talk about anger.
Speaking of Emotions: Conceptualisation and Expression. An-
geliki Athanasiadou and Elzbieta Tabakowska (eds.), 191 -226.
Berlin/New York: Mouton de Gruyter.
Thomason, Richmond (ed.)
1974 Formal Philosophy. Selected Papers ofRichard Montague. New
Haven: Yale University Press.
Tomasello, Michael
2003 Constructing a Language. A Usage-based Theory of Language
Acquisition. Harvard: Harvard University Press.
Traugott, Elisabeth C.
1989 On the rise of epistemic meanings in English: an example of
subjectification in semantic change. Language 65: 31-55.
'Conventional' and 'dead' metaphors revisited. In The UbiqUity
of Metaphor, Wolf Paprotte and Rene Dirven (eds.), 17-56.
Amsterdam/Philadelphia: John Benjamins.
1999
Trier, Jost
1931 Der Deutsche Wortschatz im Sinnbezirk des Verstandes. Die
Geschichte eines Sprachlichen Fe/des. Heidelberg: Winter.
Tsohatzidis, Savas L. (ed.)
1989 Meanings and Prototypes. S'tudies on Linguistic Categorization.
London: Routledge and Kegan Paul.
Tuggy, David
1993 Ambiguity, polysemy, and vagueness. Cognitive Linguistics 4:
273-290.
Linguistic evidence for polysemy in the mind: A response to
William Croft and Dominiek Sandra. Cognitive Linguistics 10:
343-368.
Tummers, Jose, Kris Heylen, and Dirk Geeraerts
2005 Usage-based approaches in Cognitive Linguistics: A technical
state of the art. Corpus Linguistics and Linguistic Theory 1: 225-
261.
References 475
Tummers, Jose, Dirk Speelman, and Dirk Geeraerts
2004 QuantifYing semantic effects. The impact of lexical collocations
on the inflectional variation of Dutch attributive adjectives. In Le
poids des Mots. Actes des tes Jourm'!es Internationales d'Ana-
lyse Statistique des Donm'!es Textuelles, Gerald Purnelle, Ce-
drick Fairon, and Anne Dister (eds.), 1079-1088. Louvain-la-
Neuve: Presses Universitaires de Louvain.
Turner, Mark
1996 The Literary Mind: The Origins of Thought and Language. Ox-
ford: Oxford University Press.
Tyler, Andrea, and Vyvian Evans
2003 Reconsidering prepositional polysemy networks: The case of
'over'. In Flexible Patterns of Meaning in Alind,
Brigitte Nerlich, Zazie Todd, Vimala Hennan, and David D.
Clarke (eds.), 96-160. BerlinINew York: Mouton de Gruyter.
UWenbeck, Eugenius M.
1967 The dynamic nature of word meaning. Actes du Xe Congres
International des Linguistes It 679-685. Bucarest: Editions de
l'Academie.
Ungerer. Friedrich, and Hans-Jorg Schmid
1996 An Introduction to Cognitive Linguistics. London: Longman.
Van Belle. William, and Willy Van Langendonck
1992 The indirect object in Dutch. Leuvense Bijdragen 81: 17-43.
Vandeloise, Claude
1985a La preposition dans et la relation contenantlcontenu. Leuvense
Bijdragen 74: 193-208.
Van der Meer. Geart
2000 Core. subsense, and the New Oxford Dictionary of English. In
Euralex 2000 Proceedings, Ulrich Heid, Stefan Evert. Egbert
Lelllnann, Christian Rohrer (eds.), 419-431. Stuttgart: Universi-
tat Stuttgart.
Van Dijk, Teun A.
1998 Ideology: A Multidisciplinary Approach. London: Sage.
Van Gijsel, Sofie. Dirk Geeraerts, and Dirk Speelman
2004 A functional analysis of the linguistic variation in Flemish spo-
ken commercials. In Le Poids des Mots. Actes des tes Jourm'!es
Internationales d'Analyse Statistique des Donnees Textuelles,
Gerald Purnelle, Cedrick Fairon, and Anne Dister (eds.), 1136-
1143. Louvain-la-Neuve: Presses Universitaires de Louvain.
Van Ginneken, Jac
1912 Het gevoel in taal- en woordkunst. Leuvense Bijdragen 9: 265-
356,10: 1-156, 173-273.
476 References
Van Hecke, Pierre
2001 Polysemy or homonymy in the root(s) r'h in Biblical Hebrew. A
cognitive-linguistic approach. ZeitschriJt fur Althebraistik 14:
50-66.
Van Langendonck, Willy
1986 Markedness. prototypes, and language acquisition. Cahiers de
l'lnstitut de Linguistique de Louvain 12 : 39-76.
Van Oosten, Jeanne
1986 The nature of subjects. topics, and agents: a cognitive explana-
tion. Bloomington: Indiana University Linguistics Club.
Vandeloise. Claude
1990 Representation, prototypes, and centrality. In Meanings and
Prototypes. Studies in Linguistic Categorization, Savas L. Tso-
hatzidis (eds.), 403-437. London, New York: Routledge.
1994 Methodology and analyses of the preposition in. Cognitive Lin-
guistics 5: 157-84.
Verdoodt, Albert (ed.)
1978 Belgium. (International Journal of the SOCiology of Language
15). The Hague: Mouton.
Verlooy Jan B. C.
1979 Verhandeling op d'Onacht der Moederlyke Tael in de Nederlan-
den. Jozef Smeyers and Jan Van den Broeck (eds.). Den Haag:
Martinus Nijhoff. Original edition 1788.
Violi, Patrizia
1997 Significato ed Esperienza. Milano: Bompiani.
Wagner, Doris
2002 Papier-Bier. Adjektive zur Beschreibung des Bieres in der deut-
schen Bierplakat-Werbung. SKY Journal ofLinguistics 15: 251-
282.
2003 Bier gut, Abend gut! Neuphilologische Mitteilungen 104: 133-
147.
Waismann, Friedrich
1952 Verifiability. In Logic and Language (First ,c.,'eries), Anthony
Flew (ed.), 117-144. Oxford: Basil Blackwell.
Weiland, Petrus. and George Landre
1825 Woordenboek der Nederduitsche synoniemen. 's-Gravenhage:
Allart.
Weinreich. Uriel
1966 Explorations in semantic theory. Current trends in lingUistics
Ill: 395-479.
References 477
Weisgerber, Leo
1927 Die Bedeutungslehre: ein Irrweg der Sprachwissenschaft? Ger-
manisch-Romanische Monatsschrift 15: 161-183.
1954 Die Sprachfelder in der geistigen Erschliessung der Welt. In
Festschrift lost Trier zu seinem 60. Geburtstag, Benno von Wie-
se and Karl Heinz Borck (eds.), 34-49. Meisenheim: Hain.
Wellander, Erich
1928 Studien zum Bedeutungswandel im Deutschen. Uppsala: Ber-
lings Boktryckeri.
Whedbee, William
1977 The Comedy of Job. S'emeia 7: 1-39.
Wierzbicka, Anna
1985 Lexicography and Conceptual Analysis. Ann Arbor: Karoma.
1987a English speech act verbs: a semantic dictionary. Sydney: Aca-
demic Press.
1987b Predict - Prophesy - Forecast. Semantics and lexicography. In
Aspects of Language. Studies in Honour of Mario Alinei 11,
Roberto Crespo, Bill Dotson Smith, and Henk Schultink (eds.),.
Amsterdam: Rodopi.
1989 Prototypes in semantics and pragmatics, explicating attitudinal
meanings in terms of prototypes. Linguistics 27: 731-769.
1990 Prototypes save: on the uses and abuses of the notion 'prototype'
in linguistics and related fields. In Meanings and Prototypes,
Savas L. Tsohatzidis (ed.), 347-367. London: Routledge.
1991 Semantic rules know no exceptions. Studies in Language 15:
371-398.
1992 Semantics, Culture, and Cognition. Universal Human Concepts
in Culture-Specific Configurations. Oxford: Oxford University
Press.
1996 Semantics: Primes and Universals. Oxford: Oxford University
Press.
Winters, Margaret
1987 Syntactic and semantic space: the development of the French
subjunctive. In Papers from the Seventh International Confer-
ence on Historical Linguistics, Anna G. Ramat, Onofrio Car-
ruba, and Giulano Bernini (eds.), 607-618. AmsterdamJPhila-
1989 Diachronic prototype theory: on the evolution of the French
subjunctive. Linguistics 27: 703-730.
Wittgenstein, Ludwig
1953 Philosophische Untersuchungen. Philosophical Investigations.
Oxford: Basil Blackwell.
478 References
Oxford
Wodak, Ruth, and Michael Meyer (eds.)
2001 Methods of Critical Discourse Analysis. London: Sage.
Woolard, Kathryn, Bambi Schieffelin, and Paul Kroskrity
1998 Language Ideologies: Practice and Theory. Oxford:
University Press.
Wundt, Wilhelm
1900 Volkerpsychologie J: Die Sprache. Leipzig: Engelmann.
Yu, Ning
1998 The Contemporary Theory of Metaphor: A Perspective from
Chinese. AmsterdamlPhiladelphia: John Benjamins.
Zgusta, Ladislav
1971 Manual ofLexicography. The Hague: Mouton.
Zlatev. Jordan
2003 Polysemy or generality? Mu. In Cognitive Approaches to Lexical
Semantics, Hubert Cuyckens, Rem: Dirven, and John Taylor
Zwicky, Arnold, and Jerry Sadock
1975. Ambiguity tests and how to fail them. In Syntax and Semantics
.t. John Kimball (ed.), 1-36. New York: Academic Press.
Index of names
Adorno, Theodor W., 280
Agirre, Eneko, 100
Aijmer, Karen, 8
Anttila, Raimo, 140
Apresjan, Juri, 386
Aristotle, 24, 106, 112, 113, 145,
232, 339, 350, 374
Armstrong, Sharon, 13, 16,41
Atkins, Beryl T.S., 362, 405
Ayto, John, 359
Bakema, Peter, 27, 74, 83, 85, 87,
149,151,194,196,198,199,
208,223,262,273,274,363,
410,416,442
Baldinger, Kurt, 78
Ballmer, Thomas, 142
Barcelona, Antonio, 227
Barere, Bertrand, 276, 288
Barlow, Michael, 363
Barnes, Jonathan, 112
Bartsch, Renate, 21
Bawin-Legros, Bernadette, 269
Beek, Henri, 231
Berlin, Brent, 13,84,258,261,406,
410
Biebuyck, Benjamin, 307
Bierwisch, Manfred, 391
Bleicher, Joser. 140
Blommaert, Jan, 274
Boas, Franz, 95
Boeve, Lieven, 307
Bosch, Peter, 123
Brandt, Per A., 307
Breal, Michel, 368, 369, 371. 372,
373,374,396,400
Bmgman, Claudia, 48, 49, 50, 133,
178,352,374
Bmner, Jerome, 43
Brysbaert, Marc, 272
Bulcaen, Chris, 274
Burke, Lucy, 272
Bybee, Joan, 7
Campbell, Lily B., 235
Carnoy, Albert, 368,400
Catlin, Jack, 121
Catlin, Jane-Carol, 121
Cervantes, Miguel de, 239
Chapman, Allan, 236
Chierchia, Gennaro, 221
Chomsky, Noam, 8, 392
Clark, Eve V., 95
Clark, Herbert H., 95
Coleman, Linda, 9, 133, 374
Colleman, Timothy, 175
Condillac, Etienne Bonnot de, 285,
286,287
Copi, Irving M., 113, 130
Cormack, Annabel, Ill, 125
Coseriu, Eugenio, 77, 386, 401
Coulson, Seana, 198
Craig, Colette, 4
Croft William, 99, 175
Crowley, Tony, 272
Cmse, D. Allan, 51,73,101, 105,
107,110,111,122,143,144,
145,146, 147,351,401
Cmttwell, Patrick, 235
Cuyckens, Hubert, 48, 49, 50, 67,
68,69,70,71,178
Danaher, David S., 307
Dannesteter, Arsene, 368, 373, 400
De Beer, Taco, 38
De Flines, Quirijn, 39
De Sutter, Gert, 272
Deane, Paul D., 48, 115, 121, 122,
123, 124, 134, 178
480 Index a/names
Den Boon, Ton, 327
Deprez, Kas, 289
Dewell, Robert B., 48
Diepgen, Paul, 231
Dilthey, Wilhelm, 140, 373, 435,
436,437,438,440,443
Dirven, Rene, 24, 152, 198, 228,
272,273,307,347,374
Dobbelaere, Karel, 269
Dowty, David, 23,386
Draper, John W., 235
DressIer, Wolfgang, 24
Dubois, Jean, 386
Dunbar, George, 99
Edmonds, Philip, 100
Einstein, Albert, 419, 420
Ekman, Paul, 242
Elffers, EIs, 367
Erdmann, Karl Otto, 381
Evans, Vyvian, 48
Feyaerts, Kurt, 198, 212, 307
Feyerabend, Paul, 421
Fillmore, Charles J., 4, 24, 130, 177,
198,362,374,405
Firth, John R., 23, 402
Fodor, Janer, 386
Fodor, Jerry A., 4, 75, 386, 390,
401,402,403
Forceville, Charles, 227
Frank, Roslyn, 272, 273
Frawley, William, 356
Freeman, Margaret H., 307
Fried, MiIjam, 175
Friedman, Hershey, 310
Friesen, Wallace v., 242
Gadamer, Hans-Georg, 140,438
Galen, 232, 235
Garrett, Merrill, 386
Geckeler, Horst, 401
Geeraerts, Dirk, 8, 23, 25, 27, 30,
40,47,50, 71, 72, 74, 80, 81, 83,
85,87,88,90,101, 103, 126,
135, 140, 149, 150, 151, 153,
169,172,178,184,193,194,
196, 198, 199,208,218,223,
227,228,248,251,258,262,
272,273,274,299,310,327,
328,333,334,345,346,351,
357, 361, 363, 367, 374, 375,
377,385,397,399,405,410,
416,442
Gevaert, Caroline, 212, 227
Gibbs, Raymond W., 198,210,248,
416
Girvin, Allan, 272
Givon, Tahny, 24
Glass, Arnold, 392
Gleitman, Henry, 13, 16,41
Gleitman, Lila, 13, 16,41
Goddard, Cliff, 405
Godderis, Jan, 231
Goldberg, Adele, 175, 177, 178,
194, 195, 196
Good, Edwin M., 310
Goodenough, Ward H., 386, 401
Goossens, Louis, 218, 219, 220
Gordon, Terrence W., 385, 398
Gregoire, Henri-Baptiste, 278, 280,
281, 288
Grice, Paul, 150
Gries, Stephan Th., 345
Grondelaers, Stefan, 27, 74, 83, 85,
87, 149, 151, 194, 196,227,228,
262,272,273,274,299,363,
410,416,442
Habennas,Jurgen, 140,276
Hailnan,John, 23, 221, 356, 379
Halliday, Michael A.K., 402
Hamilton, Craig, 307
Hanks, Pat rick, 359, 362
Harder, Peter, 227
Harkins, Jean, 227
Harvey, William, 234
Hasan, Ruqaiya, 402
Haser, Verena, 367
Hawkins, Bruce, 273, 374
Hecht, Max, 378, 379, 397
Heesterrnans, Hans, 332
Heider, Eleanor R, 4, 13
Herder, Johann Gottfried, 282, 285,
286,287
Herringer, Hans-Jiirgen, 107, 110
Heylen, Kris, 345
Hippocrates, 228, 229, 231
Hiraga, Masako K., 307
Hoffmann, Robert R, 137
Hoffmann, Yael, 310, 313, 314, 315,
322
Hofmann, Thomas R, 90
Holland, Dorothy, 4, 273
Holyoak, Keith, 392
Horkheimer, Max, 280
Horn, Laurence R., 134
Hiillen, Werner, 4
Husserl, Edmund, 436
Hie, Cornelia, 273
Irwin, James R, 231
Itkonen, Esa, 140,227
Jackendoff, Ray, 404, 408
Jackson, Stanley W., 239, 254
Jake!. Olaf, 252, 307, 367
Jakobson, Roman, 220
Janssens, Guy, 327
Jensen de L6pez. Kristina, 227
Johnson,Mark, 205.248, 258, 309,
352,374,378,405,413
Joseph,John, 274,284
Kail, Aubrey, 235
Katz, Jerrold 1.,4, 7, 75, 386, 390.
391,401,402,403,404,406,413
Kay, Paul, 9,13.133,177,198,374,
410
Kemmer, Suzanne, 363
Kempson, Ruth, Ill, 125, 150
Kerkhofs, Jan, 269
Kilgarriff, Adam, 359
Index ofnames 481
Kleiber, Georges, 25, 80, 100, 102,
104, 136, 141,262
Klibansky, Raymond, 231, 235, 239
Knox, Israel, 310
Koch, Peter, 413
Kovecses, Zoltan, 227. 230, 240,
241,242,243,244.245,246,
247,248,249,250,251
Kristiansen, Gitte, 228
Kronasser, Heinz, 381
Kroskrity, Paul, 274
Kuhn, Thomas, 44, 140, 375, 421
Labov, Williall1, 19,27,376
Lakatos, Ill1re, 140, 375, 421
Lakoff, George, 4, 10, 13, 14, 17,
18,23,24,25,26,42,49,50,74,
81, 101, 106, 121, 130, 133, 138,
178,191,205,221,240,248,
258,273,274,309,333,334,
352,374,378,380,381,391,
395,405,413,421
Landheer, Ronald, 150
Landre, George, 38
Langacker, Ronald W., 4,8,46,51,
74.76.80, 135, 152, 176, 188,
262,309,315,374,379,380,
391.395,405
Lara, Luis Fernando, 356
Leacock, Claudia, 99
Lehmann, Winfred P., 4, 23
Lehrer, Adrienne, 255
Levenson, Robert W., 242
Lewis, David, 392
Leys, Odo, 72, 73
Lindeboom, Gerrit A., 231
Lindner, Susan 1., 374
Lipka, Leonhard, 388
Lounsbury, Floyd, 386,401
Lutzeier, Peter R., 367
Lyons, John, 387, 390,401.403
Maalej, Zouhair, 227
Machado de Assis, Joaquim, 417
482 Index ofnames
MacLaury, Robert E., 138,227
Major, Ralph H., 231
Mangasser-Wahl, Martina, 349, 405
Martin, Laura, 95, 272, 438
Marty, Anton, 380
Mbense, Thandi G., 227
McCawley, James D., 145,391
McConnell-Ginet, Sally, 221
Medin, Doug L., 4, 392
Melc;;uk, Igor, 401
Melis, Ludo, 135
Mervis, Carolyn B., 3, 4,9, 10, 11,
28, 116, 374, 376
Meyer, Michael, 274
Michelson, Albert, 419, 420, 432,
442
Mikolajczuk, Agnieszka, 227
Miller, Scott R, 4
Minsky, Marvin, 43
Moder, Carol L 7
Montague, David, 386, 392
Moon, Rosamund, 363, 402
Morley, Edward, 419, 420, 432, 442
Nagel, Richard, 386
Nathan, Geoffrey S., 7
Neisser. Ulrich, 4
Nerlich, Brigitte, 99, 367, 398, 400
Newmeyer, Frederick 1.,391
Norrick, Neal R, 105, 109, 122
Nunberg, Geoffrey, 115, 121, 124,
216
Nyrop, Kristoffer, 368, 400
Olivier, D.e., 13
Ortony, Andrew, 249
bstman, Jan-Ola, 175
Pagels, Elaine, 322
Palmer, Gary, 227, 273, 347
Panofsky, Erwin, 231, 235. 239
Panther, Klaus-Uwe, 405
Paul, Hermann, 368, 381, 400
Peeters, Bert, 99, 356
Piaget, Jean, 43
Pinkal, Manfred, 142
Plato, 22,25,416, 417,430, 439
Pluim, Teunis, 38
Pope, Maurice, 235, 236, 245
Popova, Yana, 307
Porzig, Waiter, 388, 390. 401
Posner, Michael, 9
Pottier, Bernard, 386. 390, 401
Pulman, Steve, 41
Pustejovsky, James, 404, 406, 408,
413
Putnam, Hilary, 20, 21. 22, 25, 357,
421,426,427,443
Quadri, Bruno, 78
Quine, Willard V.O., 14, 106, 108,
350,428
Quinn, Naomi, 4, 273
Radden, Giinther, 405
Ravin, Yael, 99
Reddy, Michael, 137, 183, 273
Regier, Terry, 416
Reichling, Anton, 23, 397
Rice, Sally, 99, 416
Ricoeur, Paul, 140
Ries, John, 307
Ripa, Cesare, 234, 235, 243
Rips, Lance 1., 392
Robertson, David, 310, 313, 314,
315, 322
Rosch, Eleanor, 3, 4, 9, 10, 11, 13,
16, 18,22,27,28,29,42, 102,
116, 117,374,376,379,405
Rosier, Irene, 144
Ross, John R, 7
Rudzka-Ostyn, Brygida, 4, 27, 45,
152, 178, 193, 194, 195, 196,
198, 367, 374
RuhL Charles, 150, 152, 168, 169,
171
Ryder, Mary E., 198
Sadock, Jerry, 14, 101, 106,337,
350
Sandikcioglu, Esra, 273
Sandra, Dominiek, 99,416
Sapir, Edward, 287
Saxl, Fritz, 231, 235, 239
Schafer,Jtirgen, 231, 233, 235, 236
Schieffelin, Bambi, 274
Schiffman, Harold, 274
Schleiennacher, Friedrich, 140
Schmid, Hans-Jarg, 152,309,347.
405
Schmidt, Wilhelm, 146
Schulze. Rainer, 4, 152
Seiler, Hansjakob, 24
Shakespeare, William, 235, 236
Shoben, Edward 1.,392
Siegel, Rudolp Eo, 230, 231
Simpson, Carol, 4
Sinclair, John, 402
Sinha, Chris, 150, 227
Skutnabb-Kangas. Tove, 283, 284,
294
Smith, Edward Eo, 4, 291, 392
Soriano, Cristina, 227
Soufas, Teresa Scott, 239
Speelman, Dirk, 272, 273, 274, 299
Sperber. Hans, 368, 372
Starobinsky, Jean, 239
Stefanowitsch, Anatol. 345
Stem, Gustav, 368. 400
Stevenson, Mark. 100
Stock, Penelope, 359
Swanepoel, Piet, 362
Sweetser, Eve Eo, 133, 198,374,383
Swiggers, Pierre, 367, 396
Tabakowska, Elzbieta, 307
Talmy, Leonard, 76, 374, 395
Taylor, John, 4, 48, 74, 80, 89, 99,
101,227,309,347,367,405,408
Taylor, Talbot, 274
Thomason, Richmond, 386, 403
Tissot, Samuel, 228, 229
Tomasello, Michael, 227
Index a/names 483
Traugott, Elisabeth c., 188,248
Trier, Jost, 379, 386, 400, 401
Tuggy, David, 99, 145,351
Tummers, Jose, 272, 345
Turner, Mark, 307
Tyler, Andrea, 48
Uhlenbeck, Eugenius Mo, 7
Ungerer, Friedrich, 309, 347, 405
Van Belle, William, 175, 177
Van der Meer, Geart, 362
Van Dijk, Teun A., 274
Van Gijsel, Sofie, 272
Van Ginneken, Jac, 368, 383
Van Langendonck, Willy, 8, 175,
177.252
Van Oosten, Jeanne, 7
Vandeloise, Claude, 48, 49, 80, 133,
150, 374
Verdoodt, Albert, 270
Verlooy, Jan B.c.. 290
Verspoor, Mmjolijn, 347
Violi, Patrizia, 347,405
Virchow, Rudolph, 234
Vos, Louis, 277, 289
Vossler, Karl, 77
Voye. Liliane, 269
Wagner, Doris, 252
Waismann, Friedrich, 13
Weiland, Petrus, 38
Weinreich, Uriel, 7
Weisgerber, Leo, 147.379, 390,
396,397,400.404
Wellander, Erich. 368
Whedbee, William, 310, 313, 314,
315,322
Wierzbicka. Anna, 4, 13, 18, 19,22,
24,25.95,117,118.119,130,
131. 144, 145, 150, 152, 169,
170, 171,227,327,328,329,
330.405,416,423,425.430,441
Winters, Margaret, 8. 135
Wittgenstein, Ludwig, 9, 24
484 Index ofnames
Wodak, Ruth, 274
Woolard, Kathryn, 274
Wundt, Wilhelm, 368, 379
Yu, Ning, 227
Zgusta, Ladislav, 334
Zlatev, Jordan, 99
Zwicky, Arnold, 14, 101, 106,337,
350
Index of subjects
alloseme, 134
ambiguity, 14,51,53,58,78,106,
111,121,126,128,130,132,
134, 142, 144,233, 337, 338,
339,351
analogy, 33, 38, 61, 133, 134, 257,
307, 317, 375,418
anaphora, 106, 121, 122, 123, 124,
222, 350
anthropology, 3, 23, 74, 377, 386,
401,436
antonymy, 107, 114, 144,209,385,
387,388,401,403,410,411
applied linguistics, 328
artificial intelligence, 404
basic level hypothesis, 84, 85, 255,
258,261,406,410,411,412,
413,414
categorization, 5, 8, 9, 10, 16, 18,
21,22,27,44,96,101,102,103,
104,113,147,176,309,327,
331,333,334,343,374,375,
380,385,387,404,410,426
classical definability, 12, 14, 17,
102, 149, 150, 153, 154, 155,
156,157,160,164, 166, 167,
168, 171,341
cluster, 8, 18, 70, 142, 252, 330,
352, 357, 383, 384
cognitive linguistics, 8, 20, 21, 22,
24,27,48,99,100,103,138,
152, 175, 176, 191, 194, 198,
227,252,253,257,258,270,
272, 273, 274, 288, 307, 309,
345,367,405,416,421,422,
427,431,435,436,438,439,
440,441
cognitive science, 6, 8, 24, 43, 44,
103,405,416
cognitive semantics, 18, 23, 24, 25,
26,42,44,47,48,49,72,74,75,
76,80,81,141, 178,221,251,
309, 345, 346, 347, 352, 353,
354, 356, 357, 362, 363, 367,
374, 375, 377, 379, 380, 382,
383,384,385,387,391,392,
393,395,398,399,401,404,
405,406,407,408,411,412,
413,414,415,416,417,441,442
collocation, 30, 33, 40, 248, 345,
363,402,414,415
colloquial language, 305
componential analysis, 4, 6, 7, 23,
151,153, 178, 181,232,386,
387,388,390,392,401,402,
403,413,433
composite expression, 198, 199,
210,211,212,214,216,218,
220,223
compositionality, 198,200,201,
203,207,211,212,215,220,
223, 387
compound, 44, 49, 95,159,198,
200,209,212,215,217,220,
222,223,355,358,359
conceptual metaphor, 252, 352, 362,
413,414
connotational meaning, 45, 46, 200
construction, 36, 55, 56, 71, 77, 84,
106,109,114,120,121,122,
124,125,126,127,129,171,
173, 175, 176, 177, 178, 179,
180. 182, 183, 185, 186, 187,
189,190, 191, 192, 193, 194,
486 Index ofsubjects
195,196,197,223,313,350,
372,422,425
construction grammar, 175, 177
convergence, 295, 297, 299, 301,
302,303,304,351,398,414
core meaning, 139, 153,341, 342,
343
corpus linguistics, 30, 34, 40, 48,
85,100,149,151, 152, 155, 162,
170,175, 179, 193, 194,272,
334,335,345,363,402,410,
415,416,417,423,424,431,
434,438
critical discourse analysis, 274
cue validity, 28, 86, 90, 92
culture, 30, 94, 95,204, 225, 227,
228,230,246,248,250,251,
253,254,258,263,265,266,
268,269,270,272,273,274,
275,276,281,283,284,287,
289,290,291,292,293,294,
295,296,297,298,299,302,
305,307,310,349,373,377,
378,380,408,436,438,440
dative, 175, 178, 184, 188, 191, 193,
195, 196
definiendum, 113, 130, 155
definiens, 130
definition, 3, 4, 5, 8, 10, 11, 12, 13,
14,16,17,18,19,20,21,39,42,
49,51, 71, 72, 76, 78, 80, 85, 99,
100,101,102,106,107,112,
113, 114, 115, 117, 120, 123,
125, 129, 130, 131, 132, 133,
141, 143, 144, 145, 148, 149,
150,153, 154, 155, 156, 157,
158, 160, 164, 165, 166, 167,
168,169,170,171,172,177,
178,180,181,195,196,197,
223,250,262,297,309,327,
328, 329, 330, 337, 339, 341,
343,345,346,349,350,351,
354, 355, 356, 358, 359, 360,
361, 362, 374, 375, 376, 378,
389,390,401,403,423,424,
425,426,427,428,439
denotationa1 meaning, 10, 11, 13,
16,83, 88, 149, 152, 168, 169
diachronic semantics, 43, 103, 179,
218,251,273,352,369,370,
383,384,396,400,406,407,
409,411
diachrony, 3, 5, 135, 179, 184,218,
222,230,239,244,250,251,
263,273,301,302,327,334,
345, 352, 370, 377, 382, 384,
385,391,395,396,397,400,
407,409,411
dialect, 237, 266, 268, 274, 276,
277,278,281,301,396
Dialektik der Aufklarung, 280, 292,
296
dictionary, 20, 23, 25, 29, 38, 39,
126, 130, 309, 327, 328, 329,
330,331,332,333,334,340,
341,343,345,346,354,355,
356, 357, 358, 359, 361, 362,
370,402
discreteness, 13, 19,22,375
distinctiveness, 5, 76, 88, 89, 90,
119, 147, 149, 153, 154, 156,
157, 158, 160, 161, 165, 168,
175,252,287,346,349,354,
359,386,388,390,409,413
ditransitive construction, 177, 178,
194
divergence, 116,299,301,302,351
division of linguistic labor, 20, 21,
26,320,347,357,426,427
emancipation, 276, 284, 285, 299,
304
empirical method, 7, 10, 74, 95,
171,178,233,234,247,262,
273,300,306,334,344,368,
370,371,379,387,389,392,
395,398,399,408,412,414,
416,432,440,441,443
empiricism, 22, 416, 441, 442, 443
encyclopedic meaning, 5,6, 10, 23,
42,47, 143, 146,205,264, 329,
346,354,356,357,358,371,
378,379,387,390,391,397,
408,436
Enlightenment, 276, 280, 282, 283,
284,288,292,296
entrenchment, 90, 92, 95,262,411,
414
epistemology, 10, 18, 22, 25, 44,
286,317,333,367,405,417,
421,422,423,430,436,440,442
etymology, 37, 44, 46, 209, 210,
237,238,268,332,370,379,
383, 396
evidence, 35, 37, 109, 115, 122,
124,131,137,154,179,184,
210,222,238,242,243,248,
249,253,270,303,308,343,
370,392,404,423,438
expert knowledge, 20, 21,95,356,
357,427
expressivity, 43, 45, 46, 47, 282,
285,295,296,372,373,374,387
extension, 3, 5, 1L 12, 13,20,21,
35,45,52,58,60,61,62,69,71,
81,87,106,112,117,132,135,
139, 147, 150, 170, 180,181,
182, 184, 187, 188, 189, 191,
194, 195, 197, 198,201,207,
208,213,215,218,343,347,
348,349,351,357,375,376,
411,412,414,424,427
falsificationalism, 140
family resemblance, 9, 11, 14, 15,
28,29,40,41,80, 102,376,384
felicity conditions, 315, 316, 318,
321
Index ofsubjects 487
figurative meaning, 15, 35, 49, 128,
132,198, 199,202,203,204,
206,207,208,209,211,221,
222,245,257,347,352,362,383
figure / ground, 29, 30, 37. 46, 54,
76,95,117,118.119,235,250,
292,299,320,327,349,395,423
flexibility, 3, 5, 7, 10, 12, 17,19,20,
24,42,43,44,45,46,72,74,99,
132,137,251,310,333,346,
377,382,384,387,404,408,
427,432
formal semantics, 6, 439, 440, 441,
443
frame semantics, 6, 24, 43, 80, 82,
109, 110, 111, 120, 125, 127,
129,177, 179, 196,294,362,
405,413
frequency, 29,40,41, 75, 85, 89,
135,158,163,164. 179, 182,
193,208,263,264,299,300.
328, 348
function, 27, 29, 33, 39, 41, 42, 43,
44,45.47, 58, 82, 87, 108, 109,
115. 125, 130, 133, 134, 154,
176,179, 183, 184, 185, 191,
194, 195,243,257,266,275,
276,278,279,280,281,294,
295,296,297,304,319,320,
336,352,360,371,377,378,
384,393,406,411,413,422,
441,443
fuzziness, 7,13.16,22, 130,354
Geisteswissenschaft, 140, 373
generality, 113, 114, 118, 145, 149,
153, 168,241,242,275,276,
277,279,280,281,349,425
generative grammar, 6, 23, 389,
390,391,392,401,402,403,404
globalization, 293, 297, 306
gradience, 27, 41
henneneutics, 140, 308, 373, 380,
396,434,435,436,437,438,
442,443
historical linguistics, 8, 370
historical semantics, 43, 103, 179,
218,251,273,352,369,370,
383,384,396,400,406,407,
409,411
historical-philological semantics.
333,368,371,372,373,374,
375,377.378,379,380,381,
382, 384, 385, 386, 387, 390,
391. 393, 395
homonymy, 122, 130, 177,208,
384, 389
humoral theory, 227, 229, 230, 231,
232,233,234,235,236,237,
238,239,240,241,242,243,
244,245,246,247,248,249,250
hyperonymy, 85, 86, 114, 155, 158,
159, 165
hyponymy, 161, 163, 164, 165,387,
388,401. 403
idealism, 22, 416, 417, 421, 430,
438,441,442,443
identity and language, 38, 50, 106,
114, 123, 124,177,215.252,
266,268,269.275,282,283,
289,290,291.292,295,296,
297, 298, 304, 337, 341, 343,
350,376,377
ideology, 272, 274, 275, 276, 280,
281,282,285,288,294
idiom, 173, 198, 199,200,201,203,
205,206,207,208,209,210,
211,212,217,222,248
indeterminacy, 13, 19,50.105. 116,
142, 343, 382
indirect object, 175, 176, 177, 178,
179,180,181, 182, 183, 184,
185,186, 187, 188, 189, 190,
191,192, 193, 194, 195, 196, 197
informal language, 301, 302, 303,
304, 305
intension, 3, 11, 12, 347, 348, 375,
386.392,424
interdisciplinarity. 6, 8, 103, 104,
416,436
interpretation, 43, 52, 55, 58, 61,
73, 79, 87, 103, 105, 109, 110,
Ill, 113, 114, 115, 120, 121,
123, 124, 127, 129, 132, 133,
134, 135, 138, 139. 140, 141,
142, 143, 145, 146, 148, 150,
154,170, 199,200,201,203,
205,207,208,209,210,211.
218,219,220,221,222,227,
234,238,239,241,243,244,
245,248,273,274,288,293,
297,307.308,310,313,314,
315,322,323,339,346.348,
373, 374, 377, 380, 384, 385,
396,401,415,420,422,433,
435,436,437,438,439,440,
442,443
intracategorial polysemy, 101, 102,
104,136, 141, 142,262
introspection, 19,25,34,37,38,40,
109, 110, 152, 169, 170, 389,
417,423,428,429,430,432.
434,442
irony, 307, 308, 309, 310, 311, 312,
313,314,315,320,321,323,
324.355,423,432
isomorphism. 199,201,202,203,
206,207,208,209,211,212,
214,215,221,222
landmark, 53, 54, 56, 57, 58, 59, 60.
63,65,67,69,70,72,73,80,81,
82,83, 195
lexical field, 49, 78, 88, 147, 254,
257,261,262,264,270,390,
401,403,406,409,413,414
lexicography, 8, 18, 309, 325, 327,
328,329,330,331,332,333,
334, 343, 345, 346, 347. 354,
359, 361, 362, 363, 370, 376,
379,389,402,429
lexicology, 74, 76, 77, 80, 90, 91,
94,95,99, 194,244,271, 272,
309,346,353,359,398,409.
411,416
lexicon, 7, 13,23, 74, 79, 88, 90,
92, 95. 104, 124, 129, 133, 135,
137,167,175, 176,208,252,
377,387,398,404,406,409,
410,415
literal meaning, 14, 112, 121, 128,
132, 184, 199,200,201, 203,
206,207,209,211,212,213,
215,217,219,221,245,246,
257,270,385
logic, 3,19,65.94,99,107,110,
113, 114, 115, 116, 117, 119,
120, 124, 125, 126, 127, 129,
131,142,143,144,146,274,
275,288,292,294,295,298,
303,306.316,321,347,381.
386.387.388,391,393,403,
404,435,440,441,443
logical semantics. 386, 387, 393,
403,441
macrostructure, lexicographical-,
328,421
markedness, 8, 46, 66, 161, 184,
233,244,249,264
membership, 3, 9, 12, 13, 16, 100,
102, 112, 116, 117, 120, 123,
142,375,376,377,443
metaphor, 5, 7, 24, 45, 46. 62. 102,
103,137,138,147,176,179,
180,182,183,191, 194, 195,
198,203,205,206,207,209,
211,212,213,214,215,216,
217,218,219,220,227.240,
241,242.243,245,246,247,
248,250,251,252,257,270,
352,375,378,381,384,385,
405,411,413
metaphtonymy, 218
methodology, 5, 6, 7, 8, 10, 20, 22,
25,27,37,48,49,50,71,88,
100,101, 136, 138, 139, 140,
141, 142, 149, 153, 156, 158,
160,168,169,171, 172, 176,
177,193,230,238,239,241,
242,246,247,248,249,250,
254,256,258,262,269,270,
273, 308, 309, 324, 328, 333.
344,363,369,370,373,374,
380, 385, 386, 387, 388, 389,
390,392,393,395,396,397,
400,401,403,404,406,408,
416,418,419,421,423,428,
429,430,431,432,433,435,
436,437,438,441,442
metonymy, 7, 17,65, 103, 176, 179,
184,186, 187, 194, 195, 198,
199,211.212,213,214,216,
217.218,220,241,245,251,
343.381,384,400,405,411
microstructure, lexicographical-,
328,330,421
monosemy, 14, 18, 20, 48, 49, 127,
130, 132, 136, 149, 150, 168,
172,338
morphology, 7,159,160, 177, 184,
220,221
motivation, 104, 199,201,202,203,
204,205,206,207,210,211,
221,222,223,238,239,240,
241,242,243,244,245,248,
250,304.431
multidimensional structure, 48, 69,
73, 175, 176, 178, 179, 192, 196,
197,272,327,331,351,352,
354, 361
multilingualism, 294, 295, 296, 297
naming, 78, 83, 84, 85, 87, 96, 185,
252,300,349,406,409,427
nationalism, 272, 275, 287, 288,
289,290,291,292,293,294,
295,296,297,298,304,306
native speaker, 14, 38, 40, 72, 329,
330, 389, 390
Naturwissenschaft, 140, 373
near-synonymy, 50
necessary and sufficient criteria, 9,
11,14,16,19,71,100,101,112,
117,119,149,333,337,359,390
neogenerativist semantics, 398, 399,
402,403,404,406,407,408,
413,414,415
neostructuralist semantics, 5, 393,
398,399,400,401,402,406,408
network, 48, 80, 92, 99, 175, 178,
191,193,195,196,409
neuroscience, 8
non-equality, 12
non-referential meaning, 409
non-rigidity, 11, 12
norms oflanguage, 274, 300, 301,
304,440
objectivism, 10, 22, 24, 26, 138,
140, 141,380,393,395,421,
422,423,437,438,443
onomasiology, 27, 44, 47, 66, 74,
75, 77, 78, 79, 80, 83, 84, 85, 86,
87,88,90,91,93,94,95, 148,
158, 159, 163, 194,261,272,
274,299,300,398,400,408,
409,410,411,412,413,414,415
paradigmatic relations, 33, 38, 44,
80,90,91,92, 140, 198, 199,
201,203,206,207,210,220,
221,375,387,388,403,421
perception, 40, 96, 395
perspectival salience, 57, 60, 63, 75,
76,77,90,91, 138, 140, 189,
190,191
philology, 48, 308, 309, 333, 368,
369,371,372,373,374,375,
377, 378, 379, 380, 381, 382,
383, 384, 385, 386, 387, 390,
391,393,395,396,435,437
philosophy, 3, 8, 9,22, 24,26,44,
138, 140,276,285, 287, 296,
333,367,368,389,399,403,
416,417,421,435,436,442,443
phonology, 7, 29, 83, 133, 134, 175,
177,228,384,388
physiology, 28, 40, 41, 42, 44, 228,
229,231,235,236,238,239,
240,241,242,244,245,246,
248,249,250,268,310,369,440
politeness, 310, 317, 318
polysemy, 3, 7,14, 18, 19,23,25,
48,49,50,51,67,73,80,87,91,
97,99, 100, 101, 102, 103, 104,
105, 106, 107, 108, 109, 110,
Ill, 112, 113, 114, 115, 116,
117, 119, 120, 121, 122, 123,
124, 125, 126, 127, 128, 129,
130, 131, 132, 133, 134, 135,
136, 137, 138, 140, 141, 142,
143, 144, 145, 146, 147, 149,
150, 153,168, 169, 172, 178,
193, 197,218,251, 327, 328,
330,331,333,339,345,346,
347,350,351,358,359,362,
363,376,380,382,383,408,
410,413,414,415,428
postmodernism, 272,275,276, 288,
293,295,296,297,298,304,
305,307,308,438
pragmatics, 75, 78, 90, 104, 134,
140,150,188,310,315,316,
317,318,329,330,331,391,
409,414,415
preposition, 21, 48, 49, 81, 84. 178
prestructuralist semantics, 23, 43,
103, 179,333,354,367.383,
395,396,398,399,400,407,
411,412,414
prismatic semantics, 199, 200, 206,
211,213,214,215,216,218,
219,220,222,223
proper name, 73, 216, 252, 255,
270, 356
prototype theory, 1, 3, 4, 5, 6, 7, 8.
9. 10, 11, 12, 13, 14, 15, 16, 17,
18, 1 ~ 21,22,23,24,25,27,28,
29.34,35,36,37,38,39,40.41,
42,43,44,46,47,48,49,65,67,
70,71,72,74,75,77,80,87,90,
91, 100, 101, 102, 103, 104, 116,
120, 125, 130, 135. 136, 138,
139, 140, 141, 142, 144, 152,
170, 171, 175. 176, 178, 179,
180, 181, 183, 184, 189, 190,
191,193.194,196,231,251,
255,258,274,310,327,328,
329,330,331,333,334,335,
337, 339, 340, 341, 343, 345.
346, 349, 352, 354, 355, 356,
357,358,360,362,374.375.
377, 379, 381. 382. 383, 384,
385,387,395,397.398,405,
411,412,413,414,421,427
psycholinguistics, 4, 8, 24, 65, 67,
74,99, 100, 102.210,248,399,
404,405,416,431
psychology, 3, 4, 6, 8, 9, 10, 13,23,
28,29,36,41,42,43,44,71,72,
103,104,110,138,179,217,
222,227,230,231,234,235,
236,239,245,250,348,368,
371,372,377,378,379,380,
382, 389, 390, 392, 393, 395,
396,397,400,401,403,404,
407,416,427,430,435,441
purism, 292
quantitative method, 300, 302
radial network, 9, 14,20,24, 25, 48,
49,101,102,175,178,191,195,
196,346,352,354
rationalism, 272, 273, 275, 280,
282,283,284,285.288,289,
290,291.292,293,294,295,
296,297,298,304,305
referential meaning, 11,28,40,41,
42,44,50,51.53,66,72,80,86,
87,89,91, 102, 103, 104, 105,
108, 124, 137, 147, 148. 150,
152, 154, 155, 156, 159, 161,
162. 164, 170, 188, 258. 263,
264, 270, 339, 348, 349, 352,
353,354,409,413,423,424,
426,433
register, 301, 302, 303, 304
reinterpretation, 124, 134, 154, 157,
161,208,209,211,222,244,
245, 248
relational semantics, 48, 53, 80,
144,178. 196, 197,201,401,
403,409
representativity, 9, 12, 13, 14, 16,
17, 19, 30, 114, 252, 255, 436
romanticism, 188, 272, 273, 275.
280,282,283,284,285,288,
289,291,292,293,294,295,
296,297,298,299.304,305,
437,438,440,443
rule/list fallacy, 135
salience, 1,3, 19,20,25,27,28,35,
38,39,40,42,49,65,71,74,75,
76, 77, 79, 82, 83, 84, 85, 86, 87,
88,89,90,91,92,93,94,95,96,
101,194,207,231,252,254,
255,257,258,260,261,262,
263,264,266,270,330.331,
348,349,376,406,410,411,426
scenes-and-frames.405
semantic feature, 5, 75, 88. 90. 252,
392
semasiology, 5,44,47,66, 74, 75,
77,78,79,80,86,87,90,91,92,
93, 148, 154, 156, 158, 159, 160,
162,178, 179, 194,274,347,
348, 349, 352, 354, 368. 398,
400,408,409,410.411,412,
413,414,433
semiotics. 79, 90, 234, 235
signifiant, 77
signifie, 77, 103
snow, 94, 95
social aspects of meaning, 21, 23,
30, 35, 36, 83, 94, 95, 130, 204.
227,230,246,250,251,265,
267,270,272,273,274,275,
276,278,279,281,282,283,
284,285,287,289,290,291,
292,293.294,295,296,297,
298,299.302,305,307,310.
323, 349. 355, 368, 372, 373,
377,378.380,408,426,436,
438,440
sociolinguistics, 20, 21, 22, 23, 45,
46,83.90,228
speech act. 20, 315, 316, 318, 323
Staatsnationalismus, 292. 297
standard language. 275, 276, 277,
279,280,281,282,283,285,
293,296,298,299,301,303,304
standardization, 272, 273, 275, 276,
279,280,281,282,283,284,
285,287,288,289,293,296,
298,299,301,302,303,304,
305, 306
statistics, 28, 40, 41, 42. 44, 94, 99.
139, 162, 163
stereotype, 20, 25, 328, 357, 358,
426
stratification of language, 83, 299,
301,302
structural salience, 75, 79, 88, 90,
91,94,252
structuralist semantics, 5, 23, 43,
77, 103, 329, 344, 353, 354, 356,
362,367,374,377.379,382,
386, 387, 388, 389. 390, 391,
392,393,397,398,399,400,
401,402,403,404,406,408,
411,412,413,414,415
subjectification, 51, 176. 179, 188,
195
subordination. 323
synecdoche,65,384
synonymy, 27, 29, 38, 39, 45, 47,
50, 83, 84, 86, 90, 91, 92. 107,
143, 144, 155, 156, 158, 159,
162, 163, 167, 195,209,237,
299,351,354,359,387,388,
389,401,403,409
syntagmatic relations, 30. 33, 34,
39,52,56,60,61.64,65,70.76,
80,81,82,90,91,93,196, 198.
199,200,201,207,209,210,
220,221,222,387,388,390,
401,402,403,413,414
syntax, 6, 7, 109, 193,390
trajector, 53, 54, 56, 57. 59, 60, 62.
63,65,69,70,72,73,80,81.82,
83
transfonnational grammar, 4, 6,
106, 350, 385, 386. 388, 389,
390,391,392,393
transparency. 20, 201, 204, 207,
208,209,211,230,240,245,
248,349,423,426,430,439
truth conditions, 108, 109, 110, 134,
143,387,393,403,428
turn-taking, 317
universality, 94, 96, 227, 289,437
univocality. 14, 18,20,48.49, 127,
130, 132, 136, 149. 150. 168,
172,338
usage, 18,20,21, 30, 35, 36. 37. 38,
39,40,41,44,46,50,62.64,66,
87, 115, 149, 150, 152, 168, 169,
171, 172,208,273,278,289,
331,377,398,414,415,427,
429,431
vagueness, 12,28, 50, 72, 80, 87,
99, 100, 101, 104, 105, 106, 107,
109, Ill, 116. 120, 121, 123,
125,129,133, 134, 136, 137,
138,141,142, 144, 146, 147,
193,257,338,339,350,369,
375, 376, 381, 382
vantage theory, 138
variation, 27, 45, 46, 74, 83, 84, 86,
133,138, 150, 154, 157, 175,
205,228,256,272,273,274,
275,282,283,285,287,288,
289,292,293,298,299,305,
341, 375
vocabulary, 30, 95, 227, 230, 237,
239,241,242,244,245,246,
247,249,250,254,257,331,
392,402,439,443
Volksnationalismus, 292, 297
Woordenboek der Nederlandsche
Taa1, 29, 327, 332, 334, 335,
340,341,342,344,361,371

Iif - KGPM - Geeraerts D. Words and Other Wonders

Cargado por

Información del documento

Descripción original:

Título original

Derechos de autor

Formatos disponibles

Compartir este documento

Compartir o incrustar documentos

Opciones para compartir

¿Le pareció útil este documento?

¿Este contenido es inapropiado?

Copyright:

Formatos disponibles

Iif - KGPM - Geeraerts D. Words and Other Wonders

Cargado por

Copyright:

Formatos disponibles

Words and Other Wonders

Cognitive Linguistics Research

También podría gustarte