A Comparative Grammar of Brtiish English Dialects

A Comparative Grammar of British English Dialects
Topics in English Linguistics

50.1
Editors
Elizabeth Closs Traugott
Bernd Kortmann
Mouton de Gruyter
Berlin New York
A Comparative Grammar
of British English Dialects
Agreement, Gender, Relative Clauses
by
Bernd Kortmann
Tanja Herrmann
Lukas Pietsch
Susanne Wagner
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
A comparative grammar of British English dialects : agreement, gen-
der, relative clauses / by Bernd Kortmann [et al.].
p. cm. (Topics in English linguistics ; 50.1)
Includes bibliographical references and index.
Contents: The Freiburg English Dialect Project and corpus / Bernd
Kortmann, Susanne Wagner Relative clauses in English dialects of
the British Isles / Tanja Herrmann Some do and some doesnt :
verbal concord variation in the north of the British Isles / Lukas
Pietsch Gender in English pronouns : southwest England / Su-
sanne Wagner
ISBN 3-11-018299-8 (hardcover : alk. paper)
1. English language Dialects Great Britain. 2. English lan-
guage Great Britain Grammar. 3. English language Relative
clauses. 4. English language Agreement. 5. English language
Gender. I. Kortmann, Bernd, 1960 II. Series.
PE1721.C66 2005
427dc22
2005001607
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 3-11-018299-8
Copyright 2005 by Walter de Gruyter GmbH & Co. KG, 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 mecha-
nical, including photocopy, recording, or any information storage and retrieval system, with-
out permission in writing from the publisher.
Cover design: Christopher Schneider, Berlin.
Printed in Germany.
Preface
Bernd Kortmann
Since the 1980s, but especially over the last ten years or so, the study of the
grammar of English dialects has been very much on the rise after more than
a century of neglect in English dialectology and dialectology, in general.
Witness, in particular, Trudgill and Chambers (1991), Milroy and Milroy
(1993), and, on a global scale, Kortmann and Schneider (2004). Apart from
these and several other publications related in spirit, however, the vast
majority of publications on the grammar of English dialects concentrates on
just one particular phenomenon in one particular dialect or dialect area, is
based on a very small database and purely descriptive. Moreover, the small
size of the available databases often makes it very difficult to formulate
valid descriptive generalizations. Virtually non-existent in English
dialectology are systematic comparative studies of individual grammatical
subsystems across a selection of dialects (like comparative studies of the
tense and aspect systems, pronominal systems, relativization or
complementation patterns, etc.). Exceptions in this respect form the
sociolinguistic studies by Tagliamonte and her research team (e.g.
Tagliamonte 1999, 2002, 2003), and the contributions, especially the
regional and global synopses, in Kortmann and Schneider (2004).
However, useful as the synopses are in providing general orientation, they
can be no more than very useful starting-points for systematic comparative
analyses of individual phenomena of dialect grammar.
The present volume, the first in a series of volumes which will be
published at irregular intervals, tries to set an example as to how this gap in
English dialectology can be filled. Secondly, it will do away with another
problem that has beset the study of English dialect syntax for many
decades, namely the lack of a sufficient amount of reliable data. The Survey
of English Dialects, for example, compiled in the 1950s and serving as the
most important data source for English dialectologists and dialect
geographers ever since, was simply not geared to the systematic collection
of data on grammar. J ust a fraction of the more than 1300 questions in the
SED questionnaire was explicitly designed to collect morphological and
syntactic information. Only since the late 1980s have efforts been made at
compiling large data collections, such as the Survey of British Dialect
vi Bernd Kortmann
Grammar (Cheshire, Edwards and Whittle 1989), the Newcastle Electronic
Corpus of Tyneside English (NECTE; see Allen et al., forthcoming), and,
largest of all, the computerized Freiburg English Dialect Corpus. It is the
latter, FRED for short, which will take centre stage in this volume. Thirdly,
all studies in this volume are informed by a typological approach to English
dialect grammar (apart from the fact that two of the dialect phenomena
investigated here, namely the Northern Subject Rule and pronominal
gender, are typologically very rare). This approach is the hallmark of the
Freiburg research group on English dialect syntax, initiated and coordinated
by Bernd Kortmann, and will be outlined in the scene-setting paper by
Kortmann and Wagner. It is in this paper, too, that the nature and design of
FRED, and its advantages for both qualitative and quantitative analyses of
dialect phenomena will be discussed in some detail.
The subject matter of the three studies forming the backbone of this
volume can briefly be characterized as follows: Tanja Herrmann examines
(adnominal) relative clauses in six dialect areas of the British Isles (Central
Midlands, Central North, Central Southwest, East Anglia, Northern Ireland,
Scotland). The results of this cross-dialectal study she relates to typological
hierarchies, particularly to Keenan and Comries (1977) Noun Phrase
Accessibility Hierarchy. The Accessibility Hierarchy is largely verified for
all relative clause formation strategies found in the data, including the zero
relative marker strategy (as in The man ___ called me was our neighbour).
From a diachronic perspective, the Accessibility Hierarchy also helps to
reveal the pattern underlying the way individual relative markers (e.g. the
relative particles as and what) enter or exit an existent relative marker
system.
Lukas Pietsch investigates, synchronically as well as diachronically, the
so-called Northern Subject Rule (NSR), a feature found in the Northern
dialects of England, but also in Scotland, Northern Ireland and the Republic
of Ireland. This rule is concerned with subject-verb agreement, and can
roughly be formulated as follows: every verb in the present tense can take
an s-ending unless its subject is an immediately adjacent simple pronoun.
(Third person singular verbs always take the s-ending, as in Standard
English). In other words, the NSR involves a type-of-subject constraint
(pronoun vs. common/proper noun) and a position constraint (+/-
immediate adjacency of pronominal subject to verb). Thus, in NSR-
varieties we get the following examples: I sing vs. *I sings, Birds sings, and
I sing and dances.
Preface vii
Susanne Wagner, finally, provides a comprehensive account of a special
semantic system of (pronominal) gender marking, which is distinctive of
the traditional dialects in Southwest England. What we encounter in
Somerset, in particular, is pronominal gender that is primarily sensitive to
the mass/count distinction and only secondarily to the animate/inanimate
and human/nonhuman distinction. It is only used for mass nouns. Count
nouns take either he or she: she is used if the count noun refers to a female
human, and he is used for count nouns either referring to male humans or to
nonhuman entities. Thus we get a contrast as in Pass the bread its over
there. (bread =mass noun) and Pass the loaf hes over there. (loaf =
count noun). Most gendered pronouns are masculine pronominal forms (he,
himand Southwestern un, en <OE hine) referring to inanimate referents.
Acknowledgments
All authors most gratefully acknowledge the generous support by the
Deutsche Forschungsgemeinschaft. Without the funding of the Projects KO
1181/1-1,2,3 over a five-year period (2000-2005) the studies published here
and the compilation of FRED, the Freiburg English Dialect Corpus, would
have been impossible.
References
Allen, Will, J oan Beal, Karen Corrigan, Warren Maguire, and Hermann Moisl
forthc. Taming Unconventional Digital Voices: The Newcastle Electronic
Corpus of Tyneside English. In Using Unconventional Digital
Language Corpora. Vol. I: Synchronic Corpora, J oan Beal, Karen
Corrigan and Herman Moisl (eds.). Basingstoke: Palgrave
Macmillan.
Cheshire, J enny, Viv Edwards, and Pamela Whittle
1989 "Urban British dialect grammar: The question of dialect levelling.
English World-Wide 10: 185-225.
Kortmann, Bernd, Edgar W. Schneider in collaboration with Kate Burridge,
Rajend Mesthrie, and Clive Upton (eds.)
2004 A Handbook of Varieties of English, Vol. 2: Morphology and Syntax.
Berlin/New York: Mouton de Gruyter.
viii Bernd Kortmann
Milroy, J ohn, and Lesley Milroy (eds.)
1993 Real English. The Grammar of English Dialects in the British Isles.
London/New York: Longman.
Tagliamonte, Sali
1999 Was/were variation across the generations: View from the city of
York. Language Variation and Change 10: 153191.
2002 Variation and change in the British relative marker system. In
Relativisation in the North Sea Littoral, Patricia Poussa (ed.), 147
165. Munich: LINCOM EUROPA.
2003 Every place has a different toll: Determinants of grammatical
variation in cross-variety perspective. In Determinants of Linguistic
Variation, Gnter Rohdenburg, and Britta Mondorf (eds.), 531554.
Trudgill, Peter, and J ack K. Chambers (eds.)
1991 Dialects of English. Studies in Grammatical Variation. London/New
York: Longman.
Table of Contents
Preface...............................................................................................v
Bernd Kortmann
The Freiburg English Dialect Project and Corpus (FRED) ..............1
Bernd Kortmann and Susanne Wagner
1. Comparative dialect grammar from a typological perspective.................1
2. The Freiburg English Dialect Corpus (FRED).........................................4
3. Linguistic consequences of using oral history material .........................13
Relative clauses in English dialects of the British Isles..................21
Tanja Herrmann
Abstract ...............................................................................................................21
1. Introduction............................................................................................21
2. Data........................................................................................................22
3. Overall distribution of relative clauses and relative markers.................24
4. Previous investigations of relative markers............................................28
5. Restrictiveness/nonrestrictiveness..........................................................38
6. Personality/nonpersonality.....................................................................41
7. Preposition placement............................................................................45
8. Accessibility Hierarchy..........................................................................48
9. Resumptive pronouns.............................................................................70
10. Whichas connector?............................................................................87
11. Conclusion..............................................................................................94
Appendix 1...........................................................................................................97
Appendix 2.........................................................................................................105
Some do and some doesnt: Verbal concord variation in the north
of the British Isles.........................................................................125
Lukas Pietsch
Abstract .............................................................................................................125
1. Introduction..........................................................................................125
2. The Northern Subject Rule: Descriptive problems..............................128
3. Data from twentieth-century northern dialects.....................................132
4. The history of the Northern Subject Rule............................................173
5. Theoretical accounts of the Northern Subject Rule..............................179
6. Discussion: Variation and usage-based theories..................................190
Table of Contents x
Gender in English pronouns: Southwest England........................211
Susanne Wagner
1. Introduction..........................................................................................211
2. Gendered pronouns..............................................................................215
3. Gender in English and elsewhere.........................................................221
4. The corpora..........................................................................................235
5. Special referent classes.........................................................................251
6. Non-dialectal studies of gender assignment.........................................261
7. Persistence of gendered pronouns........................................................275
8. SED Basic Material ...........................................................................285
9. TheSED fieldworker notebooks data...................................................292
10. Southwest England oral history material..............................................319
11. Material from Newfoundland...............................................................339
12. Overall summary..................................................................................346
Appendix............................................................................................................353
(Additional) corpus material...............................................................................353
Index..............................................................................................368
The Freiburg English Dialect Project and Corpus
(FRED)
Bernd Kortmann and Susanne Wagner
1. Comparative dialect grammar from a typological perspective
The Freiburg project started in the late 1990s and has received major
funding from the Deutsche Forschungsgemeinschaft from spring 2000 until
spring 2005.
1
Its basic approach to the study of dialect grammar is informed
by the theoretical and methodological framework of functional (or:
Greenbergian) typology, which is primarily concerned with the patterns and
limits of morphological and syntactic variation across the languages of the
world. The basic idea of the Freiburg project is to adopt functional
typology as an additional reference frame for dialectological research that
fruitfully complements existing approaches. Among other things, this
means that, in a first step, we determine the cross-dialectal variation
observable in individual domains of grammar (in the present volume:
negation, relative clauses, pronominal and agreement systems) before, in a
second step, judging it against the cross-linguistic variation described in
typological studies. Both dialect syntacticians and typologists are bound to
profit from this kind of approach (cf. Kortmann, ed. 2004 for a collection of
studies on dialects in Europe conducted in this spirit). On the one hand,
dialectologists can draw upon a large body of typological insights in,
hypotheses on, and explanations for language variation. Dialect data can
thus be looked at in a fresh light and new questions be asked. On the other
hand, typologists will get a broader and, most likely, more adequate picture
of what a given language is like if they no longer ignore dialectal variation.
In fact, non-standard dialects (as varieties which are almost exclusively
spoken) are bound to be a crucial corrective for typological research, which
is typically (especially for languages with a literary tradition) concerned
with the written standard varieties of languages. Standard British English,
for example, is anything but representative of the vast majority of English
dialects if we think, for instance, of the absence of multiple negation or the
strict division of labour between the Present Perfect and the Simple Past.
2 Bernd Kortmann and Susanne Wagner
In the present volume, the typological perspective is most prominent in
Herrmanns study on relativization (not surprisingly so, given the fame of
Keenan and Comries NP Accessibility Hierarchy in language typology).
(1) Accessibility Hierarchy (AH)
subject >direct object >indirect object >oblique >genitive >
object of comparison
According to the AH, if a language can relativize any NP position further
down on the hierarchy, it can also relativize all positions higher up, i.e. to
the left of it. This constraint applies to whatever relativization strategy a
language employs. For the relativization strategy known as zero-
relativization (or: gapping) there is thus a clear prediction that the
relativized NP is most likely to be gapped if it is the subject of the relative
clause, next most likely if it is the direct object of the relative clause, etc.
However, this is clearly not the case for Standard English: the direct object
position can be gapped (2a), whereas the subject position cannot (2b):
(2) a. The man I called _____ was our neighbour. (direct object)
b.
*
The man _____ called me was our neighbour. (subject)
English dialects, on the other hand, conform to the AH prediction.
Examples like (2b) are nothing unusual, at all; in fact, gapping of the
subject position is an extremely widespread phenomenon in non-standard
varieties of English in and outside the British Isles:
(3) a. I have a friend ____ lives over there.
b. It aint the best ones ____ finish first.
So here we have a striking instance of the situation where the non-standard
varieties of English conform to a typological hierarchy whereas the
standard variety does not. For further ways in which the AH is relevant
when looking at other relativization strategies used in English dialects
compare Herrmanns comparative study of relative clauses in six English
dialect areas in this volume.
Another rewarding area for anybody investigating dialects from a
typological perspective is the study of negation markers and strategies. This
subsystem of English dialect grammar has been investigated in depth by
Lieselotte Anderwald, a senior member of the Freiburg research team (cf.
The Freiburg English Dialect Project and Corpus 3
especially Anderwald 2002a, 2003). For one thing, multiple negation (or:
negative concord) is another striking proof of the typological well-
behavedness of non-standard varieties of English (and other Germanic
languages), since multiple negation is the rule for many standard languages
in Europe. Only the standard varieties of Germanic (e.g. Standard English,
Standard Dutch, Standard German) are the exceptions. Furthermore, the
invariant supraregional negation markers dont (i.e. also for he/she/it dont)
and aint are in full accordance with the powerful typological concept of
markedness: as Greenberg found for many languages, morphological
distinctions tend to be reduced under negation. As Anderwald (2002b,
2004) has also nicely shown, the alleged amnt gap (Hudson 2000) in
almost all varieties of English (*I amnt vs. I amnot, arent I) is anything
but a gap and can indeed be considered an extreme case of local (or:
reversed) markedness. Whereas for all auxiliary verbs negative contraction
(e.g. havent, hasnt, wont) is vastly preferred over auxiliary contraction
(e.g. ve not, d not, ll not), we get the reverse picture for be. Even isnt
(12.5%) and arent (3.5%) are used very rarely in the British Isles, so that
the near absence of amnt in standard as well as non-standard varieties is
not a striking exception, but simply the tip of the iceberg.
The motivation for this striking preference of be-contraction over
negative contraction for all other auxiliaries is most likely a cognitive one,
namely the extremely low semantic content of be. This leads on to another
typological concept which can be usefully applied to the interpretation of
dialect facts: iconicity. In the case of be-contraction we find an extreme
formal reduction of a semantically near-empty auxiliary. In other words,
the amount of coding material matches the semantic content to be coded.
Another case in point is the fact that quite a number of non-standard
varieties in the British Isles and, in fact, around the world have made new
use of the number distinction for Past Tense be, i.e. the was-were
distinction (cf. Anderwald 2002a). These varieties use was for all persons
in the singular and plural in affirmative sentences, while using werent for
all persons in singular and plural in negative sentences, thus
remorphologizing the number distinction of Standard English as a polarity
distinction. What we have here is a showcase example of iconicity: a
maximal difference in form (was vs. werent) codes a maximal semantic
and cognitive difference (affirmation vs. negation). The relevant non-
standard varieties of English have clearly developed a more iconic polarity
pattern than Standard English has.
Having outlined and illustrated the typological approach to comparative
dialect grammar established by the Freiburg research group, we need to
mention at least briefly that there is of course also a generativist perspective
from which especially the dialect grammars of Italian, Dutch and German
(much less so English dialects) have been investigated. As a matter of fact,
generativists discovered the significance of dialect syntax for linguistic
theorizing and models of syntax much earlier than typologists. This
generativist interest in microparametric syntax began with the advent of the
Principles and Parameters approach in the 1980s and has steadily increased
ever since (cf., for example, Black and Motapanyane 1996 or various
contributions to Barbiers, Cornips and van der Kleij 2002), finding its way
even into generative theories based on Optimality Theory (cf., for example,
the Stochastic OT account of morphosyntactic variation in English by
Bresnan and Deo 2001). In this volume, the generativist perspective will
come in only where relevant publications on the dialect phenomena
investigated here exist.
2. The Freiburg English Dialect Corpus (FRED)
Given the aims of the Freiburg project it was first of all necessary to
compile a database which would allow to conduct serious qualitative and
quantitative morphosyntactic research across English dialects. The result is
the computerized Freiburg English Dialect Corpus (FRED), which has
been compiled over a period of roughly five years (including the
digitization of some 120 hours of audio material). FRED consists of
approximately 2.5 million words, with representative subsamples for all
English dialect areas including data from Scotland and Wales. The data in
FRED are orthographically transcribed interviews collected for the most
part during the 1970s and 1980s in the course of oral history projects all
over the British Isles. The majority of the informants are born between
1890 and 1920, i.e. are roughly a generation younger than the generation of
informants who were recruited for the Survey of English Dialects (SED).
2.1. Principles of compilation
Firstly, the corpus was designated to permit the investigation of phenomena
of non-standard morphosyntax (rather than analyses of phonetic or
phonological details). Features of syntax are almost by definition much
rarer than features of phonetics and phonology and very large quantities of
text are therefore necessary. (According to some estimates, about 40 times
the amount of text is needed for a syntactic analysis as opposed to a
phonetic one.) This considerably restricted the practicality of collecting our
own corpus from scratch. Instead, we decided to try to compile a corpus
from materials that were already available. We decided against collecting
material with the help of questionnaires in the first phase of the project.
Questionnaires were however designed and distributed in the second phase
of the project when, on the basis of extensive corpus analyses, interesting,
transitional or rare phenomena became apparent that could not be further
investigated with the help of FRED alone.
Secondly, we decided to collect material that would best be classified as
traditional dialect data. This means that we explicitly tried to find material
from speakers who grew up before the Second World War, as this date
seems to be the major cataclysmic event after which wide-ranging social
and economic changes (with concomitant linguistic changes) came into
effect. For example, highly increased mobility after WWII led to dialect
levelling on a hitherto unknown scale (see for example Williams and
Kerswill 1999: 149); mass affluence resulted, amongst other things, in
television sets becoming easily available and spreading at least passive
knowledge of the standard language; increased public spending made sure
that education changed not only qualitatively but also quantitatively, such
that children leaving school at age 11 or 12 not unusual for lower class
children only 60 or 70 years ago is no longer possible, and so on. Only by
concentrating on speakers born before WWII could we at least have a
chance that our data would still be dialectal in a regional sense, and be
comparable to older dialect descriptions and dialect data (on the
background of speaker selection for the SED, see Orton 1962: 14). There
are a number of other arguments and a priori considerations which
contributed to this decision: We had established contact with various
researchers, research groups and private individuals who were either in
possession of similar materials or were already working with such data, and
who had kindly offered us access to them. Moreover, the only existing
sources on variation in morphosyntax are based on traditional material,
most importantly the SED (Orton et al. 19621971). To guarantee
comparability between these materials, it was essential that FRED should
also consist of traditional dialect material without having to take into
account factors like mobility or the influence of mass media.
Due also to time constraints, but mainly for the reasons detailed above,
it was considered impossible from the outset to record, digitise and
transcribe all data that should make up FRED ourselves. Based on our
research objectives, we were looking for large quantities of traditional
regional speech, preferably by older local speakers with strong family
affiliations in the area, that would record the use of speakers who grew up
before WWII, or even better, before WWI. This meant that we were
looking for material preferably from the 1970s and 1980s, recording older
speakers, or from the 1990s, if these recorded very old speakers. Our
material had to be recorded in acceptable quality for linguistic analysis,
ideally even including transcripts that were reliable on a word-by-word
basis, and most important of all the material had to be more or less
freely available to us as researchers who had not originally been part of the
research design. These criteria suggested a new source that has so far not
or hardly been used for dialectological purposes, namely tape recordings
and transcripts from oral history projects.
2.2. The role of oral history
As defined by the Oral History Society, [o]ral history is the recording of
peoples memories. It is the living history of everyones unique life
experiences (Oral History Society at http://www.oralhistory.org.uk). Oral
history collections sometimes originate from projects (short- or long-term)
undertaken by an individual (sometimes also a group of individuals or an
institution), typically lay persons, not professional historians, with an
interest in a specific theme or topic, often just recording life memories.
Such a focus has certain implications concerning the content
2
and general
circumstances of an interview. Interviewees are generally pensioners in
their 60s or older, and only rarely do we find projects that have as many
female as male speakers.
3
The recording situation makes oral history material ideal for linguistic
investigation. The interviewers were usually true insiders, coming from the
area, often still speaking the dialect themselves, which tends to relax the
interview situation considerably. A second advantage is that the speakers
attention was genuinely on what was being said, rather than on how it was
being said. Fortunately, the Oral History Society advises all potential
interviewers to give a copy of the tapes to their local library or archive
4
,
and these are the places where oral history material can be found today
across Great Britain.
5
2.3. From original recording to text in FRED the steps and processes
Members of the Oral History Society are advised to at least [w]rite a
synopsis of the interview which briefly lists in order all the main themes,
topics and stories discussed;
6
verbatim transcripts are not explicitly
mentioned. But, of course, for anyone thinking about long-term work with
the material, a transcript is a very good way of allowing people from
outside to get an impression of the content of the interview without actually
having to listen to the tapes, which is a very time-consuming business. The
intentions for the future use of transcripts largely determine how the
interview was transcribed, how here referring particularly to the
(unfortunately very common) practice of normalizing the speakers
language. Since oral history projects as a rule do not involve the
employment of a professional transcriber, this is the usual course of events,
which is of course perfectly justified for oral history purposes. J ust to give
one example consisting of several actually occurring utterances, consider
(4) which could end up as (5):
(4) That pot? Oh, I, I dont know, I dont remember what I made he for.
I dont collect no pots now.
(5) I dont remember what I made that pot for. I dont collect pots now.
Normalization here has eradicated three morphosyntactic dialect features
(he=pot; hehere used in an oblique context; double negation dont ... no),
not to mention all the superfluous data (repetitions and so on) that are
simply left out. This kind of standardized re-written text is of course much
more useful to the general public than a transcript that uses so many
instances of eye-dialect to represent non-standard pronunciation as well
as dialect that it is difficult to follow the line of argument.
Despite the obvious linguistic drawbacks, the Freiburg research group
was very glad to have transcripts of at least some of its material. Although
these were highly deficient from a dialectological point of view, they at
least solved such difficult problems for us as deciphering correctly some
specialist vocabulary, unusual place names, personal names and so on, such
as the names of different apple sorts used for cider making. We then
carefully compared existing transcripts with the original tapes and re-
inserted all morphological, syntactic and discourse features, taking out
irrelevant phonetic or phonological features and features of pure eye-dialect
(compare 6 with 6'):
(6) And the farmer wot my gran used ter ee used ter have a white high
healed collar.
(6') #And the farmer what my gran used to, he used to have a white high
healed collar. (FRED Wil_011)
For the rest of the material where no transcripts were available, we
transcribed the original tapes ourselves, mostly with the help of native
speakers who either worked on the project or were associated with it in
related research projects. In addition, all transcripts were carefully checked
by dialectologically trained research assistants.
As a result, the actual transcripts used for FRED are verbatim
equivalents of the spoken versions: hesitations, repetitions, false starts of
the same sentence and so on are all included.
(7) #Oh well now, I tell you, when I first made my will, Mr (gap name)
my lawyer, (unclear) oh yes (/unclear) hes still alive, (trunc) I-
(/trunc) I told himI I I says, I want to leave the Salvation Army a bit
of money, and I have done. (FRED Nott_016)
In addition, as stated above, and most importantly for our research
purposes, all morphosyntactic dialect features have been reinserted
(indicated in bold print below).
(8) there used to be a Ginnet what we used to call was a Ginnet, he
were a nice eating apple, a nice sweet apple and a good apple for
cider. #When them apples were ripe you could pick themup and
could press themlike that and youd see your thumb mark in themor
any apple really when hes ripe, wadnt it, but when hes not ripe
hes hard, isnt he, (unclear) you aint gonna, (/unclear) well,
anything at all. (FRED Som_001)
Among the features likely to have been corrected in the original
transcript are, as in (8) for example, a what-relative, demonstrative them
and gendered pronouns.
A variety of phonological features were also kept, either if they were
already represented in the original transcripts, or if we suspected that they
might interact with morphosyntax, for example contracted forms like
wanna, gonna, spose and so on. It should be noted that we use the semi-
phonetic form mi for /mi:/ used as the possessive pronoun not as eye-
dialect but in order to facilitate searches. (For severe criticism of
gratuitous eye-dialect, see for example Preston 1985, 2000). The
orthographic form me, although widespread in other corpora, not only
suggests a certain etymology for this form (at worst, a substitution of the
object form of the personal pronoun for the possessive function), but also
complicates computer-based searches considerably, as all instances of the
object case of the personal pronoun (He saw me) have to be manually
excluded, at least as long as the corpus is not tagged for word class yet.
We also represent certain paralinguistic features like laughter, long
pauses, or indistinct stretches of conversation (marked as gaps, unclear
passages or truncated words; see also the examples above). All these
features are indicated in the transcripts by specific tags to minimize the risk
of ambiguities.
7
This opens up the possibility for analyses on a pragmatic or
discourse level. In this way we have tried to remedy the linguistic
shortcomings of the original oral history material as far as possible.
As mentioned above, extralinguistic variables in FRED are constrained
by intention FRED is not designed to be a representative sociolinguistic
corpus, but a regionally representative corpus of as broad a dialect speech
as possible. As has already been pointed out, our oral history projects
concentrated on interviewing older people. These older people are typically
very local, that is they still live in the place where they were born, without
having moved outside the region for any considerable stretch of time. Also,
typical FRED speakers usually left school about the age of fourteen, often
much earlier, certainly not progressing to higher education. Finally, most of
our speakers are male as is well known, women tend to use more
prestigious, in many cases more standard forms of speech where these are
available to them (see for example Chambers and Trudgill 1998: 30). In
other words, most of our speakers would qualify in dialectology as typical
NORMs (see Chambers and Trudgill 1998: 29), that is non-mobile old
rural male speakers with little education. Although this restricts the range
of investigations that can be conducted with the help of FRED in
sociolinguistic terms, it represents exactly the same bias as in earlier
dialectological work, where we find a preponderance of NORM speakers as
well, so that results from work on FRED will be comparable to earlier
studies or to material from earlier investigations.
2.4. Advantages and disadvantages of orthographic transcripts
FRED is transcribed orthographically, as are most computerized corpora. A
number of factors besides a simple realistic evaluation of our resources
had made it clear from the beginning of the project that orthographically
transcribing the dialect material would be the only viable (short-term)
procedure. First, we had been granted only a restricted amount of time (and
funding) to complete the compilation and transcription of the corpus, and
there has to be a natural trade-off between the detail of a transcription
(depth) and the coverage (breadth). Our aim was a large corpus that would
cover a number of dialect regions, and so we had to trade in phonetic detail.
Moreover, since our explicit focus was on morphosyntactic variability, for
all relevant features of dialect grammar that we expected to investigate and
that are discussed in the dialectological literature, a phonetic or phonemic
transcription would not only have been unnecessary, but even
counterproductive in many cases. For example, one major drawback of a
non-orthographic transcript concerns comparability. A non-orthographic
transcript would dramatically hamper the feasibility of searching for all
tokens of a certain type (for instance be, personal pronouns, and so on), as
the researcher would have no clue which forms to look for without
knowing which realisations actually occur in a given interview (or even
across all interviews). As a result, one would have to return to the
procedure that was common in corpus linguistics before the advent of
computers: reading through the texts and marking all forms of interest in
the process certainly not an ideal situation. Finally, only an orthographic
transcription of the data met the other requirements of our corpus: the
finished corpus was intended to be machine-readable, enabling easy access,
a variety of searches with various tools, and, most importantly,
comparability with other corpora/projects.
As has been mentioned above, research ties between the Freiburg team
and similar projects had been established. Since most of these projects were
working with orthographic transcripts, this lent additional support to the
decision to use orthographic transcription for FRED. Moreover
orthographic transcription would allow us to compare our data with older
collections and enable us to make comparisons between different speakers,
different dialects, different dialect areas, and different corpora. A further
advantage of orthographic transcription is the concentration on real
(morphosyntactic) dialect features, as phonetically exceptional forms do not
distract the analysts eye from the task at hand.
All transcription conventions have of course been documented. Thus in
many cases phonetic peculiarities may be traced from the transcription and
the additional databases alone without having to return to the sound files.
An alignment of sound and transcripts is planned for the near future.
2.5. FRED corpus design and area coverage
2.5.1. Word counts and areal distribution
FRED consists of 370 texts, which total roughly 2.5 million words of text or
about 300 hours of speech, excluding all interviewer utterances (see Table 1).
TheFRED material is broadly subdivided to cover nine major dialect areas,
following Trudgills modern dialects division of Great Britain (see Trudgill
1999: 65).
Table 1. FRED word counts and areal distribution
dialect area size (in thousands of words) % of total
Southwest (SW) 571 23
Southeast (SE) 643 26
Midlands (Mid) 359 15
North (N) 434 18
Scottish Lowlands (ScL) 169 7
Scottish Highlands (ScH) 23 1
Hebrides (Heb) 151 6
Isle of Man (Man) 10 1
Wales (Wal) 89 4
total 2,449 100%
Each dialect area is subdivided into different counties. A detailed
breakdown of counties can be found on the project website
(http://www.anglistik.uni-freiburg.de/institut/lskortmann/FRED/).
2.5.2. Speakers
FRED contains data from 420 different speakers (excluding interviewers):
268 (63.8 per cent) are male, and 127 (30.2 per cent) are female (gender is
unknown for the rest). In all, 77.2 per cent of the textual material in FRED
is produced by male speakers, and 21.4 per cent by female speakers.
The age of speakers included in FRED ranges from six years to 102
years, with a mean age of 75.2 years. A breakdown of age groups,
according to the amount of text produced by them, is given in Table 2. As
can be seen, about three quarters of the textual material in FRED is
produced by speakers older than 60 years.
Table 2. FRED speakers ages
age group number of speakers
% of textual material in
corpus produced
0 14 years 9 0.5%
15 24 years 14 1.2%
25 34 years 2 0.2%
35 44 years 2 0.1%
45 59 years 14 3.8%
60+years 233 74.8%
unknown 145 19.4%
The oldest of FREDs speakers was born in 1877. Overall, 14 speakers (3.3
per cent) were born between 1880 and 1889, 60 speakers (14.3 per cent)
were born between 1890 and 1899, 96 speakers (22.9 per cent) were born
between 1900 and 1909, and 64 speakers (15.2 per cent) were born between
1910 and 1919. This means that 89 per cent of all speakers in FRED were
born before 1920.
2.5.3. Recordings
The material included in FRED was recorded between 1968 and 1999. A
detailed breakdown of recording dates can be found in Table 3. Over two
thirds of all interviews were thus conducted in the 1970s and 1980s,
guaranteeing comparability with much dialectological work conducted at
that time.
Table 3. FRED interview recording dates
recording date number of speakers % of all speakers
196169 2 0.5%
197079 122 29.1%
198089 163 38.9%
199099 61 14.6%
unknown 71 16.9%
3. Linguistic consequences of using oral history material
The decision to base the FRED corpus predominantly on sources of oral
history projects has had a range of linguistic consequences, some of them
foreseen, others not predictable at the outset. Perhaps the most clearly
predictable linguistic consequences stem from the fact that oral history
material necessarily involves the speaker talking about his or her life story
at great length very often, in fact, the speakers are actively encouraged to
talk almost exclusively about their past. In the realm of tense and aspect, a
predominance of past time narratives implies a predominance of past tense
contexts (although not infrequently, of course, stretches of past time
narratives are narrated in the historical present tense as well). This is an
advantage for studies concentrating on past tense paradigms (for example
Anderwald in progress), but a clear disadvantage for any investigation into
the present tense, as the data typically yields too few examples to make a
regional comparison reliable (see Anderwald 2004). It also means that any
features that are linked to the present tense domain can be expected (and
indeed shown) to be underrepresented: for example use of the (present)
progressive vs. the simple form; forms for the recent past (for example
the after-perfect in Hiberno-English); uses, if any, of a habitual present
and so on.
A second feature one would expect, considering the fact that FRED
speakers tend to tell their own life stories, is a skewing in pronoun
frequencies. Based on the monologic nature of many of the interviews in
FRED, we might expect first person singular and first person plural
contexts to be over-represented. However, a comparison with the more
balanced demographic part of the British National Corpus (BNC) that
records everyday spontaneous conversations reveals that this is not the case
(see Table 4).
Table 4. Personal pronouns in FRED and the BNC
FRED BNC spoken
Pronoun occ. % of total occ. % of total
I 61,458 23.4 309,797 26.8
he 29,733 11.3 75,442 6.5
she 9,418 3.6 42,879 3.7
it 41,776 15.9 254,049 22.0
we 27,240 10.4 108,698 9.4
you 54,163 20.6 268,642 23.2
they 38,608 14.7 96,672 8.4
total 262,396 1,156,179
Despite the impression one gets when reading through FRED transcripts,
first person contexts are not over-represented in the corpus, but account for
roughly one third of all personal pronoun contexts in both FRED and the
spoken part of the BNC. Although there are slight deviations in frequencies
for individual third person contexts (which can easily be explained on the
basis of the nature of the recording situations), the overall frequency of first
versus third person contexts is surprisingly similar at 33.8 per cent versus
45.5 per cent in FRED and 36.2 per cent versus 40.6 per cent in the BNC
(spoken). Based on these figures, we expect that comparative analyses of
FRED and other corpora of spoken English involving the category person
will produce representative results.
Finally, in the realm of discourse, it has to be stressed that FRED does
not contain genuinely spontaneous, everyday conversations, as for example
theBNC does. In the worst case, some (but fortunately only a tiny minority
of) speakers actually read from prepared notes, as witnessed by pages
rustling in the background and distinctive pauses where pages are turned.
Although this worst case is mercifully rare, many interviews are
nevertheless monologic understandably, the interviewers tried to keep in
the background most of the time. FRED for this reason would probably not
lend itself well to the investigation of discourse strategies. However, this
limitation is probably not specific to FRED, but applies to dialectological
and sociolinguistic interviews alike, as the main objective is always to
record the speakers speech, rather than ones own (see Feagin 2002).
The nature of the data in FRED influenced the choice of the phenomena
which have been investigated in, so far, four Ph.D. theses and about a
dozen Masters theses. The former include those by Herrmann, Pietsch and
Wagner, which are presented in revised and shortened versions in the
present volume. In all these studies the focus has been on high-frequency
morphosyntactic phenomena. Moreover, the machine-readability of FRED,
which allows analyses via automatic text retrieval programmes like TACT
or WordSmith, has also influenced the methodology, in that for the first
time it is possible to conduct not only qualitative, but also quantitative
studies of dialect morphosyntax applying established corpus-linguistic
techniques.
Comparisons across the whole FRED material have not been possible
for very long yet, so most truly comparative projects by members of the
Freiburg research team are currently still work in progress. These include
cross-dialectal comparisons of multiple negation (Anderwald to appear
2005), past tense paradigms (Anderwald in progress), pronoun systems
(Hernndez in progress), complementation patterns (Kolbe in progress),
and for several areas of dialect morphosyntax the phenomenon of priming
(Szmrecsanyi 2004 and 2005). In addition, a whole range of Masters theses
have been completed or are currently under way on the basis of material
from FRED. For further information on FRED, the Freiburg project and
future perspectives of (English) dialect syntax, especially from a
typological perspective, compare Kortmann 2002, 2003, 2004; Anderwald
and Wagner 2005.
Notes
1. During the funding period (2000-2005) of Project KO 1181/1-1,2,3 a dozen
Masters theses and four Ph.D. theses were completed. Two doctoral theses,
two postdoctoral theses and several Masters theses based on FRED are well
under way. A selection of studies which has grown out of this project is given
in the references (marked with a superscript * preceding the publication year).
For further information on the Freiburg project and FRED, please consult:
http://www.anglistik.uni-freiburg.de/ institut/lskortmann/index.html.
2. There are some points to cover in every interview: date and place of birth,
what their parents and their own main jobs were. And whatever the topic, it
usually helps to get the interviewee talking if you begin with their earlier life:
family background, grandparents, parents and brothers and sisters (including
topics such as discipline), then onto childhood home (housework, chores,
mealtimes), leisure (street games, gangs, sport, clubs, books, weekends,
holidays, festivals), politics and religion, schooling (key teachers, friends,
favourite subjects), early relationships, working life (first job, a typical
working day, promotion, pranks and initiation, trade unions and professional
organisations), and finally later family life (marriage, divorce, children,
homes, money, neighbours, social life, hopes). Most people find it easier to
remember their life in chronological order, and it can sometimes take you two
or three sessions to record a full life story. (Preparing questions from
http://www.oralhistory.org.uk/advice/).
3. The advantages and disadvantages of using oral history material for linguistic
studies will be discussed in detail in section 3.
4. How to do ORAL HISTORY After the interview at http://www.oral-
history.org.uk/advice/.
5. However, this procedure also raises the unfortunate problem of copyright.
Especially for our older material, where many of the speakers have already
passed away, it is near impossible to gain copyright clearance. Museums and
archives are also often reluctant to provide copyright clearance, so that in the
foreseeable future FRED will only be accessible to academic researchers, and
cannot be published in its present form.
6. How to do ORAL HISTORY After the interview at http://www.oral-
history.org.uk/advice/.
7. The tags used in FRED include pauses, different types of non-verbal
elements, truncations, editorial corrections, dubious items, uncertain
transcriptions, gaps in transcription.
References
* Asterisks indicate publications or work-in-progress by members of the
Freiburg research team on the grammar of British English dialects.
Anderwald, Lieselotte
*2002a Negation in Non-Standard British English: Gaps, Regularizations,
Asymmetries. (Studies in Germanic Linguistics) London/New York:
Routledge.
*2002b *I amnt sure Why is there no negative contracted form of first
person singular be? In Anglistentag 2001 Vienna: Proceedings,
Dieter Kastovsky, Gunther Kaltenbck, and Susanne Reichl (eds.),
717. Trier: Wissenschaftlicher Verlag.
*2003 Non-standard English and typological principles: The case of
negation. In Determinants of Linguistic Variation, Gnter Rohden-
burg, and Britta Mondorf (eds.), 507529. Berlin/New York: Mou-
ton de Gruyter.
*2004 Local markedness as a heuristic tool in dialectology: The case of
amnt. In Dialectology Meets Typology, Bernd Kortmann (ed.), 47
67. Berlin/New York: Mouton de Gruyter.
*2005 Unexpected regional distributions: Multiple negation in FRED. In
Aspects of Negation, Yoko Iyeiri (ed.). Tokyo: Yushodo Press.
*in progress Naturalness and dialect grammar: Evidence from non-standard past
tense paradigms. Post-doctoral dissertation (Habilitationsschrift).
English Department, University of Freiburg.
Anderwald, Lieselotte, and Bernd Kortmann
*2002 Typology and dialectology: A programmatic sketch. In Present-day
Dialectology. Problems and Findings, J aap van Marle, and J an
Berns (eds.), 159171. Berlin/New York: Mouton de Gruyter.
Anderwald, Lieselotte, and Susanne Wagner
*forthc. The Freiburg English Dialect Corpus (FRED) Applying Corpus-
Linguistic Research Tools to the Analysis of Dialect Data. In Using
Unconventional Digital Language Corpora. Vol. I: Synchronic
Corpora, J oan Beal, Karen Corrigan and Herman Moisl (eds.).
Basingstoke: Palgrave Macmillan.
Barbiers, Sjef, Leonie Cornips, and Susanne van der Kleij (eds.)
2002 Syntactic Microvariation. Amsterdam: SAND.
(http://www.meertens.knaw.nl/projecten/sand/ synmic)
Black, J ames R., and Virginia Motapanyane (eds.)
1996 Microparametric Syntax and Dialect Variation. Amsterdam/Phila-
delphia: Benjamins.
Bremann, Rolf
1984 Soziolinguistische Untersuchung zum Englisch von Cornwall.
Frankfurt: Lang.
Bresnan, J oan, and Ashwini Deo
2001 Grammatical constraints on variation: Be in the Survey of English
Dialects and (Stochastic) Optimality Theory. Unpublished manu-
script. http://www-lfg.stanford.edu/bresnan/download.html
Chambers, J ack K.
2004 Dynamic typology and vernacular universals. In Dialectology Meets
Typology, Bernd Kortmann (ed.), 127145. Berlin/New York: Mou-
ton de Gruyter.
Chambers, J ack K., and Peter Trudgill
1998 Dialectology. Cambridge: Cambridge University Press.
Feagin, Crawford
2002 Entering the community: Fieldwork. In The Handbook of Language
Variation and Change, J ack K. Chambers, Peter Trudgill, and Nata-
lie Schilling-Estes (eds.), 2039. Oxford/New York: Blackwell.
Fischer, Andreas
1976 Dialects in the South-West of England: A Lexical Investigation.
Bern: Francke.
Hernndez, Nuria
*in progress Pronouns in dialects of English: A corpus-based study of non-
standard phenomena. Ph.D. thesis. English Department, University
of Freiburg.
Hickey, Raymond
2003 A Source Book for Irish English. Amsterdam/Philadelphia: Benja-
mins.
Hudson, Richard
1999 Subject-verb agreement in English. English Language and
Linguistics 3: 173207.
2000 *I amnt. Language 76: 297323.
Kastovsky, Dieter, Gunther Kaltenbck, and Susanne Reichl (eds.)
2002 Anglistentag 2001 Vienna: Proceedings. Trier: Wissenschaftlicher
Verlag.
Klemola, J uhani
2002 Continuity and change in dialect morphosyntax. In Anglistentag
2001 Vienna: Proceedings, Dieter Kastovsky, Gunther Kaltenbck,
and Susanne Reichl (eds.), 4756. Trier: Wissenschaftlicher Verlag.
Kolbe, Daniela
*in progress Complementation patterns in English dialects. Ph.D. thesis. English
Department, University of Freiburg.
Kortmann, Bernd
*1999 Typology and dialectology. In Proceedings of the 16th International
Congress of Linguists, Paris 1997, Bernard Caron (ed.). CD-ROM.
Amsterdam: Elsevier Science.
*2002 New prospects for the study of dialect syntax: Impetus from
syntactic theory and language typology. In Syntactic
Microvariation, Sjef Barbiers, Leonie Cornips, and Susanne van der
Kleij (eds.), 185213. Amsterdam: SAND.
*2003 Comparative English dialect grammar: A typological approach. In
Fifty Years of English Studies in Spain (1952:2002). A
Commemorative Volume, Ignacio M. Palacios Martinez, Mara J os
Lpez Couso, Patricia Fra, and Elena Seoane (eds.), 6583.
Santiago de Compostela: University of Santiago.
Kortmann, Bernd (ed.)
2004 Dialectology Meets Typology. Dialect Grammar from a Cross-
Linguistic Perspective. Berlin/New York: Mouton de Gruyter.
Mair, Christian
2002 Three changing patterns of verb complementation in Late Modern
English: A real-time study based on matching text corpora. English
Language and Linguistics 6: 105131.
Mair, Christian, Marianne Hundt, Geoffrey Leech, and Nicholas Smith
2002 Short-term diachronic shifts in part-of-speech frequencies: a
comparison of the tagged LOB and F-LOB corpora. International
J ournal of Corpus Linguistics 7: 245264.
Montgomery, Michael
1994 The evolution of verb concord in Scots. In Studies in Scots and
Gaelic: Proceedings of the Third International Conference on the
Languages of Scotland, Alexander Fenton, and Donald A. Mac-
Donald (eds.), 8195. Edinburgh: Canongate Academic.
Orton, Harold
1962 Survey of English Dialects Introduction. Leeds: Arnold.
Orton, Harold, Michael V. Barry, Eugen Dieth, Wilfrid J . Halliday, Philip M.
Tilling, and Martyn F. Wakelin (eds.)
196271 Survey of English Dialects. Leeds: Arnold.
Preston, Dennis R.
1985 The Lil Abner syndrome: written representations of speech.
American Speech 60: 328336.
2000 Mowr and mowr bayud spellin: confessions of a sociolinguist.
J ournal of Sociolinguistics 4: 614621.
Siemund, Peter
2002 Animate pronouns for inanimate objects: pronominal gender in
English regional varieties. In Anglistentag 2001 Vienna: Procee-
dings, Dieter Kastovsky, Gunther Kaltenbck, and Susanne Reichl
(eds.), 1934. Trier: Wissenschaftlicher Verlag.
Szmrecsanyi, Benedikt
*2004 Persistence phenomena in the grammar of spoken English. Ph.D.
thesis. English Department, University of Freiburg.
*2005 Creatures of habit: A corpus-linguistic analysis of persistence in
spoken English. Corpus Linguistics and Linguistic Theory 1: 113
149.
Tristram, Hildegard L. C.
1997 DO-periphrasis in contact? In Language in Time and Space,
Heinrich Ramisch, and Kenneth Wynne (eds.), 401417. Stuttgart:
Steiner.
Trudgill, Peter
1999 The Dialects of England. Oxford: Blackwell.
Wakelin, Martyn F.
1975 Language and History in Cornwall. Leicester: Leicester University
Press.
Weltens, Bert
1983 Non-standard periphrastic do in the dialects of South West Britain.
Love and Language 3: 5674.
Williams, Ann, and Paul Kerswill
1999 Dialect levelling: Change and continuity in Milton Keynes, Reading
and Hull. In Urban Voices: Accent Studies in the British Isles, Paul
Foulkes, and Gerald Docherty (eds.), 141162. London: Edward
Arnold.
Relative clauses in English dialects of the British
I sles
Tanja Herrmann
Abstract
This cross-dialectal analysis of adnominal relative clauses relates results from a
variety of English dialects to typological hierarchies, particularly the Noun Phrase
Accessibility Hierarchy (AH; Keenan and Comrie 1977). The AH with a few
reservations is verified for all relative clause formation strategies found in the
data, including the zero relative marker strategy. From a diachronic perspective,
the AH also helps to reveal the pattern underlying the way individual relative
markers enter or exit an existent relative marker system. The pronoun retention
strategy, however, does not necessarily follow a reversed order of syntactic
positions on the AH, as claimed by Keenan and Comrie. At least resumptive
pronouns in nonrestrictive relative clauses, that type of relative clause in which the
majority of resumptive pronouns in the dialect data are found, thwart AH-based
expectations due to their prevalence of the subject position. Moreover, due to their
explicating function, resumptives give rise to further embedded relative clauses
and other nonstandard relative constructions (nonreduction; connector which).
1. I ntroduction
This study is concerned with prototypical relative clauses, i.e. adnominal
relative clauses. A relative clause is a subordinate clause that modifies an
antecedent with which a relative marker in the relative clause is
coreferential. An adnominal relative clause forms part of a constituent of
the matrix clause, as in the examples in (1) below. In these examples, as
throughout this chapter, the following notational conventions are used: The
relative clause is put in square brackets and the antecedent is marked in
boldface. The zero relative marker is indicated by the symbol . The initial
three capital letters indicate the region from which the dialect data
originate: CMI =Central Midlands, CNO =Central North, CSW =Central
Southwest, EAN =East Anglia, NIR =Northern Ireland, SCO =Scotland
(cf. Kortmann and Wagner, this volume: Trudgills [1999] 2003 Map 18:
Modern Dialect areas). This information on the source region is followed
22 Tanja Herrmann
by the text code, the speakers identification code (where available), and
(for the BNC texts) the sentence number (see section 2):
(1) a. ... they knew just where to stop and start, especially the last
pony [ I had]. (EAN-K65<S: 0806>)
b. ... this Billy [that used to go round all the district] and, and
[buy up all these old cast horses] and [bring themup there]
.... (SCO-GYS<Person: PSSCO6><S: 136>)
c. ... they used to perhaps have competitions for the childrens
[what used to want to go on]. (CMI-FYD<S: 303>)
Adnominal relative clauses are a central syntactic phenomenon in every
dialect, taking however different forms in different dialects. Although
adnominal relative clauses have been a fair center of interest in theoretical
linguistics, including some works on individual dialects, a cross-dialectal
study within a typological framework has never been undertaken. This
framework apart, all formation strategies of adnominal relative clauses
investigated in this chapter will be subjected to a cross-dialectal analysis
with the aim of identifying salient properties that individual dialects have in
common, and those properties in which they differ from one another and
from the standard variety.
In a comparative study, it can be determined which standard features
have made inroads into traditional dialects, and to what extent (e.g. whether
there is a predominance of wh-pronouns). On the other hand, dialects also
converge toward one another in a process of dialect levelling. General
nonstandard features of informal speech have developed or are developing
from traditional dialect features (e.g. the nonstandard relative marker what)
and may in turn affect the future shape of Standard English. A comparative
view can identify these supra-regional features of informal speech, thus
allowing some prognosis as to their future development and whether they
might find entry into Standard English.
2. Data
The material used for the present investigation constitutes largely a
subcorpus of the Freiburg English Dialect Corpus (FRED; cf. Kortmann
and Wagner, this volume). The data are taken from six areas (East Anglia,
Relative clauses in dialects of English 23
the Central Southwest, the Central Midlands, the Central North, Scotland,
and Northern Ireland) and from several sources:
The East Anglian, the Central Midland, and the Scottish data are
dialectal texts extracted from the British National Corpus (BNC),
published in 1995. They were reformatted into easy-to-read texts, while
selective tags were kept. The texts result from oral history projects and
are based on recorded interviews with elderly, working-class people
speaking naturally. The East Anglian data were recorded in Suffolk and
Eastern Cambridgeshire, the Central Midlands data in Nottinghamshire,
and the Scotland data in Midlothian, Ayrshire, Selkirkshire,
Lanarkshire, and Invernesshire.
A part of the Central Southwest data is J uhani Klemolas Somerset
Rural Life Museum data (texts SRLM 105, SRLM 107, SRLM 108,
SRLM 109, SRLM 122, SRLM 123; SRLM 132).
1
The other part of the Central Southwest data and the Central North data
are taken from the FRED Corpus.
All Central Southwest data originate from Eastern Somerset. The
Cumbrian data were recorded in the pre-1974 counties Westmorland,
Cumberland, and Lancashire, in particular in the Ambleside area,
formerly Westmorland.
The Northern Ireland data are part of J ohn Kirks Northern Ireland
Transcribed Corpus of Speech (NITCS), in a re-edited form.
2
Only texts
with elderly informants (age group 3) were considered. The Northern
Ireland data were recorded in Antrim, Londonderry, Tyrone,
Fermanagh, Armagh, and Down.
The dialect corpus analyzed in this chapter totals approximately 480,000
words, divided into some 80,000 words for each of the six dialect regions.
The word dialect here should be taken with a grain of salt, as it is rather a
convenient means to partition the map than any claim to the existence and
location of firm dialect boundaries. The reason why these regions were
chosen was that they cover the British map areally, that means, they are
areas which are sufficiently wide apart geographically and have an identity
of their own. For the sake of convenience, the regional speech used in these
areas is referred to as dialect. Each region is named after a larger
geographical area, largely following Map 18 Modern Dialect areas
published in Trudgill (2003).
24 Tanja Herrmann
The six regional corpora are not equally dialectal. As a whole, the
Central Southwest, the Central North, and the Northern Ireland data
represent broader dialectal speech than do the data from the East Anglian,
Central Midland, and Scottish subcorpora. In fact, the Central Southwest
and the Northern Ireland corpora exclusively consist of data by broad
speakers, while the composition of speakers of the other four corpora is
more heterogeneous.
3. Overall distribution of relative clauses and relative markers
Table 1 presents the overall frequencies of adnominal relative markers in
the six investigated regions (Central Southwest, East Anglia, Central
Midlands, Central North, Scotland; Northern Ireland). Absolute numbers
are typed in boldface; percentages are given in square brackets. A look at
the totals of relative markers across the six regions reveals that the total of
occurrences (1874) with relative particles (zero, that, what; as) outnumbers
the total of occurrences (638) with relative pronouns (who, which, whom;
whose) by almost 3:1. The most frequent relative marker in the corpus is
that (39%), followed by zero (28.1%), which (15.1%), who (10.1%),
nonstandard what (6.8%), and as (0.8%). The case-marked wh-pronouns
whom(0.2%) and whose (0.1%) are very unusual in dialectal speech.
The regional differences in relative marker distribution are indicated by
the order and the font size in Map 1, which is based on Trudgills (2003)
Map 18 Modern Dialect areas, supplemented by a map of Ireland and
Northern Scotland. Leaving wh-pronouns as a standard and supra-regional
feature aside for the moment, Figure 1 presents the percentages of relative
markers when moving from north to south.
Table 1. Areal distribution of relative markers
CSW EAN CMI CNO SCO NI R TOTAL
(Eastern
Somer-
set)
(Suffolk;
Eastern
Cam-
bridge-
shire)
(Nott-
ingham-
shire)
(Cum-
bria:
Cumber-
land,
West-
morland,
Northern
Lanca-
shire)
(Lothian,
Borders,
Strath-
clyde,
Inver-
nesshire)
N [%]
3
N [%] N [%] N [%] N [%] N [%] N [%]

zero 84
[28.9]
86
[20.4]
80
[17.7]
142
[34]
123
[23.6]
191
[46.9]
706
[28.1]
that 77
[26.5]
93
[22]
182
[40.3]
182
[43.5]
241
[46.2]
204
[50.1]
979
[39]
what 65
[22.3]
67
[15.9]
26
[5.8]
10
[2.4]
2
[0.4]
- 170
[6.8]
as - - 11
[2.4]
6
[1.4]
- 2
[0.5]
19
[0.8]
who 26
[8.9]
65
[15.4]
57
[12.6]
30
[7.2]
70
[13.4]
5
[1.2]
253
[10.1]
which 39
[13.4]
111
[26.3]
91
[20.1]
48
[11.5]
84
[16.1]
5
[1.2]
378
[15.1]
whom - - 2
[0.4]
- 2
[0.4]
- 4
[0.2]
whose - - 3
[0.7]
- - - 3
[0.1]

total 291
[100]
422
[100]
452
[100]
418
[100]
522
[100]
407
[100]
2512
[100]
26 Tanja Herrmann
Map 1. Areal distribution of relative markers
that what
whichwho
0 50 100
that what
which who
that what
as which who
whose whom
that
what as
which who
that what
which who
whom
that
as
which who
zero that what as
North Northern Ireland: 46.9% 50.1% - 0.5%
Scotland: 23.6% 46.2% 0.4% -

Central North: 34% 43.5% 2.4% 1.4%

Central Midlands: 17.7% 40.3% 5.8% 2.4%
East Anglia: 20.4% 22% 15.9% -
South Central Southwest: 28.9% 26.5% 22.3% -
Figure 1. Distribution of relative markers along the North-South axis in
percentages
The Central Southwest data show rather even frequencies of zero (28.9%),
that (26.5%), and what (22.3%). East Anglia presents a similar picture: that
(22%), zero (20.4%), and what (15.9%) are quite evenly distributed. As we
move northward, that steadily gains strength. In the Central Midlands, that
(40.3%) is the predominant relative marker, at the expense of zero (17.7%)
and what (5.8%), which both have less importance (in absolute numbers
and percentages) when compared to the southern areas. What is more than
twice as strong as the relative particle as (2.4%), which is not found in the
south and is, percentage-wise, weak in the Central Midlands. In the Central
North, that accounts for 43.5%, although zero (34%) is also prominent.
What (2.4%) and as (1.4%) have about halved their percentages in
comparison to the Central Midlands. Scotland is even more clearly
dominated by that (46.2%). Zero (23.6%) is a much weaker second; what
(0.4%) is almost nonexistent (one clear case in Glasgow and one dubious
instance), and as is absent. Finally, in Northern Ireland, that (50.1%) is
used in about half of all instances. The other half is almost taken up by zero
(46.9%). While what is unknown, as (0.5%) is hovering around half a per
cent.
With particular regard to the dialectal variants what and as, what is by
far the stronger one, the more so the farther south we go. In the south (East
Anglia; Central Southwest), what has a substantial number of instances,
whereas in the north (Central North; Scotland), it plays a marginal role; in
Northern Ireland what plays no role at all.
As has its stronghold in the Central Midlands. It is used, though not
often, in the Central North and in one county (Tyrone) of Northern Ireland.
28 Tanja Herrmann
According to the present data, what has its strongest position in the Central
Southwest.
All existing wh-pronouns have found their way into the investigated
dialects. However, depending on the dialectal broadness of the individual
subcorpus, their proportion varies from 2.4% in the very broad Northern
Ireland subcorpus to 41.7% in the East Anglian subcorpus, which, as a
whole, is closest to the standard variety. In other words, the composition of
speakers and the dialectal quality of the data of the corpora are reflected in
the frequency of wh-pronouns. In between, there are the broad subcorpora
from the Central North (18.7% wh-pronouns) and the Central Southwest
(22.3% wh-pronouns) and the less broad subcorpora from Scotland (29.9%
wh-pronouns) and the Central Midlands (33.8% wh-pronouns). In addition
to the overall percentage of wh-pronouns, the presence or absence of case-
marked wh-forms is also indicative of how standardized or how dialectal
(i.e. broad) a subcorpus is: whom(two instances in Scotland; two in the
Central Midlands) and whose (three in the Central Midlands) only appear in
two of the three less broad subcorpora. Thus, the frequency of wh-pronouns
serves as a yardstick for the degree of standardization or traditionality of
speech, respectively. While traditional dialect only comprises relative
particles, (written) Standard English abounds in wh-pronouns, which are
indeed a typical trait of Standard English. Thus, the more wh-pronouns,
particularly of the case-marked variant, a corpus of dialect data contains,
the closer this corpus is to the standard variety and the further away from
traditional dialect.
4. Previous investigations of relative markers
4.1. Supra-regional studies and/or national surveys
Before giving a summary of earlier works on the six dialect regions in
section 4.2, it is in this section that more detailed information will be given
on the three supra-regional and national surveys conducted for the dialects
of England in the late 19th and, roughly, first half of the 20th century.
4.1.1. Wrights English Dialect Grammar
In J oseph Wrights English Dialect Grammar ([1905] 1961), the relative
marker what is said to occur in some of the north-midland counties and in
nearly all the counties south of the north midlands (Wright 1961: 77). As
is occasionally used (Wright 1961: 77) in Westmorland and generally
used in Nottinghamshire, East Anglia, and East Somerset, while at is
generally used in Scotland and Ireland. The zero relative marker is a
recurrent phenomenon in dialect also in subject position whereas whom
is never employed (cf. Wright 1961: 77).
Map 2. Map 207 31.1: (a man) thats poor (Lowman Survey)
4.1.2. Lowman Survey
The Lowman Survey of Middle and South England was carried out in
1937/38 and supplemented by data from Henry E. Collins for the Southeast
in 1950. For the Central Southwest, Map 207 and 208, which are published
30 Tanja Herrmann
in Vierecks (1975) atlas, yield the following results: Map 207, featuring
question 31.1: a man) thats poor (i.e. linguistic environment: restrictive
relative clause; subject position; personal, indefinite antecedent), displays
as in all counties except for Dorset, where the relative marker is that. Map
208 (cf. next page) gives a similar picture. It reproduces question 31.2: hes
a boy) whose father, which asks for a genitive relative marker. It turns out
that the periphrastic genitive as his father is used in all counties except for
Dorset, where a paratactic continuation
4
(his father) is used.
Map 3. Map 208 31.2: (hes a boy) whose father (Lowman Survey)
In East Anglia the survey investigated four localities in Suffolk, three in
Cambridgeshire, (and three localities in Norfolk and three in Essex). Map
207 shows nearly an even distribution of what (two in Suffolk; two in
Norfolk; one in Northern Essex) and that (two in Suffolk; two in
Cambridgeshire; one in Norfolk; two in Southern Essex) (cf. also Viereck
1980: 27). Map 208 yields similar findings: Periphrastic what his father
(two in Suffolk) and that his father (one in Western Cambridgeshire; two in
Norfolk), or its reduced variant thats father (one in Suffolk), occur about
equally often. A paratactic continuation
5
(his father) was chosen as an
alternative in all four counties (one in Suffolk; one in Eastern
Cambridgeshire; one in Norfolk; three in Essex). This suggests that East
Anglia, particularly Western Suffolk, is the heartland of relative what.
6
That was the major alternative, while as had moved further to the west.
4.1.3. Survey of English Dialects
TheSurvey of English Dialects (SED) was mainly conducted in the 1950s
with non-mobile, old, rural males and (marginally) females in 313 localities
in England.
7
The informants were given a questionnaire which contained
three questions asking for relative marker usage: Question III.3.7: If I
didnt know what a cowman is, you would tell me: He is the man ... looks
after the cows. (i.e. linguistic environment: restrictive relative clause;
subject position; personal, nonspecific, definite antecedent; that-frame),
Question IX.9.5: The woman next door says: The work in this garden is
getting me down. You say: Well, get some help in. I know a man ... will do
it for you. (i.e. linguistic environment: restrictive relative clause; subject
position; personal, specific, indefinite antecedent; who-frame), and
Question IX.9.6: That mans uncle was drowned last week. In other words,
you might say, thats the chap ... (uncle was drowned). (i.e. linguistic
environment: restrictive relative clause; genitive position; personal,
specific, definite antecedent).
8
Additional, non-elicited information on the
questions was noted down as incidental material. Responses to Questions
IX.9.5 and IX.9.6 are presented cartographically as S5 and M81 in The
linguistic atlas of England (LAE), while Vierecks The computer developed
linguistic atlas of England 1 (1991) presents the responses (and incidental
material) to all three SED questions as S8a and S8b, S9, and S10. All six
maps are reproduced in Appendix 2.
In the Central Southwest, the SED investigated thirteen localities in
Somerset, seven of which in Eastern Somerset, eight in Wiltshire, five in
Dorset, four in West Berkshire, six in Oxfordshire, one in South
Gloucestershire, and three in Western Hampshire.
In response to Question III.3.7, Dorset indeed seems to be an outsider in
the Central Southwest in being a that and zero area, although what and as
appear once in the incidental material. In the lower core counties of the
Central Southwest, namely in Somerset, Wiltshire, and West Berkshire,
32 Tanja Herrmann
what, that (and its phonemic variant at), and zero are about equally
frequent. Who and as, on the other hand, are rather infrequent. When
including the northern counties Oxfordshire and South Gloucestershire, as
catches up with what, that, and zero. Western Hampshire, on the periphery
of the Central Southwest, already leans toward the Southeast in displaying
who and zero. In the incidental material, however, what (above all in
Eastern Somerset) and particularly as dominate. While both what and as do
not transcend the Eastern Somerset/Western Somerset borderline, as has its
stronghold in the more interior counties Wiltshire, (South) Gloucestershire,
(West) Berkshire, and Oxfordshire to the north.
In response to Question IX.9.5, who is the predominant relative marker.
As gains strength again, as one moves northward and further into the
mainland, i.e. in (West) Berkshire, (South) Gloucestershire, and
Oxfordshire. The standard genitive relative marker whose prevails in the
entire Central Southwest, although in Eastern Somerset and Oxfordshire
potential genitives are also promoted to other (higher) syntactic positions,
such as indirect object (dative) and subject (nominative) (cf. section 8 and
8.4.). Altogether, the wh-pronouns who and whose are prevalent (in
Questions IX.9.5 and 9.6). What, that, zero, and as are all documented, yet
as seems to be rather restricted.
In East Anglia, the SED data have been collected in five localities in
Suffolk, one in Eastern Cambridgeshire, thirteen in Norfolk, as well as
fifteen localities in neighboring Essex.
In response to SED Question III.3.7, the relative marker as dominates in
Cambridgeshire and in the area to the west, whereas Norfolk, Suffolk,
Eastern Cambridgeshire, and Essex are part of a what area which comprises
the entire Southeast (cf. Viereck 1991: S8b). However, what seems to
alternate with that in this area; the latter is not as frequent as what though
(cf. Viereck 1991: S8a and S8b).
In response to Question IX.9.5, that occurs only once as a second choice
in Norfolk. In all probability, at is no separate relative marker in East
Anglia but a weakened form of that, whose initial th- was dropped, which
is an occasional phonological feature there (cf. also Poussa 1996: 529;
Peitsara 2002: 169). When looking at Map S9 in Appendix 2, the
predominance of who in East Anglia (and surroundings, as well as the
entire South) becomes evident. That is conspicuously rare in all England
(althoughat is frequent in the north). As is absent in East Anglia, but starts
to be very frequent in Bedfordshire, Northamptonshire, Buckinghamshire,
and farther to the west (cf. Viereck 1991: S9). The synthetic genitive
relative marker whose outnumbers the periphrastic genitive constructions
by far.
In sum: While the dialectal relative markers what and as and the relative
particlesthat and zero are preferred in SED Question III.3.7 He is the man
... looks after the cows (subject position; nonspecific, definite antecedent),
the standard wh-relative pronouns who and whose are preferred in IX.9.5 I
know a man ... will do it for you (subject position; specific, indefinite
antecedent) and IX.9.6 thats the chap ... (uncle was drowned) (genitive
case; specific, definite antecedent), respectively. Of the 46 instances of
what accompanying Question III.3.7 in the incidental material, about twice
as many instances of what were found in nonspecific (and definite)
environments than in specific environments. Of the four instances of that,
all seem to occur in nonspecific (and definite) environments. The eight
instances of as do not seem to be affected by syntactic environment. That is
to say that (non)specificity seems to be the critical syntactic variable (cf.
Poussa 1988: 447448 and 465), in the sense that specific antecedents
prefer who(such as in SED Questions IX.9.5 and IX.9.6), while nonspecific
antecedents are left for the nonstandard relative particle what (instead of the
standard relative particle that in the that-frame Question III.3.7).
9
In the Central Midlands, the SED investigated four localities in
Nottinghamshire, one in Eastern Derbyshire, and four in North-Western
Leicestershire. As occurs in the great majority of responses to all three SED
questions, including the incidental material. The Central Midlands are part
of a large as area, which extends over the entire Midlands and reaches up to
Lancashire in the north, roughly Eastern Somerset in the south, and
Cambridgeshire/Buckinghamshire in the east, excluding the Southeast (cf.
Viereck 1991: S8a, S8b, S9, and S10). There are also sporadic instances of
zero, that, what, and who, while whose is regularly used in genitive
position.
In the Central North, the SED investigated six localities in Cumberland,
four in Westmorland, and two in North Lancashire. The Central North
belongs to an at area in the north of England, although the more dominant
relative marker in North Lancashire appears to be as. (In the present study,
at, ut, and t are considered as phonemic/phonetic variants of that.) There is
a single occurrence of as in central Cumberland (i.e. in Cu4 =Threlkeld),
and the single occurrence of what in the Central North occurs no farther
north than North Lancashire (cf. Orton and Halliday 19621963: 243 and
10821085; cf. also Viereck 1991: S8a, S8b, S9, and S10).
34 Tanja Herrmann
4.2. Areal distribution of relative markers in previous investigations
With respect to the six dialect regions and, especially, the individual counties
investigated here, the major findings of previous studies are summarized in
Tables 1520 in Appendix 1. Each table provides an overview of relative
marker distribution in the respective region in chronological order. Below,
only the main similarities and differences to the areal distribution of relative
markers in the present data will be pointed out.
4.2.1. The Central Southwest Eastern Somerset
When comparing the results of the present study on the Central Southwest
to those of previous investigations, it becomes clear that in the past, as
reached as far south as Gloucestershire, Berkshire, Wiltshire, and Eastern
Somerset. In Western Somerset as did not occur, in Dorset just once in the
incidental material of Question III.3.7 in combination with all (cf. Orton
and Wakelin 1967: 291). However, it has to be conceded that as
occasionally reappears in the Lower Southwest (Devon; Cornwall) in SED
material in response to Question III.3.7 and in the incidental SED material
(cf. Orton and Wakelin 1967: 291). According to the present data, as
nowadays seems to have receded further north in relative clause formation.
Apart from that, Ihalainens (1980) distribution of relative markers for
Somerset is very similar to the present distribution. In contradistinction to
that, van den Eyndens (1992, 1993, 2002) frequencies on Dorset depart
from it: While the percentage of the zero marker is still relatively similar to
the percentage of the distribution found here, the percentage of that is
remarkably higher, and that of what very much lower, attesting that Dorset
is a distinct dialect area.
The nonstandard relative marker what has been part of the Central
Southwestern dialect from the earliest citation onward, i.e. it dates back
there at least to the middle of the second half of the nineteenth century,
when Elworthy (1877) mentioned it for West Somerset. Some authors
(Barth 1968 for Gloucestershire and SED Informant So[merset]1) hint at an
antagonism between what and as, in which as is felt to be the older variant
of the two, which was already superseded by the younger combatant at the
time of the SED. Although who is the dominant relative marker in the SED
questionnaire in Eastern Somerset, what is very frequent in the incidental
material.
4.2.2. East Anglia Suffolk and Eastern Cambridgeshire
While as is described as in general use for East Anglia in Wrights
English Dialect Grammar (cf. Wright 1961: 77), its occurrences became
very sporadic after that time. In the Lowman Survey, as is not recorded for
East Anglia any more, while the SED still records as as the dominant
relative marker for Cambridgeshire in response to Question III.3.7 and in
the incidental material of that question (cf. Orton and Tilling 1969: 301).
Later on, as apparently retracted even further west than Cambridgeshire,
where it is found only once by Ojanen (1982: 74) in Mid Cambridgeshire.
In the present East Anglian data from Suffolk and Eastern Cambridgeshire,
no instance of as was encountered.
On the other hand, what is almost unanimously recorded as the
dominant marker in East Anglia, which can be regarded as the heartland of
what. Even in Cambridgeshire, what is reported as the most frequent
relative marker by Ojanen. Not surprisingly, the present data mainly from
Suffolk are mostly in line with Keklinen/Peitsaras (1985, 2002)
findings for Suffolk: While Ojanens higher number of zeros may be partly
attributable to the inclusion of non-relative clause constructions, the
percentages for that are almost the same (ranging between 20.4% and
23%). Keklinen/Peitsaras higher figures for what (29.1% and 21.2%
versus 15.9% in the present data) are to be explained by the difference in
broadness between their corpus and the one used here. The fact that the
present corpus comprises less broad dialect speech is also reflected by the
much higher figures for who (15.4% in the present data versus 2.9% and
7.4%) and, to a lesser extent, which (26.3% in the present data versus
21.4% and 16.9%), as well as by the total absence of archaic as.
Ojanens findings for (South) Cambridgeshire and Poussas (1994,
2001) and Francis (1999, 2001) for Norfolk are different from the present
findings in two major respects: Their frequencies of the zero marker are
considerably higher, although the high number of zero subject relative
clauses should be taken with caution in Ojanens study. It contains
instances which share a common surface structure with zero relative clauses
yet are no relative clauses, such as nonfinite clauses, resultatives, and pro-
drop constructions.
10
In contrast to that, their figures for that are very low
(1.6% for Cambridgeshire; 7.9% in Poussas and 3.7% Francis corpus for
Norfolk versus 22% in the present data).
36 Tanja Herrmann
4.2.3. The Central Midlands Nottinghamshire
All previous studies on the Central Midlands and even its neighboring
counties report as either as the dominant or the only relative marker. Zero
and what were also known. In the present data, by contrast, the relative
marker as plays only a very minor role (2.4%). Partially this has to be seen
as a consequence of the less broad dialect spoken in the present corpus. On
the other hand, this is to be interpreted as a real time change. In the data
investigated here, as has been driven back by other relative markers,
particularly by that (40.3%), the wh-pronouns (32.7%), and zero (17.7%).
The relative marker what is used, but not frequently (5.8%).
4.2.4. The Central North (Cumbria = Cumberland, Westmorland, and
North Lancashire)
All previous authors agree on at as the regular relative marker in Cumbria,
except for Wright ([1905] 1961), who is not very detailed on his recordings
for the Central North. There is a disagreement among authors, however,
whether at is a separate relative particle of Scandinavian origin or whether
at alongside ut or t is a phonemic variant of that with the initial th-
being dropped. Although some scattered occurrences of at were found in
the present Central North data, this dispute concerning its origin cannot be
solved here. The suggestion offered on that point in Romaine (1982a),
citing R. Girvans (1939) Ratis Raving and other early Scots poems on
morals, sounds enticingly plausible: Even if at was a separate conjunction
and later a relative particle in Northern England and Scotland in the past, it
has become mentally merged with that over time, so that nowadays at no
longer figures as a separate relative particle (cf. Romaine 1982a: 70).
11
Interestingly, authors never argue about the current nature of at as a
conjunction, but it is commonly assumed that the conjunction at is
equivalent to the conjunction that, i.e. at = that. Since at and that
appear(ed) side by side, in the Central North and Scotland as well as in
other areas of regular or occasional initial th-dropping (when unstressed),
like East Anglia (cf. Poussa 1996: 529 and 531; cf. also Peitsara 2002:
169), the two were not formally distinguished, but the few transcribed
instances of at were subsumed under that. Hence, that is the most frequent
relative marker in the Central North data (43.5%), followed by zero (34%).
Occasional or frequent occurrences of zero are also mentioned by all previous
authors after 1905. While as seems to be a typical Lancashire feature (e.g. in
the SED), it was also (occasionally) observed in Westmorland and
Cumberland by Wright ([1905] 1961) and Reaney (1927) in the early
twentieth century. In the present late twentieth century data, as only amounts
to 1.4%, coming from the Ambleside area, formerly Westmorland. What is
recorded as an occasional relative marker for West Cumberland by Brilioth
(1913) and is mentioned also in the SED and by Wright (1979). What (2.4%)
was found somewhat more frequently than as(1.4%).
4.2.5. Scotland Lothian, Borders, Strathclyde, and Invernesshire
Like the present results, earlier findings document that/at (or its
phonetic/phonemic variants ut or t) as the prevalent relative marker in
Scotland (46.2% in the present data), followed by the zero relative marker
(23.6% in the present data). Wherever a reduced form occurs, authors take
it for a phonemically shortened that, with the exception of Dieth (1932:
153), who is uncertain on the matter, and Murison ([1977] 1980: 39). Grant
(1931: 78) is the only author who cites two literary instances of as
occurring around 1900, which he notes as rare. Neither in the other
studies, nor in the present investigation were there any instances of as,
whilewhat is very rare and seems to be restricted to Lower Scotland and
Glasgow, in particular. Whilk, the Scottish equivalent to English which
cited by Murison (1980: 39), rather seems to pertain to an earlier stage of
Scottish English.
4.2.6. Northern Ireland
Zero and that (or at) dominate in (Northern) Ireland in previous studies,
which conforms to the present findings (46.9% and 50.1%, respectively in
the present data). The extreme conservativeness of the present Northern
Ireland data can be read off the extremely low frequencies for who and
which.
38 Tanja Herrmann
5. Restrictiveness/nonrestrictiveness
As is well known, adnominal relative clauses can be either restrictive or
nonrestrictive modifiers of noun phrases. Restrictive relative clauses give
determinative or essential information, typically narrowing down the range
of possible referents of noun phrases (cf. Keenan and Comrie 1977: 63
64). Nonrestrictive relative clauses, by contrast, give supplementary, non-
defining information about noun phrases (cf. Quirk et al. 1985: 1239; cf.
also Huddleston 1984: 400). In contradistinction to restrictive relative
clauses, nonrestrictive relative clauses are characteristically delimited by
intonation breaks in speech and by commas (or dashes) in writing (cf.
Quirket al. 1985: 1258 and 366).
There is general agreement on the fact that, in the standard variety,
nonrestrictive relative clauses are formed by means of relative pronouns of
thewh-group. Nonrestrictive that may occur but is very rare (Quirk et al.
1985: 1258), while nonrestrictive zero cannot occur (Quirk et al. 1985:
1258).
In the present study, the dichotomy between restrictiveness and
nonrestrictiveness is determined on semantic grounds, since criteria based
on prosody, punctuation, and syntax (whether prescriptive or described on
the basis of written Standard English) are linguistic ideals that often fail to
apply in informal speech (cf. J acobsson 1994: 182183 and 191; cf. also
Newbrook 1997: 4547; Biber et al. 1999: 602). Nonrestrictiveness itself is
rather to be seen as a characteristic of (written) Standard English than of
spoken dialect. Hence, the use (and frequency of use) of nonrestrictive
relative clauses in a dialect or an idiolect is to be rated as an indicator of the
relative move toward the standard variety of that dialect or idiolect. Table 2
gives an overview over the distribution of restrictive and nonrestrictive
relative clauses across the six investigated regions.
Overall, restrictive relative clauses are four times as frequent as
nonrestrictive relative clauses (80.1% vs. 19.9%) in dialectal speech.
Generally speaking, restrictive relative clauses are predominantly formed
with invariant relative particles, whereas nonrestrictive relative clauses are
primarily constructed with wh-pronouns, especially with which. The
nonstandard relative marker what falls out of line since it takes up a
disproportionally high percentage (11.4%) of all nonrestrictives. As found
in previous dialectal studies (e.g. Ihalainen 1980, Keklinen 1985, Miller
1993, Miller and Brown 1982, Peitsara 2002; van den Eynden 1992 and
1993), which is the a priori choice for (mainly nonpersonal; see section 6)
nonrestrictives (61.9%), while personal antecedents of nonrestrictive
relative clauses are primarily relativized by who (14.8%).
Table 2. Restrictiveness/nonrestrictiveness across regions
restrictive (r) nonrestrictive (nr)
N % % r N % % nr
total
zero 687 34.2 97.3 19 3.8 2.7 706
that 942 46.8 96.2 37 7.4 3.8 979
what 113 5.6 66.5 57 11.4 33.5 170
as 19 1 100 0 0 0 19
who 179 9 70.8 74 14.8 29.2 253
which 68 3.4 18 310 61.9 82 378
whom 0 0 0 4 0.8 100 4
whose 3 0.2 100 0 0 0 3
total
6 %
2011 100.0
80.1
501 100.0
19.9
2512
100.0
In the present data, the zero relative marker and that are not confined to
restrictive relative clauses: 2.7% of all instances of zero and 3.8% of all
instances of that appear in nonrestrictive relative clauses. Below, examples
(2ac) illustrate nonrestrictive zero relative clauses; examples (2df)
illustrate nonrestrictive that relative clauses:
(2) a. ... there was Mr M. and B. W. from K. [ came round buying
horses]. ... (CNO FRED Wes_015)
b. Yes she had her aunt [ was a widow there at the time [when
she came to <gap cause=anonymization desc=address>]]
and she just lived about a year. (SCO-K6N<Person:
PSSCO28><S: 0114>)
c. ... because theres the father [ was a cook], and, and my
mother was a good cook, too, .... (NIR NITCS A19.3)
d. ... I seen E. A. [that lived there], he said it come one Sunday
dinnertime, .... (CNO FRED Wes_014)
e. And my my aunt [that I] my grandmother [that I stayed with],
their neighbour down the stair was quite indignant .... (SCO-
K6L<Person: PSSCO24><S: 121>)
f. ... J . M., [thats married to my niece there] {ahah}, his bog,
we used to cut it. (NIR NITCS A51.3)
40 Tanja Herrmann
Nonrestrictive zero relative clauses are particularly recurrent in there- and
have-existentials, which is the syntactic environment generally favored by
zero relative clauses (compare 8.5.). Whereas nonrestrictive zero is found
in four of the six regions, namely the Central Midlands, the Central North,
Scotland, and Northern Ireland, nonrestrictive that occurs in all dialects. An
older stage of English is preserved in the nonrestrictive use of that in
dialectal speech: Nonrestrictive that was common in Middle English (cf.
Mustanoja 1960: 190) and was still part of Early and Late Modern English
(cf. Chevillet 1996: 26; cf. also Mustanoja 1960: 197). Nonrestrictive
usages of the zero marker in Old and Middle English are arguable, since
they may be interpreted as instances of parataxis (cf. Mustanoja 1960: 121
and 203204).
Compared to the other relative particles, what shows a much stronger
propensity toward nonrestrictive environments. Its distribution resembles
that of the wh-pronounwho (70.8% restrictives vs. 29.2% nonrestrictives),
and the percentage of nonrestrictive what (33.5%) exceeds the percentage
of the overall nonrestrictives (19.9%).
Overall ratio of restrictives nonrestrictives: 4 : 1 (80.1% 19.9%)
vs.
Ratio of what restrictives what nonrestrictives: 2 : 1 (66.5% 33.5%)
Figure 2. Ratio of (non)restrictives versus ratio of what (non)restrictives
As will be demonstrated in section 6, what is not gender-marked and thus
can relativize personal and nonpersonal antecedents alike (cf. Table 3
below). On closer examination, however, there are only nine instances
(15.8%) of nonrestrictive personal what relative clauses in the data. In other
words, what mainly competes with nonrestrictive which for the same
syntactic environment (viz nonpersonal antecedents), since the vast
majority (84.2%) of all nonrestrictive instances of what are nonpersonal.
By and large, in those regional dialects that have what as a relative marker,
broad idiolects employ nonrestrictive what, while idiolects that are closer to
the standard opt for nonrestrictive which.
The few instances of the relative particle as (19 occurrences; a mere
0.8% of all relative markers) are all found in restrictive environments. This
may be a result of the extremely low text frequency of as, i.e. as is
numerically receding in all syntactic environments, including the
confinement to exclusively restrictive environments (cf. also 8.3.2).
Personal relative pronoun who prefers restrictive environments (70.8%),
whereas its (mainly) nonpersonal counterpart which runs counter to all
other relative markers by greatly favoring nonrestrictive environments
(18% restrictives vs. 82% nonrestrictives). Finally, with regard to the
highly marked use of whomand whose in our data (as few as four and three
instances out of a total of 2512 relative clauses) it may be noted that whom
seems to prefer nonrestrictive environments, while whose seems to follow
the general trend in preferring restrictive environments.
6. Personality/nonpersonality
With regard to the personality/nonpersonality variable, the six regions show
the following behavior:
Table 3. Personality/nonpersonality across regions
personal (+p) nonpersonal (-p)
N % % +p N % % -p
total
zero 251 25.9 35.6 455 29.5 64.5 706
that 392 40.5 40.0 587 38.0 60.0 979
what 49 5.1 28.8 121 7.8 71.2 170
as 10 1.0 52.6 9 0.6 47.4 19
who 244 25.2 96.4 9 0.6 3.6 253
which 16 1.7 4.2 362 23.5 95.8 378
whom 4 0.4 100 - 0 0 4
whose 3 0.3 100 - 0 0 3
total
6 %
969
38.6
100.0 1543
61.4
100.0 2512
100.0
All six dialects conform to Standard English in restricting who and its case-
marked forms whomand whose to personal antecedents. The nine cases of
who relativizing nonpersonal antecedents belong to the sanctioned
borderline cases, such as personalized animals and things (cf. Quirk et al.
1985: 12451246 and 314318): Three instances refer to domestic animals
like cows and horses, as in (3), one to a car, and five to collective nouns.
(3) you had to have a cowthe top side [who d make the other one stay
her own side]. ... (CSW-SRLM 109<T 3320>)
42 Tanja Herrmann
However, the conventional gender contrast between which and who (and its
case-marked forms) is overridden in dialectal English insofar as the gender
concord constraint is lifted, or at least loosened, in the case of which: Apart
from two instances denoting collective nouns and two referring to older (!)
children, there are twelve instances of personal which outside the group of
licensed borderline cases of personal which.
12
Consider the examples in (4):
(4) a. ... And the boy [which I was at school with] G. O., .... (CNO
FRED Wes_019)
b. Then, course then they used to <trunc> hav when the man
stood in green, I knew a fella named <gap
cause=anonymization desc=last or full name>[which we
was talking about a little while ago] and .... (EAN-
H5H<Person: PSEAN4><S: 501>)
c. ... And then there was C., [which caught his hand in the
machinery up here] and he had his hand off, being severed (at
the) wrist. ... (CSW FRED Som_009)
Personal which occurs in five of the six regions: in the Central Southwest,
East Anglia, the Central Midlands, the Central North, and Scotland. (The
data from Northern Ireland, which hardly has any wh-pronouns, only
contains five instances of nonpersonal which.) In other words, the relative
marker which is not confined to nonpersonal antecedents in dialects. In
allowing nonpersonal and personal antecedents, which seems to have lost
its status as a relative pronoun in dialect: Which relies on gender concord as
the only kind of rigorously applied agreement between antecedent and
relative marker for being considered a relative pronoun, as which shows
neither case nor number agreement with the antecedent. Thus, in dialect
which qualifies rather as a relative marker (relativeparticle): It is open to
any type of antecedent. In this respect, dialectal English has maintained a
prior stage of English, since Middle English dialects permitted nonpersonal
which (cf. Moss [1952] 1991: 62; cf. also Mustanoja 1960: 195), while in
the sixteenth century which started to be confined to nonpersonal referents
(cf. Nevalainen and Raumolin-Brunberg 2002: 117 and 119). However, as
illustrated in examples (5a,b), non-gender-marked relative marker which
can still be governed by a preposition:
(5) a. ... and we had a meeting at headquarters, to which about forty
or fifty people turned up, [of which two had probably blown a
<pause> an instrument in the past], ... (EAN-K69<Person:
PSEAN13><S: 130>)
b. But the yard one of the yard inspectors came to me and said, I
wonder if you would make up a roster for the supervisors
(UNCLEAR). (SCO-K6M<S: 304>)
[Of which there were six, seven]. (<S: 305>)
So I made a roster out for the (trunc) ro for the supervisors
which meant that every week they got a rest day. (<S: 306>)
Being able to be governed by a preposition is a defining criterion of nouns
and pronouns, in contrast to conjunctions/complementizers (relative
particles), which cannot be governed by a preposition (e.g. *of
PREPOSITION
that
RELATIVE PARTICLE (CONJ UNCTION)
). Hence, non-gender-marked relative marker
which has to be attributed the status of relative pronoun.
The occurrence of nonpersonal which in combination with a partitive
genitive appears to be triggered by two possible (and perhaps mutually
reinforcing) factors: First, according to Christian Mair (personal
communication), which is chosen in relativizing human antecedents to
activate its ascriptive property, i.e. when speakers want to express the
kind of person, instead of the identity (cf. Huddleston and Pullum 2002:
10481049; cf. also Quirk et al. 1985: 1246).
13
For example:
(6) Remember that they have a house-keeper, [which we dont have].
(Huddleston and Pullum 2002: 1049)
Mair (1998: 130132) focuses on the relaxation of a grammatical
agreement rule concerning the who/which opposition, in analogy to the
who/which distinction with collective nouns: The use of which [in these
cases] de-emphasises the status of members of a particular group as
individuals and presents them as a collective (1998: 130). This is
illustrated in (7):
(7) It [the proposed law] is aimed at the external hacker [of which
there are far fewer than press reports suggest] .... (Mair 1998: 130
[emphasis and bracketing mine])
Second, this combination of partitive genitive which with personal
antecedents is rather due to an analogy mechanism: In interrogative clauses
and nominal relative clauses, the relative pronoun which contrasts with the
44 Tanja Herrmann
relative pronouns who and what in denoting a (personal or nonpersonal)
limited set, which is semantically close to the concept expressed by the
partitive genitive of which. For example:
(8) a. indefinite interrogative pronoun who:
Who is your favourite conductor?
b. definite interrogative pronoun which referring to a limited set:
Which is your favourite conductor? (Von Karajan or
Stokowsky?) (Quirk et al. 1985: 369)
c. indefinite nominal pronoun what:
You can take [what you want].
d. definite nominal pronoun which referring to a limited set:
You can take [which you want]. (X or Y)
By contrast, adnominal which is restricted to nonpersonal antecedents, even
if it describes a definite, limited set of referents in a partitive genitive (cf.
Quirk et al. 1985: 369370; cf. also Huddleston 1984: 394). In this way,
the semantic property limited set is transferred from nominal relative or
interrogative clauses to adnominal relative clauses by analogy, as in the
examples in (5), repeated here as (5'a,b):
(5') a. ... and we had a meeting at headquarters, to which about forty
PSEAN13><S: 130>)
b. But the yard one of the yard inspectors came to me and said, I
wonder if you would make up a roster for the supervisors
(UNCLEAR). (SCO-K6M<S: 304>)
[Of which there were six, seven]. (<S: 305>)
So I made a roster out for the (trunc) ro for the supervisors
which meant that every week they got a rest day. (<S: 306>)
Where the dialectal relative marker as occurs (i.e. in the Central Midlands,
the Central North, and Northern Ireland), it is not sensitive to the
personality/nonpersonality variable, but combines with personal and
nonpersonal antecedents almost in equal numbers. Below, example (9a)
contains a personal antecedent, while example (9b) contains a nonpersonal
antecedent:
(9) a. ... I was on this er bottle washing stunt and (trunc) o one chap
[as lived next door to us, back at er at <gap
cause=anonymization desc=last or full name> Road] he
got me his this job on the farm. (CMI-FYE<S: 166>)
b. ... theres a lot of houses [ would be up that road] [as wasnt
there]. (NIR NITCS A41.3)
Unlike the nominal relative pronoun what in (10a), adnominal nonstandard
what is not limited to nonpersonal antecedents in any region. About one
third of all instances (28.8%) refer to a personal antecedent, as in (10bd):
(10) a. nonpersonal relative pronoun what in nominal relative clause:
prepositionin is elided (cf. 10.1. below):
I believe in wearing [what you feel comfortable [in]], you
work better and ... (EAN-K65<S: 1059>)
b. ... And the lady [what was driving], she said, Excuse me
young man, she said, where s your dog? ... (CSW-SRLM
132<T 1280>)
c. See he was the man [what brought in decasualization during
the war]. (EAN-H5H<Person: PSEAN4><S: 767>)
d. And er she used to sell corned beef and er pickled onions or
anything like that for people [what was working round there]
[what couldnt get home for dinner]. (CMI-FYD<S: 201>)
7. Preposition placement
When a prepositional complement is relativized, the preposition can either
be moved to initial position together with the wh-marker and govern the
relative pronoun (preposition fronting or pied-piping), or the preposition is
left behind in its normal clause position without its complement
(preposition stranding). Both types are illustrated in turn:
(11) a. fronting:
Now they were provided with a meal [for which the police, at
that time paid sixpence], .... (EAN-K68<S: 030>)
b. stranding:
... our shovels [what they used to feed the boiler with] were
all steel shovels. (EAN-H5G<S: 505>)
46 Tanja Herrmann
Table 4 presents the distribution of fronted versus stranded prepositions
across regions.
Table 4. Preposition fronting/preposition stranding
fronted N stranded N
tokens % tokens %
total
zero 0 - 61 35.5 61
that 0 - 70 40.7 70
what 0 - 14 8.1 14
as 0 - 1 0.6 1
who 0 - 5 2.9 5
which 11 91.7 20 11.6 31
whom 1 8.3 1 0.6 2
whose - - - - -

total
6 %
12
6.5
100.0 172
93.5
100.0 184
100.0
While pied-piping is typical of (written) Standard English, dialects vastly
prefer preposition stranding (172 of 184 prepositions are stranded; 93.5%).
To a large extent, this result is a corollary of the relative markers used:
Only relative pronouns allow both preposition fronting (11a) and
preposition stranding (11b). Relative particles demand preposition
stranding, because in relative clauses neither conjunctions can be governed
by prepositions (e.g. *of that, *of what; *of as) nor gaps (e.g. *of
'): recall the discussion concerning the word class status of relative
markers in section 6. Accordingly, if adnominal what were found to be
governed by a preposition, i.e. in an example of preposition fronting, this
would attest to its status of a relative pronoun, parallel to its nominal
counterpart what.
In all instances where the relative particles that, what, as and the zero
marker figure as prepositional complements, the prepositions are invariably
stranded (altogether 146 instances). Of the remaining 38 relative pronouns
(i.e. the prepositional complements who, which, and whomtaken together),
twelve (31.6%) show preposition fronting while 26 (68.4%) show
preposition stranding. In other words, even where preposition fronting is
permitted, preposition stranding is preferred in dialectal speech. Regarding
the relative pronoun who, the absence of fronted prepositions (and the
presence of five stranded prepositions) is conspicuous. Apparently, the
relevant speakers (two in East Anglia, one in the Central Midlands, and one
in the Central North) neither wanted to apply case-marking (i.e. whom), nor
did they want to have a non-case-marked relative pronoun governed by a
preposition (e.g. with who). For example:
(12) a. J anet posted a letter for me last week to a friend [who I
workedwith at Ipswich], ... (EAN-HDK<S: 072>)
b. Well the agent, that be either <gap cause=anonymization
desc=last or full name> or any agent who, [who the ship
belongedto], ... (EAN-H5H<Person: PSEAN4><S: 372>)
There are two instances involving whom. In example (13a) the preposition
for is fronted, whereas in example (13b) it is stranded:
(13) a. But anyway there was so many people and one chap who he
he was, as a matter of fact, he was organizer with Communist
Party [for whom Ive got the very greatest respect], the very
greatest respect. (CMI-FYH <S: 300>)
b. And er this bicycle well it would go out of fashion and was put
in a, a loft in one of the, it must have changed hands fromMr
<gap cause=anonymization desc=last or full name>
[whom it was made for]. (SCO-GYW<S: 093>)
Out of the 31 relative clauses introduced by the relative pronoun which,
eleven instances (35.5%) show preposition fronting, while the majority
(64.5%) shows preposition stranding.
Of the eleven fronted prepositions governing which, six are partitive
genitive of (governing also personal which, as in example (14b) below).
(Partitive) genitive of should always be fronted, according to a
prescriptive rule in (written) Standard English. J espersen (1927: 188)
phrases this rule as there are certain cases in which it is unnatural to have
of at the end, thus if it is the equivalent of a genitive and [p]artitive of is
generally placed first (J espersen 1927: 188), while Huddleston and Pullum
(2002: 1041) describe it as [p]artitive of resists stranding. For example:
(14) a. Most of the textbooks [ were handed to us] came from
cupboards of storage, [of which they must have had about
48 Tanja Herrmann
sixty each], those classes were always, fromthen on, sixty
boys in a class for one teacher. (CMI-FXU<S: 035>)
b. ... and we had a meeting at headquarters, to which about forty
PSEAN13><S: 130>)
In dialectal speech, however, we do find examples of stranded of-genitives,
in combination with which (15a) or even the zero relative marker (15b),
which requires preposition stranding anyhow:
(15) a. resumptive personal pronoun them after stranded partitive
genitive preposition of (preferred analysis):
Now the first job he did was to get you well acquainted with
the tools, [which we had quite a number of them]. (CMI-
FY2<S: 003>)
b. Well you know they used, did you know there used to be a
timber [ they used to make cog wheels of]? (CNO FRED
Wes_004)
8. Accessibility Hierarchy
8.1. Preliminaries and overall Accessibility Hierarchy frequencies
The strong form of the Noun Phrase Accessibility Hierarchy (in the
following short Accessibility Hierarchy or AH) and the Hierarchy
Constraints was put forward by Keenan and Comrie in 1977, stating
universal principles in relative clause formation (cf. Keenan and Comrie
1977: 6667):
Accessibility Hierarchy (AH):
SUBJ >DO >IO >OBL >GEN >OCOMP
Spelled out as follows:
SUBJ ECT >DIRECT OBJ ECT >INDIRECT OBJ ECT >OBLIQUE
CASE >GENITIVE >OBJ ECT OF COMPARISON
The Accessibility Hierarchy reflects the accessibility of various syntactic
positions of noun phrases to relativization (in restrictive relative clauses
with definite head noun phrases) in descending order from left to right. In
psychological terms, the further a position is located to the left, the easier it
is to relativize. The Hierarchy Constraints are phrased as follows:
Hierarchy Constraints (HCs):
1. A language must be able to relativize subjects.
2. Any R[elative]C[lause]-forming strategy must apply to a
continuous segment of the AH.
3. Strategies that apply at one point of the AH may in principle cease
to apply at any lower point.
Hierarchy Constraint 1 bars relativization on a lower position than subject
without also having subject relative clauses in a language. Hierarchy
Constraint 2 forbids leaving out a position. Hierarchy Constraint 3
maintains that any position is a potential cut-off point for a relative clause
formation strategy (cf. Keenan and Comrie 1977: 68).
Later in the article, Keenan and Comrie restricted the Constraints to
primary strategies (in a language). A relative clause formation strategy is
calledprimary if it can relativize the subject position. If a primary strategy
can relativize a low position on the Accessibility Hierarchy (e.g. GEN
position), this strategy will be able to relativize all higher positions (cf.
Keenan and Comrie 1977: 6869).
Over the years the formulation of the Hierarchy Constraints has
undergone several revisions, while trying to come to grips with various
(real and apparent) counterexamples to the Accessibility Hierarchy and its
Constraints. However, the ranking of syntactic positions on the
Accessibility Hierarchy in terms of accessibility of individual relative
clause formation strategies to these positions has remained valid,
provided that a language has this particular syntactic position in its
grammatical make-up (cf. Keenan and Comrie 1977: 66).
For the present investigation of dialects of English, some minor
modifications to the Accessibility Hierarchy were made on theoretical and
also practical grounds:
The position OCOMP[ARISON] was eliminated for two reasons: First,
than can function as a conjunction (particle) and as a preposition (cf.
Maxwell 1979: 367; cf. also Comrie and Keenan 1979: 662), as in than
whom (cf. Simpson and Weiner 1989, OED, 17: 861). Hence,
50 Tanja Herrmann
OCOMP[ARISON] could be subsumed under the position
PREPOSITIONAL COMPLEMENT (cf. below). In addition, OCOMP is
not sensed as a grammatically identifiable position in the same way as the
relative clauses that are higher than OCOMP (Keenan and Hawkins 1987:
64) but rather conflated with OBL (cf. also Keenan and Hawkins 1987: 82
83). Second, even Keenan and Comrie (1977: 90 and 74) grant that
examples like the man who J ohn is taller than are of marginal acceptability.
In the present dialect data, there was no instance of an object of
comparison.
Although indirect objects are used in English (in dialects as well as in
the standard variety), this position is commonly taken up by prepositional
objects (i.e. objects of prepositions (OP)) in declarative clauses: He gave
her
IO
the book he gave the book to her
OP
. Resulting from that and from
the fact that English is one of the languages which [f]or purposes of
relative clause formation, ... assimilate indirect objects to the other oblique
cases (Keenan and Comrie 1977: 72), indirect objects are very
infrequently relativized (six instances in the entire data). Therefore, the
syntactic position IO was subsumed under a bracketed group called
OTHER, whose miscellaneous members are outside the scope of this
version of the Accessibility Hierarchy.
The syntactic position OBL (oblique case) was renamed PCOMP
(prepositional complement). In the initial set-up, there was a distinction
between prepositional complements functioning as objects of prepositions
(PCOMP(OP)) and prepositional complements functioning as adverbials
(PCOMP(A)). In addition to that, a third subgroup of prepositional
complements was distinguished, viz prepositional complements functioning
as (of)-genitives (PCOMP(GEN)). Since, in the course of the analysis, the
three subgroups seemed to behave alike, all three were collapsed into one
cover group PCOMP. At the same time, adverbial relative clauses were
considered as a separate type of adnominal relative clauses, which do not
enter the Accessibility Hierarchy as they are structurally and conceptually
different. Proper adnominal relative clauses can postmodify noun phrases
with any kind of lexical meaning and their relative markers can serve all
sorts of grammatical functions in the relative clause (i.e. subject, object,
complement; adverbial). Adverbial relative clauses, by contrast, post-
modify noun phrases with the lexical meaning of a spatial, temporal,
causal, or modal adverbial; moreover, their relative markers (including
relative adverbs such as where, when, or why) only serve adverbial function
in the relative clause.
Furthermore, infrequent positions like subject complements (SCOMP),
as in (16a), and object complements (OCOMP), as in (16b), were grouped
with indirect objects and so-called nonprepositional adverbials (A), as in
(16c), to constitute the cover group OTHER:
(16) a. ... You can charcoal any timber [ there is] regarding
different species. (CNO FRED Wes_004)
b. Quarterly meetings [which they call them] that was held in er
the Odd Fellows Hall in Forest Road .... (SCO-K7G<S: 359>)
c. elided stranded preposition with makes relative marker that
function as A (instrumental) instead of PCOMP(A):
And never in my wildest dreamdid we imagine that Labour
would get in with the resounding majority [that they did get in
with]. (SCO-K6M<S: 010>)
In conclusion, the modified version of the Accessibility Hierarchy adopted
here is outlined in Figure 3:
SUBJ >DO >PCOMP >GEN (OTHER)
PCOMP(OP) PCOMP(A) PCOMP(GEN) SCOMP OCOMP IO A
Figure 3. Modified version of Accessibility Hierarchy
The overall frequencies given in Table 5 confirm the Accessibility
Hierarchy in all six regions (Central Southwest, East Anglia, Central
Midlands, Central North, Scotland; Northern Ireland). (Frequencies of the
group OTHER, which is not part of the Accessibility Hierarchy, are given
for the sake of completeness). In all regions, subjects are by far most
frequently relativized (mean 63.2%), followed by direct objects (mean
27.2%), prepositional complements (mean 7.3%), and genitives. The latter
are found in only one region, where they account for less than 1% (mean
0.1%). The heterogeneous group OTHER varies between 1% and 3.6%
across regions: Its 54 tokens are composed of 21 subject complements,
eleven object complements, six indirect objects, and 16 nonprepositional
adverbials.
52 Tanja Herrmann
Table 5. Overall frequencies
SUBJ > DO > PCOMP > GEN (OTHER)
N N N N N
% % % % %
total
RCs
CSW 165
56.7%
94
32.3%
27
9.3%
0
0%
5
1.7%
291
EAN 247
58.5%
134
31.8%
34
8.1%
0
0%
7
1.7%
422
CMI 298
65.9%
117
25.9%
25
5.5%
3
0.7%
9
2
452
CNO 285
68.2%
97
23.2%
21
5%
0
0%
15
3.6%
418
SCO 332
63.6%
128
24.5%
48
9.2%
0
0%
14
2.7%
522
NIR 260
63.9%
114
28%
29
7.1%
0
0%
4
1%
407
total 1587
63.2%
684
27.2%
184
7.3%
3
0.1%
54
2.2%
2512
8.2. Individual relative clause formation strategies
While Keenan and Comrie (1977: 76) recognize only two strategies
(+case/-case) for (Standard) English (cf. also Comrie and Keenan 1979:
656), more precise subdivisions were made for the present data. Each
relative marker of the two major groups declinable relative pronoun (wh-
relative pronouns) and indeclinable relative particles accounts for a
distinct strategy (with the exception of whomand whose). In contrast to
that, Keenan and Comries pronoun retention (in combination with a
relative marker at the beginning of the relative clause) amounts to a
separate relative clause formation strategy, which will be investigated in
section 9. Wh-pronouns were split into the two gender-opposed components
who and which, which surface in all six regions in the data. The case-
marked wh-pronouns whomand whose only appear in Scotland and the
Central Midlands (SCO: two whom; CMI: two whom; three whose). Whom
and whose indeed have to be seen as the case-marked forms of who, since
all seven instances refer to personal antecedents. In Tables 6 and 7 below,
the occurrence of a relative marker in a syntactic position of the
Accessibility Hierarchy is indicated by a plus sign [+]; nonoccurrence by a
minus [-]. Absolute numbers of the total occurrences of who and its case-
marked forms in all regions are given thereafter. The feature [+p] indicates
that who, whom, and whose are wh-pronouns marked as referring to
personal antecedents only. Featuring whose as [rp] would be permissible in
principle but is not borne out by the present data:
Table 6. (Non)occurrence and frequencies for wh-pronoun [+p]
SUBJ > DO > PCOMP > GEN
who whom whom whose
wh-pronoun [+p] + + + +
total 244 1 3 3
On the other hand, who is not restricted to nominative case/subject position.
In dialects, it can (even though only very rarely) relativize all syntactic
positions (SUBJ , DO; PCOMP) except GEN (if the antecedent is personal).
Table 7. (Non)occurrence and frequencies for who
who + + + -
total 244 4 5 -
(17a and b) show who in direct object position and examples (17c and d) in
prepositional complement position:
(17) a. Mr H. went in to (gap place name), and he met an old man
there [who
DO
he knows]: How today, Tommy? ... (CNO FRED
Wes_019)
b. All those positions were going for want of an application, so I
applied and because <gap cause=anonymization desc=last
or full name> brought in people [who
DO
he knew], certain of
themwere automatically filled .... (EAN-HDL<S: 335>)
c. ... The chap [who
PCOMP
I got it off], he says, I ll come up on
Monday .... (CNO FRED Wes_003)
d. J anet posted a letter for me last week to a friend [who
PCOMP
I
worked with at Ipswich], ... (EAN-HDK<S: 072>)
54 Tanja Herrmann
Which (predominantly but not exclusively nonpersonal) exhibits the
following distribution on the Accessibility Hierarchy, according to regions:
Table 8. (Non)occurrence and frequencies for which
which + + + -
CSW 17 15 4 -
EAN 80 17 12 -
CMI 64 18 5 -
CNO 36 10 1 -
SCO 58 11 7 -
NIR 4 0 1 -
The numerical distribution of which supports the Accessibility Hierarchy
very well (disregarding DO in Northern Ireland, whose overall number of
which relative clauses is too low anyway). Since the occurrence and
frequency of wh-pronouns serve as a yardstick of traditionality or degree of
standardization of speech, respectively, the remarkable variance in subject
position (ranging between 80 instances in East Anglia and four instances in
Northern Ireland) is to be ascribed to the varied quality of the corpora used
here in this respect. Whereas our corpora for East Anglian, the Central
Midlands, Scotland, and the Central North also contain data from speakers
whose dialect is less broad, the Central Southwest and Northern Ireland
corpora only consist of data from broad dialect speakers.
A look at the frequencies of the relative particle that also corroborates
the Accessibility Hierarchy. That can relativize all positions except GEN
(see Table 9).
Although various dialectologists (e.g. Romaine 1980: 227, Seppnen
and Kjellmer 1995, Aitken 1979: 105, Newbrook 1997: 41; Harris 1993:
150151; Comrie 1999: 87) and surveys (SED, Lowman Survey; Survey of
Anglo-Welsh Dialects) cite thats as a dialectal GEN relative marker, the
blank in the paradigm cannot be filled by the present data. Thats
developed from the invariant relative particle that +the reduced form of the
possessive pronouns his or its into a general genitive relative pronoun
thats representing also feminine and plural antecedents (cf. Seppnen and
Kjellmer 1995: 397398). At least this seems to apply to thats in Scottish
English, as in The woman thats sister mairriet the postie (from Grant and
Murison 1974, The Scottish National Dictionary, IX: 265) or the people
thats houses were demolished (from Aitken 1979: 105).
Table 9. (Non)occurrence and frequencies for that
that + + + -
CSW 58 14 4 -
EAN 58 24 10 -
CMI 125 49 7 -
CNO 145 32 4 -
SCO 157 54 28 -
NIR 137 49 17 -
The zero marker demands a more differentiated treatment, since it is
ungrammatical as the subject of a relative clause in Standard English (cf.
Quirk et al. 1985: 1250). In dialects, however, the zero marker can function
as subject. At the same time, Table 10 shows that it is only in two of the
northern regions, the Central North and Northern Ireland, that the subject
position is easiest to relativize in all regions and thus corresponds to the
Accessibility Hierarchy.
Table 10. (Non)occurrence and frequencies for zero
+ + + -
CSW 29 43 12 -
EAN 17 63 5 -
CMI 27 46 7 -
CNO 68 48 14 -
SCO 45 62 12 -
NIR 112 65 11 -
Occurrence and frequency of zero subject relative clauses can be tied to
traditionality and/or colloquiality of speech. Belonging to the more
traditional corpora, the Central North and the Northern Ireland data
illustrate that traditional dialects can reestablish the proper order of
positions along the Accessibility Hierarchy. Less traditional dialects cannot
do that but they also remedy the situation on the Accessibility Hierarchy by
filling the gap in SUBJ left by Standard English.
The relative particle what occurs in all regions except in Northern
Ireland. In Scotland, there are only two occurrences of what, one of which
is doubtful; the other is a clear case from Glasgow. The slight discontinuity
56 Tanja Herrmann
in the series between DO and PCOMP in CMI must be attributed to the
very low figures. Analogous to that, what serves all positions except GEN
in the data. What and whats in genitive position were recorded in the SED
(cf. Upton, Parry, and Widdowson 1994: 490), while whats was also
recorded in a survey conducted at schools in the 1980s (cf. Cheshire,
Edwards, and Whittle 1993: 69). In the present data, however, there are no
instances of what or whats in genitive position.
Table 11. (Non)occurrence and frequencies for what
what + + + -
CSW 36 21 7 -
EAN 32 28 4 -
CMI 19 1 2 -
CNO 5 4 1 -
SCO 2 0 0 -
NIR 0 0 0 -
The traditional relative marker as is still to be found in three of the six
regions (Central Midlands, Central North, and Northern Ireland). On
account of its very low totals, the sheer (non)occurrence of as in a position
via a +/- sign should be focussed on.
Table 12. (Non)occurrence and frequencies for as
as + + + -
CSW - - - -
EAN - - - -
CMI +(7) +(3) +(1) -
CNO +(3) +(2) - -
SCO - - - -
NIR +(2) - - -
8.3. Interpretation
8.3.1. Who, which, that, and zero
In the present data, the wh-pronoun strategy who is the only strategy which
can relativize all positions on the Accessibility Hierarchy including GEN. It
either emerges in the shape of the case-marked forms who, whom, or whose
or in the shape of the non-case-marked form who in SUBJ , DO, and
PCOMP. That is to say, GEN is the only position that requires an explicit,
case-marked form in the very few cases when it appears (cf. also 8.4.). Both
personal and nonpersonal which are well represented in all syntactic
positions except GEN. Nevertheless, Romaines and Dekeysers results for
Middle Scots and Early Modern English, respectively, are not borne out by
the present dialect data: With the exception of GEN, wh-pronouns do not
dominate the lower end of the Accessibility Hierarchy; likewise relative
particles (that and zero) are not restricted to the higher positions (cf.
Romaine 1980: 228; cf. also Dekeyser 1984: 76). Actually, the frequencies
involving relative particles outnumber those involving wh-pronouns by far
from twice the number in the Central Midlands to almost 40 times as
many in Northern Ireland. In particular, that and zero prevail over wh-
pronouns in the less accessible positions DO and PCOMP. Whereas that is
also prevalent in SUBJ , certain allowances have to be made for zero in
subject position. Although zero relativizes subjects in all regions, it does
not always have its stronghold there. In Keenan and Comries framework,
zero is rehabilitated as a primary relative clause formation strategy in
dialects, because of its ability to relativize subjects. (The least broad East
Anglian corpus, by contrast, is an exception to the rule in various ways: Its
ratio of wh-pronouns to relative particles is just 174 : 243. Although that is
also frequent in SUBJ and PCOMP, which is the most frequent relative
marker in these two positions in East Anglia.)
8.3.2. Change in progress in terms of the Accessibility Hierarchy: what
andas
The nonstandard relative particles what and as seem to be following
opposite developments (once again, East Anglia is a special case): What is
on the rise (as a supra-regional nonstandard relative marker), while as is on
the decline (as a regional dialect relative marker).
58 Tanja Herrmann
From its southeastern heartland East Anglia including Essex (cf. Poussa
1988: 448; cf. also Viereck 1975: Maps 207 and 208), what has been
radiating out through the adjoining Midlands and the Home Counties,
especially London, to the Southwest and, eventually, to the north (cf.
Poussa 1988: 448 and 450; cf. also Cheshire, Edwards, and Whittle 1993:
64; Viereck 1991: S8b, S9, and S10; Poussa 1991: 311).
The frequencies in the present data reflect the process of dissemination.
What is most frequent in the South, less frequent in the Midlands, and least
frequent in the North. In our day, what spreads via the big cities (cf.
Cheshire, Edwards, and Whittle 1993: 68), for which Glasgow is a case in
point. In short, what has infiltrated all investigated areas, except for the
countryside of Northern Ireland.
The Central Midlands data suggest that what enters the Accessibility
Hierarchy by the subject position and gradually works its way down the
hierarchy. Cheshire, Edwards, and Whittle (1993: 6970) come to the same
conclusion on the basis of results from questionnaires in a nation-wide
survey at schools. They observe an implicational hierarchy, such that all
schools reporting the occurrence of what as a genitive pronoun also
reported what as object pronoun, and all schools reporting what as object
pronoun also reported what as subject pronoun (Cheshire, Edwards, and
Whittle 1993: 69). Being a hallmark of Present-Day Standard English, the
wh-pronouns entered the Middle English relative marker system by the low
positions on the Accessibility Hierarchy in formal and complex written
language (cf. Romaine 1980: 234; cf. also Dekeyser 1984: 76). By contrast,
the nonstandard relative marker what introduces itself into the Accessibility
Hierarchy via the top end, because it is part of an informal straightforward
spoken code, which has greater affinity to the simpler positions of the
Accessibility Hierarchy (cf. Cheshire, Edwards, and Whittle 1993: 70).
What appears to have an even firmer grip on the Central Southwest (i.e.
Eastern Somerset). Of all relative markers, what is recorded to be second
strongest in the lower positions DO and PCOMP, as well as in SUBJ .
Althoughwhat originated in East Anglia, it is said not to be thriving in
its place of origin and the prognoses of its future are not favorable. Unlike
in urban centers, where it carries covert prestige among the younger
generation, what has the stigma of an old and vulgar relative marker in East
Anglia, which makes it unpopular among the present young and middle-
aged generation, or even older people (cf. Poussa 1988: 443444; cf. also
Poussa 1996: 530; Poussa 1994: 419422). However, among the traditional
dialect speakers investigated here, what still proves to be the indigenous
relative marker of East Anglia: Its overall number is fairly high and what can
relativize lower positions on the Accessibility Hierarchy with equal ease.
In opposition to what, as has become a relic or completely disappeared,
such as in the Central Southwest and East Anglia. Where it is still present,
as seems to have retreated to the higher positions on the Accessibility
Hierarchy, from which it will probably exit via the subject position. In
addition, as appears to be confined to restrictive syntactic environments
nowadays. The relative marker as once was almost the only relative marker
existing in the Midlands (cf. LAE: Map S5). Hence, it is not surprising that
the only instance of as in PCOMP in the data is found in the Central
Midlands. In comparison to a total of 26 instances of what in the Central
Midlands, there are just eleven instances of as. In the Central North, ten
instances of what are faced with six instances of as. That is to say, what
seems to be driving as out.
8.4. Genitive avoidance
Only three instances (0.1%) of whose are recorded in the data, all of them
in the Central Midlands. In other words, synthetic genitives are avoided in
dialect. Only when prompted for a genitive relative marker as in the
questionnaire of the Survey of English Dialects speakers tend to use the
standard relative pronoun whose. The use of whose requires a coreference
relationship between the antecedent and a determiner (possessor) of a noun
phrase in the relative clause, which is more complex than a simple
antecedent relative noun phrase relationship (cf. Givn 1993: 133).
According to Keenan and Comrie (1977: 69), unrelativizable NPs can be
systematically promoted to higher positions on the Accessibility Hierarchy,
whence they can be relativized. Thus, dialect speakers may circumvent
synthetic genitives by promoting a potential genitive to a higher position on
the Accessibility Hierarchy. The following possibilities are available:
Possibility 1: In theory, speakers could promote a potential synthetic
genitive to PCOMP (PCOMP(GEN) position) by using an analytical of-
genitive. With possessive genitives, however, this was never done in the
investigated dialect material. J ohansson (1993) analyzed the spoken part of
the Birmingham Corpus (1,3 million words of (nondialectal) British
English; mainly from 19601981) with regard to whose and of which with
nonpersonal antecedents (cf. examples [18a] and [18b], taken from Quirk et
60 Tanja Herrmann
al. 1985). Her findings showed that whose is unpopular but of which is
even more unpopular (cf. J ohansson 1993: 112).
(18) a. The house [whose roof was damaged] ... (Quirk et al. 1985:
12491250 [emphasis and bracketing mine])
b. The house [the roof of which was damaged] ... (Quirk et al.
1985: 12491250 [emphasis and bracketing mine])
c. The house [of which the roof was damaged] ...
Possibility 2: In the literature on dialects, analytical genitives, which are
formed by means of a resumptive pronoun (cf. section 9.1. below), figure
very prominently (e.g. Edwards, Trudgill, and Weltens 1984: 27, Ihalainen
1985: 66, Miller 1993: 111 for Scottish English, Harris 1993: 150151 for
Irish English; Comrie 1999: 87; cf. Tables 1520 in Appendix 1). In these
cases, the relativized noun phrase is promoted from the genitive to the
indirect object position. The possessor is expressed by a resumptive (i.e.
possessive) pronoun. In the present data, an analytical genitive appeared
just once:
(19) a. But (trunc) y youd got to watch, there again, that er you
didnt exceed the width of er of your waggon, [which
DATIVE
its
POSSESSIVE PRONOUN
maximumlimit was er would be er eight
foot three, or er eleven foot six, high]. (CMI-FY2<S: 026>)
The fact that the antecedent in this example is nonpersonal may
additionally have favored the employment of nonpersonal relative pronoun
which + neuter possessive pronoun its instead of whose, which is
predominantly applied with personal antecedents (cf. Quirk et al. 1985:
366).
Possibility 3: (Re)constructing a proposition as a possessive have or get
construction (example [20b]) or as a paratactic attributive with construction
(example [20c]) are two well-described genitive avoidance strategies in the
literature (e.g. Quirk et al. 1985: 1249, Keenan and Comrie 1977: 9091,
J ohansson 1993: 112, Ihalainen 1985: 66, Elworthy 1877: 42; Kruisinga
1905: 38). In possessive have or get constructions, potential genitives are
promoted to subject position while a resumptive pronoun expresses the
possessor. Attributive with constructions avoid relative clause formation
altogether.
(20) a. The house [whose roof was damaged] ... (Quirk et al. 1985:
12491250 [emphasis and bracketing mine])
b. The house [that had its roof damaged] ....
c. The house with the damaged roof ....
Whereas no clear case of a possessive have or get construction avoiding
GEN position was encountered in the present data, there was a handful of
attributivewith constructions. For example:
(21) instead of: Had a gun, you see, [whose cartridge was blank]:
Had a gun, you see, with a blank, blank cartridge in. (EAN-
K65<S: 0495>)
It can be concluded from the present investigation that dialect speakers
usually resort to paratactic constructions to describe a genitive sense
relation, rather than promote a relativized noun phrase up the Accessibility
Hierarchy. Common parataxes, and-coordinations, and left dislocations are
used in the place of a genitive relative clause (cf. 22ac below). Of course,
these paratactic constructions are not exclusively applied to evade genitive
relative clauses, but subordinate clauses in general, including relative
clauses, particularly of the nonrestrictive type.
(22) a. parataxis:
And the <gap cause=anonymization desc=last or full
name> used a little further past the <gap
cause=anonymization desc=last or full name> and then
there was another one further up near Cranfield well I cant
remember the name of that. (EAN-H5H<Person:
PSEAN4><S: 479>)
instead of: ... there was another one further up near
Cranfield well [whosename I cant remember].
b. and-coordination:
... L. R., and her father was a farmer down at E. Hall, and he
had rather an unusual Christian name. (CNO FRED
Wes_017)
instead of: ... L. R., [whose father was a farmer down at E.
Hall], ....
62 Tanja Herrmann
c. left dislocation the grandfather:
Now the boatmen they used to erm<UNCLEAR> the name,
the family of a name of <gap cause=anonymization
desc=last or full name>, and the old man, the grandfather
his name was <gap cause=anonymization desc=last or full
name>and .... (EAN-H5H<S: 345>)
instead of: ... the grandfather [whose name was <gap
cause=anonymization desc=last or full name>] ....
8.5. Interaction of zero subject relative clauses with topicalization
hierarchies
Topicalization structures including clefts, pseudo-clefts, and all-pseudo-
clefts (explained below) are not counted as involving relative clauses but
form a distinct type of clause. In contradistinction to relative clauses,
topicalization structures do not revolve around the modification of an
antecedent, but the focussing of an antecedent (cf. Lehmann 1984: 363; cf.
also Quirk et al. 1985: 13861387; Huddleston and Pullum 2002: 1035;
Keenan 1985: 170; Stockwell, Schachter, and Partee 1973: 421422). The
subordinate clauses of these topicalization structures are related to relative
clauses and can be seen as assuming positions along a continuum of
relative clauses, with topicalization structures and relative clauses proper
at opposite ends (see Figure 4).
adnominal relative clauses
lexically empty antecedent
cleft all-pseudo existential relative clause
topicalizationmorelative
structures clause
proper
pseudo-cleft
nominal relative clauses
Figure 4. Continuum of relative clauses: from topicalization structures to
relative clause proper
The closer the types are located toward the right pole, the more similar to
relative clauses proper they are. In Figure 4, all clause types resembling
adnominal relative clauses are arranged above the line, while those
resembling nominal relative clauses are below the line. In contrast to clefts,
pseudo-clefts, and all-pseudo-clefts, all of which are placed at or toward the
topicalization end of the continuum, those subtypes in bold type, i.e. the so-
called lexically empty
14
antecedent relative clause and the existential, are
included in the count of all relative clauses in the present analysis, as they
sufficiently meet the defining criteria for relative clauses put forward in
section 1.
Although resembling an adnominal relative clause, the cleft sentence is
a topicalization structure that splits one proposition into two separate
clauses for the sake of giving focus to an element (for the following
discussion cf. Quirk et al. 1985: 13831387). The it-cleft follows the
pattern: it +be +focussed NP +subordinate clause, which resembles a
restrictive relative clause. However, there are various syntactic differences
between an annex clause (Quirk et al. 1985: 1387) of a cleft and a
relative clause, in addition to the prosodic difference arising from its
focussing nature (cf. also Lehmann 1984: 362).
15
One of the most striking
differences between relative clauses and clefts is that the co-occurrence
restriction between proper names (or other nouns or pronouns with specific
or unique reference) and restrictive (!) zero or that clauses is lifted in cleft
sentences, both in dialectal speech and in Standard English (cf. Quirk et al.
1985: 1241 and 1387; cf. also Huddleston and Pullum 2002: 10561057).
Consider the examples in (23):
(23) a. Isnt it High House we re talking about? (CNO FRED
Wes_011)
b. [...] I think it was Bristol Engineering that they did a lot of
work for themduring the war. (CNO FRED Wes_009)
c. [...] its you its up to. [...] (CSW-SRLM 123<T 1100>)
As shown in (24), clefts can also be introduced by the demonstrative
pronounthat (or this):
(24) a. Thats the boss shes speaking about. (SCO-G62<Person:
PSSCO13><S: 1070>)
b. No me elder brother was, not me second brother, that was the
Second World War he was in. (CMI-FXW<S: 037>)
64 Tanja Herrmann
c. And that was J . H. that they had from E. (CNO FRED
Wes_015)
Like dialectal relative clauses, dialectal clefts also lend themselves to
having a gap (zero marker) in subject position in the subordinate clause:
(25) a. It was J . H. lived in there. Her father raised in that house
next, it was A. H. lived there. (CSW FRED Som_012)
b. [...] It was my grandmother owned this bit of land [...]. (NIR
NITCS A13.3)
c. [...]t werent everyone had a binder. (CSW-SRLM 108<T
1440>)
The subordinate clause of the all-pseudo-cleft sentence looks like a
restrictive relative clause. The all-pseudo-cleft, as illustrated in (26),
parallels the pseudo-cleft in structure, but differs from it by having the
overt preceding noun phrase all in the matrix clause, followed by a
subordinate clause.
16
In the examples in (26) all and the subordinate clause
are underlined:
(26) a. [...] All as we could get for this milk was four pence a gallon,
[...]. (CSW-SRLM 105<T 1160>)
b. [...] Practically {PAUSE}, you practically kept the house on it,
you know {I know}[insertion in curly brackets, T.H.], all
was bought was bread. [...] (NIR NITCS A61.3)
c. all-pseudo-cleft focussing a verb:
[...] Id go the rounds with himand all I used to do was to er
take the peoples [sic] things that theyd bought up the entry
you see because they were all entries then. (CMI-FY5<S:
415>)
d. all-pseudo-cleft focussing a verb:
[...] so all as he had to do were go round in a circle all the
time, [...]. (CSW FRED Som_001)
While the all-pseudo-cleft in example (26c) also occurs in Standard
English, examples (26a,b and d) involve the dialectal particle as and the
zero marker in subject position, which are typical dialect features.
Topicalization structures incorporate particles (e.g. as) and/or maintain
usages (e.g. the zero marker in subject position) which have dropped out of
the relative marker system (in an area). In the Central Southwest, for
example, the relative particle as has retracted from the relative marker
system to semantically and syntactically related niches and exits the dialect
via topicalization structures, such as it-clefts and all-pseudo-clefts.
17
Historically, topicalization structures like all-pseudo-clefts served as as a
stepping-stone from the comparative construction (such as) into the relative
marker system in (Early) Middle English (cf. Smith 1984: 99100 and
134). As seemed to have developed a stronghold in topicalization structures
until the rising wh-pronouns also started infiltrating topicalization
structures like it-clefts in Late Middle English (cf. Ball 1994: 183). In other
words, these particles appear to enter a linguistic system via such
topicalization structures and also appear to leave it through them.
Mustanoja (1960: 191) reports all what as occurring in Old English, while
in Middle English, what is said to mostly relativize so-called antecedents
of less definite character, like all andnothing (Mustanoja 1960: 194). If a
dialect (or an idiolect) shows a specific particle in a relative clause, such as
what in the Central Southwest, it also shows this particle in topicalization
structures like clefts and all-pseudo-clefts. For instance, in the Central
Southwest, what as a relative marker occurs in each of the investigated
texts. So it also occurs in all-pseudo-clefts, as in example (27):
(27) [...] so all what we had was oil lamp, oil lamp and a telephone up
there for the police to keep ringing us up to see if we were all right.
(CSW FRED Som_012)
Hence, the following implicational tendency emerges, which holds true for
the vernacular particles what and as(in the Central Southwest) and the zero
marker in subject function overall (cf. existentials and lexically empty
antecedent relative clauses):
particle in topicalization structures > particle in relative clauses proper
18
A variety of syntactic structures that introduce new information or
participants into the discourse, such as existential sentences and lexically
empty antecedent relative clauses (i.e. relative clauses that modify
semantically bleached antecedents like the (only, best, first, last, next)
thing, the (only, best, first, last, next) one, one thing, something, anybody,
etc.), possess a focalizing aspect in addition to a weightier modifying
function. Consequently, have- and be-existentials of the types subject +
66 Tanja Herrmann
have +NP-[RC] and there +be +NP-[RC], respectively, are regarded as
involving relative clauses. Being a topicalization construction to some
extent as well, however, existentials evince zero relative clauses after
specific nouns and proper names (compare Lumsden 1988: 199201).
Hence, these clauses are nonrestrictive zero relative clauses syntactically,
which, of course, are ungrammatical in Standard English (cf. Quirk et al.
1985: 1258; cf. also Huddleston and Pullum 2002: 1056 and 1059). For
example:
(28) a. there-existential:
[...] there was Mr M. and B. W. from K. [ came round
buying horses]. [...]. (CNO FRED Wes_015)
b. there-existential:
(UNCLEAR) of course you there would just been my father
and mother [d be speaking Gaelic all the time in the house]
you see. (SCO-K6N<Person: PSSCO28><S: 0473>)
c. there-existential:
[...] there was only him {you know} [insertion in curly
brackets, T.H.] [ used to preach], [...]. (CMI-FYH<S: 631>)
d. have-existential:
Yes she had her aunt [ was a widow there at the time [when
she came to <gap cause=anonymization desc=address>]] and
she just lived about a year. (SCO-K6N<Person: PSSCO28><S:
0114>)
In addition, the zero relative marker is ungrammatical in subject position in
(written) Standard English (cf. Quirk et al. 1985: 1250; cf. also Biber et. al.
1999: 619). Supposedly, this restriction on the use of the zero marker in
subject function is due to the difficult processing of constructions in which
the zero relative clause modifies an initial noun phrase (subject of the
matrix clause) (cf. Bever and Langendoen 1971: 444). The following
example is taken from Bever and Langendoen (1972: 91):
(29) a. *Anyone owns a fleet of six cars deserves to be taxed at the
highest rate.
b. sentence is bracketed as if there were two coordinate clauses at
first blush:
*[Anyone owns a fleet of six cars] deserves to be taxed at the
highest rate.
c. sentence is rebracketed as a matrix clause and a relative clause
modifying the matrix subject, after coming across the second
finite verb (deserves), which needs a subject:
*[Anyone [ owns a fleet of six cars]
RC
deserves to be taxed at
the highest rate]
matrix
At first sight, the finite verb occurring first is, according to their analysis,
bracketed together with the preceding noun phrase, in the treacherous
assumption that the first clause is a main clause. Only after coming upon
the second finite verb, the former misbracketing [NP-V-O]
main clause
-V-O is
corrected to [NP-[NP
-V-O]
RC
-V-O]
matrix clause
(cf. Bever and Langendoen
1972: 66). If an overt subordinator (relative marker) indicates the beginning
of a subordinate clause, such erroneous analyses are ruled out from the start
(cf. Huddleston and Pullum 2002: 1055). Bever and Langendoen (1972: 78
and 91) claim that this perceptual constraint was grammaticalized, i.e.
extended to situations where perceptual problems would not arise (cf. also
Huddleston and Pullum 2002: 1055).
Nevertheless, descriptive syntax makes some concessions to the
appearance of zero in subject function: Zero subject relative clauses are
either clearly non-standard, between very informal and non-standard
(Huddleston and Pullum 2002: 1055), or a marginally non-standard usage
that may occur in some conversational varieties (Biber et. al. 1999: 619)
or in very informal speech (Quirk et al. 1985: 1250) under certain
conditions (Huddleston and Pullum 2002: 1055).
In dialectal speech, the constraint on zero subject relative clauses is
overridden the more a clause type moves away from the modification of an
antecedental noun phrase (relative clause proper) toward the
topicalization of a noun phrase (topicalization structures). In other words,
zero subject relative clauses pattern along the continuum of relative clauses
in Figure 4 above. They occur with increasing ease (in a given dialect or
idiolect), as one goes from existential and related presentational clauses like
lexically empty antecedent relative clauses to clear topicalization clauses
like clefts and all-pseudo-clefts, while they are very scarce in pure
modification structures (relative clauses proper).
Looking at previous dialectal investigations, the overwhelming majority
of zero relative clauses in subject function is reported to occur in the
following syntactic environments (or those syntactic environments can be
deduced from cited examples), irrespective of the region investigated:
There- and have-existentials, existential-like constructions like copular be
68 Tanja Herrmann
sentences or equational sentences, lexically empty antecedent relative
clauses, and clefts (e.g. cf. Elworthy 1877: 41 and 58; cf. also Elworthy
[1886] 1965: xxxiv, 660, and 688; Grant and Dixon 1921: 102; Harris
1993: 149150; Ihalainen 1980: 189; Ihalainen 1987: 84 and 9091;
Keklinen 1985: 355; Miller and Weinert 1998: 107, 128, 301, and 347
348; Ojanen 1982: 7578; Peitsara 2002: 171172 and 174; Poussa 1999:
96; Tagliamonte 2002: 157158 and 160161; van den Eynden 1993: 170,
175, and 178187; van den Eynden Morpeth 2002: 186188). Moreover, a
handful of examples occur in have-existential-like constructions involving
the verb to know or in other constructions which are not (or not easily)
subsumed under topicalization structures (e.g. cf. Corrigan 1997: 335; cf.
also Ojanen 1982: 77; Wilson 1915: 91). Shnukal (1981), who encountered
all these various types of zero subject relative clauses in topicalization
structures in a variety of Australian English, sketches the following
implicational scale, which parallels the frequencies of occurrence of the
nonstandard variant as well as the history of obligatory
S[ubject]R[elative]P[ronoun] presence in written Standard English
(Shnukal 1981: 324):
19
Type IV Type III Type II Type I
According to Shnukal, speakers who have zero subject relative clauses in
Type IV sentences may also show zero subject relative clauses in the other
three syntactic environments but not vice versa. Type IV consists of a
relative clause modifying anybody (here, lexically empty antecedent
relative clause type), Type III consists of the present have-existential-like
construction involving the verb to know, Type II comprises there- and
have-existentials, and Type I clefts and equational sentences. Except that
here, Shnukals Type III is grouped with (have-)existentials and that
equational sentences are subsumed either under lexically empty
antecedent relative clauses or under existentials (depending on the type of
antecedent), her implicational scale mirrors the continuum of relative
clauses sketched above. All types of zero subject relative clauses are
illustrated below:
(30) a. there-existential:
[...] theres nobody [ does that now]. (NIR NITCS A44.3)
b. multiple relative clause: first relative clause: have-existential-
like construction involving the verb to know; second relative
clause: have-existential; third and forth relative clause: there-
existentials:
No, no, no, no, eh, they went over, they would have been... I
know two, or three [ went over], but they went till the
building. They had, they had friends [ was in the building
trade in Scotland], they went over to, there, to... But there was
more [ went till the States], [ went till America], fromthis
country {ahah}, in the young days. They had ones with uncles
and aunts in America, and brought themout when they were
young. (NIR NITCS A31.3)
c. existential-like construction involving the semantically weak
verb to see:
[...] I seen a chap at Broughton Moor, [ got his leg took off].
[...] (CNO FRED Wes_014)
(31) a. lexically empty antecedent relative clause:
[...] Yet, at themtimes, anybody [ wanted to learn] could
learn, and I see anybody [ doesnt want to learn yet],
doesnt learn, so I dont know. [...]. (NIR NITCS A39.3)
b. lexically empty antecedent relative clause:
[...] The ones [ was here] was all going back to France.
(NIR NITCS A49.3)
(32) a. relative clause proper:
Aye, cogs, not cut up, a cog cut up, but cut up and sort of work
themin together you know, they could make a cog wheel [
would last quite a long time], of wood, you wouldnt think that
possible. [...] (CNO FRED Wes_004)
b. relative clause proper:
[...] and he put in the paper about these sheep [ would live
under t snow], [...]! [...] (CNO FRED Wes_015)
The total of 298 zero relative clauses in the present dialectal data
corresponds to Shnukals findings: There are 228 there-existentials, 24
have-existentials, 9 equational sentences, 9 existential-like constructions
involving the verbs to see, to hear, or to know, 18 lexically empty
antecedent relative clauses, and 10 relative clauses proper, which do not
fit into any particular category. It can be observed that not only does the
overall number of zero relative clauses correlate with the broadness of the
70 Tanja Herrmann
individual corpora (e.g. of the overall 228 there-existentials, 89 and 54
instances, respectively, originate from the broad NIR and CNO corpora,
respectively) but broadness of speech also determines the occurrence or the
frequency of zero relative clauses in particular syntactic environments:
Broader corpora have (more) instances of zero in relative clauses proper
and lexically empty antecedent relative clauses than less broad corpora,
while existentials and existential-like constructions are open to all corpora.
Thus, the implicational scale (supporting the continuum of relative clauses
in Figure 4 above) looks as follows:
clefts > existentials > lexically empty antecedent relative clause > relative
clause proper
Total: 270 18 10
While resembling a nominal relative clause, the topicalization structure
pseudo-cleft is a S-V-SCOMP sentence in which a subordinate clause,
usually introduced by what, functions as subject (basic pseudo-cleft). In
reversed pseudo-clefts, the matrix clause, normally starting with
demonstrative that (or this), precedes the subordinate clause, which
functions as subject complement in the sentence (cf. Quirk et al. 1985:
13871388; cf. also Huddleston 1984: 462). Since pseudo-clefts give focus
to the referent of the subordinate clause (cf. Huddleston 1984: 466) they are
distinct from nominal relative clauses (cf. Lehmann 1984: 361). For
example:
(33) a. basic pseudo-cleft: Verb is highlighted:
[...] what we used to do were, book a drop less milk and make
a little tiny one, [...]. (CSW-SRLM 122<T 1320>)
b. reversed pseudo-cleft:
[...] No doubt that was what did it. [...] (CNO FRED Wes_015)
9. Resumptive pronouns
Resumptive pronouns represent the antecedent within the relative clause
on top of a representation of the antecedent by a relative marker at the
beginning of the relative clause. Thus, resumptive pronouns have to be
distinguished from so-called copies, which represent the antecedent in the
continuation of the matrix clause. Resumptive pronouns surface as
pronouns (personal, possessive, demonstrative; existential there) or adverbs
(e.g. demonstrative there), usually in the position they would have in a
corresponding declarative clause. They point out or reinforce the
grammatical function of the relativized noun phrase in the relative clause
by case-marking and position, and they strengthen the coreference relation
between the relativized noun phrase and the antecedent by agreement in
gender and number.
9.1. Interpreting the pronoun retention strategy in terms of the
Accessibility Hierarchy
The examples below illustrate resumptive pronouns in the syntactic
positions subject, direct object, prepositional complement, genitive, and
(nonprepositional) adverbial, respectively:
(34) a. resumptive personal pronoun it in subject position:
Well, it s what they fed, you used to put it [i.e. treacle, T.H.]
on hay [that it was mouldy], you know, bad hay, and just
sprinkle it on to give a better taste for t cow to eat, you see.
(CNO FRED Wes_008)
b. resumptive personal pronoun it in direct object position;
antecedent and resumptive pronoun are not strictly
coreferential as number-markers do not match:
And I have lots of letters which I discovered about five years
ago, which I thought were probably letters [that my father had
written it], Adam<gap cause=anonymization desc=last or
full name>, because his name and my grandfathers name
were exactly the same. (SCO-GYW<S: 101>)
c. resumptive personal pronoun themin prepositional complement
position:
... it was, you know, looked upon then you were, were public
transport and the public team [that you belonged to them].
(EAN-HDL<S: 101>)
d. resumptive possessive pronoun its in genitive position:
But (trunc) y youd got to watch, there again, that er you
didnt exceed the width of er of your waggon, [which its
maximumlimit was er would be er eight foot three, or er
eleven foot six, high]. (CMI-FY2<S: 026>)
72 Tanja Herrmann
e. resumptive demonstrative adverb there in (nonprepositional)
adverbial position; preposition elision of at (cf. 10.1. below):
they had the stallions down there then you see, they had
stallions at the stud. (EAN-HYC<Person: PSEAN17><S:
0207>)
[[at] Which we were there often]. (<S: 0208>)
Based on Keenan and Comries findings, two predictions are made for the
present data:
1. The distribution of resumptive pronouns in simple relative clauses
mirrors the accessibility of syntactic positions (Accessibility Hierarchy)
such that resumptive pronouns will be more likely to occur the lower a
position ranks on the Accessibility Hierarchy (GEN >PCOMP >DO >
SUBJ ) (cf. Keenan and Comrie 1977: 92).
2. Resumptive pronouns combine more often with invariant relative
particles than with relative pronouns because particles are less explicit in
terms of structural markedness. In other words, in contrast to relative
pronouns, relative particles lack case-marking, agreement in gender or
animacy (i.e. with regard to the personality/nonpersonality opposition), and
agreement in number with the antecedent (to some extent, relative pronouns
show agreement in number with collective nouns). High structural
markedness eases the retrievability of the relativized noun phrase (and thus
the identification of the antecedent) and helps to identify the syntactic
function of the relative marker in the relative clause.
Table 13. Resumptive pronouns in simple relative clauses across regions
SUBJ DO PCOMP GEN (Other:A) total
zero - 1 - - - 1
that 1 2 1 - - 4
what 1 1 - - - 2
as - - - - - -
who 1 - - - - 1
which 5 1 3 1 (3) 13
whom - - - - - -
whose - - - - - -
total 8 5 4 1 (3) 21
Neither prediction is borne out by the data. In general, resumptive pronouns
are extremely rare, as shown in Table 13. Indeed, the data seem to suggest
the opposite of what was predicted. First, instead of a steadily increasing
number of resumptive pronouns from subject to genitive position, the
number of resumptive pronouns decreases with every position. Second,
when all invariant relative particles and both wh-pronouns are taken
together, twice as many relative pronouns co-occur with resumptive
pronouns as relative particles do. In addition, the least explicit relative
particle, i.e. the zero marker, combines only once with a resumptive
pronoun.
How is this to be explained, given that the Accessibility Hierarchy
should also apply to the pronoun retention strategy, which it clearly does in
Keenan and Comries (1977: 9293) cross-linguistic investigation?
Admittedly, Keenan and Comrie were primarily concerned with the
occurrence/nonoccurrence of a resumptive pronoun in a position within a
language, not with the number of occurrences in each position. Statements
on frequencies of resumptive pronouns in individual positions they made
only in passing (cf. Keenan 1985: 146148; cf. also Comrie and Keenan
1979: 663). Second language acquisition, however, seems to provide
evidence for an inverse relation between resumptive pronouns and the
Accessibility Hierarchy (except for the genitive position) (cf. Hyltenstam
1984: 4751; cf. also Gass and Selinker 1994: 113114).
For one thing, the overall number of 21 resumptive pronouns is not
particularly impressive. In fact, the frequency of resumptive pronouns in
the investigated 480,000 words corpus is as low as 4.4 occurrences per
100,000 words, so the scarcity of the phenomenon might be responsible for
a skewed distribution.
Secondly, new starts in favor of a paratactic construction might be
mistaken for applications of a resumptive pronoun. The two phenomena are
difficult to distinguish, especially in the face of missing information on
prosody and speakers intentions. This is particularly true for resumptives
in subject position:
(35) a. new start after er, starting with they, which introduces a new
parataxis:
That was our marking, all our ships used to have the blue and
er I think blue <UNCLEAR> and yellow in the square, cos
they hired these <UNCLEAR> the people who do er you know
suppose hire themoff now would be the ermthe Dutch people
74 Tanja Herrmann
cos theyre the people [what] er, they deal in all that type of
thing, big dredging, thats how Rotterdamwas built <vocal
desc=clears throat> (EAN-H5G<S: 846>)
b. new start after er, starting with it, which introduces a new
parataxis:
We then got the, the agency for Morris, which was a very
popular car and ermwe also got the agency for Austins,
[which] er, it came about with father being interested in a
hiring car, the Austin Twenty, he thought it was, it was the
best that could be got. (SCO-GYW<S: 138>)
c. new start after truncated belong, starting with that, which
introduces a new parataxis:
And we had ermwe, there was a few changes during that time,
we used to get troops coming in and occupying the classroom,
and we (UNCLEAR) maybe, at the very early days I can
remember having to go to the <gap cause=anonymization
desc=address> Street School and also the <gap
cause=anonymization desc=address> Street Hall [which
(trunc) belong] that was a church hall. (SCO-GYW<S: 169>)
All examples lacking an indication of a new start, such as punctuation,
pause fillers, or self-correction, were counted as instantiating resumptive
pronouns. This policy might be (partially) responsible for an inflated
number of resumptives in subject position. Below, four instances of a
resumptive pronoun in subject position will be cited and two instances of a
resumptive pronoun in PCOMP position:
(36) a. resumptive personal pronoun they in subject position;
antecedent and resumptive pronoun are not strictly
coreferential: Semantics and number concord do not match;
resumptive they refers to the M. clan, living at the M.
place, which is represented by Mrs M.:
Yeah. I was getting ready to go to school one morning, he
said, Where you going today? I said, Going to school,
dad. No, he said, you gotta come with me. And we
walked fromBarton St David down, round, well, ??? down
round that corner and well on down to Catsham. Up through
Southwood, up to Lottisham. And Mrs M., [what they live up
here to, where the, (CSW-SRLM 132<T 1140>)
Q: MUSEUM IS....
museumis]. Er, to her father that did, had the farmsale. And,
er, course we went up there and stayed there till the sale was
finished. Well then father took on some cattle then, to deliver,
and we finished up out to Mr A.s farmsale. We took some
cattle and we finished up out to North Wootton, that night.
b. existential there is the resumptive pronoun in subject position
(preferred analysis); alternatively, this might exemplify
preposition elision of stranded partitive GEN preposition of;
compare with example (36f) below:
I can remember quite vividly the old tramcars running there er
day and night, with the last service leaving the outskirts of
Edinburgh around about er twelve er eleven thirty and you
<trunc> g have about ten minutes or so to reach the depots
[which there were many and varied <UNCLEAR> at this
particular time]. (SCO-GYU<S: 224>)
c. resumptive personal pronoun it in subject position, pointing
out singular number of relativized noun phrase in relative
clause, in combination with relative particle that:
Well, it s what they fed, you used to put it [i.e. treacle, T.H.]
on hay [that it was mouldy], you know, bad hay, and just
sprinkle it on to give a better taste for t cow to eat, you see.
(CNO FRED Wes_008)
d. resumptive demonstrative pronoun that in subject position
(preferred analysis); alternatively, that could also be analyzed
as a vacuous conjunction in the Middle English relative
marker relic which that:
So I went and left I got a job at <gap cause=anonymization
desc=last or full name> [which that was one of <gap
cause=anonymization desc=last or full name>s]. (CMI-
FYE<S: 228>)
e. resumptive personal pronoun them after stranded partitive
genitive preposition of (preferred analysis); alternative
analysis: change to parataxis in the middle of the clause, i.e. to
consider we had as a clause element which belongs to two
adjacent clauses. It is the subject and verb in the preceding
relative clause (which we had) and simultaneously the subject
and verb of an ensuing main clause (we had quite a number of
76 Tanja Herrmann
them). Bracketing would look as follows: ... the tools, [which
[we had] quite a number of them]:
Now the first job he did was to get you well acquainted with
the tools, [which we had quite a number of them]. (CMI-
FY2<S: 003>)
f. resumptive personal pronoun us after stranded partitive
genitive preposition of (preferred analysis); alternative
analysis: Existential there is resumptive pronoun in subject
position; compare to example (36b) above:
So after a fortnight, thereabouts, the union had a meeting and
decided that everybody in the industry like, [which of course
there was only about a thousand of us any way], everybody
would (trunc) s would stop until this man could start at work.
(CMI-FYH<S: 556>)
The preponderance of resumptive pronouns involving which, in general,
and in subject position, in particular, is conspicuous (compare Peitsaras
[2002: 175] examples in her Suffolk corpus and Ihalainens [1980: 190
191] examples in his Somerset corpus).
However, (36d) and another example involving which that in subject
position are not excluded from an analysis as resumptive pronouns,
although they could in principle be analyzed as instances of the relative
marker relic which that found in Middle English (where that functioned as
a vacuous conjunction). The occurrence of analogous combinations of a
relative marker and a resumptive pronoun in subject position (like which it,
who they, what they, or that it in the examples above) makes a special
interpretation for which that unwarranted. In such cases an analysis of the
second element as a conjunction is not possible, since the resumptive
pronoun takes the shape of a personal pronoun. Nor is the occurrence of a
demonstrative pronoun like that figuring as a resumptive pronoun, as in
which that, unusual. Resumptive pronouns can, after all, take on the shape
of personal, possessive, or demonstrative pronouns.
In contradistinction to what J espersen (1927: 188) says about Standard
English, namely that pied-piping is mandatory in partitive genitives (cf.
also Huddleston and Pullum: 1041), examples (36e) and (36f) seem to
document that stranding the preposition of in partitive genitives is permitted
in dialect.
9.2. (Non)Restrictiveness as a factor
Checking the restrictiveness/nonrestrictiveness of the which examples in
subject position, and then of the which examples in all other positions, it
turns out that, without a single exception, they are all nonrestrictive. Thus,
the preponderance of which with a resumptive pronoun (contradicting
Prediction 2) appears to correlate with the nonrestrictiveness of the relative
clauses. In dialect, the relative particle what, which combines with personal
and nonpersonal antecedents, is also available for syntactically
nonrestrictive environments (compare [36a] above; the other what example
in direct object position is also nonrestrictive). Which, however, is the
dominant relative marker in nonpersonal nonrestrictives and a viable option
in personal nonrestrictives (compare [36f] above), whereas the relative
markers who, that, and zero are only minor options in nonrestrictive
relative clauses (cf. Table 2 above). By contrast, the restrictive subject
relative clause in example (36c) selects the relative particle that. The three
restrictive examples in DO and the one in PCOMP equally opt for relative
particles (three instances of that; one instance of zero).
Recalling that Keenan and Comrie set up the Accessibility Hierarchy on
the basis of restrictive relative clauses with definite antecedents (cf. Keenan
and Comrie 1977: 64), the prediction about the distribution of the pronoun
retention strategy does not hold if, as in the present corpus, the majority of
relative clauses involving the pronoun retention strategy are nonrestrictive.
If all nonrestrictive examples of resumptives were discarded, we would be
left with a mere handful of restrictive examples (one zero relative clause;
four that relative clauses), which would neither save nor thwart Prediction
1. The question arises why resumptive pronouns in the present data mainly
appear in nonrestrictive relative clauses. It seems that resumptives have a
predilection for appearing in difficult and complex syntactic environments
(cf. Keenan and Comrie 1977: 92). Since nonrestrictives are atypical of
spoken language and uncommon or even alien to traditional dialect, their
application demands the help of a resumptive pronoun, more than
restrictives do. In sum, dialect speakers resort to resumptive pronouns in
constructing nonrestrictive relative clauses in order to cope with a linguistic
complexity with which they are rather unfamiliar.
With respect to Prediction 1, that the distribution of resumptive
pronouns mirrors the accessibility of syntactic positions, the question
remains why resumptives in nonrestrictive relative clauses, particularly
those involving which, do not follow the hierarchy for pronoun retention in
78 Tanja Herrmann
restrictive relative clauses (GEN >PCOMP >DO >SUBJ ). Or, to be more
precise, what causes this abundance of nonrestrictive which resumptives in
the subject position?
Dialect speakers who have less practice in constructing a nonrestrictive
relative clause, since nonrestrictive relatives are rather a feature of written
Standard English, fall back on the most basic type of (nonrestrictive)
relative clause: the one which relativizes the subject position in Keenan
and Comries (1977: 95) words, subject relatives are psychologically
simpler than nonsubject relatives. That is to say, speakers start a
nonrestrictive relative clause by employing the standard relative marker
which in subject position and additionally reinforce the syntactic position
of the relativized noun phrase in the relative clause via a resumptive
pronoun. Hence, the predominance of the subject position within the
pronoun retention strategy or even a reverse order of syntactic positions
from what was expected (compare Table 14 below) arises from the
dominance of the subject function in the general Accessibility Hierarchy,
which also applies to the pronoun retention strategy. In this sense, the
general Accessibility Hierarchy (SUBJ >DO >PCOMP >GEN) overrides
the pronoun retention Accessibility Hierarchy (GEN >PCOMP >DO >
SUBJ ) in nonrestrictive environments.
Table 14. Resumptive pronouns in nonrestrictive relative clauses, with special
emphasis on which
SUBJ DO PCOMP GEN (Other:A) total
what 1 1 - - - 2
which 5 1 3 1 (3) 13
total 6 2 3 1 (3) 15
In sum, Keenan and Comries Accessibility Hierarchy, which was set up
for restrictive relative clauses with definite antecedents, is not invalidated
by these unexpected result with regard to the pronoun retention strategy. It
is the inclusion of nonrestrictive relative clauses which is responsible for
Predictions 1 and 2 not being borne out by the present data. The pronoun
retention strategy reflects the Accessibility Hierarchy in reverse order in
restrictive contexts, yet it is not or not necessarily transferable to
nonrestrictive relative clauses.
9.3. Further embedded relative clauses via resumptive pronouns
According to Keenan and Comrie (1977: 92), pronoun retention enables or
facilitates relativization in difficult environments such as coordinate
NPs, other relative clauses, indirect questions, and even sentence
complements of NPs (Keenan and Comrie 1977: 92; cf. also Comrie 1993:
140141 and 147; Comrie 1999: 89; Keenan 1985: 155156; Hawkins
1999: 265). In spelling out the relativized noun phrase in its normal clause
position, resumptive pronouns maintain the coreference relation between
the antecedent and the relative marker after intervening clauses or in
otherwise deviant coordinate relative clauses.
(37a) and (37b) exemplify multiple relative clauses in which the relative
marker serves different grammatical functions within the various
coordinated relative clauses. Without a resumptive pronoun this would be
prohibited by a coordinate construction constraint. With the aid of a
resumptive pronoun, which explicates the changed grammatical function of
the relative marker, the coordinate construction constraint is overcome so
that these disparate juncts can be linked via and-coordination:
(37) a. The zero relative marker serves direct object function in the
first coordinated relative clause, whereas it serves
prepositional complement function in the second coordinated
relative clause; the resumptive pronoun it indicates the change
of grammatical function of the zero marker from the first to
the second relative clause when it surfaces as prepositional
complement after the stranded preposition with:
Er things like er <pause> crowbars and bull croppers and er
rescue ropes and lines and things of that kind, which are very
very simple, stuff [ you would buy in a hardware shop and
probably be able to manage with it]. (EAN-K69<Person:
PSEAN13><S: 065>)
b. Which serves prepositional complement function in the first
two relative clauses (stranded preposition about is omitted in
the second relative clause); which changes to direct object
function in the third relative clause, where the resumptive
pronounit explicates the changed grammatical function of the
relative marker:
... you know the decontamination side, the clearing up, the
protecting of the environment from toxic chemicals er [which
80 Tanja Herrmann
weve all heard about in newspapers, and read reports and
seen it on television], .... (EAN-K69<Person: PSEAN13><S:
038>)
In (38a) the presence of a resumptive pronoun in prepositional complement
position keeps up the coreference relation between antecedent and relative
marker after an intervening clause parenthesis (Imnow retired). Example
(38b) exhibits a resumptive pronoun in a further embedded coordinate to-
infinitive clause.
(38) a. ... so that that was quite an interesting area [which, even now
[Imnow retired] I still have a little hand in that] .... (EAN-
K69<Person: PSEAN13><S: 031>)
b. resumptive pronoun them in further embedded and-
coordinated to-infinitive clauses indicates change of
grammatical function from prepositional complement
(PCOMP(OP)) to direct object; relative marker that functions
as subject in the relative clause, which is reinforced by
resumptive pronoun they after interrupting subordinate clauses
(after a false start in which relative marker that would function
as direct object: that theyd have done):
... But, eh... {PAUSE}therere quite a few old sayings, and
one thing and another, in the country, [that [if you had time
[to, to think about {mm}and study them]], theyd have, theyd
have done, gone by the board] {ahah}. ... (NIR NITCS A64.3)
In (38b), the resumptive pronoun themindicates the different grammatical
function of the relativized noun phrase in the second to-infinitive clause:
While in the first to-infinitive clause the relativized noun phrase serves
prepositional complement function, in the second to-infinitive clause it
serves direct object function (if you had time [to, to think about
RELATIVIZED
NP: PCOMP
{mm} and study them
RELATIVIZED NP: DIRECT OBJ ECT
]). After this
intervening complex conditional clause, an additional resumptive pronoun
they ensures the coreference relation between the antecedent and the distant
relativized noun phrase and it indicates the subject function of the
relativized noun phrase in the remainder of the relative clause (quite a few
old sayings, and one thing and another, in the country that [if you had
time [to, to think about {mm}and study them]], theyd have, theyd have
done, gone by the board).
Resumptive pronouns allow the construction of complex sentences by
relativizing into otherwise inaccessible syntactic positions involving
different levels of subordination. Thus, the presence of a mental prop in the
shape of a resumptive pronoun gives rise to a type of complex relative
clause in dialect which is not possible in Standard English (cf. Quirk et al.
1985: 1299; cf. also Huddleston and Pullum 2002: 1091).
(39a) is an instance of self-correction: from a further embedded relative
clause occurring in Standard English, in which the conjunction that is
obligatorily omitted because the relative marker that assumes subject
function in the relative clause (our own [that
SUBJ ECT
we know THAT are])
(cf. Quirk et al. 1985: 1298; cf. also Huddleston and Pullum 2002: 1083
1085; Comrie 1993: 162
20
; Comrie 1998: 65 and 80; van den Eynden 2002:
186 and 191), to a further embedded relative clause only occurring in
nonstandard speech, namely one which shows relativization into an indirect
question clause. The latter relative clause explicates the sense relation
between the superordinate clause (we know) and the subordinate clause
(howthey are bred) by the conjunction how and spells out the subject of
the deep-embedded clause via a resumptive pronoun (they). While there is
relativization across one clause boundary in (39a), there is relativization
across two clause-boundaries in (39b) first into an interrogative clause,
then into a conditional clause with the resumptive pronoun themserving as
prepositional complement:
(39) a. false start: Standard omission of conjunction that after we
know since relative marker that serves subject (passive)
function (subject in partially object environment [cf.
below]); however, this false start is abandoned for an
interrogative clause after we know; new start (self-correction):
starting with relative marker that; resumptive pronoun they is
subject (passive) in a further embedded interrogative clause:
Yes, aye, or breed off our own [that
SUBJ ECT
we know THAT
are] [that we know [how they are bred]]. (CNO FRED
Wes_002)
b. resumptive pronoun themin prepositional complement (object
of preposition) function in further (first interrogative clause,
second conditional clause) embedded conditional clause:
... When I went to school we had all sorts of games, [which the
children of now wouldnt know [what you were talking about
[if you talked about them]]]. .... (CNO FRED Wes_019)
82 Tanja Herrmann
The examples in (40) to (42) exemplify resumptive pronouns in so-called
subject in partially object environments (cf. Quirk [1957] 1968: 101).
That means, the relativized subject noun phrase in the deep-embedded
clause itself functions as the direct object of the superordinate clause and
surfaces as a relative marker at the beginning of this complex structure:
(40) a. relative marker that in subject (passive) function is resumed as
it (resumptive pronoun) after interjection you know; relative
marker that introduces a nonrestrictive relative clause:
... He used to tell me about his country [that [you know] it was
taken over by the Russians] and then itd be taken over by the
Germans and You know what I mean. (SCO-K6L<S: 064>)
b. further embedding of object clause after insertion you know of
superordinate clause; conjunction that is obligatorily omitted
due to subject function of relative marker that; instead, there is
the resumptive pronoun it in subject (passive) function:
... He used to tell me about his country [that [you know]
[THAT it was taken over by the Russians]] and then itd be
taken over by the Germans and You know what I mean. (SCO-
K6L<S: 064>)
c. relative marker that as direct object of you know; resumptive
pronoun it has subject (passive) function in further embedded
object clause:
... He used to tell me about his country [[that you know] [it
was taken over by the Russians]] and then itd be taken over by
the Germans and You know what I mean. (SCO-K6L<S: 064>)
Different ways of bracketing render different interpretations plausible: In
(40a) you know is analyzed as an interjection after which the relative clause
is continued; the relative marker that has subject function, as well as the
resumptive pronoun it. This analysis of resumptive pronoun placement
without further embedding would receive more support if there were some
indication, like commas or dashes, that the speaker had paused before and
after you know. In (40b) you know is taken to be an insertion in the relative
clause on which the following object clause is dependent. The relative
marker that and the resumptive pronoun it in the further embedded object
clause assume subject function. In (40c) the relative marker that is felt to be
the direct object of you know. The resumptive pronoun it in subject
function is further embedded in the object clause.
Syntax and/or semantics might lend support to the interpretations in
variant (40b) or (40c). The analysis of example (40) advocated here is
instantiated by variant (40b). In (41) below, an interpretation of the relative
marker functioning as object is ruled out, since the superordinate clause he
was sure requires a prepositional complement instead of a direct object:
(41) relative marker which has subject function; he was sure is
bracketed as an insertion; bracketing which as the direct object of
he was sure is prohibited because he was sure demands a
prepositional complement: he was sure of +PCOMP:
... And he come home one night, and he had two wee canaries,
[which [he was sure] [was hens]]. ... (NIR NITCS A45.3)
Irrespective of the issue whether the relative marker functions as the subject
of the relative clause or whether there is something object-like about it,
the redundant resumptive pronoun, which here is analyzed as an overt
subject in the deep-embedded object clause, makes these complex
sentences more explicit, as in:
(42) a. resumptive pronoun it as overt subject in further embedded
object clause:
And they had a what they called the (UNCLEAR) twenty four
girls dancing you know, and also they, they had the, a film in
the show [which probably I think [it was a Bing Crosby film
[that was on]]]. (SCO-GYW<S: 462>)
b. resumptive pronoun it as overt subject in further embedded
object clause:
I was the only one who stood. (SCO-GYW<S: 512>)
In that second ward. (SCO-GYW<S: 513>)
Which was a ward [which I would say [it was built about the
(trunc) ninet the eighteen seventies]]. (SCO-GYW<S: 514>)
In addition, nonstandard speech may disregard the prescriptive rule that the
conjunction that has to be omitted if the relative marker serves subject
function (cf. note 26), presumably to prevent that that be mistaken for the
subject of the further embedded clause (cf. Quirk et al. 1985: 1050; cf. also
Comrie 1999: 8485). This is illustrated in example (43), taken from Quirk
et al. 1985:
84 Tanja Herrmann
(43) a. They pointed out the damage [which they supposed [had been
done by last nights storm]]. (Quirk et al. 1985: 1050
b. *They pointed out the damage [which they supposed [that had
been done by last nights storm]]. (Quirk et al. 1985: 1050
In example (44) below, the relative clause modifying the first one is
constructed according to the rules of further embedded subject in partially
object environment relative clauses in Standard English. The second
relative clause modifying the one appears to ignore two prescriptive rules at
once: First, the relative clause seems to lack a subject, i.e. the zero relative
marker is ungrammatical in subject function (cf. Quirk et al. 1985: 1250);
second, the conjunction that is not omitted. While the occurrence of the
zero relative marker as subject in such further embedded relative clauses is
also noted for Standard English (cf. Quirk 1968: 102 for Standard British
English; cf. also Huddleston and Pullum 2002: 1085; Graf 1996: 5859 for
Standard American English), it is regarded as an undesirable weak point in
Standard English, which is concealed by a push-down [i.e. further
embedding] element (Schmied 1993: 87). The presence of the conjunction
that when the relative marker has subject function is ungrammatical in
Standard English, but recorded in dialectal speech (e.g. van den Eynden
1993: 162 for Dorset), which has preserved a historically older, less
restrictive way of using the conjunction that (cf. van den Eynden 1993:
224). Alternatively, if that were taken for a demonstrative pronoun in (44),
the situation of a missing subject in the relative clause would be redeemed
by the presence of a resumptive pronoun in subject position in the further
embedded object clause.
(44) first relative clause: Relative marker that is counted as subject in
subject in partially object environment; Herbie thought introduces
a dependent, further embedded object clause in which a
conjunction that is obligatorily omitted since the relative marker
that functions as subject; second relative clause: The zero relative
marker fills the subject position in subject in partially object
environment; when the relative marker has subject function, the
presence of the conjunction that in an object clause is
ungrammatical in Standard English, but not uncommon in dialect;
if that were taken for a demonstrative pronoun instead of a vacuous
conjunction, the relative clause would receive a subject in the shape
of a resumptive pronoun:
... So I dont know whether now I have two males instead of the, the
first one [that Herbie thought [was a male]], were not so sure, but
the one [ he thought [that wasnt a male]], is a male. ... (NIR
NITCS A45.3)
9.4. Nonreduction
A special variant of resumptive is the nonreduction of the relativized
noun phrase, that is, the antecedental noun phrase, or part of it, surfaces
within the relative clause. Obviously, a nonreduced noun phrase is even
more explicit than a resumptive pronoun or adverb:
(45) a. ... These are the points, [what we used to stand themon the
points], yes. ... (CSW FRED Som_009)
b. And then they, the one where J ohnny Weismuller was, it was a
lake in front of a great arena [which they did a lot of you
know, swimming about in the lake and the different
formations]. (SCO-GYW<S: 484>)
c. ... well then there used to be wire and used to have a big wheel
in top, [which you couldnt go over the top] and .... (EAN-
H5G<S: 883>)
d. multiple relative clause: further embedded that relative clause
consisting of two coordinated juncts linked via or, inside which
relative clause:
... cos you got a radio tower on the dock now [which every
ship [that come into port or leave the port] have to go through
the radio tower] and .... (EAN-H5H<Person: PSEAN4><S:
425>)
e. stranded (or even fronted: at which there was sufficient room)
prepositionat was extracted from superordinate relative clause
across one bracket into the further embedded to-infinitive
clause together with the nonreduced noun phrase:
And of course the, with the boxes being all round the side, the
dressing boxes, this was outmoded, we needed a building at the
side, [which there was sufficient roomat [to build something at
86 Tanja Herrmann
the side] [because we had a park at the side of it then]]. (SCO-
GYW<S: 552>)
Nonreduced noun phrases should be subject to the same principles as
resumptive pronouns. In other words, Prediction 1 and 2, set out for
resumptive pronouns in 9.1., should hold for them, too: First, the frequency
of occurrence of nonreduced noun phrases should reflect the Accessibility
Hierarchy in the reverse order (GEN >PCOMP >DO >SUBJ ). Second,
nonreduced noun phrases should co-occur more often with relative particles
than with relative pronouns. On the basis of only a handful of clear
examples of nonreduced noun phrases, no valid interpretation can be made
regarding these predictions. However, all examples relativize the
prepositional complement position a position located at the lower end of
the Accessibility Hierarchy which seems to lend support to Prediction 1.
The nonreduced noun phrases fill the gaps created by relativizing on
PCOMPs with stranded prepositions. Either the whole antecedental noun
phrase re-emerges as a nonreduction or just part of the antecedent re-
emerges: the head noun with its determiner in (45d), or the prepositional
attribute in (45c). Examples (45d) and (45e) are complex relative clauses
which involve another embedded relative clause, consisting of two
coordinated relative clauses in (45d) and a further embedded to-infinitive
clause, followed by a causal clause in (45e). Thus, the distance between the
relative marker and its governing (stranded) preposition is considerable,
created by an intervening medial relative clause in (45d) and by an
infinitival clause in (45e). The appearance of the redundant nonreduced
noun phrase makes the laborious recovering process of the relativized noun
phrase unnecessary and that of its grammatical function in the relative
clause in such environments. In (45e), the stranded preposition at was
extracted from its proper position at the end of the (superordinate) relative
clause and moved into the further embedded infinitival clause together with
its governed nonreduced noun phrase.
Four examples contain the relative pronoun which and one example the
relative particle what, thus contradicting Prediction 2 that resumptive
pronouns combine more often with invariant relative particles than with
relative pronouns. This can, however, be attributed to the nonrestrictiveness
of all examples. It is the cause of the application of the nonrestrictive
relative markers which and what, on the one hand, and of the recourse to a
resumptive in the form of a nonreduced noun phrase, on the other hand, as
could be seen in 9.2. above.
10. Whichas connector?
In his investigation of Somerset English, Ihalainen (1980: 190) argues that
which can function as a so-called connector not as a relative pronoun. In
this function it simply connects two statements in cases in which there is
no obvious antecedent for which (Ihalainen 1980: 190). For example:
(46) a. ... And, er, you had a great big chap up in between the hooves.
Which the cow did go crippled. ... (CSW-SRLM 107<T
3004>)
b. Then we moved to the <gap cause=anonymization
desc=address> and then I went to <gap cause=anonymization
desc=address>Road School, which I finished my time there,
fourteen. (CMI-H4B<S: 561>)
c. And of course the, with the boxes being all round the side, the
dressing boxes, this was outmoded, we needed a building at the
side, which there was sufficient roomto build something at the
side because we had a park at the side of it then. (SCO-
GYW<S: 552>)
d. And they had addresses and I ran around, I went to get a job at
the pit, which er I had no more sense so Imglad now that I
didnt get on. (CMI-FXU<S:159>)
e. ... the one down Taunton left his pottery to his son, which all he
wanted was drink. (CSW FRED Som_009)
In a similar vein, Miller and Weinert (1998: 110) remark in their study of
spoken Lothian Scottish English that in these cases [t]he function of which
is to signal a general link between the material that precedes it and the
material following. Instead of a relative pronoun, which is analyzed as a
general discourse linking element or complementizer (Miller and
Weinert 1998: 111). This analysis appears to receive support when the
subsequent clause shows no grammatical gap, i.e. when which does not
seem to have a grammatical function in the clause, as all syntactic positions
are already taken by other noun phrases (cf. Miller and Weinert 1998: 110
111). Miller and Weinert observe parallel historical developments of that
and which, in the course of which the former relative pronoun which has
caught up with that in having developed a second function as general
conjunction.
88 Tanja Herrmann
Nevertheless, analyzing which as a general conjunction is unnecessary
as well as unwarranted. The element of (post)modification is still strongly
felt in these cases even if the relative marker cannot readily be assigned a
grammatical function in the clause at first sight. As will be shown below, in
all alleged connector instances which introduces either a nonrestrictive
adnominal relative clause or a sentential relative clause, which is
nonrestrictive by nature. The relative pronoun which is the nonrestrictive
relative marker par excellence and the only available option in sentential
relative clauses. Besides, all instances of alleged connector which involve
nonpersonal antecedents, except for one instance in text CSW- FRED
Som_009: Example (46e) above contains a personal antecedent his son. In
section 6 it was demonstrated that which does not exclusively modify
personal antecedents, although this is its highly preferred environment. In
the Central Southwest and in the idiolect of one CSW speaker in particular,
which also combines with personal antecedents. These occurrences of
personal which are clearly instances of adnominal relative clauses:
(47) a. ... And then there was C., [which caught his hand in the
machinery up here] and he had his hand off, being severed (at
the) wrist. ... (CSW FRED Som_009)
Accordingly, which does not seem to have parted with its properties as a
relative pronoun but still seems to operate in its favorite syntactic
environment, i.e. in nonrestrictive relative clauses modifying nonpersonal
antecedents. A close look at all putative examples of connector which
reveals that they can be analyzed more satisfactorily as instances of either
preposition elision, resumptives, or new starts.
10.1. Preposition elision
The following instances can be explained as cases of preposition elision:
(48) a. formerly example (46e): stranded preposition by elided
(preposition by could also have been elided in fronted
position: [from/by] Which the cow did go crippled); relative
pronoun which functions as adverbial (A) instead of
prepositional complement in adverbial function (PCOMP(A)):
... And, er, you had a great big chap up in between the
hooves. [Which the cow did go crippled [from/by]]. ... (CSW-
SRLM 107<T 3004>)
b. multiple relative clause: relative clause embedded in relative
clause; preposition with is elided; stranded preposition with
is of questionable acceptability in the position above, but after
the second coordinate noun phrase any haulage it would cause
an obstruction between the antecedent any haulage and the
following zero relative clause: Carriages ......... well, my father
had two horses [which he used to do the local emptying of
dustcarts and any haulage[with] [ there was to be done in
the village]]; (preposition with could also have been elided in
fronted position: [with] which he used to do the local emptying
of dustcarts and any haulage [ there was to be done in the
village]); problematic positioning of preposition might have
enforced preposition elision; relative pronoun which functions
as A instead of PCOMP(A):
Carriages ......... well, my father had two horses [which he used
to do the local emptying of dustcarts [with] and any haulage
[ there was to be done in the village]]. .... (CNO FRED
Wes_018)
c. preposition elision of stranded to in the face of a second
preposition (off) which precedes to; relative pronoun which
functions as indirect object (IO) instead of prepositional
complement functioning as object of preposition
(PCOMP(OP)):
So we were really answerable to the Ipswich Borough Council,
rather than to private enterprise [which some people really
wanted to sell us off [to]] as being a, you know, a weight round
their necks because if we didnt make a lot of money after the
war, ... (EAN-HDL<S: 277>)
Examples like (48ac) are the result of preposition elision. Which was
supposed to function as a prepositional complement (PCOMP) with a
stranded preposition. (Recall that preposition stranding is the rule in
dialectal speech). However, stranded prepositions, which are separated
from their governing relative markers, are often forgotten. Elision of
(stranded) prepositions is a recurrent dialectal feature (also in nominal
relative clauses; compare example (10a) in section 6) as well as a typical
90 Tanja Herrmann
feature of spontaneous spoken language. Preposition elision occurs with
various relative markers and also outside relative clauses, for instance in to-
infinitive clauses (cf. Miller and Weinert 1998: 106 and 108 for spoken
Lothian Scottish English; cf. also Miller 1993: 112). Preposition elision is
frequently found after massive and/or complex intervening material (as in
example [48b]) and after another preposition (as in example [48c]). If the
governing preposition is missing, which cannot function as a prepositional
complement (PCOMP), of course. Instead which assumes the functions
adverbial (A) or indirect object (IO), whereas with a preposition which
would function as PCOMP(A) or PCOMP(OP), respectively.
Elision of prepositions is not limited to relative clauses introduced by
which. It also occurs with other relative markers:
(49) elided stranded preposition with after another preceding
prepositionin:
And never in my wildest dreamdid we imagine that Labour would
get in with the resounding majority [that they did get in with].
(SCO-K6M<S: 010>)
10.2. Resumptives: Resumptive pronouns (or adverbs) and nonreduced
noun phrases
The following examples of connector which can be explained as instances
of resumptives:
(50) a. formerly example (46b): For claritys sake, preposition at is
portrayed as having been elided in fronted position
(preposition at would rather have been elided in stranded
position: which I finished my time [at]); instead of a
preposition, resumptive demonstrative adverb there
unambiguously indicates the grammatical function of relative
pronounwhich; which functions as A instead of PCOMP(A):
Then we moved to the <gap cause=anonymization desc=
address> and then I went to <gap cause=anonymization
desc=address>Road School, [[at] which I finished my time
there], fourteen. (CMI-H4B<S: 561>)
b. resumptive personal pronoun us after stranded partitive
genitive preposition of; relative pronoun which and resumptive
pronoun in PCOMP(GEN) function; (alternative analysis:
Existential there is resumptive pronoun in subject position):
So after a fortnight, thereabouts, the union had a meeting and
decided that everybody in the industry like, [which of course
there was only about a thousand of us any way], everybody
would (trunc) s would stop until this man could start at work.
(CMI-FYH<S: 556>)
c. sentential relative clause: relative marker which and resumptive
pronounthat in subject function; (less preferable analysis: that
could also be analyzed as a vacuous conjunction in the Middle
English relative marker relic which that [cf. 9.1. above]):
Because once the ship come up thats still a certain amount of
water in the hold [which that must be], cos then once you
heave your door up then of course you load your ship again
and ... (EAN-H5H<S: 684>)
d. conflation of sentential relative clause modifying bought this
and adnominal relative clause modifying this: slightly
reconstructed version of: which to get I were lucky; resumptive
pronoun it as direct object in further embedded to-infinitive
clause saves the sentence from ambiguity:
... They sold this and some at Cary and I jumped in and bought
this, [which I were lucky in a way [to get it]], you know, had
it, being a tenant you did get it cheaper, you see? (CSW-
SRLM 105<T 1420>)
e. formerly example (46c): nonreduced noun phrase in
PCOMP(A) function in further embedded to-infinitive clause:
And of course the, with the boxes being all round the side, the
dressing boxes, this was outmoded, we needed a building at
the side, [which there was sufficient room[to build something
at the side] [because we had a park at the side of it then]].
(SCO-GYW<S: 552>)
Examples (50ad) involve resumptive pronouns (or adverbs), while (50e)
involves a nonreduced noun phrase. Due to the redundant nature of
resumptives, the respective grammatical function is filled twice in a relative
clause: once by the relative marker and a second time by the resumptive.
Thus, the relative pronoun which is not deprived of its grammatical
function in the relative clause by the presence of a resumptive, but the
92 Tanja Herrmann
resumptive steps in to explicate its grammatical function in some syntactic
environments rather than in others (cf. 9.1. above). For instance:
1. In nonrestrictive adnominal and sentential relative clauses, which
are difficult and rather unknown to dialect speakers (compare 9.2.
above).
2. In cases of preposition elision, as in (50a): Instead of a preposition,
the resumptive adverb there is inserted, which has taken over the
task of unambiguously indicating the adverbial locative function in
the normal clause position.
3. In inaccessible relativization environments, such as in (50b) and
(50d): In (50b), relative pronoun which relativizes the PCOMP
position, while in (50d), relative pronoun which relativizes a
position in a further embedded clause (cf. 9.3. above).
In example (50e), inaccessibility of position (PCOMP) in terms of the
Accessibility Hierarchy combines with further embedding (specifically, a
further embedded to-infinitive clause).
Furthermore, resumptives are found with all kinds of relative markers
(cf. Table 13 above). Yet the zero marker, that, what, and who are not
analyzed as connectors instead of real relative markers when they take
a resumptive. For example:
(51) resumptive personal pronoun themin prepositional complement
position, co-occurring with relative marker that:
... it was, you know, looked upon then you were, were public
transport and the public team [that you belonged to them]. (EAN-
HDL<S: 101>)
10.3. New starts
The last group of examples of so-called connector which can be explained
as new starts:
(52) a. formerly example (46d): sentential relative clause: abandoned
after relative marker which; pause filler er, followed by new
parataxis with its own subject (I):
And they had addresses and I ran around, I went to get a job
at the pit, [which] er I had no more sense so Imglad now
that I didnt get on. (CMI-FXU<S:159>)
b. sentential relative clause: abandoned after relative marker
which; pause filler well, followed by new parataxis:
I dont really know, but er they managed to keep us looking
nice and clean and tidy er because one respected teachers and
elders [which], well, I dont think Ive ever changed anyway,
but today they dont]. (CMI-FXV<S: 090>)
c. formerly example (46e): adnominal relative clause (e.g. which
only wanted (to) drink): abandoned after relative pronoun
which in subject function, in favor of new start with all-pseudo-
cleft:
... the one down Taunton left his pottery to his son, [which all
he wanted was drink]. (CSW FRED Som_009)
Examples (52ac) belong to a third category which comprises abandoned
relative clauses after the relative marker which. A new start is made using a
paratactic construction, often preceded by a filler element like er, erm,
well, etc. and/or punctuation (like a comma in [52b]) to indicate a pause.
Both elements are indicative of reorientation in sentence planning (cf. 9.1.
above). Consequently, this type of connector which is due to an
involuntary midclause reconstruction, which proves the dominance of
parataxis over hypotaxis, particularly in difficult and unfamiliar
environments, such as nonrestrictive adnominal and (nonrestrictive)
sentential relative clauses. Dialect speakers tend to start the nonrestrictive
relative clause as a subject relative clause, which is the most basic type of
relative clause (cf. Keenan and Comrie 1977: 95).
In contrast to object relative clauses, subject relative clauses demand no
change of the normal SVO word order. According to the relative marker
fronting rule (cf. Downing 1977: 181), the relative marker in subject
function is already in front position. Relativization of a prepositional
complement with a stranded or elided preposition also avoids disrupting the
normal (paratactic) SVO word order; the fronting of the relative marker in
PCOMP position corresponds to the peripheral clause position of PCOMPs
in declarative main clauses, where prepositional complements can precede
or succeed the central SVO group. This may explain why most so-called
connector which examples either relativize nonrestrictive subjects or
(primarily adverbial) prepositional complements. (Some) dialect speakers
94 Tanja Herrmann
plan to construct a nonrestrictive relative clause, but owing to online
production constraints and unfamiliarity with nonrestrictives as well as its
structural markers, i.e. wh-pronouns, these relative clauses are finished in
an unconventional nonstandard way.
11. Conclusion
The relativization strategies found in the dialectal data are grammatically
less constrained than those of Standard English. Often they represent earlier
stages of English, containing grammatical and syntactic features which are
disfavored or banned from Present-day Standard English. While traditional
dialects use a pure relative particle, wh-pronouns are a hallmark of (written)
Standard English. Hence, the infiltration of dialects by Standard English
can be read off from the number of wh-pronouns in this dialect, and
particularly from the number of case-marked wh-pronouns, since
grammatical concord is generally absent from traditional dialects. In the
present data, relative particles prevail over relative pronouns at a ratio of
3:1, whereas case-marked relative pronouns are very scarce. In contrast to
Standard English, preposition stranding is generally preferred in the
dialectal data, even with wh-pronouns. There are further supra-regional
characteristics of traditional dialect: The relative marker which also refers
to personal antecedents; nonrestrictive relative clauses may also be
introduced by relative particles like what, that, and zero; zero subject
relative clauses are legitimate; resumptive pronouns and nonreduced noun
phrases are used for clarification, partially enabling complex constructions
unknown in Standard English; and prescriptivisms like the omission of the
conjunction that when the relative marker serves subject function in
further-embedded relative clauses (push-down elements) do not apply.
However, as traditional dialects are declining in favor of a common
nonstandard variety, the truly vernacular relative markers like as are
receding: As has retracted from East Anglia and the Central Southwest.
Nowadays as is still retreating toward the interior of the country, while its
number has decreased dramatically since the days of the compilation of the
Survey of English Dialects (in the 1950s). Even in the (Central) Midlands,
where as used to be the dominant relative marker, as has diminished
drastically in numbers. At the same time, as has backed out of several
syntactic environments like nonrestrictive relative clauses and is moving
toward relativizing only the subject (that is, the topmost) position on the
Accessibility Hierarchy, while it has largely withdrawn from the lower
positions. As is superseded by what, the novel general nonstandard relative
marker, to the mutual exclusion of the two, as they both compete for the
whole range of syntactic environments. What is spreading across Britain
geographically, both numerically and in terms of syntactic environments. It
has developed from a typical East Anglian relative marker toward a supra-
regional nonstandard marker which may eventually find its way into
spoken Standard English. Due to its affinity to informal non-posh spoken
language and thus, to simpler syntactic positions, what enters a (new)
region by the topmost position of the Accessibility Hierarchy (subject
position) and percolates down the Hierarchy. If it were found governed by
a fronted preposition (e.g. the man [for what J ack worked]), this would
even signal the beginning of a change in word class, from conjunction
(relative particle) to pronoun (relative pronoun). In addition, the relative
particles that (or its phonemic variant at) and zero proves to be prominent
in each investigated dialect. While that increases its number as one moves
farther north, zero appears to be independent of region.
The Accessibility Hierarchy was slightly modified for the present study:
SUBJ >DO >PCOMP >GEN. This Accessibility Hierarchy is confirmed
for all relative clause formation strategies in each of the investigated
regions, with some reservations for the zero relative marker strategy. In
contrast to Standard English, zero can relativize the subject position in all
dialects, yet only in two of the broad corpora (Northern Ireland and Central
North) does zero relativize most often on subjects. However, by its sheer
presence in SUBJ in four regions and its supremacy in the remaining two,
the zero marker strategy is rehabilitated as a primary relative clause
formation strategy, as the gap in subject position on the Accessibility
Hierarchy left by Standard English is remedied in dialectal English.
Relating the Accessibility Hierarchy (formulated on the basis of
restrictive relative clauses) to the pronoun retention strategy yields
unexpected results when including instances of resumptive pronouns in
nonrestrictive relative clauses: First, instead of reversing the Accessibility
Hierarchy (GEN >PCOMP >DO >SUBJ ), resumptive pronouns appear
most often in subject position, followed by direct object, prepositional
complement, and genitive position. Regarding the ease of retrievability of
the relativized noun phrase and the identification of the syntactic function
of the relative marker, relative particles are less explicit than relative
pronouns because they lack grammatical concord and case-marking.
Nevertheless, instead of co-occurring with less explicit relative particles,
96 Tanja Herrmann
resumptives rather combine with more explicit relative pronouns,
particularly which. The predominance of which is caused by the
nonrestrictiveness of these examples, as which is the nonrestrictive relative
marker par excellence. (The few restrictive relative clauses indeed select a
relative particle.) Recalling that resumptives tend to occur in difficult
syntactic environments and bearing in mind that nonrestrictive relative
clauses are characteristic of written Standard English, but uncommon or
even alien to traditional dialect speech, it becomes apparent that dialect
speakers apply resumptives when moving on unfamiliar ground, such as
nonrestrictive relative clauses. Dialect speakers use the principal
nonrestrictive relative marker which, seeking the help of a resumptive to
construct a nonrestrictive relative clause, which they construct in the most
basic way possible, that is by relativizing the subject position. In other
words, the reverse order of the Accessibility Hierarchy in the pronoun
retention strategy is not necessarily transferable from restrictive relative
clauses, for which the Accessibility Hierarchy was originally set up, to
nonrestrictive relative clauses, in which resumptives mainly occur in the
present data.
Connector which turns out to be an artifact to explain nonstandard
constructions of which relative clauses. Such examples are the product of
online production and analyzable as instances of preposition elision,
resumptives, and new starts, which are recurrent in spontaneous speech. In
all instances, the modifying character inherent in relative clauses is still
pre-eminent.
Appendix 1
Abbreviations: + = reported, - = reported as not occurring, GEN =
genitive, poss pro =possessive pronoun
Table 1. Areal distribution of relative markers in previous investigations on the
Central Southwest and Eastern Somerset in particular
Central Southwest
author/study zero that what as who which whose, whom
Elworthy
(1877, [1886]
1965): West
Somerset
+ + + only in
East
Somer-
set
- - ; GEN
avoidance
Lowsley
(1888):
Berkshire
+ no whom;
whosen
Dartnell and
Goddard
(1893):
Wiltshire
+
Kruisinga
(1905): West
Somerset
+/ [hAt] + no whose;
GEN
avoidance
Wright
([1905]
1961): East
Somerset
occurs gener-
ally
used

Wilson
(1913):
Burley, New
Forest,
Hampshire
+
Lowman
(1937/38):
Eastern
Somerset
+ as+poss pro
98 Tanja Herrmann
Table 1. cont.
Central Southwest
author/
study
zero that what as who which whose,
whom
SED
(1950s):
Eastern
Somerset
occurs occurs/ at occurs occurs
(older)
domi-
nant
whose;
as+
poss
pro;
whats;
GEN
avoid-
ance
Barth
(1968):
Naunton,
Glou-
cester-
shire
+
(newer)
+
(older)
Rogers
(1979):
Wessex
+ + + -; what
his +
poss
pro)
Ihalainen
(1980,
1985,
1987):
Somerset
+
[29.1%]
+
[25.2%]
+
[17.5%]
+
[5.8%]
in East
Somerset
only
+
[1.9%]
+
[20.4%]
1
-
J ones and
Dillon
(1987):
Wiltshire
+ + + +
van den
Eynden
(1992,
1993,
2002):
Dorset
+
[24.9%]
+
[37.1%]
+
[7.1%]
+
[13.9%]
+
[11.6%]
+
[0.6%]
2
;
zero +
poss
pro;
GEN
avoid-
ance
Table 2. Areal distribution of relative markers in previous investigations on
East Anglia and Suffolk and Eastern Cambridgeshire in particular
East Anglia zero that what as who which whose,
author/ study whom
Forby ([1830]
1970)
predomi-
nant

Wright ([1905]
1961)
occurs general-
ly used

Lowman
(1937/38):
Suffolk, East
Cambridgeshire
+ + what +
poss pro;
thats
SED (1950s):
Suffolk, East
Cambridgeshire
occurs occurs
once as
at
dominant occurs
only in
Cam-
bridge-
shire
occurs whose;
as+poss
pro;
GEN
avoidanc
e
Francis SED
Corpus [1956] in
Poussa (1999,
2001): Northern
Norfolk (Nf 19)
+
[40%]
+
[2.5%];
at [1.2%]
+
[47%]
+
[8%]
+[1.7%] no whom
Claxton (1968) + hardly
whom
Ojanen (1982):
(South)
Cambridgeshire
+
[39.5%]
3
+[1.6%] +
[53.3%]
+[0.2%] +
[3%]
+
[2.3%]
-;
GEN
avoidanc
e
Keklinen/
Peitsara (1985,
1988;
2002
4
): Suffolk
+
[24.3%]
[30.7%]
5
+/ at
[20.4%]
[23%]
+
[29.1%]
[21.2%]
+[1.9%]
[0.9%]
+
[2.9%]
[7.4%]
+
[21.4%]
[16.9%]
-;
who+
poss pro
Poussa (1994):
North-East
Norfolk
+
[31%]
+/ at
[7.9%]
+
[25.6%]
+
[19.4%]
6
+
[16.2%]
very rare
Poussa (2001):
Docking, North-
West Norfolk
very
frequent
rare frequent occurs occurs -
100 Tanja Herrmann
Central Midlands and Nottinghamshire in particular
Central Midlands
author/study zero that what as who which whose,
whom
Evans and Evans
(1881): Leicestershire
+
Pegge (1896):
Derbyshire
+
Wright ([1905]
1961):
Nottinghamshire
occurs generally
used

SED (1950s):
Nottinghamshire
occurs + whose; as+
poss pro
Storr (1977): Selston,
Erewash Valley,
Nottinghamshire
+
Central North (Cumbria = Cumberland, Westmorland, and North
Lancashire)
Central North
author/study zero that what as who which whose,
whom
Peacock ([1862/63]
1968a, [1867]
1968b):
Westmorland,
Central
Cumberland, North
Lancashire
at
Ellwood ([1895]
1965): Cumberland,
Westmorland,
North Lancashire
at
Wright ([1905]
1961):
Westmorland
occa-
sional

Hirst (1906):
Kendal,
Westmorland
occurs at -
Brilioth (1913):
Lorton, West
Cumberland
occurs
often
commonest:
at
much
less
com-
mon
-
Reaney (1927):
Penrith,
Cumberland
frequent commonest:
at
occa-
sional

Round (1949):
Broughton-in-
Furness, Lancashire
often commonest:
at
some-
times

SED (1950s):
Cumberland,
Westmorland,
North Lancashire
occurs
+
at is
dominant;
ut; t
occurs
once in
North
Lanca-
shire
occurs
once in
Cum-
berland;
occurs
in North
Lanca-
shire
oc-
curs
once
whose; at
+poss
pro, ats;
as his;
GEN
avoidance
Wright (1979) some-
times
usual:
at
occa-
sional

102 Tanja Herrmann
Scotland and Lothian, Borders, Strathclyde, and Invernesshire in
particular
Scotland
author/study zero that what as who which whose, whom
Wright ([1905]
1961)
at
Wilson (1915):
Lower
Strathearn
District of
Perthshire
+ ut +
with
all
no whose but ut +poss
pro; GEN avoidance
Grant and
Dixon (1921)
+ +/ at /
t
at +poss pro; hardly
whom
Wilson (1926):
Central
Scotland
+ +/ ut - - -
Dieth (1932):
Buchan,
Aberdeenshire
at
Grant (1931);
Grant and
Murison
(1974)
+ +/ at rare at +poss pro/ thats
Murison
([1977] 1980)
+ +/ at - obsolete
whilk
-; that +poss pro
Aitken (1979) no whose but thats;
GEN avoidance
Romaine
(1982a, b,
1984)
+ +/ at rare + no whose but thats;
GEN avoidance
Table 5. cont.
Scotland
author/
study
whom
Miller and
Brown
(1982);
Miller
(1983);
Miller and
Weinert
(1998):
Edinburgh,
West and
East
Lothian
+ typical rare + no whose
but that +
poss pro;
nowhom
Macafee
(1983):
Glasgow
+ most
common
occa-
sional in
West of
Scotland

Macaulay
(1985,
1989,
1991): Ayr
second
frequent
domi-
nant
very
rare
very rare no whose
but GEN
avoidance
McClure
(1994)
+ +/ at thats / ats
104 Tanja Herrmann
Northern Ireland
Northern
I reland

author/
study
whom
Wright
([1905]
1961):
Ireland
at
Todd
(1971):
Tyrone
+ +
Policansky
(1982):
Belfast
most
fre-
quent
fre-
quent
occurs occurs no whose;
zero +
poss pro;
GEN
avoidance
Harris
(1984,
1993):
Irish
English
(North and
South)
+ + rare rare that &
zero +
poss pro;
thats
Henry
(1995):
Belfast
+
Corrigan
(1997):
South
Armagh
most
fre-
quent
among
older
people
+ compara-
tively rare
compara-
tively rare
GEN
avoidance
Appendix 2
Map 1. S5 (LAE) (SED Question IX.9.5 who)
106 Tanja Herrmann
Map 2. M81 (LAE) (SED Question IX.9.6 whose)
Map 3. S 8a: III.3.7 that (The computer developed linguistic atlas of England 1)
108 Tanja Herrmann
Map 4. S 8b III.3.7 that (The computer developed linguistic atlas of England 1)
Map 5. S 9 IX.9.5 who (The computer developed linguistic atlas of England 1)
110 Tanja Herrmann
Map 6. S 10 IX.9.6 whose (The computer developed linguistic atlas of England 1)
Notes
1. I would like to thank J uhani Klemola for kindly giving me permission to use
his SRLM material.
2. I would like to thank J ohn Kirk for kindly allowing the English Department at
the University of Freiburg to work with the NITCS.
3. Percentages were rounded to one digit after the point, which might cause
occasional incongruence with the total sum of percentages.
4. Alternatively, the continuation with his father could also be interpreted as an
analytical genitive, i.e. as a combination of the zero relative marker +the
possessive pronoun his.
5. Alternatively, the continuation with his father could also be interpreted as an
analytical genitive, i.e. as a combination of the zero relative marker +the
possessive pronoun his.
6. Based on Poussas (1988, 1991) investigations of the three SED questions on
relative clauses, relative what seems to originate from Essex (cf. Poussa 1988:
448). In addition, Albrecht (1916: 136) in his treatise on the dialect poet
Charles E. Benham of Colchester in north Essex, records what (r p) as the
major relative marker, alongside as.
7. The results of the SED appeared in several publications: The original four
Survey of English Dialects (B) Basic Material volumes, edited by Harold
Orton et al., were published between 1962 and 1971. Dividing England into
four large geographical areas, the volumes reproduce the SED questions and
responses, including incidental material (i.e. additional, nonelicited
information on the questions), according to counties. The linguistic atlas of
England (LAE) (1978), edited by Orton, Sanderson, and Widdowson,
presents selected SED questions as maps on which isoglosses are drawn as
well as individual responses indicated. Wolfgang Vierecks comprehensive
The computer developed linguistic atlas of England (1991, 1997) illustrates
SED questions and responses, including incidental material, in great detail. It
lists individual responses and marks them via symbols on the maps. The
Survey of English Dialects: The dictionary and grammar (1994), edited by
Clive Upton et al., is an exhaustive reference work which arranges SED
questions and responses according to feature occurrences in individual
counties.
8. Occasionally, informants gave no answer or more than one answer.
9. (In)definiteness, however, did not prove to be a critical variable, neither in
investigations based on SED material nor in the present investigation, but
tends to be a by-product of the variables (non)restrictiveness and
(non)specificity, for which reason the variable (in)definiteness was dropped in
the course of this analysis.
112 Tanja Herrmann
10. E.g., We used to have niggers go along in front of us. [bare infinitive clause]
We had a combine come here way fromHolland. [causative resultative
construction (We made a combine come here way fromHolland.)] (cf.
Ojanen 1982: 77).
11. In A Middle English syntax (1960: 191), Mustanoja remarks on at: At occurs
by the side of that in the North, where it is common in the 14th and 15th
centuries. It becomes rare after 1500, being supplanted by quhilk.
12. For details on licensed borderline cases of personal which, please see Quirk et
al. 1985: 12451246, 1260, and 367; cf. also Huddleston and Pullum 2002:
1048.
13. Mair found eight definitely human antecedents among some 90 downloaded
examples when looking for the partitive genitive of which in the British
National Corpus (personal communication). For instance:
The hereditary sultans (of which there are only ...) ... (BNC-FEP: 426)
... the very special women, of which there are few, ... (BNC-GV8: 717).
14. Lexically empty antecedents involve semantically bleached noun phrases
like anybody, somebody, the best, the only thing, the only ones, the
one, the one thing, them, etc., being relatively close to focus constructions
semantically. Quirk et al. (1985: 1387) use the term "general antecedent"
instead of lexically empty antecedent.
15. For details, please see Quirk et al. 1985: 1387; cf. also Huddleston and
Pullum 2002: 1046 and 10561057.
16. Peitsara (2002: 174) also calls them cleft sentences, but unfortunately
includes them in her analysis of relative clauses nevertheless.
17. Compare also to Ihalainens (1980: 189 and 191; 1985: 6869) findings.
18. The symbol >in X >Y means Y implies X.
19. In Middle English, the frequency of zero subject relative clauses was much
higher and they were not constrained in use (cf. Moessner 1984: 71).
20. In English, non-subjects of subordinate clauses are freely relativizable, while
subjects can be relativized only if there is no conjunction:
the girl [that you think (that) I love]
the girl [that you think (*that) loves me] (Comrie 1993: 162).
21. Total might include up to three sentential relative clauses.
22. Plus 4.8% sentential relative clauses and so-called connector which clauses
to add up to 100%.
23. It contains instances which share a common surface structure with zero
relative clauses, yet are not relative clauses.
24. Her 1985 study was based on the data of only nine of the 19 informants, while
her 2002 analysis exploited the whole of the Suffolk data.
25. The figures of her 2002 analysis are slightly re-worked here, such as to
exclude adverbial relative clauses, sentential relative clauses, and clefts;
however, there might still be an unknown number of inappropriate
constructions in the data.
26. The actual absolute number might differ slightly, since Poussa does not
distinguish between who and its case-marked forms; on page 425 she records
three whose and one whom, but her Figure 1 shows 2 who in genitive
function.
References
Aitken, A. J .
1979 Scottish speech: a historical view, with special reference to the
Standard English of Scotland. In Languages of Scotland, A. J .
Aitken and Tom McArthur (eds.), 85119. Edinburgh: W & R
Chambers.
Albrecht, Theodor
1916 Der Sprachgebrauch des Dialektdichters Charles E. Benham zu
Colchester in Essex. (Palaestra CXI.) Berlin: Mayer & Mller.
Ball, Catherine N.
1994 Relative pronouns in it-clefts: The last seven centuries. Language
Variation and Change 6: 179200.
Barth, Ernst
1968 The Dialect of Naunton (Gloucestershire). Zrich: P. G. Keller.
Bever, T. G. and D. T. Langendoen
1971 A dynamic model of the evolution of language. Linguistic Inquiry 2
(4): 433463.
1972 The interaction of speech perception and grammatical structure in
the evolution of language. In Linguistic Change and Generative
Theory. Essays fromthe UCLA Conference on Historical Linguistics
in the Perspective of Transformational Theory. February 1969,
Robert P. Stockwell and Ronald K. S. Macaulay (eds.), 3295.
Bloomington/London: Indiana University Press.
Biber, Douglas, Stig J ohansson, Geoffrey Leech, Susan Conrad, and Edward
Finegan
1999 Longman Grammar of Spoken and Written English. Harlow: Pearson
Education.
Brilioth, Brje
1913 A Grammar of the Dialect of Lorton (Cumberland). Historical and
Descriptive. Inaugural diss., Uppsala: Philosophical Faculty of
Uppsala.
114 Tanja Herrmann
Cheshire, J enny, Viv Edwards, and Pamela Whittle
1993 Non-standard English and dialect levelling. In Real English. The
Grammar of English Dialects in the British Isles, J ames Milroy and
Lesley Milroy (eds.), 5397. London/New York: Longman.
Chevillet, Franois
1996 Relativization strategies in Early Middle English (and after). In
Speech Past and Present. Studies in English Dialectology in Memory
of Ossi Ihalainen, J uhani Klemola, Merja Kyt, and Matti Rissanen
(eds.), 1832. Frankfurt am Main / Berlin / Bern / New York / Paris /
Wien: Peter Lang.
Claxton, A. O. D.
1968 The Suffolk Dialect of the 20th Century. 3rd ed. Ipswich: Norman
Adlard.
Comrie, Bernard
1993 Reprint. Language Universals and Linguistic Typology. Syntax and
Morphology. 2nd ed. Oxford/Cambridge: Blackwell. Original
edition, 1989.
1998 Rethinking the typology of relative clauses. Language Design 1: 59
86.
1999 Relative clauses. Structure and typology on the periphery of standard
English. In The Clause in English. In Honour of Rodney Huddleston,
Peter Collins and David Lee (eds.), 8191. Amsterdam/Philadelphia:
J ohn Benjamins.
Comrie, Bernard and Edward L. Keenan
1979 Noun phrase accessibility revisited. Language55: 649664.
Corrigan, Karen Patrice
1997 The syntax of South Armagh English in its socio-historical
perspective. Ph.D. thesis, University College Dublin.
Dartnell, George Edward and Edward Hungerford Goddard
1893 A Glossary of Words Used in the County of Wiltshire. (English
Dialect Society 69.)
Dekeyser, Xavier
1984 Relativicers [sic] in early Modern English: A dynamic quantitative
study. In Historical Syntax, J acek Fisiak (ed.), 6187. Berlin/New
York/Amsterdam: Mouton.
Dieth, Eugen
1932 A Grammar of the Buchan Dialect (Aberdeenshire). Descriptive and
Historical. Vol. I: Phonology Accidence. Cambridge: W. Heffer &
Sons.
Downing, Bruce T.
1977 Typological regularities in postnominal relative clauses. In Current
Themes in Linguistics: Bilingualism, Experimental Linguistics and
Language Typologies, Fred R. Eckman (ed.), 163194.
Washington/London: Hemisphere.
Edwards, Viv K., Peter Trudgill, and Bert Weltens
1984 The Grammar of English Dialect. A Survey of Research. London:
Economic and Social Research Council.
Ellwood, T.
1965 Reprint. Lakeland and Iceland Being. A Glossary of Words in the
Dialect of Cumberland, Westmorland and North Lancashire. Vaduz:
Kraus. Original edition (English Dialect Society 77), 1895.
Elworthy, Frederic Thomas
1877 An Outline of the Grammar of the Dialect of West Somerset
Illustrated by Examples of the Common Phrases and Modes of
Speech Now in Use among the People. (English Dialect Society 19.)
1965 Reprint. The West Somerset Word-Book. A Glossary of Dialectal
and Archaic Words and Phrases Used in the West of Somerset and
East Devon. Vaduz: Kraus. Original edition (English Dialect Society
50), 1886.
Evans, Arthur Benoni and Sebastian Evans
1881 Leicestershire Words, Phrases, and Proverbs. (English Dialect
Society 31.)
Forby, Robert
1970 Reprint. The Vocabulary of East Anglia. Vol. 1. Newton Abbot,
Devon: David & Charles. Original edition, 1830.
Fox, Barbara A.
1987 The noun phrase accessibility hierarchy reinterpreted: Subject
primacy or the absolutive hypothesis? Language 63: 856870.
Gass, Susan M. and Larry Selinker
1994 Second Language Acquisition. An Introductory Course. Hillsdale,
New J ersey/Hove/London: Lawrence Erlbaum.
Givn, Talmy
1993 English Grammar. A Function-Based Introduction. Vol. II.
Amsterdam/Philadelphia: J ohn Benjamins.
Graf, Dorothee
1996 Relative clauses in their discourse context. A corpus-based study.
M.A. thesis, University of Freiburg.
Grant, William (ed.)
1931 The Scottish National Dictionary. Vol. I. Edinburgh: The Scottish
National Dictionary Association.
Grant, William and J ames Main Dixon
1921 Manual of Modern Scots. Cambridge: University Press.
116 Tanja Herrmann
Grant, William and David D. Murison (eds.)
1974 The Scottish National Dictionary. Vol. IX. Edinburgh: The Scottish
National Dictionary Association.
Harris, J ohn
1984 Syntactic variation and dialect divergence. J ournal of Linguistics 20:
303327.
1993 The grammar of Irish English. In Real English. The Grammar of
English Dialects in the British Isles, J ames Milroy and Lesley
Milroy (eds.), 139186. London/New York: Longman.
Hawkins, J ohn A.
1999 Processing complexity and filler-gap dependencies across grammars.
Language 75: 244285.
Henry, Alison
1995 Belfast English and Standard English. Dialect Variation and
Parameter Setting. New York/Oxford: Oxford University Press.
Hirst, T. O.
1906 A Grammar of the Dialect of Kendal (Westmoreland). Descriptive
and Historical. Heidelberg: Winter.
Huddleston, Rodney
1984 Introduction to the Grammar of English. London: Cambridge
University Press.
Huddleston, Rodney and Geoffrey K. Pullum
2002 The Cambridge Grammar of the English Language. Cambridge/New
York/Melbourne/Madrid/Cape Town: Cambridge University Press.
Hyltenstam, Kenneth
1984 The use of typological markedness conditions as predictors in
second language acquisition: The case of pronominal copies in
relative clauses. In Second Languages: A Crosslinguistic Per-
spective, R. Andersen (ed.), 3958. Rowley, MA: Newbury House.
Ihalainen, Ossi
1980 Relative clauses in the dialect of Somerset. Neuphilologische
Mitteilungen 81: 187196.
1985 Synchronic variation and linguistic change: Evidence from British
English dialects. In Papers fromthe 4th International Conference on
English Historical Linguistics, Amsterdam, 1013 April, 1985,
Roger Eaton, Olga Fischer, Willem Koopman, and Frederike van der
Leek (eds.), 61-72. Amsterdam/Philadelphia: J ohn Benjamins.
1987 Towards a grammar of the Somerset dialect. A case study of the
language of J .M. In Neophilologica Fennica. Socit
Nophilologique 100 Ans, Leena Kahlas-Tarkka (ed.), 7186.
Helsinki: Socit Nophilologique.
J acobsson, Bengt
1994 Nonrestrictive relative that-clauses revisited. Studia Neophilologica
66: 181195.
J espersen, Otto
1927 A Modern English Grammar on Historical Principles. Part III:
Syntax, 2nd vol. London/Copenhagen: George Allen & Unwin/Ejnar
Munksgaard.
J ohansson, Christine
1993 Whose and of which with non-personal antecedents in written and
spoken English. In Corpus-Based Computational Linguistics, Clive
Souter and Eric Atwell (eds.), 97117. Amsterdam/Atlanta, GA:
Rodopi.
J ones, Malcolm and Patrick Dillon
1987 Dialect in Wiltshire and its Historical, Topographical and Natural
Science Contexts. Trowbridge, Wiltshire: Wiltshire County Council.
Keenan, Edward L.
1985 Relative clauses. In Language Typology and Syntactic Description.
Vol. II: Complex Constructions, Timothy Shopen (ed.), 141170.
Cambridge: Cambridge University Press.
Keenan, Edward L. and Bernard Comrie
1977 Noun phrase accessibility and universal grammar. Linguistic Inquiry
8 (1): 6399.
Keenan, Edward L. and Sarah Hawkins
1987 The psychological validity of the accessibility hierarchy. In
Universal Grammar: 15 Essays, Edward L. Keenan (ed.), 6085.
London/Sydney/Wolfeboro, New Hampshire: Croom Helm.
Keklinen, Kirsti
1985 Relative clauses in the dialect of Suffolk. Neuphilologische
Kruisinga, E.
1905 A Grammar of the Dialect of West Somerset. Descriptive and
Historical. Bonn: P. Hanstein.
Lehmann, Christian
1984 Der Relativsatz: Typologie seiner Strukturen, Theorie seiner
Funktionen, Kompendiumseiner Grammatik. Tbingen: Narr.
Lowsley, B.
1888 A Glossary of Berkshire Words and Phrases. (English Dialect
Society 56.)
Lumsden, Michael
1988 Existential Sentences. Their Structure and Meaning. London/New
York/Sydney: Croom Helm.
118 Tanja Herrmann
Macafee, Caroline
1980 Characteristics of non-standard grammar in Scotland. Manuscript.
1983 Glasgow. Amsterdam/Philadelphia: J ohn Benjamins.
Macaulay, Ronald K. S.
1985 The narrative skills of a Scottish coal miner. In Focus on: Scotland,
Manfred Grlach (ed.), 101124. Amsterdam/Philadelphia: J ohn
Benjamins.
1989 He was some man him: Emphatic pronouns in Scottish English. In
Synchronic and Diachronic Approaches to Linguistic Variation and
Change, Thomas J . Walsh (ed.), 179-187. Washington, D. C.:
Georgetown University Press.
1991 Locating Dialect in Discourse. The Language of Honest Men and
Bonnie Lasses in Ayr. New York/Oxford: Oxford University Press.
Mair, Christian
1998 Man/woman which ... Last of the old, or first of the new? In
Explorations in Corpus Linguistics, Antoinette Renouf (ed.), 123
133. Amsterdam: Rodopi.
Maxwell, Daniel N.
1979 Strategies of relativization and NP accessibility. Language 55: 352
371.
McClure, J . Derrick
1994 English in Scotland. In The Cambridge History of the English
Language. Vol. V: English in Britain and Overseas: Origins and
Development, Robert Burchfield (ed.), 2393. Cambridge/New
York/Melbourne: Cambridge University Press.
Miller, J im
1993 The grammar of Scottish English. In Real English. The Grammar of
English Dialects in the British Isles, J ames Milroy and Lesley
Milroy (eds.), 99138. London/New York: Longman.
Miller, J im and Keith Brown
1982 Aspects of Scottish English syntax. English World-Wide 3: 318.
Miller, J im and Regina Weinert
1998 Spontaneous Spoken Language. Syntax and Discourse. Oxford:
Clarendon Press.
Moessner, Lilo
1984 Some English relative constructions. La Linguistique 20 (1): 5779.
Moss, Fernand
1991 Reprint and translation. A Handbook of Middle English.
Baltimore/London: The J ohns Hopkins University Press. Original
edition, 1952.
Murison, David
1980 Reprint. The Guid Scots Tongue. Edinburgh: William Blackwood.
Original edition, 1977.
Mustanoja, Tauno F.
1960 A Middle English Syntax. Part I: Parts of Speech. Helsinki: Socit
Nophilologique.
Nevalainen, Terttu and Helena Raumolin-Brunberg
2002 The rise of the relative who in Early Modern English. In
Relativisation on the North Sea Littoral, Patricia Poussa (ed.), 109
121. Mnchen: Lincom Europa.
Newbrook, Mark
1997 Relative THAT, that isnt always restrictive. I.T.L. Review of
Applied Linguistics 115116: 3960.
Ojanen, Anna-Lisa
1982 A syntax of the Cambridgeshire dialect. Licentiate thesis, University
of Helsinki.
Orton, Harold, Stewart Sanderson, and J ohn Widdowson
1978 The Linguistic Atlas of England. London: Croom Helm.
Peacock, Robert Backhouse
1968a Reprint. A Glossary of the Dialect of the Hundred of Lonsdale,
North and South of the Sands, in the County of Lancaster, Together
with an Essay on Some Leading Characteristics of the Dialects
Spoken in the Six Northern Counties of England (Ancient
Northumbria). (Transactions of the Philological Society.) Original
edition, 1867.
1968b Reprint. On Some Leading Characteristics of Northumbrian; and on
the Variations in its Grammar fromthat of Standard English, with
their Probable Etymological Sources. (Transactions of the
Philological Society: 232264.) Original edition, 1862/63.
Pegge, Samuel
1896 Two Collections of Derbicisms. (English Dialect Society 78.)
Peitsara, Kirsti
1988 On existential sentences in the dialect of Suffolk. Neuphilologische
2002 Relativizers in the Suffolk dialect. In Relativisation on the North Sea
Littoral, Patricia Poussa (ed.), 167180. Mnchen: Lincom Europa.
Policansky, Linda
1982 Grammatical variation in Belfast English. Belfast Working Papers in
Language and Linguistics 6: 3766.
120 Tanja Herrmann
Poussa, Patricia
1988 The relative WHAT: two kinds of evidence. In Historical
Dialectology. Regional and Social, J acek Fisiak (ed.), 443474.
1991 Origins of the non-standard relativizers what and as in English. In
Language Contact in the British Isles. Proceedings of the Eighth
International Symposiumon Language Contact in Europe, Douglas,
Isle of Man, 1988, P. Sture Ureland and George Broderick (eds.),
295315. Tbingen: Max Niemeyer.
1994 Norfolk relatives (Broadland). In Verhandlungen des internationalen
Dialektologenkongresses, Bamberg 29.7.4.8. 1990. Regional-
sprachliche Variation, Umgangs- und Standardsprachen. Vol. 3,
Wolfgang Viereck (ed.), 418426. Stuttgart: Franz Steiner.
1996 Gender-specific occupations and speech among North-East Norfolk
working-class informants born 18991915. In IX International Oral
History Conference, Gteborg 1316th J une, 1996. Vol. 3: Peoples
History; Political and Popular Movement, 524533. From a pre-
printed collected conference papers.
1999 The Flemings in Norman Norfolk: Their possible influence on
relative pronoun development. In Thinking English Grammar. To
Honour Xavier Dekeyser, Professor Emeritus, Guy A. J . Tops, Betty
Devriendt, and Steven Geukens (eds.), 8999. Leuven/Paris: Peeters.
2001 Syntactic change in north-west Norfolk. In East Anglian English,
J acek Fisiak and Peter Trudgill (eds.), 243259. Cambridge: D. S.
Brewer.
Quirk, Randolph
1968 Reprint. Essays on the English Language. Medieval and Modern.
London/Harlow: Longmans. Original edition, 1957.
Quirk, Randolph, Sidney Greenbaum, Geoffrey Leech, and J an Svartvik (eds.)
1985 A Comprehensive Grammar of the English Language. London/New
York: Longman.
Reaney, Percy H.
1927 A Grammar of the Dialect of Penrith (Cumberland). Descriptive and
Historical. Manchester: University Press and London/New York:
Longmans, Green & Co.
Rogers, Norman
1979 Wessex Dialect. Bradford-on-Avon, Wiltshire: Moonraker Press.
Romaine, Suzanne
1980 The relative clause marker in Scots English: Diffusion, complexity,
and style as dimensions of syntactic change. Language in Society 9:
221247.
1982a Socio-Historical Linguistics. Its Status and Methodology.
Cambridge: Cambridge University Press.
1982b The reconstruction of language in its social context. Methodology
for a socio-historical linguistic theory. In Papers from the 5th
International Conference on Historical Linguistics, Anders Ahlqvist
(ed.), 293303. Amsterdam/Philadelphia: J ohn Benjamins.
1984 Some historical and social dimensions of syntactic change in Middle
Scots relative clauses. In English Historical Linguistics. Studies in
Development, N.F. Blake and Charles J ones (eds.), 101122.
Sheffield: The Centre for English Cultural Tradition and Language,
University of Sheffield.
Round, Inabel G.
1949 The living dialect of Broughton-in-Furness in the county of
Lancashire. M.A. thesis, University of Leeds.
Schmied, J osef
1993 Qualitative and quantitative research approaches to English relative
constructions. In Corpus-Based Computational Linguistics, Clive
Souter and Eric Atwell (eds.), 8596. Amsterdam/Atlanta, GA:
Rodopi.
Seppnen, Aimo and Gran Kjellmer
1995 The dog thats leg was run over. On the genitive of the relative
pronoun. English Studies 26: 389400.
Shnukal, Anna
1981 Theres a lot mightnt believe this .... Variable subject relative
pronoun absence in Australian English. In Variation Omnibus,
David Sankoff and H. Cedergren (eds.), 321328. Carbondale, IL:
Linguistic Research.
Simpson, J . A. and E. S. C. Weiner
1989 Oxford English Dictionary, R. W. Burchfield (ed.). Vols. 120. 2nd
ed. Oxford: Clarendon Press.
Smith, Evan Shreeve
1984 Relative that and as: A Study in Category Change. Ph.D. thesis,
Indiana University, 1982. University Microfilms International, Ann
Arbor, MI.
Stockwell, Robert P., Paul Schachter, and Barbara Hall Partee
1973 The Major Syntactic Structures of English. New York: Holt,
Rinehart and Winston.
Storr, J eremy George
1977 Survey of the dialect of Selston in the Erewash Valley. M.A. thesis,
University of Sheffield.
122 Tanja Herrmann
Tagliamonte, Sali
2002 Variation and change in the British relative marker system. In
Todd, Loreto
1971 Tyrone English. Transactions of the Yorkshire Dialect Society 13:
2940.
Trudgill, Peter
2003 Reprint. The Dialects of England. 2nd ed. Malden, MA/Oxford, UK:
Blackwell. Original edition, 1999.
Upton, Clive, David Parry, and J . D. A. Widdowson (eds.)
1994 Survey of English Dialects: The Dictionary and Grammar.
London/New York: Routledge.
Van den Eynden (Morpeth), Nadine
1992 Relativization in the Dorset dialect. In History of Englishes. New
Methods and Interpretations in Historical Linguistics, Matti
Rissanen, Ossi Ihalainen, Terttu Nevalainen, and Irma Taavitsainen
(eds.), 532555. Berlin/New York: Mouton de Gruyter.
1993 Syntactic Variation and Unconscious Linguistic Change. A Study of
Adjectival Relative Clauses in the Dialect of Dorset. Frankfurt am
Main/Berlin/Bern/New York/Paris/Wien: Peter Lang.
2002 Relativisers in the Southwest of England with special emphasis on
subject contact clauses, preposition placement with particle and
pronominal relative strategies and pushdown relatives. In
Viereck, Wolfgang
1975 Lexikalische und grammatische Ergebnisse des Lowman-Survey von
Mittel- und Sdengland. 2 vols. Mnchen: Wilhelm Fink.
1980 The dialectal structure of British English: Lowmans evidence.
English World-Wide 1: 2544.
Viereck, Wolfgang (ed.), in collaboration with Heinrich Ramisch
1991 The Computer Developed Linguistic Atlas of England. Vol. 1.
Tbingen: Niemeyer.
Wilson, J ames
1913 The Dialect of the New Forest in Hampshire (as Spoken in the
Village of Burley). London: Oxford University Press/Amen Corner.
1915 Lowland Scotch as Spoken in the Lower Strathearn District of
Perthshire. London: Oxford University Press/Humphrey Milford.
1926 The Dialects of Central Scotland. London: Oxford University
Press/Humphrey Milford.
Wright, J oseph
1961 The English Dialect Grammar. Oxford: Oxford University Press.
Original edition, 1905.
Wright, Peter
1979 Cumbrian Dialect. Clapham: Dalesman Books.
Some do and some doesnt: Verbal concord
variation in the north of the British I sles
Lukas Pietsch
Abstract
Among the chief characteristics of the northern dialects since Middle English times
has been the so-called Northern Subject Rule, a systemic split in the verbal
concord system which allows for invariant verbal -s forms everywhere except
when the verb is directly accompanied by a simple personal pronoun. This study
provides a geographical and comparative survey of the reflexes of this pattern in
the northern dialects, drawing attention to their variability and to their interaction
with other related and/or competing patterns of concord variation. A corpus
investigation reveals that over and above the hard constraints that define the
Northern Subject Rule as such, there exist a number of soft probabilistic
constraints governing its effects which are also near-universally shared between
the varieties in question. I then go on to discuss the likely paths of historical
development that have given rise to this grammatical pattern, and critically review
some attempts that have been made to account for it in terms of formal syntactic
theories. I show that existing formal models fail to account for the range of
variability of this pattern, both in a comparative, diatopic perspective and on the
level of individual speakers. I finally argue that variation phenomena of this kind
can theoretically be better accounted for in a usage-based model in the vein of
current functionalist and emergentist theories.
1. I ntroduction
The verbal concord system is an area of particularly rich regional variation
in English. Several quite distinct regional types of non-standard agreement
systems exist in the traditional dialects. Some of them are also well attested
historically and can be traced as far back as the Middle English period.
Today these patterns coexist, and sometimes compete, with other types of
non-standard agreement forms, which have spread through the modern
vernaculars. Some of these even appear to have the status of vernacular
universals (Chambers 2004).
126 Lukas Pietsch
Within this complex field of variation, one pattern stands out as
particularly interesting, both from a historical and from a theoretical
perspective: the so-called Northern Subject Rule. According to this rule, the
Standard English contrast between verbal -s in the third person singular and
zero forms elsewhere is observed only where the subject is one of the
closed set of simple personal pronouns: he/she/it goes; I/you/we/they go.
All other subjects can take an invariant -s form of the verb. Moreover, the
agreement contrast in the pronominal subjects is found consistently only
when the subject directly precedes or follows the verb; in other cases, even
I, you, we and they may take the -s form. Agreement contrasts are thus
organized not or not exclusively along the lines of the subjects person-
number features, but instead they are sensitive to the morphological type
and syntactic position of the subject. This intriguing system has been
among the chief grammatical characteristics of the dialects of northern
Britain, including Scots, since the Middle English period. It is also found in
Irish English, above all in the varieties of Ulster. Some of its reflexes,
especially with regard to the usage of was and were, can be found in
overseas vernacular varieties all over the English-speaking world (cf. for
instance Montgomery 1988, 1997, Tagliamonte 2002). The latter varieties
must remain outside the scope of the present paper, which will concentrate
on those varieties where the pattern originated historically, namely those of
northern Britain, as well as their immediate geographical neighbour and
offspring, the dialects of Ulster.
Verbal concord variation in English nonstandard varieties has received
some more attention in past research than many other phenomena of
grammatical, especially syntactic, variation. This is certainly due to the fact
that it is a high-frequency phenomenon, whose occurrences in texts can
relatively easily be identified and counted. It has been given some
treatment in traditional dialectology, at least occasionally (e.g. Murray
1873; Wright 1892, 1905; Orton et al., ed. 19621971). The Northern
Subject Rule, in particular, is routinely mentioned in dialect descriptions,
though often only in a summary fashion (Beal 1993, 1997, 2004; Miller
and Brown 1982; Miller 1989, 1993, 2004; Macafee 1983, 1994; Harris
1993; Ihalainen 1994; Robinson 1997, Filppula 1999, Shorrocks 1999).
Some of the major historical patterns of verbal concord have also been
documented for older varieties (e.g. Macafee 1993, McIntosh, Samuels and
Bensik 1986). Recently, verbal concord variation has become a stock-in-
trade of corpus-based studies in variation and change. Some of them have
been based on diachronic corpora (e.g. Meurman-Solin 1992, Kyt 1993,
Verbal concord variation 127
Montgomery 1994, Ogura-Wang 1996), while most studies used
synchronic local corpora from individual speech communities, either
historical (e.g. Bailey and Ross 1988, Bailey, Maynor and Cukor-Avila
1989, Montgomery, Fuller and DeMarse 1993; Schendl 1996, 2000, Wright
2002, McCafferty 2003) or contemporary (e.g. Tagliamonte 1999, Smith
and Tagliamonte 1998, Godfrey and Tagliamonte 1999, Britain 2002,
Schreier 2002, Peitsara 2002).
The present chapter reports on a new corpus-based study (Pietsch 2003,
2005) which attempts to complement this existing research in two ways.
With respect to empirical description, a mostly diatopic-comparative
perspective was chosen. Being based on dialectal speech recordings
sampled over relatively large areas, the study traced the distribution of
related variation patterns through space and also (in some parts) through the
apparent-time dimension. The aim was to identify commonalities and
differences in the linguistic conditioning of inherent, quantifiable variation
across related varieties. The main result is that over and above the
defining constraints constituting the NSR itself there exist a number of
recurrent types of constraints which act as conditioning factors on concord
variation. They can be detected statistically in the form of probabilistic
effects, and they can best be described in terms of prototypical syntactic
environments which idiomatically favour a certain morphological
realization of the verb over the other.
The second contribution lies in an attempt to forge a closer link between
this descriptive, empirical work on the one hand and recent grammatical
theorizing on the other. Most of the existing studies of verbal concord have
treated the phenomenon of quantifiable variation mainly under its societal
(in Chomskyan terms: E-language) aspects. On the other side, those
among the studies of dialect grammar that have been inspired by theoretical
concerns about individual linguistic competence (I-language) have
tended to concentrate on variation in the sense of differences between
lects, but have rarely addressed the issue of variability within lects (see
papers in Black and Motapayane 1996 and Barbiers, Cornips and van der
Kleij 2002; for a notable recent exception see Henry 2002). There are only
a few studies offering specific analyses regarding a formal theoretical
characterization of concord phenomena of the northern type (Henry 1995,
Corrigan 1997, Brjars and Chapman 1998, Hudson 1999), and none of
them addresses inherent variability. I will therefore propose a fresh attempt
to relate the empirical observations regarding variable grammatical
performance to the theoretical question of how variation is anchored in
128 Lukas Pietsch
grammatical competence. In doing so, I will seek explanations in a
theoretical framework inspired by functionalist and cognitive research
traditions (Bybee 1985, Bybee and Hopper 2001, Langacker 1987; Croft
1995, 2001).
This chapter is organized as follows. Section 2 gives a descriptive
survey of the phenomena in question, and raises some of the problems
involved in an adequate synchronic analysis. Section 3 reports on empirical
corpus findings with respect to the patterns of quantitative variation and
their geographical distribution as found in a range of twentieth-century
dialects. Section 4 presents an excursus into the history of the northern
concord pattern and the hypothetical diachronic processes of its first
emergence in early Middle English. Section 5 presents a critical discussion
of several theoretical proposals that have been made to account for the
effects of the Northern Subject Rule in modern dialects. Finally, Section 6
presents a concluding discussion and an outlook on possible theoretical
solutions to the problems presented in the earlier sections.
2. The Northern Subject Rule: Descriptive problems
The most concise descriptive definition of the phenomenon in question can
be given as follows:
(1) The Northern Subject Rule (Version A): concord verbs
1
take the -s
form with all subjects, except with the personal pronouns I, we, you
andthey when they are directly adjacent to the verb.
However, this is only a somewhat idealized statement, describing a
hypothetical, pure northern concord system. Indeed, varieties that come
reasonably close to this have been identified, for instance in Older Scots
and northern Middle English (Montgomery 1994; McIntosh, Samuels and
Bensik 1986). In these older varieties, verbal -s in fact occurred with near-
categorical regularity in all environments where the rule licensed it. In
Modern English varieties, however, the system is always a variable one. It
has therefore been customary in the literature since Montgomery (1994) to
describe the variation patterns observed in terms of two separate
constraints, the first of them most often called the Type-of-Subject
Constraint, the second variously Position-of-Subject Constraint, Sequence
Constraint, Proximity-to-Subject Constraint or similarly.
(2) The Northern Subject Rule (Version B):
a. All third singular subjects (and, where preserved, the old
second singular thou) always take verbal -s.
b. The Type-of-Subject Constraint: All other subjects except
the personal pronouns I, we, you, they (and, where it exists,
youse) take verbal -s variably.
c. The Position-of-Subject Constraint: Non-adjacency of
subject and verb favours verbal -s.
Condition (2c) may apply, in principle, to all types of subjects. This means,
on the one hand, that the prohibition of verbal -s with I, we, you and they
may be overridden if the verb and the pronoun are not adjacent. With
respect to the other environments, on the other hand, it means that there is
commonly a quantitative effect further increasing the likelihood of verbal
-s. However, the nature of this constraint is difficult to define exactly. Some
of the effects in question appear to be cross-dialectal universals that can be
observed even in varieties not directly affected by the Northern Subject
Rule. There seems to be a strong tendency in many dialects of English that
clauses that diverge from the canonical structure, of a syntactically simple
subject immediately followed by the verb, may display lack of agreement.
For instance, it has often been observed that complex subjects consisting of
two conjoined singular noun phrases may trigger an invariant third singular
verb form, as in the following Early Modern English example (Visser 1963:
80):
(3) I and my company was arrested ij days at Dunckyrke.
Another environment that seems to stand out is subject-verb inversion, for
instance in questions and under locative inversion. Non-agreement in the
latter type of clause seems to have been common in English already during
the Old English period (4). The same has been true for almost all forms of
English in the case of existential there clauses, a special clause type that
developed through grammaticalization out of the more general schema of
the locative inversion. Non-agreement in this type of clause is found quite
independently of the Northern Subject Rule (5).
(4) On m selfan hrgle ws eac awriten a naman ara twelf
heahfdra (On that same garment was also written the names of the
twelve patriarchs) [lfred, C.P. 6,15, quoted after Visser 1963: 73]
130 Lukas Pietsch
(5) There is two or three lords and ladies more married [Shakespeare,
Mids. IV, II, 16, quoted after Visser 1963: 74]
Interestingly, it will be found that just among those modern dialects that
otherwise follow the Northern Subject Rule in allowing many non-standard
verbal -s forms, there are some that exhibit variation in exactly the opposite
direction with respect to the existentials: here, they allow non-standard,
seemingly plural verb forms even with singular subjects.
Whereas concord variation in existentials is thus largely independent of
the Northern Subject Rule, the situation is different with subject-verb
inversion in questions. This type of clause is neatly integrated with the
general pattern of the Northern Subject Rule in the northern dialects. Thus,
under interrogative inversion, full noun phrases are clearly distinguished
from pronominal subjects just as in canonical subject-verb order. Whereas
full noun phrase subjects in inversion may trigger a fairly strong effect in
favour of verbal -s, even more so than full noun phrase subjects in
canonical position, inverted pronoun subjects faithfully follow the Type-of-
Subject Constraint, displaying agreement no less regularly than in other
positions. The northern dialects thus regularly display forms such as have
they, not *has they. Indeed, it will be argued in section 4 that exactly these
environments, of verbs with inverted adjacent pronouns, may in fact
represent the historical core and point of origin of the Northern Subject
Rule as a whole, and that the occurrence of the forms without -s in these
environments has always been one of its central features.
Relative clauses, especially those following clefting its or existential
there, are another environment where non-agreement is often observed, and
these relative clause environments will be found to play a major role for the
dialects affected by the Northern Subject Rule too.
A case that is problematic for the statement of the Northern Subject
Rule is the one where subject and verb stand in the canonical order of the
declarative clause but are separated from each other by intervening, clause-
internal material. In Modern English, this may apply to either adverbs or
so-called floating quantifiers, as in I often go or they all go. The Position-
of-Subject Constraint as stated above predicts forms like I often goes or
they all goes. Indeed, such forms occur, but only in the most purely
northern, older varieties such as Old Scots do they reach a high amount of
regularity:
(6) a. that we lely heichtis and grantis [Montgomery 1994: 89]
b. weall hes mater to thank God [Montgomery 1994: 89]
In present-day dialects such as Northern Irish English (cf. section 3.3.4
below), similar forms are also attested but much rarer:
(7) a. Oh never, they never was so strict, at that time, anyway
[NITCS: L10.2]
b. And they, they both was yoked onto it. [NITCS: L19.3]
In other dialects such as Yorkshire and Lancashire English, similar forms
with adverbs like often, never, always are fairly common, but it has been
argued (Shorrocks 1999: 112, 116117) that this usage is governed not so
much by the syntactic environment but by the temporal semantics of the
adverb, as verbal -s in these varieties also functions as a marker of
habituality.
The last remaining type of environment where verbs typically occur in
positions non-adjacent to their pronominal subjects is found in co-ordinated
verb phrases. The second and any subsequent members of a series of verbs
sharing the same subject will regularly take s in typical northern dialects.
Clauses of this type, they sing and dances, are often quoted as prototypical
instances of the Northern Subject Rule. Indeed, of all the subtypes of
Position-of-Subject effects, this appears to be the one that is most
characteristic and specific to the northern dialects.
Summing up, it may be said that whereas the Type-of-Subject
Constraint has been a stable and fairly unmistakeable feature of northern
dialects, the Position-of-Subject Constraint may be regarded as the
composite effect of several different patterns, of varying degrees of
regularity, only some of which are specifically characteristic of the northern
dialects whereas others are shared with many varieties elsewhere. Only in
those older dialects which were least affected by standardizing influences
from the south or other similar dialect contact or levelling effects, can it be
said that the Position-of-Subject Constraint was a unified, tightly integrated
feature of a consistent grammatical system.
132 Lukas Pietsch
3. Data from twentieth-century northern dialects
3.1. Data and methods
In the following sections, I will report on empirical findings regarding the
quantitative variation patterns and the geographical distribution of reflexes
of the northern concord system and some related phenomena in twentieth-
century dialects. These data are based in part on the Survey of English
Dialects (SED, Orton et al., ed. 19621971); partly on the Northern Ireland
Transcribed Corpus of Speech (NITCS, Kirk 1991); and partly on a subset
of a preliminary version of the Freiburg English Dialect Corpus (FRED;
see Kortmann and Wagner, this volume). These data are of rather different
kinds, and caution must be exercised in interpreting them in order to make
results commensurable. Nevertheless, taken together they do give a fairly
comprehensive picture of a range of grammatical varieties across a large
geographic area.
The NITCS is a corpus of some 230,000 words, collected across a
geographically regular grid of 38 mostly rural locations in Northern
Ireland, sampling (ideally) one speaker from each of three age groups (9
12, 3545, and 6575) from each location. It is based on unscripted
interviews conducted in the context of a dialect atlas project, the Tape-
Recorded Survey of Hiberno-English Speech (TRS), during the 1970s (cf.
Barry 1981b). In representing speech from different age groups, this project
went an important step beyond the traditional design of dialect atlas
surveys. Hence, it can be used for analyses not only along a diatopic but
also along a diachronic, apparent-time dimension. For the purposes of this
study, some 17,000 tokens of clauses with concord verbs were extracted
and tagged. Obviously, many of these actually, their great majority are
not of much interest for this study, as they exemplify environments that
display no or only marginal variation. In general, this goes for all third
person singular tokens, and for collocations with immediately adjacent
personal pronoun subjects. The main focus of the analysis therefore were
the approximately 1,000 tokens of third-person plural verbs with non-
pronominal subjects. A separate study was made of existential thereclauses
(approx. 2,400 tokens). Tokens were classified for a range of different
environment variables (such as syntactic constellation or morphological
type of subject, in addition to the obvious variables of person and number).
Multivariate analyses were then conducted, using the well-known Varbrul
system, examining these intra-linguistic context variables as well as a
number of social, extra-linguistic variables such as age, sex, and religious
denomination.
These data, which provide for a fairly detailed view of the linguistic
situation within the small geographical area of Northern Ireland, were
complemented with data from FRED, illustrating a much wider
geographical range of varieties. The subcorpus of FRED selected for the
present study consisted of approximately 300,000 words of transcribed
speech from 63 informants. It represented Scottish predominantly
Lowlands speakers as well as different regions within the north of
England. Most of the texts were conversations recorded for purposes of
local oral history projects. They were recorded between the 1970s and the
1990s, and typically consist of interviews between a fieldworker and an
informant, centring about topics such as life in the old days, working
conditions, war experiences and so forth. The informants were typically
elderly people, of predominantly working-class background. Owing to the
interview situation, speech styles tended to be relatively formal, but the
speech nevertheless displayed a considerable range of local non-standard
linguistic features. On the whole, these recordings are thus fairly similar in
style to the interviews with the older age group in the NITCS. The Scottish
recordings exemplify different speech styles along the continuum between
broad Lowlands Scots and Standard Scottish English; similarly, the English
recordings range from strongly local, containing conservative dialectal relic
forms, to something fairly close to the standard. To these texts were added
six additional recordings from the TRS, which for technical reasons could
not be included in the NITCS. These are from Counties Donegal, Leitrim,
and Louth. While these data, taken together, represent a much larger area
than those of the NITCS, they evidently lack the systematicity of sampling
characteristic of the latter, which means that they cannot be used with the
same degree of reliability for investigations of social or fine-grained
geographical variation. Nevertheless, multivariate analyses of this material
revealed highly interesting patterns that matched or complemented findings
from the NITCS in a number of ways.
To add a further dimension of real-time depth, as well as more fine-
grained geographical information, the corpus data were supplemented with
data from the Survey of English Dialects (SED). Obviously, these cannot be
analysed with the same quantitative methods and in direct comparison with
corpus data of the former kind. The SED data, collected during the 1950s,
consist of isolated lists of elicited tokens, not exhaustive records of all
forms produced in natural speech, and hence do not lend themselves to a
134 Lukas Pietsch
quantitative assessment of intra-speaker variation. Nevertheless, the SED
data provides valuable evidence for the geographical distribution of some
morphosyntactic variables in the traditional dialects, and, if interpreted with
the necessary caution, also for some of their quantitative aspects. The SED
has repeatedly been used for studies of subject-verb agreement (Orton,
Sanderson and Widdowson 1978; Viereck 1991/1997; Trudgill 1990);
Ihalainen 1991, 1994; Klemola 1996, 2000; Bresnan and Deo 2001; Wright
2002; and Britain 2002). However, none of these studies has fully
exhausted its possibilities with regard to a study of the Northern Subject
Rule and related phenomena.
Most of the relevant questionnaire items in the SED deal with the
morphology of the primary verbs be, have and do. Some others deal with
the agreement morphology of lexical verbs after various pronouns. Only
two questions are specifically aimed at morphosyntactic environments for
the Northern Subject Rule, namely, lexical verbs after full noun phrase
subjects (III.10.7 bulls bellow, and VIII.7.5 burglars steal them). A
number of other questions were primarily aimed not at morphosyntactic but
at various lexical, phonological or idiomatic targets, but nevertheless
provide material that contains tokens of present-tense subject-verb
combinations which can be included in the analysis. As for most other
questions eliciting verbs as answers, the material unfortunately does not
regularly include subject forms and is therefore unusable.
Apart from these, the SED fieldworkers recorded a large number of
additional utterances produced spontaneously by the informants during the
interviews, whenever they felt these utterances illustrated interesting dialect
features. These recorded tokens are known as incidental material in the
SED. In cases where the features illustrated by these utterances were also
the topic of one of the systematic questionnaire items, incidental material
evidence was sometimes included in the published Basic Material of the
SED (Orton et al. 19621971, henceforth SED-BM). However, this
coverage of the incidental material is far from complete. For instance, the
list of incidental tokens of plural verbal -s given under questionnaire item
iii.10.7 (bulls bellow) is mostly though not quite consistently
restricted to tokens that match the elicited grammatical context in a rather
narrow sense: subject NPs headed by lexical nouns in canonical, declarative
SV clauses. Some of the most interesting grammatical environments with
respect to the Northern Subject Rule are therefore missing: relative clauses,
verbs after plural demonstrative and indefinite pronouns, to name but a few.
Many incidental material tokens that illustrated verbal -s in these
environments are either scattered across a large number of other headings
inSED-BM, or have up to now not been accessible at all. Thus, a wealth of
additional data is still enclosed in the original hand-written fieldworker
notebooks and has never become available in publications. For this reason,
the geographical picture that has so far been derivable from published SED
data alone is in some respects distorted.
To remedy this situation, it was necessary to return to a study of the
original fieldworker notebooks, held at Leeds University Library. This
investigation was conducted for 139 of the 311 SED locations, covering the
northern half of England roughly down to the Chester-Wash line and
somewhat beyond. All incidental material tokens illustrating subject-verb
agreement were excerpted from the notebooks, converted from the original
phonetic to an orthographic transcription, and later collated with the data
found under various headings in SED-BM. (The resulting token lists, which
form the basis for the analyses in the following sections, can be found in an
appendix to Pietsch 2005) They include approximately one thousand tokens
of plural verbal -s related to the Northern Subject Rule, and another
thousand tokens representing various other related phenomena.
In the following sections, I will first give a brief overview of the
geographical distribution of several variation phenomena related to verbal
concord in the north. I will then give a more detailed account of each of
them in turn, starting with some phenomena of minor importance,
proceeding through the special areas of was/were variation and of
existential there clauses, and finally dealing with the reflexes of the
Northern Subject Rule proper, found in the shape of plural verbal -s with
non-pronominal subjects.
3.2. An overview
For the situation in the traditional dialects of England, a good overview can
be gained from the data in the SED. It can be summed up as follows (cf.
Map 1). There is a central northern area covering the three northern
counties of Cumberland, Westmorland, Durham, and the southern half of
Northumberland.
2
They are characterized by a fairly consistent,
homogeneous subject-verb agreement system, which conserves many
features of northern Middle English. Its most important features are:
preservation of the thou versus you distinction, with thou always taking
verbal -s forms; generalized use of is as the singular present tense form of
136 Lukas Pietsch
be in all three persons; preservation of the was versus were number
distinction as in the standard; hence parallel paradigms of be in both tenses
(I/thou/he is, we/you/they are, I/thou/he was, we/you/they were);
application of the Northern Subject Rule to all verbs, including be. This
bundle of features can conveniently be called the Central Northern
3
agreement system.
In the northern half of Northumberland, we find more or less the same
features, but with less consistency, be paradigms more similar to the
standard, and a lack of thou. This area can conveniently be labelled the
Upper North.
Towards the northwest Midlands, more heterogeneous and variable
agreement systems can be found. Verbal -s conforming to the Northern
Subject Rule exists here too, but competes with inflectional forms in -n,
another conservative relic form from Middle English. Unlike in the Central
North there is a strong tendency to neutralize the was/were distinction in
favour of generalized were forms. The second singular form of be is
generally thou are or thou art. These three features together are
characteristics of an area between Derbyshire and southern Lancashire, also
reaching into the southwestern border areas of Yorkshire, and may be
labelled the Northwest Midlands agreement system. To the north of it,
one finds a gradual shift from the Northwest Midlands features towards the
Central Northern features. It can be visualized as a bundle of successive
isoglosses stretching through Yorkshire and northern Lancashire. These
isoglosses are grouped roughly around a line from Morecambe Bay to the
mouth of the Humber. This line, also called the Humber-Lune-Ribble line,
has long been known as one of the most important and oldest boundaries in
the traditional dialects of English (Trudgill 1990: 34; Ihalainen 1994: 219;
Samuels 1988). For the present purposes, the area spanned by the whole
isogloss bundle, i.e. most of Yorkshire and northern Lancashire, can
conveniently be called the Lower North.
Further southwest into the Midlands, more variation in the paradigm of
be is found, involving present indicative forms such as be, bist, and bin.
Also, the area preserving inflectional -n overlaps in the south with an area
that has generalized verbal -s independent of the Northern Subject Rule.
The East Midlands differ from the West Midlands mainly in that present-
tense paradigms of be are more similar to the standard, and second singular
thou is generally absent. The two areas are divided by a line that runs
between the counties of Derbyshire and Staffordshire in the west, and
Nottinghamshire and Leicestershire in the east. However, one feature that
both the west and the east Midlands share is the tendency to neutralize the
was versus were number distinction, in the one or other direction.
Map 1. Important subject-verb agreement isoglosses in the SED
L. Pietsch 2004
0 50 100km
Cheshire
Cumberland
Derbyshire
Durham
Lincolnshire
Lancashire
Leicestershire
Northumberland
Norfolk
Nottinghamsh.
Northsh.
Staffordsh.
Yorkshire
Westmorland
Shropshire
Cambridgesh.
Huntsh.
Herefordshire Worcestersh.
Warwicksh.
I am
I is
you are
thou is
he was
he were (occ.)
I keep
I keeps (occ.)
he were (freq.)
he were
I is
I am
thou is
you are burglars steals
burglars steals
burglars steal
burglars steal
thou art
you are/thou bist
they keep
they keep(en)
they keeps
they keepen
they keep(s)
1
2
3
4
5
6
7
8
9
1
2
3
4
5
6
1
2
3
4 5
6 1
2 3
4
1
2
3
4
5
6
7
8
9
10
11
12
13
14
1
2
3
4
5
6
7
8 9
10
11
12
13
14 15
16
17
18 19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
1 2
3
4
5
6
1
2
3
4 5
6
7
1
2
3
4
1
2
3
4
5
6
7
8
9 10
11
12
13
14
15
1
2
3
4
5
6
7
8
9
10
11
1
2
3
4
5
6
7
8
9
10
11
1
2 3 4
5
6
7
8
9
10
1
2
1
2 3
4
5
7
1
2
3
4
5
6
7
1
2
3
4
5 6
7
1
2
3
4
5
1
2
1
2
1
4
7
8
138 Lukas Pietsch
Plural verbal -s with lexical noun subjects, as licensed by the Northern
Subject Rule, is found everywhere north of a line running from Merseyside
to the Wash. In Map 1 the isogloss is labelled burglars steals (named after
a relevant item in the SED questionnaire). It cuts right through both the
West and East Midlands with their otherwise very different systems.
For Scotland and the north of Ireland, data of similar geographical quality
are not available. Data from the TRS/NITCS show a strong influence of the
Northern Subject Rule in the northern parts of Ireland. Data from FRED and
from other dialectological and sociolinguistic studies suggest that the dominant
situation in Scotland for the most part resembles that of the Upper North.
While modern Scottish varieties share with many English dialects a tendency
for neutralization of the was/were contrast independently of the Northern
Subject Rule (as in we was, you was), such tendencies were not discovered in
the northern Irish data. On the other hand, one pattern that is found only in
Ulster and in some parts of Scotland is the tendency of using neutralized
singular there were, and possibly there are, in existential clauses only.
3.3. Minor patterns
3.3.1. Verbal -n
Verbal forms in -n are relic forms of the Middle English plural -en
paradigms typical of the Midlands (Wright 1905: 296; LALME I: 467). In
theSED, such forms are still well attested in one compact relic area in the
northwest Midlands, covering southern Lancashire, Cheshire, Derbyshire,
Shropshire, and Staffordshire, and reaching also into the southwestern
corner of Yorkshire (Map 2; see also Pietsch 2005: ch. 4). They are also
documented for some more recent dialects of that area (cf. Shorrocks 1999
on Bolton, Lancashire). The area overlaps with that affected by the
Northern Subject Rule, which means that nonstandard forms in -n and in -s
compete with each other in some areas.
Map 2. Verbal -n forms in the SED
0 50 100km
L. Pietsch 2004
Present-Tensebin
Generalizedam/-m
Verbal -n
(Other Verbs)
Cheshire
Cumberland
Derbyshire
Durham
Lincolnshire
Lancashire
Leicestershire
Northumberland
Norfolk
Nottinghamsh.
Northsh.
Staffordsh.
Yorkshire
Westmorland
Shropshire
Cambridgesh.
Huntsh.
Warwicksh.
1
2
3
4
5
6
7
8
9
1
2
3
4
5
6
1
2
3
4 5
6 1
2 3
4
1
2
3
4
5
6
7
8
9
10
11
12
13
14
1
2
3
4
5
6
7
8 9
10
11
12 13 14
15
16
17
18
19
20
21
22 23
24 25
26
27
28
29
30
31
32
33
34
1
2
3
4
5
6
1
2
3
4 5
6
7
1
2
3
4
1
2
3
4
5
6
7
8
9 10
11
12
13
14
15
1
2
3
4
5
6
7
8
9
10
11
1
2
3
4
5
6
7
8
9
11
1
2 3
4
5
6
7
8
9
10
1
2
1
2 3
4
5
1
4
5
1
3
7
1
2
3
4
5
1
2
1
2
1
4
7
8
10
7
3
6
7
2
4
5
6
140 Lukas Pietsch
The examples in (8) show that verbal -n forms occur in all plural
environments, with the pronouns we, you, and they, as well as occasionally
with plural noun phrase subjects. However, the latter type appears to be
rare. The example quoted under (8f) is the only one found among 335
tokens of -n forms in the SED, all the others having pronoun subjects.
Shorrocks (1999: 114) states that -n occurs only in the first, second and
third persons plural, after pronominal subjects that do not trigger -s. This
type of distribution, where an -s versus -n alternation behaves according to
the same pattern as -s versus - in more purely northern varieties, may
have been typical of some parts of the border zone between the north and
the Midlands as early as in late Middle English (Laing 1978: 244; McIntosh
1988: 117).
(8) a. We never sayen that. [SED: Y29]
b. Wecallen it. [SED: Db1]
c. Youmowen. [SED: Db6]
d. Theytaken more pulling than a cart. [SED: Db1]
e. They think they knowen it. [SED: Ch1]
f. Burglarsthieven them. [SED: St2]
In the singular, verbal -n forms seem to occur only in the paradigm of be
(in the form I/he bin), and of have (I/he han). In addition, he done is
recorded, but only once (Sa5). The bin forms occur only in the
southwestern half of the area in question, namely in Shropshire and some
closely adjacent areas in the neighbouring counties. Areas yet further south,
especially Herefordshire and Worcestershire, tend to have generalized be
forms instead of bin (not charted in Map 2). Another phenomenon
geographically related to the verbal -n forms seems to be the existence of
amforms generalized to other persons and numbers than the first singular
in the paradigm of be. Especially youmand theymare often recorded, and
seem to be used as a contracted form both of you/they are and sometimes
you/they have. These forms compete with both the bin and the be
paradigms, in an area overlapping with the general -n area in Shropshire
and southern Staffordshire.
4
In the FRED data used for the present study, verbal -n was not
recorded, but this may well be due to the area not being well represented in
this part of the corpus.
3.3.2. I is
As stated above, throughout the Central North north of a line reaching
from Morecambe Bay to the mouth of the Humber, but excluding the Upper
North the present tense of be generally has is in the first and second
singular. In this area be has isomorphic paradigms in the present tense and
past tense, with singular -s forms contrasting with plural -r forms. This
means that the first singular present tense of be, just like its past tense in
most other areas, falls by necessity outside the scope of the Northern
Subject Rule.
While forms like I is (or Is) are abundantly attested in the SED, they are
almost completely absent in the more recent FRED data. Only two
speakers in the corpus show residual I is as a relic form. Interestingly, one
of these informants (Yks3:SL, a farmer from the Teesside area, born
c.1910, and recorded during the mid-1980s) used Is only in direct speech
quotations in the context of narratives set in the old days, using standard
Imelsewhere. For this speaker Is apparently functions as a signal of
vernacular speech employed to give a certain stylistic touch to his narrative.
The other speaker (Wes4:HL, a retired forest worker from the Lake
District, Cumbria) used twice Is and twice Imin his recording.
(9) a. And t porter says, well, I s about sick o this. [FRED:
Yks3:SL]
b. And mi mother says, I s not goin to have roomto work in
here. [FRED: Yks3:SL]
c. And now, so I got this one done and I thought, This is mi
last morning, I s not going to bother. [FRED: Yks3:SL]
d. I used to dread to have to go down to give my boss a message
when he was among all t boozey folks, cause they all knew
me: Fetch hima drink in. Fetch hima drink! I scoming
back. Ill just deliver this message! [FRED: Yks3:SL]
e. I s going on for eighty. [FRED: Wes4:HL]
f. He says, I s goin to have a real good go at it [FRED:
Wes4:HL]
142 Lukas Pietsch
3.3.3. Verbal -s with thou
The old second singular pronoun thou (in a variety of forms: thou, thee,
[], [t] etc.) is found preserved in the SED in a large coherent area
covering all of the north, with the exception of the East Midland counties.
Only in northern Northumberland is it found less frequently, and ye/you is
generally used instead (cf. Trudgill 1990: 86, Beal 2004). Where they exist,
thethou forms almost invariably command agreement with -sforms (rarely
also -st forms) of the verb, irrespective of the Position-of-Subject
Constraint. Hence they do not pattern together with you under the Northern
Subject Rule but rather align themselves with third person singular
he/she/it. Variation in the verbal forms occurs only in the paradigms of be.
As for the past tense, the variation between thou was and thou were will be
dealt with together with that in the other persons and numbers in the next
section below. As for the present tense, three fairly distinct regional types
can be distinguished (Map 3). In the Central North one regularly finds thou
is, alongside I is and he/she/it is. Further southwest, in southern Lancashire,
southwestern Yorkshire, Derbyshire, Cheshire, and northern Staffordshire,
the forms are thou are or thou art, in inversion sometimes contracted to
art. Going yet further south into the West Midlands, one finds forms such
as thou bis(t), as well as other forms with the verbal stem be in which the
plural forms have been generalized to the second singular (thou be, thou
bin).
Like the I is forms, the thou forms are clearly obsolescent in the more
recent data from FRED. Only four tokens of subject thou (thee/tha) with
concord verbs are found in the data. One example from the Teesside data,
and one from the Scottish Borders, again occur in narrated direct speech
situated in the old days (10). These two tokens have the old -s form of the
verb and happen to be both in interrogatives. The other two tokens are in
stereotypical collocations of the you see / you know type, and have been
recorded in northeast England and southwest Scotland respectively (11). In
these two tokens the pronoun thou has taken over the affixless verb form
used also with you.
Map 3. Verbal agreement forms with thou in the SED
L. Pietsch 2004
0 50 100km
thou is
thou bist
thou are/art
thou be/been
thou -s, other verbs
Cheshire
Cumberland
Derbyshire
Durham
Lincolnshire
Lancashire
Leicestershire
Northumberland
Norfolk
Nottinghamsh.
Northsh.
Staffordsh.
Yorkshire
Westmorland
Shropshire
Cambridgesh.
Huntsh.
Warwicksh.
3
4
5
6
8
9
1
2
3
4
5
6
1
2
3
5
6 1
2 3
4
1
2
3
4
5
6
7
8
9
10
11
12
13
1
2
3
4
5
6
7
8
9
10
11
12 13
14
15
16
17
18 19
20
21
22 23
24
25
26
27
29
30
31
32
33
34
1 2
3
4
5
1
2
3
4 5
6
7
1
3
4
1 3
4
5
6
7
8
9 10
11
12
13
14
15
2
3
5
7
8
9
10
11
1
2
3
5
7
8
10
11
1
2 3
4
5
6
7
8
9
10
1
2
1
2
3
4
7
2
3
4
5
6
1
2
3
4
5 6
7
1
2
3
4
5
1
2
1
2
1
4
7
8
1
2
7
4
28
14 14
2
2
6
4
6
9
1
7
1
4
6
144 Lukas Pietsch
(10) a. And uh, an old woman over North Gate said, where 's thou
come fromthen? Says, fromMiddlesbrough. She says, what a
bloody muck-hole that is. [FRED: Yks1:WF]
b. And they used to go down and cook their meals, down in the,
on the fire, hm-hm, and never anybody said, what doest thou,
they never were put out or anything then. [FRED: Pee2:MT]
(11) a. Tha see it? See this here? [FRED: Dur1:ML]
b. Oh, they, thee ken, there was no, ye see, there were, there was
a lot come in at what we call Huggins up there. [FRED:
Dfs1:WH]
3.3.4. Verbal -s with I/we/you/they
Apart from the old use of I is, described above, and the widespread
was/were variation which will be the subject of section 3.4 below, non-
standard -s with the pronouns I, you, we and they occurs occasionally in
many but not all northern varieties. We can distinguish several types of
usage.
First, there are those instances which are covered by the Position-of-
Subject Constraint, most typically in the coordinated structures of the they
sing and dances type. This usage is characteristic of the more conservative
varieties. In the relatively recent data of FRED it is not attested, but it
appears with some regularity in data from the SED (12-13), even as far
south as Nottinghamshire and Lincolnshire, and it is also found in a number
of Ulster speakers in the NITCS (1417).
5
(12) a. They gang and never speaks. [SED: Du4]
b. They peel emand boils em. [SED: La6]
c. They break into houses and steals. [SED: La7]
d. They go in and cuts emdown. [SED: Y17]
e. They throw that down and picks another sheaf up. [SED: Y17]
f. They feel over emand weighs em, does butchers. [SED: Y22]
g. They rope emand then pulls emin. [SED: Y29]
h. They run at one another and brods one another. [SED: Y31]
i. They cut emdown low and lays em. [SED: Nt2]
j. They lead bulls wi staffs, puts a band on their horns, puts it
undernien their hind legs. [SED: L4]
(13) a. You stack it up and carts it up to where you want it. [SED:
Y17]
b. You put middling of water into it and pulls it down. [SED:
Y17]
c. You go fair down t middle and pulls one each way. [SED:
Y22]
d. You take that there up and shakes it out. [SED: Y14]
(14) The women goes out and rickles, dear. I rickle my own turf. After I
do my work in the daytime, I go out and rickles my turf. When I come
home, I go away and rickles my turf. [NITCS: L17.3]
(15) a. You have, you shovel off the, makes the top of it smooth, you
know. [NITCS: L17.3]
b. And then you just cut down, and makes the shape of the turf
[NITCS: L22.1]
c. You pull a wee drawer there, hey, and puts the maybe you
have one of them? [NITCS: L2.2]
d. Like, you, if you go out to speak to anyone, know, ch-, has to
challenge themabout, if theyre doing any harm, or d-, your
property or anything, they just answer you back. [NITCS:
L4.2]
(16) a. But we lived, we lived about, about four mile out of the town
here, and was taught at Ballyreagh school. [NITCS: L25.3]
b. We were all good neighbours, and is yet, I hope. [NITCS:
L17.3]
c. Mm, we just t-, plays swinging on the ropes, and in the bars,
you know, climbing bars, and swinging round on them.
[NITCS: L17.1]
(17) a. And they season and gets lighter there then. [NITCS: L22.2]
b. So they closed it and sends their milk to Manorhamilton.
[NITCS: L31.2]
c. So they go on that way and takes about ten minutes for that
[NITCS: L8.1]
Note, incidentally, how in example (15b) the presence of an intervening
adverb in the first conjoined phrase (you just go) fails to trigger a Position-
146 Lukas Pietsch
of-Subject effect according to the Northern Subject Rule, whereas the
conjoining of verb phrases does trigger -s in the second verb.
A second, apparently quite unrelated pattern involving non-standard
verbal -s with pronoun subjects is found in the so-called historic present. A
tendency to mark narrative clauses in the present tense with generalized
verbal -s forms is commonly reported for many modern varieties of
English. Such verbal -s forms occur very commonly also with pronoun
subjects where they are not licensed by the Northern Subject Rule.
The prototypical usage condition for such forms is in clauses
introducing direct speech within the narrative. In the FRED data, 141
tokens from 21 informants in all parts of the survey area were recorded in
this textual function; with one exception (I shouts) all using the verb says.
In the NITCS data there were 20 tokens from 8 informants, all of them
usingsays. One clear piece of evidence for the restricted, formulaic nature
of the I says idiom is that for some of the older speakers it is associated
with a special rule of subject-verb inversion (says I; also: says he etc.),
which is likewise restricted lexically and pragmatically to exactly these
formulae of introducing direct speech. 16 of the tokens mentioned above
were of this type, and they were found in a few older speakers in Scotland
and in Ulster.
Only 18 tokens were found of other verbal -s forms used in other textual
functions within a narrative in the FRED data (and none at all in the
NITCS), of which 13 came from a single informant and were produced
during two longer narrative sequences. Apart from this, there is little
evidence in the data of the present study that speakers have a productive
stylistic rule of using verbal -s as a marker of the historic present as such,
over and above the prototypical, idiomatic use of says (or its semantic
equivalents).
6
However, such wider usage of the narrative present has
repeatedly been reported elsewhere, for instance by Harris (1993: 154
156), Robinson (1997: 127) and Henry (1995: 18) for Irish English. Henry
in a formal, generativist discussion of verbal concord uses the existence
of this usage as an argument for a formal analysis which involves a T
(i.e. tense) node playing a central role in a feature checking mechanism that
is responsible for licensing verbal -s. (1995: 27). The reference to the T
node comes close to suggesting that the simple present and the narrative
present must actually be two distinct morphological tenses, as that would
presumably be the case by definition within a formal theory such as
Henrys if this node contained different features in each case. (For further
discussion of Henrys analysis, see section 5.1 below).
The attestation of verbal -s with I/you/we/they in environments other
than those described so far, i.e. where it violates the Northern Subject Rule,
is quite marginal in both the NITCS and FRED. This is somewhat
astonishing, since the existence of such forms is fairly well documented in
some of the older dialects in the SED. There it is attested occasionally in a
large area of Northern England, overlapping with the central northern I is
area in Yorkshire and Lancashire but also extending further south, and
excluding most of Cumberland, Westmorland and Durham in the north
(Map 4). It should be noted that the phenomenon of occasional verbal -s
usage in these areas must be distinguished geographically from the much
more general usage of neutralized verbal -s in a different area towards the
southwest (beginning in Map 4 in the counties of Herefordshire and
Worcestershire; cf. Klemola 1996: 5052). In this southwestern area, verbal
-s is traditionally found quite regularly in all environments, irrespective of
type and position of subject. Both areas are separated from each other by a
broad belt in which non-standard verbal -s in the pronominal environments
seems to have been largely absent in the traditional dialects, just as it was
also absent further north. These observations confirm those by Wright
(1905: 296), who likewise stated that exceptions to the Type-of-Subject
Constraint occurred occasionally in parts of Yks. Lanc. and Lin. but not
further north.
As for the function of this Yorkshire/Lancashire type of occasional
verbal -s usage, nothing can be stated on the basis of the SED material,
since the tokens are mostly documented without context. However, it seems
likely that the phenomenon reflected in the SED attestations can be
identified with that described for more recent Lancashire dialects, where
verbal -s in such cases is reported to serve as a marker of habitual semantics
(Shorrocks 1999: 112).
It should be noted in passing that many of the SED tokens that can be
counted as potential examples of habitual -s occur with one of several
typical adverbs between the pronoun and the verb: never, always, often, etc.
According to the ideal northern system, the presence of such an adverb
would also in itself count as a condition for the Position-of-Subject
Constraint to apply. Hence, it is sometimes difficult to decide whether
verbal -s in this cases is triggered by the temporal semantics, or by the
Northern Subject Rule.
148 Lukas Pietsch
Map 4. Verbal -s with I/you/we/they in the SED
L. Pietsch 2004
0 50 100km
Verbal -s recorded with:
I
we
you
they
(largesymbols: total
of 4 tokens or more)
1
2
3
4
5
6
7
8
9
1
2
3
4
5
6
1
2
3
4
5
6 1
2 3
4
1
2
3
4
5
6
7
8
99
10
11
12
13
14
1
2
3
4
5
6
7
8
9
10
11
12 13
14
15
16
17
18 19
20
21
22 23
24
25
26
27
28
29
31
30
32
33
34
1 2
3
4
5
6
1
2
3
4 5
6
7
1
2 3
4
1
2
3
4
5
6
7
8
9 10
11
12
13
14
15
1
2
3
4
5
6
7
8
9
10
11
1
2
3
4
5
6
7
8
9
10
11
1
2 3
4
5
6
7
8
9
10
1
2
1
2
3
4
5
7
1
2
3
4
5 7
6
1
2
3
4
5 6
7
1
2
3
4
5
1
2
1
2
1
4
7
8
However, to the degree that the same dialects also exhibit occasional
verbal -s in cases of no intervening adverb, with comparable frequency, the
pattern is better described as independent of the Northern Subject Rule, at
least in the case of Yorkshire and Lancashire. Matters are somewhat
different in the northern Irish data of the NITCS: here, verbal -s with
pronominal subjects is quite rare throughout; but it is, relatively speaking, a
good deal less rare in cases not adjacent to the pronoun than in others. In
this dialect, then, it is indeed the Northern Subject Rule that is responsible
for the marginal option of verbal -s in these cases.
3.4. Was/were neutralization
Variation between was and were in the northern dialects is a highly
complex field. This is due to the fact that it tends to follow only partly the
pattern defined by the Northern Subject Rule, with was and were behaving
like the -s and zero forms of other verbs. This pattern is often overlaid with
other, complementary or competing, rules of variation specific to was and
were alone. The high potential for irregularity and variation that
characterizes this verb can be linked historically to two factors.
First, the was/were paradigm patterned differently from the present-
tense verbs from the outset, and of course it still does so in Present-Day
Standard English. The same form, was, is used for both the first and third
person singular, whereas in present-tense verbs the -s form is unique to the
third person. This is the last remnant of the Old English concord paradigms
of the past tense, in which first and third singular regularly patterned
together as against the rest. The breakdown of this system can best be seen
when the old second person singular is taken into account too. Its original
Old English form (u wre) resembled the plural form (wron) more than
the other singular forms (ws), and it predictably fell together with the
plural form in many dialects of Middle English (thou were). Elsewhere,
however, analogy with other verb classes gave rise to a variety of other
forms such as thou wert, thou was, thou wast, before the second person
singular finally became obsolete in Modern English. It is not surprising that
tendencies of analogical extension or levelling of forms have continued to
operate on the was/were paradigms of different dialects, leading to a
number of different outcomes.
Secondly, it must be noted that both was/were and is/are/amdid not
originally fall under the scope of the Northern Subject Rule at all. As will
150 Lukas Pietsch
be described in more detail in section 4, the diachronic development in
these verbs went exactly in the opposite direction than with all others. In
the lexical verbs, the (now non-standard) -s forms represent an older,
conservative form both in the singular and in the plural, while the (now
standard) zero forms in the plural are an innovation in Middle English. In
contrast to this, the forms is and was are truly and exclusively singular in
origin, and were only extended to plural use under the Northern Subject
Rule by way of analogy with the other -s forms later. Apparently, the use of
was in plural environments as licensed by the Northern Subject Rule never
became quite as regular as the corresponding use of other verbal -s in the
same environments, even in the purest older dialects of the northern type
(cf. Montgomery 1994 on Older Scots). But even though the forms of be
were relatively slow in picking up the Northern Subject Rule pattern, they
have also been slower than other verbs in giving it up again and replacing
the northern with the standard pattern, under conditions of dialect levelling
in more recent times. This results in a situation found in some modern
varieties where be apparently tends to be the only verb to conserve reflexes
of the northern pattern (see e.g. Tagliamonte 1999 on the English of York).
In the traditional dialects of northern England as reflected in the SED,
the following situation obtains. In the Central North, the use of indicative
was and were is, as a rule, parallel to that of is and are. The singular of all
three persons (including thou) is invariably was, while the plural (including
you) mostly has were but allows for was in accordance with the Northern
Subject Rule. This is mainly true for the four northern counties of
Cumberland, Northumberland, Westmorland and Durham. The SED data
suggest a near-categorical validity of the number contrast with adjacent
pronoun subjects in this area.
All the rest of the survey area shows a tendency of was/were levelling,
either generalizing was to the plural, or were to the singular. Four regional
clusters of locations can be distinguished in this respect. First, in a compact
area centring around southern Lancashire, southwestern Yorkshire (i.e.
southwest of a line from Morecambe Bay to the Humber) and Derbyshire,
there is a strong preference for generalized singular were forms (cf. also
Shorrocks 1999: 168). Second, in a broad transitional belt from this area
into the Central North, covering northern Lancashire and the northeastern
half of Yorkshire, singular were forms are also occasionally recorded but
less frequent. Third, in a smaller area in the east Midlands, especially in
northern Lincolnshire, the preference seems to be for neutralization in was
rather than were. In the Northwest Midlands there are also occasional
attestations of plural weren. Finally, most other locations in the Midlands,
further south, have highly variable or hybrid systems where both singular
were and plural was may co-occur. A common tendency in many of these
dialects seems to be that were is preferred in negated environments (cf.
Anderwald 2002, Britain 2002). This effect is discernible in all parts of the
SED data except the Central North and those parts of the NW Midlands and
Lower North where were levelling is predominant in all environments. To
the north of the central generalized were area, this negation constraint tends
to affect only the singular forms (he was vs. he werent), while were
remains near-categorical in the plural, except where the Northern Subject
Rule licenses was. The existence of systems of this kind is consistent with
findings by Tagliamonte (1999), who reports the combined effects of
singular were generalization, the negation constraint, and the Northern
Subject Rule (though with no plural verbal -s usage preserved in the lexical
verbs) in the local dialect of York city. In contrast to the SED findings,
Beal (1993: 194) attests plural was levelling also for Tyneside and
Northumberland.
In varieties where was/were levelling and the Northern Subject Rule
compete with each other, it is also often found that the third person plural
differs from the first and second person plural. Levelling typically results in
we was and you was, whereas the third person plural tends to conserve
near-categorical they were in conformity with the Northern Subject Rule
(Tagliamonte 1999, Chambers 2004).
152 Lukas Pietsch
Map 5. Was/were neutralization in the SED
L. Pietsch 2004
0 50 100km
I
we
you
they
1
2
3
4
5
6
7
8
9
1
2
3
4
5
6
1
2
3
4
5
6 1
2 3
4
1
2
3
4
5
6
7
8
99
10
11
12
13
14
1
2
3
4
5
6
7
8
9
10
11
12 13
14
15
16
17
18 19
20
21
22 23
24
25
26
27
28
29
31
30
32
33
34
1 2
3
4
5
6
1
2
3
4 5
6
7
1
2 3
4
1
2
3
4
5
6
7
8
9 10
11
12
13
14
15
1
2
3
4
5
6
7
8
9
10
11
1
2
3
4
5
6
7
8
9
10
11
1
2 3
4
5
6
7
8
9
10
1
2
1
2
3
4
5
7
1
2
3
4
5 7
6
1
2
3
4
5 6
7
1
2
3
4
5
1
2
1
2
1
4
7
8
Map 6. Thewas/werent negation constraint in the SED
L. Pietsch 2004
0 50 100km
I
we
you
they
1
2
3
4
5
6
7
8
9
1
2
3
4
5
6
1
2
3
4
5
6 1
2 3
4
1
2
3
4
5
6
7
8
99
10
11
12
13
14
1
2
3
4
5
6
7
8
9
10
11
12 13
14
15
16
17
18 19
20
21
22 23
24
25
26
27
28
29
31
30
32
33
34
1 2
3
4
5
6
1
2
3
4 5
6
7
1
2 3
4
1
2
3
4
5
6
7
8
9 10
11
12
13
14
15
1
2
3
4
5
6
7
8
9
10
11
1
2
3
4
5
6
7
8
9
10
11
1
2 3
4
5
6
7
8
9
10
1
2
1
2
3
4
5
7
1
2
3
4
5 7
6
1
2
3
4
5 6
7
1
2
3
4
5
1
2
1
2
1
4
7
8
154 Lukas Pietsch
Regional diversification in this respect, similar to what is evidenced in
theSED, can also be seen in the FRED data. Table 1 displays the rates of
non-standardwas in the three pronominal standard were environments (we,
you, they) across eight regional clusters of texts. For comparison, the table
also shows the rates of standard I was as opposed to non-standard I were.
The regional clusters are defined as follows. Ulster comprises the six TRS
recordings in the northern part of the Republic of Ireland, all in close
neighbourhood to the six Northern Ireland counties covered by the NITCS
(the label is used somewhat loosely, since not all of the locations are
actually in the province of Ulster four are in County Donegal, one in
Leitrim and one in Louth). Northern Scotland comprises one recording
from Banffshire and six from the Scottish Highlands (Inverness, Eastern
Ross and Sutherland). Mid Scotland refers to a cluster of interviews from
the east coast area, between Kincardineshire and West Lothian. Southern
Scotland comprises recordings from the Scottish Borders as well as one
from Dumfriesshire. A group of Northumberland recordings represent the
traditional English Upper North. Recordings from County Durham and
northeast Yorkshire are grouped under Teesside and represent part of the
traditional Central Northern area. Another cluster, labelled as Cumbria,
represents data from near Ambleside in the Lake District and also forms
part of the traditional Central North. Finally, grouped under Lower North
are two recordings from southern Lancashire and southwest Yorkshire. The
data in Table 1 suggest that speakers in Ulster, northern Scotland, as well as
in the Upper and Central North of England, tend to have categorical
was/were concord with pronoun subjects, identical to the standard. This is
Was with pronoun subjects in FRED
They We You I
Ulster 0/22 0% 0/5 0% 0/15 0% 11/11 100%
N Sco 1/113 1% 0/23 0% 0/8 0% 66/66 100%
M Sco 1/188 1% 8/68 12% 60/104 58% 115/116 99%
S Sco 4/114 4% 4/15 27% 2/6 33% 48/48 100%
Northumb. 0/39 0% 1/15 7% 0/14 0% 47/49 96%
Cumbria 0/66 0% 1/27 4% 0/22 0% 76/77 99%
Teesside 6/264 2% 3/128 2% 3/171 2% 331/344 96%
Lower N 9/104 10% 15/61 25% 3/7 43% 99/115 86%
Total 21/910 2% 32/342 9% 68/347 20% 793/826 96%

Table 1.
also confirmed in the data from the NITCS, where non-standard was with
adjacent pronoun subjects is extremely rare. In Middle and Southern
Scotland, there is variable non-standard we was and you was, but hardly
any they was. It must be stressed that these differences are not due simply
to an uneven distribution of standard and dialectal speakers across the
corpus. All clusters, including those that show no was/were variation at all,
do contain informants who otherwise use strongly non-standard features,
including heavy use of non-standard verbal -s according to the Northern
Subject Rule. These data therefore fit in with the earlier findings in the SED
and the NITCS, which indicated that central Northern England and Ulster
do not share in the pronominal was/were variation in the same way as so
many other English vernaculars, and that this type of was/were variation is
historically unrelated to the Northern Subject Rule.
The Lower North speakers, as can be expected on the basis of the earlier
SED findings, have some use of non-standard were in standard was
environments, represented in the table by the figures in the I column (only
84% of standard was; the situation is similar in third person singular
environments.) However, at the same time they also have variable plural
was in standard were environments. These speakers even have they was,
but they too use it less frequently than you was or we was. In all areas that
allow plural was at all, you was seems to be more frequent than we was,
and both are more frequent than they was. This fits in with observations
made in other studies (e.g. Smith and Tagliamonte 1998, see also Chambers
2004). It is only in the speech of the Teesside informants that plural was is
found with no marked differences between the three environments, but it is
only of marginal frequency in all three. In a common Varbrul model for all
areas that have plural was, factor weights of .26, .61, and .88 were
calculated for they, we and you respectively.
Whereas there is no discernible Position-of-Subject effect in present-
tense verbs with pronoun subjects in the FRED data, some effect of this
type may be in evidence with was/were, although owing to small absolute
token counts the evidence is hardly conclusive. While the overall rate of
non-standard was with adjacent we/you/they is 115/1579 (7%), with non-
adjacent pronoun subjects it is 6/20 (30%). Varbrul selects the factor group
as significant at a 0.009 level, with factor weights of 0.49 versus 0.84 for
adjacent and non-adjacent subjects respectively. The negation constraint
had no discernible effect on plural was in the data.
156 Lukas Pietsch
3.5. Existential there clauses
In all data discussed in the previous sections, clauses with existential there
have been excluded because they require special consideration. Concord
variation in existentials and concord variation in canonical clauses have
followed quite different paths of historical development, they have different
sociolinguistic status in the dialects in question (cf. also Wilson and Henry
1999: 12), and their distribution is governed by different sets of linguistic
constraints.
As is well-known, existential clauses in formal Standard English require
the verb to agree in number not with the syntactic dummy subject there, but
with the so-called notional subject, the NP that follows it. This constitutes a
systemic anomaly, as the morphosyntactic properties of subjects are
divided between two constituents. The dummy subject there acts as the
subject of the clause as far as word order is concerned, but the following
NP acts as the subject in so far as it controls agreement. Many varieties of
English share the tendency to level out this irregularity, by allowing
invariant singular verb forms after there. Non-agreement in clauses
beginning with there or similar adverbs have been a structural option in
most forms of English even since the Old English period (Visser 1963: 62).
In the modern dialects, the strong tendency for non-agreement means that
the residual subject status of the following noun phrase is effectively lost.
Historically this can be seen as part of a long-term trend of
grammaticalization of the there construction, in which there has gradually
changed its status from being originally a deictic adverb to being a subject
(cf. Breivik and Swan 2000; pace Van Gelderen 1997: 88109).
It is not surprising that evidence for non-agreement in existentials was
also found everywhere in the data of the present study. However, there is a
second, competing non-standard pattern in effect in some parts of Northern
Ireland and Scotland, which works in just the opposite direction. In this
pattern the distinction between singular and plural existentials is also
neutralized, but the neutralization is in favour not of the normal singular
verbal forms is/was, but of the verb forms that are otherwise used in the
plural: are/were. These forms appear to be favoured in negated clauses, but
are by no means restricted to them. The present-tense form of this pattern is
often a contracted form in which the verb has lost all or most of its
segmental phonological substance: there < therere < there are, but full
forms of are are also attested (18c). In the following discussion I will refer
to these forms collectively as -r forms, as opposed to the -s forms is, s and
was.
(18) a. Och, aye, there no blacksmithing work now, except odd wee
bits. [NITCS: L7.3]
b. Therere no school. [NITCS: L7.3]
c. The young would rather go away till Dungiven, or, or
Strabane, or some place where there are a big, eh, dance-hall.
[NITCS: L10.2]
In the NITCS data, -r forms are found strongly concentrated in the speech
of informants from the north and northwest of Ulster, whereas only a bit
further towards the southeast (county Down and the southern half of county
Antrim) such forms were not found or only very marginal. This finding is
interesting from a dialect-geographical perspective, as this apparent
division cuts right across what has long been recognized as an important
dialect boundary within Ulster, that between Core Ulster Scots and Mid
Ulster English (Gregg 1972; see Pietsch 2005: ch. 5.4.1 for details). In the
FRED data, singular -r forms in the present tense are found only in the
speech of six informants, two of them located in Ulster and four in southern
Scotland. Some informants in the northwestern part of the Core Ulster
Scots area went on record using singular -r in about two thirds of all cases.
However, the existence of the competing patterns of neutralization in -s and
neutralization in -r means that for some speakers all eight of the options
shown schematically in (19) and (20) seem to be possible.
(19) a. Theres a house now
b. Theres houses now
c. There was a house then
d. There was houses then
(20) a. There a house now
b. There houses now
c. There were a house then
d. There were houses then
The following authentic examples (21) give a first impression of the
amount of variability observed in the corpus. They were all produced by
one Ulster speaker within a single short stretch of discourse, in this order:
158 Lukas Pietsch
(21) a. There was fairs.
b. Well, there was a fair in Kilrea.
c. There were no pastime for you.
d. There were no fences.
e. There were better sheepdogs than there is now. [NITCS: L6.3]
Across the whole survey area, singular were in past tense existentials is
more widespread than singular are in the present. All of the six FRED
informants who used singular bare there in the present tense also had
singular there were in the past. Additionally, singular there were was also
found in the speech of twenty other informants. Many of these informants
are also found in Ulster and southern and middle Scotland. But, not
surprisingly, singular there were is also found in the Lower North of
England, that is, with those speakers who have singular were levelling in
non-existential clauses as well. In the Central and Upper North of England
and in northern Scotland, singular were in either type of environment is
only marginally present or, in many speakers, not attested at all. The two
phenomena can thus best be described as distinct in principle, and probably
in origin: speakers either have singular -r forms in existentials (the Ulster
Scots type: there are a house, there were a house); or they have singular
were in all clause types (the Yorkshire type: he were going; there were a
house); but no speakers have both. Structures of the type there were a
house, representing a point of incidental overlap between the two patterns,
seem to have spread across an area much larger than either of the focal
areas of these two patterns, as a marginal option at least. This distribution
can be seen in Table 2, cross-tabulating figures for the three usage types
across the eight regional clusters within the FRED data.
Unlike the FRED data, the data of the NITCS allowed also for a study of
the effects of sociolinguistic variables, such as age and sex (Pietsch 2005:
ch. 5.4.2). Looking at the three age groups sampled in the NITCS, it was
found, first, that the middle age group tended to observe standard-
conforming number concord in existentials more often than either the
children or the older informants in the corpus. The youngest age group
showed a strong tendency to use generalized -s forms, approaching near-
categoricity in some groups. In contrast to this, the use of generalized -r
forms was strongly associated with the oldest age group. Moreover, across
all age groups, male informants consistently used -r more often than
women and girls.
Table 2. Nonstandard -r forms in FRED
Pres. Sg. Ex. Past Sg. Ex. Past 3Sg. / 1Sg. Total
therere +Sg. there were +Sg I/he/she/it were
Ulster 9/43 21% 13/48 27% 2/110 02% 24/201 12%
S Scot. 9/42 21% 26/92 28% 9/678 01% 44/812 05%
M Scot. 1/45 02% 13/175 07% 9/1257 01% 23/1477 02%
N Scot. 0/21 00% 1/68 01% 4/455 01% 5/544 01%
Cumbria 0/22 00% 0/80 00% 4/428 01% 4/530 01%
Northumb. 0/11 00% 1/58 02% 6/272 02% 7/341 02%
Teesside 0/82 00% 8/362 02% 35/1850 02% 43/2294 02%
Lower N 0/22 00% 22/73 30% 72/661 11% 94/756 12%
Total 19/288 07% 84/956 09% 141/5711 02% 244/6955 04%

All of this clearly indicates that generalized existential -r has the status of a
socially marked, conservative local variant in Northern Ireland and that
there is a change in progress replacing this pattern with that of generalized
-s. The gender differentiation can be seen as an expectable side effect of
such a development, as it confirms the long-established rule in
sociolinguistics (cf. Trudgill 2000: 73) that women tend to follow
prestigious standard models of speech more than men do, and that men in
rural societies tend to be more conservative in preserving local non-
standard forms. In this case, women were found both to adapt more to the
standard of formal English than men do, and simultaneously to be leading
the trend, apparent in younger speakers in general, to move away from the
more specifically local non-standard pattern of generalized -r forms,
towards the supraregional colloquial norm of generalized -s forms. Table 3
displays the distribution of -s and -r tokens across the different social
groups in two geographical sub-regions in the Ulster corpus.
In addition to this, there seem to be indications in the NITCS data that
Protestant speakers in the north may have a stronger preference for the
conservative -r forms than Catholic speakers. Although the findings are
somewhat inconclusive in statistical terms (Pietsch 2005: ch. 5.4.2), such
an effect would fit in plausibly with a hypothesis that linguistic features of
traditional Ulster Scots serve as a marker of Protestant identity in Northern
Ireland, whereas Catholics show more of an orientation towards varieties of
English used in other parts of Ireland (Harris 1991: 46, cf. also McCafferty
1998a, 1998b, 1999).
160 Lukas Pietsch
Table 3. Existential -r in the NITCS, by area, sex and age group
Area Sex Age Group Singular Plural Total
North m 6575 82/188 44% 78/126 62% 160/314 51%
3545 38/158 24% 51/93 55% 89/251 36%
912 13/38 34% 5/19 26% 18/57 32%
f 6575 8/30 27% 5/15 33% 13/45 29%
3545 1/63 02% 15/28 54% 16/91 18%
912 2/56 04% 5/46 11% 7/102 07%
South m 6575 10/150 07% 15/92 16% 25/242 10%
3545 4/129 03% 7/67 10% 11/196 06%
912 0/75 00% 1/40 03% 1/115 01%
f 6575 3/137 02% 2/69 03% 5/205 02%
3545 0/71 00% 14/49 29% 14/120 12%
912 0/73 00% 3/44 07% 3/117 03%

Verbal concord in existentials is also influenced by linguistic
environment factors. One of them is negation. With singular notional
subjects, the use of -r is three times more frequent in negated than in non-
negated clauses (27 per cent as opposed to 9 per cent); in plural clauses the
difference is slightly less marked but the direction is the same (45 per cent
as opposed to 28 per cent). This effect reflects the pattern already discussed
in the previous section, of generalized werent being preferred over wasnt
in many dialects (cf. Anderwald 2002: 180182). In Ulster, such an effect
was found only in existentials, not in canonical clauses, and it seems to
apply in parallel fashion both to was/were and is/are. The data suggest that
for many speakers the negation constraint is a more powerful factor in
determining the choice of form than the grammatical number of the
notional subject. A Varbrul model shows a slightly larger span of factor
weights, of 0.45 versus 0.71 for non-negated and negated clauses
respectively, as opposed to a span of only 0.42 versus 0.64 for singular
versus plural subjects. Another factor that seems to constrain the choice of
verb forms in existentials is the syntactic environment in which it occurs.
Table 4 shows token counts tabulated across three types of structures:
Type A, the bulk of the canonical existentials with the verb immediately
preceding the notional subject; Type B, clauses where the verb and the
notional subject are separated by intervening material such as adverbials;
and Type C, all other clauses in which the element order between the verb,
the notional subject and the existential marker there differs in some way.
This group most notably comprises tag clauses, questions, and relatives. It
can be seen that both of the special environment types have an interesting
effect. In the cases with intervening adverbials, the proportion of -r and -s
is almost exactly the same in the singular and in the plural. In these cases,
the choice of verb form thus appears to be governed entirely by factors
other than grammatical number. In other words, agreement between the
verb and the notional subject which, as seen earlier, is disregarded by
many speakers much of the time anyway is practically non-existent
whenever other words intervene between the two. However, in the Type C
cases, those with inverted or other non-canonical word orders, the picture is
different: whereas for the plural the rate of -r usage is just about average,
surprisingly few instances of non-standard -r were found in the singular. A
definitive explanation for this effect cannot be given at the present moment,
but it may be related to the fact that in these clause types (tags, questions
and so on) the verb is usually stressed. Apparently, the use of singular were
andare in Ulster English tends to be restricted to unstressed environments.
Table 4. Existential -r in the NITCS, by number and clause type
Clause Structure Singular Plural Total
Type A: Adjacent NP 152/1134 13% 196/612 32% 357/1755 20%
Type B: Adv.
Intervening
14/76 18% 10/51 20% 24/127 19%
Type C: Others 1/53 02% 14/44 30% 15/97 15%
Total 167/1263 13% 229/716 32% 396/1979 20%
3.6. Verbal -s with plural NPs
Having so far discussed the distribution of phenomena that have no or only
a marginal relationship to the Northern Subject Rule, we can now proceed
to a more detailed discussion of the pattern that stands at its core and is
most characteristic of the traditional dialects of the northern type: non-
standard verbal -s with third person plural subjects other than the pronoun
they.
The geographical distribution of this pattern in the traditional dialects of
England can be traced well in the SED data. The SED questionnaire
contained two items that were designed to elicit structures of the relevant
type (III.10.7 bulls bellow, and VIII.7.5 burglars steal them), and atlas
162 Lukas Pietsch
charts based on these data have been printed in Viereck (1991/1997: II,
M27 and I, M40); Klemola (2000: 334), and Viereck, Viereck and Ramisch
(2002: 84); see also the isogloss labelled burglars steals in Map 1 above.
However, as was pointed out earlier, the published material these charts are
based on is in some sense deficient, as it excludes a great number of
incidental material tokens that did not match the elicited environment in a
narrow sense, but still provide important evidence for the effects of the
Northern Subject Rule.
When this additional material, recovered from the original fieldworker
notebooks, is taken into account, two important observations can be made.
First, the area affected by the Northern Subject Rule in the traditional
dialects reaches a good deal further south into the East Midlands than
shown in those maps based exclusively on the published SED material. The
data suggest that there is a broad transitional zone in which plural verbal -s
occurs but becomes progressively rarer the further one goes to the south.
The outer limits of this transitional zone are still virtually identical to the
limits of the northern system of six hundred years earlier (see section 4
below). This can be seen by comparing the attestations in the SED (Map 7)
with those in the Linguistic Atlas of Late Middle English (LALME,
McIntosh, Samuels and Bensik 1986), displayed in Map 8.
Second, the distribution of tokens in the border zone reveals an
interesting concentration of some special environment types. Whereas in
the core area of the northern dialects proper, plural verbal -s can be found
in all types of clauses, with all types of subject noun phrases and all types
of verbs, towards the transition zone in the south plural verbal -s is
documented almost exclusively in a set of special environments. Among
them are: relative clauses; clauses with demonstrative pronoun subjects
such as themor those; and clauses with indefinite pronoun subjects such as
some, some of themetc. Unfortunately, these are by and large just the types
that were not documented systematically in the published SED material.
These types of environments together make up for almost 80 per cent of all
the (relatively sparse) tokens of plural verbal -s in the transitional zone. In
the northern dialects proper, their predominance is less strong, but with
roughly 50 percent of all recorded tokens they still appear over-represented.
To these environment types that seem to be particularly favourable to
verbal -s can be added cases of subject-verb inversion, as in questions and
tag clauses.
Map 7. Plural verbal -s in the SED fieldworker notebooks
7
(all NP subjects)
L. Pietsch 2003
0 50 100km
fieldworker
S. Ellis others
no tokens
1-2 tokens
3-5 tokens
6 or more
no tokens
1-4 tokens
5-10 tokens
11 or more
Cheshire
Cumberland
Derbyshire
Durham
Lincolnshire
Lancashire
Leicestershire
Northumberland
Norfolk
Nottinghamsh.
Northsh.
Staffordsh.
Yorkshire
Westmorland
Shropshire
Cambridgesh.
Huntsh.
Warwicksh.
1
2
3
4
5
6
7
8
9
1
2
3
4
5
6
1
2
3
4
5
6 1
2 3
4
1
2
3
4
5
6
7
8
9
10
11
12
13
14
1
2
3
4
5
6
7
8
9
10
11
12 13
14
15
16
17
18 19
20
21
22 23
24
25
26
27
28
29
30
31
32
33
34
1 2
3
4
5
6
1
2
3
4 5
6
7
1
2 3
4
1
2
3
4
5
6
7
8
9 10
11
12
13
14
15
1
2
3
4
5
6
7
8
9
10
11
1
2
3
4
5
6
7
8
9
10
11
1
2 3
4
5
6
7
8
9
10
1
2
1
2
1
2
3
4
5
7
1
2
3
4
5
6
7
1
2
3
4
5 6
7
1
2
3
4
5
1
2
1
4
7
164 Lukas Pietsch
Map 8. The NSR in Late Middle English (based on data in LALME I: 467, IV:
110111)
Shown below are typical SED attestations of relative clauses, including
presentational relative clauses after existential there, most typically with
zero relativizer and with non-concord in the preceding existential
construction (23), as well as after cleft its (24).
(22) a. You dont see many has holes now [SED: La7]
b. Themwhats got a few [SED: Lei5]
c. Hedges that hasnt been done [SED: Lei9]
d. The ones that goes across was braces [SED: R2]
L. Pietsch 2003
0 50 100km
Linguistic Profiles
Regular Verbal -s
Occasional Verbal -s
Rare Verbal -s
e. It kills the thorns as grows round it [SED: Nth2]
f. I know several signs as is pretty sure [SED: Nth4]
g. People what was used to it [SED: Hu1]
(23) a. Theres a lot of the people now as doesnt talk like they used to
[SED: L9]
b. Theres any amount takes cattle on [SED: Nb4]
c. Theres not so many fills a ten-quart tin [SED: Cu3]
d. Theres a lot of people kills em[SED: L15]
e. Theres two or three comes up at five oclock in the morning
[SED: Lei7]
(24) Were both right, arent we, its tothers ats wrong [SED: Y16]
Typical attestations with indefinite pronouns are shown under (25), and
examples with demonstratives are shown under (26). The indefinite items
includesome, most, many, a lot, etc., often with postmodifiers like of them,
sometimes also in determiner function followed by nouns. As for the
demonstratives, the most typical item is them, but standard these/those also
occur; in the Ulster and Scottish data a demonstrative determiner they (also
spelled thae in written Scots, and not to be mistaken for the simple
pronoun) occurs and has a similar effect. Examples such as these are
extremely common in all parts of the data.
(25) a. some uses a jug and gets it broke [SED: R1]
b. some says lop but Id say slat [SED: L13]
c. some on emreckons it int [SED: Lei1]
d. some on ems red [SED: L13]
e. most on emhas one [SED: L14]
f. a good many keeps hens [SED: Lei10]
g. some folks steals em[SED: L9]
(26) a. thems ourn [SED: Lei9]
b. thems gisters in the top field [SED: Lei9]
c. these is the front of these [SED: Lei2]
d. I wonder if themtwos married [SED: L11]
Plural verbal -s in inversion is exemplified in (27). Note that the list
includes several instances where em, a weak form of demonstrative them,
166 Lukas Pietsch
is used as a tag subject just like a simple pronoun they, i.e. apparently
without emphasizing or deictic force. It nevertheless takes verbal -s,
retaining its status of a full NP, and not of a simple pronoun, in terms of
sensitivity to the Type-of-Subject Constraint. This can be taken as a piece
of evidence that the Type-of-Subject Constraint is defined in terms of
particular items and the constructions they are part of, not in terms of some
abstract semantic or formal features defining pronoun-ness.
(27) a. Is themtwo married? [SED: Nb2]
b. Whos them? [SED: Nb3]
c. Is thy teeth warking? [SED: Cu5]
d. Wheres my yorks at? [SED: We4]
e. Doesnt em? [SED: La4]
f. Is em? [SED: La4]
g. Has thi taties comed up yet [SED: Y7]
h. Is both you women wed? [SED: Y7]
i. Has themhorses been served? [SED: Y27]
j. Wheres themcome fra? [SED: Y28]
A special idiomatic type of inversion environment which appears to be
particularly favourable to plural verbal -s, found mostly in Yorkshire data,
is in tag clauses of the kind exemplified under (28) and (29), used as a
postponed expression of a sentence topic. Note that these are typically
preceded by a clause with a co-referential pronoun they, and that this first
clause invariably has the appropriate verbal form without -s.
(28) a. Theyre real hard gossips, is them. [SED: Y2]
b. Theyre rough mutton, is tups. [SED: Y3]
c. Theyre rumthings, is the pigs. [SED: Y11]
d. Theyre a bit queer, is pigs to manage. [SED: Y12]
e. Theyre very affectionate, is pigs. [SED: Y25]
f. Theyre laced boots, is these of mine. [SED: Y18]
g. Theyre not worth bringing up, isnt little pigs. [SED: La6]
(29) a. They vary, does stee-steps. [SED: Y18]
b. Theyve recently comed, has them. [SED: Y18]
c. They feel over emand weighs em, does butchers. [SED: Y22]
d. They always crawl upwards, does lice. [SED: Y25]
Two tokens of this type, both from the same informant, are also recorded in
the more recent FRED data, both using was (and both closely preceded by
rule-conforming they were):
(30) a. They were proper slaves, was women, in them days.
[FRED:Yks9]
b. They were wicked, was farmers, for playing nap.
[FRED:Yks9]
As was mentioned above, the Northern Subject Rule also interacts in
interesting ways with the concurrent pattern of was/were levelling. Tokens
of plural was with plural NP subjects are found throughout the area affected
by the Northern Subject Rule. To the degree that they was is rarer or absent
in these dialects, these forms can be ascribed to the Type-of-Subject
Constraint. Such tokens occur even in areas which otherwise have strong
preferences for generalized were even in the singular, such as the core
generalized were area in south Yorkshire and the northwest Midlands.
Examples are shown under (31). Note that many of these tokens also
exemplify the common occurrence of subjects involving demonstrative
themand other demonstrative pronouns.
(31) a. Shops was open while ten. [SED:Y20]
b. Thems sideboards what was cut off. [SED:Y25]
c. Themthat was on that ship at went down. [SED:Y27]
d. Half onemwas apt. [SED:Y28]
e. Horses was baiting. [SED:Y28]
f. These was like that. [SED:Y32]
g. Thempigeons was there. [SED:Ch2]
These findings fit in with observations reported by Wright
8
as early as in
1892: in the dialect of Windhill in the West Riding of Yorkshire, plural
verbal -s had largely become restricted to relatives and to the forms of have
and be, and it occurred in other environments only rarely as a relic form
(1892: 156). Wright also includes several examples with the subject them
among lists of typical occurrences:
(32) a. Thems the men that does their work best.
b. Themmens been very good to me.
c. Us thats done so much for him.
168 Lukas Pietsch
d. Me thats so poorly.
e. The coals isnt done yet.
(33) a. Ive done.
b. Theyre at it again.
c. Themmen do their work very well.
Wrights location is situated well north of the burglars steals isogloss in the
SED. Although the SED data does attest the continued presence of verbal -s
in that same area even outside the special favouring environments, more
than half a century after Wright, his observation suggests the existence of a
strong favouring tendency, and as such it fits in nicely with the observation
in the SED that verbal -s was preserved longer, and further in the south, in
just the same set of favouring environments.
The impression that environments of this type are over-represented
among the occurrences of verbal -s cannot be tested statistically in the SED
data, because the SED does not fully record all utterances of any informant
through a given stretch of discourse. However, very similar effects are also
found in the corpus data of both the NITCS and FRED. In both corpora, the
relative clause constraint is solidly in evidence, as relative clauses take
verbal -s up to twice as often as other clauses. This goes especially for that
relatives and for the non-standard zero relatives that are common in these
dialects. Wh-relatives, on the other hand, tend to co-occur more often with
standard concord behaviour, apparently due to their stylistic status of being
more characteristic of formal Standard English. A similar stylistic
differentiation can also be found with respect to the demonstratives.
Whereas the non-standard demonstratives themand thae have a very strong
favouring effect on verbal -s, the effect of the standard demonstratives
these and those is rather the opposite.
Potential candidates for plural verbal -s in inverted environments are
quite rare in the corpus (since the overwhelming majority of inverted
clauses have pronoun subjects and/or modal verbs), but in the few cases in
point, verbal -s is strongly over-represented (7 out of 9 cases in the NITCS,
and 11 out of 12 in FRED). Inverted clauses must therefore be counted as
one of the most strongly favouring environments.
As for the indefinite pronouns, effects are difficult to quantify, as the
class of constructions that exemplify this type is a rather open one. Its
typical ingredients (some, a lot, ones, of them, etc.) can occur in a wide
variety of combinations, and they also often co-occur with some of the
other factors mentioned above. For instance, many instances of a lot are
followed by a relative clause; ones is often preceded by a demonstrative
them ones, etc. It is therefore not easy to find the most appropriate
classification when it comes to testing the effects of these and similar items
statistically, as the statistic method always requires the counting of tokens
across a set of distinct, clearly defined environment types. The
classifications used for the statistics below are therefore somewhat
arbitrary. However, intuitively it seems to be the case that the effect is
linked not so much to a precisely circumscribed set of structural
environments anyway, but rather to a set of prototypicality conditions. It
will be discussed in more detail in section 6 how effects of this type can be
integrated into a theoretical model of grammatical variation. Going through
the attested examples of plural verbal -s, one is left with the impression that
sentences such as theres a lot of people kills emrepresent something close
to a common prototype, and that a sentences likelihood of having verbal -s
depends on its relative degree of similarity or dissimilarity with this or a
small number of other prototypes. Among the conditions that seem to play
a role here is the subjects role of being focussed, unlike in the most
common constellation of a clause with a pronoun subject, where the subject
is typically topical and unstressed.
However, apparent effects of favouring verbal -s may also be linked
quite arbitrarily to specific items in the syntactic or semantic environment.
For instance, it was found in the NITCS data that sentences with the
subjects times and days displayed a far higher than average proportion of
verbal -s, as in the examples below (3435). This usage type must be
characterized as a special stylistic idiom.
(34) a. The times is better in a way, like, as regards money. [NITCS:
L4.3]
b. Oh, aye, times is a whole lot better now. [NITCS: L15.2]
c. Oh, the times is a-changed very much. [NITCS: L36.3]
d. The times was very bad. [NITCS: L8.3]
e. If the people had sense, the times is perfect. Thats what the
times is. [NITCS: L8.3]
(35) After all, my young days was far better. [NITCS: L10.3]
Quite prominent among the factors favouring or disfavouring verbal -s is,
not surprisingly, the type of verb involved. In the FRED data, a moderate
170 Lukas Pietsch
overall preference for plural was (used in 51 per cent of all cases of past-
tense be) is in evidence in all parts of the corpus. Is is also used slightly
more often (35 per cent) than the -s forms of other verbs (31 per cent). It is
perhaps remarkable that this effect is not even stronger. J udging from some
recent descriptions of northern varieties e.g. Tagliamonte (1999) on York
English, or Miller (2004) on the current spoken language in the Central
Lowlands of Scotland one might expect the trend towards a restriction of
the Northern Subject Rule to was/were to have progressed much further.
These studies attest effects of the Northern Subject Rule to be preserved
only for be. As for modern Scots, Grlach (2002: 95) suggests that the
Northern Subject Rule may have become a victim of social stigmatization
in recent decades, despite its long tradition in Scots, because of its overlap
with the non-standard verbal -s in other colloquial varieties of English. He
reports that it is even avoided in some modern Scots writing. This
stigmatization may explain the presence of some speakers in the FRED
corpus who consistently lack verbal -s but otherwise display many quite
distinctively Scots grammatical features in their speech. The very solid
presence of Northern Subject Rule effects both with be and with lexical
verbs in many other speakers in the corpus confirms that FRED, on the
whole, represents a range of predominantly very conservative types of
speech.
In the NITCS data the situation is more complex, since internal
geographic diversification was found within Ulster in this respect. Towards
the northeast, in what has been known as the area of traditional Core
Ulster Scots dialects (Gregg 1972), the Northern Subject Rule seemed to
apply without much distinction to all verbs, with relative frequencies of
verbal -s at a uniform but only moderate level (around 21 per cent) for all
verbs alike. Along the southern border of Ulster, plural was was found
much more frequently (56 per cent) than other -s forms, including is (24 per
cent). The verb have seemed to play another special role, showing
exceptionally low levels of verbal -s in this area (5 per cent). Interestingly,
the highest levels of verbal -s usage overall were found in neither of these
two areas, but in the transitional zone between them, in central and
southeast Ulster (see Pietsch 2005: ch. 5.3.1 for details).
Table 5 and Table 6 show the relevant statistics for the most important
linguistic environment factors in terms of two (partial) Varbrul models
calculated separately for the FRED and NITCS data. For technical
purposes, the definition of the factors and factor groups is not exactly
identical in both models, but the similarity of the results with respect to the
principal effects discussed so far can easily be seen. (For details regarding
the design of the models, interaction tests between the various factor
groups, and the statistical integration of the social and geographic factor
groups with the linguistic factor groups shown here, see the discussion in
Pietsch 2005).
Table 5. Partial Varbrul model for NSR in FRED
Factor Group Factor Tokens
-s/Total
% Factor
Weight
Verb Was/Were 215/420 51% 0.58
Is/Are 32/92 35% 0.42
Other Verbs 52/167 31% 0.34
Clause Inverted 7/9 78% 0.91
Zero/That Relative 64/90 71% 0.78
Canonical SV (Adjacent) 198/475 42% 0.46
Wh- Relative 9/26 35% 0.37
Canonical SV(Non-Adjacent) 21/79 27% 0.35
Subject Them 19/29 66% 0.53
Indef. Pronouns 61/129 47% 0.56
Other Demonstratives 13/43 30% 0.31
Other NPs 206/478 43% 0.50
Total 299/679 44%
Summing up, there seems to be strong evidence that the use or non-use of
verbal -s in the variable grammars of present-day dialect speakers is
governed by a set of prototypicality conditions. Certain types of
environments, defined in syntactic, lexical, or possibly also semantic terms,
are associated with a relative preference for the use of the conservative
dialectal option of verbal -s. The likelihood for any particular clause to
display verbal -s seems to depend on its degree of closeness to one of these
prototypes. The relevant conditions may range from very general,
productive patterns (such as the relative clause or inversion constraints) to
highly specific (such as the times constraint detected in the NITCS data). In
diachronic terms, these effects may be described as a gradual loss of
productivity of a once general, universal pattern, which may in the long run
lead to its becoming fossilized in a highly restricted set of environments. In
synchronic terms, the effects in the grammars of individual speakers may
best be characterized as structural idioms: arbitrary properties associated
172 Lukas Pietsch
with specific construction types, which range somewhere in between fully
general syntactic rules on the one end, and individual lexical properties on
the other.
Table 6. Partial Varbrul model for NSR in the NITCS
Factor Group Factor
Tokens
-s/Total
%
Factor Weight
Clause Type
Inverted 11/12 91% 0.97
Zero relatives 37/60 61% 0.79
That relatives 38/79 48% 0.58
Canonical SV 227/751 30% 0.46
Wh- Relatives 3/36 8% 0.23
Subject-Verb
Distance
Non-adjacent 52/122 42% 0.62
Adjacent (and
zero subjects)
253/804 31% 0.48
Determiner of
Subject NP
Demonstrative
them/thae,
interrogatives
30/38 78% 0.86
Quantifiers +of
them
42/95 44% 0.69
Others 235/746 31% 0.47
Demonstrative
these/those
9/59 15% 0.31
Nominal Head
of Subject NP
Times/Days 18/21 85% 0.86
Ones 19/34 55% 0.69
Other nouns 192/643 29% 0.47
Total 316/938 34%
4. The history of the Northern Subject Rule
So far I have dealt with the variable grammars of recent dialects, and
centrally with the reflexes they preserve of the Northern Subject Rule,
which was shown to be one of the main characteristic patterns of the north.
In the present section I will explore what can be reconstructed of the
historical development of this pattern. Unfortunately, much of its historical
origins lie in the dark. Its first appearance in the dialects of northern
England must be dated to the time of early Middle English, a period from
which no written documents of northern provenance are extant. In the latest
surviving Northumbrian Old English documents from before this gap (the
Lindisfarne Gospels, the Rushworth Gloss, and the Durham Ritual, all mid-
tenth century), there are signs of an ongoing change that can be understood
as a corollary and prerequisite of the emergence of the later rule, namely
the change from the verbal -e/-a/-ia/-is affixes to neutralized -es. In the
earliest reliable Middle English documents from after the gap, from c.1300
onwards, the Northern Subject Rule is already fully in place.
It has been suggested that the emergence of the Northern Subject Rule
may have been a result of earlier contact of English with Brythonic Celtic
(Klemola 2000). The idea is tempting because of certain typological
parallels between the northern system and the concord systems of Welsh
and related languages. J ust like northern English, Welsh displays non-
agreement with full NP subjects but agreement with accompanying
personal pronouns. (For the different case of a rather more elusive parallel
between the northern English concord system and that of Irish see the
discussion in Corrigan 1997: 190225 and Pietsch 2005: ch. 2.1.2.)
However, I have argued elsewhere (Pietsch 2005: ch. 3.4) that a transfer
explanation of the kind envisaged by Klemola is not tenable. This is mainly
due to problems with the relative timing between the supposed period of
contact and certain dateable linguistic changes such as the affix
neutralization mentioned above, which are logical prerequisites of the
emergence of the Northern Subject Rule and must therefore predate it. In
what follows, I will instead sketch a possible alternative model of how the
pattern could have evolved. It is based primarily on language-internal
causal factors such as analogy and frequency-induced change.
It is, first of all, crucial to point out that the verbal -s forms in those
environments where they do not match modern Standard English are not an
innovation. They were not, as one might think, an intrusion into these
positions from the third person singular. Rather (for most verbs, at least)
174 Lukas Pietsch
these -s forms are a conservative retention. Etymologically, verbal plural -s
is no less genuine a reflex of an original agreement marker (northern OE
plural -a>-as>-s) than third person singular -s (northern OE -e>-es>-s).
The main innovation thus lies not in a spread of the suffix but in the spread
of the suffixless forms, in the environments with adjacent pronoun subjects.
This innovation can plausibly be seen as part of the general drift of affix
loss that affected all the Germanic languages. It may appear somewhat
unexpected that it should have been the northern dialects that eventually
preserved more of the traditional affixal system than others, in this
particular domain of the grammar. After all, it is widely agreed that the
trend towards affix loss was, on the whole, particularly vigorous in the
north (possibly owing to a situation of intensive language contact with
Scandinavian). But it must also not be overlooked that the apparent
conservatism of the northern dialects in this particular respect only applies
to a last remnant of the affixal system, and not to its actual function of
agreement marking. Paradoxically, in keeping the affixed verb forms,
northern English eventually preserved more of the phonological substance
of the old agreement system than the standard did, but only at the price of
having in effect less agreement.
This paradoxical outcome may be interpreted as the result of a
conspiracy of two independent developments originating in different
dialects. One of them was the weakening and subsequent neutralization of a
set of previously distinct but phonologically similar affixes (-e/-a/-ia/-is
> -s). This development originated in the north and was well advanced by
late Old English. The other was the innovation of affixless, so-called
syncopated forms, at first only in a certain restricted set of syntactic
environments adjacent to pronouns. This development was apparently
headed by the southern dialects and only began to reach the north at some
time during late Old English. At this point, the previous neutralization
process had already brought the older, fuller agreement system to the verge
of breakdown in that dialect, having obliterated almost all distinguishing
contrasts in the verbal paradigm. The new affixless forms that were the
output of the second innovation were therefore apt to be reinterpreted and
pressed into service as carriers of a new agreement contrast.
A somewhat more detailed description of the whole process, as far as it can
be reconstructed, is in order at this point (see Figure 1). The northern
dialects of Old English originally shared the common Old English
agreement system (Brunner and Sievers 1965: 352378), which can be
summed up as follows. The three persons in the plural had already been
Syncopation
Midl. repl. ->-n
Affix
Spread of -sto
Northern Subject Rule
syncretized in pre-Old English and were -a or -ia in the present
indicative of lexical verbs, -en or -un in other paradigms. The singular
forms in the present indicative were originally the common Germanic
forms -u, -is, -i (in the case of the strong verbs; the endings in the weak
classes were similar.) They underwent a process of vowel weakening to -e,
-is, -e during Old English. Despite this, the plurals remained distinct from
the third singulars for some time (strong verbs -a versus -e; weak verbs
-ia versus -e or -a respectively). In the course of the Old English period,
the second singular added a -t affix after the -s, owing to a reanalysis of the
cliticized subject pronoun u.
Figure 1. Development of agreement paradigms in Old and Middle English
During the transition to Middle English, the three principal dialect areas of
England behaved differently. The Midlands dialects replaced the plural -a
endings with -en, apparently through analogy with the past and subjunctive
paradigms. Through later phonological reduction of the endings, these -en
North
South
Spread of affixless forms
Middle English Modern English
sing-e
sing-
es(t)
sing-e
sing-a
sing-a
sing-e
sing-est
sing-e
sing-a
sing-a
sing-a
sing we
sing ge
sing
sing-est
sing-e
sing-en
sing-en
sing-en
sing
sing-est
sing-s
sing
sing
sing
sing-e
sing-es
sing-es
sing-es
sing-es
sing-es
sing-s I sing
sing-s
sing-s
sing-s we sing
sing-s you sing
sing-s they sing
sing
sing-s
sing-s
sing-s
sing-s
sing-s
sing we
sing ge
sing ai
we sing
ge sing
ai sing
176 Lukas Pietsch
forms developed directly into the modern affixless forms of Standard
English (cf. Figure 1, bottom). The south retained the - forms throughout,
only to replace them with universal generalized -s much later in the modern
dialects.
The north replaced the - forms, both in the plural and in the third
singular, with -s, a change that occurred already during the Old English
period. This change has repeatedly been linked to Scandinavian influence
(Keller 1925, Samuels 1988). This idea has met with considerable
skepticism (for a survey of the older literature see Brunner 1962: 177), but
it seems to have gained ground again in recent decades in the light of
current language contact theories (Bailey and Marold 1977: 45, Thomason
and Kaufman 1988, Stein 1986; for a recent sceptical view see Ferguson
1996: 178). The older northern documents, such as Cdmons Hymn, the
Leiden Riddle, and with one notable exception the runic inscriptions, all
have the old - forms. The late Northumbrian documents from the mid-
tenth century are witness to a stage where - and -s were used variably. The
variation in these documents has repeatedly been the object of quantitative
analyses (most thoroughly by Berndt 1956, but see also Holmqvist 1922,
Ross 1934, Blakeley 1949 and most recently Stein 1986). These studies
suggest that the variation was conditioned, among other things, by phonetic
environment factors, but also already by a tendency to treat pronoun
subjects differently from full NP subjects; hence in some way
foreshadowing the later Type-of-Subject Constraint.
During the same time when - changed to -s in the north, the vowels in
the two endings also lost their contrasts. The new -st ending in the second
singular appears variably in the late Northumbrian documents, but seems to
have been only a temporary intrusion from the south, and was then again
replaced by -s. The -e ending in the first singular became mute. Taken
together, these changes meant that by the time of early Middle English the
present tense paradigms of lexical verbs must have contained only two
distinct forms, - and -(e)s. In the present subjunctive, the neutralization
had gone even further, through the loss of final -n, so that only one form -
was left. In the present indicative, a hypothetical Late OE or Early ME
stage must be posited in which all forms except the first singular had been
completely neutralized in -s (cf. Figure 1, top left).
At about the same time, the second of the two innovation processes
mentioned earlier seems to have set in: new affixless forms must have
developed in the pronominal environments in the plural. The exact nature
of this process is not known, but the most likely source is found in the
southern dialects of some centuries earlier. In the West Saxon standard
form of Old English, the original plural -a and -enaffixes were sometimes
deleted before the first and second person plural pronouns we and ge, when
these were in an immediately post-verbal, probably clitic, position (cf.
Figure 1, bottom left). Brunner and Sievers (1965: 360) suggest that these
so-called syncopated forms first occurred in the subjunctive. Here they
would be explained easily enough as an early application of the overall
trend of deleting final -n. They would then have spread by analogy into the
indicative. Why in southern Old English these syncopated forms remained
restricted to the first and second person plural and did not also spread to the
third person is another open question but need not concern us here. What is
important is the fact that the syntactic positions these forms occurred in are
a subset of those where the affixless forms occur in the later northern
dialects.
Curiously, northern Old English, although more progressive in matters
of affix reduction in most other respects, seems to have been slow to pick
up these syncopated forms at first. In the surviving tenth century
manuscripts they are attested, but marginal (Berndt 1956: 213216). It can
be no more than a matter of speculation what exactly triggered the sudden
and much wider spread of these affixless forms during the following three
centuries. In any case, once the affixless forms did start to appear in the
north, they must have spread fairly quickly into all positions with
pronominal subjects, irrespective of their pre- or postverbal position.
Moreover, the affixless forms were also adopted in the third person plural,
where they had not occurred in the older southern dialects, and where the
northern dialects had in the meantime also innovated a new personal
pronoun, the Scandinavian loan they. (cf. Figure 1, top centre). This process
had approached completion by 1300 in the north, when written
documentation of that area sets in again.
The functional motivation for the quick and thorough adoption of the
affixless forms seems to have been that they were co-opted to serve as
genuine agreement forms, re-introducing and stabilizing a formal contrast
that was no longer signalled reliably by the older set of affixes. However,
the newly introduced contrast remained restricted to pronominal
environments, as it had been a subset of the pronominal environments
where the affixless forms had originally emerged. Once they had assumed
their new role as a carrier of agreement marking in these environments,
they stopped spreading further across the system. Thus had emerged a
relatively stable new system with an effective split of the once uniform
178 Lukas Pietsch
system of subject-verb agreement into two distinct paradigms valid for two
complementary sets of environments: a contrast of -s vs. - with pronoun
subjects, and neutralized -s everywhere else. In a certain sense, this
paradigm split can be seen as a major structural innovation in the history of
English, comparable perhaps to the common Germanic split of the adjective
inflection into the strong and weak paradigms: the occasional clear
instance of an innovation that goes contrary to the main drift and persists
along with it for long periods of time (Ferguson 1996: 189).
Finally, the reorganization of the concord system was completed by two
further developments. The first person singular which had had no
consonantal affixes from the outset introduced new and unetymological -s
forms in positions not adjacent to the subject, obviously by way of analogy
with the other persons and numbers. Then, also by way of analogy, the
usage of is/are and of was/were was partly aligned with that of the other
verbs, as was and is gradually began to be used in plural environments in
the same way as the -s forms of other verbs (Montgomery 1994). At this
point, the ideal Northern Subject Rule system as described in definition
(1) above had been reached.
By the late Middle English period, the Northern Subject Rule had
established itself as a more or less categorical pattern across all of northern
England and Scotland (Mustanoja 1960: 481482). There is some evidence
that the pattern had been spreading from the north into the northeast
Midlands, especially Lincolnshire, during the thirteenth century (Berndt
1982: 131). From then on, its southern boundary, roughly along a Chester-
Wash line, was to remain stable up to the traditional dialects of the
twentieth century, as was shown earlier in Section 3.6 (see Map 7 and Map
8 on pages 163164).
Laing (1978: 244247) and McIntosh (1988) draw attention to several
types of intermediate systems that occurred along the boundaries of the
northern area. In all of them, there was an alternation between two forms
that was governed by similar constraints as in the northern system proper,
but involved different morphological material (-s, -th, -en, or -). It seems
not entirely clear in all cases to what degree each of the affixes in these
intermediate systems represents etymologically regular reflexes of older
affixes in the respective dialects, or later analogical reintroductions
triggered through contact with the northern system proper (cf. also Schendl
1996: 149).
Plural -s forms apparently conforming to the Northern Subject Rule are
also widely attested in the emerging Early Modern English standard
language, including the language of Shakespeare (Knecht 1911). This is
usually interpreted as a dialect contact effect brought about by the massive
amount of migration from the north into the London area (Schendl 1996,
2000; cf. also Bailey, Maynor and Cukor-Avila 1989; Bailey and Ross
1988; Montgomery, Fuller and DeMarse 1993; Wright 2002).
The Northern Subject Rule was brought to Ireland, especially to Ulster,
by settlers both from Scotland and England (McCafferty 2003, 2004).
Later, it was to be transported by Scottish, Irish and English settlers to
many of the new overseas varieties of English. There has been some debate
about just to what degree certain commonalities in verbal concord variation
discernible across a wide range of overseas Englishes today can historically
be ascribed directly to a common, specifically northern, dialectal source, or
to what degree these similarities reflect universal, functionally explainable
trends in English (see e.g. Montgomery 1988, Bailey and Ross 1988,
Montgomery, Fuller and DeMarse 1993, Tagliamonte 2002, Chambers
2004). While strong northern input seems historically plausible for many of
the varieties in question, there is also some evidence that variation patterns
resembling the northern rule may have developed independently in some
places, even within England. For instance, some studies of present-day
southwestern varieties have found quantitative (though non-categorical)
constraints similar to the Type-of-Subject Constraint to be active (Godfrey
and Tagliamonte 1999, Peitsara 2002). The apparently independent rise of
such structures strengthens the case for some dialect-universal mechanism
(vernacular primitives in the terms of Chambers 2004).
5. Theoretical accounts of the Northern Subject Rule
There have been several attempts to characterize the northern concord
system within formal theories of syntactic competence, using various
models in the tradition of formal grammar. Besides some very brief
discussion in Roberts (1993) and Van Gelderen (1997), and a somewhat
inconclusive treatment in Corrigan (1997), I am aware of two more
elaborate proposals: Henry (1995), and Brjars and Chapman (1998). A
third and rather different approach, also within a rigidly formalized
framework but with a radically different stance regarding linguistic
universals, can be found in a brief sketch in Hudson (1999). However, each
of the existing formal proposals is burdened with problems on the level of
descriptive adequacy.
180 Lukas Pietsch
The difficulty which formal theories have had in dealing with the
Northern Subject Rule is demonstrated by the fact that the existing
proposals have mostly concentrated on either the Type-of-Subject
Constraint or the Position-of-Subject Constraint, according to some
descriptive definition or other, but none of them has succeeded in
integrating both in a unified account. A part of the problem seems to be due
to a somewhat puzzling feature of the Northern Subject Rule itself, which I
have elsewhere called the markedness paradox (Pietsch 2005: ch. 1.2.4).
All the existing formal analyses implicitly operate with the concept of
marked and unmarked forms (even when they do not explicitly use that
term). However, which of the two forms involved in the dialectal concord
system (-s, -) is the marked member of the paradigm and which is the
unmarked one?
In Standard English, the -s form unambiguously encodes one particular
person-number value, the third singular, whereas the - form acts as the
default for all other persons and numbers. It is therefore fairly
unproblematic to characterize the -s suffix as an agreement morpheme in
the normal sense, and hence, the -s form as the marked member of the
paradigm. This is a straightforward analysis even though the system may be
a rather uncommon one in terms of a cross-linguistic comparison, as the
third singular is generally the most likely candidate for zero marking within
agreement paradigms (Bybee 1985). Matters are rather different in the
northern system, even though in personal-pronoun environments it is
identical to the standard, encoding the same grammatical information with
the same means. In all non-pronominal environments it is the generalized -s
form that leads to a complete neutralization of all person-number agreement
contrasts. It can therefore be argued that the -s form in the NSR system,
though formally carrying the overt agreement morpheme, acts as a
functionally featureless form devoid of person-number information. The -s
morpheme has repeatedly been characterized instead as a mere tense-mood
marker. In contrast to this, the formally unmarked - form, where it occurs,
has the effect of upholding agreement oppositions, particularly that
between singular and plural in the third person. It is therefore usually
regarded as the one that functionally does carry genuine person-number
agreement features. This is a plausible synchronic analysis for the modern
system, even though at the earliest stage, at the time when the zero forms
first emerged, the system must have been just the reverse. As described in
section 4, the zero forms were originally a product of the erosion of the
agreement morphology. They were reanalysed as genuine plural agreement
forms, taking on a new functional load as carriers of agreement
information, only after the two formerly distinct endings -e/-es and -a/-as
happened to fall together and were re-analysed as default singular forms.
The resulting, somewhat paradoxical distribution of markedness properties
will be seen as a recurrent problem that has affected several synchronic and
diachronic accounts of the Northern Subject Rule.
5.1. Henry (1995)
Henrys (1995) discussion of plural verbal -s (singular concord, in her
terminology) is part of an extensive analysis of the syntax of contemporary
Belfast English, based on a late principles-and-parameters version of
generative grammar. As such, it is an example of the type of variation
studies that has come to be known as microparametric syntax in recent
generative work. Henrys data are based on acceptability judgements
elicited in interviews with Belfast informants. While Henry devotes
extensive discussion to assumed parametrical links between the
phenomenon of optional plural verbal -s and some other non-standard
grammatical phenomena in this dialect, she makes no attempt at integrating
her analysis of the narrowly local variety of Belfast with a historical or
larger geographical account, with respect to the continuity of the
phenomena in question across a wider range of varieties. As to her
methodology, it must be noted as unfortunate that Henry fails to give any
detailed account of the number and choice of informants, her methods of
sampling, or the set of test sentences used. Neither does she give any
detailed quantitative or qualitative account of how her informants reacted to
each test sentence.
Henrys account of the reflexes of the Northern Subject Rule in Belfast
English differs descriptively in some respects from descriptions found
elsewhere in the dialectological literature, and it is sometimes difficult to
judge whether these apparent differences reflect genuine divergence
between the dialects in question, or whether they are merely due to
diverging research methodology. Most strikingly, Henrys analysis does
not deal with the Position-of-Subject Constraint in the form it is usually
stated in the descriptive literature. Neither the relative clause structures of
the type they that goes, nor the classical northern pattern they sing and
dances are mentioned in her analysis; nor does she say anything about the
effect of intervening quantifiers or adverbs, as in they both goes, they often
182 Lukas Pietsch
goes (the latter structures admittedly being rare in Northern Irish English
but by no means non-existent, at least in the traditional dialects surrounding
Belfast). Interestingly, Henry does report (1995: 19, 26) on an apparent
effect in exactly the opposite direction: with adverbs intervening between a
full-NP subject and an operator verb (raising verb in her terminology,
that is, a verb that raises to INFL before spellout), verbal -s is reported not
to be licensed. This means that contrary to the common pattern observed
elsewhere, non-adjacency of subject and verb in this case would favour the
standard agreement pattern. Thus:
(36) a. The children really are late.
b. The children is late.
c. *The children really is late.
Henry explains (1995: 19) that it seems to be the case that the adverb
position between the subject and the topmost projection of INFL which
exists in English is unavailable in singular concord. She does not state
explicitly whether the same kind of environment has any influence on
agreement with personal-pronoun subjects (they really
?
is/
?
are late), but
from her statement it would seem that a fortiori the -s forms should be
ruled out in that case too. A comparison of her observations with some data
from the NITCS confirms the possibility that verbal -s in some of these
environments may indeed be rare or avoided (Pietsch 2005: ch. 2.1.1).
However, since the adverb position Henry is speaking of tends to be
reserved for a rather restricted set of semantic-pragmatic environments in
English (Quirk et al. 1991: 8.18), it does not become clear from Henrys
discussion whether her informants reluctance to accept verbal -s in these
positions really was caused by some property of the phrase structure as
such, or by other factors such as, for instance, stress conditions on the verb
form, or possibly by the mere rareness of the construction type. It would
seem that Henrys set of test sentences, and her methodology in reporting
informants actual reactions to them, are somewhat too limited to draw far-
reaching structural conclusions on this particular issue.
A second, interesting constraint that Henry reports concerning the
position of subjects is a prohibition of verbal -s under inversion (1995: 16
18). This would exclude sentences of the type does the children sing? in
Belfast English. This constraint is not confirmed in the data of the present
study: as pointed out in section 3.6 above, sentences of this type are
regularly attested in conservative varieties of the northern type both in
Northern Ireland and in Britain. They were even found to be among the
environments most favourable to verbal -s. While the overall textual
frequency of potential application environments of this type in the corpora is
quite low the bulk of all question clauses actually produced contain either
pronoun subjects, or modal non-concord verbs, and thus fall outside the scope
of the rule the relative frequency of verbal -s in the few tokens produced is
extremely high. Again, it seems unfortunate that Henry does not give any
detailed account of her test procedure and results. The question whether these
contrasting findings represent a genuine difference between present-day
Belfast English and other dialects of the northern type, or just an artefact of
diverging research methodologies, cannot be discussed in more detail here.
As far as the Type-of-Subject Constraint is concerned, Henry (1995: 23)
finds that verbal -s outside the third person singular is possible in Belfast
English whenever subjects are not openly marked as nominative. This
formulation of the Type-of-Subject Constraint covers the prohibition of
verbal -s with the simple personal pronouns I, we, and they. The pronoun
you must be treated as an exception, as it fails to have an overt distinction
between subject and object case but nevertheless patterns together with the
other three pronouns with respect to the Northern Subject Rule. Henry
takes as a strong piece of evidence for her claim that, according to her
observations, the prohibition of verbal -s with these items does not apply
only to the standalone pronouns but also to co-ordinated NPs involving any
of them. The strong forms of the pronouns, which are identical to the
accusative forms, are not subject to such a constraint. Co-ordinated NPs
involving these can freely take singular concord:
(37) a. You and I are going.
b. You and me are going.
c. *You and I is going.
d. You and me is going.
Henry reports that her informants found sentences of type (37c)
completely ungrammatical to a degree which, as she argues, excludes the
explanation that the ungrammaticality judgment could be due only to a
sociolinguistic mismatch with singular concord being a non-standard
feature, and thus sounding strange when placed together with the formal,
prestige form involving co-ordinate pronouns in the nominative case
(1995: 24). Consequently, Henry makes the nominative constraint the
focus of her analysis.
184 Lukas Pietsch
In the principles-and-parameters framework adopted by Henry,
agreement phenomena are assumed universally to involve checking in a
specifier-head configuration between the two elements. Therefore, the
question of agreement or non-agreement logically leads to the question
whether the subject is in fact in SPEC/AGR
S
P in this structure (1995:
21). Henry discusses, and rejects, an analysis according to which subjects
of singular concord clauses (those with non-standard verbal -s) would
remain in VP, while those of standard agreement clauses would raise to
SPEC/AGR
S
P. Her own proposal is similar but involves a different set of
nodes in the tree: while subjects and verbs of standard agreement clauses
raise to SPEC/AGR
S
P and AGR
S
respectively, just as in Standard English,
subjects and verbs of singular concord clauses raise only to one node
lower in the tree, SPEC/TP and T (tense) respectively. This reflects the
intuition that the generalized verbal -s of the singular-concord clauses is a
mere tense marker, and that only tense features but no agreement features
are involved in this position. Following common assumptions in generative
work on English, the movement of the verb is supposed to happen overtly
(at spell-out) only with the raising verbs be and have, while all other
verbs undergo the raising operation only at the level of LF. As for the
subject, the T node is also assumed to be unable to assign nominative case
to the subject in its specifier position, as this is an exclusive property of
AGR
S
. However, Henry stipulates that T in this dialect has the property of
being able to assign non-nominative or default case. Therefore, all
subjects except those morphologically marked for nominative can appear in
this position, and this explains the nominative constraint. The question of
why and when the elements are forced to raise to either of the two target
positions is handled, following the preferred mode of thought in recent
generative work, by assuming that features are either weak or strong. In
this case, the NP features of AGR
S
can optionally be either weak or strong
in the dialect, whereas in Standard English they are always strong.
One argument Henry adduces for her analysis (1995: 29) is the
behaviour of negative-polarity items (NPIs) in this dialect, which she
claims to be parametrically linked to the verbal concord behaviour. She
finds that non-standard structures such as (38) are possible in Belfast
English:
(38) a. Anybody wouldnt be able to do it.
b. I was surprised that anybody didnt go.
However, in the plural such negative-polarity subjects can only occur if
there also is singular agreement:
(39) a. Any friends isnt coming.
b. *Any friends arent coming.
Following the generative tradition, Henry explains the licensing of
negative-polarity items in terms of their being c-commanded by a raised
negative operator on the level of LF. Her point is that for theory-internal
reasons raising of the negative element above the negative-polarity subject
is easily conceivable if the subject is in SPEC/TP, but not so easily if it is in
SPEC/AGR
S
P. Again Henry seems to exclude the possibility that the
perceived co-occurrence restriction between singular concord and negative-
polarity subjects could be due simply to stylistic mismatches of elements
from different registers:
That there is a link between singular concord and NPI [i.e. Negative
Polarity Item] licensing in subject position seems clear from the the [sic]
fact that the two phenomena seem to go together in speakers grammars;
those speakers who allow singular concord also permit NPIs in subject
position, and conversely non-users of singular concord find NPIs in that
position strongly ungrammatical. This is a clear case where careful
examination of dialects or sub-dialects can help to show whether proposed
connections are real or not, and where it is important to check what co-
occurrence constraints there are on dialect features. (1995: 2930)
It is not my intention to comment on the technical, theory-internal merits of
Henrys proposal. While it is the most comprehensive and probably the
empirically best argued formal analysis so far, it must be pointed out here
that it fails to provide a unified analysis of the whole range of
(contemporary or older) concord systems of the northern type. The
presumed parametrical link between the Northern Subject Rule and the
negative-polarity scope behaviour is just such a case where more careful
examination of dialects or sub-dialects (to use Henrys own words) would
have been a welcome addition: Henry makes no attempt at establishing
whether the two phenomena do indeed go together in other related varieties
outside Belfast. A similar and even stronger argument holds for the
nominative constraint. Even if that rule is descriptively adequate for those
Type-of-Subject effects which Henry finds in her data, there does not seem
to be a straightforward way how her analysis should carry over to varieties
where the Position-of-Subject Constraint played a more prominent role.
186 Lukas Pietsch
Such varieties, after all, do allow overtly nominative pronouns to co-occur
with verbal -s, in non-adjacent positions. It is difficult to see how Henrys
analysis could be adapted to account for varieties of this type, even by
positing a couple of additional parameter settings. This, of course, does not
logically invalidate Henrys analysis as such. It may in principle be the case
that present-day Belfast English differs fundamentally from northern
systems in other (or older) dialects; in that case, Henrys analysis of the
former would stand but explanations along rather different lines would
have to be sought for the latter. However, an argument in the other
direction holds. If there was an explanation that could account for the
classic older systems with their stronger Position-of-Subject effects, such
as Older Scots or late northern Middle English, then this explanation would
very likely be able to account also for the more restricted range of
nonstandard verbal -s in Belfast English in a unified way. Such a universal
explanation would then certainly be preferable to the parochial one Henry
proposes for Belfast English.
5.2. Brjars and Chapman (1998)
As we saw in the last section, at the heart of Henrys proposal is the idea
that the collocations of pronouns and affixless verb forms represent real
agreement whereas the constructions with the -s form do not. This is
entailed by the assumption that the former but not the latter involve
checking of identically marked features of pronoun and verb in a Specifier-
Head constellation under AGR
S
P. The proposal by Brjars and Chapman
(1998) implies the exact opposite. They interpret the co-occurrence
restriction on pronouns and affixed verbs in terms that might be described
as anti-agreement (in the terms of Corrigan 1997: 203, quoting Roberts
1997: 109), that is, the avoidance of double marking of some feature. In
this perspective, then, it is the -s forms which represent the real agreement
marking, and the - forms represent the lack of such. Obviously, the
feasibility of these two diametrically opposed approaches is a consequence
of the markedness paradox as described earlier. Yet, none of the authors
involved seems to be aware of the possibility of the opposite view.
Brjars and Chapman (1998) discuss the northern concord system from
the perspective of Lexical-Functional Grammar (LFG). They propose a
formal model based on the intuition, first developed in Chapman (1998),
that pronoun subjects and adjacent verbs stand in a closer syntactic relation
with each other than other subjects and verbs, and that together they form
an integrated unit in syntactic processing in the words of Chapman (1998:
39): a syntactic unit which is interrupted if additional information is added
in the form of a second pronoun, for example, or some sort of modifier
between the pronoun and verb. This proposal, then, differs from the
preceding one in explicitly focussing on the effects of the Position-of-
Subject Constraint, as illustrated in (40).
(40) a. They often talks.
b. They talk.
When seeking a formal explanation for the contrast in (40), one must resort
to either of two principal strategies: either posit that the two instances of
they in (40a) and (40b) are in two different structural positions in the
syntactic tree (which would not be the case in the corresponding sentences
in Standard English), or posit that the two instances of they are somehow
not the same thing. The analysis proposed by Brjars and Chapman
combines both these options. As for the tree positions, they choose an
analysis essentially opposite to Henrys: whereas Henry assumes the
subjects of the -s verbs to be structurally closer to the (original) position of
the verb in the tree configuration, namely the T node instead of the Agr
S
node, Brjars and Chapman ascribe a closer position to the subjects of the
affixless verb forms. As for categorial status, they assume the existence of
two partly homonymous sets of pronouns with different roles: those which
co-occur with an -s ending on the verb: he, she, it, I
1
, you
1
, we
1
, they
1
; and
those that occur without an -s ending: I
2
, you
2
, we
2
, they
2
. The central point
of their proposal is that this latter set has the categorial status of verb
inflections, more exactly: pronominal inflections they are really not
part of the syntax but part of the verb morphology (1998: 76, 83). In what
follows I will adopt the authors convention of writing these constructions
with hyphens to indicate this presumed affixal status: they=talk.
Brjars and Chapman go on to specify the necessary properties of the
two sets of pronouns in terms of feature settings in the formal mechanism
of Lexical-Functional Grammar. These properties entail the three most
important aspects of their behaviour: that the second set must always be
adjacent to the verb, that it cannot co-occur with another competing
inflectional affix on the other side of the verb stem (namely, the -s affix)
and that neither of the two sets can co-occur with each other or with
another overt subject. In sentences with pronominal inflection, the affixal
188 Lukas Pietsch
pronoun is generated under the V node, and these sentences therefore lack a
subject in the canonical subject position, the sister of the VP. They are
therefore, technically speaking, pro-drop structures. This generates the
following grammatical structures:
(41) a. [
NP
they
1
] [
VP
often [
V
talks]]
b. [
NP
e ] [
VP
[
V
they
2
=talk]]
Structure (42), on the other hand, containing both the -s ending and a
pronominal inflection they
2
, is correctly predicted to be ungrammatical:
(42) *[
NP
e ] [
VP
[
V
they
2
=talks]]
There are various problems with this analysis. First, as pointed out by
Shorrocks (1999: 112), the assumption of affixal or clitic status intuitively
conflicts with a central property usually connected with affixes (or clitics,
for that matter): phonological dependence. While it may be true that
preverbal pronouns in connected speech often tend to be realized as
phonologically reduced forms (Brjars and Chapman 1998: 97), these
pronouns can nevertheless also take contrastive stress, and will then still
occur with the non-inflected verb form in NSR dialects. A second problem
concerns the descriptive adequacy of the proposal with respect to the
variability of the system. It seems somewhat unfortunate in this respect that
the authors base their analysis as they admit themselves on rather
sketchy and incomplete data, collected from some cursory descriptions of
concord variation in the dialectological literature. This is particularly
serious since the Position-of-Subject Constraint features so prominently in
their discussion. As mentioned earlier, the effects of the Position-of-Subject
Constraint in most modern dialects are rather marginal. Brjars and
Chapmans proposal predicts, incorrectly, that the presence of -s after
intervening adverbs should be categorical, i.e. that they often talk should
actually be ungrammatical but it certainly is not, in any of the modern
varieties I have looked at. Even if one concedes that the authors goal is
only to describe an idealized system and that their generalizations may
hold only for a subset of subjects or a subset of verbs in each dialect
(1998: 75), this constitutes a weakness of the proposal that cannot easily be
glossed over.
Moreover, problems may remain even within the proposed formal
mechanism itself. To name but the two most important ones: first, I fail to
see how the model, in and of itself, accounts for even as much as the
ungrammaticality of *they talks. After all, instead of the ungrammatical
structure in (42) above, repeated here as (43b), one could just as well
construe this surface string by means of the structure in (43c), which does
not seem to violate any of the principles posited by the authors.
(43) a. *they talks
b. *[
NP
e ] [
VP
[
V
they
2
=talks]]
c. (?) [
NP
they
1
] [
VP
[
V
talks]]
The second problem stems from the, by now well-known, markedness
paradox inherent in the northern system. As stated earlier, the proposal is
based on the assumption that the affixless verb forms are unmarked in
terms of agreement features. This may work for lexical verbs, but it hardly
works for the forms of be. Forms like amand are following the adjacent
pronouns cannot easily be regarded as non-inflected in the same way as e.g.
talk can. This is particularly obvious with I=am, where identical person-
number information is clearly encoded twice, in just the way the LFG
mechanism is designed to preclude.
5.3. Hudson (1999)
The most recent proposal for a formal analysis of the northern concord
system is only a cursory, tentative sketch within a larger discussion of
verbal concord in Standard English, found in Hudson (1999: 204). Hudson
puts forward an analysis within his own brand of formal dependency
grammar, called Word Grammar (cf. Hudson 2001). As his contribution is
not centrally concerned with the northern dialects, the descriptive
information on which his account of them is based is even sketchier than in
the case of Brjars and Chapman the only authority cited is a passage in
Harris (1993). Nevertheless, it may be argued that his proposal constitutes a
step forward over those discussed up to now, as it avoids the
methodological imperative common to most branches of formal grammar,
of having to deduce observed grammatical behaviour from a small set of
postulated universal, innate elements. In Word Grammar, features and
syntactic categories can be language-specific and can be freely assigned by
ad-hoc rules. Words and grammatical entities form a taxonomic hierarchy
(a network of so-called isa relations), in which all elements can either
190 Lukas Pietsch
inherit default properties from their superordinate nodes or override these
by specific rules. To account for some special cases in the agreement
behaviour of subject noun phrases in Standard English, Hudson introduces
a novel ad-hoc feature of agreement-number that exists side by side with
the traditional number feature. Both are linked through the default rule of
identity, but may differ from each other in special cases (e.g., the noun
family may have singular number but, by way of exception, plural
agreement-number). A similar mechanism of exceptional feature
assignments specified for individual words is used to account for the
behaviour of the pronouns I and you, and of the verb be.
In this framework, the formal description of the Type-of-Subject
Constraint becomes surprisingly simple: in the northern dialects, all nouns
by default have either singular agreement-number features or no
agreement-number at all, and only the pronouns I, you, we, they have an
exceptional plural agreement-number. The main problem with this
proposal seems to be that it does not provide for a way to deal with the
Position-of-Subject Constraint, of which Hudson does not seem to be
aware. As all the formal distinctions in this model are associated with
individual items in the lexicon, there seems to be no easy way to account
for differences in behaviour that are triggered only by accidental syntactic
environment factors other than the choice of the verb or subject themselves.
6. Discussion: Variation and usage-based theories
As shown above, all the formal analyses that have been proposed to
account for the northern concord systems so far have run into problems of
descriptive adequacy. None of them is quite successful even with respect to
their primary aim: descriptively integrating the conditioning factors of the
Northern Subject Rule under some unified principle. The combinatory
effect of the Type-of-Subject Constraint and the Position-of-Subject
Constraint, this curious amalgamation of lexical and syntactical conditions,
apparently resists such generalizations. Moreover, none of the proposals
even attempts to deal with those conditioning factors that affect the choice
of form in the modern dialects, over and above the core defining constraints
of the Northern Subject Rule itself: the preference for relative clauses, the
effects of indefinite and demonstrative pronouns, of is/are and was/were,
and so on. None of the proposals addresses the question of the quantitative,
probabilistic patterns of variation observed with respect to these latter
factors. In addition, none of the proposals deals with the full range of
diatopic variation and the various geographical sub-types of the northern
system, especially with the cases of overlap or competition between the
Northern Subject Rule and the other, peripheral patterns of concord
variation, such as was/were levelling. And finally, none of the proposals
can easily deal with the diachrony of the Northern Subject Rule, with its
apparently gradual processes of spread and diffusion. In short, these formalist
accounts cannot fully account for the northern concord system, either in its
reconstructed nascent state of early Middle English, or in its ideal form in
the older northern dialects, or indeed in its highly variable forms affected by
dialect levelling and dialect attrition, in the present-day varieties.
In this discussion, I will take for granted one thing that may sound like a
truism in the context of variationist and sociolinguistic studies, but which in
the context of formal grammatical theory requires some justification: that
the morphosyntactic variation observed in the dialectal corpora is indeed of
the type that has come to be called inherent variation. It cannot be
reduced to variation between speakers with different grammars, nor to
effects of code-switching between competing, distinct grammatical systems
within the production of individuals. The choice between one form and the
other within the performance of each individual speaker will vary
unpredictably, at any point in discourse, often even within a single sentence
as in the example that forms the title of this chapter (44):
(44) Some do and some doesnt. [SED: Y10]
Variation of this type can only be described with the help of stochastic
models. Even though it may be intuitively evident that much of the
observed patterns in the present case can be described in terms of a
competition between two distinct rule systems the rules of the ideal
northern system on the one hand, and those of Standard English on the
other such an account may in effect be no more than just a convenient
descriptive abstraction. There is little evidence in the behaviour of speakers
that these competing grammars are represented as distinct, separate
entities and that any wholesale switching between them is involved in
actual production. A description involving competing individual rules
within one unified grammatical system seems more plausible.
The idea that quantitative variability may need to be taken into account
as part of core syntax has recently been acknowledged by Henry (2002),
a rare exception for an author coming from a generative research tradition.
192 Lukas Pietsch
Note, incidentally, that theoretical attempts at integrating statements of
competing grammatical regularities with statements about the probability of
their respective use basically in a fashion similar to the Variable-Rules
Theory of the Labovian school in the 1970s have recently also been
brought forward again in the shape of Stochastic Optimality Theory
(Boersma 1998, 2000; Boersma and Hayes 2001; Bresnan and Deo 2001).
While probabilistic variation has often been regarded as an awkward
problem in formalist theories of grammar, functionalist and emergentist
theories have for a long time seen inherent variation as a natural and
necessary part of linguistic knowledge. Quantitative variation is easy to
accommodate for any theory that accepts the idea that grammars are not
stable from the completion of language acquisition in childhood, but can
continuously change, as a result of the language encountered throughout
the lifetime of a speaker. In this process, a central role is commonly
attributed to the relative frequencies with which different linguistic
structures are encountered in discourse (Krug 2003). Different quantitative
exposure to linguistic structures in discourse is directly mirrored by
quantitatively different representation of these structures in linguistic
knowledge. The principal cognitive mechanism invoked in modelling
frequency effects on mental representation is entrenchment, strengthening
of representations through their activation in use (Langacker 1987: 59).
I would like to suggest that frequency-based theories are highly relevant
not only in accounting for the state of variation observed in the present-day
dialects, but also in accounting for the historical process that led to the
emergence of the Northern Subject Rule in the first place. At the very heart
of the northern concord pattern are constructions whose principal
distinguishing property is their high discourse frequency: combinations of
personal pronouns and finite verbs. Pronouns have, by their very nature, a
much higher discourse (token) frequency than any lexical noun has. The
relatively much higher degree of entrenchment and routinization that must
be involved in the processing of recurrent collocations of adjacent pronoun
subjects and their verbs as opposed to combinations involving lexical
noun subjects can plausibly be seen as causally involved in any
diachronic tendency that leads to morphological reduction or irregularity
associated with these constructions. As was explained in section 4, it was
exactly such a process that seems to have provided one of the main triggers
for the emergence of the Northern Subject Rule in early Middle English:
the phonological reduction of the verbal inflexional suffixes in the high-
frequency environment with adjacent clitic pronouns. In this perspective,
the emergence of the Northern Subject Rule can be seen as a case of
morphosyntactic irregularity associated with a high-frequency construction,
induced by mental routinization.
We can finally turn again to the question of how the present-day
northern concord systems can best be characterized descriptively, with their
curious sets of multi-level conditioning factors (lexical, syntactic,
idiomatic, semantic). I would like to suggest that a usage-based form of
construction grammar (Croft 2001, Kemmer and Israel 1994, Langacker
1987) may offer the most natural account. A central tenet of all brands of
construction grammar is the continuity between lexicon and syntax.
Properties of grammatical constructions, on all levels of abstraction, are
assumed to be represented in essentially the same way as idiosyncratic
properties of lexical items. Constructions form a multiple taxonomy,
ranging from highly schematic to highly specific, where the more specific
constructions can inherit but also override the grammatical behaviour
specified in the more general schemata of which they are instantiations.
The behaviour of constructions at all levels is non-universal and arbitrary,
which means that there is no need to search for explanations in the sense
of the highly abstract generalizations common in the microparametric
syntax of the Chomskyan tradition. Applied to subject-verb agreement in
English, such an approach can easily describe, for instance, the
combination they+verb as a construction that instantiates the more general
schema subject+verb, and which, while sharing some or most of its
properties with this more general construction by default, may also partly
differ from it in some arbitrary way, for instance with respect to its concord
behaviour. Such a construction-based approach is conceptually similar
though not formally identical to that sketched out by Hudson (1999) in the
framework of Word Grammar. Word Grammar too makes extensive use of
hierarchies of more general and more specific rules providing defaults and
overrides, but locates these in the lexical entries of specific words, not of
constructions as such. I would like to suggest that the construction-based
description can give a more natural account of those cases where identical
items show different behaviour determined only by the word-order
constellation they occur in, as it is the case with the Position-of-Subject
Constraint.
Following Kemmer and Israel (1994: 165171), I shall assume that
different construction schemata can compete with each other during
production of an utterance, and that variability in a speakers production
can be explained by this competition. The probabilistic results of this
194 Lukas Pietsch
competition provide the basis for a natural account of the sort of structured
variability traditionally modelled by variable rules (Kemmer and Israel
1994: 165). The more heavily entrenched a constructional schema is in
memory, the higher is its probability of being selected as the relevant
categorization unit for the production of a specific usage event.
The following will give a short and tentative sketch of how this kind of
competition might be modelled. Consider first a schematic representation of
the agreement system of Standard English (see Figure 2). From the
perspective of a maximally parsimonious grammar, a minimum of only
three construction schemata, i.e. nodes in the taxonomic hierarchy of stored
grammatical representations, need to be posited (ignoring, for the moment,
the special cases of be and that of the non-concord verb forms). One is a
maximally abstract schema specifying the relationship of subject and verb
in general. It has two daughter nodes, one for third person singular subjects,
specifying the use of the verbal -s morphology, and one for all other
combinations, specifying the use of the base forms of the verbs. In a usage-
based account, it may be further assumed that some additional, subordinate
nodes might exist for more specific, frequent combinations, involving for
instance pronoun subjects and/or high-frequency verb forms (shown in
dotted lines in Figure 2). These may be stored and processed as separate
units, having attained unit status through their entrenchment as high-
frequency combinations, but they are not associated with any observably
different formal behaviour and therefore formally redundant.
Compare this system with that of a northern dialect, considering first the
case of a hypothetical, pure system in which the Northern Subject Rule is
categorical throughout (Figure 3). Here, the structure of the inventory is
somewhat different. There must be a set of nodes representing the
exceptional behaviour of the pronouns I, we, you and they (taking the -
forms), in opposition with a more abstract schema representing the use of
the default verbal -s form for all the rest. Again, there might also be further,
more specific units representing other high-frequency combinations, but
these are formally redundant. In particular, it seems quite likely that those
pronoun combinations that take the verbal -s forms (he/she/it +V-s) may
also be stored and processed as independent units just like the other
pronouns, owing to their high frequency of occurrence.
Figure 2. A construction inventory for Standard English agreement
Figure 3. A construction inventory for NSR agreement
We can now turn to the representations needed in the case of a hybrid
dialect involving variation between standard-conforming and agreement of
the northern type. I assume that each of the formal options available to a
speaker of such a dialect is represented by a construction schema of its
own, and that these schemata can compete with each other in production.
Subj Verb
Subj
+3Sg
V-s
...
it V-s
she V-s
he V-s
...
Subj
+3Sg
has
Subj
+3Sg
was
Subj
+3Sg
is
Subj
3Sg
V-
...
they V-
you V-
we V-
I V-
...
Subj
3Sg
have
Subj
3Sg
were
Subj
3Sg
are
Subj Verb
Subj V-s
it V-s
she V-s
he V-s NP V-s
they V-
we V-
you V-
I V-
196 Lukas Pietsch
Figure 4 shows a partial network structure of the construction inventory
needed for such a system. The schemata NP V-s (from the northern system)
and NP
Pl
V- (from the standard system) will compete with each other.
Any usage event of a particular subject and a particular verb will be
processed as instantiating either of these two, whichever happens to be
more salient when the construction is being processed. Depending on which
of the two is more deeply entrenched, either the one or the other will have
the higher likelihood of winning out.
Figure 4. A construction inventory for variable NSR agreement
The structured variation effects of the type captured by variable-rules
models can be integrated in a system of this kind by again assuming
further, subordinate nodes in the network. In a construction grammar
approach, separate representation as a unit in the structured inventory of a
speakers knowledge is commonly ascribed to [a]ny construction with
unique, idiosyncratic morphological, syntactic, lexical, semantic,
pragmatic, or discourse-functional properties (Croft 2001: 25; emphasis in
original). In order to capture a system of structured, inherent variation, a
further factor can be added to this list: any type of construction that is
associated with distinct, quantifiable behaviour with respect to a linguistic
variable must also have unit status and be separately represented. In this
way, nodes in the construction network may correspond to what would be
modelled as factors in a variable-rule model.
Figure 5 shows a partial network model of such a system. Both the
competing nodes from Figure 4 (NP V-s and NP
Pl
V-) are here shown as
having a number of daughter nodes. The subnodes shown represent
schemata associated with individual verb forms. Yet others, not shown
here, might represent different types of subjects, and so on. Each of the
Subj Verb
NP V-s
it V-s
she V-s
he V-s
they V-
we V-
you V-
I V-
NP
Sg
V-s NP
Pl
V-
schemata represents either a standard-conformant or an NSR-conformant
use. Each of them may be thought of as characterized by its individual
degree of strength of entrenchment (symbolized in the graph by boxes of
different thickness). During production, there will be a pairwise
competition between relevant nodes. In this way, a situation can be
modelled in which, for instance, plural use of is as opposed to are is
relatively more likely than plural use of goes as opposed to go, just as it
was found in the corpus data from many northern dialects.
Figure 5. A partial construction inventory for structured variation in NSR
agreement
In construction grammar, it is also commonly assumed that units in the
taxonomic network of constructions can have multiple parents, and
individual usage events can instantiate more than one construction
schemata simultaneously. Each of the schemata activated for the production
of any individual utterance will be responsible for a different aspect of its
formal behaviour. Transferring this idea to the modelling of structured
variation, it may be assumed that more than one set of competing nodes
may be activated simultaneously in the production of an utterance, in such
a way that each of them represents a different conditioning factor. For
instance, it was shown in the previous sections that in many northern
dialects the choice between the competing sentences themis going and
themare going is influenced by two independent conditioning factors
(represented as members of two factor groups in a variable-rules model):
a relatively stronger preference for is over are as compared with other
verbal -s forms; and a factor favouring the use of verbal -swith the specific
Subj Verb
NP V-s
NP
Pl
V-
NP is NP was NP has NP goes
NP
Pl
are NP
Pl
were NP
Pl
have NP
Pl
go
198 Lukas Pietsch
item themas a subject. Such combinatorial effects of several independent
conditioning factors can be modelled in a construction grammar approach
by assuming that the production of a sentence such as themis going
involves the simultaneous competition between a pair of schemata NP +is
and NP
Pl
+are, as shown in Figure 5; and between a second pair of
schemata, them+V- and them+V-s.
A usage and schema based approach such as this can account in a
natural way for the existence of the two types of constraints on subject-verb
agreement that have been reported for many varieties of English, not only
those that are directly related to the northern ones: Type-of-Subject and
Position-of-Subject constraints. I hypothesize that a Type-of-Subject
constraint may arise whenever combinations of verbs and personal
pronouns, owing to their high discourse frequency, are entrenched
separately and attain unit status. A Position-of-Subject constraint will be a
corollary of the fact that both the more specific and the more general
schema may compete with each other in the production of a usage event. I
hypothesize that the more specific schema in memory a gestalt consisting
of a particular pronoun and a verb will be more salient, and hence more
likely to be activated as the relevant categorizing unit, if the utterance that
is being formed involves a direct collocation of the pronoun and the verb. If
both items are not adjacent, the construction less closely matches the
gestalt prototype of the stored schema. It will then be less likely to be
categorized as an instance of that specific schema, and by default the more
abstract schema will be more likely to win out as the relevant categorizing
unit.
I also hypothesize that the emergence of a Type-of-Subject Constraint is
particularly likely in a situation where a verbal agreement paradigm has
previously undergone heavy erosion and reduction. As was shown in
section 4, such a state of affairs held in the historical situation in which the
Northern Subject Rule first emerged in northern early Middle English. The
present tense verbal system had been reduced, by a series of changes, to
only one formal opposition between -(e) and -(e)s, so that only the first
person singular was reliably distinguished from all the rest of the paradigm.
If language learners are confronted with such a heavily eroded
typologically and functionally implausible agreement system, then a
distinction such as that between pronominal and non-pronominal subjects
may become cognitively more salient in processing than the person-number
distinction, and a re-structuring of the agreement system along these
dimensions may be the long-term consequence. It may also be argued that a
similar state of affairs has again been reached, in a different form, in
Modern Standard English, where only the third singular is distinguished
from all the rest of the paradigm (and where, in addition, the contrast
betweenwas and were patterns according to yet a different rule, contrasting
first or third singular as against the rest). This may be a reason why
variation phenomena resembling the northern system in some respects have
emerged, independently as it appears, in some modern varieties outside the
north.
With the foregoing discussion, I hope to have provided some basis for
the claim that a usage-based, cognitive approach to grammatical theory can
and should be fruitfully combined with variationist research using methods
from the Labovian sociolinguistic tradition. Such a combination has
repeatedly been envisaged in recent work. However, while sociolinguistic
research has dealt descriptively with phenomena such as verbal concord
variation in English quite thoroughly, much of the more theoretically
oriented work has so far been restricted to more narrowly local phenomena
of grammatical organization, such as inflectional paradigms, phonological
reduction phenomena in specific word chunks, and so forth. It seems fair to
say that this line of research has been somewhat reluctant to tackle
questions of variation in the domain of syntax proper (cf. Bybee and
Hopper 2001). The morphosyntax of agreement would appear to be a
promising domain for an extension of such studies. Especially with respect
to a language whose social and geographical variation is so well
documented as English, there is still much fruitful ground for further
research aiming to bridge the gap between theory and description.
Acknowledgments
I would like to thank all organisations and individuals who contributed the
speech recordings which were used in FRED and form much of the
empirical basis of the present article, especially Roger Leitch, J ane Petrie,
the Ambleside Oral History Group, Pennine Heritage (Hebden Bridge) and
the Oral History Group at Huyton (Merseyside). Thanks are also due to
J ohn Kirk for providing access to the NITCS, to him and Raymond Hickey
for their efforts at recovering material of the TRS, and to Clive Upton for
assistance with the unpublished material of the SED. I would also like to
thank Kevin McCafferty, Sali Tagliamonte, Peter Auer, and Ingo Plag for
valuable feedback and advice during the preparation of this work, and
200 Lukas Pietsch
Anthea Gupta, Clive Upton, Esther Asprey and Kate Wallace for special
assistance with transcription problems.
Notes
1. For the purposes of this study, the term concord verb refers to all verbal
forms which have the morphological potential of displaying an agreement
contrast. This includes all finite present tense verbs except for modals and
subjunctives, plus the past tense forms was and were. The term verbal -s is
used to include, where not otherwise specified, the irregular forms is and was.
2. Following the usage in the SED and much of the dialectological literature, I
am using the historical (pre-1974) counties as a geographical reference frame.
3. Designations such as Central North, Lower North etc. mirror the well-
known, partly homonymous labels introduced by Trudgill (1990) for the
modern dialect areas, but are not intended as equivalent to them. As used
throughout this chapter, these terms designate areas defined exclusively in
terms of the subject-verb agreement data in the SED.
4. Three isolated tokens of mforms were also found in the incidental material
in southwestern Yorkshire, but they can all be explained as products of a mere
phonetic assimilation of han before labial consonants.
5. The examples from the NITCS include some tokens where the second verb is
was or is. Since there is otherwise hardly any variation involving these verbs
with these pronoun subjects in the Ulster data, these tokens can safely be
attributed to the Northern Subject Rule together with those involving lexical
verbs.
6. See, for instance, the common extension of the idiom to other verbs of
saying, such as the modern vernacular I goes, attested for instance for
Glasgow in Macafee (1983).
7. Owing to the different data collection habits of certain fieldworkers, it was
necessary to use different weighting criteria for tokens recorded by one of
them, S. Ellis, and those recorded by the rest. On average, Ellis tended to
record only about half as many tokens per location than Wright did in the
same geographical area. Locations whose fieldworker notebooks were not
consulted are excluded in this map.
8. The examples from Wright (1892) are quoted after Chapman (1998: 38).
References
Aissen, J udith, and J oan Bresnan
2002 Optimality and functionality: objections and refutations. Natural
Language and Linguistic Theory 20: 4380.
Anderwald, Lieselotte
2002 Negation in Non-Standard British English: Gaps, Regularizations
and Asymmetries. London: Routledge.
Bailey, Charles, and Karl Maroldt
1977 The French lineage of English. In Langues en Contact Pidgins
Creoles Languages in Contact, J rgen M. Meisel (ed.), 2153.
Tbingen: Narr.
Bailey, Guy, and Garry Ross
1988 The shape of the superstrate: morphosyntactic features of Ship
English. English World-Wide 9: 193212.
Bailey, Guy, Natalie Maynor, and Patricia Cukor-Avila
1989 Variation in subject-verb concord in Early Modern English.
Language Variation and Change 1: 285300.
Barbiers, Sjef, Leonie Cornips, and Susanne van der Kleij (eds.)
2002 Syntactic Microvariation. http://www.meertens.nl/books/synmic/.
Barry, Michael V.
1981a The southern boundaries of Northern Hiberno-English speech. In
Aspects of the English Dialects in Ireland, Michael V. Barry (ed.),
5295. Belfast: Queens University, Institute of Irish Studies.
1981b The methodology of the tape-recorded survey of Hiberno-English
speech. In Aspects of the English Dialects in Ireland, Michael V.
Barry (ed.), 1846. Belfast: Queens University, Institute of Irish
Studies.
Beal, J oan
1993 The grammar of Tyneside and Northumbrian English. In Real
English: The Grammar of English Dialects in the British Isles,
J ames Milroy, and Lesley Milroy (eds.), 187213. London:
Longman.
1997 Syntax and morphology. In The Edinburgh History of the Scots
Language, Charles J ones (ed.), 335377. Edinburgh: University
Press.
2004 The morphology and syntax of English dialects in the North of
England. In Varieties of English: The British Isles, Bernd Kortmann,
and Clive Upton (eds.). Berlin: Mouton de Gruyter.
Berndt, Rolf
1956 Formund Funktion des Verbums imnrdlichen Sptaltenglischen:
eine Untersuchung der grammatischen Formen und ihrer
202 Lukas Pietsch
syntaktischen Beziehungsbedeutungen in der groen sprachlichen
Umbruchsperiode. Halle: Niemeyer.
1982 A History of the English Language. Leipzig: VEB Verlag Enzyklo-
pdie.
Black, J ames R., and Virginia Motapanyane (eds.)
1996 Microparametric Syntax and Dialect Variation. Amsterdam: Ben-
jamins.
Blakeley, Leslie
1949 The Lindisfarne s/ problem. Studia Neophilologica 22: 1547.
Boersma, Paul
1998 Functional Phonology: Formalizing the Interactions between
Articulatory and Perceptual Drives. The Hague: Holland Academic
Graphics.
Boersma, Paul, and Bruce Hayes
2001 Empirical Tests of the Gradual Learning Algorithm. Linguistic
Inquiry 32: 4586.
Brjars, Kersti, and Carol Chapman
1998 Agreement and pro-drop in some dialects of English. Linguistics 36:
7198.
Breivik, Leiv Egil, and Toril Swan
2000 The desemanticisation of existential there. In Words: Structure,
Meaning, Function, Christiane Dalton-Puffer, and Nikolaus Ritt
(eds.), 1934. Berlin: Mouton de Gruyter.
Bresnan, J oan, and Ashwin Deo
2001 Grammatical constraints on variation: Be in the Survey of English
Dialects and (Stochastic) Optimality Theory. Unpubl. Ms., Stanford
University.
Britain, David
2002 Diffusion, levelling, simplification, and reallocation in past tense be
in the English Fens. J ournal of Sociolinguistics 6: 1643.
Brunner, Karl
1962 Die englische Sprache, Vol. 2. Tbingen: Niemeyer.
Brunner, Karl, and Eduard Sievers
1965 Altenglische Grammatik. 3rd edition. Tbingen: Niemeyer.
Bybee, J oan
1985 Morphology: A Study of the Relation between Formand Meaning.
Amsterdam: Benjamins.
Bybee, J oan, and Paul Hopper
2001 Introduction. In Bybee and Hopper (eds.), 126.
Bybee, J oan, and Paul Hopper (eds.)
2001 Frequency and the Emergence of Linguistic Structure. Amsterdam:
Benjamins.
Chambers, J ack K.
2004 Dynamic typology and vernacular universals. In Dialectology Meets
Typology, Bernd Kortmann (ed.). Berlin: Mouton de Gruyter.
Chapman, Carol
1998 A subject-verb agreement hierarchy: evidence from analogical
change in modern English dialects. In Current Issues in Linguistic
Theory: Historical Linguistics 1995. Vol. 2, Richard Hogg, and L.
van Bergen (eds.), 3544. Amsterdam: Benjamins.
Clark, Brady Z.
2003 On stochastic grammar. Unpubl. ms., University of Stanford.
http://www.stanford.edu/~bzack/papers/stotgrammars.pdf.
Corrigan, Karen P.
1997 The syntax of South Armagh English in its socio-historical
perspective. PhD thesis, University College Dublin.
Croft, William
1995 Autonomy and functionalist linguistics. Language71: 490532.
2001 Radical Construction Grammar. Oxford: Oxford University Press.
Ferguson, Charles A.
1996 Variation and drift: loss of agreement in Germanic. In Towards a
Social Science of Language: Papers in Honor of WilliamLabov.
Vol. I: Variation and Change in Language and Society, Gregory
Guy (ed.), 173198. Amsterdam: Benjamins.
Filppula, Markku
1999 The Grammar of Irish English: Language in Hibernian Style.
London: Routledge.
Godfrey, Elizabeth, and Sali Tagliamonte
1999 Another piece for the verbal -s story: evidence from Devon in
southwest England. Language Variation and Change 11: 87121.
Grlach, Manfred
2002 A Textual History of Scots. Heidelberg: Winter.
Gregg, Robert J .
1972 The Scotch-Irish dialect boundaries in Ulster. In Patterns in the Folk
Speech of the British Isles, Martin F. Wakelin (ed.), 109139.
London: Athlone Press.
Harris, J ohn
1991 Ireland. In English around the World: Sociolinguistic Perspectives,
Cheshire, J enny (ed.), 3750. Cambridge: Cambridge University
Press.
1993 The grammar of Irish English. In Real English: The Grammar of
English Dialects in the British Isles, J ames Milroy, and Lesley
Milroy (eds.), 139186. London: Longman.
204 Lukas Pietsch
Henry, Alison
1995 Belfast English and Standard English: Dialectal Variation and
Parameter Setting. Oxford: Oxford University Press.
2002 Variation and syntactic theory. In Handbook of Language Variation
and Change, J ack K. Chambers, Peter Trudgill, and N. Schilling-
Estes (eds.), 267282. Oxford: Blackwell.
Holmqvist, Erik
1922 On the History of the English Present Inflections, Particulary -th
and -s. Heidelberg: Winter.
Hudson, Richard
1999 Subject-verb agreement in English. English Language and
Linguistics 3: 173207.
2001 Word Grammar. http://www.phon.ucl.ac.uk/home/dick/hcl.htm.
Ihalainen, Ossi
1991 On grammatical diffusion in Somerset folk speech. In Dialects of
English: Studies in Grammatical Variation, Peter Trudgill, and J ack
K. Chambers (eds.), 104119. London: Longman.
1994 The dialects of England since 1776. In The Cambridge History of the
English Language. Vol. V: English in Britain and Overseas: Origins
and Developments, Robert Burchfield (ed.), 197276. Cambridge:
Cambridge University Press.
Keller, Wolfgang
1925 Skandinavischer Einfluss in der englischen Flexion. In Probleme der
englischen Sprache und Kultur: Festschrift J ohannes Hoops, W.
Keller (ed.), 8087. Heidelberg: Winter.
Kemner, Suzanne, and Michael Israel
1994 Variation and the usage-based model. In Papers from the 30th
Regional Meeting of the Chicago Linguistics Society, Katherine
Beals (ed.), 165179. Chicago: Chicago Linguistics Society.
Kirk, J ohn M.
1991 Northern Ireland Transcribed Corpus of Speech. Vol. 1 (with S.
West and S. Gibson): Textfile. Colchester: Economic and Social
Research Council Data Archive, University of Essex, Colchester.
Klemola, J uhani
1996 Non-standard periphrastic do: a study in variation and change. PhD
thesis, University of Essex.
2000 The origins of the Northern Subject Rule: a case of early contact? In
The Celtic Englishes II, Hildegard Tristram (ed.), 329346. Heidel-
berg: Carl Winter Verlag.
Knecht, J akob
1911 Die Kongruenz zwischen Subjekt und Prdikat und die 3. Person
Pluralis Prsentis auf -s imElisabethanischen Englisch. Heidelberg:
Winter.
Krug, Manfred
2003 Frequency as a determinant in grammatical variation and change. In
Determinants of Grammatical Variation, Gnter Rohdenburg, and
Britta Mondord (eds.). Berlin: Mouton de Gruyter.
Kyt, Merja
1993 Third-person present singular verb inflection in early British and
American English. Language Variation and Change 5: 113139.
Laing, Margaret
1978 Studies in the Dialect Material of Medieval Lincolnshire. Unpub-
lished PhD dissertation, University of Edinburgh.
Langacker, Ronald W.
1987 Foundations of Cognitive Grammar, Vol. I: Theoretical Prerequi-
sites. Stanford: Stanford University Press.
Macafee, Caroline
1983 Glasgow. Amsterdam: Benjamins.
1993 A short grammar of Older Scots. Scottish Language 11/12: 1036.
1994 Traditional Dialect in the Modern World: A Glasgow Case Study.
Frankfurt: Lang.
McCafferty, Kevin
1998a Shared accents, divided speech community? Change in Northern
Ireland English. Language Variation and Change 10: 97121.
1998b Barriers to change: ethnic division and phonological innovation in
Northern Hiberno-English. English World-Wide 19: 735.
1999 (London)Derry: between Ulster and local speech class, ethnicity
and language change. In Urban Voices: Accent Studies in the British
Isles, Paul Foulkes, and Gerard Docherty (eds.), 246264. London:
Arnold.
2003 The Northern Subject Rule in Ulster: how Scots, how English?
Language Variation and Change 15: 105139.
2004 [T]hunder storms is verry dangese in this countrey they come in less
than a minnits notice...: The Northern Subject Rule in Southern
Irish English. English World-Wide 25: 5179.
McIntosh, Angus
1988 Present indicative plural forms in the later Middle English of the
North Midlands. In Middle English Dialectology: Essays on Some
Principles and Problems, Angus McIntosh (ed.), 116122.
Aberdeen: Aberdeen University Press.
206 Lukas Pietsch
McIntosh, Angus, Michael L. Samuels, and Michael Bensik
1986 A Linguistic Atlas of Late Mediaeval English. 4 vols. Aberdeen:
Aberdeen University Press.
Mather, J ames Y., and Hans-Henning Speitel
197586 The Linguistic Atlas of Scotland: Scots Section. 3 Vols. London:
Croom Helm.
Meurman-Solin, Anneli
1992 On the morphology of verbs in Middle Scots. In History of
Englishes: New Methods and Interpretations in Historical
Linguistics, Ossi I. Rissanen, Terttu Nevalainen, and Irma
Taavitsainen (eds.), 611623. Berlin, New York: Mouton de
Gruyter.
Miller, J im E.
1989 The grammar of Scottish English. In Regional Variation in British
English Syntax, J ames Milroy, and Lesley Milroy (eds.). Swindon:
Economic and Social Research Council.
1993 The grammar of Scottish English. In Real English: The Grammar of
English Dialects in the British Isles, J ames Milroy, and Lesley
Milroy (eds.), 99138. London: Longman.
2004 The morphology and syntax of Scottish English. In Varieties of
English: The British Isles, Bernd Kortmann, and Clive Upton (eds.) .
Berlin: Mouton de Gruyter.
Miller, J im E., and Keith Brown
1982 Aspects of Scottish English syntax. English World-Wide 3: 317.
Montgomery, Michael
1988 The Roots of Appalachian English. Clevedon: Multilingual Matters.
1994 The evolution of verb concord in Scots. In Studies in Scots and
Gaelic: Proceedings of the Third International Conferences on the
Languages of Scotland, Alexander Fenton, and Donald McDonald
(eds.), 8195. Edinburgh: Canongate.
Montgomery, Michael, J anet M. Fuller, and Sharon DeMarse
1993 The Black men has wives and sweet hearts (and third person plural
-s) jest like the white men: evidence for verbal -s from written
documents on 19th-century African American speech. Language
Variation and Change 5: 335357.
Murray, J ames A. H.
1873 The Dialect of the Southern Counties of Scotland. London: Philo-
logical Society.
Mustanoja, Tauno F.
1960 A Middle English Syntax. Helsinki: Socit Nophilologique.
Newmeyer, Frederick J .
2000 Language Formand Language Function. Cambridge/Mass.: MIT
Press.
2002a Optimality and functionality: a critique of functionally-based
Optimality Syntax. Natural Language and Linguistic Theory 20: 1
42.
2002b A rejoinder to Bresnan and Aissen. Natural Language and
Linguistic Theory 20: 8195.
2003 Grammar is grammar and usage is usage. Language 79: 682707.
Ogura, Mieko, and William S. Wang
1996 Snowball effect in lexical diffusion: the development of -s in the
third person singular present indicative in English. In English
Historical Linguistics 1994: Papers from the 8th International
Conference on English Historical Linguistics, Edinburgh 1994,
Derek Britton (ed.), 119142. Amsterdam: Benjamins.
Orton, Harold, Steward Sanderson, and J ohn Widdowson
1978 The Linguistic Atlas of England. London: Croom Helm.
Parry, David
197779 Survey of Anglo-Welsh Dialects. Vol. I: The South-East; Vol. II: The
South-West. Swansea: University College.
Peitsara, Kirsti
2002 Verbal -s in Devonshire: the Helsinki Dialect Corpus evidence. In
Variation Past and Present: VARIENG Studies on English for Terttu
Nevalainen, Helena Raumolin-Brunberg (ed.), 211230. Helsinki:
Socit Nophilologique.
Pietsch, Lukas
2003 Subject-verb agreement in northern dialects of English. PhD thesis,
University of Freiburg i. Br.
2005 Variable Grammars: Verbal Agreement in Northern Dialects of
English. Tbingen: Niemeyer.
Poplack, Shana
2001 Variability, frequency, and productivity in the irrealis domain of
French. In Bybee and Hopper (eds.), 405430.
Quirk, Sidney G., Sidney Greenbaum, Geoffrey Leech, and J an Svartvik
1991 A Comprehensive Grammar of the English Language. 9th revised
edition. London: Longman.
Roberts, Ian
1993 Verbs and Diachronic Syntax: A Comparative History of English
and French. Dordrecht: Kluver.
208 Lukas Pietsch
1997 Comparative Syntax. London: Arnold.
Robinson, Philip
1997 Ulster Scots: A Grammar of the Traditional Written and Spoken
Language. Belfast: Ullans Press.
Ross, Alan S.
1934 The origin of the s-endings of the Present Indicative in English.
J ournal of English and Germanic Philology 33: 6873.
Samuels, Michael L.
1988 The great Scandinavian belt. In Middle English Dialectology: Essays
on Some Principles and Problems, Angus McIntosh (ed.), 106115.
Aberdeen: Aberdeen University Press.
Sankoff, David, and William Labov
1979 On the uses of variable rules. Language in Society 8: 189222.
Schendl, Herbert
1996 The 3rd plural present indicative in Early Modern English: variation
and linguistic contact. In English Historical Linguistics 1994:
Papers fromthe 8th International Conference on English Historical
Linguistics, Edinburgh 1994, Derek Britton (ed.), 143160.
Amsterdam: Benjamins.
2000 The third person present plural in Shakespeares first folio: a case of
interaction of morphology and syntax? In Words: Structure,
Meaning, Function, Christiane Dalton-Puffer, and Nikolaus Ritt
(eds.), 263276. Berlin: Mouton de Gruyter.
Schreier, Daniel
2002 Past be in Tristan da Cunha: the rise and fall of categoricality in
language change. American Speech 77: 7099.
Shorrocks, Graham
1999 A Grammar of the Dialect of the Bolton Area. Vol. II: Morphology
and Syntax. Berlin: Lang.
Smith, J ennifer, and Sali Tagliamonte
1998 We was all thegither, I think we were all thegither: was regulariza-
tion in Buckie English. World Englishes 17: 105126.
Stein, Dieter
1986 Old English Northumbrian verb inflection revisited. In Linguistics
across Historical and Geographical Boundaries. Vol. 1: Linguistic
Theory and Historical Linguistics, Dieter Kastovsky, and Alexander
Szwedek (eds.), 637650. Berlin, New York, Amsterdam: Mouton
de Gruyter.
Tagliamonte, Sali
1999 Was/were variation across the generations: view from the city of
York. Language Variation and Change 10: 153191.
2002 Comparative sociolinguistics. In Handbook of Language Variation
and Change, J ack K. Chambers, Peter Trudgill, and N. Schilling-
Estes (eds.), 729763. Oxford: Blackwell.
Thomason, Sarah G., and Terrence Kaufman
1988 Language Contact, Creolization and Genetic Linguistics. Berkeley:
University of California Press.
Trudgill, Peter
1990 The Dialects of England. Oxford: Blackwell.
Van Gelderen, Elly
1997 Verbal Agreement and the Grammar behind its Breakdown:
Minimalist Feature Checking. Tbingen: Niemeyer.
Viereck, Wolfgang
1991/97 The Computer-Developed Linguistic Atlas of England. 2 vols.
Tbingen: Niemeyer.
Viereck, Wolfgang, Karin Viereck, and Heinrich Ramisch
2002 Dtv-Atlas englische Sprache. Mnchen: Deutscher Taschenbuch
Verlag.
Visser, Frederic T.
1963 An Historical Syntax of the English Language. Vol. 1. Leiden: E. J .
Brill.
Weinreich, Uriel, William Labov, and Marvin I. Herzog
1968 Empirical foundations for a theory of language change. In Directions
for Historical Linguistics: A Symposium, W. P. Lehmann, and Y.
Malkiel (eds.), 95188. Austin: University of Texas Press.
Wilson, J ohn, and Alison Henry
1998 Parameter setting within a socially realistic linguistics. Language in
Society 27: 121.
Wright, J oseph
1892 A Grammar of the Dialect of Windhill in the West Riding of
Yorkshire. London: Kegan Paul, Trench, Trbner and Co.
1905 The English Dialect Grammar. Oxford: Henry Frowde.
Wright, Laura
2002 Third person plural present tense markers in London prisoners
depositions, 15621623. American Speech 77: 242263.
Gender in English pronouns: Southwest England
Susanne Wagner
Abstract
In Dem, ees a er an ers a ee, all cept th aud Tom cat an even ees a er!
[In Devon, he is a her and she is a he, all except the old Tom cat and even he is a
her!] (Marten 1973: 9)
This saying, attributed to a number of counties in Southwest England, nicely
summarises one of the regional characteristics of the morphosyntax of
Southwestern dialects. Traditionally, these dialects employ masculine pronominal
forms (corresponding to StE he and him) to refer not only to humans and animals,
but also to count nouns. It, on the other hand, is never used for anything except
mass nouns. The feminine forms (sheand her) are rare, generally referring only to
women. This intricate system of gender assignment is made even more complex by
the presence of the form [], which is equivalent to StE him, not her.
Different from other features of non-standard morphosyntax, these gendered
pronouns do not seem to be stigmatised. This may be one of the reasons why the
feature could be successfully exported from the West Country to Newfoundland,
and could survive in these varieties to the present day. Another reason for the
relative strength of the feature today is the tendency of spoken varieties of English
in general to employ non-neuter forms in contexts where emotional involvement or
attitudes are expressed.
The interaction of at least three different systems of gender assignment is
investigated in traditional dialect data from Southwest England (Survey of English
Dialects) and more modern corpus data for both Southwest England and
Newfoundland. The factors responsible for gender assignment, it will be argued,
differ with regard to their distributional preferences in the respective varieties,
resulting in different systems of gender assignment in these varieties today.
1. I ntroduction
At a first glance, [...] it might seem that gender in modern English is a
relatively straightforward category to discuss, in comparison with the
phenomenon in many other languages; indeed, in many text-books for both
212 Susanne Wagner
native and non-native speakers of English it is barely mentioned, if at all.
(Wales 1996: 134)
Katie Wales expresses here what almost everyone, be they laymen or
linguists, native or non-native speakers, would agree to without a second
thought gender is an at most marginal category of Present-Day English
(PrDE). For centuries, it has been largely equalled (and confused) with the
biological category of sex, making it difficult to speak of gender in
English at all. However, gender assignment in English is by no means as
straightforward as grammarians describe it. Spoken and regional varieties
of English in particular employ systems that are far more intricate than a
superficial analysis would reveal.
This study originated from a long-standing interest in one particular
feature found in the system(s) of gender assignment of the traditional
English dialects of Southwest England and Newfoundland, which the
author had first been encountered in her early student days. As with many
interesting features of morphosyntax, literature on the subject was scarce to
non-existent, and actual examples were hard to come by. However, over the
years it was possible to gather what literature was available and remedy the
lack of data by collecting a corpus of actual dialect speech (FRED; for
details, see Kortmann and Wagner, this volume; also Anderwald and
Wagner forthcoming). The following examples shall illustrate the core
issue of this study, namely the seemingly confusing use of pronominal
forms in basically all varieties of (spoken) English. All of the examples in
(1) share one feature: the personal pronoun forms used (in bold print) are
supposedly reserved for reference to human or at least animate entities.
(1) a. That is a dead teat with no milk into en[n]. (38 Do 3, book III)
b. We call en[PB] a peeth [well]. (36 Co 6, book IV)
c. He used to say Put un [candle] where ye can zee im[candle] and I
can zee im[candle] as well. (FRED Wil_024)
d. <u Int>Those flat irons interested me too because they had a handle
that would come off, removable handle, why was that?
<u Inf>Das for you put one on da stove now and you put two on da
stove and when dey get warmyou put da handle in you take and ah ah
when he[iron] when he[iron] get cold off you put he[iron] on you put
your handle in t udder onehe[iron] be warmsee, hed [iron] be hot
and youd take en [iron] and youd iron your clothes den and while
one be, while one be warmin da t udder be coldin, time be coldin
off see. (MUNFLA 70-003: C0626)
Gender in English pronouns 213
e. I joined in the first hall was down dere and den they build this one is
down there now and they build he [hall/house] in 19-, in 1921 I
believe they build it, I joined in the other one, I joined in 1917, in the
ole lodge. (MUNFLA 71-131: C1034)
f. press themlike that and youd see your thumb mark in themor any
apple really when hes [apple] ripe, wadnt it, but when hes [apple]
not ripe hes [apple] hard, isnt he[apple], (FRED Som_001)
g. Ok, crack er up! (From the movie Titanic [USA 1997]; the speaker is
(presumably) an American male, talking about the safe being brought
up from the ocean floor.)
h. Where is she? If she will give us the pleasure there she is! (From
the movie The Red Violin; the speaker is [presumably] a Canadian
male, talking as an auctioneer about the violin that is to be sold next.)
i. shes up/down (in reference to a share [price]; Graham Shorrocks,
personal collection);
shes up 30 today; shes off $2 today (reference to the market [e.g. the
Dow, Toronto, NASDAQ] and to an individual stock [price];
Newfoundland brokers; Graham Shorrocks, personal collection
j. This is another pot and saucer. A bit dusty! You see that one isnt
exactly glazed proper, burnt proper is she[pot]. (FRED Som_009)
At a first glance, the target nouns seem to have nothing in common: a
cows teat, a well, a candle, an iron, an assembly hall, an apple, a safe, a
violin, stock prices, a pot this list could be continued, but suffice it to
note for now that there seem to be more contrasts than shared properties
among the referents of the relevant nouns.
As varied as the target nouns are the forms of the personal pronouns
employed in the examples: en, un, im, he, she. While the last three forms
can be readily identified as (reduced) object pronoun (masculine) and
subject forms (masculine and feminine), the origin(s) of the first two forms
may not be as obvious. Realized in numerous ways in speech, ranging from
[n] to [nB], there is no uniform orthography for the item in question. Most
commonly spelled either enor un, it is analyzed as a reduced form of Old
English (OE) hine(acc. sing. masc.; cf. OED un 1., being used as early as
1633), thus being equivalent to PrDE him. This enor unis a form typical of
the Southwest. Its extension, based on data from the Survey of English
Dialects (SED), is shown by Wakelin (1986: 35).
Scholars have not (yet) agreed on a suitable label for the phenomenon
terms that have been applied include gender diffusion, animation
214 Susanne Wagner
(Siemund 2001, 2002; forthcoming) and individuation (Pawley 2002).
For reasons that will be discussed in detail in later sections, a more neutral
term will be used here: gendered pronouns, referring to instances of
personal pronouns which are marked for gender (i.e. either masculine or
feminine) but which refer to inanimate (count) nouns.
Different from many other features typical of non-standard
morphosyntax, gendered pronouns do not seem to be stigmatized, clearly
operating below the level of consciousness. When asked about them, native
speakers are generally aware of the weird use of pronouns, and readily
offer all types of explanations. As can already be seen from the examples
cited above, one should distinguish at least two systems of gendered
pronouns: one where feminine forms are used and a second one that mainly
employs masculine forms. As will be shown below, the two systems have
different origins and different regional distributions. Where they interact,
various factors have to be taken into account when discussing gender
assignment in the relevant varieties.
In the following sections, gendered pronouns will be analyzed from a
number of different perspectives. Section 2 will provide a brief overview of
existing research on the subject. Also, when discussing gender assignment
in non-standard varieties, the standard system deserves a closer look
(Section 3). The corpora this study is based on will be presented in Section
4. Special pronominal referent classes such as animals are discussed in
various studies on standard and non-standard varieties of English around
the world and are thus also included here (Section 5). Studies of gendered
pronouns in non-dialectal contexts are at the heart of Section 6, while
Section 7 focuses on possible persistence phenomena (priming effects) in
the use of gendered pronouns.
Sections 8 to 11 constitute the empirical part of the study. The actual use
of gendered pronouns will first be analyzed in the material from the SED
(Basic Material Section 8 and fieldworker notebooks Section 9),
followed by detailed accounts of those gendered pronouns occurring in the
oral history material from Southwest England (Section 10). For lack of
space, the analyses of the Newfoundland material will only be briefly
summarised here (Section 11), pointing out the most interesting differences
between the English and the Canadian material.
1
An overall summary
concludes the study (Section 12).
2. Gendered pronouns
The study of gendered pronouns has a long tradition in dialectology. In the
late 19th and early 20th century, they are mentioned in most descriptions of
dialects, both regionally restricted ones and general accounts (e.g. Wrights
English Dialect Grammar [1905]), as one of the few non-lexical features
mentioned (but generally not explained) in descriptions largely dominated
by lexical material. Interestingly, gendered pronouns never reached any
considerable fame outside dialectology, and although native speakers
readily offer examples when confronted with the issue, they do not seem to
be aware of the extension of the phenomenon. Only few studies dealing
with gendered pronouns are not restricted to dialect evidence, most
extensively Morris doctoral thesis on Gender in English (1991). Because
the evidence is scattered throughout the dialectological literature, mostly
accumulating in the various publications of the English Dialect Society, the
most important accounts will be summarized in the following sections.
2.1. Gendered pronouns in (traditional) English dialectology 1879 to
the present
Probably the earliest mentioning of gendered pronouns can be found in
William Marshalls Provincialisms of the Vale of Glocester, where the
following is said about pronominal usage:
[T]his quarter of the island affords, among others, one striking deviation in
GRAMMAR in the use or abuse of the pronouns. The personal pronouns are
seldom used in their accepted sense [...] sometimes he [is used] for she; as,
he was bulled he calved; and almost invariably for it; all things
inanimate being of the masculine gender. (1789: 56; emphasis original,
boldface SW)
The excerpt shows the authors clearly negative attitude towards the variety
of English he encountered in Gloucestershire, which may explain why he
offers neither concrete examples nor a more detailed description of the
phenomenon.
To this day, Frederic Elworthys work on the traditional dialect of
Somerset is unsurpassed in detail and richness of documentation. Although
their authenticity is debatable
2
, Elworthys examples offer researchers the
opportunity of catching at least a glimpse of what 19th-century pronominal
usage may have been like in (West) Somerset (cf. Siemund forthcoming,
216 Susanne Wagner
who used Elworthys 1886 publication as the main corpus for his analysis)
an opportunity that is not available for any of the other varieties. In partly
re-stating and summarizing his own earlier ([1875] 1965a) work, the author
offers the most detailed description of gendered pronouns to be found
anywhere in the literature in his 1877 Outline of the Grammar of the
Dialect of West Somerset (Elworthy [1877] 1965a). It is this account that
has influenced all later studies, and the distinctions and classifications made
therein have been taken as reference points ever since.
Every class or definite noun, i.e. the name of a thing or object which has a
shape of its own, whether alive or dead, is either masculine or feminine, but
nearly always the former; indeed, the feminine pronouns may be taken as
used only with respect to persons. (Elworthy 1965a: 32; emphasis SW)
Examples here include a pitcher as well as the nouns tool, book, house,
coat, cat, letter etc., that are all spoken of as he (Elworthy 1965a: 33).
On the other side, [i]t is simply an impersonal or ABSTRACT pronoun, used
to express either an ACTION or a noun of the UNDEFINED SORT, as cloth in
the quantity, water, snow, air, etc. Weather, hay, and beer are used to
exemplify this use (cf. Elworthy 1965a: 33; small capitals SW).
With this, Elworthy established a system that could be described in
modern terms as a semantic gender system based on a mass-count
distinction in nouns, with count nouns taking feminine or masculine
pronouns, while mass nouns are neuter. Elworthy himself uses a similar
wording in his 1886 West Somerset Wordbook (Elworthy [1886] 1965b),
from which the following quotation (entry for he) is taken
3
:
The universal nominative pronoun to represent all things living or dead, to
which the indefinite article can be prefixed. [...] He is used in speaking of a
cow or a woman, but not of corn, water, wool, salt, coal, or such things as
are not individual, but in the mass. (Elworthy 1965b: 328; boldface SW)
A contemporary of Elworthys, the Dorset poet William Barnes, whose
poems contain numerous examples of dialect, also wrote a short treatise on
the grammar of his home county. His Dissertation on the Dorset Dialect of
the English Language (Barnes [1844] 1994) was originally published
together with his first collection of poems, Poems of Rural Life in the
Dorset Dialect. Although gendered pronouns are mentioned only very
briefly in this sketch, Barnes poems show that he was aware of the
phenomenon. More than 40 years later, his Glossary of the Dorset Dialect
with a Grammar of its Word Shapening and Wording offers more detail on
the matter. The heading two classes of things contains the following
observations on pronominal usage:
Whereas Dorset men are laughed at for what is taken as their misuse of
pronouns, yet the pronouns of true Dorset, are fitted to one of the finest
outplannings of speech that I have found.
[There are two classes of things:]
1. Full shapen things, or things to which the Almighty or man has given
a shape for an end; as a tree, or a tool: and such things may be called the
Personal Class: as they have the pronouns that belong to man.
2. Unshapen quantities of stuff, or stuff not shapen up into a form fitted to
an end: as water or dust: and the class of such things may be called the
Impersonal Class, and have other pronouns and those of the personal class.
(Barnes [1886] 1970: 17; emphasis SW)
It is not difficult to equate Barnes class I (personal class) with Elworthys
count nouns and Barnes class II (impersonal class) with Elworthys mass.
According to Barnes (1970: 17), heis the pronoun of the personal class, with
en serving as objective form. En is explained as deriving from the Saxon-
English accusative (<he-ene >hine >hin). The impersonal class, on the
other hand, uses it. The author illustrates the resulting contrast with the help
of examples referring to a tree (personal, thus: hes a-cut down, J ohn velld
en) and to water (impersonal, thus: its a-dried up). In addition, the same
noun can be classified differently according to context. Thus, en has to be
used when referring to a brick bat (take en up), but it for a lot of brick-
rubbish: take it up (Barnes 1970: 18).
Work by Perry (1921) on North Somerset, Kruisinga (1905) on West
Somerset, J ennings (1869) on the West of England (with special emphasis
once more on Somerset), and Dartnell and Goddard (1893) on Wiltshire
shows a mixture of implicit and explicit references to gendered pronouns,
while explanations for the phenomenon are generally absent from these
accounts. Very often, the given examples contain gendered pronouns, but
their use is either only worth a side remark of the type that un stands for
him, or it (Dartnell and Goddard 1893: 124), or not commented on at all
(e.g. J ennings 1869).
Despite its title, Alexander Ellis On Early English Pronunciation
provides a source for much more than phonological analysis. Ellis district
4, the Southern division, constitutes the focal area of the present
investigation, consisting of speaking in terms of modern dialect areas (see
Trudgill 1999
2
: 65) the central Southwest as well as the southern part of
218 Susanne Wagner
the upper Southwest, with Somerset, parts of Dorset, Hampshire, and
Gloucester forming the core area.
In his description of special grammatical constructions of this district,
the author makes the following comments on gendered pronouns:
This (mQ) is very widely spread in the S[outhern] div[ision], and is also used
where it is said in received speech, on account of the general use of he
applied to inanimate objects [...] (Ellis 1889: 43).
This statement seems rather insubstantial and unsatisfactory as an
explanation. That Ellis was aware of the construction can be deduced from
his specimens, one of which is full of gendered forms (see Ellis 1889: 151
152), including various masculine references to a tree (he6x, un7x, er 1x),
a ravine, an oven, a roof (un), and a gate (en2x).
Gendered pronouns are also encountered in the prose and poetry of
West Country authors, most prominently, as already mentioned above, in
the work of William Barnes and Thomas Hardy, who in particular has been
subject of various studies focusing on his language use. However, most of
these studies either emphasize quantity rather than quality (e.g. Hirooka
1980, which contains long but uncommented lists of examples), or
comment only very insufficiently on the dialect features found in Hardys
dialogues (e.g. Elliott 1984).
Unfortunately, then, Ossi Ihalainens summary of the 18th- and 19th-
century dialect literature seems more than adequate in light of the material
presented above: Generally speaking, the picture that emerges from the
early evidence is patchy, difficult to interpret and open to conjecture.
(1994: 197).
2.2. Gendered pronouns in modern dialectological investigations
Two scholars in particular, namely Ossi Ihalainen and Martyn Wakelin,
contributed to a better understanding of dialects in the Southwest,
publishing extensively on West Country English. Ihalainens focus is
primarily on grammar, while Wakelins two monographs on Cornwall
(1975) and the Southwest in general (1986) are mainly concerned with
phonology and lexicography, but also include sections on morphology and
syntax.
Ossi Ihalainen generally agrees with the accounts of his predecessors in
evaluating gendered pronouns. However, since a century of language
development must have had an impact on the traditional system of
attributing gender, the author modifies Elworthys earlier rather strict
system of mass versus count referents as follows:
On the whole, the evidence suggests that Elworthys description is basically
correct. However, rather than say that it is used for mass referents and the
personal forms for thing referents, the correct generalization today seems
to be that it can be used for thing and mass referents, although it
predominantly occurs with mass referents, whereas the personal forms do
not occur with mass referents at all. (Ihalainen 1985a: 158)
This adaptation of the system is based to some extent on the observation
that native speakers themselves seem to vary in their judgements of what is
or is not a thing: informants may disagree about thingness (Ihalainen
1985a: 158).
Ihalainens main interest lies in the analysis of two details of
pronominal distribution: i) what is the relationship between standard it and
non-standardheforms, and ii) what is the distribution of subject and object
forms in this context? Ihalainen uses a partly typological approach to
explain the results he observed in his corpus. When looking for gendered
pronominal forms, he noticed that their distribution across grammatical
cases, i.e. basically subject and object position, was by no means identical.
Based on his findings, the author suggests that standard forms establish
themselves in the dialect system in less accessible positions of the Noun
Phrase Accessibility Hierarchy (subject > direct object > non-direct/
indirect object > possessor/oblique; cf. e.g. Keenan and Comrie 1977;
Comrie 1989: 155163) and spread from there. Thus, it is more frequent in
object position than one might expect, at least in contexts where heis also
possible in dialect (cf. Ihalainen 1985b: 6970). Or, in Ihalainens own
words:
[T]he most prominent linguistic contexts are also the most favourable to
dialectal forms. Interpreted diachronically, this means that the changes
concerned arose in non-prominent contexts and are spreading to more
salient ones. (Ihalainen 1991: 105)
For the present investigation, this means that standard it forms first
invaded the territory of personal forms in object position, later spreading
to the more prominent subject contexts as well.
4
This hypothesis is in fact
borne out in the corpus data investigated in detail in Wagner (2004b).
220 Susanne Wagner
Contrasting with Ihalainens studies, which for the most part is based on
authentic speech samples, Martyn Wakelins three major publications
(1975; 1986; 1981) are all based on the SED or at least on SED material,
which is the main drawback of the respective monographs.
Largely following the traditional SED vein, Wakelin (partly) shifts the
focus from lexicology to phonology and also includes general background
information on the variety in question, from geographical to settlement
information (e.g. paragraphs about the historical situation of English in the
respective areas, including in the case of Cornwall some remarks on the
situation of Cornish in relation to English, which is necessary for a full
understanding of the language situation in present-day Cornwall).
The 1986 monograph contains a valuable compilation of texts from the
core counties of the Southwest, including material from the 16th century to
the 1970s. Unfortunately, not much information on morphological or
syntactic peculiarities of the area is included. In the book on Cornwall,
based on his Ph.D. thesis, Wakelin introduces the chapter headed
Morphological features as follows:
The small collection of morphological features from SED assembled under
several headings in this chapter is intended to support the phonological
material in the preceding chapter and the lexical material in that following.
(Wakelin 1975: 175)
Wakelin emphasizes that the respective SED material was selected to
support his main hypothesis of the book, namely that Cornwall is divided
into two parts: From a linguistic point of view, the East is close to its
traditional Southwest neighbours, while the West is closer to StE, mainly
because of the late introduction of English there (Cornish being the
traditional first language until its extinction
5
). Probably due to this bias in
the selection of the material, gendered pronouns are not even mentioned
among the morphological features (four features are used, one from the
system of personal pronouns, two from verbal morphology, and one from
noun phrase morphology; cf. Wakelin 1975: 175179). One cannot help
but wonder why Wakelin chose not to include gendered pronouns, whose
use should be very indicative of the assumed distribution of features. One
would expect the traditional system for the East, with personal forms being
used for count/thing nouns and it being used for mass referents, while
in the West the StE system or at least something close(r) to it should
dominate. Strangely enough, Wakelin chose the other peculiarity of
Southwestern pronominal systems, pronoun exchange, to make his point.
That the author is by no means unaware of the phenomenon can be seen
from his description of gendered pronouns in his likewise SED-based
monograph on English dialects:
En is used for it (object) as well as for him in the south-west of England,
beside which the forms he, him and occasionally she, her may also be used
to denote an inanimate object over a rather wider area in the west but in
more scattered examples [...] The full implications of the use of he, she,
him, her for inanimate objects have not yet been explored. (Wakelin 1981:
113)
Although showing some phonological bias as well, The Southwest of
England (1986) is much more balanced. An eight-page section on
phonological features stands against a four-page overview of grammatical
features. However, even though comments on the pronominal systems
make up about half of the section, gendered pronouns are again only
mentioned in passing. While the formal peculiarities (existence of old
accusative n in a number of variant realizations/spellings) are listed, the
slash giving himandit equal status in the example is not commented on (cf.
1986: 34).
6
The present authors own investigations on pronoun exchange,
a feature that does not avail itself easily to detailed study for a number of
reasons (cf. Wagner 2001), show that it is far less frequent than gendered
pronouns. Although Wakelins procedure is more than surprising, it is
unfortunately not unique, as a look at some other publications shows:
gendered pronouns do not figure prominently in any of the admittedly few
modern publications on West Country dialect (cf. e.g. Attwell 1987;
Downes 1986; J ones and Dillon 1987; Marten 1973).
3. Gender in English and elsewhere
This section gives a short overview of the history of gender in English,
both in isolation and in comparison with other languages. Questions that
need to be answered in this context include the following:
1. What is gender? typological classification
2. How do gender systems develop and change? diachronic
evidence, focus on English
3. Based on the definitions and classifications in 1. and 2., where
does gender fit in English today?
222 Susanne Wagner
These issues will be addressed from a functional-typological point of view,
taking English as the focal point, but also taking a glimpse at both related
and unrelated languages as dictated by the subject matter.
3.1. Gender in the worlds languages basic terminology and classi-
fications
Two major, sometimes competing systems for assigning gender exist in the
worlds languages. On the one hand, there are semantic systems, where
semantic factors are sufficient on their own to account for assignment
(Corbett 1991: 8). Various features are used as the basis for gender
assignment in such systems. Systems where masculine gender is attributed
to males and feminine gender to females are often called natural gender
systems (Corbett 1991: 9). Criteria for such systems are widespread; often,
the general division is one between human and non-human, and humans are
divided into male and female in turn (cf. Corbett 1991: 11). Sometimes the
dividing line is animate inanimate instead of human non-human.
English might be an example of this kind of semantic system, as animals
(particularly domestic animals) are usually masculine or feminine
according to sex; however, there are other factors that may influence
pronoun choice (e.g. conventions of childrens stories; cf. Corbett 1991:
12). Cross-linguistically, [t]he feature animate is particularly pervasive in
semantic gender systems (Corbett 1991: 31).
A more complex system can be found in Algonquian languages: Most of
these have two genders, with a basic animate inanimate contrast. An
additional factor for gender assignment is power: powerful and/or
dangerous things (although inanimate) usually belong to the animate
gender, i.e. are grammatically animate (cf. Corbett 1991: 2021). In
Caucasian languages, a count non-count distinction seems to play a role;
for example, liquids and abstracts (non-count, non-rigid) belong to the
same gender (cf. Corbett 1991: 2430).
On the other hand, there are formal systems. Here, gender is determined
to a large extent by formal criteria (usually phonological, e.g. in French, or
morphological, e.g. in Russian; cf. Corbett 1991: 3739). It is important to
note that neither strict semantic nor strict formal systems seem to exist.
Most of the worlds languages make use of mixed systems, but even in
formal systems gender always has a basis in semantics (Corbett 1991:
63). Thus, when conflicting rules exist, semantic considerations normally
take precedence (Corbett 1991: 66). This can be illustrated with the help of
a German example.
German has a relatively formal morphological gender system (cf.
Corbett 1991: 4950) which usually assigns gender on the basis of
derivations. Lexemes with the suffix -chen (indicating diminutives) are
assigned to the neuter gender. With the lexeme Mdchen girl, there is a
clash between semantics and morphology, as semantics clearly demands
feminine gender rather than the neuter assigned on the basis of a formal
rule. While it is ungrammatical to use a feminine article (i), feminine
endings on attributive adjectives (ii), or a feminine relative pronoun (iii), it
is possible to use sieshe as an anaphoric pronoun (iv), although neuter es
it is also possible. The Agreement Hierarchy (AgrH; cf. v) nicely predicts
this possibility (cf. Corbett 1991: ch. 8):
i. *die Mdchen ist jung the
NEUT
girl is young
ii. *eine junge Mdchen a

FEM
young
FEM
girl
iii. das Mdchen, *die ich meine the
NEUT
girl who(m)
FEM
I mean
iv. das Mdchen ist jung; ich kenne sie/es
the
NEUT
girl is young
; I know her/it
v. The Agreement Hierarchy
attributive <predicate <relative pronoun <personal pronoun
As we move rightwards along the hierarchy, the likelihood of semantic
agreement will increase monotonically (that is, with no intervening
decrease).
7
(from Corbett 1991: 226)
It is interesting to see that the AgrH also works in the reverse scenario of a
basically semantic system where nouns can exceptionally be assigned a
category other than the expected gender. English boat nouns offer nice
examples in this context (cf. Corbett 1991: 180181; 236238):
vi. the QEII is a beautiful ship
vii. the QEII, on *whom I sailed recently, is a beautiful ship
viii. the QEII, on which I sailed recently, is a beautiful ship
ix. I sailed on the QEII recently; she/it is beautiful
While it is impossible to use the relative pronoun who(m) withship, the use
of sheas illustrated in (ix) is possible, thus supporting the AgrH.
8
Corbett
defines agreement as the determining criterion of gender (1991: 4) and
adds that [t]his is the generally accepted approach to gender (1991: 4).
Agreement can manifest itself to various degrees in numerous categories.
Adjectives can agree with the nouns they classify, verbs with their subjects
or even objects, etc.
224 Susanne Wagner
Control of anaphoric pronouns by referents is usually included as part of
agreement; thus, a language like PrDE, where gender only becomes evident
in personal pronouns, should be said to have gender as well (cf. Corbett
1991: 5). Corbett calls such a system a pronominal gender system (1991:
5). In light of the AgrH, attributive modifiers and personal pronouns seem
to be two opposite poles of a single hierarchy, i.e. they should be treated
as part of the same phenomenon (Corbett 1991: 112). As illustrated in the
above examples, personal pronouns (at the end of the AgrH), are the most
likely stage at which semantic factors overrule formal ones. This could
already be observed in Old English, where it was possible for a non-neuter
noun to take a neuter anaphoric pronoun, especially when some spatial
distance between pronoun and referent was involved (cf. x):
x. t u one wisdom e e God sealed,
that you that wisdom which to you God gave,
r r u hiene befstan mge, befste.
there where you it
MASC
implant may, implant.
Geenc hwelc witu us a becomon
think what punishments to us then came
for isse worulde, a a we hit nohwer
for this world when we it
NEUT
neither
ne selfe ne lufodon, nee ac orum
NEG ourselves NEG loved, NEG also each other
monnum ne lefdon
men NEG allowed
(Dekeyser 1980: 101; Corbett 1991: 242243)
In the example above, the masculine hieneis used in the immediate vicinity
of wisdom(masculine in OE), but the next reference, spatially removed
from its antecedent, is neuter hit. One can thus conclude that personal
pronouns seem to be the major initiator of changes in the balance between
syntactic and semantic gender (Corbett 1991: 242). English obviously
took this direction in its history, developing from a language with a fairly
formal gender system comparable to that of Modern German to a language
with a semantic (or even natural) gender system.
3.2. Gender in English the history
It is generally known that English inherited a formal gender system
9
from
its Germanic parent language which, roughly between the 10th and 14th
century, was gradually replaced by the semantic natural or logical
gender system that is known from PrDE, the change having been completed
in early Middle English (ME) (cf. Kastovsky 2000: 709; Dekeyser 1980:
102). The dissolution and ultimate demise of the formal system is usually
associated with phonetic changes in the syllable that eventually led to the
loss of most inflectional endings on the noun. However, it has been noted
that it is probably not entirely correct to classify the OE gender system as a
purely formal category. Although some derivational endings showed a clear
one-to-one match of ending to gender (e.g. -dom masculine, -ness
feminine), many suffixes were affiliated to more than one gender
(Kastovsky 2000: 712). Also, for underived nouns, next to nothing could be
predicted about their gender from their inflectional class, thus making it
basically impossible to speak of a formal gender system in OE (cf.
Kastovsky 2000: 712). Moreover, the distribution of nouns into the three
genders masculine, feminine, and neuter had an almost semantic basis in
OE already in that (in West Saxon) most male nouns were masculine, most
female nouns feminine, and the majority of neuter nouns were asexual
(Ross 1936: 321; cf. also J ones 1988: 35; Moore 1921: 89).
10
Kastovsky (2000) illustrates how various stages and processes that took
place in the inflectional morphology of the noun phrase were intimately
linked to the dissolution of the formal category gender. It turns out that in
OE already, very few combinations of an agreement-marking item
(determiner, adjective) +noun were really unambiguously gender-specific
(cf. Kastovsky 2000: 715717). Thus, wrong gender assignment not only
existed in OE times (cf. [x] above; J ones 1988: 10), but was presumably
rather frequent.
11
In addition, semantic and/or pragmatic factors can be
expected to have overruled formal ones at that stage as well, particularly
where formal gender and sex clashed (e.g. masculine wifman, neuter
mgden; cf. Kastovsky 2000: 711712; J ones 1988: 3638; Wales 1996:
137139). Consequently, it seems appropriate to look for the beginning of
the end of gender in English at a much earlier time, and it is likely that
226 Susanne Wagner
Moore (1921: 91) was correct in saying that natural gender did not replace
grammatical gender in ME but survived it.
12
Anne Curzan (2000) adds another important factor to support this view,
a factor which, judging from traditional accounts on gender in English, has
largely been overlooked or at least neglected. She points to the status of
English as a non-literary language when its first grammars were written,
inheriting all systems and their classifications from Latin, a highly
synthetic language with an elaborate grammatical gender system:
The early English grammarians were beginners, stumbling through
English grammar with only Latin grammar and its terminology as their
guide; despite the obvious inapplicability of many Latin categories for
English, they often retained them in the English grammars either because
they wished to adhere to tradition or because they could not conceive of
other possibilities. (Curzan 2000: 563)
J udging from the inflectional paradigm of OE nouns, feminine noun
endings were generally more distinctive than either masculine or neuter
ones. This could support the view that feminine gender lingered longer than
the other two, and that this continued association led to the persistence of
feminine pronominal reference with some nouns to the present day (cf.
Fennell 2001: 64). This is a fact worth remembering when the
predominance of feminine referents is investigated in Section 6. Matters
were not simplified by the gradual case syncretism of dative and accusative
personal pronouns. Visser (1963: 427) notes that the reason for the change
from hinetohimhas as yet not been satisfactorily accounted for. Possible
explanations include the extension of the dative from rather frequent verbs
taking dative forms as objects after speakers were no longer aware of the
case distinction(s) (cf. Visser 1963: 427), or an economically motivated
choice of the pronoun most distinct from the nominative (cf. Howe 1996:
114115). Although dative-accusative levelling took place in most
Germanic languages, some languages lost the dative forms, while others
generalized the dative and lost the accusative forms. Howe (1996: 111) thus
concludes that no theory is at present able to account fully satisfactorily
for both these directions. Although the entries for it and himin the OED
describe the concrete developments in detail, the only conclusion to be
drawn from these observations is that there was a rather extended period of
time in the history of the English language when the choice of a supposedly
masculine personal pronoun (him) said nothing about the gender or sex of
the referent. It could be masculine, male, neuter, or asexual and every
combination of those four.
3.3. Gender in English today
Although rather marginal in PrDE, gender has always been a category
(formal or functional) of English, and no one dares (yet) to let go of it
completely. In the following, the view that grammarians have on gender in
PrDE will be investigated more closely.
3.3.1. Gender in modern references grammars
As has already been indicated, modern scholars writing on PrDE are highly
influenced by traditional accounts of gender in Indo-European languages.
They adopt the traditional terminology which was developed for those
languages, and no one seems to consider it inappropriate for describing a
language like PrDE that is almost as far removed from a true gender
language as it can possibly be. Quirk et al.s Comprehensive Grammar of
the English Language (1985) is no exception in this respect:
By gender is meant a grammatical classification of nouns, pronouns, or
other words in the noun phrase, according to certain meaning-related
distinctions, especially a distinction related to the sex of the referent. In
English, unlike many other related languages, nouns, determiners and
adjectives have no inflectionally-marked gender distinctions. Some 3rd
person pronouns and wh-pronouns do, however, express natural gender
distinctions:
it, which etc [nonpersonal] contrasts with the following:
who, whom etc [personal]
he, himself etc [masculine, chiefly personal]
she, herself etc [feminine, chiefly personal]
(Quirk et al. 1985: 314; original emphasis)
This definition of gender sounds suspiciously non-specific; yes, gender is a
grammatical category, but is it really a grammatical category of PrDE? The
authors choose the unfortunate path of equating gender with sex in talking
about meaning-related distinctions, in particular those referring to the sex
of the referent. The next paragraph tries to clarify matters, but one cannot
help but feel slightly confused by the varied terminology:
228 Susanne Wagner
Gender in English nouns may be described as notional or covert in
contrast to the grammatical or overt gender of nouns in languages such
as French, German, and Russian; that is, nouns are classified not
grammatically, but semantically, according to their coreferential relations
with personal, reflexive, and wh-pronouns. We use the term male and
female in reference to the covert gender of nouns, as distinct from the
overt gender of pronouns. (Quirk et al. 1985: 314; original emphasis)
While the previous paragraph spoke of natural gender distinctions, now
they are notional or covert. This in itself poses no real problem, but
equating the existence of gender with the existence of gender-specific
pronouns does. Claiming that English has gender because it has pronouns
that show gender distinctions is a dangerous garden path the traditional
argumentation would rather be that the distinctions in the pronominal
systems only exist because the nominal referent carries the feature gender,
which in turn has to be mirrored in the pronominal system(s).
What is even more unfortunate in the terminology used in the above
paragraphs is the use of male, female, masculine and feminine. Quirk et al.
equate the gender terms masculine and feminine with the overt, natural
gender of pronouns, while nouns with their covert, notional gender
are male or female, which are biological categories referring to the sex of
the referent. This seems to be a contradiction in terms. Why should
different terms which describe essentially the same phenomenon be
used for different word classes? A personal pronoun referring to a male
human has either both features, [+male] and [+masculine], or, if one were
to accept that English gender is a purely semantic category, only [+male].
Also, it is not clear why the authors suggest that nominal gender is an overt
category in German grammatical, yes, but the categories are only rarely
expressed overtly (although some derivational endings allow certain
predictions about gender assignment, the rules for underived nouns are far too
complex to call them overt). As Corbett (1991: 49) summarizes, gender
assignment rules in German consist of a complex interplay of overlapping
semantic, morphological and phonological factors. The gender system Quirk
et al. derive from these observations must look confusing, if not wrong, to
most speakers of English (cf. Fig. 5.104 in Quirk et al. 1985: 314).
According to this system, English has nine (!) gender classes (a term
that is never explained nor defined) which reveal a high degree of overlap
with each other. None of the sub-categories is sufficient for assigning
gender to a noun. At first glance, inanimate seems to be sufficient, but if
the classification were based on pronoun co-reference exclusively (as the
authors claim it is), all categories that allow which it (i.e. categories d,
e, f, g, h, i) should be in the same sub-gender. Obviously, the authors
applied some sort of hierarchy to arrive at the above classification in which
humanity >sex >animation >pronoun choice, but the basis of such a
system, if it exists, is never explained.
On the category of animals, the following can be found: Male/female
gender distinctions in animal nouns are maintained by people with a special
concern (for example with pets) (Quirk et al. 1985: 317). No reference is
made to the fact that animals are generally he in spoken language (cf.
section 5.2). The authors concern here is with professional language use
that includes terminological differentiation, e.g. dog bitch etc. Another
exceptional noun class mentioned is that of country names, which can be
used with neuter (emphasising the geographical unit) or feminine
(political/economic units) pronouns (cf. Quirk et al. 1985: 318). Oddly
enough, it is stated that in the latter case gender is either class (b) or (g)
(ibid.) (b) is acceptable, but it seems more than doubtful that a country
could be classified as a higher female animal (class g)
On the well-known use of feminine pronouns referring to ships, we are
told that
[i]nanimate entities, such as ships, towards which we have an intense and
close personal relationship, may be referred to by personal pronouns, eg:
Thats a lovely ship. What is shecalled?
In nonstandard and Australian English, there is extension of she references
to include those of antipathy as well as affection, eg:
Shesan absolute bastard, this truck.
(Quirk et al. 1985: 318, original emphasis)
With the generalized we in the first sentence the authors overdo things
slightly. While it cannot be denied that a group of professionals (fishermen,
ship crews, yacht owners, etc.) certainly have intensive and close personal
relationships with the ships they are sailing on, it seems difficult to claim
such a relationship for the average person. The following statement on
Australian English sounds rather unprofessional, the wording and inclusion
itself a bit unreasonable and rather unfounded. Although the co-ordination
of non-standard and Australian English seems awkward (and is, as will
be shown later, wrong), the authors at least mention that feminine forms are
the pronouns of choice in speech (cf. Section 6). However, they offer
neither comments nor explanations for this type of use.
230 Susanne Wagner
A more contemporary (or updated) approach to gender in English can be
found in Longmans Grammar of Spoken and Written English (Biber et al.
1999), incorporating, as the title suggests, considerable amounts of material
from spoken language for the first time and thus complementing rather than
substituting Quirk et al. (1985).
Gender is a less important category in English than in many other
languages. It is closely tied to the sex of the referent and is chiefly reflected
in cooccurrence patterns with respect to singular personal pronouns (and
corresponding possessive and reflexive forms). The main gender classes
are:
example noun pronoun
personal/human:
masculine Tom, a boy, the man he
feminine Sue, a girl, the woman she
dual a journalist, the doctor he, she
non-personal/neuter: a house, a bird it
(Biber et al. 1999: 311312)
The system presented here seems much closer to reality than the one
proposed by Quirk et al. Biber and his colleagues are well aware of the fact
that gender is a problematic category in PrDE. Thus their insightful caveat
about oversimplifying matters: However, gender is not a simple reflection
of reality; rather it is to some extent a matter of convention and speakers
choice and special strategies may be used to avoid gender-specific
reference at all (Biber et al. 1999: 312).
After this general introduction, Biber et al. largely discuss pragmatic
motivations for pronoun choice, such as the use of specifically gender-
marked forms on the one hand or avoidance on the other (e.g. chairman/-
woman vs. chairperson). They note a continuing sex-bias in English
language use and society more generally (1999: 313) towards masculine
terminology, which is not only reflected in a much higher occurrence of
masculine forms in pairs like the one mentioned above, but also in the
distribution of third person singular personal pronouns in general:
masculine forms are more frequent in all registers, occurring 1.5 to more
than 3 times as frequently as feminine forms (Biber et al. 1999: 333335).
Of major relevance for the present discussion is the section about personal
vs. non-personal reference (Biber et al. 1999: 317318). The authors state
that [p]ersonal reference expresses greater familiarity or involvement. Non-
personal reference is more detached (1999: 317). Items falling into the
category that offers a three-way choice (personal he, she; non-personal it) are
expressions for young children (infant, baby, child) and animals (pets in
particular; cf. Biber et al. 1999: 318). An exceptional status is once again
attributed to nouns denoting countries and ships, which offer a two-way
choice (personal she, non-personal it). Although Biber et al.s account comes
much closer to the actual facts observed in the realization of gender English
today, like Quirk et al. they fail to offer explanations for the exceptional
cases, most of which are not even mentioned (e.g. why a ship can be she).
Huddleston and Pullums Cambridge Grammar of the English
Language(2002), which contains the most recent comments on the subject,
will be introduced by way of quoting their introductory remarks on
Gender as a grammatical category.
[...] it is important to distinguish carefully between the grammatical terms
masculine and feminine and the semantic or extralinguistic terms male and
female. Until relatively recently it was usual to make a parallel distinction
between gender (grammatical) and sex (extralinguistic) [...]. In the social
sciences, however, sex came to be used to refer to biological attributes
and gender to the social construction of sex, and this usage has been
incorporated into linguistics. A book on language and gender will
therefore not be primarily concerned with gender as a grammatical
category, but will cover such matters as differences between the speech of
men and women. Our concern in this section, however, is with gender in the
old, strictly grammatical sense [...] (Huddleston and Pullum 2002: 484485;
italics SW)
For the first time, the change in terminology and the resulting confusion
is topicalized in a grammar, which in itself points to a radical restructuring
in the respective areas of research. The authors continue their introduction
by justifying their decision to treat gender as a grammatical category, using
an argumentation very much in Corbetts vein (agreement as the defining
criterion of gender English does show agreement, though in a very
restricted sense English has gender, though it is not an inflectional
category and not as strongly grammaticalized as in other languages; cf.
Huddleston and Pullum 2002: 485486). Although their use of the term
less grammaticalized is debatable
13
, the authors take a very clear position
in assessing the category, which is refreshing and helpful compared with
the earlier descriptions.
Typical wording can be found in the actual distributional properties of
masculine he, feminine she, and neuter it. Heandshereferring to males and
232 Susanne Wagner
females respectively, it referring to entities which are neither male nor
female, are identified as the core uses of he, she, and it (cf. Huddleston
and Pullum 2002: 484). As this definition of it would exclude its use with
both animal and human antecedents, which do exist, there is an extra
section on these exceptional uses. In the case of animate nonhuman (i.e.
animal
14
) antecedents, the authors state the following:
It is generally used when the sex is unknown (cf. Huddleston and
Pullum 2002: 489);
heandsheare more likely with pets, domestic animals, and
creatures ranked high in the kingdom of wild animals (2002: 489;
e.g. lions, tigers, ...);
the use of heor sheindicates a somewhat greater degree of
interest in or empathy with the referent than does it (2002: 489).
It is the third factor that is remarkable, as this is what every native speaker
would say in an impressionistic account and what has been part of socio-
pragmatic gender studies for a long time, but what has not been taken up in
grammars so far.
15
As for the reverse scenario, the use of it with human
antecedents, the authors again combine a traditional commonplace (it can
be used for babies) with modern specifications (used in such a manner, it
tends to suggest resentment or antipathy; 2002: 489).
The other special section concerns the use of she with non-females.
According to the authors, such usage is possible with two categories,
namely with countries (when considered as political, not as geographical
entities) and ships and the like(Huddleston and Pullum 2002: 484):
Ships represent the classical case of this extended use of she, but it is found
with other kinds of inanimates, such as cars. There is considerable
variation among speakers as to how widely they make use of this kind
of personification. It is often found with non-anaphoric uses of she: Here
she is at last (referring to a ship or bus, perhaps), Down she comes (with
she referring, say, to a tree that is being felled). (2002: 484; boldface SW)
The (in)appropriateness of the label personification will be discussed in a
later section (cf. section 5.1). However, it should be clear from the
examples themselves that we cannot be dealing with personification if the
pronoun is not used anaphorically what ispersonified if the referent is not
known? Perhaps an abstract idea or situation (as this is what most of the
relevant instances of she seem to refer to; cf. section 5.3)? Although it is
admirable that the authors mention such uses at all, which are more
widespread than generally assumed, they fail to explain them.
All in all, the descriptions of gender in modern reference grammars are
unsatisfactory in a number of respects. They either do not reflect actual
language use, or they mix traditional with modern interpretation, which
adds to the confusion rather than helps clarify it. If anything, the authors
largely describe an idealized version of gender assignment in written StE.
3.3.2. Descriptions of gender in various works
Gender is not a category that figures prominently in descriptions of PrDE.
Usually it is listed among those categories that have been weakened over
time. Thus, Leisi and Mair (1999
8
: 140141) state that gender in English
has lost much of its weight, primarily because it was a purely grammatical
category without any solid basis in reality. Exceptional feminine and
masculine nouns include names of countries and machines men have a
close emotional relationship with (e.g. motorbike), classified as adopted
natural (psychological) gender (Leisi and Mair 1999
8
: 141, translation
SW). This category is traditionally known as metaphorical gender.
Additionally, there is the allegorical gender of abstract nouns, which,
according to the author(s), is largely based on the gender associated with
the noun in the original classical language. Thus, lovecan be masculine (<
Ltn. amor), peacefeminine (<Ltn. pax).
Figure 1. Gender categories in Brinton (2000)
More recently, Brinton (2000: 105106) follows the dominating view that
PrDE has natural gender as opposed to its earlier grammatical gender.
She notes that gender is generally a covert category in nouns, while a
related category of animacy (animate/inanimate) is not only expressed in
personal, but also in interrogative and relative pronouns (what vs. who;
which vs. who; 2000: 105). Interesting in her classification are the
postulated animacy groupings: humans and higher animals on the one,
he, she; who
humans higher animals
it; which
lower animals inanimates
?
234 Susanne Wagner
lower animals and inanimates on the other hand. Animals thus appear on
both sides of the scale. Although it is not mentioned explicitly by the
author, it should be clear that the cut-off point can vary on all levels of
lectal variation (dia-, socio-, idiolect), depending on the situation, context,
addressee, etc. Figure 1 is an attempt to illustrate Brintons categories.
A description in an almost forgotten grammar of English is encountered
time and again in publications on the status of gender in PrDE. George
Curmes English Grammar was first published in 1925, and to the present
authors knowledge, it served as a basic college grammar.
16
Curmes
introductory remarks on gender read as follows: Gender is a distinction in
the form of words to indicate sex. There are two kinds of gender in English
natural gender and the gender of animation (1962: 209). In the following
paragraphs, the author illustrates natural gender, which rests upon the
conception of sex in nature (1962: 209), with the help of the three
strategies that English employs:
a) use of distinctive terms (generally relational opposites such as man
woman, husband wife),
b) use of affixes (e.g. -essvs. -or),
c) use of pronouns (personal and possessive, i.e. he him his vs. she
her her)
In the context of the latter, Curme notes a tendency to regard animals as
masculine, although they are generally treated as neuter (cf. Curme 1962:
211). What is really noteworthy about Curmes description is his category
of a gender of animation, which he explains as a sort of remnant from OE
times (and its grammatical gender) that has developed a life of its own:
The old habit of associating lifeless things with sex continued and in our
playful moods with their animated feeling still has strong sway (1962:
212). Curmes main point, the contrasting function of animated pronouns,
will figure prominently in later parts of this study. The author distances
himself strongly from identifying animation with personification:
This gender does not as most grammars and rhetorics falsely suppose
rest upon vivid personification, but is merely an animated formto serve as
a contrast to the scientific precision of our normal expression, which treats
as neuter all living and lifeless things which lack personality. (Curme 1962:
213; emphasis SW)
David Crystal is an exemplary case of a scholar who uses personification as
a way out of the dilemma that exceptional pronoun usage poses. He tackles
a problem that will be discussed in detail in the main section of this study
namely why (at least in certain varieties) feminine pronouns are much more
frequently found in neuter contexts than their masculine counterparts:
Many nouns are given variable gender, depending on whether they are
thought of in an intimate way. Vehicles and countries are often called she as
well as it (She can reach 60 in 5 seconds; France has increased her exports).
Pets are often he or she. A crying baby may become it. It is not obvious
why some entities are readily personified while others are not. Nor is it
obvious why most entities are given female personifications. It is not simply
a matter of feminine stereotypes, for she is used in aggressive and angry
situations as well as in affectionate ones: guns, tanks and trucks which
wont go remain she. The only consistently male trend in personification
which the author has heard in recent years is in computing, where word
processors and other devices are widely given male pet names and
pronouns. Why this should be so is beyond him [...] (Crystal 1995: 209)
Although it will be shown in the relevant section (see 5.1) that Crystal
jumps too readily to the conclusion that all of these are cases of
personification, his helplessness seems indicative of most scholars feelings
towards the issue at hand. It is hoped that the discussion presented here will
help shed some light on a largely uninvestigated (and underestimated) area
of PrDE personal pronoun usage.
4. The corpora
The data used for this study stem from two very different sources, namely
(1) the basic and incidental material collected by the SED fieldworkers and
(2) various collections of an oral-history type from the Southwest of
England and Newfoundland. While the previous material consists of
unconnected utterances of sometimes just a couple of words, the latter is in
interview form, so that it is obvious from the outset that different types of
questions need to be addressed to these different sets of data. This section
will provide some basic background information on the individual sources.
In addition, problems that emerge in connection with the composition of
the data (e.g. issues of comparability) will be discussed.
236 Susanne Wagner
4.1. The Survey of English Dialects (SED)
The fieldwork for the SED was undertaken between 1950 and 1961 (cf.
Orton 1962: 14). 311 localities all over England were selected according
to their geographical position isolatively (sic) and relatively to each other
(Orton 1962: 15); agricultural communities with a population of
approximately 500 were preferred. The informants selected are by and large
of the type that in dialectal studies is known as NORMs, i.e. non-mobile,
older, rural males (cf. Chambers and Trudgill 1998: 29). As Orton states,
informants were very rarely below the age of 60, which sets their birth
dates at the end of the 19th century or earlier (cf. Orton 1962: 15). Speakers
who had spent a considerable amount of time away from their home
community were constantly regarded with suspicion (if only in terms of
their linguistic authenticity; Orton 1962: 16). The method chosen for data
elicitation is the questionnaire-based interview, which is one of the direct
methods used in data collection (cf. Francis 1983: 7880). The responses
were taken down in phonetic script in the fieldworkers notebooks. An
example of such a notebook page can be found in Francis (1983: 98).
The sheets [of the notebooks] were divided down the middle. The left side
was reserved for the informants responses and for any remarks or
explanations about them. [...] The right side of the page was intended for
the fieldworkers transcriptions of any significant expressions from the
informants conversation that had relevance to problems under investigation
in the Questionnaire. Relatively unconditioned by the somewhat artificial
circumstances of the interview, this incidental material is particularly
valuable for confirming, supplementing, amplifying or even contradicting
the evidence of the responses themselves. All the fieldworkers made a point
of collecting as much of this material as was feasible in the situation. (Orton
1962: 1718)
The fieldworker notebooks provide an incredible wealth of material for a
study concerned with personal pronouns, a fact that may come as a surprise
considering the rationale behind the SED: of the 1322 virtual questions
that constituted the questionnaire, 387 concern phonological issues, 730 are
concerned strictly with lexical differences, and only 205 questions (128 +
77 or 15.5%) directly addressed morphological or syntactic phenomena (cf.
Orton 1962: 15). It may thus seem inappropriate to use SED material for a
study that is ultimately concerned with a morphosyntactic phenomenon.
However, the make-up of the SED data (both the basic and incidental
material) allows a number of analyses which, although not envisioned by
its makers, are of a morphological and/or syntactic nature. A prerequisite
for such a study is the salience of the feature under investigation.
Except for agreement phenomena, personal pronouns are probably the
only class of items that can be found in the vast majority of responses. This
is largely due to the fortunate circumstance that informants usually did not
restrict themselves to one-word responses, but replied in complete
sentences. Unfortunately, the basic material does not always reflect this. In
the majority of cases, just the tokens relevant to the question are presented
in the published material. Luckily though, the original responses are
preserved in their entirety in the fieldworkers notebooks, which are
accessible to researchers in situ at the University of Leeds.
In the basic material, care must be taken not to overlook possibly
interesting questions/responses due to the overall make-up of the published
data. Although there is a section in the questionnaire that deals with
morphosyntactic items exclusively (book IX), this does not mean that all
other questions are worthless for non-lexicological investigations. J ust to
give one example relevant to the present study, question I.11.6 has the form
How do you empty the cart the quickest way?. The expected response is
to tip. Most informants, however, did not simply reply tip or to tip,
but used a whole sentence, generally in the form (I/we/you) tip it (up),
thus including a personal pronoun (object form) referring to cart.
For this particular question, a pronoun is present in 45 of 66 responses
(68%) in the Northern Counties, while no pronoun was recorded in only 21
cases.
17
For the West Midlands, out of a total of 86 localities, speakers at 54
localities included a pronoun in their response (63%; no pronoun at 34
localities). In the East Midlands and East Anglia, 58% of speakers used a
pronominal form (55 out of 95 localities; no pronoun at 40 localities), and
in the Southern Counties, the figure even climbs to 89% (67 of 75 localities
with, eight without pronoun). The method introduced here will be taken up
in the actual analysis of examples to support certain claims or to give
details of a particular observation.
Used in such a manner, the basic material has more to offer in terms of
morphosyntactic content than may first meet the eye. Still, however fruitful
a search for particular forms in the basic material may be, it is nothing in
comparison with the unsurpassed wealth of data that the fieldworkers
notebooks provide. For the present investigation, a search through almost
80 notebooks from 11 counties yielded a total of roughly 700 pronominal
forms of interest. Although critics claim that studies based on the incidental
material fail to take the context of the utterance into account, such criticism
238 Susanne Wagner
is unfounded in this case.
18
The following paragraphs will explain why this
should be the case with the help of some examples from the notebooks.
Question IX.3.2 belongs to the irregular verbs section, in this
particular case trying to elicit the present, simple past and past participle
forms of find. It is a so-called completing question (cf. Orton 1962: 48)
where the fieldworker is supposed to read out the stimulus sentence and
pause before the key word (in bold print below) to allow the informant to
complete the sentence. This and all further questions are reprinted here
from the published version of the questionnaire (cf. Orton 1962); the
respective contexts are:
He was looking for his knife but couldnt ... findit.
Next day he looked for it again and this time he ... foundit.
He came back looking pleased and told us that he had ... foundit.
Irrespective of the provided context, informants very often chose to
reformulate their response in a variety of manners. Such a reformulation
could result in a more personal format (I found it) instead of keeping to
the third-person singular context, or in a repetition of the trigger sentence,
but once more in the first person (I/we was/were looking for the knife but
I/we could not find it). No matter what the change, the fieldworkers
usually put everything down that did not conform to the prescribed
response 100%. Thus, the following responses to question IX.3.2 can be
found in the notebooks (see Table 1a).
For the published material, only the token actually asked for was
extracted from the response, while all additional information was omitted
i.e. it has been lost to everyone who does not have access to the notebooks.
The policy of inclusion or non-inclusion of a personal pronoun (usually in
the direct object position) in the basic material is unclear. For example, all
of the following questions yielded a comparatively high output of
pronominal forms when the notebooks were consulted, but none of them
made it into the basic material, where only the required form was reprinted
(see Table 1b; the underlined noun is the one that reappears in pronominal
form in the responses).
Table 1a. Responses to question IX.3.2 in SED fieldworker notebooks
county location
infor-
mant
response
19
Cornwall St. Ewe J J D,I48QQ
I found it.
Devon Blackawton J W I #can #find en [Q].
I cant find it.
Devon Kennford J W e #told us ed a-#found en [Q].
He told us (me) hed found it.
Devon Swimbridge GY e #told I ed a-#foun en [Q].
He told me hed found it.
Somerset Horsington TFW He gid en [Q] to en [Q].
He gave it to him.
Somerset Montacute EP 8VNQQQHYUYo,Q Q
We can never find it.
Somerset Montacute EP 8VI48QGQCZHU8VS8WQCG48QWX
We found it where we put it down [to].
Somerset Pitminster CMM e was lookin for en [QC] but e couldn find
en [QC].
He was looking for it but he couldnt find it.
Somerset Wedmore WF (I) couldn(t) see en [QC] nowhere.
I couldnt see it anywhere.
Wiltshire Burbage RH e couldn find en [QC ].
He couldnt find it.
Wiltshire Burbage RH Ive found en [QC].
Ive found it.
Table 1b. SED questions containing pronominal forms
VI.5.11 When I have an apple, I ... eat it.
VIII.5.8 Whats a grave filled in with? Earth.
VIII.7.3 A rubber ball thats punctured wont ... bounce.
IX.1.3 A picture not hanging straight is hanging ... askew.
IX.2.7 A door left like this, you say is ... ajar.
IX.2.9 If a door has been made of unseasoned wood, before long it will be
sure to ... warp.
On the other hand, pronominal forms were included in the following
questions, which also yielded numerous pronominal forms of interest:
240 Susanne Wagner
VIII.7.6 A dog buries a bone because he wants to ... hideit.
IX.2.6 And now [stand sideways in front of it (door, SW)] ... in front of it.
IX.2.8 If the door blew open on a cold day, youd get up at once and ... shut it.
The decision to include or not include items with the same referent (here
e.g. questions IX.2.59, which all concern a door) seems to be based on the
presence or absence of a pronoun in the expected response. However, this
principle is not always obeyed, as question VI.5.11 shows it doesinclude
a pronoun that is not included in the basic material.
Most of the pronominal forms from the fieldworker notebooks that are
relevant for the present study are unambiguous. Although the referent of
the pronoun is not always obvious at a first glance, it can usually easily be
deduced from the respective question or its context. The process of
disambiguating unclear referents involved the following steps:
a) The formulation D,o](,GQCVN,QQo I always ate en skin and
all was encountered at St. Ewe (Cornwall), in book VI;
b) the question that was recorded closest to this remark on the left side
of the notebook page is VI.5.11, which is about an apple;
c) considering the content of the utterance, it seems impossible for the
[QC] to refer to anything other than an apple;
d) as a precaution, the immediately preceding and following questions
were also considered. In this particular case, question VI.5.10 reads
this, where the roots of the teeth are, enquiring about gums, and
VI.5.12 When, in eating, we crush apples or biscuits noisily with
our teeth, we say we ... crunch them. There is no possibility for the
[QC ] recorded to refer to either gums or to have anything to do with
crunch them, making it a 99% certain example of a gendered
pronoun referring to an inanimate noun, namely apple.
Although some ambiguous or problematic cases remain unclear even with
such careful scrutiny, the results from the fieldworker notebooks are overall
very transparent. Unclear cases will be pointed out in the discussion of
examples, usually in connection with issues that have been debated in the
respective literature. From the present authors own experience with the
notebooks, at most 10% of the examples are problematic or ambiguous.
More often than not, referents are even included in the utterance, so that the
possibility of misinterpretation does not even arise. Table 2 lists the
notebooks that have been used for this study.
Table 2. SED fieldworker notebooks used in this study
county ID in SED county locations
36 Co Cornwall 1. Kilkhampton
2. Altarnum
3. Egloshayle
4. St. Ewe
5. Gwinear
6. St. Buryan
7. 7. Mullion
37 D Devon 1. Parracombe
2. Swimbridge
3. Weare Giffard
4. Chawleigh
5. Gittisham
6. South Zeal
7. Kennford
8. Peter Tavy
9. Widdicombe
10. Cornwood
11. Blackawton
31 So Somerset 1. Weston
2. Blagdon
3. Wedmore
4. Coleford
5. Wootton Courtenay
6. Stogursey
7. Stogumber
8. Withypool
9. Brompton Regis
10. Stoke St. Gregory
11. Horsington
12. Pitminster
13. Merriott
(14. Montacute
20
)
38 Do Dorset 1. Handley
2. Ansty
3. Whitchurch Canonicorum
4. Portesham
5. Kingston
242 Susanne Wagner
Table 2. cont.
32 W Wiltshire 1. Ashton Keynes
2. Sutton Benger
3. Avebury
4. Burbage
5. Steeple Ashton
6. Netheravon
7. Sutton Veny
8. Fovant
9. Whiteparish
In addition to the notebooks from the core Southwest, the notebooks for
Hampshire (6 localities
21
), Gloucestershire (7 localities), Herefordshire (6 +
1 localities Lyonshall, only four miles from the Welsh border, was
already visited by Peter Wright in 1952, but was later excluded from the
Basic Material, probably because the fieldworker noted that the Welsh
influence was rather strong), Worcestershire (7 localities) and Oxfordshire
(6 localities) were consulted, resulting in a total of 79 notebooks. Although
the 5 notebooks for Berkshire were also consulted, it was decided not to use
them as issues of legibility made it impossible to draw any clear findings
from the material. A map detailing the exact locations of the SED localities
can be found in Orton (1962: 30).
4.2. Collected material
The second major source of material that was tapped for this study are
interviews from various oral history projects all over the Southwest of
England. Advantages and disadvantages of using oral history material for
linguistic purposes are discussed in detail in Anderwald and Wagner
(forthcoming). For Newfoundland the major source was soon identified in
the archive at Memorial University (Memorial University of Newfoundland
Folklore and Language Archive MUNFLA), where interviews with
natives are stored that were conducted for a range of different studies, both
individual (generally students papers and theses) and project-oriented (e.g.
a collection of Folktales, which was also utilised in the present study).
22
4.2.1. Somerset material
The Oral Archive of the Somerset Rural Life Museum (SRLM) consists of
some 350 interviews to date, recorded between approximately 1973 and the
present, and is continuously expanded whenever someones interest in a
certain subject is peaked. The material for this study was selected in
reading through transcripts and listening to stretches of the tapes to find out
if the speakers were dialectal enough for the project at hand. In the end,
the Somerset corpus consisted of 30 interviews from the SRLM Oral
Archive and one interview from the private collection of one of the Friends
of the Museum. The statistics for the SRLM data can be found in Table 3.
Table 3. SRLM material
number of interviews 31
number of speakers
23
36
number of words (total) 174,500
number of words per interview (average) 5,629
number of words per interview (maximum) 17,291
number of words per interview (minimum) 1,671
interviews recorded from to 19681996
recording dates (details)
24
1960s (1x); 1970s (7x); 1980s
(20x); 1990s (3x)
informants ages from to 7089
informants ages (decades) 70s (10x); 80s (15x)
informants born from to 18841918
informants born (decades) 1880s (3x); 1890s (7x); 1900s
(10x); 1910s (5x)
4.2.2. Wiltshire material
Although the Wiltshire Folklife Society had recently been dissolved when
our interest in suitable dialect material started, we were able to get access to
some of the material. Three of the interviews in very traditional Wilts
dialect are used here (details in Table 4). As copies of the tapes were not
available, this material has to be taken with a grain of salt. Although
speakers seem to be typical NORMs, born around the turn of the century,
they show features that were rare even in SED times, such as the use of thic
as demonstrative pronoun. Thus, these interviews are not taken to be as
244 Susanne Wagner
authentic as the rest of the material and will be treated accordingly in the
analyses.
Table 4. WFLS material
number of speakers 4
The second source for Wiltshire material was found at the Trowbridge
Museum and its oral archive. The archive primarily holds a collection of
interviews about the history of weaving, the primary industry in
Trowbridge until recently. The interviewees are former workers in the
weaving mills in and around Trowbridge, and this is one of only few
locations for which at least some data for females (three speakers) is also
available, as many of the jobs in the mills typically were womens. The
dialect is very strong in most speakers selected for this study; others who
showed a more standard pattern of speech were disregarded. Most of the
traditional features associated with Wiltshire or more general West Country
speech are present in those speakers. Details for the Trowbridge material
can be found in Table 5.
Table 5. Trowbridge material
recording dates (details) 1987; 1992 (2x); 1994 (2x)
informants ages (details) 66, 71, 81, 85
informants born (details) 1907, 1909, 1913, 1923, 1926
4.2.3. Devon material
The material from Devon used in this study stems from the oral history
material of the Totnes Community Archive at the Totnes Elizabethan
Museum. The collection of tapes is large (ca. 300), but as no
comprehensive catalogue exists and only about 10% of the interviews are
transcribed, deciding what to select was difficult. Interviews with existing
(standardized) transcripts were preferred in the initial stages of the selection
process, replaced by time-consuming listening-in on the interviews at later
stages. Details for the interviews, mostly recorded in and around Totnes, that
have been selected for this study are provided in Table 6.
Table 6. Devon material
recording dates (details) 1984; 1985 (2x)
informants ages (details) 76, 79, 82
informants born (details) 1902, 1906, 1909, 1910, 1913
4.2.4. Cornwall material
When first researching West Country speech, it appeared that Cornwall was
a negligible area in terms of traditional dialect to be found there. No
accounts from the 19th century comparable to Elworthys studies of
Somerset speech exist, and secondary literature in general is scarce to non-
existent. The successor of the Institute for Cornish Studies, the Cornish
Audio and Visual Archive, held a large amount of material suitable for the
present study. Unfortunately, almost none of it was transcribed, again
making the selection process very time-consuming. In addition, the
transcription of the material turned out to be a particularly difficult
enterprise, as those interviews prove that modern and traditional Cornish
dialect is much better than its reputation the material from Cornwall is
much more dialectal than is to be expected from the literature. It has always
been claimed that Cornwall is much closer to StE, as the result of the much
shorter history of English in the area. Accordingly, dialects did not have
time to develop, and the area was much more influenced by school English
246 Susanne Wagner
rather than traditional West Country speech recall Wakelins position on
this (Wakelin 1975: 100, quoted in full in note 5 on page 372).
In reality, nothing could be further from the truth. Being familiar with
the relevant literature, one gets the distinct impression that researchers
tended to over-generalize the situation they found in the phonological
system and the lexicon, where not as many traditional dialect features /
words were found as was expected see, for example, Fischers
lexicological description (1976: 298) or Bremanns socio-phonological
study (1984: 153; cf. Wagner 2004a). An overview of the interviews
selected here, all from West Cornwall (Pendeen, Gurnards Head, Zennor
and St. Ives) is presented in Table 7.
Table 7. Cornwall material
interviews recorded from to 19781980s
recording dates (details) 1978 (2x); 1979; 1980s
informants ages (details) 74, late 70s, 80s, 86
informants born (details) 1892, 1895, 1901?, 1904
4.2.5. No material fromDorset?
Although every attempt was made to get hold of material from Dorset, none
was successful. The Dorset Record Office holds a lot of material that
seemed very promising on paper, judging from catalogue card copies. The
reality, however, was rather disappointing all transcripts that were
consulted were StE, and none of the tapes listened to were more than
slightly dialectal. Despite serious efforts to remedy the lack of data from
Dorset (all libraries and local museums were contacted in search of relevant
material, without any result), no good source for comparatively modern or
even traditional dialect material could be found. At that time, the
disappointment was great. However, it is relativized to quite some extent
when considering the equally small in comparison with the other
Southwest counties almost non-existent amount of really dialectal data
that the SED fieldworkers were able to gather (cf. the relevant paragraphs
in Sections 8 and 9).
4.3. Newfoundland material
4.3.1. Material fromthe Memorial University of Newfoundland Folklore
and Language Archive (MUNFLA)
MUNFLA is listed as a major resource for research in Michael Linns
collection of archives (1993: 444), and a first visit to assess the available
material confirmed this. The sheer amount of material at MUNFLA is
amazing: according to a list prepared by an archival assistant in 2000, there
are more than 2,000 transcribed interviews, collected from the early 1960s
to the present day. The interviews stem from various sources, but most of
them are the results of students and staffs research for papers, theses, or
other publications.
The Folklore Department encourages students to deposit their tapes and
papers at MUNFLA, and as it is the policy of MUN that all students,
regardless of their major, should take at least one class in Folklore, the
variety of material is unsurpassed. Topics range from gardening to ghost
stories or a recording and description of traditional songs and music this
short list should suffice to give an impression of the wide array of tackled
topics. Students usually went to their home communities to do the
interviews, which is probably the greatest advantage of the MUNFLA
material: The interviewers were almost always insiders, often talking to
relatives or at least acquaintances. The informants were thus not as
inhibited as is often the case when an outsider, least of all a non-native,
intrudes on them.
Another major advantage of MUNFLA is that many of the interviews
held in the collection are transcribed
25
, and although some are standardized,
the transcription policies at MUNFLA have long been such that dialect
features are to be presented as accurately as possible. Moreover, the
transcripts were all done by professional transcribers, though not trained
linguists. Even if an earlier version of the transcript existed (generally done
by the student who had originally submitted the paper), the professional re-
transcribed it. Thus, the standard of transcription is very high, and in the
present authors experience does not leave much to wish for.
248 Susanne Wagner
Originally, material from 17 collections had been requested, but as not
all interviews were equally suitable for the present project, some were
disregarded in the final corpus of Newfoundland English. The present
corpus consists of 31 interviews with 34 speakers, totalling about 130,000
words, as detailed in Table 8.
Table 8. MUNFLA material
interviews recorded from to late 1960s to early 1980s
informants ages (details) 27, 31, 49, 60s (5x), 70s (5x), 80s
(6x)
informants born (details) 1880s (4x), 1890s (6x), 1900s
(5x), 1920, 1938, 1942
4.3.2. Folktales of Newfoundland
The collection of Mrchen published in 1996 in the two-volume Folktales
of Newfoundland is another valuable source for studies on Newfoundland
English. It differs from the rest of the material used in this study in one
important aspect: the telling of a tale constitutes a different discourse level
than an oral-history-type interview. It will be interesting to see how far the
genre influences the choice of language used when investigating gender
diffusion in detail. The two authors/editors, Herbert Halpert and J ohn
Widdowson, are also the primary collectors of the more than 150 tales
included in the books. While Halperts background is in folklore,
Widdowson is a linguist with a long-standing interest in dialects. The
authors became interested in the oral tradition of Newfoundland, which
ultimately resulted in the publication of Folktales. For practical purposes,
they ultimately used tales that had been collected between 1964 and 1979
(cf. Halpert and Widdowson 1996: xii). Although neither Halpert (born in
NYC) nor Widdowson (an Oxford graduate) were natives, they were able
to record the tellers without much inhibiting influence.
The reasons for peoples openness lies primarily in the relative isolation
of most of the communities that were visited, where easy access to mass
transportation was still a couple of decades away. The major advantage of
the tales is clearly their presentation in orthographic transcription. As the
authors point out, and as should be obvious to anyone who has ever dealt
with transcribing non-standard speech, the final results can only be a
compromise between all possible extremes. For some issues, the folklorists
point of view was prone to clash with the linguists, and vice versa. But
however difficult the actual production was, the published versions of the
tales are a dream for anyone interested in morphology or syntax.
Despite these and other necessary compromises in our editing procedure,
we have constantly borne in mind our intention to present a text as close to
the original speech as is both possible and practicable, always erring on the
side of accurate representation rather than on the kind of editing which, in
both obvious and more subtle ways, changes the original text radically in its
insistence on presenting a text more acceptable from the literary viewpoint.
We have retained the original lexis, grammar, and syntax because these are
essential to the dialect, particularly since they also reflect the regional
character of the tales. (Halpert and Widdowson 1996: lixlx; emphasis
SW)
An actual stretch of a tale is reprinted below:
[Int. B: Who were the best storytellers around?]
Uh well I dont know who the best one I dont know who the BEST
storyteller is. Well ol S[mith] now Eli Smith hes up there to Port
Anson. I think he was about so good a feller as ever I heard (could) tell
stories you know.
[Int. A: He used to tell these about J ack as well did he?]
Oh yes. He used to tell em (right) bout J ack as well.
[Int. A: An youve picked them up when you heard people in the woods
and elsewhere have you?]
Oh yes. When I whenever whenever I hear a story told see that I pick
un up Id knowed un. I could tell un then right on after he was
finished. (Tale No. 32, p. 342343)
The original tapes are held at MUNFLA today, and the authors are
considering making them accessible for researchers. Detailed background
250 Susanne Wagner
information is generally available on the tellers, making it easy to single
out informants with a West Country family background. The storytellers
are traditionally males over 60 who work(ed) in the fishery or lumber
industry, and thus could be termed the Newfoundland equivalent to
NORMs. The procedure for selecting tales did not differ much from that of
assessing other oral history material. Statistical details for the tales corpus
are provided in Table 9.
Table 9. Folktales material
number of words per interview (minimum) 401
4.4. Summary
Some of the issues that may seem problematic due to differences in type,
selection, or general make-up of the data need to be addressed. First, it may
be argued that the SED material (both the basic and notebook material) is
not comparable to the oral-history-type material, as the SED was
questionnaire-based while the oral history data are usually one-on-one
interviews. It should be mentioned initially that the type of analysis pursued
here does not depend on long stretches of discourse. Often enough, a single
phrase (and sometimes even an individual form) exemplifies the type of
language use under scrutiny here. In addition, the differences are by no
means as great as is generally thought. The atmosphere of responding to the
SED questionnaire could not have been that different from a more general
interview. In both cases, the interviewer(s) sat face-to-face with the
informant(s). The only difference is that the contributions of the
interviewer(s) are pre-determined in the case of the SED, but free
(conversation) in case of oral history interviews. Also, the primary interest
here is in the notebooks contents, which are usually side remarks that have
nothing to do with the actual questionnaire answering. In style, these
remarks come very close to an oral-history-type setting, which justifies a
comparison of these data with those extracted from oral history material.
26
Second, it could be argued that the time frames of the SED recordings and
the oral history material differ too much for them to be comparable. As has
been mentioned above, fieldwork for the SED took place between 1950 and
1961, and speakers were generally in their sixties or seventies then, setting
their birth dates in the 1880s to 1900s. As will be recalled from the respective
tables listing the birth dates of the oral history informants, these are identical
or at least very close to those of the SED informants. Thus, we are essentially
dealing with the same generation of speakers, an issue that will become
important in later sections. As for the different oral history data, there is no
obvious reason why they should not be comparable. Informants are generally
NORMs, female speakers are scarce. Table 10 gives an overview of the
corpus material that will be used in the analyses in this study.
Table 10. Corpora used in this study
area / source number of interviews number of words
Newfoundland
Folktales 55 156,000
MUNFLA 31 132,000
Southwest
Wiltshire 9 88,800
Somerset 31 174,500
Devon 5 51,100
Cornwall 4 18,900
total 135 621,300
5. Special referent classes
In the course of this study, it has been and will be mentioned time and
again that certain nouns which may trigger gendered pronouns deserve a
special status. In addition to those nouns, a specific use of (particularly
feminine) pronouns merits a closer investigation. The two major categories
to be identified in this respect are instances of personification and
references to animals. The specific use is subsumed under the label non-
referential she. The reasons for excluding those two categories and
assigning special status to the third use are manifold and complex. This
section will provide background information on all three and explain why
252 Susanne Wagner
they are not treated here at all (personification and animals) or analyzed
separately (non-referential she).
5.1. Personification
By definition, personification is classified as a figure of speech which
attributes human qualities to non-humans and things (animals, plants,
elements of nature, and abstract ideas). Although superficially easily
mistaken for instances of personification, almost none of the examples in
the present study really are. The entry for personification in The New
Fowlers Modern English Usage (Burchfield and Fowler 1998; available
online) is interesting insofar as it links the loss of grammatical gender to the
rise of personification, giving examples from the OED:
Personification arises partly as a natural or rhetorical phenomenon and
partly as a result of the loss of grammatical gender at the end of the Anglo-
Saxon period. In Old English a pronoun used in place of a masculine noun
was invariably he, in place of a feminine noun heo ( =she), and in place of
a neuter noun hit ( =it). When the system broke up and the old grammatical
cases disappeared, the obvious result was the narrowing down of he to refer
only to a male person or animal, she to a female person or animal, and it to
nearly all remaining nouns. At the point of loss of grammatical gender,
however, he began to be applied illogically to some things personified as
masculine (mountains, rivers, oak-trees, etc., as the Oxford English
Dictionary has it), and she to some things personified as feminine (ships,
boats, carriages, utensils, etc.). For example, the Oxford English Dictionary
cites examples of he used of the world (14c.), the philosophers stone
(14c.), a fire (15c.), an argument (15c.), the sun (16c.), etc.; and examples
of she used of a ship (14c.), a door (14c.), a fire (16c.), a cannon (17c.), a
kettle (19c.), and so on. At the present time such personification is
comparatively rare, but examples can still be found: e.g. Great Britain is
renowned for her stiff upper lip approach to adversity; I bought that yacht
last year: she rides the water beautifully; (in Australia and NZ) shes right;
shes jake; shes a big country, etc. (The New Fowlers Modern English
Usage, personification; boldface SW)
One has to distinguish between personification in its own right and
personification as sub-component of metaphor (as in the mouth of the
river). It should be obvious that, while the second use occurs frequently in
everyday speech, particularly in idioms and proverbs, the discussion here
only concerns the first use which is, as mentioned above, rare.
27
In the corpus data used here, clear examples of personification are
generally restricted to the telling of myths and legends, where they are
typical of the genre. Personification is an extremely unlikely explanation
when a cider maker uses heto refer to an apple, when a watchmaker refers
to one of his watches, or a house mover to one of his houses. These are
examples of true dialect use, based on a system that has nothing to do with
personification. This can also be deduced from the provenance of
masculine pronouns in these domains, while personification has been
associated with feminine forms (recall Crystal 1995: 209) and the Spoken
Standard system(s) described in Section 6.
According to this and other sources (e.g. The Oxford Companion to the
English Language, online at www.xrefer.com; Wales 2002: 331; McArthur
1992, personification), the representation of ships as female is also an
example of personification, probably based on the imagery of a ship as a
womb-like container. However, one should hesitate to classify all of the
literally hundreds of examples of feminine (though not female) ships in the
Newfoundland data of this study as instances of personification. A
Newfoundland fisherman would simply never use it to refer to his ship, and
personification is more than unlikely when no standard forms are used.
Other arguments are more plausible and convincing than personification,
and it does definitely not suffice as an explanation accounting for gendered
pronouns in general. Summarizing various researchers work on gender
marking, Wales (2002: 333) argues similarly:
Personification is obviously too general a label to cover what seem to be
quite complex analogical or metaphorical hierarchies of salience according
to such value(s) as occupation, local environment and climate and general
relevance to human needs, as well as subtle forms of gender symbolism.
5.2. Animal referents
At first glance, the major (and only) reason for excluding nouns referring to
animals from the discussion of gendered pronouns is fairly simple, though
maybe not obvious: Although most grammars of modern and earlier stages
of English claim that it is the appropriate pronoun to use when referring to
an animal, except for cases where the sex of the animal is known (in which
case the pronoun corresponding with the sex of the animal may be used
alternatively), actual language use could not be further removed from this
254 Susanne Wagner
prescriptive statement. When looking at speakers behaviour, it seems that
even at a very superficial level forms of heand sheby far outnumber
instances of it in everyday casual speech.
In the corpora used here, there is essentially not a single example of it
referring to an animal, while literally hundreds of masculine and feminine
pronouns can be found. In the West Country corpora, for example, there are
about 500 references to animals with masculine or feminine pronouns, but
only a handful with neuter forms. Though surprising at first, a more
detailed investigation of additional corpus data and a number of studies
dealing with the issue reveals that the observed pattern is the rule rather
than the exception.
MacKay and Konishi (1980) investigated the use of what they call
human pronouns (i.e. he, him, she, her) to refer to non-human
antecedents. Though outlined as a study dealing with personification, it
soon turned out that personification played only a minor or no role at all in
those cases where non-standard pronouns were chosen.
The authors based their study on a database of approximately 35,000
pronouns collected from an anthology of childrens literature (cf. MacKay
and Konishi 1980: 151). They distinguished three large classes of
antecedents, namely animals (including real, imaginary, and toy animals),
fantasy creatures (including imaginary beings such as fairies, ghosts, giants,
and trolls), and things (including abstractions such as thought and time).
The first major result from these counts was highly unexpected in light
of prescriptive grammarians eyes: Of the approximately 450 pronominal
references to animals, more than 80% were masculine or feminine he
(including object him, possessive his and reflexive himself) occurred in
62% of cases, she (her, hers) in 20%, and it (its) in only 18% of the
examples (cf. MacKay and Konishi 1980: 151). Next the authors classified
the pronouns according to whether or not the antecedent was personified,
assuming that personification would play a significant role in triggering
non-neuter pronouns. Although this was found to be true in general, the
figures for the non-personified instances are striking (compare the
percentage columns in Table 11).
Within the class of animals, personification could be held accountable
for the use of a human pronoun in only about half of the cases (234 of
452). In the non-personified cases, a human pronoun was recorded in
more than two thirds (69%) of the examples a figure that clearly shows
how rarely it is really used to refer to animals, here in only 31% (68
examples) of cases. The figures for the other two classes, on the other hand,
are clearly within the expected norm. All examples of fantasy creatures
being referred to by heor sheare instances of personification, and in only
six cases did speakers use a gendered pronoun to refer to things.
Table 11. The use of he and she vs. it for non-personified antecedents (from
MacKay and Konishi 1980: 152)
Pronoun used
Total he andshe it
Nature of antecedent N N % N %
Animals 218 150 69 68 31
Fantasy creatures 0 0 0 0 0
Things 26 6 23 20 77
Total 246 156 64 88 36
In another study, Marcoux (1973) investigated students use of personal
pronouns in tag questions. Tested noun classes included countries, ships,
animals and humans. Like MacKay and Konishi, Marcoux found
surprisingly high figures of human pronouns referring to animals of
unknown sex. Some of the sentences that were used in this study are
reprinted below (2ad), together with the pronominal forms that were
recorded in the tags.
(2) a. My dog will eat anything. he 88, it5, she3, aberrant 12
b. That cat looks hungry. it 46, he 43, she 9, he/she 2, aberrant 8
c. This canary sings beautifully. it 69, he 23, she 7, he/she/it 1, aberrant 8
d. Tweety, my parakeet, is sick. she 42, he 40, it 14, he/she 2, aberrant 10
Two conclusions can be drawn from these observations: First, [t]he
presence of a proper noun seems to encourage the use of either a masculine
or a feminine pronoun rather than the neuter form (Marcoux 1973: 104).
And second, the masculine pronoun is highly favoured over the feminine
one.
28
Table 12. Pronouns for antecedent dog in a sample from the BNC (spoken)
N %
masculine form 162 56.6
feminine form 23 8.0
neuter form 101 35.3
total 286 99.9
256 Susanne Wagner
Table 13. Pronouns for antecedent cat in a sample from the BNC (spoken)
N %
masculine form 88 52.7
feminine form 38 22.8
neuter form 41 24.6
total 167 100.1
A cursory analysisof personal pronouns referring to animal antecedents in
the spoken part of the BNC reveals the same pattern
29
: In the contexts of a
search for dog and cat, all pronouns referring to the keyword were
marked. Results can be found in Table 12 and Table 13 (percentages are
rounded to first decimal so that figures may not add up to 100%)
Although slightly different from Marcouxs results, the overall
emerging pattern is identical: Masculine pronouns are the unmarked choice
when referring to a pet such as a cat or dog. While it can be assumed that
most of the instances of feminine pronouns referring to dogs are used by
speakers who know the sex of the dog in question, cats are more likely to
beshes generically, based on the biological-semantic pattern (dog=neuter
or +male, vs. bitch; cat =neuter or +female, vs. tom-cat).
Pronoun switches are typical, and a number of emotive factors play a
role in the choice of pronouns when referring to animals. For example, the
a cat owner is very likely to refer to a dog chasing their cat as it rather than
heor she, signifying their emotional attitude or intimacy towards their cat,
but at the same time signalling distance towards the dog (see also Mathiot
and Roberts 1979 and their idea of up- and downgrading in section 6.1.1).
The reverse pattern naturally holds for dog owners. Some representative
examples from the BNC can be found in (3ae).
All of the examples in (3) highlight certain aspects of pronominal use
connected with animal referents. In (3a), a police officer (PS1SF) is being
questioned about dogs on the force. He himself has never owned such a
dog, which, in addition to the rather formal situation of the discussion,
should explain his four uses of it. Once he gets emotionally involved
though, talking about a dog becoming a member of the family of the
leading officer, he switches to hein the two final references.
In (3b), a farmer (PS2VX) is talking about hunting foxes. Although
reporting a rather general procedure (One dog would go in ...), the
speaker obviously has one specific dog in mind, which explains his use of
shein all instances.
Speaker (PS555) in (3c) has an obvious antipathy towards small dogs,
such as a friends Chihuahua. Both the negative feelings and the animals
size are responsible for the choice of pronoun it in all but one instance,
where the speaker uses a masculine pronoun, most likely referring to the
true sex of the dog in question.
In (3d), the owner of the cat (PS1D1) uses masculine pronouns
exclusively, while her friend (PS1CX) only uses neuter pronouns, a typical
pattern signalling familiarity or ownership. Similarly, the speaker in (3e)
only uses feminine pronouns, most likely in accordance with the sex of his
cat.
(3) a. ... Alright? Next question. Yes young man.
[PS000]: What was it like when you had your police dog?
[PS1SF]: I have never had a police dog. Ive never had, never been on
er the special course
30
. A lot of people like it ... because basically th er
when you look after a police dogit becomes your pet as well, you take
it home with you and you take it to work with you, and the u youll
have a police dog for sort of like its working life of seven to eight
years, so basically youre gonna have himfor seven to eight years and
hebecomes a fa like a family pet. Ive never been on the course (i.e.
force; SW) so Ive never had a police dog. FM7 (257)
b. [PS2VX]: Aye. Aye. And ermsay the fox had been in the ground, and
the [...] and the the young cubs, for about three or four days. And we
used to hear somebody saying there was a vixen there and some and
some young ones. [...] we went up there with the dogs and let themin
in to the burrow. Block everywhere, let theminto the burrow. One dog
would go in, and shed just shake her tail and come back, and you
couldnt get her in afterwards because she knew that theyd cleared
off. HER (217)
c. [PS555]: I couldnt stop laughing. The little dogs going [yelping
sound] [panting]
[...] This little dog was mad, man, did you see it? I t was so ugly I
wouldve
[PS55A]: Yeah.
[PS555]: kickedit if I saw it. Same as Chriss chihuahua. Id, Id love
to kick it. Id love to kick her dog. Hes so tiny! I feel so sorry for it
you know, up at that house with all thembig fat balls of, of fat.
Theyve probably stepped on it enough times.
258 Susanne Wagner
[PS55A]: [...]
[PS555]: Andhave you heard it crying at night?
[PS55A]: Mhm. KPG (4043)
d. [PS1D1]: Come on puss, shh, shh, shh
[PS1CX]: Wheresit gone Rebecca? Wheres pussy cat?
[PS1D1]: puss, puss, puss, puss
[PS1CX]: [laughing] Wheresit gone [...]?
[PS1D1]: ishethere?
[PS1CX]: Canyou see him? ... Can you see him?
[PS1D1]: Wheres the cat?
[PS1CX]: Go on out, out cat [shooing away]
[PS1D1]: [laugh] whereshegone? KB9 (1084)
e. [PS0H9]: I know, woke up this morning she was, she was obviously
cold, cat was right under the covers, snuggled right up to me and got
her, her chin on me armlike that, I was asleep ... KCY (438)
Clearly, pets are more likely to be hes than shes or its when their sex is
unknown, hethus (still) serving as generic pronoun despite the arguments
of recent feminist linguistics theories. Generally, researchers agree that
personal involvement seems to be the most relevant factor in pronoun
choice.
[T]he use of he and she seems to signal personal involvement or
empathy for the referent in the case of [...] an owner of an animal,
someone who is emotionally attached or values the referent, [...] or
someone attached to a specific animal. By the way of contrast, the use of it
seems to signal lack of involvement or empathy with the referent in the
case of [...] [a speaker] who is not personally attached to the referent or
wishes to devalue it, an entity which is acted upon, and finally a nonspecific
animal or class of animals with which personal involvement is out of the
question. (MacKay and Konishi 1980: 155156; boldface SW)
The cut-off point within the class of animals differs from speaker to
speaker, depending on profession, environment, or similar factors. For
someone who grew up in the city and has never lived in the countryside, it
is highly probable that only pets, or even just dogs and cats, can be heor
she, whereas a badger or fox, never having been encountered in their
natural habitat, will be an it. On the other hand, it is extremely likely that a
farmer will refer to the animals on his farm as heor she, that a hunter will
refer to the hunted animal, the fisherman to the fish in his catch as he.
It should thus be concluded that the prescriptive rules in grammars
concerning anaphoric pronouns to be selected to refer to animals are not
reflected in everyday conversations. As some degree of personal
involvement is usually present when speakers talk about animals, neuter
pronouns are the least expected forms. Pets will be its only derogatorily or
to signal detachment, while the status of wild animals depends to a large
extent on the speakers background. Other factors that may influence
pronoun choice are saliency of the animal in the discourse (centrality in
MacKay and Konishi 1980: 155), size (the bigger the more likely he;
MacKay and Konishi 1980: 153), and various (supposed or real) character
attributes (brave, wise =male; weak, passive =female etc.; cf. MacKay and
Konishi 1980: 154 and also section 6.1.1 on Mathiot and Roberts 1979).
5.3. Non-referential she
This category includes many examples that are mentioned elsewhere in this
study. When investigating the use of gendered pronouns, examples like
those in (4) occur with a regularity that warrants a closer analysis.
(4) a. Okay! J ulia yelled. Get ready. Here shegoes!
Timber!
The post toppled slowly and as it landed on the grass with a thump,
they both cheered.
Woman (J ulia) and her daughter are removing the posts that held a
rope rail; reference seems to be to the situation as a whole rather than
an individual post; Nicholas Evans, 2002 (Corgi ed.), The Smoke
J umper, p. 401.
b. Watch out! Here shecomes!
(speaker is sea-sick; Svartengren 1928, ex. 139)
c. Hereshecomes!
(Paddock 1991: 30, referring to an approaching weather front)
d. Shes blowing hard out there. It almost blew over the tree.
(Mathiot and Roberts 1979: 38)
31
e. Well HE done it well say, th (other) one had to do it. He climbed up
you know an just as he got on the top part away shegoes helter to
skelter all over the ki ... all over the place.
(Folktales 141; reference to situation of a pile of chairs falling)
260 Susanne Wagner
f. I only come down to help pull en [house]. But nevertheless I said, all
right boys, straighten out, and away dey goes, my son, straightened
out, we took dat house and here shecome.
(MUNFLA 72-089: C1187; cataphoric reference to what happens)
g. Stay the night with us? Ah, shes right. all is in order
(R. Stow, 1963, cited in Ramson 1988: 531)
h. Shes fine; shes cool; shell be joe. synonyms of It doesnt matter.
(Orsman 1997: 717)
i. Well ... it rolled in at my feet and hed pulled t pin out! I got out o
that hole faster than I went in, and up shewent!
(FRED Yks_003; explosion caused by a grenade it)
All of these examples have one thing in common: The referent of the
personal pronoun is either difficult to identify or cannot be named at all.
Very often, sheseems to refer to the (abstract) situation, circumstances, or
side effects of the utterance rather than to a concrete thing. Items of this
type can be found in all varieties of English, pointing to the fact that this
use is restricted neither regionally nor socially. For that reason, non-
referential she does not fall within the framework of this study.
Nevertheless, some of the characteristics of these forms will briefly be
outlined, since descriptions of this use are difficult to come by.
One of the major characteristics shared by most of these constructions is
word order: More often than not, extraposition results in an output of the
form X-S-V instead of standard S-V-X. X is usually a spatial or
demonstrative adverb, most often here or there. Alternatively, the
preposition of a prepositional verb is extraposed, resulting in patterns such
asup she V or down she V.
J udging from the relevant literature, this type of fronting seems to be
rare in English. Birner and Ward (1998), who analyzed pre- and postposed
non-canonical word order patterns, do not mention this construction type.
An analysis in terms of theme/rheme or given/new information is difficult
in most cases. The fronted element, although usually containing new
information, is generally not the topic of the respective utterance. Matters
are further complicated by the fact that expressions such as here/there/PP
she V seem to assume an almost idiomatic meaning, making it impossible to
attribute any type of information status such as theme/rheme or
topic/comment to the individual elements at all.
Moreover, the verb is always in the present tense, although the action
described would generally demand a progressive form. It seems that the
need for fronting/extraposition overrides any aspect requirements. The
origins of this type of use remain in the dark flattering (or unflattering)
though it would be, it is definitely not enough to claim that the situations in
question show some feminine characteristics, as folk belief has it. It is
probably true that most speakers who use non-referential shes are not aware
of it, but that the construction seems to have established itself in everyday
English conversations, and be it only because There she goes sounds much
better than There it goes. From the present state of affairs, one can only
conclude that further investigation is needed.
6. Non-dialectal studies of gender assignment
6.1. American English
6.1.1. Sociological perspective (M. Mathiot)
In their 1979 article, Madeleine Mathiot and her assistant Marjorie Roberts
investigated the use of referential gender in American English. They
assumed that patterns of speakers pronoun use reveal certain sex roles as
manifested in language. Their approach is a sociological or even
psychological rather than a purely linguistic one, and attitudes and mental
representations are used to explain language use. viewpoint
Data were collected for a period of 10 years, resulting in two subsets,
one from the Los Angeles (years 1-3.5) and one from the Buffalo area
(years 3.5-10). The examples stem from informal face-to-face
conversations (cf. Mathiot and Roberts 1979: 5). The authors do not specify
whether they elicited their examples with the help of some priming
sentences, or whether they occurred in natural discourse.
Mathiot and Roberts distinguish the standard pattern of referential gender
(normative pattern) from the intimate pattern, which allows the use of he
or she for an inanimate entity
32
or of it for a person. While the normative
pattern predicts constant use of one pronominal form, in the intimate pattern,
the same entity may be referred to with either one of the three pronominal
forms by the same speaker (Mathiot and Roberts 1979: 7).
At a very early stage, even before the actual analysis, the authors offer a
generalization which they think explains the differing uses of he, she andit
in the intimate pattern: The choice of he, she or it depends on the
speakers general attitude towards the entity referred to or his feelings of
the moment towards that entity. (Mathiot and Roberts 1979: 7; emphasis
SW). Although details of Mathiots and Roberts analysis are debatable
262 Susanne Wagner
(some minor differences as well as some major contrasts in their
assumptions compared to the present authors analysis will be pointed out
below), it should be mentioned here already that the general tenor of the
article is very similar to findings from this study.
6.1.1.1. The intimate pattern
As in the standard variety, the intimate pattern manifests two basic
oppositions in pronominal gender: heand shevs. it on the one and within
that opposition he vs. she on the other hand. According to Mathiot and
Roberts, the first contrast can be attributed to (semantic) upgrading(it he,
she) or downgrading(he, she it). While the present author does not share
the authors sentiment that upgrading in general corresponds to
personification (cf. section 5.1), their association of positive involvement
on the part of the speaker seems a good means of tackling the issue (cf.
Mathiot and Roberts 1979: 11).
33
Similarly, negative involvement is said to
underlie instances of downgrading, which also extends to those cases of
previously upgraded items (i.e. return to the standard pattern).
Mathiot and Roberts evidently did not expect to encounter instances of
the intimate pattern at such a high frequency. As a consequence, they may
over-generalize to some extent when they say that [i]t seems that any
nonhuman entity can be referred to as either he or she, i.e. upgraded,
without regard to its nature (Mathiot and Roberts 1979: 11; emphasis
SW). While the contrast between it on the one hand and heandsheon the
other is relatively straightforward, much more variation occurs within the
intimate pattern, i.e. between heandshe. The authors differentiate between
mens and womens usage, as they assume that certain patterns of thought
manifest themselves in the intimate pattern. A detailed description of mens
and womens mental images of both themselves and the other sex can be
found in the respective tables of the article (Mathiot and Roberts 1979: 14
and 16). A brief summary of their findings will follow.
In general, Mathiot and Roberts adopt a rather archaic picture of men
and women and the values and ideas that are part of their heritage. Each of
the sexes occupies one of the opposing ends of a continuum in each of the
relevant categories, such as beautifulugly, incompetentcompetent,
challenge/rewardbrave, prized possessiongood-natured, and mature
infantile. While the first are traditionally associated with women, the
second apply to men, at least according to the popular clich.
Although many examples are provided to illustrate all of the categories
that Mathiot and Roberts mention, their work is rather impressionistic, and
it seems as if theirs is not a clear-cut system of pronominal use, but rather
an interpretation of more or less incidental facts. Under different
circumstances, most of the examples could be interpreted differently. Also,
the article discussed here was written and researched in the 1970s, and both
authors are women, a fact one should not neglect. Although not (yet)
established then, the sentiments expressed here are reminiscent of feminist
linguistics concepts. Reading between the lines, a certain dissatisfaction
with the current state of affairs can be recognized in statements such as the
following:
Men define themselves both independently of their relationship to women
and in terms of it. Women define themselves only in terms of their
relationship with men. [...] Men regard themselves as intellectually superior
to women. Women regard themselves as emotionally superior to men.
(Mathiot and Roberts 1979: 2728; italics original)
6.1.1.2. Summarizing the results
One of the major drawbacks of Mathiots and Roberts study is its lack of
quantitative or statistical information. It is unclear from the article how
many instances of gendered pronouns (or of the intimate pattern) the
authors actually found in their data. The examples given in the analysis
itself add up to approximately 130, with masculine and feminine forms
distributed fairly evenly. That figure itself should cause suspicions as to the
overall figures, as it is highly unlikely that gendered pronouns are
distributed evenly between male and female speakers. Taking together the
relevant forms of the appendices (excluding animals), the following picture
emerges: Men use sheor her about 40 times to refer to an inanimate entity,
while not a single use of a masculine pronoun is mentioned. Women, on the
other hand, use masculine pronouns about 60 times to refer to something
inanimate, but there are also approximately 10 examples of women using
feminine pronouns in the same context.
The most interesting conclusion to be drawn from these data is the
following: If taking the examples presented here as representative of male
and female use respectively, a clear picture emerges. Pronominal use in the
intimate pattern is primarily dependent on the sex of the speaker where
males prefer feminine pronouns, females will generally use masculine ones.
264 Susanne Wagner
Only very rarely will a woman use a feminine pronoun that is not an
instance of personification, nor are men very likely to use masculine
pronouns for any inanimate entity. The use of he representing womens
inherent image of men is obviously entirely limited to women
(Mathiot and Roberts 1979: 41).
The second conclusion to be drawn from Mathiots study is a rather
sobering one: Although there might be a certain pattern in the use of non-
neuter pronouns, this use is by no means systematic.
34
This must be
concluded from the numerous examples where speakers switch pronouns
without any observable pattern (cf. examples in 5 from Mathiot and
Roberts 1979: 34):
(5) a. Do you realize how many times I have picked himup? He keeps
slipping off the shelf. Next time this happens Imgoing to leave it on
the floor. See how helikes it! [towel]
b. This one has been around long enough. I say, get rid of it! He is A
season [out of fashion], get rid of it! [bedspread]
c. What the hell is the matter with this thing? I t just wont work for me!
Heusually isnt like this! [typewriter]
Mathiot and Roberts explain all of the above shifts as instances of attaching
negative attributes to things that are usually upgraded (cf. Mathiot and
Roberts 1979: 33). However, the explanative value of such an assumption
is rather low. In (5a), for example, the speaker is clearly annoyed why
then not use it in all slots? The switch back to heis rather unexpected.
The pattern Mathiot and Roberts observed for everyday language use in
Los Angeles and Buffalo in the 1970s is by no means exceptional.
Examples of this type can be found aplenty in modern (American) fiction
and movies and have been overheard in conversations of Newfoundlanders.
From those examples it can be concluded that the pattern of pronominal
use as described by Mathiot and Roberts is rather prototypical of non-
standard spoken English in general. It will be shown in the following
sections that their study arrives at results that are very similar to those of
Svartengren (1930s, fiction; section 6.1.2), Morris (1990s, Canadian
English; section 6.2). On the other hand, a stark contrast exists between the
results from these studies and those from the analyses of the West Country
and Newfoundland dialect corpora (cf. Sections 10 and 11).
6.1.2. Vernacular perspective (H. Svartengren)
6.1.2.1. Introduction
In three essays very similar in content (Svartengren 1927, 1928, 1954),
Hilding Svartengren investigated the use of feminine pronouns used for
inanimate referents. His study differs from most other studies mentioned here
in that it is based on fiction, i.e. non-natural language use, for the most part
from US writers. Premature as it may seem, the first conclusion from
Svartengrens studies can be drawn here already: Although obviously puzzled
by the weird use of feminine forms for inanimate entities, Svartengren does
not mention a similar phenomenon for the masculine counterparts.
Two explanations for this come to mind: a highly unlikely one would be
that the author considered the use of masculine forms nothing unusual, thus
thinking it unnecessary to mention it. The second (much more likely)
explanation is that strange masculine forms were encountered at a far lower
frequency than feminine ones, thus not deserving any comment.
The non-existence (or at least extreme rarity) of masculine pronouns
referring to inanimate entities in American fiction supports one of the major
claims of this study very strongly: For the average native speaker of
English, the gendered pronoun of choice is feminine, while it is masculine
for a speaker with a West Country dialect background. This could already
be seen in the previous section(s) and sections, and will be supported by the
analyses presented in the following sections as well.
Of course, this rule is not without exception. One factor which is
apparently strong enough to overrule the general system is the speakers
sex (cf. Mathiot and Roberts 1979), resulting in females using masculine
pronouns in some situations. In addition, it should be clear that in our
modern times pure dialect systems do not really exist any longer. In this
particular case, the mixture of lects (true dialect vs. spoken Standard)
results in outwardly conflicting uses of pronouns, as the dialect predicts
masculine forms where the spoken variety would rather use a feminine
form. This seems to be one of the rare instances in which neither of the
most likely choices in casual style corresponds to the written standard
language which might be the explanation behind scholars puzzlement
over the phenomenon. In the following paragraphs, a brief overview of
Svartengrens studies will be presented, largely based on his 1927 and 1928
papers, which are basically identical in terms of content. Because of the
completely different outline and aims of Svartengrens work in comparison
266 Susanne Wagner
with the present study, evaluation and adaptation of Svartengrens
observations are possible only to a very limited extent. Emphasis will thus
be put on those points which either add new information to or support the
assumptions made here.
6.1.2.2. Svartengrens database
According to his 1928 article (Svartengren 1928: 79), Svartengren based
most of his analysis on 79 texts of contemporary, mostly American authors
(written ca. 19001925), among them such well-known names as J ack
London (with 10 texts) and Mark Twain (one text). In another 37 books
written by Americans or describing American life nothing, or nothing
worth quoting, has been found (Svartengren 1928: 8), giving the
impression that this particular use of feminine pronouns may be ideolectal
(i.e. restricted to some authors) rather than universal.
Another noteworthy feature (which is probably at the heart of
Svartengrens explanation for the use of these feminine forms) is that
almost all of the specimens stem from males, occurring either in direct
speech or some sort of internal dialogue, or simply because the author is a
man. In addition, Svartengren remarks that [m]any novels dealing with
upper and middle class life have contributed very little to our collection
(Svartengren 1927: 113). For him, the phenomenon is obviously a) not
geographically restricted and b) vernacular and rural at heart, but he is
aware of the bias of his database in this respect:
Examples show clearly that it is a distinct colloquialismat home chiefly
among men familiar with the stern realities of life and whose speech is
uninfluenced by literature this practically all over the United States and
Canada. Most of the material [...] hails from the fur, the timber, the mining,
and the cow countries, which may, or may not, represent the actual state of
things, for, we must add, works describing life in the industrial centers have
been drawn upon only to a limited extent. (Svartengren 1927: 113;
emphasis SW)
The high number of disregarded books (37 out of 116; roughly 32%) also
relativizes the overall frequency of the feature in the first place. As the
author himself states, [t]he two hundred and odd ex[amples] are drawn
from some 175 books, chiefly novels (Svartengren 1928: 14; Siemund
2001 counted a total of 268 examples, based on Svartengrens 1927
article). The resulting average is one to two examples per book, which is,
based on the present authors reading experience of the past years by no
means exceptional (cf. examples from modern popular fiction elsewhere).
6.1.2.3. Classes of nouns
Not very surprisingly after the prologue above, the noun classes Svartengren
identifies as being capable of triggering feminine pronouns are largely based
on research by earlier authors. Due to the diversity of referent nouns, very
often real classes cannot be identified at all, but rather represent a
accumulation of nouns that often share no more than one semantic feature.
Svartengren himself is well aware of this (Svartengren 1927: 110): It will be
seen, then, that every attempt to confine to certain categories of nouns the
instances when the feminine is to be used, must be abortive. The major
classes Svartengren lists are (cf. Svartengren 1927, 1928):
1. Concrete things made or worked upon by man
(a) Machinery, industrial plants
(b) Hollow things, receptables
i. Rooms, houses, and their uses
ii. Musical instruments
(c) Other things made, created, worked, or worked upon by man
i. Various small objects not tools
ii. Large scale undertakings
iii. Picture, film, newspaper
iv. Clothing, wooden leg
v. Food and drink
vi. Coins, money, amount of money, amount generally
vii. Organized bodies
viii. Districts
ix. Road, trail, distance
x. Natural resources exploited by man
2. Actions, abstract ideas
(a) Actions
i. Expressions containing an imperative
ii. Other expressions denoting actions
(b) Abstract ideas
i. Pronoun referring to substantive mentioned
ii. No substantival propword
3. Nature and natural objects not worked upon by man
(a) Nature
(b) Celestial bodies
(c) Geographical appellations
268 Susanne Wagner
(d) Material nouns
(e) Seasons, periods
(f) Fire, temperature, weather conditions, ice, snow
(g) Human body and its parts
The first category is very reminiscent of what Elworthy and Barnes
described as man-made objects, hardly a coincidence. It is the second
category in particular that commands our interest. Svartengren is, to my
knowledge, the only author among those investigating gendered pronouns
who proposes this category, which has been labelled non-referential she
earlier in this study (cf. 5.3). Non-referential here stands for real
instances (i.e. pronouns without antecedent) as well as abstract nouns
referring to situations, events, etc. Svartengren found many examples
illustrating this type of use, some of which are reprinted in (6ae) below.
(6) a. Let her go! Let her went (I am ready) (Svartengren 1928, ex. 130)
b. Start her off (ref. to making pancakes) (Svartengren 1928, ex. 134)
c. Watch out! Here shecomes! (speaker is sea-sick)
(Svartengren 1928, ex. 139)
d. Thereshegoes! (undertaking) (Svartengren 1928, ex. 59)
e. How do you like it, Tim? Shes alright.
(Svartengren 1928, ex. 161)
One major sub-category of the action category is connected with the use
of an imperative, again a use that is encountered with high frequency in
everyday conversations (such as Fill er up! referring to refuelling a
vehicle). Although not explicitly mentioned, another syntactic peculiarity
of this category is its frequent use of fronting or extraposition (examples of
the type There she goes!).
Example (6e) is reminiscent of the use of shes apples, shes right, shell
be right, shes sweet, etc. to refer to the general situation or circumstances,
supposedly restricted to Australian/New Zealand English (cf. Ramson
1988: 14, 31, 577, 656; Orsman 1997: 717), maybe hinting at the origins of
that particular construction.
The comparatively high overall frequency of this category is another
argument supporting the claim that we are essentially dealing with two
completely different, largely unrelated systems. The dialectal system(s)
described in Sections 10 and 11 do not allow the use of a gendered
pronoun referring to something as abstract as a situation or, even more
extreme, a gendered pronoun without any antecedent. We saw earlier that
for the traditional dialect systems the theories put forward by 19th-century
scholars still hold, at least to a certain extent. Most people will agree that it
is difficult to attribute a high degree of individuality or human traits (which
would justify an interpretation as personification) to a situation. In fact, this
raises a problem in Svartengrens own study: He himself claims that the
items referred to with a feminine pronoun must be capable of assuming at
least some degree of individuality (Svartengren 1927: 110). No comment
is made as to how this is achieved with his category 2. While this statement
supports the theory that Siemund (2001) proposes (roughly: only highly
individuated nouns can be animated), the data clearly contradict it, with
more than 20% of abstract referents pointing out that Svartengrens
material is of a completely different nature, illustrating a different system,
than the rest of the data that Siemund analyzed.
Not much needs to be said about the rest of Svartengrens categories,
which include numerous items capable of triggering feminine pronouns
even in the standard language (e.g. nature, celestial bodies, cities, ). As
has been pointed out before, they are not systematic in any way a fact that
the author himself acknowledges in saying that none of the restrictions (e.g.
artificial vs. natural objects; big vs. small, etc.) proposed by previous
researchers hold for his data (cf. Svartengren 1928: 41).
6.1.2.4. Origins and explanations
As he had earlier advocated the vernacular status of gendered pronouns,
Svartengren turns to the influence of other vernacular varieties as one of the
conceivable origins of the phenomenon after having dismissed possible
influence by foreign languages. Conveniently ignoring the fact that the
Southwest of England is he-territory and that large portions of settlers,
particularly working-class people, came from that area, he notes parallels
between the use he observed for America and the one documented for
Northern and Celtic English(es) (Svartengren 1927: 108). From the short
and rather cursory statements in various reference works (the EDD among
them) Svartengren concludes that, while it may have its origins in Great
Britain, the phenomenon now is American at heart and is, no doubt, rather
slowly invading British English as well, aided possibly by northern
dialectal influence (Svartengren 1927: 113).
Even though thinking of gendered pronouns as a feature found only in
the lower (working) classes, Svartengren does not automatically dismiss it
as wrong or a result of imperfect learning. Rather, he assumes that the
270 Susanne Wagner
emotional character is the distinguishing feature of the phenomenon
(Svartengren 1928: 51) and subsumes it under the more general label of
personification, an error easily justified by recalling the sex-bias of his
database and the general prejudices of that time.
35
The importance of
emotions in connection with the use of gendered pronouns is undeniably
one of the deciding factors (possibly even thedeciding factor) in triggering
the phenomenon in the first place.
6.2. The Canadian perspective (L. Morris)
In her doctoral thesis (Morris 1991), Lori Morris investigated gender in
modern Canadian English, drawing on both spoken and written data.
However, hers is not a corpus study, but is rather based on impressionistic
evidence and observations. The clear advantage of the study lies in its
breadth of focus. Morris considers all possible types of referents, from
humans to animals to inanimates, also including personification and other
relevant sub-categories or factors that may influence pronominal usage. To
my knowledge, this is the only work of this kind to date.
6.2.1. Animal denotata
Morris criteria for assigning gender are very similar to those already
identified as crucial in other studies and our own analyses: Animals playing
a (particular) role in discourse will be referred to by heor sherather than it.
Table 14 shows Morris three categories (cf. Morris 1991: 112139).
In her data, animals are much more frequently hethanshe, a pattern that
is expected based on knowledge from other varieties. For animals,
examples involving variation between sheand it are much more difficult
to find (Morris 1991: 124). Figure 2 shows the hierarchical system of
assigning gender to animals according to Morris (1991: 125).
Table 14. Gender assignment for animal denotata according to Morris (1991)
it background, non-individual; generally accepted behaviour of
species
he foreground, specific; individual; behaviour different from expected
norm/peculiar
she behaviour typical of species
Figure 2. Gender assignment for animals in Morris (1991)
The categorizations in Table 14 are typical; two traits that occur again and
again when investigating animal references are highlighted:
The major division is between neuter and non-neuter, or animate
and inanimate (it vs. he/she).
The factor that is mostly responsible for a change in the assignment
pattern is pragmatic rather than grammatical: an animal that is
foregrounded as the topic of a conversation will very likely be
animated.
In addition, Morris attributes the choice between she and he to the
behaviour of the animal in question, while stating at the same time that
feminine pronouns referring to animals are rare in her data. The one
example that puzzles the author fits in with our explanation of non-
referential she in an earlier section (cf. 5.3): There she blows!, uttered
about a spouting whale, would usually receive a masculine anaphoric form
(cf. Morris 1991: 135). However, the utterance (or rather the pronoun)
could also be interpreted as non-referential. Many of the non-referential
examples show fronting or preposing of certain elements, which is
otherwise rare Here she comes! or There she goes! are similar in this
respect. Only rarely is it possible to identify a referent; more often, the
speaker seems to refer to the situation in general.
clearly animate inanimate
he it
+female - female (neutral)
he she
inanimate
272 Susanne Wagner
6.2.2. Biologically inanimate denotata
Contrasting with animal pronominalization, Morris found that for inanimate
entities, she is favoured over he (cf. Morris 1991: 139). In her opinion,
speaker familiarity is responsible for many of the she-pronominalizations
in her data (Morris 1991: 146). Very often, the feminine pronoun is part of
a (short) imperative; if it were used, it would feel like a simple order; she,
on the other hand, has an inviting, attenuating effect (Morris 1991: 159
160; e.g. Let er rip!). Such an attenuating effect can easily be assumed
as an explanatory factor for the occurrences of non-referential feminine
forms in general. In addition, the author contrasts feminine and neuter
pronominal forms with the help of a criterion that we will encounter again
in Pawleys analysis of Tasmanian Vernacular English (cf. Pawley 2004;
Morris 1991: 163):
she particular denotatum, particular impressions of a given denotatum
it concept/norm of that type of denotatum
What plays a particular role in choosing personal pronouns is the
prototypicality of the referent in question: While an average denotatum
will generally be it, the speaker is bound to shift to a feminine form as soon
as anything peculiar or noteworthy about the referent is to be emphasized.
Contrasting with the use Morris observed for animate denotata, and also
almost diametrically opposed to the situation described for West Country
and Newfoundland dialects, masculine pronouns are basically non-existent
for inanimate referents: While masculine reference to any type of
inanimate denotatumis extremely rare, no examples at all were found in
which a native English speaker used heto represent an intangible, difficult-
to-identify type of denotatum (Morris 1991: 164; emphasis SW).
Based on the few examples of masculine pronouns referring to
inanimate entities that the author was able to collect
36
, she establishes the
following contrasts between she on the one hand and he on the other (cf.
Morris 1991: 168):
she familiarity, well-known; predictable, foreseeable
he maintains features of the unknown; less familiar, unpredictable,
more individualistic
Taking together the criteria Morris establishes for using she, heandit, the
speaker has to make a number of choices when referring to any kind of
noun. An attempt at showing all relevant relations is made in Figure 3. The
data stem from Morris (1991: 175177).
According to Morris, the primary function of pronoun gender is to
represent and express the manner in which a speaker has formed his mental
image of the denotatum (Morris 1991: 175). Overall, pronoun choice is
thus largely based on discourse-pragmatic factors, and in Morris system,
generalizations or predictions are difficult to impossible to make, as it is
predominantly the speakers worldview that is responsible for the choice of
a pronominal form. Although some patterns influencing this choice are
obvious and well-known, this is by no means as systematic a procedure as
those observed for Newfoundland and West Country dialects.
Figure 3. Overall system of gender assignment in Morris (1991)
6.3. Summary
The preceding sections have hopefully clarified a number of issues.
Although the varieties (and methodologies) investigated could not have
been more different, the results of the studies presented here are very
similar. Taken together, all of the studies mentioned in this section support
yes
no
+animate - animate IT
general use, no
individual
denotatum
+feminine
comparing against
other members of
category/ individual
denotatumselected
corresponds to general image?
HE SHE
274 Susanne Wagner
one of the major arguments of this study very strongly: sheis the gendered
pronoun of choice for non-dialectal English.
Mathiot and Roberts (1979) investigated a social variety of American
English, contrasting womens and mens language use with regard to their
use not only of gendered pronouns (what they called upgrading), but also
of the reverse phenomenon, downgrading (e.g. the use of it referring to a
human being). Their major result was that the sex of the speaker has
considerable influence on pronoun choice, with men using feminine
pronouns in many situations where women would prefer masculine ones.
Svartengren (1927, 1928, 1954) analyzed more than 100 novels and
other texts, primarily by American and Canadian authors, investigating the
use of feminine pronouns referring to inanimate entities. As his main result,
he identified a strong connection between the phenomenon and vernacular
speech, and also proposed that this was an essentially male feature, as
almost all of his examples stemmed from men. Svartengren could not
confirm any of the restrictions on noun classes allowing gendered pronouns
which had been proposed in earlier research.
The most important result of Morris research on Canadian English
(Morris 1991) for this study was the rareness of non-standard masculine
pronouns referring to inanimates in her database. While feminine forms
were relatively frequent and have been explained as representing the mental
image that the speaker has of the referent, masculine ones were almost non-
existent.
In his analysis of Tasmanian Vernacular English, Pawley (2002, 2004)
obtained very similar results: The Tasmanian (Vernacular English) universe
is largely feminine. Based on oral-history-type interviews, Pawleys data
stem largely from men.
The conclusions to be drawn from these analyses are obvious: In
everyday, casual spoken English, possibly world-wide, the pronoun of
choice when referring to an inanimate noun and wishing to add extra
information is (and, as Svartengrens studies indicate, has been for some
time) a form of she. Mostly, this extra has been identified as some sort of
emotional information, either positive or negative. In contrast, the pronoun
signifying non-involvement or simply disinterest is it.
The sex of the speaker may influence the pronominal form in so far that
women are more likely to use masculine forms in a number of contexts
where male speakers will use feminine ones, particularly in domains
associated with gender or gender-biased behaviour (e.g. cars, tools, etc.).
Although concrete nouns receive gendered reference more often than
abstract ones, there seem to be no restrictions, semantic or otherwise, on the
type of noun that can take a feminine form in anaphoric references.
An interesting category within the abstract domain is the use of sheand
her referring to a referent which is hard to identify, a situation or general
circumstances, which is shared by male and female speakers alike.
Although often identified as a peculiarity of Australian and New Zealand
English, examples are reported from basically all major varieties of
English.
The system outlined here stands in sharp contrast to the traditional
dialect system(s) of Southwest England and Newfoundland, where heand
the corresponding object form(s) occur in a large number of slots that are
occupied by shein the spoken Standard described in this section.
7. Persistence of gendered pronouns
7.1. Introduction and previous studies
The phenomenon under discussion here is known as persistence or
priming, including various sub-forms. Very generally, theories behind
these phenomena postulate that similar or identical forms cluster in speech.
Clustering effects are a well-known feature and have been studied in
variationists accounts for a long time.
Priming effects have been observed in a number of syntactic areas, with
particular emphasis on subject-verb-agreement (cf. Poplack 1979, 1980;
Poplack and Tagliamonte 1993, 1996; Scherre and Naro 1991). All of these
studies found that in those dialects which allow a zero marker on nouns
and/or verbs where StE would require a non-zero marker (usually third
person or plural -s), sequences including only marked or only unmarked
forms are much more likely to occur than a mixture of marked and
unmarked forms:
[] if a plural is going to be realized, the tendency will be for it to be
realized on the first element; if it is not, subsequent developments will not
tend to rectify this in a functional way. What follows might either be all
markers or all zeroes, so that a case like s turns out to be virtually non-
existent. (Poplack 1980: 6465)
These results show quite clearly that a parallel marking process is
occurring. Although isolated or first occurrences of verbs reflect no special
276 Susanne Wagner
influences, verbs that follow other verbs tend to mimic the marking of the
previous occurrence. (Scherre and Naro 1991: 25)
[W]e observe a concord effect, whereby lack of marking on a preceding
reference verb leads to a greater probability of zero marking on the current
verb (at .68 for Saman English and .66 for the Ex-Slave Recordings),
while overt marking leads to more marking. (Poplack and Tagliamonte
1993: 190)
Other studies focus on the reasons for clustering, which are found in the
way humans are assumed to process language. It is in this area of research,
often situated within the field of psycholinguistics, that the idea of
priming originated. In speaking, the use of one form activates that form,
and a recently activated form will be easier to re-activate in subsequent
discourse than an alternative (new) form. Also, less effort will have to be
made to re-activate a form that has been used more recently than to re-
activate one used much earlier in the conversation.
Potter and Lombardi (1998) used an immediate recall task to investigate
double object constructions in English, assuming that the structure of the
primes would influence the output of the recalled target, which was indeed
the case (cf. Potter and Lombardi 1998: 270271). Tree and Meijer (1999)
obtained parallel results in a similar study, again focusing on double object
constructions. In addition to an immediate recall task, they also compared
structures differing in the complexity of the phrases investigated:
In two experiments we observed that sentences with the same major
constituent structure shared syntactic routines. These syntactic routines are
likely to be stored in a hierarchical structure and to be activated
hierarchically. That is, major constituents are activated first after which
subroutines are called to build the structures with these constituents [...]
[S]yntactic priming occur[red] across conditions varying in complexity [and
was] equally frequent across conditions. (Tree and Meijer 1999: 8990)
Types of priming include lexical, morphological, form and syntactic
priming (for definitions of these, see e.g. Szmrecsanyi in progress). The
boundaries between the different types are not always clear-cut; for reasons
detailed below, the phenomenon will simply be called priming here.
7.2. Primed gendered pronouns?
7.2.1. Introductory background
The studies dealing with priming so far have usually analyzed syntactic
phenomena that fell within clearly definable linguistic categories: Poplack
(1979, 1980) investigated the use of plural -s in Puerto Rican Spanish
(Vernacular), Scherre and Naro (1991) noticed a priming effect in the
agreement behaviour of verbal strings in Brazilian Portuguese (Vernacular),
and Poplack and Tagliamonte (1993, 1996) analyzed priming in past tense
marking in varieties of AAVE and Nigerian Pidgin English.
All of these studies focus on variety-internal variation, usually those
cases where both a standard and a non-standard marker can be used.
Typically, no conflict arises between standard and non-standard markers,
which are often realized as zero: The non-standard marker is not identical
with another standard marker of the same paradigm, nor is the standard
marker identical with another non-standard marker of the same paradigm.
For gendered pronouns, the situation is slightly more complex.
In truly traditional dialect, be it the West Country or Newfoundland
variety, variation between it (in all its forms) and he (in all its forms)
referring to count nouns should not exist at all. In those varieties, only a
masculine form can be used to refer to count and/or concrete nouns, while
it is restricted to mass and/or abstract nouns. Standard Written English
would use it for both of these noun classes, while masculine forms are
restricted to animate entities.
In more modern dialects, such as those that form the basis of the
present study, the traditional dialect paradigm and the standard paradigm
overlap and interact. Thus, it is possible to use it with reference to a count
noun, but hecan also be used (cf. Figure 4).
278 Susanne Wagner
Standard English
____________ _ _ _ _ _
modern dialects
______________________ _ _ _ _ _ _ _ _ _
traditional dialects
________________________________________
humans animals count nouns mass nouns
Figure 4. Extension of he across noun classes in 3 varieties of English
(simplified)
Priming can affect the choice between a neuter or a masculine form for
these referents. What is problematic about this choice is that it may be
switching registers or lects (namely dialect vs. standard) rather than an
intra-lectal choice. Although none of the earlier studies involved such a
switch, it seems appropriate nevertheless to use the concept of priming in
this context. It has to be assumed that speakers are at least bi-lectal
nowadays, i.e. have the capability to use the standard (or at least near-
standard) variety of the language and at least one additional (regional,
social, ethnic, ) variety. The speakers concerned here will use a regional
lect in addition to a more standard one. The regional variety can use
masculine pronouns for count nouns, while the standard variety cannot.
Depending on the situation, speakers will switch between the codes
available to them.
7.2.2. Problems and pitfalls
A number of issues have to be considered when analyzing possible priming
effects in a corpus of spoken language:
Each interview has to be looked at in isolation, as priming effects
obviously cannot occur across different interviews.
In oral-history-type interviews, the possible influence of the
interviewer (if not a dialect speaker) must be checked and
accounted for.
In interviews with more than one dialect speaker (dialogues,
discussions between speakers), priming between speakers is
possible and can be analyzed accordingly.
The current investigation faces one problem in particular: standard forms
(i.e. its) are necessary for an analysis to be possible at all. Thus, only those
interviews (or better: speakers) can be used that show a comparatively high
influence of the standard variety, i.e. those where it rather than a masculine
pronoun is used at least occasionally to refer to count nouns. In addition,
the interviews have to contain a considerable number of tokens in order for
statistical analyses to be possible at all. For example, an interview with 10
instances of masculine vs. two instances of neuter pronouns is of no use.
Another pitfall mentioned in connection with other studies is not
applicable here, but will be mentioned for the sake of the argument. Certain
constructions or forms do not occur in regular intervals throughout
discourse. Tenses may be given as an example: Although in a narrative
most stretches will be told in the simple past, the general narrative tense
employed in English, there may also be passages that are told in past
perfect or, for dramatic effect, simple present (historic present). When
analyzing the clustering effect of simple past forms, there will thus be
stretches of speech without a single simple past form simply based on the
fact that the narrative situation is different in that passage. Similarly, when
analyzing future marking and possibly competing future markers (e.g. be
going to vs. will/shall), only those passages can be analyzed where future is
marked. Consequently, priming effects can only be studied in those
stretches where the forms under scrutiny occur.
However, as it is basically impossible to carry on a conversation without
using personal pronouns referring to both mass and count nouns, this issue
is irrelevant for the present study. Although there will be certain passages
where only count or only mass referents occur, the overall distribution of
pronominal forms is fairly even within the texts that will be investigated
below. Keeping these restrictions and problems in mind, we should be able
to study possible priming effects on pronoun choice.
7.3. Priming in the corpus
In order to be able to investigate priming effects, a number of modifications
needed to be made in the corpus material. First and most importantly, all
occurrences of it had to be disambiguated according to their referents or
280 Susanne Wagner
rather the type of noun they referred to, i.e. mass or count nouns. Moreover,
the three issues listed above had to be considered.
Each interview will be looked at individually, taking care of the first
issue. Secondly, the interviewer influence is minimal to non-existent in the
interviews selected. For the chosen Newfoundland material, the
interviewers themselves are natives, so that only intra-speaker code-
switching occurs. The Somerset speakers who were selected are of the very
talkative kind, so that the interviewer only rarely intervenes to ask further
questions, generally just making supporting noises to encourage the
informant to continue. As only one-on-one conversations will be analyzed,
the third issue is irrelevant.
Prerequisite for the selection of texts was a high overall frequency of
gendered pronouns. After the disambiguation of all occurring its, the
modified text was used as an input for the TACT text retrieval programme.
Counting switches (i.e. sequence of either neutermasculine or masculine
neuter forms) and non-switches (i.e. sequence of either neuterneuter or
masculinemasculine forms) produced a two-by-two table of the following
kind:
it he
precedingit A B
precedinghe C D
If no priming effect were observed, as many switches (cells B and C) as
non-switches (cells A and D) would be expected, the null hypothesis being
that each scenario is equally likely (i.e. 25% probability for each cell).
7.3.1. Priming in the Somerset data
Five texts from the Somerset sub-corpus were chosen for an analysis of
possible priming effects. This comparatively low output of interesting texts
is based on the rather low overall frequency of gendered pronouns in the
Somerset texts, making statistical analyses difficult or even impossible in
many cases. The results of the analyses are presented in Tables 15 to 18.
The tables should be interpreted in the following manner: The likelihood
of switches is much lower (between 14% and 30%) than the likelihood of
non-switches. The bold figures show the likelihood of a masculine form
following another masculine form, ranging between 64% and 82%.
Although the chi-square values for the four tables above are not always as
significant as hoped for, a tendency for priming or clustering of identical
forms can be observed. In fact, it is highly unlikely that a speaker would
switch pronouns in any given passage without outside influence. In
comparison, Klemola (1996: 237ff) tested priming effects of unstressed
periphrastic do in his thesis. His likelihood values for identical forms to
occur subsequently only range between 10.5% and 46.1%, averaging 30%.
Table 15. Priming in Som_001
text code neuter form masculine form
Som_001 n ratios n ratios
preceding neuter form 32 0.86 5 0.14
preceding masculine form 4 0.27 11 0.73
significant at 0.002%
significant at 1.2%
significant at 1.3%
7.3.2. Priming in the Newfoundland data
As the interviews from MUNLFA are more traditional in the sense that
they contain more gendered pronouns, more texts could be investigated
than in the Somerset sub-corpus. Also, the overall results should be more
282 Susanne Wagner
relevant because the speech of most of the informants is highly
conservative. The results for texts that were investigated for priming effects
can be found in Tables 19 to 23.
Table 19. Priming in C626
C626 n ratios n ratios
The generally higher frequency of gendered pronouns contributes to overall
more significant statistics. The five texts tested here once again show high
likelihoods of sequences of identical forms (61% to 86%). In fact, the
overall distributions indicate that Newfoundland speakers are even more
susceptible to priming than their Somerset counterparts.
7.3.3. Generalizations and summary
The scattergrams in Figure 5 and Figure 6 clearly show that priming is a
factor in the tested interviews, thus nicely summarizing the results observed
in the tables for the individual interviews (cf. Sankoff and Laberge 1978] for
the procedure/methodology and background assumptions).
37
Figure 5 plots
the proportions of masculine to neuter switches against the overall
proportions of neuter forms. Figure 6 plots the reverse option, i.e. the
proportions of neuter to masculine switches against the overall proportions of
masculine forms. If the null hypothesis held, as many switches as non-
switches would be expected, i.e. all the dots should fall on the diagonal. The
real distribution, however, shows all dots below the diagonal, which indicates
that identical forms in sequence are highly unlikely to be switched.
Figure 5. Scattergram 1: MUNFLA and Somerset texts
284 Susanne Wagner
Figure 6. Scattergram 2: MUNFLA and Somerset texts
Note that even those interviews which were non-significant employing the
chi-square test conform with the general tendency. The 13 interviews
included in the scattergrams result in an overall picture that is fairly clear
and rejects the null (non-priming) hypothesis. It can thus be concluded that
priming is indeed a factor in pronoun choice. Speakers behaviour as
observed for selected texts from both the Southwest of England as well as
Newfoundland material confirms tendencies that have been reported for
other grammatical categories: If confronted with a choice of two competing
forms expressing the same category, speakers will generally prefer the
already activated form over the alternative one, resulting in sequences of
same forms rather than a mixture of forms A and B without a pattern.
Although without further research and more detailed studies of priming it
would certainly be premature to make any far-reaching predictions
concerning pronominal choice, a tendency for conservatism (i.e. clusters
of identical forms) has been observed for the investigated texts.
8. SED Basic Material
As mentioned in section 4.1, the Basic Material of the SED holds a lot of
material that, although it may not be immediately obvious, contains
information relevant for the present study. Based on the investigations of the
fieldworker notebooks, it was possible to isolate certain SED questions that
could evoke answers containing gendered pronouns. Those questions were
checked in the Basic Material publications, not only for the Southwest but for
all locations. The following paragraphs present the results of that investigation.
Some of the problems connected with such an endeavour were already
mentioned in section 4.1. First of all, the policy for recording or not
recording a pronoun is unclear. For certain questions, the pronoun was part
of the expected response, and was thus written down by the fieldworkers.
In those cases it is to be expected that the pronoun will also appear in the
published Basic Material. For other questions, however, where a pronoun
was not an essential part of the response, policies of reprinting pronouns in
the Basic Material vary widely. Sometimes pronouns are included,
sometimes they are not. Some of the entries in the Basic Material note that
where recorded, the personal pronouns are included in the response.
However, it becomes clear that not all fieldworkers actually took notes of
such uses, as additional pronouns are only rarely recorded for the North or
East, while they are ubiquitous in the Southwest. The only alternative, but
highly unlikely, explanation which would account for such a difference is that
informants in the North simply use(d) fewer pronouns than those in the
South. As a result of these fieldworker preferences, the Southwest shows
by far the highest rate of pronoun retention, as will be seen from the data
presented below. The codes that will be used in the following sections follow
general SED conventions. The questions that were considered relevant can be
found in Table 24; the formulations are taken from Orton (1962).
The different levels of pronoun inclusion are obvious in the third column
(expected response). The part that was actually sought is in boldface, and
while a(n additional) pronoun was obviously expected in some questions (e.g.
IX.2.6), it was not in others (e.g. I.11.2). The following sections will discuss
the responses to these questions as reprinted in the respective sections of the
Basic Material, following its outline in the order of discussion. Questions will
be identified by their number in the questionnaire, which can be found in the
first column of Table 24. Not all of the 10 questions will be discussed for all
regions, depending on the relevance of the recorded forms. If only standard
286 Susanne Wagner
responses were given or no pronouns were recorded at all, the respective
question will not be discussed.
Table 24. SED questions likely to contain instances of gendered pronouns
book question
expected
response
I.7.1 If you want to know how heavy a thing is, what do
you do?
To weigh it
I.11.2 What do you use to prevent your cart going
backwards when you stop on a hill?
Prop/chock
I.11.6 How do you empty a cart the quickest way? To tip
VIII.7.6 A dog buries a bone because he wants to hide it
IX.2.6 And now [stand sideways in front of it =door] in front of it
IX.2.8 If the door blew open on a cold day, youd get up at
once and
shut it
IX.3.1 If theres a hole in the pocket where you keep your
knife, youre almost certain to
lose it
IX.4.4 You cant have my spade today because I want it,
but you can have it tomorrow, because then (I)
(I ) shant want
it
IX.8.2 J ack wants to have Tommys ball and says to him,
not: Keep it!, but [gesticulate]:
Give it me!
IX.9.3 You have something to give away and before
deciding on the person to be given it, you might ask
yourself: I wonder
to whom I
shall give it?
8.1. The Northern Counties
Of the 10 questions listed in Table 25, the Northern Counties show
exceptional pronominal usage in only one question. Responses to question
I.7.1, IX.8.2 and IX.9.3 are it in all 75 localities; no pronouns were
recorded for question I.11.2, IX.3.1, and IX.4.4. For questions VIII.7.6,
IX.2.6 and IX.2.8, the recorded response either contained a form of it (it, t)
or no pronoun at all. Thus, the only question of interest is I.11.6.
At 40 of the total 75 locations, standard it was recorded; the seven
exceptional uses are listed in Table 25; for the other locations, no pronouns
were recorded. Although all non-standard variants appear in Yorkshire, no
pattern can be detected. Locations vary from the far North (loc. 3, Skelton,
is on the northern coast) to the West (loc. 14, Grassington), East (loc. 25,
Newbald) and extreme South (loc. 32, Ecclesfield, and loc. 33, Tickhill).
Also, although the preferred non-standard form is [or], [im] is recorded at
loc. 32, thus making it impossible to say anything about the distribution of
masculine or feminine forms. It is highly unlikely that that both pronouns
can be attributed to a semantic contrast between the referent nouns (cart vs.
load). While an explanation for those seven exceptional forms is thus
difficult (at least for the time being), it should be noted that the preference
for the feminine form is to be expected: Traditionally, feminine pronouns
can be used to refer to vehicles, even in StE, at least metaphorically (cf.
Section 3). Also, as will be recalled from earlier sections, feminine forms
were identified as the general unmarked choice of gendered pronouns in
spoken standard varieties, excluding Southwest England.
Table 25. Non-standard pronoun forms, question I.11.6, Northern Counties
county locality response
Y 3 lip il up, lip oi up
Y 4 sLcI oi up
Y 14 lip oi up
Y 25 sLcI oi up
Y 32 ju:l im up |ju:l oi up shoot her (=the load) up]
Y 33 lip oi up
8.2. The East Midland Counties and East Anglia
Based on our knowledge about the distribution of gendered pronouns, we
do not expect any real surprises (i.e. non-standard forms) from this area,
which includes the region whose dialect has become Standard English.
Interesting responses are given in Table 26.
Table 26. Non-standard pronoun forms, East Midlands and East Anglia
question county locality response
I.11.6 Nt 1 lip o+ up
L 9 luip o+ (tipe her)
Lei 6 also: luip o+ up (tipe her up)
IX.2.6 Nth 2 o (in front on er)
IX.2.8 Nt 1 o (shut er)
Nth 2 o (shut er)
288 Susanne Wagner
Loc. 1 in Nottinghamshire is geographically close to loc. 32 and 33 in
Yorkshire, where almost identical pronominal forms ([r]) were recorded
(cf. responses to question I.11.6 above). Not much can be said about the
other exceptional forms. It should be noted, however, that non-standard
gendered pronouns seem to be completely absent from East Anglia.
8.3. The West Midland Counties
As most of the counties included in this region border on the traditional
Southwest (Herefordshire, Worcestershire, Gloucestershire, Oxfordshire), a
higher number of non-standard pronoun tokens is expected. Also, the
frequency of masculine forms should be higher than that of feminine forms
(which should approach zero). For better legibility, the questions will be
presented in groups and individual distributions will be commented on; the
detailed tables this discussion is based on can be found in the respective
volumes of the SED Basic Material.
As was expected, the West Midlands present a picture that is radically
different from either the Northern or the Eastern region. For question I.7.1,
non-standard forms have reached a total of 10% (63 responses it, seven
responses some form of he; no pronoun in four locations), all of them
masculine. Although non-standard responses to question I.11.2 are too rare
to include them generally, the two examples from Worcestershire and
Gloucestershire stand for the general trend (increase in masculine forms)
and are thus listed here.
Even more dramatic are the figures for question I.11.6. For the total 74
localities, no pronouns were recorded in 33, leaving a total of 41. In 23 of
these, the pronoun is standard it, while we find 18 instances of masculine
and two cases of feminine forms (i.e. 46.5% non-standard forms;
occasionally, the total of standard and non-standard forms is higher than the
grand total, as at some locations more than one response was recorded).
Both feminine forms occur in comparatively northern locations (South
Cheshire and Mid-Staffordshire). The distributions of the other forms show
a very nice wave pattern depending on how far removed from the core
Southwest they were recorded. The lone standard it in Worcestershire
occurs in the northernmost location in the county, with the greatest possible
distance to the Southwest. The same holds true for the only recorded
pronoun in Warwickshire (also it).
The two forms in Oxfordshire look like a textbook case: standard it was
recorded in Islip on the eastern border of the county, while non-standard
[,P] was chosen in Eynsham, close to the western border. In both
traditional and modern maps of major dialect areas, the isogloss separating
the Southwest from other dialect areas usually cuts vertically through the
middle of Oxfordshire. Eynsham would thus be considered part of the
Southwest, while Islip rather belongs to the Southeast. With five instances
of masculine forms and no occurrence of it, Gloucestershire is, at least in
the context of this question (referent =cart), truly Southwestern territory.
A complete picture of this wave structure will be presented below, also
including the Southern Counties. The overall frequency of masculine forms
for this question is 0% in Warwickshire (one it), 50% in Oxfordshire (one
of two forms total), 66% in Worcestershire (4 of 6), 75% in
Monmouthshire (3 of 4), 83% in Herefordshire (5 of 6), and 100% in
Gloucestershire (5 of 5).
Although not as pronounced as for question I.11.6, the general tendency
and patterning of forms is the same for question VIII.7.6: The only
feminine form is found in the North, while standard it occurs frequently in
border zones, be it to other dialect areas or the coast. Masculine forms, on
the other hand, are the rule for all localities closest to core Southwest
territory. Percentages of masculine forms range from 17% in
Worcestershire (1 of 6) and 50% in Gloucestershire (3 of 6) to 66% in
Herefordshire (4 of 6).
The two questions referring to door once again show the by now
familiar pattern: The closer one moves towards the core Southwest, the
more likely it is to encounter masculine forms. In both questions, no
masculine forms were recorded in Warwickshire or Oxfordshire (nor
Monmouthshire for IX.2.6). The ranking of the other counties goes from
Monmouthshire via Worcestershire and Herefordshire (which change
positions in the two questions) to Gloucestershire, which shows the highest
ratios of masculine forms in both questions.
38
Overall, it is recorded in 51
localities for question IX.2.6, while a masculine form appears in five
localities (8.9%). Much higher ratios can be found for question IX.2.8,
whereit occurs in 30, a masculine form in 14 localities (31.8%). Even the
comparatively high 1 : 1 ratio in Shropshire fits the general pattern: the
masculine form is located in the South of the county, while standard it
appears in the North.
The referent of question IX.3.1 is knife, which shows even higher
overall masculine ratios than the previous questions. No pronouns were
290 Susanne Wagner
recorded in 42 out of 76 localities, attributing much more weight to the 10
masculine forms (one feminine form is found in Derbyshire) in relation to
the 23 neuter forms, resulting in a 32% likelihood to encounter a non-
standard form (11 of 34 forms).
39
In question IX.8.2, which primarily
investigates the order of constituents in a double object construction, the it
refers to a ball. Non-standard pronouns were only recorded at three
localities in Gloucestershire, with speakers at all other localities using
standard it. This low frequency of non-standard forms is striking in
comparison with the figures for the same question recorded in the Southern
counties, as will be seen below.
8.4. The Southern Counties
The Basic Material of the SED does not distinguish between Southwest and
Southeast, which seems surprising considering the stark contrasts between
those two dialect areas. On the other hand, this offers a nice opportunity to
analyze transition zones between the Southwest and the Southeast with
little effort. As in the previous section, the data will be presented in groups
of questions, followed by brief discussions of the results.
Even when the Southeastern counties Sussex, Kent and Surrey are
included, masculine forms outnumber neuter forms for question I.7.1 (38 of
73; 52%; for details for this section, see the respective tables of the Basic
Material). Without those three counties, where only standard it was
recorded, the percentage of masculine forms climbs to 69% (38 of 55
forms).
40
With such high ratios of masculine forms even for a
comparatively undefined referent (something), it comes as no surprise that a
more concrete referent (cart in question I.11.6) should only increase the
figures for masculine pronouns. Out of 63 pronominal forms, 47 (74.6%)
are masculine. Excluding the three Southeastern counties, this figure climbs
to 86.8% (46 of 53 forms).
41
Similar results are obtained from question
VIII.7.6 (ref.: bone), with 46.7% (28 of 60) masculine forms including the
Southeast, 59.6% (28 of 47) without it.
42
The door, referent in question IX.2.6 and IX.2.8, is [QC] or [Q] with
some of the highest frequencies of all items. Overall ratios are between
46.9% (IX.2.6, incl. Southeast) and 81.4% (IX.2.8, excl. Southeast).
43
Ratios of masculine pronouns as part of the responses to question IX.3.1
(ref.: knife) are never below 50%, with overall totals between 70.5% (31 of
44, incl. Southeast) and 79.5% (31 of 39, excl. Southeast).
44
Noteworthy
here is the sentiment of the informant at So 5 who considers the form [QC ]
older than standard it.
Question IX.8.2 is one of the few questions where it is an actual
keyword. It is not impossible that fieldworkers checked informants
responses by asking them to repeat their formulations, which in certain
instances might have led to a highly stylized interview situation resulting in
the use of more standard language. This might be an explanation for the
low to non-existent frequency of masculine pronouns in the other regions in
comparison with the Southern counties (recall that there were only three
instances of masculine pronouns in the West Midlands, all in Gloucester-
shire). For the Southern counties, however, overall ratios of masculine
forms as part of the response are as expected, ranging between 46.8% (36
of 77, incl. Southeast) and 61% (36 of 59, excl. Southeast).
45
Not surprisingly, a number of non-standard forms occur in the Southern
counties in responses to questions where pronouns were generally not
recorded in the other regions probably because the response would have
been standard it there anyway. This includes question I.11.2 (ref.: cart),
IX.4.4 (ref.: spade) and IX.9.3 (ref.: something).
8.5. Summary and results
The results of this section are noteworthy in two respects: First, the clarity
of the emerging pattern of the distribution of non-standard pronominal
forms is striking. Figure 7 summarizes the results of all analyses in this
section, showing a clear core Southwest area with adjoining transition
zones. Only some localities fall out of the picture, which is generally
textbook-coherent, starting from highest frequencies of masculine pronouns
in the deep Southwest and slowly petering out the further east one moves.
Interesting to note is also the belt of generally higher frequencies close to
the eastern border of the West Country. Why this vertical strip exists is
unclear at the moment. One might speculate that, from the equally high
frequencies for the whole area, certain localities were more influenced by
the standard than others. Those localities then exercised consecutive
influence on the neighbouring ones. Somerset seems to be leading in this
change towards the standard.
Although this may not look surprising, this west-east cline contains the
second noteworthy aspect: It has been accepted as common knowledge that
West Cornwall is not really to be considered part of the West Country
292 Susanne Wagner
dialect region. The analyses presented here, however, prove that the
opposite is true. All Cornish localities show frequencies of masculine
pronouns of between 80%100%, which is definitely not to be expected
from the picture drawn in the literature.
Figure 7. Distribution of masculine pronominal forms (SED Basic Material)
9. The SED fieldworker notebooks data
TheSED fieldworker notebooks provide a yet largely untapped data source
for dialectological studies. Although by no means suitable for just any
morphological or morphosyntactic investigation, the notebooks contain a
wealth of examples in the domain of personal pronouns. Despite the
obvious problems connected with using the fieldworker notebooks for a
corpus study (or any study), it was felt that here the sheer quantity of data
by far outweighs any problems of a qualitative nature. This does not mean
that these problems will simply be ignored where they are of particular
interest and where it seems appropriate, they will be discussed, especially
since unclear transcriptions can often actually provide the key to solving
the theoretical issues. The following paragraphs will illustrate the system of
pronominal gender assignment as found in traditional West Country
dialects in the 1950s and 1960s.
9.1. Referent types
Based on the literature on gender assignment in the core Southwest, we
expect to find a system that is (still) very similar to the one described by
Elworthy and his colleagues in the 19th century. Speakers of such a
traditional variety would use he and him, un, en, n etc. to refer to
inanimate count nouns, while feminine pronouns are restricted to female
animates (i.e. humans and animals) exclusively. Instances of masculine
pronouns referring to mass nouns and instances of feminine pronouns
referring to inanimate entities should not be encountered.
It should be clear from the beginning that the array of possible referents
is heavily restricted by the design of the SED, most examples involving one
of the words asked for in the questionnaire. As most of those items are
concrete count nouns, the chances of encountering personal pronouns
referring to highly abstract nouns are rare from the start. The overall
predictive value of this SED-based section will therefore be rather low.
First and foremost, the results obtained here are intended to serve as a
reference frame for the investigations in the following sections, where more
modern material will be at the core of the analysis.
9.1.1. Masculine referents
As has already been pointed out, the very nature of the SED fieldworker
notebooks (and, indeed, the SED itself) is responsible for a very
homogeneous picture regarding the kinds of nouns that occur with gendered
pronouns. Although the term homogeneous may be misleading, it is
appropriate when comparing the rather restricted number of referents from
the fieldworker notebooks with the wide variety of possible referents that was
found in the collected corpora from Southwest England and Newfoundland.
294 Susanne Wagner
The nouns to which the non-standard pronouns refer can be divided into
three large classes (1 to 3 below); not very surprisingly (considering the
nature of the SED), class 1, man-made objects, contains by far the largest
number of entries. The following paragraphs will show typical examples of
each class and its sub-categories. Also, overall frequencies of occurrence
will be given for each class.
1. MAN-MADE OBJ ECTS
(a) buildings, their parts and contents (e.g. chimney, furniture)
(b) containers (e.g. box, cup)
(c) tools and instruments (e.g. farming utensils)
(d) vehicles
(e) creature comforts
i. clothing
ii. accessories of modern life (e.g. watch, pen, photograph, toys)
iii. food and drink (excluding naturally occurring things)
(f) nature re-modelled (e.g. mine, lane, curb stone)
2. NATURE
(a) trees and plants
(b) other (e.g. fruit, vegetables; ground, hill, pond, river)
3. BODY PARTS (of animals and humans)
Although the class labels are slightly different from those proposed in
earlier research, the reader will immediately discover similarities between
Classes 1 to 3 and the traditional categories postulated by 19th-century
writers such as Elworthy or Barnes and also Svartengrens suggestions. The
first class of man-made objects contrasts with the second one, which
subsumes all naturally occurring things under it. The third class, on the
other hand, could also be labelled man and beast, as its entries largely
refer to animals or humans.
9.1.1.1. Man-made objects
BUILDINGS
This category comprises more than 40 relevant forms. Typical referents are
door, house or chimney; representative examples can be found in (7).
46
Those that occur with more than two examples (see Table 27) are generally
either key words in questions or mentioned therein and taken up in the
response by the informant.
(7) a. L(GWcSZLSWcVC WXQ (36 Co 5, book V)
It had a top with a tassel on it. ref. =bed
b. MUQC (31 So 14, book VIII)
(Do you) hear it? ref. =bell (church bell)
c. KLG,GQCJ(WNOL
,
QG48WEWZQVM]L (32 W 6, book V)
It was cleaned out only once a year, you see. ref. =chimney
d. ZLGG=(QLVHL]DIZHRSPC (36 Co 4, book IX)
We generally say its half-way open. ref. =door
e. ]ZLS'YOX,IM<ED,QJZDQZo
U
6Q (31 So 9, book V)
Sweep the floor if you arent going to wash it. ref. =floor
f. 'DV'SXVWL]41GWX (36 Co 7, book IV)
Thats the post its hung on. ref. =gate
g. ,Q(Y],GQLDQW],GQ (24 Gl 7, book 9)
I havent seen it; he hasnt seen it. ref. =house
Table 27. Masculine pronouns referring to buildings, their parts and contents
(Fieldworker notebooks)
referent #of examples #of speakers
bed 2 1
bell 1 1
building, house 15 14
chimney 5 4
church 1 1
door 17 14
floor 2 2
gate 2 2
lock (hasp) 1 1
oven 2 2
roof 2 1
table 1 1
WC 1 1
total 51 38
The referents are usually typical count nouns which are expected to trigger
masculine forms. WC may be a bit unusual, but according to the context, a
sort of outhouse is discussed, which belongs in the same category as house.
The status of floor may be debatable; one could argue that floor in the sense
that it is used here (layer on which people tread) is a singular noun, as
there is only one floor in a room. Nevertheless, it should be clear that the
296 Susanne Wagner
word is +count in general, which would justify the use of a masculine
pronoun in any of its uses.
CONTAINERS
When the different categories of noun classes were devised, a separate class
was reserved for containers. The rationale behind this decision was based
on the (folk) belief that containers have some features in common with
women womb-like is one of the terms one hears mentioned. On such a
basis, at least some feminine pronouns that are used to refer to such a noun
are expected. However, the data tell a different story: Not a single feminine
pronoun referring to a container is found in the material. Containers, like all
other man-made objects, are masculine pronoun territory, just as expected
from the traditional literature. Some typical examples can be found in (8),
detailed distributions of all container-like nouns in Table 28.
Table 28. Masculine pronouns referring to containers (FN)
bag 1 1
basket 2 2
bin 3 2
bucket, kettle, trough, tub 7 7
can 1 1
coffin 2 1
cup, bowl, glass 4 4
grave 7 7
parcel 3 2
pocket 7 7
pot 1 1
tumbler 10 9
vase 5 5
total 53 36
(8) a. LZ]PHGE,Z,',] (36 Co 4, book V)
It was made by (with) withies. ref. =basket
b. M<NQVWDQGQcQ](QG (37 D 9, book I)
You can stand it on its head. ref. =bin (corn bin)
c. S8GWoY'ENWQ(PSQ48BW (36 Co 1, book V)
Put it over the bucket and empty it out. ref. =bucket
d. I: bI 'bo:id 'in (31 So 6, book IV)
It is dented in. ref. =can (tin can)
e. vi o 'ando lu: on (31 So 14, book V)
... with a handle on it. ref. =bowl dish
f. I: 'sLd :p o 'g:s on I: 'bnL lo 'pI:soz (36 Co 5, book IX)
He shattered the glass and it broke to pieces. ref. =glass
g. o:, 'n
i
: 'in on (37 D 2, book VII)
Theres nothing in it. ref. =pocket
h. oi 'didn 'bc:L n (38 Do 5, book IX)
I didnt break it. ref. =vase
Referents occurring with some frequency are pocket andtumbler, which are
either key words or mentioned in the respective question, and different
articles of hollow- and flatware. The items all fall within the expected
norm.
TOOLS AND INSTRUMENTS
Based on the outline and purpose of the SED, we would expect this
category to contain a high number of non-standard pronominal forms.
Details for the actual referents found in the fieldworker notebooks are
summarized in Table 29, and illustrative material is provided in (9). Also,
as many tools are keywords in the questionnaire, the number of examples
for a specific referent (e.g. spade) should be higher than in many of the
other categories, where we often find just a single example per referent.
One such item is knifeas part of questions IX.3.1 and IX.3.2, which are
primarily concerned with the different forms of the verb find. These two
questions include various sub-questions, which altogether yield no less than
22 references to knife with the help of a masculine pronoun. Plough (or
parts thereof) is, not very surprisingly, also among the high-frequency
items. Another favourite is spade, which once more figures prominently in
one question of the grammar section (IX.4.4).
(9) a. vI: 'pul n 'nn o 'voio: (32 W 3, book V)
We put it on the fire. ref. =fire grid
b. I: cd 'v
i
: 'pnpz l.: on (37 D 7, book I)
It had four prongs to it. ref. =fork
c. o 'bni? gnl 'ipz nn on (32 W 9, book I)
A beetle has got rings on it ref. =beetle (hammer types)
298 Susanne Wagner
d. DYoVQxLYDG`<P,ScNW (36 Co 2, book IX)
Ive lost it; it fell through my pocket. ref. =knife
e. LGN,S',QVY`PYoQ8W (38 Do 4, book I)
It keeps the linch from falling out. ref. =linch-pin
f. oW'Z41'xPHN1`QG,Y`Q (37 D 5, book I)
Alter the wing there. Make it run differently. ref. =plough
g. LYJcGRGLQVD,JcWQC (37 D 5, book V)
It has become old since I got it. ref. =poker
h. LEc`,]LY`LoIPC (36 Co 4, book VIII)
He borrows it very often. ref. =shovel
Table 29. Masculine pronouns referring to tools and instruments (FN)
bar 3 2
brake shoe 2 1
The fireplace 13 7
fork 6 4
Horses & Co. 12 9
knife 26 19
linch-pin 3 2
mallet, hammer 2 1
plough (or part of) 14 9
post 3 2
roller 3 2
scarecrow 1 1
screw 1 1
scythe 4 3
shovel 3 3
spade 21 20
stake 1 1
stretcher 1 1
string 1 1
various tools 25 18
whetstone 2 2
total 148 65
A sub-category is related to various items which can broadly be
summarized as having something to do with horses; 11 examples belong
here (belly band, brush, clog, harness, etc.). Another subgroup consists of
items from the fireplace (book V.1) such as bellows, grate, or poker (eight
examples). Moreover, there are numerous references to one tool or other,
which will only be listed collectively (various tools) below; clearly
identifiable items include coulter, flail, rake, or rope-twister.
In line with our expectations, the category tools and instruments is,
with almost 150 examples (or almost 25% of all gendered pronouns in the
SED notebooks), well-represented in the data. Many instances of non-
standard masculine pronouns relate directly to questions in the
questionnaire in that informants pick up a noun included in the stimulus
question and use a personal pronoun referring to that noun anaphorically.
VEHICLES
If no neuter pronoun is used to refer to vehicles, the pronominal form of
choice is generally feminine rather than masculine (cf. also Sections 3 and
6), which suffices as an argument in favour of treating vehicles as a
separate category here. However, it turns out once more that the SED data
are very traditional indeed only one instance of a feminine pronoun
referring to a vehicle could be found (10a). The reasons for such scarcity of
feminine references are unclear. It is extremely unlikely that none of the
informants used the occasional sheor her to refer to a tractor etc. What is
more likely is that such a use was not remarkable in the eyes of the
fieldworker(s), as the use of feminine pronouns for vehicles is sanctioned
by grammarians, if with reservations (cf. the ubiquitous fill er up at petrol
stations). One gets the distinct impression that only vehicles with a motor
can be she apart from sailing vessels, that is. At SED times, there were
not that many motor vehicles around. Rather, horse-drawn cart and wagon
were the norm, the advent of tractors is only occasionally mentioned. As a
consequence, not a single she-tractor made it into the fieldworker
notebooks. In addition, the one feminine reference to car does not stem
from the core Southwest. It was recorded in Hatherden (Hampshire), which
is in the North of the country, close to the Wiltshire and Berkshire borders.
The absence of feminine pronouns from the core Southwest counties
does not mean that masculine forms do not occur either in fact, the
opposite is true. Table 30 shows the details for this category; examples are
given in (10). The absolute number of denotata is relatively small,
indicating that (motorized) vehicles did not yet play the central role in SED
times that they play today.
(10) a. ZLDGW`Q6LRP (39 Ha 2, book VIII)
We had to run it home. ref. =car
300 Susanne Wagner
b. j.: Lon ': 'mai baiL ] ai 'ja:n vnn? n (37 D 5, book IX)
You can have my bike. I shant want it. ref. =bike
c. ju: gnl 'ip 'op ':p (36 Co 5, book I)
Youve got to trig/scotch it up. ref. =cart
d. L: on 'qug o: 'qag ] vilj ju:m 'main lu: (36 Co 5, book I)
Call it drug or drag, which(ever) you want to. ref. =sledge
e. 'qc:v on 'rn zu: I: 'va:n in 'baL (37 D 6, book IX)
Drive it round so it wont run back. ref. =vehicle
Table 30. Gendered pronouns referring to vehicles (FN)
referent #of masc. ex.s #of fem. ex.s #of speakers
bike 2 2
bus 1 1
car 1 1 2
cart 34 23
sledge 1 1
vehicle 6 5
total 46 1 30
CREATURE COMFORTS
With a label such as creature comforts, it will be clear from the beginning
that this is not a homogeneous class. Rather, it is an accumulation of items
belonging to three main sub-categories, namely clothing, accessories of
modern life (e.g. watch, pen, photograph, toys), and food and drink
(excluding naturally occurring items). To everyone familiar with the SED it
will be obvious that these groups are inspired by frequently occurring
referents in the questionnaire.
Articles of clothing Although few, a number of clear references to articles
of clothing and a couple of questions that contain clothing items as possible
referents justify the existence of this class. Detailed distributions and
examples follow below (Table 31 and examples in 11). Once again, it is a
question in book IX that yields the highest output of masculine forms.
Although the context of IX.3.3 is plural (... put them ref. collar studs), the
pronominal forms that are used by the informants are clearly singular
(mostly [n]). Another question in book IX has suit as its referent, again with
a comparatively high ratio of masculine forms.
Table 31. Masculine pronouns referring to articles of clothing (Fieldworker
notebooks)
bonnet, hat 6 4
boot 2 2
article of clothing 4 3
collar 1 1
sock 2 2
collar stud 12 10
suit, jacket 8 8
total 35 27
(11) a. XNZ]DVPEcQWZ,cWVI`L]cQQ (36 Co 6, book VI)
A gook was a summer bonnet with lots of frills on it. ref. =bonnet
b. KL]E`DQQMXQ (32 W 8, book VI)
Its a brand-new one. ref. =boot
c. S8,PcI (15 He 7, book VI)
Pull it off. ref. =hat
d. D,DQ]LGQo
U
',NcGxZ'LSGQ (37 D 1, book IX)
I havent seen anything of your collar stud. Where have you put it?
ref. =collar stud
e. <PHGQC (37 D 10, book IX)
Who made it? ref. =suit
Accessories of modern life Included here are all those items that might be
classified as tools, but which have nothing to do with everyday work on a
farm. Examples are watchor pen. Moreover, things that could be classified
as fun tools (i.e. toys) such as ball, are also included in this category.
Table 32 and the examples in (12) provide details.
Ball figures prominently as it is part of a number of questions,
particularly VIII.7.3 A rubber ball thats punctured wont ... bounce
(Orton 1962: 93). Informants usually form an independent clause and use
an anaphoric pronoun to refer to the ball, resulting in formulations as in
(12a).
Money, or rather coins and notes, are also frequently he or en, as in
(12g). Part of question IX.1.3 (A picture not hanging straight is hanging ...
askew; Orton 1962: 95), pictureis another frequent denotatum (12h).
302 Susanne Wagner
Table 32. Masculine pronouns referring to accessories of modern life (Field-
worker notebooks)
ball 20 13
book 1 1
clock, watch 9 5
coin, note 6 4
lamp 1 1
pen(cil) 3 2
photograph 1 1
picture 10 9
plate 1 1
see-saw 1 1
song 1 1
total 54 38
(12) a. I: vanl gIans (36 Co 1, book VIII)
It wont bounce. ref. =ball
b. 'ulj o cnd on lu i (31 So 14, book IX)
I will lend it to you. ref. =ball
c. i 'Ludnl +I:d im (16 Wo 7, book VIII)
I couldnt read it. ref. =book
d. I:z oboul '!aiv minils '!a:sl (15 He 7, book VI)
Its about five minutes fast. ref. =clock
e. ju: Ln 'loin on on 'o:pon on on 'jul n (31 So 14, book VII)
You can turn it and open it and shut it. ref. =watch
f. im did 'Lu: im l (15 He 5, book V)
He called it that. ref. =lamp
g. 'ju: 'o: on gin o 'ail (36 Co 6, book VII)
You hold it against the light. ref. =note
h. I:z 'hapin 'ul 'in o: (38 Do 3, book IX)
Its hanging out (askew), isnt it? ref. =picture
Food and drink This category only includes countable items of food and
drink mass referents are treated in section 9.1.3. Also excluded are
naturally grown things (such as apple), which are covered in section
9.1.1.3. As a result, this is a rather small class, with only 11 examples.
Table 33 shows details; examples can be found in (13).
In (13a), the speaker uses beer as a count noun, making masculine
reference possible. In two other cases, speakers refer to breador pastryashe,
assuming an individual loaf of bread or pastry as referent (V.6.12 When
your bread or pastry has not risen, you say it is ... sad. Orton 1962: 72).
Example (13e) may not really belong in this category, but oil can be
subsumed under food. In addition, the semantic make-up of the example
is very similar to others in this category: A mass noun is converted into a
count noun by using a unit of measurement that is +count. Drop is used
here, while the beer example is obviously meant to include glassor pint.
(13) a. L]J8]QxLG,GQC`,1N,]E,xLJ8]GQ (36 Co 6, book VI)
Hes guzzling. He didnt drink his beer. He guzzled it. ref. =beer
b. 'DVNW6LNHNxLGE,GRL]L (36 Co 2, book VI)
Thats clutchy cake. Itd be doughy, see. ref. =cake
c. NHNVE,WNRVxLDQW`D,]GS (39 Ha 2, book V)
The cakes a bit sad. It hasnt risen (up). ref. =cake
d. 1MQQW(GLSDVWLx'PDGE,QE,GP(,W,QQ (37 D 6, book VII)
Onion and potato pasty. There might have been a bit of meat in it.
ref. =pasty
e. (,YS`cSc,QN(W6QYD, (36 Co 2, book V)
Heave up a drop of oil and catch it afire. ref. =drop of oil
Table 33. Masculine pronouns referring to food and drink (Fieldworker
notebooks)
(glass of) beer 1 1
bread/pastry, cake, pasty, pie 9 9
crumb, piece, loaf of bread 6 6
drop of oil 1 1
total 17 16
NATURE RE-MODELLED
This category represents a sort of compromise or intermediate step between
the class of man-made objects and the next bigger category, nature. Items
included here are usually natural substances or similar things which man
has changed to his liking. A hedge, for example, is usually left alone to
grow once it has been planted, and often not maintained in any way but it
has to be planted in the first place. Similarly, a lane may consist of gravel
304 Susanne Wagner
or mud, i.e. natural substances, but someone had to build it. The same goes
for a haystack, a rick, etc. Examples are given in (14); Table 34 shows
details of distribution.
Table 34. Masculine pronouns referring to nature re-modelled (Fieldworker
notebooks)
bundle of straw, stook 2 2
cesspit 1 1
dike/ditch 1 1
drain 1 1
field, meadow 11 8
furrow 3 3
ridge 1 1
garden 1 1
hedge 12 9
lane, path, road 4 4
rick, sheaf, stack 5 4
stick 3 3
tombstone 1 1
wall 4 3
yard 1 1
total 51 31
(14) a. 'lciL 'ul iL 'b:nd o 'slou on 'bip n ':n (31 So 13, book II)
Take out that bundle of straw and bring it along. ref. =bundle of straw
b. I:z 'siLslI c:Loz al mcdo (36 Co 6, book IV)
Its sixty acres that meadow. ref. =meadow
c. c: do 'b
i
:L p 'pr on on 'lI: n o'gc:n (36 Co 7, book II)
They break it, plough it and till it again. ref. =field
d. 'I:z 'p.ljl :p 'bc:v 'al v:n iz (36 Co 2, book II)
Thats pitched up well, that one is. ref. =ridge
e. 'va:loi on bi 'nil (31 So 14, book IX)
Water it by night. ref. =garden
f. hI:z :I goud ouvo: (31 So 4, book IV)
It is all grown over. ref. =hedge
g. I:z 's:m 'no 'io: o: (36 Co 4, book IV)
It is very narrow here, Sir. ref. =path, road
h. MXNGM1(PVW,NxLZ8GE4)B G4)B Q4QVP (36 Co 4, book II)
You cut a young elm stick, it would bow down handsomely (very
well). ref. =stick
i. 6LS8GJRRYKL (38 Do 1, book IV)
A sheep wouldnt go over it. ref. =wall
Fieldis included here as the OED states that the use of fieldmeaning open
land as opposed to woodland; a stretch of open land is obsolete (OED,
field 1.a.). Moreover, SED informants are generally referring to their
arable fields, clearly meaning the worked land, not some untouched stretch
of land. Interestingly, the semantic change took just the opposite path with
meadow, which originally only referred to the worked land, but now also
stands for grass-covered land in general. There is just one clear masculine
reference to a meadow in the SED fieldworker notebooks (14b), where the
reference to its size shows that it is probably used for either hay production
or as pasture.
Furrowandridgeare intrinsically connected in any field (see ridge 5.
in the OED), and speakers refer to both with the help of masculine
pronouns as in (14d), for example. Hedgeis part of questions IV.2.34, and
is generally taken up in the response with a masculine form, as in (14f).
The walls included here (see 14i) are not parts of buildings, but are used to
separate fields, fulfilling the same function as a hedge.
9.1.1.2. Man-made objects summary
Summarizing the different sub-categories of man-made objects, we can
conclude that the SED informants from the West Country still use a system
of pronoun reference that is very close to the traditional system as
described in 19th-century dialectological literature. Denotata are, almost
without exception, concrete count nouns (cf. section 9.1.3 for non-standard
referents). Table 35 summarizes the results from the different sub-
categories (excluding the one feminine form referring to car).
As has been pointed out at the beginning of this section, man-made objects
constitute the overwhelming majority of the masculine pronoun referents in
the fieldworker notebooks. With far over 400 instances, approximately two
thirds of all examples belong to this category.
306 Susanne Wagner
Table 35. Masculine pronouns referring to man-made objects (Fieldworker
notebooks)
referent subclass #of examples #of speakers
buildings 51 38
containers 53 36
tools & instruments 148 65
vehicles 46 29
creature comforts
articles of clothing 35 27
accessories of modern
life
54 31
food & drink 17 16
nature re-modelled 51 31
total 455 110
9.1.1.3. Nature
A category nature is postulated for the same reasons that were given for
the classes of containers and vehicles respectively. Nature and
naturally occurring things are often given feminine pronoun representations
in StE, partly based on mythological explanations (e.g. Latin arbor tree
grammatically masculine, but feminine agreement based on the belief that
nymphs inhabit trees), partly on patterns of metaphorical personification
(nature she).
Based on the make-up of the SED questionnaire and its general concern
with things important in farming communities, a number of nouns one can
subsume under the heading nature are part of the questions or even key
words. In the following, non-edible denotata (basically trees and plants
47
)
will be distinguished from others (edible ones such as fruit and
vegetables, but also other features of nature such as river or moon). This
distinction is based on categorizations in other studies (see, e.g., Siemund
2001 or Pawley 2002). Whether or not it is justified will be discussed at the
end of this section.
TREES AND PLANTS
Treeor parts of a tree such as boughor branchare among the most frequent
referent nouns, occurring 20 times in the context of six questions. In
particular, trees are mentioned or at least implied in all questions in section
IV.10 Trees, bushes and IV.12 Parts of a tree. In question VIII.7.4, it,
referring to a tree, is part of the expected response (If a boy wanted to get
to the top of a tree, hed have to ... climb it. Orton 1962: 93). Examples of
this category are given in (15) and summarized in Table 36.
(15) a. pul o big bu n on I:d bo:n :I vI:L (16 Wo 7, book V)
Put a big bough on and it would burn all week. ref. =bough
b. I:z '!c:os iz 'v:d o'b:d (37 D 10, book II)
Its face is wide open. ref. =flower
c. I:z 'b:do: 'in o: (32 W 8, book II)
Its broader, isnt it? ref. =leaf
d. I: adn? :?o g+ou i (25 O 2, book II)
It isnt supposed to grow here. ref. =plant
e. i! I: 'bc:L ':! I:z 'brn lo I:v o 'slu: biain (36 Co 4, book IV)
If it breaks off, its bound to leave the stump behind. ref. =tree
Table 36. Masculine pronouns referring to trees and plants (Fieldworker
notebooks)
bush 1 1
flower 1 1
grass 1 1
leaf 2 1
plant 6 6
tree (or part of) 21 17
total 32 25
OTHER
Referents belonging to this category are rather heterogeneous: apples are
represented, as are other naturally occurring things such as fire, hill or
river. Various references to the moon are interesting. Although traditionally
associated with feminine rather than masculine pronouns (cf. OED, moon
1.b.), dialects use masculine pronouns almost exclusively in reference to
the moon. Two potential explanations come to mind: for one, it seems
possible that the traditional Germanic gender distinctions with a feminine
sun and a masculine moon (as still found in Modern German) survived in
the dialects and did not give way to the classic system of Latin/Romance
languages, which has a masculine sun and a feminine moon (Lat. sol m.
sun and luna f. moon).
308 Susanne Wagner
Alternatively or possibly reinforcing the first scenario, another folk
belief took over in dialects, namely that of referring to the patches of
darkness and light as a mans face, thus the man in the moon (cf. OED,
moon 1.a.). In the latter case it looks as if we would be well-advised to
classify these examples as personifications. However, the OED definition
explicitly states that no personification need be involved when using a
feminine pronoun so why should we necessarily assume it for masculine
ones? Table 37 shows details of the referents found in this category,
illustrative material is provided in (16).
Table 37. Masculine pronouns referring to other natural features (Fieldworker
notebooks)
fruit, vegetables etc.
apple 6 5
berry 2 2
nut 1 1
onion 3 2
potato 2 2
egg 2 2
forms of water
pond 4 3
puddle 2 1
river, stream 8 8
well 1 1
ford 1 1
other
fire 8 7
hill 2 1
stone 5 5
moon 7 5
thunder 1 1
total 56 42
(16) a. D,o](,GQCVN,QQo (36 Co 4, book VI)
I always eat/ate? it skin and all. ref. =apple
b. E41N1S (31 So 3, book V)
Bank it. ref. =fire
c. 'I: gnd o 'vI:o un on (38 Do 3, book VII)
It has got a wheel around it. ref. =moon
d. al 'union coi ju:d 'panl im ju:d gcl 'ljiboz !inm im (25 O 4, book V)
That onion there, (if) youd plant it youd get chibols from it. ref. =onion
e. 'idn no: 'dcp0 in on ] I:z 'jao (36 Co 7, book IV)
(There) isnt any depth in it. Its shallow. ref. =pond
f. !ipon nv on (31 So 14, book VIII)
Flinging it. ref. =stone
A by now familiar picture emerges when summarizing the results from the
two sub-categories that constitute the superordinate category nature:
Once more, not a single feminine pronoun is encountered. No matter if the
denotata in question are big or small (e.g. treevs. bush), dangerous as a fire
or completely innocent, edible or inedible, moving or unmoving (e.g. river
vs. pond), the non-standard pronouns that are chosen by the SED
informants are masculine without exception.
9.1.1.4. Body parts
Various questions in the SED refer to body parts, both of humans and
animals. Occasionally, one could expect a metaphorical link in the form of
a part-whole relationship that would justify the use of gendered pronouns
even in StE. Individual cases will have to be put under scrutiny to see if
such a condition applies. This is the only category dealing with animates
(human beings and animals) all others discussed so far strictly concerned
inanimate entities. Relevant referents are divided into two groups (human
animal) in Table 38, with representative examples in (17).
(17) a. aiv 'bin 'iLl p (36 Co 4, book VI)
Ive been (and) ricked it. ref. =ankle
b. 'd+:mo'io+ 'LII:nin im aul vi o 'pin (15 He 7, book VI)
Drum of the ear, cleaning it out with a pin. ref. =(drum of the) ear
c. ji 'loi on 'u! (31 So 14, book III)
She tore it off. ref. =hoof
d. 'al 'iz o 'd:d 'lcl vi 'no: 'miL in lu: on (38 Do 4, book III)
That is a dead teat with no milk in it. ref. =teat
310 Susanne Wagner
Table 38. Masculine pronouns referring to body parts (Fieldworker notebooks)
human body parts
ankle 4 4
belly 2 2
blister, boil, bump, swelling, wound 15 11
disease 1 1
ear (& parts) 4 4
finger, nail, part of hand 5 4
head 3 3
tongue 1 1
tooth 4 3
animal body parts
bone 15 13
hoof 3 3
pelt 2 1
skin (of bacon) 1 1
teat 5 4
total 65 39
Bone is one of the most frequently recorded referents, owing to one
question in particular in which it is part of the expected response (VIII.7.6
A dog buries a bone because he wants to ... hide it. Orton 1962: 93). The
decision to include the one reference to diseaseunder human body parts
may be debatable, but seems justified as the question is part of the general
section on the human body in the questionnaire and because diseases befall
human (and animal) bodies. As far as one can tell from the short excerpts in
the fieldworker notebooks, all of the 57 examples of masculine pronouns
used for body parts are true instances of gendered pronouns, without any
personification or metaphor involved. Referents are generally typical count
nouns, and the use of a masculine pronoun would be expected according to
the traditional system of gender assignment in West Country dialects.
9.1.1.5. Over-generalized cases masculine female animals
It is assumed in the traditional literaturethat West Country speakers from
time to time seem to over-generalize their masculine pronouns. As a result,
a cow may become he, a reference that is usually too remarkable to go
unnoticed by the SED fieldworker. It is unclear in how far such uses are
really instances of over-generalization, but it seems as good an explanation
as any. Table 39 lists the animals that receive masculine reference by SED
informants. Unsurprisingly, cowis the most frequent denotatum (cf. 20ac).
As pronominal reference to animals is outside the scope of this study (see
section 5.2), these cases are only mentioned here for their own sake and
will not play any role henceforth.
Table 39. Masculine pronouns referring to female animals (Fieldworker
notebooks)
cow 12 9
ewe-cat 1 1
hen 1 1
total 14 11
(18) a. li 'I:v bin 'miLl (37 D 7, book III)
Till it has been milked. ref. =cow
b. 'al 'Lrz IrLin 'po:qI ':o 'idnl I: (37 D 8, book VII)
That cow is looking pretty thin, isnt it? ref. =cow
c. 'iL 'Louz ':! ] gn? o lji oboul on (32 W 5, book VI)
That cow is wet; its got a chill. ref. =cow
9.1.2. Feminine referents
Not unexpectedly, feminine pronouns are only rarely used to refer to
inanimate nouns in the fieldworker notebooks. In fact, rarely is much too
tame an expression to describe the situation: Among the almost 700
interesting examples, there are only three, with two from the same speaker,
which are clearly feminine. They can be found in (19).
(19) a. #Lot o #copper in #she. When #shewent in was #bad #thing.
Lot of copper in it. When it was liquidated ... (36 Co 6, book IV)
b. 'jI: lciL s:m 'puIin
It takes some pulling. (21 NF 6, book VI)
The first reference is to a copper mine, the second to pulling out an eye
tooth. The first example contains the only feminine references to an
312 Susanne Wagner
inanimate entity in the Southwestern notebooks. It is unclear why the
speaker chooses this form here, although mines are traditionally among the
entities which can take feminine pronouns. Also, it should be noted that St.
Buryan, the locality in question, is at the westernmost tip of Cornwall,
which is supposedly not as strongly influenced by West Country dialect as
the more easterly parts. Any of these would suffice as an explanation.
The second example stems from Norfolk, which is location-wise of
course neither part of the West Country nor example-wise really relevant to
the discussion. However, it shows the basic west-east chasm, as three West
Country speakers used a masculine pronoun in the same context in four
instances (referring to an eye tooth; see Table 38).
9.1.3. Problematic cases
This section discusses those cases which should not exist in the
traditional system, that is, masculine forms referring to abstract nouns,
mass nouns, etc. It has already been mentioned that the fieldworkers
transcriptions are not always consistent and thus not reliable. Most of the
problematic cases are classified as such based on these inconsistencies.
The count-mass distinction is traditionally assumed to be at the heart of
the principle of gender assignment in West Country dialects, with count
nouns taking forms of he (and she), while all other nouns take it
exclusively. The problem about this paradigm is its interaction and partial
overlap with that of abstract vs. concrete nouns. At its most extreme, the
system predicts that only concrete count nouns employ masculine forms,
while all other nouns can only use neuter forms as anaphoric pronouns.
48
The following examples illustrate the wide array of unexpected
referents found in the SED fieldworker notebooks. It should be mentioned
here already that exceptional referents are amazingly rare totalling about
60 examples, they do not even amount to 10% of the grand total of almost
700 non-standard pronominal forms in the fieldworker notebooks.
The non-count referents encountered can be subdivided into four basic
categories: a) liquids, food, b) weather phenomena, c) abstracta and d) (other)
mass nouns. Referents of class a) include nasty medicine, gravy, fat,
bacon, milk and butter, among others (examples in 20).
(20) a. '4W'(P,NVJcQxLZ(QWW8NUo
8
G (31 So 14, book V)
That milks gone, it went to a crud (got sour). ref. =milk
b. 7,N1CQSE,W (36 Co 7, book V)
Thicken it up a bit. ref. =gravy
c. ,YGW6,Q (32 W 5, book V)
Ive had to churn it. ref. =butter
The referents for class b) are more difficult to identify. Fog or snow
can be deduced as in (21a) and (21b), but often, the referent seems to be
day or weather in general, as in the example in (21c). It is interesting to
note that we are almost exclusively confronted with bad weather the
negative feelings and implications might be responsible for gender choice.
The importance of emotive factors in gender assignment were already
mentioned in Section 6.
(21) a. 'DWVPD,QYcJxc,JcWoVW,Q,Q (32 W 2, book VII)
Thats a heavy fog. I got lost in it. ref. =fog
b. c,NQPc,QWc,PZLMXVWEPCSQL] (36 Co 5, book VII)
I can remember a time (when) we used to have it up to our knees. ref. =snow
c. L]SPZ('WG(, (36 Co 5, book I)
Its damp weather today. ref. =weather
Another small subgroup of non-count referents is made up of a number of
abstracta, mostly in response to two questions of the questionnaire which
involve a chance to go to college (IX.3.7) and a job (IX.4.11; 22ab).
(22) a. LQ(Y8NQC (36 Co 7, book IX)
He didnt take it. ref. =chance to go to college
b. M<PVJ,GQCGQWPc` (37 D 8, book IX)
You must get it done tomorrow. ref. =job
It is to be assumed that most of the exceptions to the traditional system of
gender assignment are based on analogical extension: each of the three
major categories (liquids/food, weather phenomena, abstracta) contains
numerous other members which are count nouns, one can conclude that the
masculine forms in the examples above were most likely used on that basis.
Also, it should be obvious that we cannot really tell in many cases what the
exact referent of the pronoun is. It might as well be a count noun expressing
the same concept just to give one example often used to illustrate this:
breadis - count (i.e. it) while loaf (of bread) is +count, thus he.
314 Susanne Wagner
Table 40 summarizes the examples belonging to this section. Three
speakers contribute more than two examples (three, four and eight
respectively), while 14 speakers only add one example each. Two examples
each are contributed by the remaining seven speakers. As the categories
noun class and number of speakers overlap (one speaker can contribute
to more than one class), the individual figures in the last column differ from
the total.
Table 40. Summary of problematic referents (Fieldworker notebooks)
noun class #of examples #of speakers
liquids, food 21 11
weather phenomena 15 7
abstracta 11 8
(other) mass nouns 9 9
total 56 26
Cornwall figures rather prominently in this section. Eleven instances of
unexpected masculine pronouns stem from St. Buryan and Mullion, and
seven from Gwinear. Taking a look at the map of the individual SED
locations, we see that all three localities are situated in West Cornwall. It
will be recalled from earlier sections that it is generally assumed that West
Cornwall is less dialectal than the more easterly regions. This assumption is
based on the theory that the existence of Cornish and, following that, the
relatively late introduction of English to the area are responsible for the
comparatively standard dialect (of English) in West Cornwall.
However, the data presented here (and in the previous section on the
SED Basic Material) make another scenario more likely: The Cornish
people either seem to have done very well in adopting the West Country
system of pronominal gender assignment in a rather short period of time
viewed in this light, the misuses we observe in the SED data are clear
cases of over-generalization. Or, a more likely explanation, speakers were
well aware of the system from the start and did not have to learn it at all,
as it was part of their English competence from the their early childhood.
9.2. Summary and results
9.2.1. Types of referents
At the beginning of this section we were interested in two issues in
particular. First, we were curious to investigate the relationship between the
fieldworker notebooks data and the Basic SED Material. Second, various
claims made in the traditional dialectological literature as to the nature of
the referents that allow gendered pronouns were to be tested.
In Section 8, we already learned that Cornwall is leading in the use of
gendered pronouns. Devon, Wiltshire and Dorset follow, but Somerset,
which is generally considered core Southwest, falls far behind for
unknown reasons, we have to admit. It is interesting to note in this respect
that Montacute, the Somerset location not included in the Basic Material,
shows figures and uses very close to those of Cornwall, resulting in a stark
contrast between this location and the other 13 which constitute the Basic
Material data for Somerset. It cannot be ruled out that Somerset speakers
are less traditional than their Cornish counterparts, already having started
the shift towards StE. This will have to be investigated more closely when
comparing the modern corpus material for the two counties (cf. 10.2.1 and
10.2.3).
The dialect found in Cornwall in the SED period is of a very traditional
Southwestern nature this fact as such is probably the most important
single result to be obtained from both the Basic Material and the
fieldworker notebooks. Speakers from the two most westerly locations, St.
Buryan and Mullion, use almost twice as many masculine pronouns
referring to inanimate entities as their colleagues at the other Cornish
locations. Based on experts views in the relevant literature, we would have
expected to find the exact opposite in West Cornwall, namely a variety
closer to StE.
When investigating which type(s) of noun(s) occur with a masculine (or
feminine) anaphoric pronoun in the fieldworker notebooks, a surprisingly
clear picture emerges. From a semantic point of view, almost 87% of
denotata fall into three basic categories: (1) Man-made objects, (2) nature,
and (3) body parts. Only 13% of the examples are referring to (in dialect
terms) non-standard, difficult-to-classify or unclear referents. Non-standard
feminine pronouns are so rare that the label non-existent seems justified.
For the sake of convenience, Table 41 summarizes the figures from the
individual tables given in this section.
316 Susanne Wagner
Table 41. Summary of masculine pronouns referring to inanimate entities
(Fieldworker notebooks)
referent class subclass #of examples
MAN-MADE OBJ ECTS
buildings 51
containers 53
tools & instruments 148
vehicles 46
creature comforts 106
nature re-modelled 51
NATURE
trees & plants 32
other 56
BODY PARTS
human body parts 39
animal body parts 26
subtotal 608

feminine referents 2
problematic referents 56
unclear referents 30
total 696
TheSED fieldworker notebooks data show a variety of English that is (still)
surprisingly close to the traditional West Country vernacular described by
Barnes, Elworthy and other 19th-century authors. Critics may say that the
material used here is atypical for a corpus study because only one of two
possible values is used. Fieldworkers only noted non-standard language
use, resulting in many exceptional masculine pronoun forms in the
notebooks. We have no way of knowing how many standardforms speakers
may have used. However, this is inconsequential here we do not want to
compare the system of gender assignment in SED times with that of other
periods. Thus, if we were to play devils advocate, it is possible that SED
speakers used it to refer to a door, one of the standardly masculine referents
in dialect, in two or even three out of four cases, resulting in a frequency of
non-standard forms as low as 25%50%. As stated, this may well be possible
but it does not influence or change the results presented here.
Some possible tendencies towards a less rigid system of gender
assignment were witnessed in the unclear and exceptional referents.
Concluding ex negativo, these uses of masculine forms referring to
unacceptable denotata (from a traditionalists point of view) may point to
the diffusion and ultimate dissolution of the traditional system. Most of the
problematic examples seem to be over-generalizations, where masculine
pronouns are used to refer to a noun that may be close in meaning to
another acceptable one, but whose semantics do not allow a masculine
anaphoric pronoun in traditional West Country dialect.
9.2.2. Differences between counties
The distribution of noteworthy pronominal forms among the investigated
counties is by no means equal. Although differences were expected, they
turned out to be much more extreme than had been assumed. Table 42
summarizes the figures for the individual counties, including all localities of
the core Southwest counties (i.e. also those not used in the Basic Material)
and all clear examples of masculine pronouns as presented in Tables 27 to 38
(figures in fourth and sixth columns rounded to first decimal).
Table 42. Frequency of gendered pronouns per county and location (SED
fieldworker notebooks)
county
#of
examples
#of
locations
examples
per location
#of
speakers
examples
per speaker
Cornwall 163 7 23.3 20 8.2
Dorset 40 5 8 8 5.0
Devon 126 11 11.5 26 4.8
Wiltshire 70 9 7.8 15 4.7
Somerset 88 13+1 6.3 28 3.1
total 487 46 10.6 97 5.0
Columns 4 and 6 are of particular interest in Table 42. Even on a very
superficial level, the picture emerging could not be any clearer: Speakers
from Cornwall produce most of the gendered pronouns by far and are
responsible for almost exactly a third (163 out of 487; 33.5%) of all
examples. Speakers from Dorset, Devon and Wiltshire are close to the
average of five forms per speaker, while Somerset lags behind once
again, we have to say (recall Figure 7).
This overall picture does not change when looking at detailed
distributions per location and per individual speaker. The order of counties
318 Susanne Wagner
is slightly different for examples per location Devon and Dorset change
places (see column 4). Examples per location range from two to 45. All but
one of the Cornish locations are above the average of 10.6 examples per
location, as are five out of nine in Wiltshire, six out of 11 in Devon, two
out of five locations in Dorset, but only one out of 14 in Somerset
(Montacute is not included in the Basic Material).
The order of counties stays the same when looking at who actually
contributes masculine forms. In Cornwall and Devon, all informants do,
while this is not so in the remaining counties. One of the nine Dorset
informants (i.e. 11.1%) does not contribute, while this percentage climbs to
25% in Wiltshire (five of 20 speakers) and to 33.3% in Somerset (12 of 36
speakers, excluding Montacute). Although the order of counties in the
detailed distribution list changes to a certain extent, individual informants
who might distort these figures cannot be identified. Contributions range
between one and 24 per speaker, with an average of five. 70 speakers are
below that average or conform to it, while 27 contribute more than their
share. Those 27 (or 27.8% of speakers) contribute 295 forms, i.e. 60.6% of
the total of 487.
Thus, we are once more faced with a now familiar picture: (West)
Cornwall is much more dialectal than has generally been assumed, at least
when it comes to the use of non-standard personal pronouns. Why earlier
researchers thought of Cornwall as a region where a rather standard variety
of English is spoken remains an open question. Data from the SED
material, both the Basic Material and the fieldworker notebooks, as well as
more modern oral history material suggest that the opposite is the case.
Somerset in the 1950s, on the other hand, does not seem to have much
in common with the Somerset of Elworthys times. While the gender
system described in his studies can be considered the epitome of West
Country dialect, the SED data show a system that is much closer to StE
than to the 19th-century one. Although it is possible in theory that the
variety changed so drastically in only one or two generations, we can
immediately dismiss such a theory based on the data from the surrounding
counties. Devon, Dorset, Cornwall and Wiltshire have only changed
minutely in comparison with Somerset. It almost seems as if the Somerset
informants language was not as dialectal as that of the rest of the West
Country informants to start with maybe better informants were simply not
available at that time. It is very unlikely that Somerset was influenced by
StE to a greater extent than its neighbouring counties. It will be interesting
to see how the individual counties developed since SED times an issue
which will be investigated in the next section.
10. Southwest England oral history material
The material on which this section is based can be described very generally as
oral history material. Although comparable in many respects, it differs from
the SED fieldworker notebooks data in very important ways: The major
difference between the two main sources for the Southwest certainly has to be
seen in the fact that the notebooks only offer clauses, phrases or sometimes
single words with minimal or even without any context at all. The oral history
material, on the other hand, consists of complete interviews, generally in the
form of a dialogue where the interviewer contributes only minimally in terms
of quantity. While the former material does not allow quantitative or
statistical analyses of any kind, the latter does.
In terms of content and types of referents, however, the SED and oral
history material are very similar, maybe surprisingly so. These similarities
are mainly caused by some of the oral history collectors interest in
agricultural or related topics, such as cider making, cheese making, or
farming in old times. Thus, it will be possible to compare the data from a
qualitative (mainly semantic) point of view, and, where relevant, to trace
changes in use.
The data from the different counties will be analyzed as a unit when
discussing types of denotata. In a second step, however, counties will be
looked at individually, partly in order to be able to point to differences
between them, partly to illustrate peculiarities of a certain region better, and
also to identify possible intra-county contrasts. The outline of this section
will parallel that of the previous one wherever possible to make cross-
referencing and comparisons easier. For easier legibility, there will be no
tables detailing individual distributions; interesting figures are included in
the text and in the summary in section 10.2.
10.1. Referent types
As in Section 9, semantically similar referents will be discussed together in
groups, using the same three basic classes that were already postulated in
that section. Accordingly, most denotata are expected to belong to class (1)
320 Susanne Wagner
of man-made objects. Class (2) is constituted by items referring to natural
features, while class (3) referents are body parts; for sub-divisions and
details, see page 294.
Problems of classification are bound to occur when a framework
constructed for one data set is used on a different one. Such problems are
expected here as well, since it is almost impossible that all referents found
in the West Country material will neatly fit into one of the categories
postulated for the SED fieldworker notebooks. Where they occur, these
problems will be discussed and the decisions for putting an example into
one or the other category will be justified wherever necessary and possible.
10.1.1. Man-made objects
10.1.1.1. Buildings
References to buildings, their parts and contents are distributed fairly
evenly throughout the Southwest corpora. All counties are represented;
seven examples stem from Cornwall, 10 from Devon, 10 from Somerset
and two from Wiltshire. Most speakers contribute more than one example,
as can be seen in (23) below; feminine examples could not be found.
(23) a. Twas a tall chimney, you see and I think they tried to bomb that
chimney, thinking perhaps when he fell, you know, hed do quite a bit
of damage. (FRED Dev_009)
b. [talking about a roof] That hasnt been on very many years, I said
Hes been on more than sixty years, he said How did you know?
I said I can mind when he was (gap indistinct) it was all thatched
then. (FRED Som_022)
c. (gap name) Lovely church that is if youve never seen im. Hes
worth looking at. Oh, father and mothers buried there.
(FRED Som_036)
10.1.1.2. Containers
This class is well-represented, partly as a result of the topics of the
interviews. Tubs, barrels and similar items are used in cider and cheese
making, as well as in mining ( coal tub). The informant of Som_009 is a
potter, so it comes as no surprise that he should make numerous references
to pots when relating the story of his life. Interestingly, there is one
instance of shereferring to a pot among those 21 masculine references, in a
non-significant position (neither at the beginning nor at the end of the
interview). What motivated this form is unclear.
Only ten of the total of 81 forms are not from Somerset; eight are used
by speakers from Cornwall and two by Devonians. While the comparative
ratio of number of words to number of examples is as expected for
Cornwall (Somerset: ca. 175,000 words ca. 70 examples; Cornwall: ca.
20,000 words eight forms), it is too low for Devon. Also, it is unclear
why there is not a single example of a masculine pronoun referring to a
container in the Wiltshire data.
Example (24b) very nicely illustrates and supports Ihalainens theory
that standard forms (i.e. the it here) are first used in the less accessible
positions of the noun phrase accessibility hierarchy. The speaker has it in
the prepositional object slot, while he (still) uses dialectal he in the more
prominent subject position (for details on and an investigation of this
theory, see Wagner 2004b).
(24) a. But the best corn would come this side and when that chute did fill up
his bag there so as you could tie un nicely, he would switch over, turn
the chute over in that bag and while that bag was filling he would pull
him back and tie him and put him back in a nice row and hang up
another empty one there. (FRED Som_010)
b. So, therefore, if they had to use a stone jar they used to get the basket
maker to put some wicker around it, so if he had a bump he wouldnt
break. (FRED Som_024)
c. Not fully, well he been fully, what I think he was glazed with a dark
glaze, like that other one I think, only he hasnt been fully burnt yes.
He would have been back one side, you know and the fire didnt
actually get to him and made enough. So they gave him away I
think (FRED Som_009)
10.1.1.3. Tools and instruments
Based on the topics of most of the interviews, a relatively large number of
examples for this class is expected. One of the Somerset interviews is
concerned with sailing, boats and their parts in particular. For that reason, a
sub-class ship parts is postulated, containing references to anchor, boom,
322 Susanne Wagner
flag, gaff, sail, rope, etc. Only those referents are included in this class that
do not occur in other classes.
A second sub-class is concerned with items used in cloth making, the
topic of many of the Wiltshire interviews. Various tools subsumes
denotata such as fork, knife or shovel under its heading. Once more, we do
not find a single feminine pronoun used anaphorically to refer to a tool or
instrument. As was expected, examples are numerous for this class.
Somerset speakers once again contribute more than their share with 78
relevant forms, while all other counties are underrepresented: 21 examples
stem from Wiltshire, three from Devon, five from Cornwall.
(25) a. Then one of themlet the bar fall. As he felled, he said, he hit both mi
hands, for I had hold of a rope, and bounced off the ground, come on
mi two feet. (FRED Som_028)
b. Ive got a barley fork out there and Ive had him there for years. I
used to have him when I thatched the ricks - push him in the rick to
keep the bottomreeds fromslipping. (FRED Som_022)
c. Ill tell you what I found - one of our old hay knives. I got him out and
cleaned him up and put a new handle on him; he looks alright too.
(FRED Som_027)
d. I drove into a lamp post. (v giggling) Theres the, the pavement, see
and your lamp post was right here on the curb. This was a great big
one and he got sort of battens all the way around en you know, this,
one of the anchored posts this was. (FRED Dev_001)
10.1.1.4. Vehicles
It may be recalled that it was comparatively surprising that the SED
fieldworker notebooks data did not contain many references to vehicles at
all. In the more modern material, on the other hand, we find numerous
references to all types of farming machines, particularly to tractors (often
referred to as engine in the corpora). This category interacts with the
previous one because many farming machines are vehicles but also tools
(e.g. thrashing machine). It was thus decided to establish a new category
farming machines for these referents, which will be discussed after this
class. As a consequence, only non-farming vehicles will be covered here.
Ships, boats and other water vehicles are included here, and it is not
surprising that reference to them is made with the help of feminine
pronouns in an overwhelming majority of cases. There are some unclear
masculine references to a tug boat and a steamer in Som_028
49
, which
could be explained in reference to size and strength. As these references
occur in the same interview where the speaker uses more than 100 feminine
pronouns referring to various boats and ships, including tug and steamer,
masculine reference seems not very likely. Thus, these masculine pronouns
could also be interpreted as references to the captain(s) of the vessels rather
than to the vessel itself. In the concrete examples as in (26a), however,
contrasts between masculine and feminine pronouns seem to reflect
emotional attitudes, supporting an interpretation in favour of semantic
contrasts between heandshe. Hestands for large size or negative emotions,
whilesheis smaller and/or positively connotated.
50
In (26b) and (26c), we
find two more examples supporting Ihalainens hypothesis. The speakers
use he in subject position to refer to the lorry and the train, but a neuter
pronoun in (prepositional) object position.
Masculine forms seem to be the pronouns of choice when referring to a
non-water vehicle. From todays perspective, we would have expected
more feminine pronouns. However, there are not enough examples to make
any far-reaching claims about the general situation in West Country dialects.
It should be mentioned though that seven of the 14 masculine pronouns not
referring to boats stem from women or very traditional dialect speakers. In
other words, these speakers behave as expected. Included are all references to
car, which is more frequently feminine these days, and three references to
coach(two examples) and train(one example).
(26) a. Thats why you had, you know why you had a pilot, but there was no
need to because youmfollowing the damn tug. Because he had to
keep there, hehad to keep behind unin any case. (FRED Som_028)
b. Oh I had a, I had a - I drove a lorry that was the same age as miself,
1926 Dennis and drove it for years and henever broke down on me
all that time, and to start it you had to crank the handle
(FRED Wil_006)
b. if they would change the time of the train so that the, we could
travel on it and get to work at uh, eight oclock, you see I think he
runned a bit later, to go, I dont think they in Totnes started until nine
in the milk factory. (FRED Dev_010)
324 Susanne Wagner
10.1.1.5. Farming machines
This section will discuss all types of farming machines, regardless of
whether they are vehicles such as tractor or rather tools such as swath
turner. The examples are nicely distributed across the counties, obviously
based on the farming bias in many of the interviews. At first glance, gender
assignment seems to be highly variable for this category. While farming
machines are consistently masculine for some speakers, they are
exclusively feminine for others. Both exclusive uses are easily explained,
as the traditional dialect requires masculine forms while the StE system
allows feminine ones. Only one speaker uses both non-neuter genders for
the same referent, probably as a result of the traditional system competing
with the (spoken) standard one. Size, though a factor in the relevant
examples, is not identified with either gender consistently enough to claim
that it is responsible for the choice of pronouns. Based on frequency of use
(three masculine, 22 feminine forms), feminine pronouns seem to be this
speakers first choice, implying that his system is closer to StE than to
the dialect for this category. Emotional involvement can be used as an
explanation for shifts from standard it to dialectal masculine forms as in
(27c). Although feminine pronouns do occur, only three speakers actually
use them. With the one exception mentioned above, intra-speaker variation
occurs only between feminine and neuter or masculine and neuter
pronouns, but not between masculine and feminine forms. The traditional
system thus seems to be fairly intact for this category under closer
scrutiny, the surface variability turned out to be quite systematic.
(27) a. Soon as the engine come over the main Bristol road, of course, down
hill like that shedroppedher nose and shebegin to pick up speed and
run. He couldnt handle her, he couldnt stop her. He jumped out of
her, let her run away, and shecrashed through the wall, and dropped
down about sixty feet in this ravine, and smashed herself all to pieces.
(FRED Som_014)
b. Ah, well now the swath turner was a marvelous machine because you
seehed turn it over on, turn it over on dry ground, hewouldnt put it
all together, hed turn two swaths, one there and one there.
(FRED Som_017)
c. you can work a tractor and it dont get tired like horses. When you
finish you can put it in the garage or in the shed, n all you got to do
is give ma drop of water n fill unup wi diesel HDC (Devon)
51
10.1.1.6. Creature comforts
ARTICLES OF CLOTHING
Considering that many of the Wiltshire conversations are concerned with
work in cloth mills, it is strange that this class, with only nine masculine
examples, is very underrepresented. It seems that here the traditional
gender assignment system has already given way to the standard for many
speakers, as can also be witnessed in (28b), where the neuter form seems to
be in free variation with two instances of traditional he.
(28) a. that were all splattered, she set he[riding habit] to the cleaners
(FRED Wil_001)
b. You know what the length of the cloth is before it goes in the machine,
say 65 and the designer says, I want that 60 yards long, see, well you
can calculate how much hewas shrinking by measuring the yard up,
see what I mean, he went in 61 inch, see what I mean, thats, how
many yards is that? (FRED Wil_001)
ACCESSORIES OF MODERN LIFE
Based on the fact that the oral history material investigated here is more
modern than the SED data, more references to this class are expected.
Referents include a wide array of things, from coin to computer, from
painting to paper. Example (29c) nicely illustrates the importance of
emotions in gender assignment. The speaker talks about an oil lamp that is
difficult to light, using it three times before shifting to he in the last
reference, where he is clearly annoyed.
Although neuter references occur regularly for this class, not a single
feminine example could be found. This is noteworthy in comparison with
the observations that other researchers and also this author have made. In
modern fiction and everyday conversations, modern technological
appliances such as computers are shes rather than hes (if not its). We
should thus conclude that the traditional system of gender assignment has
been substituted by the StE system in the Southwest, at least for nouns
referring to accessories of modern life.
(29) a. you used to get these Litmus papers, you know what I mean to say.
These what do change the colour, and if he did show green, see, he
[cloth] were alright, see, but know up to you got to a certain point he
326 Susanne Wagner
might be red, purple, blue and all of a sudden hegoes green. As soon
ashegoes to green you know hes [cloth] alright (FRED Wil_001)
b. Now, oh, you aint got en[tape recorder] on, have ye?
(FRED Wil_010)
c. Little oil-lamp, I had to get a match and light it and as I was coming
home it blew out, and there was I by there trying to light it and I, he
wouldnt light with nearly a whole box of matches. (FRED Wil_023)
FOOD AND DRINK
This section features a number of irregular referents belonging to the
class of liquids, with cider as the expectedly most common one. It may be
recalled that the SED fieldworker notebooks contained a number of
references to non-count nouns of this class as well. In general, however, the
number of entries for this class is, as expected, small. Most nouns referring
to items of food and drink are non-count nouns, which should receive
neuter reference, and obviously do so in most instances. Non-dialectal
references in the Southwest corpora are few; denotata include butter, cider,
milkand a type of cow feed.
In (30b), the informant describes the process of cheese making. While
the first references are still to milk and consequently neuter, the change
from milk to cheese is paralleled in the change from it to masculine forms.
A period of indecisiveness as to the status of the milk-to-become-cheese
mixture is obvious in the use of both forms in the middle of the passage.
Cheese, both in its milk-based and apple-based variants, is a frequently
occurring referent. The classification of cheese is problematic in some
respects (cf. also 10.1.1.7). Cider and cheeseare by far the most common
and most evenly distributed referents in the Southwest material.
(30) a. you just simply make the butter and put your hand in the salt and
put so much on en, like. (FRED Con_005)
b. And tip it back in when you had it the right temperature. Well, she
knowed shed got electric steamers, boilers and the cheese stuff was
all heating, you see. (gap indistinct) Like when youd finished she
used to have to lift it out to help it on, lift it out then cut it up in chunks
and take it out and then keep it so long wrapped up because it was
warm. Well then you had to grind it, salt it, put it in these vats, put it
in the press. Next day youd take un out, turn it over, put a fresh cloth
aroundit, turn it over, press un again and then the day after that you
would take un out and youd sew, put a cloth over the top again and a
canvas around the side. Sew it up and pin it through. And then youd
put un back in the press for the next morning and that press was tight.
Well then youd take un out, put un in the cheese roomon a tray, and
you would turn emevery day. (FRED Som_013)
c. They used to put potatoes in the, the loaf of bread, you know, and
when you cut un, you cut half, half a potato and half a bread.
(FRED Som_028)
d. [Int.: what did you do with the milk?] Oh, separated en.
(FRED Con_005)
10.1.1.7. Nature re-modelled
Once more based on the agricultural bias of many of the interviews, we
expect to find numerous examples illustrating this class. References to ricks
and other forms of bundling cereal plants (sheaf, bundle) as in (31a+b) are
particularly frequent. The distribution of referents across counties is fairly
even, with 30 examples from Somerset, six from Devon and three from
Cornwall. Wiltshire speakers only contribute one example, probably based
on the fact that agricultural topics are not discussed as often as in the other
counties. Feminine forms do not occur.
(31) a. And when youd dig a ditch, if you hadnt done himright for the water
to run, that wasnt right. (FRED Som_013)
b. Oh, Ive thatched a rick, built a rick and thatched him.
(FRED Som_22)
c. Course if one sheaf was a little bit out or anything Id patch himand
hed go right up. (FRED Som_22)
10.1.1.8. Man-made objects summary
The most noteworthy conclusion to be drawn from the data presented in this
section is the following: Although examples of both standard pronouns (neuter
it) and spoken standard forms (generally she) do occur in the modern oral
history material, the traditional dialect system of gender assignment is alive
and well. Masculine pronouns are frequently used anaphorically in reference
to count nouns associated with man-made objects. The sheer quantity of
examples is also an indicator for this. When comparing the SED fieldworker
notebooks data with the modern Southwest material, we would have expected
fewer examples for the latter, based simply on the amount of material and
328 Susanne Wagner
number of speakers. It seems that the questionnaire style of the SED did
indeed inhibit speakers from using their local lect, resulting in a variety much
closer to the standard than it would have been in free conversation.
Moreover, we were able to observe in many examples that Ihalainens
accessibility hypothesis seems to be a very good explanation to account for
the presence of both neuter and masculine forms referring to the same entity
in a single utterance. While the subject form he still defends its territory,
positions further down (to the right) in the accessibility hierarchy are losing
ground. Thus, even though masculine forms are still very frequent in
comparison with other traditional dialect phenomena, we have to assume that
the dialect system is slowly giving way to the system of StE. The path it
follows is very often the one outlined earlier: The StE forms invade the
territory of gendered pronouns from the less accessible positions along the
NP Accessibility Hierarchy. Thus, many speakers still use he in subject
contexts, but have already shifted to it in oblique contexts. Table 43
summarizes the results from the tables presented above.
Table 43. Gendered pronouns referring to man-made objects (SW)
referent class subclass #masc. ex.s #fem. ex.s #of sp.
buildings 29 13
containers 81 1 17
vehicles 32 142 12
farming machines 65 23 21
creature comforts
articles of
clothing
9 4
accessories of
modern life
37 14
food & drink 49 14
total 449 166 68
10.1.2. Nature
A considerable amount of the Somerset material stems from a project on
cider making. It is thus not surprising that apple is one of the most
frequently occurring referents. All examples stem from Somerset, two of
which can be found in (32a) and (32b). Example (32a) shows yet again the
variability of the system, with instances of both it and he referring to the
apple. It should be noted that this example again conforms with Ihalainens
hypothesis concerning the stages of change from dialect to standard.
Classification is difficult for cheese, which occurs as cider cheese and
regular cheese made from milk. Both cheeses are ultimately man-made
and serve as food (the cider cheese is often used as pig feed), and it was
thus decided to include cheesein the food and drink section (see above).
(32) a. [talking about the right time to pick apples]
Well, like if you picked one an cooked it early heisnt same as when
hes been picked and kept, is he? (FRED Som_013)
b. but the black withy is crips. Brittle. Well hes crips, he aint no
good, break himup like that. (FRED Som_022)
c. Oh, I remember when that were planted that tree, they cut it down just
before they built the house. I went one day wi the car, theyd cut it
down, they had roots and all out, and I remember when it were a little
one about as big as mi little finger, I think it were planted about the
turn of the century there, but when they cut it down hewere about that
big round, yes. I didnt know how big it was not til after I come down
round there with the car, and I saw they were cutting the Walnut tree
down, but when we were kids it was about as big as mi thumb and he
were up about six or seven foot probably, hadnt been planted very
long, when they cut it down then I should think hewere about that big
round, cause they had a timber carriage and that there to load imup
on, and thats how big it got in sixty or seventy years.
(FRED Wil_003)
A closer look at example (32c) gives some interesting input for the
discussion on the division of tasks between gendered and standard
pronouns. The topic of the conversation is a walnut tree that the speaker has
known for his whole life. He remembers its being planted and relates his
emotions when he saw it being felled. It seems that two or even three
systems interact in producing the personal pronouns of this passage.
The first five references are it, three in direct object and two in subject
position. We can explain all its most easily by claiming that the speaker is
using StE here. However, staying within the dialect framework, another
explanation is possible. Ihalainens hypothesis can be held responsible for
the three instances in object position. For the two examples in subject slots,
330 Susanne Wagner
we can come back to a factor that was thought relevant in much of the
literature on this topic size (cf. e.g. Mathiot and Roberts 1979). Although
studies could not establish a direct link between size and gender
assignment, folk belief uses size frequently as an explanatory factor, for
example with animal referents a tiger is much more likely a heor shethan
an ant, which will probably be it.
The speaker recalls how small the tree was when it was planted. As soon
as he refers to the grown tree that is being felled, he shifts to hein the sixth
reference, either because of its size or because of the subject position (or
both). The next two references are neuter once more, again based on small
sizeor for the first example possibly also on the atypical syntactic context
(subject complement). The next reference is to the young tree in subject
position, resulting in a masculine form. The following it is again in object
position, referring to the grown tree, as is the masculine pronoun in subject
position after that. Following this is the only masculine form in object
position, obviously a remnant of an earlier stage of the speakers dialect.
Finally, the speaker uses another neuter pronoun in non-subject position.
This excerpt nicely illustrates the factors that are competing for gender
assignment. Although it may seem unsystematic at first glance, the gender
assignment here is very systematic, if highly complex.
With 49 examples, Somerset speakers are once more the most prominent
contributors to this class. Nine examples stem from interviews from
Wiltshire, three from Devon and two from Cornwall. It is also obvious
from the numerous examples that natural features are either neuter or
masculine in the Southwest, but never feminine.
10.1.3. Body parts
Denotata of this class are rare, which seems surprising insofar as various
aches and ailments are expected to feature prominently in conversations
with elderly people. A closer investigation of the contrasts between neuter
and masculine forms may reveal that West Country speakers abandoned the
traditional system and have adopted the StE one for this category, although
this does not seem very likely.
(33) a. I cant clench that hand see for one thing, hewont go back no more
than that. (FRED Som_016)
b. [what to do when you had a toothache in the old days]
You did go into the doctor and he did get hold of un and drag un out
with a pair of pinchers or something, you see. (FRED Som_031)
10.2. Individual counties
As laid out at the beginning of this section, we will now change our
viewpoint from a macro- to a microperspective. Although the material
constituting the Southwest corpus stems from comparable sources, we
cannot assume unquestioningly that they are equal. Numerous factors can
influence the type of variety that is found in an oral history interview
situation. Without having access to all the relevant background
information, however, we can only make certain assumptions about these
factors based on the general degree of non-standardness that we observe
in the interviews. As many of these factors are connected with the
behaviour of the interviewer, and as many of the interviews in the
Southwest corpus were conducted by only a handful of interviewers per
county, a closer look at the individual counties might help to identify such
disturbing factors.
Moreover, the comparison of the counties is interesting from a statistical
point of view. Based on the total word number and total of gendered
pronouns, combined with the overall frequency of pronominal forms, it
should be possible to make claims about the degree of genderedness of
certain counties (and/or speakers). Some speakers may use an over-
proportional number of non-standard pronouns, while others rank far below
the average. A comparison of the four main counties under investigation
could also help to pinpoint certain changes that may have occurred since
SED times. Table 44 summarizes the figures the following sections will be
based on, showing only those examples that stem from one of the core
Southwest corpus texts. One of the Somerset texts will be excluded from
the calculations most of the time, as its unusually high frequency of
gendered pronouns (179 in an interview of ca. 13,500 words, largely owing
to more than 100 references to boats and 25 references to coal tubs) distorts
the overall impression in this group of texts. Figures in brackets refer to the
totals when this text is excluded.
332 Susanne Wagner
Table 44. Summary of gendered pronouns in the Southwest corpora
referent class #of masc. ex.s #of fem. ex.s #of speakers
buildings 22 (18) 0 10 (9)
containers 78 (53) 1 14 (13)
tools 93 (69) 0 24 (23)
vehicles 32 (27) 109 (1) 9 (8)
farming machines 36 23 14
creature comforts 72 (62) 0 22 (21)
nature re-modelled 32 0 11
nature 59 (56) 0 16 (15)
body parts 6 0 4
total 430 (359) 133 (25) 50 (49)
10.2.1. Cornwall
Speakers from the main Cornwall corpus contribute 42 examples out of the
total of 384, i.e. 10.9%. This is almost exactly twice as many as we would
have expected based on the percentage of total words for this county
(18,900 of 319,800, i.e. 5.9%). Based on our observations in the previous
section, we can thus say that English in Cornwall and West Cornwall in
particular is much more West Country like than generally assumed.
One speaker is responsible for more than half of the examples (23), which
are spread fairly evenly across the different semantic domains. This informant
shows the most traditional lect of all Cornish speakers, using many typically
Southwestern and also specifically Cornish dialect features.
52
It is this speaker
as well who uses a striking construction when talking about a situation or
general circumstances: lets put en this/that way.
A second speaker contributes eight examples, the third six and the
fourth speaker five gendered pronouns. All speakers contribute to the
domains containers and creature comforts, two speakers to those of
buildings, tools and nature re-modelled respectively, and one speaker each
to the categories vehicles and nature. No speaker uses a masculine or
feminine form to refer to a body part in the Cornish interviews. An
additional text that is not part of the main corpus provides the four
remaining examples, adding up to a grand total of 46.
10.2.2. Devon
Although they contribute 16% of the words constituting the Southwest
corpus (51,100), speakers from Devon use only 16 non-standard
(masculine) pronouns (or 4.2% of the total). The material from Devon is
relatively standard in comparison with that from Cornwall or Somerset,
which may explain this ratio. It should also be mentioned in this context
that the Devon interviews are not primarily concerned with agricultural
matters, thus contrasting thematically as well with the data for Cornwall
and Somerset. Although universal non-standard features such as non-
standard agreement, regularized irregular verbs and reflexive pronouns,
non-standard use of relative and demonstrative pronouns, etc. are
frequently found in the interviews, typically Southwest patterns are rare.
The distribution of non-standard forms between the five speakers is
relatively even. One speaker contributes five, one four, two three and one
speaker one example. The denotata can be categorized as buildings, tools,
vehicles, farming machines or creature comforts. No examples belong to
the domains of containers, nature re-modelled, nature, or body parts.
When comparing the modern Devon material with data from other
dialect corpora (e.g. the Helsinki Dialect Corpus), it turns out that the
material is by no means as exceptional as assumed. The Devon part of the
Helsinki Corpus, recorded in the mid-1970s, contains even fewer gendered
pronouns proportionally than the modern Devon material. Thus, judging
from the Helsinki material, one should also consider the possibility that the
Devonian dialect has already advanced further on the path towards StE than
some of its neighbouring dialects since SED times.
10.2.3. Somerset
The 30 interviews from Somerset constitute almost exactly half of the total
material from the Southwest in terms of word number (161,000 of 319,800,
i.e. 50.3%). Based on the almost standard SED data for this county, a much
lower percentage of non-standard pronouns was to be expected. J ust the
opposite is true, however.
With 258 examples, Somerset speakers provide more than two thirds
(67.2%) of all examples a result that comes as a surprise after the
disappointing figures from the SED data. Given the numerous examples, a
relatively even distribution across semantic domains is to be expected.
334 Susanne Wagner
Table 45 shows the absolute and relative frequencies of denotata for the
Somerset material.
Table 45. Distribution of gendered pronouns across semantic domains (modern
Somerset material)
referent class
#of examples
Somerset
#of examples
total
% Somerset data of
total
buildings 6 18 33.3
containers 46 54 85.2
tools 43 69 62.3
vehicles 10 28 35.7
farming machines 48 59 81.4
creature comforts 26 62 41.9
nature re-modelled 28 32 87.5
nature 46 56 82.1
body parts 5 6 83.3
total 258 384 67.2
Only in three domains, namely buildings, vehicles and creature comforts,
do Somerset speakers contribute fewer than expected forms. In all other
categories, they come either close to the overall ratio of 67.2% as for tools,
or even surpass it with ratios of around 80% for the remaining categories.
Contributions per speaker range from one to 45 examples, with eleven
speakers above and 19 below the mean of 8.6 examples per Somerset
speaker. Those eleven speakers use a total of 186 non-standard pronouns
between them (72% of the total), but only 37.9% (61,000) of the Somerset
word total.
53
Only four speakers, among them the three with the most forms
(41, 2x 25), use significantly moreforms than their colleagues.
54
Even
when we exclude these four texts from the calculations, Somerset speakers
still use more gendered pronouns than expected (153 examples =59.3% in
134,100 =42% of words).
Based on these figures, any distorting influence of individual speakers
can be ruled out. Explanations for the unexpectedly high frequency of non-
standard pronouns in the Somerset material thus have to be sought
elsewhere. The informants of the modern corpus material are approximately
one generation younger, aged 70-80 in the 1980s, than the SED informants,
who had reached the same age in the 1950s. Although the modern material
does not cover all of Somerset as evenly as the SED did, locations are more
numerous and more widely spread than for any other area, and a certain
representativeness should be guaranteed. We are thus forced to fall back on
the same conclusion as in Section 8, namely that the Somerset speakers
selected for the SED are not representative of the countys dialect. The only
alternative analysis would be that, judging from the modern corpus
material, Somerset in the 1980s is more dialectal than in the 1950s an
extremely unlikely scenario given what we know about mobility, social and
historical developments, and dialectology in general.
10.2.4. Wiltshire
As has already been emphasized earlier, the modern corpus material for
Wiltshire stems from two different sources. While background information
is available for the material from Trowbridge, next to nothing is known
about the informants, recording details, etc. of the Wiltshire Folk Life
Society data. It can be assumed, however, that the WFLS material was
recorded in the 1970s or early 1980s, while most of the recording dates of
the Trowbridge material are in the 1990s. In addition, the Trowbridge
material is concerned with the history of cloth weaving, the primary
industry of the region until recently, and not with farming, as is most of the
Somerset and Cornwall material. In this respect, the Trowbridge and Devon
material are comparable. Furthermore, many of the Trowbridge informants
live in towns rather than villages.
Much of what has been said about the material from Devon also holds
for the Trowbridge material: Although universal non-standard features are
frequently encountered in the data, and some typically Southwestern
features (unemphatic do in particular) are also ubiquitous, the dialect in
most of the interviews is much closer to StE than that of the Cornwall and
Somerset data. This is consequently also reflected in the figures concerning
gendered pronouns: The Trowbridge material constitutes almost a quarter
of the corpus material (75,100 words, i.e. 23.5%), but only 47 examples (or
12.2%) of non-standard pronouns are found in these data.
Contributions per speaker range from one to 13 (none of these
differences is statistically significant), with a heavy bias towards tools as
the favoured semantic domain, represented by 19 examples from four
speakers. Other domains where Trowbridge speakers use masculine
pronouns to refer to inanimate entities are buildings (two examples),
vehicles (one example), creature comforts (18 examples), and nature
(seven examples).
336 Susanne Wagner
Only in the data from the Wiltshire Folk Life Society do total word
number and number of examples nearly match (13,700 words, i.e. 4.3% of
the total, and 21 examples, i.e. 5.5%). J udging from certain features in the
speakers dialect (e.g. demonstrative thick(y), relative particle as), the
WFLS material is very traditional in nature. One speaker uses two, two
three each, and one speaker 13 exceptional pronominal forms. With 10
examples, denotata stem most frequently from the domain of farming
machines, to which all but one speaker contribute. Another four pronouns
refer to car, leaving only one or two occurrences each for the domain of
tools (one example), creature comforts (two), nature re-modelled (one),
nature (two) and body parts (one example). When comparing the results
from the two Wiltshire sub-corpora, we may be witnessing an apparent or
real time change. WFLS informants use a variety of English that is (still)
very close to that recorded by the SED fieldworkers. Trowbridge
informants, on the other hand, still use many universal dialect features, but
their inventory of Southwest features seems already reduced in comparison
with traditional accounts and also the SED data. The development thus
seems to be one towards StE, which we would expect based on the location
of Wiltshire on the (present-day) border of the West Country.
10.3. Summary developments and changes
The first result from the corpus analyses is striking in its clarity. When
comparing ratios of word numbers and numbers of examples of non-
standard pronouns, the following picture emerges: The ratio of pronouns to
word numbers forms a West-East continuum, with Cornwall at the top,
followed by Somerset. Both counties lie above the expected average ratio
(>1), while Wiltshire and Devon fall below it. Details can be found in
Table 46. Even when the WFLS material is excluded, Wiltshire is still third
on the list, with a ratio of 0.52. In other words: The further westwards we
move, the more non-standard pronouns we encounter. The status of Devon
does not seem to fit into this image, but when we consider a second
continuum, namely North-South, new possibilities emerge. As the Devon
data come from South Devon exclusively (Totnes and surroundings), it
cannot be ruled out that data from the North would be closer to the
Cornwall and Somerset data. Whether this North-South continuum has
developed recently, i.e. over the past few decades, or has simply become
stronger since SED times is impossible to say. But one can clearly observe
the gravitational pull of StE or London, which is ultimately responsible for
both continua.
Table 46. Ratio of word numbers to number of examples (modern corpora)
county % of examples % of words ratio ex.s to words
Cornwall 10.94 5.90 1.85
Somerset 67.19 50.34 1.33
Wiltshire 17.70 27.77 0.64
Devon 4.17 15.98 0.26
total 100 99.99 1.00
When consulting a map of the modern Southwest, it will be noticed
immediately that almost all locations with a high percentage of gendered
pronouns are located north of the modern access route to the Southwest,
A30. In addition, Trowbridge is at the Wiltshire-Somerset border, as far
west as one can move in the county. Totnes, on the other hand, is almost as
far south as possible in Devon. These observations are supported by
evidence from other areas of (linguistic) research: First, when looking at
urbanized areas in England around 1950 (>400 people/sqm), it becomes
obvious that the region around and particularly South of Exeter was
urbanized in the 1950s already, while North Devon, Cornwall and most of
Somerset (except Bristol) remained essentially rural in character. Second,
even when looking at analyses of completely different phenomena, (for
instance the absence of third person singular -s; cf. Viereck, Viereck and
Ramisch 2002: 98), the same picture emerges: South Devon behaves
heterogeneously, shifting from complete absence to 100% presence and
back to complete absence of -s forms, and in Cornwall, including West
Cornwall, we find zero forms very homogeneously. When comparing this
map with Figure 7, definite similarities will immediately be identified.
Moreover, a quantitative study of lexical contrasts based on SED data
shows that bundles of heteroglosses separate the Central Southwest (more
or less at the border of Dorset) from the Eastern Southwest more strongly
than the Southeast is separated from the Southwest a result that once
more ties in with our observations (cf. Viereck, Viereck and Ramisch 2002:
94). The same analysis also shows that the contrasts between Devon and
East Cornwall are more pronounced than those between East and West
Cornwall, pointing towards the homogeneity rather than heterogeneity of
English in Cornwall.
55
338 Susanne Wagner
In comparison with the analyses in the previous section, the following
similarities emerge: (West) Cornwall should be considered core Southwest
territory in all respects, as the investigations of both the SED data and the
modern oral history data have shown. J udging from the modern corpus
material for Somerset, the suspicion we had in the preceding section is
confirmed: TheSED informants for Somerset are not representative of the
dialect of the time and area. The situation for Devon and Wiltshire is not as
straightforward. While most of Devon clearly belonged to the core
Southwest at SED times, the modern situation is unclear. Our data indicate
that the county is moving towards StE much faster than its neighbours, but
we cannot dismiss the claim that our Devon material may not be
representative of the county as a whole. As our data for Wiltshire largely
stem from the western part of the county, it could be argued that the overall
results pointing towards Southwest status of the county could be equally
skewed. However, when comparing the modern data with the SED data for
Wiltshire, it should by no means be considered impossible that the material
used here describes the present situation fairly well.
Apart from these noteworthy facts, another fact must be acknowledged
as remarkable, namely the stability of semantic domains to which gendered
pronouns generally refer. Even though nothing can be said about general
frequencies due to basic differences in the make-up of the two data pools,
most denotata of non-standard pronouns were and are tools and
instruments. It seems that Elworthys and Barnes class of man-made
objects is still very relevant in gender assignment. The corpus analyses
have also shown that feminine pronouns are (still) quite rare in West
Country dialect. The only entities which are frequently feminine in
anaphoric references are vehicles, a use that has most certainly been taken
over from StE. In light of the facts from modern studies on spoken English
in general and scholars general opinion as presented in grammars of PrDE,
we would have expected a much higher frequency of feminine forms than
we actually encountered. Based on all of these aspects, we can conclude
that the traditional West Country system of gender assignment has
remained largely intact to the present day, at least in elderly (rural)
speakers. It will be interesting to see how far the New World and its
dialects have influenced this system. An answer to this question will be
attempted in the following sections, where corpora of Newfoundland
English will be analyzed.
11. Material from Newfoundland
11.1. MUNFLA material
For the category of man-made objects, a number of conclusions can be
drawn from the analyses of the oral history material from MUNFLA (for
details and figures, see Wagner 2003). First, the traditional (West Country)
system of gender assignment seems to have survived fairly intact in
Newfoundland. In comparison of total word numbers and total number of
gendered pronouns for the Southwest and MUNFLA corpora, the
Newfoundland data lead by far close to 900 forms in 132,000 words.
56
Second, however, in comparison with the Southwest England material,
it looks as though feminine forms are slowly invading the system. In some
semantic domains (in particular accessories of modern life), these forms
already constitute almost 30% of the total number of forms (the domain of
water vehicles can be neglected here as it obviously always has been
feminine). Table 47summarizes the individual figures of forms referring to
man-made objects in the MUNFLA material.
Table 47. Gendered pronouns referring to man-made objects (MUNFLA)
buildings 160 3 14
containers 33 1 12
tools & instruments 116 4 18
vehicles 1 6 3
water vehicles 1 322 24
creature comforts
accessories of
modern life
144 41 24
food & drink 13 5
total 508 377 38
There is one domain of referents that deserves special attention in the
MUNFLA material, basically because it contains items of a class that is
much more prevalent in the MUNFLA material and which was either not at
all or only marginally encountered in the corpora analyzed so far: abstract
referents.
340 Susanne Wagner
The label abstract referents is used here to refer to pronouns denoting
abstract count nouns such as storyor song. In addition to these denotata, we
also find a handful of pronouns of the non-referential type described in
section 5.3. Examples (34a) to (34c) illustrate the first, (34d) and (34e) the
second of these uses. The speaker of (34d) describes how dynamite was
used to make a passageway for a boat through ice. Although the feminine
pronoun could refer to the boat or the stick of dynamite, an interpretation as
non-referential is equally likely, particularly since the structure of the
elements involved resembles the one identified as prototypical in such
cases (X-S-V). The same structure is also used in (34e), where the feminine
pronoun seems to be cataphoric, referring to the following events, rather
than anaphoric.
57
(34) a. Ohhes the hell of a long story. (MUNFLA 70-003: C631)
b. I learned, if I hear a song, well say somebody sings all night and if I
mind to know en, (gap indistinct) I could sing entoday.
(MUNFLA 70-037: C751)
c. And hes [song] all there but theres different words in it you know.
(MUNFLA 70-037: C751)
d. he used to put the dynamite down by the side of the ship to loosen
up the ice fer to get back and forth see, and they used dynamite at that
time and they used to heat up the poker and heave out the red hot
poker to put to the fuse and when they would touch the fuse it would
be as long as that and the powder used to go right thru the cap and
fling shed go, and so (gap name) he had two men killed with the
dynamite (MUNFLA 73-046: C1958)
e. I only come down to help pull en [house]. But nevertheless I said,
all right boys, straighten out, and away dey goes, my son, straightened
out, we took dat house and here shecome. (MUNFLA 72-089:
C1187)
Five speakers use 17 masculine forms to refer to abstract referents (poem,
one example; song, 15 examples; story, one example), while five instances
of non-referential sheare used by three speakers.
11.2. Traditional versus modern uses of pronouns the past and future of
gender assignment in Newfoundland English
The analysis of the MUNFLA material allows two basic conclusions: First,
Newfoundland speakers adhere to the traditional West Country system of
gender assignment more strongly than their fellow speakers in the mother
country. This result is based on the comparison of the relative frequencies
of gendered pronouns in the two corpora, with a higher frequency of non-
standard pronominal forms the Newfoundland data. Southwest speakers, on
the other hand, use proportionally fewer gendered pronouns. In other
words, more Southwest English speakers have adopted the StE system.
Second, the Newfoundland data show a tendency not observed in the
Southwest material. While British speakers seem to be shifting (or have
already shifted) from the traditional dialect system ( masculine forms)
towards the (written) standard system ( neuter forms), Newfoundland
speakers seem to prefer the spoken standard system ( feminine forms)
rather than the StE (neuter) system. This can be concluded from the higher
frequency and more even distribution of feminine pronouns in the
MUNFLA corpus. While feminine pronouns occurred in only three of the
nine semantic categories in the Southwest material, the MUNFLA material
contains feminine forms referring to all but two of the categories. Those
categories, nature re-modelled and body parts, are represented with
only very few examples, however, thus possibly precluding the occurrence
of feminine forms simply though lack of examples. Feminine forms
constitute only about 5% of all examples (25 forms out of over 500,
excluding references to boats) in the Southwest corpora, but they make up
approximately 10% of all gendered pronouns in the MUNFLA data (60
forms out of 621, excluding references to boats).
Based on what is known about Newfoundland settlement history, the
present state of gender assignment in West Country Newfoundland
dialect(s) can be explained as follows: Settlers brought with them the
(strict) system that has been described in 19th-century literature on
(English) West Country dialects. Although contact between them and other
settler groups was relatively restricted, it must have existed to a certain
extent. None of the varieties the West Country settlers came into contact
with used masculine forms in the same manner as their mother lect. Where
the pronominal system did offer a choice, it was between neuter and
feminine rather than neuter and masculine forms.
342 Susanne Wagner
In some domains of usage, contact with speakers of other dialects as
well as with standard speakers resulted in a (or possibly two) shift(s) in the
traditional paradigm: Neuter forms became the unmarked choice in some
domains, and only when a special need for emphasis was felt to play a role
( discourse-pragmatic factors), non-neuter forms were selected
( approaching StE). For some semantic domains, feminine forms as they
were used in other dialects and were also possible in some instances in StE,
began to compete with the traditional masculine forms. As a result,
Newfoundlanders today use standard it alongside dialectal she and West
Country he. All these factors come together to form a unique system that is
conservative in some aspects and modern in others.
Where the change from the traditional to the StE system is in progress, it
seems to follow the path outlined by Ossi Ihalainen: The standard form (it)
is first used in non-subject positions. The form he, which may also be used
in emphatic object position ( pronoun exchange), is the only remnant of
the traditional system in many ideolects. This theory is supported by
evidence from individual utterances as well as more general observations
from whole interviews (cf. Wagner 2004b).
If one wanted to make any predictions about the future of gender
assignment in Newfoundland English, it would be wise not to
underestimate the strength of the traditional West Country system in
claiming that a change towards StE is already dawning on the horizon. In
comparison with Southwest England, Newfoundland English is much more
conservative and successful in preserving and sometimes re-introducing or
spreading features that have long since died out in the mother country.
11.3. Folktalesmaterial
Although not as well represented as in the corpora analyzed so far, non-
standard pronominal references to man-made objects constitute the majority
of gendered pronouns in the Folktales corpus as well. As was expected, the
traditional nature of the variety of English used for the telling of tales is
reflected in the data. Masculine pronouns are generally employed to refer to
man-made objects. Only one sub-class contains more than impressionistic
evidence of feminine forms ( accessories), the use of which conforms
with StE rules (gun =she). As in the MUNFLA data, ships and boats are
exclusively feminine for Newfoundland speakers, regardless of genre.
Table 48. Gendered pronouns referring to man-made objects (Folktales)
buildings 30 13
containers 54 1 10
ships and boats 77 8
creature comforts
accessories of
modern life
73 3 7
food & drink 22 4
total 241 81 13
As in the MUNFLA data, the class of abstract referents deserves special
attention in the analysis of the Folktales. While gendered pronouns
referring to man-made objects are very frequent in those varieties which
employ them, they become less and less frequent the further one moves
away from the concrete towards the abstract denotata. The traditional
system of gender assignment as described in the 19th-century literature
does not allow such uses at all. Some instances of speakers using
masculine forms to refer to abstract denotata were already encountered in
the MUNFLA corpus. The figures from the Folktales corpus, however, are
even more striking.
Part of the ritual of story-telling is the fieldworkers question about
the origin of the tale (i.e. when and from whom the teller first heard it).
When relating the tales history, the teller usually refers to the tale
repeatedly. The pronoun that is used in those contexts is generally
masculine (he or un), resulting in some 90 masculine references to story.
Some illustrative examples can be found in (35).
(35) a. Head Card Player o the World. Uh? You [i.e., Int. B.] heardun.
[recorder off] Now this uh ... I just as well tell un. I think I ... I think I
can manage to tell unalright. (Tale 012)
b. [Int: Where did you get THAT one Mr. Snook?]
Hey? Thats one I got fromUncle J ohn Martin. Fella is dead now.
[Int: Where was he? Here in...]
He belonged Harbour Breton an I got he[ref. =man or story] up erm
... in Connaigre Bay. I only wants to hear un once an that settled -
344 Susanne Wagner
an Ill get un. Now ... song I got to have unwrote off, I got to look un
over hebout twenty-five times. (Tale 007)
c. [Int: How many times did you hear it?]
Only heard it once.
[Int: An you ... an you ve been tellin it ... ]
Yes. Only hear ... only hear it once, I only want to hear a story once
an I knowed unall then.
[Int: Mm-hm. When did you first start to tell it?]
When did I first start [to] tell un? Oh uh ... oh a long spell ago uh .. (I
started) ... first start to tell un. Long spell ago, must be thirty-five year
ago I say ... start to tell un. Now Ive aknowed that story ever since I
was about ... about forty year old. An I can still remember un.
(Tale 038)
d. [Tellers wife]: (Now be) hes goina have her all fooled up. [aside:] I
got un mixed up. I ... yes I got un all fooled up now. I got un all
fooled up now. [recorder off] (Tale 005)
In (35a), the teller uses masculine pronouns to refer to song as well as to
story. (35c) contains neuter alongside masculine forms. The context makes
it likely that the neuter (non-dialectal) forms are triggered by the
fieldworkers use of it. As soon as some distance to the neuter forms has
been established, the teller switches (back) to his usual masculine forms.
(35d) is noteworthy as it supports one of the (by now) commonplaces of
sociolinguistic research: Women tend to use a variety closer to the
standard, while men will use the regional / vernacular variety (longer) (cf.
e.g. Labov 1972: 243). The tellers wife uses a feminine pronoun, that is
the spoken standard form, to refer to the story, while her husband uses
traditional (regional) masculine forms.
Another abstract denotatum found in the Folktales corpus is bet, around
which one of the stories evolves. As already mentioned, one speaker refers
to songwith masculine pronouns. Yet another story involves mathematics:
One of the protagonists is tested by having to solve a sum, which is
frequently, but, as example (36a) shows, not always masculine.
In addition to these repeatedly occurring denotata, one speaker uses un
to refer to a trip (36b), and in (36c) and (36d), we encounter a by now
familiar use of she non-referential. In (36c), she is used to refer to the
event of the pile of chairs falling down. Although identical in wording, the
interpretation of (36d) is less obvious, but a non-referential interpretation or
an unspecified reference seems most likely.
(36) a. Oh she said Ive got a sumhere she said the teacher gi me to
do she said an I cant do it. She said I cant do that sum she
said tis too hard for me. He said Let me have a look at un.
Hah! she said Whats the good to show un to you she said
Hard Head? She said No good to show unto you /she/ said you
knows YOU cant do un. Well he said theres no harmfor me to
lo[ok] ... SEE unI dont suppose. (Tale 037)
b. Well he said J ack youre goina finish your trip? Yes J ack
said. Imgoina finish un. (Tale 010)
c. ... was in the kitchen preparin ... a meal for us. I was a devil-may-care
an mi father didnt care what I do. I piled all the chairs one an top o
the other. Well they were right to the loft. An I said to the sailor
Now if you cant climb up one side an come an the other youre no
kind of a sailor. Well up goes one of emyou know an he come down
alright an th other side. Well HE done it well say, th (other) one
had to do it. He climbed up you know an just as he got an the top part
away she goes helter to skelter all over the ki ... all over the place.
(Tale 141)
d. Now he said thats all youre gettin. An now dont think youre
goin to serve me like you served the ... like ... [laughing tone] Bill he
said because youre not. Uh ... uh ... th old man said Thats what
you thinks. Huh! [laughs] An he jumped up. An when he did J ack
jumped up too an he picked up the hot wa[ter] ... the boil water, he
dumpedun down over his head an away shegoes. [laughs] Cap hair
the whole lot down on the floor. [laughs] An th ol man took off an
J ack after un. Very good. (Tale 003)
Summarizing the results from the Folktales, the distribution of gendered
pronouns among the different semantic categories generally does not differ
drastically from those observed for the other corpora. While the category of
tools and instruments held the highest proportion of non-standard forms in the
Southwest corpora, masculine (or feminine) forms referring to those denotata
are comparatively rare in both Newfoundland corpora. When excluding
vehicles from the calculations, tools and instruments reach 24.7% in the SED
fieldworker notebooks and 23% in the modern material of all gendered
pronouns in the Southwest corpora, followed by non-standard forms referring
to natural features (15.9% and 16.6% respectively).
In the Newfoundland corpora, on the other hand, the category of
accessories contains most instances of non-standard pronouns (29.1% in the
346 Susanne Wagner
MUNFLA corpus, 24.9% in the Folktales data). The next four places on
this frequency list are occupied by the same categories in both corpora,
with tools ranking third, natural features fourth, and buildings and
containers ranking second and fifth respectively. (The second place of
buildings in the MUNFLA material is probably caused by the house
movers frequent reference to houses, which slightly skews the picture.)
The close parallels between the corpora from the same area are
surprising and rather unexpected in their clarity. The different priorities
are very likely based on the contrasts in the major occupations of the
informants. While most of the Southwest speakers are farmers, the
Newfoundlanders are fishermen.
In one domain, however, the Folktales corpus deviates greatly from the
other corpora: With over 140 examples, abstract denotata constitute more
than a quarter of all non-standard pronominal references in the corpus (145
out of 527, i.e. 27.5%). Although it is possible that the Folktales simply
contain more instances of anaphoric references to abstract nouns, another
explanation is more likely. Storytelling is a very traditional genre, which
includes the use of certain formulaic expressions as well as the use of a very
traditional (conservative) language variety. Even though the use of masculine
forms referring to abstract nouns is not really a traditional feature, the use of
gendered pronouns as such certainly should be classified so. How and why
this extension (from only count nouns to abstract nouns) occurred is unclear.
12. Overall summary
or: Theres never two people speaks the same
(16 He, Lyonshall, book VI)
The analysis of gender assignment in a number of different varieties of
(present-day) English proved to be enlightening beyond the expected
aspects of the investigated features. The results from these analyses can be
divided into different categories, depending on the type of variety, the
region, and/or the genre being analyzed or combinations of these factors.
The results we obtained from our analyses of gender assignment
primarily concern the following varieties of English:
traditional varieties of Southwest England
traditional varieties of West Country Newfoundland (=those
parts of Newfoundland that were primarily settled by people of
West Country origin and that were only marginally influenced by
other varieties of English or other languages)
spoken standard English (= modern varieties of English in
countries with English as a first language and only minimal input
from other languages; primarily Great Britain, USA, Canada,
Australia, New Zealand)
As this study only dealt with varieties of spoken English, not much will be
said about Standard English. As far as the practise of assigning gender in
StE is concerned, the myths still presented in grammars should be
relativized from what we know about actual speech behaviour. What
follows is a brief list of dos and donts in gender assignment, where the
donts are generally not spelt out in grammars, but rather follow by
deduction from the dos.
Based on the evidence presented in the preceding sections, it will be
obvious to the reader that speakers either obey these rules only to a certain
extent or not at all. Rather than the clear-cut division of tasks that StE
postulates for the gendered personal pronoun forms, different varieties use
different cut-off points along a continuum that ranges from (concrete) count
nouns via mass nouns to abstract ideas and situations.
Dos (according to grammars) Donts (according to grammars
Use he and she when referring to
human beings; it may only be used
when the sex of a very young child is
unclear
Use it when referring to animals; he
andshemay be used to refer to pets or
animals the speaker has a close
relationship with
Dont use heandsheto refer to animals
in general
Refer to all things with it Dont use he or she to refer to an
inanimate object
Useit to refer to abstract denotata Dont use he or she for purposes of
emotional emphasis
348 Susanne Wagner
Figure 8 is an attempt at illustrating these cut-off points for the different
varieties of English that were analyzed in this thesis. Letters only stand for
the principal availability of the gender for that category in the variety in
question, not for overall frequencies of occurrence.
humans animals
count nouns
(concrete)
mass nouns abstract nouns
StE mfmfmfmfNfmNNNfmNNNNfNNNNNNfNNNNNNNfNNNNN
spoken standard mfmfmfmfmmNfmmNfmNNmNNfNNNfNNNmNNNfNNfNNN
trad. West Country mfmfmfmmfmmmmmmmmmmmNNNNNNNNNNNNNNNNN
mod. West Country mfmfmfmfmNfmfmNmNmfmNmNNNmNNfNmNNfNNmNNN
trad. West C. NF mfmfmfmmfmmmfmNmmNmNmfmNmNNmNNNmNNfNNmN
modern West C. NF mfmfmfmmfmfmNmNmfmmNmfmNmNNmNNNmNNmNfNN
m=masculine forms
f =feminine forms
N =neuter forms
Figure 8. Gender assignment in varieties of English (based on data from this
study)
Certain difficulties emerge when it comes to categorizing the referent types.
The major problem is posed by the interaction of the continua count mass
and concrete abstract. Count nouns can be both concrete and abstract,
resulting in a three-dimensional rather than two-dimensional model.
However, for reasons of practicality and based on the observations made in
the corpora, it was decided to reduce these categories to those which
actually play a role in gender assignment. Thus, when count nouns are
mentioned below, only concrete count nouns are meant, while abstract
nouns are generally abstract count nouns.
Although necessarily very simplified, the most important contrasts
between the varieties are nicely illustrated in Figure 8. Let me comment on
the details and background assumptions implied:
(Written) Standard English (as presented in grammars)
humans masculine or feminine (according to sex)
animals supposedly neuter, but also masculine and feminine (not
always according to sex)
count nouns boats, countries, etc. can also be feminine ( metaphorical
gender)
mass nouns neuter; if anything else, then feminine abstract nouns
neuter; regionally also feminine (AusE)
Spoken Standard (based on studies presented here)
animals generally not neuter, but masculine and feminine (not
always according to sex)
count nouns can be masculine and feminine, although feminine forms
seem to be more frequent (cf. studies by Svartengren,
Mathiot/Roberts and also Pawley)
mass nouns can be feminine, masculine and neuter, but are
predominantly neuter
abstract nouns although generally neuter, often feminine in special
constructions (non-referential sheetc.)
Traditional West Country dialects (19th century)
animals generally masculine (often even when female cows in
SED)
count nouns masculine
mass nouns neuter
abstract nouns neuter
Modern West Country dialects (oral history)
generally a mixture of the traditional West Country and the Spoken Standard
system
animals generally masculine, but also feminine and occasionally
neuter
count nouns still predominantly masculine, but also neuter and
occasionally feminine
mass nouns diffusion of traditional system into mass nouns can be,
though rarely, masculine (and feminine Spoken
Standard) as well as neuter
abstract nouns see mass nouns, although to an even lesser extent
350 Susanne Wagner
Traditional West Country Newfoundland dialect (Folktales)
in general, the system is almost identical to that of the traditional West Country;
all cut-off points diffuse to the right
neuter
count nouns can also be feminine (rare) in addition to masculine and
neuter
mass nouns can also be masculine (in addition to neuter)
abstract nouns can also be masculine (probably also by extension from
concrete (count) nouns)
Modern West Country Newfoundland dialect (oral history)
generally close to traditional Newfoundland dialect, with neuter forms diffusing
left
neuter
count nouns masculine and neuter, but also feminine
mass nouns
abstract nouns
generally neuter, but also masculine and occasionally
feminine abstract nouns
In light of the findings presented here and in the preceding sections, it
seems appropriate to return to the term gender diffusion, which was used
as a label for what was later called gendered pronouns, non-standard
pronominal forms, etc.
The comparisons of the modern data with the more traditional material
revealed certain tendencies both in Southwest England and Newfoundland:
The traditional system is slowly dying out, witnessed by fewer and fewer
masculine forms in those domains which at one time were their exclusive
territory. The formerly obligatory system has developed into an optional
system. But on the other hand it seems as if gendered pronouns spread to
domains they were formerly not used in (abstract referents, particularly
in the Folktales corpus). Although such uses will probably not be
continued, the current stage seems to be a mixture of one, two or all three
of the following systems:
1. traditional West Country system
masculine forms for all count nouns;
mass and abstract nouns =neuter;
females (humans, rarely also animals) =feminine
2. Spontaneous Spoken (Standard) system (and elsewhere?)
feminine forms for everything that deserves closer attention, no
matter what its (semantic/biological) status;
animals generally masculine; in addition,
non-referential sheto refer to situations;
masculine forms in reference to inanimate objects are rare to
nonexistent
3. (Written) Standard English gendered references to nouns also
acceptable in StE (e.g. feminine forms for boats, guns, cars) are
more easily/readily accepted in spoken language than those non-
standard pronouns that are out according to prescriptivists
views.
System (1) is the most endangered one, as it is under pressure from both
systems (2) and (3). The co-existence of both feminine and masculine
forms denoting the same (inanimate) referent(s) leads to mixed diffused
systems. The Spontaneous Spoken Standard and the StE system slowly
invade the lects of those speakers who only would have known their
regional dialect some decades ago, but who now cannot help but be
influenced by the supraregional standard(s).
Those features found in the non-regional (weaker) but not in the regional
(stronger) system may also lead to the extension of the stronger forms to
cover all areas of use (e.g. employing masculine forms to refer to abstract
referents). Alternatively, speakers may simply employ forms from different
systems to cover the whole semantic spectrum (e.g. masculine forms for
count nouns, feminine forms for non-referential situation-type uses).
Not very surprisingly, the Spoken Standard system in Southwest England
seems much closer to StE than the English spoken in Newfoundland. As a
result, West Country speakers are losing their traditional system of gender
assignment more rapidly than their Newfoundland counterparts. The West
Country system is being substituted by the StE one (this can be deduced from
the differences between the SED material and the modern West Country oral
history data), without the intermediate Spoken Standard step that is
352 Susanne Wagner
obviously employed in Canada, which leads to an overall slower demise of
the traditional system there (and a higher frequency of gendered pronouns).
The one feature that has accompanied us throughout this study is that
native speakers of English are obviously not very happy about the
ubiquitous status that it has assumed in their mother tongue. Although we
can only make educated guesses about the exact reasons behind the
individual phenomena, all speakers choose to employ personal pronouns
other than neuter, non-distinct, semantically empty it if they want to add
feeling to an utterance, no matter if the emotion to be conveyed is
positive or negative. The natural gender system of StE thus seems to be on
its way to becoming a pragmatic gender system in the Spoken Standard,
where forms marked for gender are used according to the different
requirements, emotionality and general circumstances of the situation.
In certain domains of use, the gender chosen by speakers of modern
Spoken Standard varieties depends on extralinguistic factors such as the sex
of the speaker (particularly relevant for cars, etc.), professional background
(gendered pronouns = non-professional), emotional attitude (generally,
neuter forms =disinterest, negative attitude), etc. The traditional systems,
on the other hand, are based on intralinguistic gender assignment rules
(semantic domains of nouns). As both systems can be employed at the same
time, a diffusion is not only likely, but expected. As a result of the
preference of traditional dialects for masculine and that of modern dialects
for feminine forms, the mixture of systems looks like free variation to an
outsider. The factors that determine pronoun choice in modern dialects of
Southwest England and Newfoundland, and probably in native varieties of
English worldwide, are manifold and complex.
Acknowledgments
None of the research presented in this article would have been possible
without the help of numerous individuals and organisations. Special thanks
are due to Treve Crago, Ann Heeley, Clare Lyall and Norman Rogers for
providing access to archives and museums, to Oliver Pickering and Clive
Upton for help with and access to the SED material, to Phil Hiscock,
Graham Shorrocks, Patricia Fulton and all the MUNFLA staff for their help
during my stays in Newfoundland, and of course to everyone in Freiburg
for providing the most essential ingredient of all a stimulating
environment where such work can be carried out.
Appendix
(Additional) corpus material
The British National Corpus (BNC)
The Freiburg English Dialect Corpus (FRED)
The Helsinki English Dialect Corpus (pre-final version)
Parts of (*) and complete transcripts of MUNFLA Tapes by the
following collectors (see References for details; partial passages
were noted down in situ at the archive)
Burton 1982
Churchill, Kean and
Stratton 1978
Du Pree 1969
Feener 1983
Fudge 1969
Hynes 1977
Ivany 1974*
Lake 1977
Lehr 1977
Mifflin 1972*
Powell 1976
Quinton 1972*
Ryan 1985
Ryan 1978
Simms 1975
Squires 1978
Welsh 1971*
Widdowson 1963*
Wiseman 1976
Wood 1971
Yother 1971
Notes
1. The reader interested in details of the Canadian material is referred to Wagner
(2003), which includes the full analyses.
2. Although Elworthy claims that all examples given in his books are authentic
utterances by native speakers (cf. 1886: viii), the modern reader has obviously
no possibility to verify this. Thus, modern narratives in dialect seem not that
far removed from Elworthys data, particularly since all of Elworthys
examples are out-of-context utterances, generally not more than one sentence,
with no detailed information on speaker background etc. Any statistical
analysis of such material cannot be taken seriously, as the comparability of
data sets is in no way guaranteed.
354 Susanne Wagner
3. Elworthys own view on the phenomenon seems much more biased in his
later work, where he states that it is unknown to us in W. Somerset as a
neuter pronoun (1886: xxi), a gross over-generalization and/or over-
simplification.
4. See also Ihalainen 1991: 114115. The link between more accessible and
more frequent is established in, e.g., Keenan (1987: 49): The frequency
with which people relativise in discourse conforms to the [AH], subjects
being the most frequent, then direct objects, etc.
5. [S]peakers of Cornish in the Modern Cornish period would learn not the
ancient Wessex dialect of east Cornwall, Devon and Somerset [...], but a
version of English taught them in schools and by the upper classes and better-
educated (note that it was the gentry who gave up Cornish and spoke English
first), an English deliberately acquired, as distinct from a regional dialect
passed on from generation to generation. (Wakelin 1975: 100)
6. Wakelins highly eclectic methods in data selection have been previously
criticized; cf. e.g. Klemola (1996: 28) on Wakelins treatment of periphrastic
do.
7. Note that German does not mark predicative agreement; thus der Mann ist
nett the man is nice
and die Frau ist nettthe woman is nice
.
8. It is debatable whether or not sheprovides better semantic agreement than it.
Thus, it would probably be more appropriate to speak of a modified
Agreement Hierarchy that can be based on numerous factors in addition to
the original likelihood of semantic agreement increases criterion, e.g. on
pragmatic or stylistic factors. Another problem is posed by the question of
what exactly connects heandshewith who(m), while it is linked with which.
Animation seems the most likely factor.
9. As this is the viewpoint adopted by most scholars, it will be assumed here,
too. However, as will become clear from the following paragraphs, this view
is at least questionable, if not outright wrong.
10. Asexual is used here and in the remainder of this study as a label for the
third sexual (i.e. biological) category, i.e. male, female, asexual = sex;
masculine, feminine, neuter =gender.
11. See, for example, Barons (1971) summary of Moores (1921) work on the
topic. Baron (1971: 119) states that Moore statistically demonstrates the
prevalence of natural gender in OE. It should also be kept in mind that there
were obviously huge regional differences in the gradual process of gender
loss, with the North leading the way (cf. Kastovsky 2000: 715; Baron 1971:
127; Dekeyser 1980: 99).
12. Original emphasis. Scholars like Mitchell (1985: 2931) argue that Moores
analysis is over-generalized, as his choice of texts is not representative of the
respective periods. Still, it should be recalled that in those cases where a
conflict between sex and gender is potentially possible, sex seems to overrule
gender more often than not.
13. As English used to have grammatical gender, its loss would have to be
classified as one of the rare cases of de-grammaticalization, as previously
highly grammatical items lost their grammatical status. On the issue of
unidirectionality in grammaticalization, cf. e.g. Hopper and Traugott 1993,
ch. 5.
14. This is debatable from a biological point of view, as plants and trees are also
animate and non-human.
15. Biber et al.s description contained a similar remark, but did not follow this
path as consequently as Payne and Huddleston. Note that specialist studies
such as that by Morris (1991) or Mathiot and Roberts (1979), which will
figure prominently in a later section, have long since argued in such a
manner.
16. This is the conclusion drawn from the fact that the umpteenth reprint (1962)
was published as trade-market paperback as part of Barnes and Nobles
College Outline Series.
17. The figure for no pronoun recorded includes irrelevant responses or
unanswered questions, i.e. the sum of the total of responses without a
pronoun plus the total of responses with a pronoun gives the grand total.
18. One major point of criticism concerns the fact that any use of the fieldworker
notebooks depends on the researchers interpretation of the usually
exclusively phonetic material. It should be pointed out in this context that
the whole SED (i.e. the published material as well) is ultimately based on
these notes, i.e. the interpretations of the individuals who were responsible
for the publication. Anyone who has worked with the notebooks in quite
some detail will agree that the conventions of transcription and orthographic
interpretation varied greatly between different fieldworkers, making it almost
impossible to claim that one standard was used for the whole SED.
19. The first line, if in phonetics, is transcribed letter for letter, as found in the
notebooks (as far as the typescript was available). Italic script is the authors
translation of what was found in phonetic script in the notebooks. The
regular script version is the authors standard translation.# stands for
word stress as indicated by the fieldworker. Some fieldworkers provided
orthographic translations while others did not.
20. For unknown reasons, Montacute was excluded from the Basic Material. The
dialect may have seemed too conservative even for SED purposes, retaining
such ancient features as utch for I.
21. Excluding the Isle of Wight, where the fieldwork was done by Michael Barry,
whose note-taking style proofed to be highly unsuitable for this investigation
most information taken from his notebooks would have been wishful
thinking rather than actual data, as his handwriting is close to illegible.
356 Susanne Wagner
22. For lack of space, a detailed description of the dialects under investigation
will not be included here. Information on these two regions and their dialects
can be found e.g. in Clarke (2004) on Newfoundland and Wagner (2004c) on
the Southwest.
23. In some of the interviews, a spouse or child was present during the recording
and gave the occasional comment. However, there is only one main speaker
in most of the interviews. The same is true for the material from the other
counties.
24. The discrepancies between the totals in the age and recording date column are
due to the fact that this information was not always available.
25. As transcription is a time-consuming task, and the amount of new material
coming in is much greater than the production of transcripts, MUNFLA has
to deal with an approximate 10-year backlog. Thus, the most recent
interviews used here stem from the late 1980s.
26. Cf. also Klemola (1996: 39), who argues in a similar vein: The formal nature
of the questionnaire interview situation probably did not encourage the use of
[non-standard] features [...], whereas the incidental material contains
utterances that the fieldworkers picked up from their more informal
conversations with the informants.
27. It is absolutely unclear how the Australian and New Zealand uses shes right
etc. can be classified as personification in the quoted dictionary entry the
personal pronoun is usually non-referential, excluding personification as an
explanatory factor.
28. Morris (1991) came to similar results; see section 6.2.1. It is unclear why the
results for the two birds (canary and parakeet) differ to such a large extent.
We can assume that a parakeet is more pet-like than a canary, which would
explain the comparatively high output of it for the latter. Why she and he
occur with almost identical frequencies in example (2d) remains unclear,
though.
29. A search for dog and cat was run, restricted to the spoken sub-corpus.
The total figures are incidental, based on the printed output of eight pages
after marking those cases where pronouns occurred in the maximum scope
setting.
30. The transcriber probably misunderstood; force is more likely in this
context, as the police units with dogs are usually part of a special force.
31. The first pronoun (she) most likely refers to the weather in general, while the
second one (it) refers to the wind specifically, which the authors identified as
referent in both cases.
32. The primary distinction made by Mathiot and Roberts is between human and
non-human, not animate vs. inanimate as assumed in the earlier sections.
However, this difference is of no consequence here. As explained in section
5.2, animals should be treated separately anyway.
33. This concept will underlie all future uses of the term involvement in this
thesis. It should be treated separately from the more pragmatic meaning (
ensuring collaboration between speakers; thus e.g. in Cheshire 1997). While
the latter involves inferencing, negotiating of meaning, etc., the former is
purely referring to the individual speakers sentiments (e.g. sympathy or
antipathy) without involving the addressee.
34. It is unclear how Siemund (2001) can detect a system in the pattern. It is not
surprising that he concludes (Siemund 2001: 104) that the mens pattern is
similar to the one found for Tasmanian English the Tasmanian informants
are almost exclusively males (cf. Pawley 2004). However, this nicely
supports the present authors view that we are essentially dealing with a
pattern of spoken English in general rather than a (regionally restricted)
variety-specific pattern.
35. [E]motional interest that is mirrored by the feminine gender (Svartengren
1927: 110); familiarity and the feeling of companionship between an artisan
and his tools (Svartengren 1927: 110), etc. What Svartengren means seems
to be some form of personal involvement as used by Mathiot and Roberts
rather than personification in its strict sense.
36. Morris database for this category is very small. Of the approximately 1,500
examples which constitute her overall database, only 80 instances of
masculine pronominalization of biological inanimates could be found. These
include 15 instances of personification and about 30 examples from other
authors studies (cf. Morris 1991: 166).
37. As mentioned above, the interviewers influence is minimal in all of the
chosen texts; accordingly, the scattergrams only include the informants
switches. Included here are C 626, C 627, C 628, C 631, C 751, C 1187/8,
and C 2914 from the MUNFLA material, and Som_001, Som_009, Som_014,
Som_021, Som_022 and Som_028 from the Somerset material. The
additional figures not included in the tables above on which the scattergrams
are based are:
text it/total switches he-it he/total switches it-he
C 626 0.53 0.34 0.47 0.36
C 627 0.57 0.39 0.43 0.27
C 628 0.42 0.29 0.58 0.375
C 631 0.54 0.32 0.46 0.27
C 751 0.38 0.33 0.62 0.45
C 1187/8 0.07 0.06 0.93 0.63
C 2914 0.31 0.14 0.69 0.3
Som_001 0.71 0.27 0.29 0.14
Som_009 0.58 0.36 0.42 0.29
Som_014 0.43 0.31 0.57 0.41
Som_021 0.51 0.30 0.49 0.30
Som_022 0.42 0.18 0.58 0.25
Som_028 0.14 0.10 0.86 0.63
358 Susanne Wagner
38. Individual rations: Monmouthshire 0-25% (1 of 4), Worcestershire 20%-40%
(1 of 5; 2 of 4), Herefordshire 17%-57% (1 of 6; 4 of 7), Gloucestershire
50%-83% (3 of 6; 6 of 7).
39. Individual county ratios: 50% for Monmouthshire and Worcestershire (1 of 2;
2 of 4), 60% in Herefordshire (3 of 5), 100% in Gloucestershire (4 of 4).
40. Individual county ratios: 20% in Berkshire (1 of 5), 25% in Somerset (3 of
12), 75% in Wiltshire (6 of 8), 80% in Dorset (4 of 5), 85.7% in Hampshire
(6 of 7), 100% in Cornwall and Devon (7 of 7; 11 of 11).
41. Individual county ratios: Berkshire 25% (1 of 4), Sussex 50%(!) (1 of 2),
Somerset and Hampshire 83.3% (10 of 12; 5 of 6), Wiltshire 87.5% (7 of 8),
Cornwall, Devon and Dorset 100% (7 of 7, 11 of 11, 5 of 5).
42. Individual county ratios: 27.3% Somerset (3 of 11), 28.6% Wiltshire (2 of 7),
77.8% Devon (7 of 9), 85.7% Cornwall (6 of 7), 100% Dorset and Hampshire
(5 of 5 each).
43. Individual ratios (first figures for question IX.2.6, second for IX.2.8): total
incl. Southeast 46.9% (15 of 32), excl. Southeast 55.6% (15 of 27); Devon
37.5% (3 of 8), Wiltshire 40% (2 of 5), Hampshire 50% (3 of 6), Somerset
83.3% (5 of 6), Cornwall and Dorset 100% (1 of 1 each); total incl. Southeast
63.2% (36 of 57), excl. Southeast 81.4% (35 of 43); Sussex 20% (1 of 5),
Berkshire 25% (1 of 4), Hampshire 50% (2 of 4), Wiltshire 71.4% (5 of 7),
Cornwall 83.3% (5 of 6), Somerset, Devon, Dorset 100% (8 of 8, 10 of 10, 4
of 4).
44. Individual county ratios: Berkshire 50% (1 of 2), Dorset and Hampshire
66.7% (2 of 3, 4 of 6), Somerset 70% (7 of 10),Wiltshire 75% (3 of 4),
Cornwall and Devon 100% (6 of 6, 8 of 8).
45. Individual ratios: Somerset 46.2% (6 of 13), Dorset 50% (3 of 6), Wiltshire
60% (6 of 10), Hampshire 71.4% (5 of 7), Devon 81.8% (9 of 11), Cornwall
100% (7 of 7).
46. Although great care has been taken in reproducing the examples from the
hand-written notes in the fieldworker notebooks, misinterpretations and
mistakes cannot be ruled out completely. However, it is hoped that those only
affect the parts that are irrelevant to the present analysis. Where certain
phonetic symbols where either not available to me or could not be interpreted
in positions relevant for the analysis, this is generally mentioned. As the
notebooks where consulted three times in as many years, certain
inconsistencies and developments on my part cannot be excluded.
Interesting (dialectal) lexical items will not be commented on in the
examples.
47. I am aware of the fact that in other cultures this distinction may be non-
existent or that the cut-off points may differ. Plant is taken in its popular
rather than scientific sense (which would include vegetables as a sub-
category).
48. Recall Elworthys criterion for gendered pronouns: the noun in question has
to be a definite noun, i.e. the name of a thing or object which has a shape of
its own, which would exclude abstracta (Elworthy 1877: 32).
49. Twas an American steamer in camouflage. Hewent by her heturned, come
back by her again, didnt know twas a trick on. The steamer (he) is
passing by a small boat with survivors from a sunken ship (she).
50. This theory is supported by data from other regions.
51. This is an example from the pre-final version of the Helsinki Dialect Corpus.
I would like to thank Kirsti Peitsara for granting us access to the material in
situ at Helsinki University.
52. Among the features that occur in the interview are the Cornish habitual
belong (see Hancock 1994), periphrastic do, purposive for to, pronoun
exchange, non-standard use of prepositions (on for of, directional
prepositions), non-standard use of relative pronouns, agreement features,
regularized verb paradigms, present tense used in the extended-now sense
usually (in StE) expressed by the present perfect (hes dead years).
53. Detailed distribution: Two speakers contribute one example each, four
speakers two, three three, four four, two five, two six, one seven, one eight,
one nine, three ten, one 12, one 14, two 15, two 25, one 41.
54. Text AH: 15.9% of examples, 6.3% of words; significant at <0.001%-level
(chi-square); SRLM 44: 5.4% of examples, 1.6% of words; significant at
<0.001%-level; SRLM 224: 9.7% of examples, 4.1% of words; significant at
0.002%; SRLM 108: 9.7% of examples, 4.7% of words; significant at
<0.05%-level.
55. 615 forms in ca. 320,000 words by 68 speakers in the Southwest England
corpus, 885 forms in ca. 132,000 words by 34 speakers in the MUNFLA
corpus. Even when excluding the 300 or so boat references, the MUNFLA
material contains proportionally more gendered pronouns than the Southwest
corpus. Table 48 does not include the 20 or so examples that stem from
interviews which do not belong to the core corpus as described in Section 4.
56. Though house seems to be the most likely referent when analyzing this
example superficially, it may be recalled that this is the speaker who uses
more than 100 masculine forms referring to houses in the remainder of the
interview. It is thus extremely unlikely that he should here use a lone
feminine.
57. Due to the nature of the data chosen for the analysis, questions of a
sociological nature such as influences of age, sex, education, urban vs. rural,
etc. could not be addressed at all. Moreover, sophisticated statistical analyses
are often not possible due to small sample size. To remedy this situation,
future studies would have to be based on material collected to fit these
requirements. As basically no other studies on the phenomenon were
available at the time of writing, the present author chose to tackle the basic
360 Susanne Wagner
issues first. A follow-up study could investigate those areas that turned out to
be a promising field for future work.
References
Anderwald, Lieselotte and Susanne Wagner
forthc. The Freiburg English Dialect Corpus (FRED) - Applying corpus-
linguistic research tools to the analysis of dialect data. In Using
Unconventional Digital Language Corpora. Vol. I: Synchronic Cor-
pora, J oan Beal, Karen Corrigan and Herman Moisl (eds.). Basing-
stoke: Palgrave MacMillan.
Attwell, J ames
1987 Dorset Dialect Days. Sherborne: Dorset Publishing Company.
Barnes, William
1886 A Glossary of the Dorset Dialect with a Grammar. Guernsey/St.
Peter Port: Toucan Press.
1844 A dissertation on the Dorset dialect of the English language. In
WilliamBarnes - Selected Poems, Andrew Motion (ed.), 117138.
Hardmondsworth: Penguin.
Baron, Naomi S.
1971 A reanalysis of English grammatical gender. Lingua 27: 113140.
Biber, Douglas, Stig J ohansson, Geoffrey Leech, Susan Conrad, and Edward
Finegan
1999 Longman Grammar of Spoken and Written English. London:
Longman.
Birner, Betty J . and Gregory Ward
1998 Information Status and Noncanonical Word Order in English.
Amsterdam/Philadelphia: Benjamins.
Bremann, Rolf
1984 Soziolinguistische Untersuchung zumEnglisch von Cornwall. Frank-
furt: Lang.
Brinton, Laurel J .
2000 The Structure of Modern English. Amsterdam/Philadelphia: Ben-
jamins.
Burchfield, R.W. and H.W. Fowler (eds.)
1998 The New Fowlers Modern English Usage. Oxford: Oxford
University Press.
Burton, Terry
1982 MUNFLA ms 82-319/C10260, 89pp.
Chambers, J ack K. and Peter Trudgill
1998 Dialectology. Cambridge: Cambridge University Press.
1970
1994
Cheshire, J enny
1997 Involvement in standard and nonstandard English. In Taming the
Vernacular. From dialect to written standard language, J enny
Cheshire and Dieter Stein (eds.), 6882. London/New York:
Longman.
Churchill, Shirley, Betty Kean, and Nita Stratton
1978 MUNFLA transcripts of Tapes 79-037/C3831-32.
Clarke, Sandra
2004 Newfoundland English: morphology and syntax. In A Handbook of
Varieties of English. Vol. II: Morphology and syntax, Bernd
Kortmann and Edgar W. Schneider (eds.), 303318. Berlin/New
York: Mouton de Gruyter.
Comrie, Bernard
1989 Language Typology an Linguistic Universals. Chicago: University
of Chicago Press.
Corbett, Greville
1991 Gender. Cambridge: Cambridge University Press.
Crystal, David
1995 The Cambridge Encyclopedia of the English Language. Cambrigde:
Curme, George O.
1962 English Grammar. New York: Barnes & Noble. [rpt.; 1925; 1947].
Curzan, Anne
2000 Gender categories in early English grammars: Their message to the
modern grammarian. In Gender in Grammar and Cognition. I
Approaches to gender. II Manifestations of gender, Barbara
Unterbeck (part I), Terttu Nevalainen and Mirja Saari (eds.), 561
Dartnell, George Edward and Rev. Edward Hungerford Goddard
1893 A Glossary of Words Used in the County of Wiltshire. London:
Oxford University Press. [Publications of the English Dialect
Society 69; Vaduz: Kraus Reprint Ltd., 1965].
Dekeyser, Xavier
1980 The diachrony of gender systems in English and Dutch. In Historical
Morphology, J acek Fisiak (ed.), 97111. The Hague: Mouton.
Downes, J ohn
1986 A Dictionary of Devon Dialect. Padstow: Tabb House.
Du Pree, Victor
1969 MUNFLA Tapes 70-003/C626-33.
Elliott, Ralph
1984 Thomas Hardys English. Oxford: Blackwell.
362 Susanne Wagner
Ellis, Alexander J .
1889 On Early English Pronunciation. Vol. V. Existing dialectal as
compared with West Saxon pronunciation. London: Trbner.
Elworthy, Frederic Thomas
1877 An Outline of the Grammar of the Dialect of West Somerset.
London: Trbner & Co. [Publications of the English Dialect Society
19; Vaduz: Kraus Reprint Ltd., 1965].
1886 The West Somerset Word-book. London: Trbner & Co.
[Publications of the English Dialect Society 50; Vaduz: Kraus
Reprint Ltd., 1965].
Feener, David Cyril
1983 MUNFLA ms 83-289/11pp.
Fischer, Andreas
1976 Dialects in the South-West of England: A Lexical Investigation.
Bern: Francke.
Francis, W. Nelson
1983 Dialectology. An Introduction. London/New York: Longman.
Fudge, J esse
1969 MUNFLA Tape 70-037/C751.
Halpert, Herbert and J ohn D.A. Widdowson
1996 Folktales of Newfoundland The Resilience of the Oral Tradition.
St. J ohns: Breakwater.
Hancock, Ian F.
1994 Componentiality and the creole matrix: The Southwest English
contribution. In The Crucible of Carolina. Essays in the Develop-
ment of Gullah Language and Culture, Michael Montgomery (ed.),
95114. Athens/London: University of Georgia Press.
Hirooka, Hideo
1980 Thomas Hardys use of dialect. no place: Shinozaki Shorin Press
(J apan).
Hopper, Paul J . and Elizabeth Closs Traugott
1993 Grammaticalization. Cambridge: Cambridge University Press.
Howe, Stephen
1996 The Personal Pronouns in the Germanic Languages. Berlin/New
York: Mouton de Gruyter.
Huddleston, Rodney and Geoffrey K. Pullum
2002 The Cambridge Grammar of the English Language. Cambridge:
Hynes, Derek
1977 MUNFLA transcript of Tape 77-187/C2654 and ms, 55pp.
1965a
1965b
Ihalainen, Ossi
1985a He took the bootle and put n in his pocket: The object pronoun it in
present-day Somerset. In Focus on: England and Wales, Wolfgang
Viereck (ed.), 153161. Amsterdam: Benjamins.
1985b Synchronic variation and linguistic change: evidence from British
English dialects. In Papers fromthe 4th International Conference on
English Historical Linguistics, Amsterdam, 10-13 April 1985, Roger
Eaton , Olga Fischer, Willem Koopman, and Frederike van der Leek
(eds.), 6172. Amsterdam: Benjamins.
1991 On grammatical diffusion in Somerset folk speech. In Dialects of
English. Studies in grammatical variation, Peter Trudgill and J ack
K. Chambers (eds.), 104119. London/New York: Longman.
Ivany, R.M.
J ennings, J ames K.
1869 The Dialect of the West of England, particularly Somersetshire.
London: J ohn Russell Smith.
J ones, Charles
1988 Grammatical Gender in English: 9501250. London: Croom Helm.
J ones, Malcolm and Patrick Dillon
1987 Dialect in Wiltshire. Trowbridge: Wiltshire County Council.
Kastovsky, Dieter
2000 Inflectional classes, morphological restructuring, and the dissolution
of Old English grammatical gender. In Gender in Grammar and
Cognition. I Approaches to gender. II Manifestations of gender,
Barbara Unterbeck, (part I), Terttu Nevalainen and Mirja Saari
(eds.), 709727. Berlin/New York: de Gruyter.
Keenan, Edward L.
1987 Variation in Universal Grammar. In Universal Grammar: 15 Essays,
Edward L. Keenan, 4659. London: Croom Helm.
Keenan, Edward L. and Bernard Comrie
1977 Noun phrase accessibility and Universal Grammar. Linguistic
Inquiry 8: 6399.
Klemola, J uhani K.
1996 Non-standard periphrastic do: a study in variation and change. PhD
thesis, Department of Language and Linguistics, University of
Essex.
Kruisinga, Etsko
1905 A Grammar of the Dialect of West Somersetshire: Descriptive and
Historical. Bonn. [Bonner Beitrge zur Anglistik, Heft 18]
364 Susanne Wagner
Labov, William
1972 Sociolinguistic Patterns. Philadelphia: University of Pennsylvania
Press.
Lake, Sharon
1977 MUNFLA transcript of Tape 78-024/C3162.
Lehr, Genevieve
Leisi, Ernst and Christian Mair
1999
8
Das heutige Englisch: Wesenszge und Probleme. Heidelberg:
Winter.
Linn, Michael D.
1993 Appendix: resources for research. In American Dialect Research,
Dennis R. Preston (ed.), 433448. Amsterdam: Benjamins.
MacKay, Donald G. and Toshi Konishi
1980 Personification and the pronoun problem. Womens Studies
International Quarterly 3: 149163.
Marcoux, Dell R.
1973 Deviation in English gender. American Speech 48: 98107.
Marten, Clement
1973 The Devonshire Dialect. Exeter: Clement Marten.
Mathiot, Madeleine and Marjorie Roberts
1979 Sex roles as revealed through referential gender in American
English. In Ethnolinguistics: Boas, Sapir and Whorf Revisited,
Madeleine Mathiot (ed.), 147. The Hague: Mouton.
Mifflin, Robert
Mitchell, Bruce
1985 Old English Syntax. Oxford: Clarendon.
Moore, Samuel
1921 Grammatical and natural gender in Middle English. Publications of
the Modern Language Association of America (PMLA) 36: 79103.
Morris, Lori
1991 Gender in modern English: The system and its uses. PhD thesis,
Universit Laval, Quebec.
Orsman, H.W. (ed.)
1997 The Dictionary of New Zealand English. Auckland: Oxford
University Press.
Orton, Harold
1962 Survey of English Dialects - Introduction. Leeds: Edward Arnold.
Pawley, Andrew
2002 Using he and she for inanimate referents in English: questions of
grammar and world view. In Ethnosyntax. Explorations in Grammar
and Culture, Nick J . Enfield (ed.), 110137. Oxford: Oxford
University Press.
2004 Australian Vernacular English: some grammatical characteristics. In
A Handbook of Varieties of English. Vol. II: Morphology and
syntax, Bernd Kortmann and Edgar W. Schneider (eds.), 611642.
Perry, F.C.
1921 The dialect of North Somerset. Transactions of the Yorkshire Dialect
Society 4 (22): 1741.
Poplack, Shana
1979 Function and process in a variable phonology. PhD thesis,
University of Pennsylvania, Philadelphia.
1980 The notion of the plural in Puerto Rican Spanish: Competing
constraints on (s) deletion. In Locating Language in Time and
Space, William Labov (ed.), 5567. New York: Academic Press.
Poplack, Shana and Sali Tagliamonte
1993 The zero-marked verb: Testing the creole hypothesis. J ournal of
Pidgin and Creole Languages 8: 171206.
1996 Nothing in context: Variation, grammaticization and past time
marking in Nigerian Pidgin English. In Changing meanings,
changing functions. Papers relating to grammaticalization in
contact languages, Philip Baker and Anand Syea (eds.), 7194.
London: University of Westminster Press.
Potter, Mary C. and Linda Lombardi
1998 Syntactic priming in immediate recall of sentences. J ournal of
Memory and Language 38: 265282.
Powell, Clifford
1976 MUNFLA Tape 76-290/C2894.
Quinton, Leith D.
Quirk, Randolph, Sidney Greenbaum, Geoffrey Leech, and J an Svartvik
1985 A Comprehensive Grammar of the English Language. London:
Longman.
Ramson, W.S. (ed.)
1988 The Australian National Dictionary. Melbourne: Oxford University
Press.
Ross, Alan R. C.
1936 Sex and gender in the Lindisfarne Gospels. J ournal of English and
Germanic Philology 35: 321330.
Ryan, Diane
1985 MUNFLA ms including transcript of Tape 85-154/C12247.
366 Susanne Wagner
Ryan, Patricia F.
1978 MUNFLA transcript of Tape 79-341/C4719 and manuscript, 71pp.
Sankoff, David and Suzanne Laberge
1978 Statistical dependence among successive occurrences of a variable in
discourse. In Linguistic Variation. Models and Methods, David
Sankoff (ed.), 119126. New York: Academic Press.
Scherre, Maria Marta Pereira and Anthony J . Naro
1991 Marking in discourse: Birds of a feather. Language Variation and
Change 3: 2332.
Siemund, Peter
2001 Pronominal gender in English. A study of English varieties from a
cross-linguistic perspective. Postdoctoral thesis, Fachbereich
Philosophie und Geisteswissenschaften, Institut fr Englische
Philologie, Freie Universitt, Berlin.
2002 Animate pronouns for inanimate objects: Pronominal gender in
English regional varieties. In Anglistentag 2001 Wien - Proceedings,
Dieter Kastovsky, Gunther Kaltenbck, and Susanne Reichl (eds.),
1934. Trier: Wissenschaftlicher Verlag.
forthc. Pronominal Gender in English: A Study of English Varieties froma
Cross-Linguistic Perspective. London: Routledge.
Simms, Garfield
1975 MUNFLA Tapes 75-164/C2829-31.
Squires, Shirley
1978 MUNFLA transcript of Tape 78-427/C4220 and manuscript, 43pp.
Svartengren, T. Hilding
1927 The feminine gender for inanimate things in Anglo-American.
American Speech 3: 83113.
1928 The use of the personal gender for inanimate things. Dialect Notes 6:
756.
1954 The use of feminine gender for inanimate things in American
colloquial English. Moderna Sprak 48: 261292.
Szmrecsanyi, Benedikt
2004 Persistence phenomena in the grammar of spoken English. PhD
thesis, Englisches Seminar, Albert-Ludwigs-Universitt, Freiburg.
Tree, J ean E. Fox and Paul J .A. Meijer
1999 Building syntactic structure in speaking. J ournal of Psycholinguistic
Research 28: 7192.
Trudgill, Peter
1999
2
The Dialects of England. Oxford: Blackwell.
Viereck, Wolfgang, Karin Viereck and Heinrich Ramisch
2002 dtv-Atlas Englische Sprache. Mnchen: dtv.
Visser, Frederikus Th.
1963 An Historical Syntax of the English Language. Part I: Syntactical
units with one verb. Leiden: Brill.
Wagner, Susanne
2001 Pronoun exchange a feature of English dialects? Unpublished
manuscript.
2003 Gender in English pronouns. Myth and reality. PhD thesis,
Englisches Seminar, Albert-Ludwigs-Universitt, Freiburg.
http://www.freidok.uni-freiburg.de/volltexte/1412
2004a English in Cornwall a re-evaluation. Paper presented at the 15th
Sociolinguistics Symposium, Newcastle upon Tyne, April 1-4 2004.
2004b Gendered pronouns in English dialects A typological
perspective. In Dialectology Meets Typology: Dialect Grammar
froma Cross-Linguistic Perspective, Bernd Kortmann (ed.), 479
2004c English dialects in the Southwest: morphology and syntax. In A
Handbook of Varieties of English. Vol. II: Morphology and syntax,
Bernd Kortmann and Edgar W. Schneider (eds.), 154174.
Wakelin, Martyn F.
1975 Language and History in Cornwall. Leicester: Leicester University
Press.
1981
2
English Dialects: An introduction. London: Athlone Press.
1986 The Southwest of England. Amsterdam: Benjamins.
Wales, Katie
1996 Personal Pronouns in Present-Day English. Cambridge: Cambridge
University Press.
Welsh, Kinsley
Widdowson, J .D.A.
1963 MUNFLA transcripts of Tapes 63-002/C0005, C0010-11.
Wiseman, Sherman
1976 MUNFLA Tapes 76-295/C2913-14.
Wood, Brian
1971 MUNFLA Tapes 72-089/C1187-88.
Yother, Larry
1971 MUNFLA Tapes 73-046/C1958-59, C1961.
I ndex
Accessibility Hierarchy 2, 21, 48
70
adverbial relative clause 50, 112
affix loss 173175
agreement
verbal seeverbal concord
analogy 177178
analytical genitive 59, 111
and-coordination 61, 79
animals (as referents) 253259
in CanE 270271
apparent-time variation 158
BE see also verbal concord
finite be forms 136, 140, 142
generalized am 140
generalized is 141
Belfast 181186
British National Corpus (BNC) 14,
23, 112, 256, 353
Cheshire 138
Chester-Wash line 138
clauses
existential 129, 138, 156161
interrogative 129, 161, 165168
locative inversion 129
relative 130, 161, 164
tags 161, 166167
cleft 6270, 112, 130, 164
pseudo-cleft 6267, 70, 93
cliticization 177
Cognitive Grammar 192193
concord seeverbal concord
connector which 21, 8893
Construction Grammar 193
contact
Old English and Brythonic 173
Old English and Scandinavian 176
coordinate construction constraint
79
copies 70, 116
Cornwall 218, 220, 239241, 245
246, 251, 291, 312322, 327,
330338, 354, 358
Cumberland 135, 141, 150
Derbyshire 138, 150
Devon 211, 239, 241, 244245,
251, 315324, 327, 330, 333
338, 354, 358
Dorset 216218, 241, 246, 315
318, 337, 358
Durham (county) 135, 150
Early Modern English 40, 57, 178
elided preposition 45, 51, 88, 90, 93
English Dialect Grammar 29
entrenchment 192193
existential 40, 6371, 75, 76, 91
focussing 6264
Folktales of Newfoundland 242,
248251, 259, 342346, 350
fossilization 171
Freiburg English Dialect Corpus
(FRED) 822, 132133, 140
142, 154, 212, 319338, 353
(dis)advantages of orthographic
transcripts 1011
corpus design 1113
principles of compilation 46
role of oral history 67, 1315
frequency 192193
fronting 4547, 93 see also pied-
piping
gapping 2 see also zero subject
relative clause
Index 369
gender
history of, in English 225226
in PrDE 227235
in the world's languages 222225
gender assignment
in American English 261270
in Tasmanian Vernacular English
274
genitive avoidance 5962
grammaticalization
of existential there 156
habitual 131
have
forms in -n 140
Herefordshire 140
Hierarchy Constraints 4849
historic present 146
in Irish English 146
idioms 171
inherent variation 191
Irish English 146
Lancashire 131, 138, 147150
Late Modern English 40
left dislocation 6162
Lexical-Functional Grammar 186
189
lexically empty antecedent 62, 65
70, 112
Lincolnshire 144, 150, 178
Lowman Survey 2931, 35, 54
markedness 180181, 186189,
256, 275, 287, 342
microparametric variation 181186
Middle English 40, 42, 58, 65, 75
76, 91, 112, 225226
dialect boundaries 162
Midlands dialects 175
Midlands verbal systems 140
northern dialects 173179
southern dialects 176
modification 62, 67, 88
MUNFLA 242, 247251, 339340
narrative present see historic present
negation 23
and verbal concord 151, 156, 160
negative polarity items 184
Newfoundland 211214, 235, 242,
247250, 253, 264, 272277,
280284, 293, 338342, 345,
350352
Newfoundland English, varieties of
seeNewfoundland
nonreduction 21, 8586
Northern Ireland see Ulster
Northern Ireland Transcribed Corpus
of Speech (NITCS) 23, 111, 132,
144, 149, 158, 170
Northern Subject Rule 124199 see
also verbal concord
Celtic substrate hypothesis 173
conserved only with forms of BE
150, 169
definition 128
extension to BE 178
extension to forms of BE 149
history 173179
markedness relations 180181
Position-of-Subject Constraint
129, 131, 144146, 185190
southern boundary 178
southern limits 162164
Type-of-Subject Constraint 129
Northumberland 135, 142, 150151
Nottinghamshire 144
noun
count 211, 214222, 277280,
293296, 303, 305, 310313,
326327, 340, 346351
mass 211, 216220, 249, 277
280, 293, 302303, 312, 314,
347351
370 Index
Old English 65, 213, 224226, 234
Northumbrian dialect 173177
West Saxon dialect 177
Old Norse 174, 176
paratactic (construction) 3031, 40,
6061, 7375, 9293
partitive genitive 4344, 4748, 75
76, 90, 112
personal(ity)/nonpersonal(ity) 41
45
personification 232235, 252254,
262, 264, 269270, 306, 308,
310, 356357
pied-piping 4546, 76 see also
fronting
Position-of-Subject Constraint see
possessivehave or get construction
6061
Present-Day English 212213, 224
227, 230, 233235, 338
Principles-and-Parameters theory
146, 181186
pronoun retention 21, 52, 7179,
9596
pronouns
demonstrative 165
indefinite 165, 168
strong subject forms 183
referent types
abstract referents 269, 339340,
343345
body parts 309310, 330331
feminine 311312
man-made objects 294306, 320
328, 339, 342
masculine 293311
nature 306309, 328330
restrictive(ness)/nonrestrictive(ness)
3841, 7778, 86, 96, 111
resumptive (pronoun) 48, 60, 70
86, 88, 9096
Scotland 154
singular therere 156159
Scots 170 see also Ulster Scots
Older 150
SED seeSurvey of English Dialects
she
non-referential 251, 259261, 268,
271272, 340, 344, 349, 351
Shropshire 138, 140
Somerset 215218, 239245, 251,
280283, 291, 315322, 327
328, 330338, 354, 357358
Southwest England 211214, 275,
287, 293, 319, 339, 342, 346,
350352, 359 see also West
Country
Staffordshire 138
Standard English 23, 22, 28, 38,
41, 4647, 50, 55, 58, 6368, 76,
78, 81, 84, 9496
standard variety see Standard
English
stochastic grammars 191
structural markedness 72
Survey of English Dialects 355
Survey of English Dialects (SED) 4,
3134, 37, 54, 56, 59, 94, 111,
132135, 138, 144, 161, 213
214, 220221, 235243, 247,
250251, 319322, 325327,
331337, 345, 349, 351, 355
Basic Material 285292
Fieldworker notebooks 292319
Tape-Recorded Survey of Hiberno-
English Speech (TRS) 132133
Teesside 141142
thae (demonstrative) 165
there (existential marker) 129, 156
161
Index 371
they
as demonstrative seethae
loan fromScandinavian 177
thouyou distinction 135136, 142
144
topicalization structures 6268
Tyneside 151
Type-of-Subject Constraint see
Ulster 144, 149, 154, 170
communal variation 159
singular therere 156159
Ulster Scots 157, 159
Ulster Scots 170
Varbrul 132, 160, 170
verbal concord
formalist analyses 146, 179190
in Central North 135, 141142,
150, 154
in Lower North 137, 155, 158
in Upper North 136, 142
in East Midlands 136
in Middle English 162, 173179
in Northwest Midlands 136142,
150
in Old English 173177
in Scotland 138, 154, 170
in Southwest England 147, 179
in Ulster 138, 149, 154, 170, 181
186
in West Midlands 136, 142
-n forms 136, 138140, 151
past tense, in Old English 149
present-tensebe 136, 140142
waswere 136, 142, 149155, 158
with coordinated verbs 131, 144
146
with demonstrative pronouns 165
with existential clauses 129, 138,
156161
with habitual clauses 131, 147
148
with historic present 146
with indefinite pronouns 165, 168
with intervening adverbs 130, 146,
160, 182
with inversion 129, 161, 165168,
182
with locative inversion 129
with negation 151, 156, 160
with relative clauses 130, 164
withthou 142144
vernacular primitives 179
Welsh 173
West Country 211, 218, 221, 244
245, 250, 254, 264265, 272
273, 277, 291, 293, 305, 310
320, 323, 330, 332, 336342,
346351 see also Southwest
England
Westmorland 135, 150
Wiltshire 217, 239, 242244, 251,
299, 315322, 325, 327, 330,
335338, 358
Worcestershire 140
Word Grammar 189190
York 151, 170
Yorkshire 131, 138, 147150, 166
zero relatives seezero subject
relative clause
zero subject relative clause 21, 29,
3539, 48, 55, 6270, 77, 79, 84,
89, 9495, 111112
W
W
W
.
M
O
U
T
O
N
-
P
U
B
l
l
S
H
f
R
S
.
C
O
M
e )BOECPPL is by far the most thor-
ough reference work on phonology and
the hrst-ever comprehensive overview of
the morphology and syntax of varieties of
English in the world. It consists of a two
volume set accompanied by an interactive
CD-ROM.
e books feature descriptive survey arti-
cles that are authored by widely acclaimed
specialists in the held and that
cover all main national standard varie-
ties, distinctive regional, ethnic, and so-
cial varieties, major contact varieties, as
well as major ESL varieties;
share a common core, which makes them
invaluable research tools for cross-lin-
guistic comparisons;
provide information on UIF historical
and cultural backgrounds as well as UIF
current sociolinguistic situations in
the respective regions;
A MOUTON MULTIMEDIA REFERENCE AND RESEARCH TOOL
Q A Handbook of
Varictics of fngIish
A MuItimcdia Rcfcrcncc TooI
VoIumc 1: PhonoIogy - VoIumc
2: MorphoIogy, Syntax
%DITEDBY"ERND+ORTMANNAND
%DGAR73CHNEIDERTOGETHERWITH
+ATE"URRIDGE2AJEND-ESTHRIE
AND#LIVE5PTON
2004. Approx. 2500 pages. Cloth. 2 books
and CD-ROM (Single User).
ISBN 3-11-017532-0
serve as state-of-the-art reports on ma-
jor issues in current research.
e CD-ROM not only supplements the
printed volumes through interactive access
to the varieties but also provides a compre-
hensive database with:
a unique collection of speech record-
ings of English from around the world;
sound samples that open new per-
spectives on the varieties of English,
as speech recordings also constitute the
central aspect of research;
interactive and synchronized maps that
allow either phonological or morphosyn-
tactic (grammatical) comparisons.
Given its accessible style and its rich audi-
tory and visual support, this )BOECPPL is
also ideally suited not only for professional
academics but also for undergraduate and
beginning graduate students.
SNfAk PRfVlfW AVAllABlf
PIcasc visit our wcbsitc at www.mouton-onIinc.com
for morc information and for a frcc sncak prcvicw.

A Comparative Grammar of Brtiish English Dialects

Cargado por

Información del documento

Descripción 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

A Comparative Grammar of Brtiish English Dialects

Cargado por

Copyright:

Formatos disponibles

A Comparative Grammar of British English Dialects

Topics in English Linguistics

ii. *eine junge Mdchen a

and die Frau ist nettthe woman is nice

También podría gustarte