Está en la página 1de 15

Using Eye Tracking to Observe

Differential Effects of Repeated


Readings for Second-Grade Students
as a Function of Achievement Level
Andrea M. Zawoyski

ABSTR ACT

Scott P. Ardoin

Repeated readings (RR) is an evidence-based instructional technique in which


students read the same text multiple times. Currently, little is known about
how effects of RR may differ based on students achievement levels. Eye
tracking provides a means for closely examining instructional effects because
it permits measurement of subtle changes that occur during RR. The current
study measured changes in the reading behavior of second-grade students
who were divided into two groups of 22 students each based on their reading
achievement levels. Participants read a grade-level passage embedded with
low- and high-frequency target words four times in a single session while
their eye movements were recorded. Findings replicated those of previous
research, suggesting that RR facilitated reading for students in both groups,
particularly on low-frequency target words. Results indicated both similarities and differences in patterns of performance between lower and higher
performing readers. Additionally, results implied that effects were greater
for lower performing readers because they made greater improvements on
high-frequency target words, whereas effects were diminished for higher
performing readers. The findings have implications for improving future eye
movement research investigating young students reading and the efficiency
of RR in the classroom.

University of Georgia, Athens, USA

Katherine S. Binder
Mount Holyoke College, South Hadley,
Massachusetts, USA

eading fluency is a key component of proficient reading and an


essential area of instruction for early elementary readers
(National Institute of Child Health and Human Development
[NICHD], 2000). Despite efforts to foster a nation of proficient readers, data from the National Center for Education Statistics indicate
that 33% of fourth graders are still reading below the basic level (Aud
etal., 2012). These disappointing findings underscore the need for a
better understanding of the specific effects that instruction has on
students reading. Knowing exactly how instructional techniques improve students reading should assist teachers in making appropriate
instructional modifications so a greater proportion of students can
develop adequate reading skills.

Repeated Readings

Reading Research Quarterly, 50(2)


pp. 171184 | doi:10.1002/rrq.91
2014 International Literacy Association.

Repeated readings (RR) is a well-known instructional technique for


improving students reading fluency. Fundamentally, RR requires
students to reread a text until they reach a criterion (Chard, Vaughn,
& Tyler, 2002; Therrien, 2004). The foundational principles of RR are
derived from LaBerge and Samuelss (1974) theory of automaticity,

171

which suggests that readers must achieve automaticity


in lower level processes (e.g., decoding, word recognition) before they can allocate enough of their attention
to higher level processes (e.g., comprehension) to permit
sufficient understanding of a text. When lower level
skills are not automatic, readers must expend a majority
of their cognitive resources on deciphering individual
words, reducing their attention to the meaning of a text.
It is thought that repeated practice of a text improves
readers automaticity with the decoding and recognition of words within the text, permitting them to focus
greater attention on text comprehension (Huey, 1908/
1968; Samuels, 1979, 2006).
Findings from applied research on RR demonstrate
that repeated practice can improve reading for numerous populations, including skilled readers (Ardoin,
Morena, Binder, & Foster, 2013; Levy, Nicholls, & Kohen,
1993; Sindelar, Monda, & OShea, 1990), readers with
learning disabilities (Chard etal., 2002), and all students
through fourth grade (NICHD, 2000). In a meta-analysis
of RR research, Therrien (2004) analyzed mean improvements on reread passages, and reported effect sizes
of .83 for fluency and .67 for comprehension.
Despite strong support for the effectiveness of RR, it
remains unknown whether the procedure improves fluency and comprehension for all readers equally or if RR
differentially impacts component reading skills, such
that lower performing readers experience greater improvements in word recognition, whereas higher performing readers experience greater improvements in
comprehension (NICHD, 2000). For example, Levy
etal. (1993) found that RR resulted in improved reading
rates, comprehension, nonword error detection (i.e.,
identifying made-up words), and higher order word error detection (i.e., identifying real words that did not
make sense in context) for high- and low-performing
elementary students on passages of varying difficulty.
Despite improvements across readings for both groups,
notable differences were observed. Findings suggested
that low-performing readers initially improved in measures of word-level processing, whereas high-performing
readers initially performed well and reached efficiency
in word-level processing but required facilitation from
rereading to improve on measures of comprehension.
Similar effects were reported by Faulkner and Levy
(1994), who had third- and sixth-grade students read
one passage followed by a second that was (a) the same
passage, (b) a passage with high word overlap, (c) a passage with high content overlap, or (d) a passage that
lacked word and content similarity. Analyses revealed
that when passages were easy for students to read, their
reading times did not decrease significantly on passages
with high word overlap but did decrease significantly on
passages with high content overlap. Results suggested
that when students read passages that were easy for them

172 | Reading Research Quarterly, 50(2)

to read, they were able to focus their attention on comprehension as opposed to word recognition. In contrast,
when reading difficult passages, benefits were observed
on both comprehension and word recognition measures.
These results were largely replicated in a follow-up study
involving fourth-grade students and undergraduates
(Faulkner & Levy, 1999).
Together, findings from Faulkner and Levy (1994,
1999) imply that RR generally improves reading but that
the difficulty level of practice passages relative to students skill levels may affect the magnitude or type of
benefits that students receive from RR. A notable limitation of these studies is that conclusions about how skill
and difficulty level impacted text processing were based
on data collected using transfer passages. Unlike studies
in which students repeatedly read the same passage,
Faulkner and Levys results may be affected by variability resulting from the different words and content in the
transfer passages. Furthermore, the rereading condition
involved only one repetition and did not reflect recommended RR procedures (Samuels, 1979; Therrien, 2004).
Another limitation present in the aforementioned
studies is that only products of reading behavior (word
reading rate and accuracy) were measured and not actual reading behavior across readings. In fact, despite
substantial evidence supporting RRs effectiveness
(Therrien, 2004), researchers have yet to determine
what underlying changes in reading behavior from RR
result in improved reading and whether the benefits
are differentiated on the basis of skill or text level.
Fortunately, recent improvements in eye-tracking technology permit data collection on students eye movements during RR (Rayner, Ardoin, & Binder, 2013),
allowing for direct observation of reading behavior
across practice readings.

Eye Movement Research


Eye Movements and
Dependent Measures
Eye movement research reveals that during reading, the
eyes actually make a series of quick, jumping movements
(saccades) and pauses (fixations; Rayner, 1998; Rayner,
Chace, Slattery, & Ashby, 2006). The eyes also make
backward saccades (regressions), which can occur between words (interword regressions) or within words
(intraword regressions). Eye-tracking systems can record
multiple parameters of eye movements during reading.
Several common parameters are first fixation duration,
gaze duration, number of intraword regressions, total
fixation time, average fixation count per word (or target
word), and number of interword regressions.
First fixation duration is the length of time required
for readers to make their first fixation on a word,

regardless of the number of times the word is fixated.


Gaze duration is the sum of all fixations made on a word
before a saccade is made to another word. The number of
intraword regressions indicates the total number of regressions made within words. Total fixation time represents the duration of all fixations made on a word. Unlike
gaze duration, total fixation time includes fixations that
occur after a regression is made to that word. Average
fixation count per word (or target word) indicates the average number of fixations made on each word (for global
analyses) or on each target word (for target word analyses). The number of interword regressions is the total
count of regressions made between words.
Research suggests that eye movement measures represent different aspects of reading behavior, grouped
into categories called early and late processing. Early
processing refers to initial stages of word recognition,
whereas late processing refers to discourse-level processes such as meaning selection and text integration
(Binder, 2003; Binder & Morris, 1995). Eye movements
considered to represent early processing are first fixation duration, gaze duration, and number of intraword
regressions. Eye movements considered to represent late
processing are total fixation time, average fixation
count per word (or target word), and number of interword regressions. Interestingly, Ashby, Rayner, and
Clifton (2005) found that arduous reading is characterized by longer fixations on words, shorter saccades, and
more frequent regressions. In contrast, efficient reading
is characterized by shorter fixations, longer saccades,
and less frequent regressions.

Eye Movement Studies on Rereading


Eye movement research conducted with adults suggests
that the effects of rereading a passage facilitate their
processing of texts, resulting in faster reading of a text
and more efficient eye movements (Hyn & Niemi,
1990; Levy & Burns, 1990; Shebilske & Fisher, 1980).
Interestingly, sensitivity to word frequency remains
across readings, with readers spending significantly
more time processing low-frequency words than highfrequency words (Raney & Rayner, 1995).
Foster, Ardoin, and Binder (2013) examined changes
in the eye movements of second-grade students during
RR. Reading behavior was analyzed globally (i.e., across
the passage) and on embedded low- and high-frequency
target words. Findings revealed that RR reduced the
amount of time required for students to read the passage and engage in early and late processing. Between
the first and second readings, immediate decreases
were observed in measures of gaze duration, total fixation time, average number of fixations per word, and
number of interword regressions. Interestingly, continued effects were observed for measures indicative of late

processing (i.e., total fixation time, number of interword regressions, average number of fixations per
word). In many ways, the findings replicated known
changes in the eye movement patterns of adults during
rereading (Hyn & Niemi, 1990; Raney & Rayner,
1995; Shebilske & Fisher, 1980), such as significant
effects of word frequency and reduced overall reading
time.
Interesting differences were also noted between
Foster etal.s (2013) study and the aforementioned adult
rereading literature. Specifically, when averaged across
embedded target words, childrens gaze duration and
total fixation time decreased across readings on lowfrequency target words but not on high-frequency target
words. In contrast, findings from adult literature (Raney
& Rayner, 1995) indicate significant decreases in gaze
duration and total fixation time on both low- and highfrequency target words. These differences suggest that
RR primarily improves childrens reading efficiency on
low-frequency words and reduces the amount of additional processing time required for children to read text.
Findings from Foster etal. were largely replicated in a
follow-up study examining the impact of RR on target
words presented in generalization passages (Ardoin,
Binder, Zawoyski, Foster, & Blevins, 2013). Together,
these studies provide a preliminary understanding of
how RR prompts changes in reading behaviors and ultimately leads to improvement captured by outcome measures (e.g., words read correctly in a minute [WRCM]).
Unfortunately, the implications of these studies are limited because the researchers did not examine changes as
a function of students skills, making it difficult to ascertain whether improvements in students reading rate
and accuracy were due to similar changes in underlying
reading behavior.
Given that the National Reading Panel suggested
that high- and low-performing readers may benefit differently from RR (NICHD, 2000), failure to address this
question precludes a comprehensive understanding of
how RR improves these students oral reading rate with
accuracy. Furthermore, eye movement research on rereading with adults (Raney & Rayner, 1995) and applied RR research with children (Faulkner & Levy,
1994, 1999; Levy et al., 1993) suggest that effects may
differ depending on a readers skill level. Despite preliminary evidence for the effectiveness of RR in improving reading for early elementary students (e.g.,
Ardoin, Binder, et al., 2013; Foster et al., 2013), eye
movement research has yet to examine potential differential effects of RR for lower and higher performing
students in this population. Therefore, the purpose of
the current study was to extend the RR and eye movement literature by evaluating the reading behavior of
lower and higher performing readers during RR of a
grade-level passage.

Using Eye Tracking to Observe Differential Effects of Repeated Readings for Second-Grade Students as a Function of Achievement Level | 173

Method
Participants and Settings
The participants were 44 second-grade students (20
males, 24 females) with a mean age of 7years 11months
(range = 7 years 4 months to 8 years 11 months).
Students in the sample identified as Caucasian (81%),
multiracial (9%), Asian (5%), or black (5%) and attended
one of three suburban public schools in the Southeastern
United States. Most second-grade students at these
schools met state standards for English/language arts
(8795%) and reading (91100%). At the time of assessment, free and reduced-price lunch rates for the schools
ranged from 18% to 29%.
Participants were selected from a group of students
participating in a larger study that involved pretesting,
10 weeks of intervention or assignment to a control
group, and posttesting. Data for the current study were
collected during the posttesting period. Of note, to ensure that the data set represented complete sessions of
RR, 55 participants who skipped portions of the text
were excluded from the larger sample. Because of requirements for the larger study, no participants in the
current study received special education, gifted, or
English for Speakers of Other Languages services.
In terms of oral reading fluency level, participants
represented the bottom 25% (lower performing readers,
n = 22) and the top 25% (higher performing readers,
n = 22) of the sample for the larger study. Rank order
was determined by each students median oral reading
fluency (ORF) scores on Formative Assessment Instrumentation and Procedures for Reading curriculum-based
measurement probes (Christ, Ardoin, Monaghen, Van
Norman, & White, 2013). On average, the lower performing readers attained median ORF scores of 74
WRCM (range = 4392, standard deviation [SD] = 13),
whereas the higher performing readers attained average
median ORF scores of 144 WRCM (range = 127224,
SD= 22). Significant differences between groups of lower
and higher performing readers were confirmed by t-test
(p < .001). The ranges of ORF scores indicate that lower
performing readers read grade-level passages near or
slightly above an instructional level and that higher performing readers read grade-level passages at the mastery
level.
All students completed the letter-word identification, reading fluency, passage comprehension, and word
attack subtests of the WoodcockJohnson III, Form A
(Woodcock, McGrew, & Mather, 2001). Broad reading
composite standard scores fell within the average range
for lower performing readers (mean [M] = 106, range =
95111, SD = 5) and the average to high-average range
for higher performing readers (M = 117, range = 106
126, SD = 5). Results from t-tests indicated that there
were significant differences (p < .001) between groups

174 | Reading Research Quarterly, 50(2)

of lower and higher performing readers for the broad


reading composite score and all subtests with the exception of word attack, for which the t-test was significant
at p=.02.

Apparatus
Eye movement data were collected with an SR Research
EyeLink 1000 system. This system has a resolution of
0.01 degrees of visual angle, a range of 32 horizontally
and 25 vertically, and a sampling rate of 1000Hz. Eye
movements were recorded by a desktop-mounted camera positioned directly in front of a computer screen,
which was either a 19 (48.26cm) ViewSonic VG930m
or a 22 (55.88cm) ViewSonic VX2268wm LCD display
monitor. The camera recorded movements from only
one eye, but participants viewing was binocular
throughout the assessment. Typically, the system recorded movements from the right eye. Data were collected from the left eye in the event of tracking
difficulties with the right eye.
The camera was positioned at the recommended
tracking distance, approximately 5055 cm from the
chin rest on which participants placed their chins during tracking. The brightness and height of the monitor
were adjusted prior to assessment. Participants indicated when they had finished reading each passage and
responded to comprehension questions by pressing the
left and right analog trigger keys on a Microsoft
SideWinder Plug & Play game pad.

Materials
The stimulus was an experimenter-created passage
(see the Appendix). It consisted of four paragraphs,
containing 16 sentences and 162 words. The Spache
(1953) readability estimate of the passage was grade
3.18. Uniform black text appeared on the computer
monitor screen in 20-point Times New Roman font
with 1.5 line spacing against a white background. The
16 sentences of the passage spanned 13 lines, with line
length ranging from 18 to 87 characters. Distributed
across the text were five low-frequency target words,
ranging from six to 11 characters each, and five highfrequency target words with four or five characters
each. Researchers defined low-frequency words as
those that appeared 10 or fewer times per million
words of text and high-frequency words as those that
appeared 50 or more million times per words of text.
Frequency of the target words was determined using
The American Heritage Word Frequency Book (Carroll,
Davies, & Richman, 1971). Target word locations were
randomly selected, but placement at the ends of sentences was prohibited to avoid inclusion of extra processing time related to sentence wrap-up (Just &
Carpenter, 1980).

Procedure
Eye tracking was conducted by two examiners trained
in eye-tracking procedures. One examiner positioned
the participant and explained testing procedures while
the other examiner adjusted the camera and ensured
that computer settings met threshold requirements.
Then, calibration and validation commenced using a
9-point grid, which was followed by a practice trial to
familiarize participants with assessment procedures
(i.e., reading text from the screen, answering questions
with the game pad). Next, an examiner informed participants that they would read several passages silently
and urged them to do their best reading. Participants
were also told that the experimenters could not assist
them during reading and that they would be required to
answer a comprehension question following each reading. Calibration and validation procedures were then
repeated, and the assessment began.
The assessment consisted of six trials. For each trial,
participants read a passage, pressed a button on the
game pad to indicate that they had finished reading,
and used the game pad to answer a comprehension
question that was presented without accompanying
text. The first two trials were related to a larger study.
Data for the current study were gathered during participants repeated reading of a single stimulus passage (see
the Appendix) presented during the third through sixth
trials. Examiners provided additional instructions
about RR procedures prior to participants first reading
of the RR passage. Specifically, participants were told
that they would read a passage four times and that after
each reading, they should press a button on the game
pad to indicate that they had finished. If participants
independently began rereading the passage after completing one reading, examiners prompted them to press
the button. After pressing the button, participants answered a comprehension question and received feedback about their reading time.
The purpose of the comprehension question was to
encourage participants to attend to the meaning of the
text. The purpose of providing feedback about reading
time after each reading was to maintain participants
motivation for continued improvement in the next repetition. There was no time limit for reading. Sessions
lasted approximately 1520 minutes, unless technical
difficulties and/or a participants excessive body movement necessitated recalibration of the eye tracker.
Participants were allowed short breaks between readings if they seemed fatigued.

Data Analyses
The effects of RR on eye movement parameters were examined via mixed analyses of variance (ANOVAs).
Analyses were conducted at the global level (i.e., across

all words in the passage) and at the target word level


(i.e., across low- and high-frequency target words) for
groups of lower and higher performing readers.
Multivariate statistics (i.e., Wilkss ) were reported
when Mauchlys test indicated a violation of sphericity.
Log(x) transformations were conducted to correct extreme outliers (i.e., z-score > 3.29) or violations of the
assumption for homogeneity of variance. One-way
ANOVAs and Bonferroni-corrected follow-up analyses
were conducted to examine significant interactions between skill level and/or rereading and target word
frequency.
Overall effect sizes for RR are reported as values of
partial eta-squared, which can be interpreted using
Cohens (1988) f benchmarks for small (p2 = .0099), medium (p2 = .0588), and large effects (p2 = .1379). Effect
sizes for RR between pairs of readings are reported as
values of Cohens d, which can be interpreted using
Cohens d benchmarks for small (0.2), medium (0.5),
and large (0.8) effects. For global and target word analyses, significant interactions that pertained to differences between lower and higher performing readers
were followed up with pairwise comparisons and
t-tests.

Results
Global Analyses
Global measures included first fixation duration, gaze
duration, total fixation time, number of interword regressions, number of intraword regressions, and average fixation count per word. A 4 (Rereading) 2 (Skill
level) mixed ANOVA was conducted for each variable.
Due to extreme outliers and violations of Levenes test
for homogeneity of variance, log(x) transformations
were conducted for gaze duration, total fixation time,
number of intraword regressions, and average fixation
count per word. Assumptions for normality were met
following log(x) transformations, with the exception of
one data point in the analysis of higher performing
readers average fixation count per word. This data
point was removed, and data were normalized. All
variables met assumptions for homogeneity of variance after transformation, except for number of
intraword regressions. Further attempts to achieve
homogeneity of variance through transformations
(i.e., square root and reciprocal methods) also failed.
Although findings for this measure must be interpreted
with caution, changes between readings occurred in
the expected direction. Of note, there were no differences in statistically significant findings between
nontransformed and transformed data. Means, test
statistics, and effect sizes for global measures are presented in Table1.

Using Eye Tracking to Observe Differential Effects of Repeated Readings for Second-Grade Students as a Function of Achievement Level | 175

TABLE 1
Summary of Global Eye Movement Parameters Across Rereading by Skill Level
Measure

Reading 1 M (SD)

Reading 2 M (SD)

Reading 3 M (SD)

Reading 4 M (SD)

First fixation duration (ms; N = 44)


Between groups: F(1, 42) = 8.916, p = .005,
p2 = .18
Main effect (Rereading): F(3, 126) = 4.91,
p = .003, p2 = .11
Interaction (Rereading Skill level):
F(3, 126) = 1.014, p = .389, p2 = .02

280 (31)

274 (32)

274 (31)

270 (31)

Lower performing readers (n = 22)

291 (26)

287 (29)

288 (26)

281 (26)

Higher performing readers (n = 22)

269 (32)

262 (30)

259 (30)

259 (32)

Gaze duration (ms; N = 44)


Between groups: F(1, 42) = 56.563, p < .001,
p2 = .57
Main effect (Rereading): F(3, 40) = 19.820,
p < .001,a p2 = .60
Interaction (Rereading Skill level):
F(3, 40) = 1.515 p = .225,b p2 = .10

430 (106)

397 (97)*

391 (90)

377 (86)

Lower performing readers (n = 22)

510 (76)

464 (77)

443 (77)

434 (78)

Higher-Performing Readers (n = 22)

350 (61)

329 (61)

320 (52)

320 (47)

Total fixation time (ms; N = 44)


Between groups: F(1, 42) = 70.531, p < .001,
p2 = .63
Main effect (Rereading): F(3, 40) = 48.337,
p < .001,c p2 = .78
Interaction (Rereading Skill level):
F(3, 40) = .998, p = .404,d p2 = .07

634 (194)

543(172)*

504 (156)*

490 (161)

Lower performing readers (n = 22)

782 (147)

674 (134)

614 (143)

601 (155)

Higher performing readers (n = 22)

486 (99)

412 (83)

393 (64)

380 (59)

Number of interword regressions (N = 44)


Between groups: F(1, 42) = 19.830, p < .001,
p2 = .32
Main effect (Rereading): F(3, 126) = 28.898,
p < .001, p2 = .41
Interaction (Rereading Skill level):
F(3, 126) = 1.108, p = .387, p2 = .02

.317 (.112)

.273 (.105)*

.243 (.095)*

.229 (.099)

Lower performing readers (n = 22)

.362 (.092)

.335 (.089)

.292 (.081)

.284 (.084)

Higher performing readers (n = 22)

.271 (.113)

.211 (.080)

.195 (.084)

.175 (.082)

Number of intraword regressions (N = 44)


Between groups: F(1, 42) = 51.161, p < .001,
p2 = .55
Main effect (Rereading): F(3, 40) = 28.205,
p < .001,e p2 = .68
Interaction (Rereading Skill level):
F(3, 40) = 2.558, p = .069,f p2 = .16

.251 (.154)

.187 (.122)*

.160 (.010)*

.150 (.108)

Lower performing readers (n = 22)

.357 (.140)

.269 (.111)

.233 (.091)

.219 (.112)

Higher performing readers (n = 22)

.146 (.074)

.106 (.065)

.088 (.033)

.080 (.038)

(Continued)

176 | Reading Research Quarterly, 50(2)

TABLE 1
Summary of Global Eye Movement Parameters Across Rereading by Skill Level (Continued)
Measure

Reading 1 M (SD)

Reading 2 M (SD)

Reading 3 M (SD)

Reading 4 M (SD)

Average fixation count per word (N = 43)


Between groups: F(1, 41) = 74.721, p < .001,
p2 = .65
Main effect (Rereading): F(3, 39) = 55.38,
p < .001,g p2 = .81
Interaction (Rereading Skill level):
F(3, 39) = 3.186, p = .034,h p2 = .20

2.05 (0.540)

1.79 (0.518)*

1.58 (0.403)*

1.55 (0.469)

Lower performing readers (n = 22) follow-up:


F(3, 63) = 31.119, p < .001, p2 = .76

2.43 (0.447)

2.18 (0.381)*

1.89 (0.315)*

1.89 (0.437)

Higher performing readers (n = 21) follow-up:


F(3, 60) = 53.95, p < .001 , p2 = .88

1.65 (0.282)

1.37 (0.283)*

1.27 (0.179)*

1.21 (0.158)

Note. M = mean; SD = standard deviation.


a
Wilkss = .40. bWilkss = .90. cWilkss = .22. dWilkss = .93. eWilkss = .32. f Wilkss = .84. gWilkss = .19. hWilkss = .80.
*Significant pairwise differences between the denoted reading and its previous reading, p < .05. Significant pairwise differences between the first and
fourth readings, p < .05.

As expected, because of the design specification of


grouping participants by achievement scores, analyses
across all measures confirmed large, significant betweengroup effects. Significant main effects of rereading were
also observed on all measures and indicated improvement in the expected direction across all four readings of
the passage, with a medium effect on first fixation duration and large effects on gaze duration, total fixation time,
number of interword regressions, number of intraword
regressions, and average fixation count per word. Pairwise
comparisons between the first and fourth readings indicated significant aggregate effects of rereading across all
measures in the expected direction (first fixation duration, p = .014; total fixation time, p = .01; gaze duration,
number of intraword regressions, average fixation count
per word, and number of interword regressions, p<.001).
Furthermore, pairwise comparisons between the
first and second readings revealed immediate, significant effects of rereading on measures indicative of both
early processing (i.e., gaze duration and number of intraword regressions, p<.001) and late processing (i.e., number of interword regressions, p=.001; total fixation time
and average fixation count per word, p<.001). Significant
effects of rereading persisted between the second and
third readings for all measures indicative of late processing (i.e., number of interword regressions, p = .04; total
fixation time, p=.001; average fixation count per word,
p<.001) but only one measure thought to represent early
processing (i.e., number of intraword regressions,
p < .001). For all measures, the effect of rereading was
not significant between the third and fourth readings.
Average fixation count per word was the only global
dependent measure to reveal a significant Rereading
Skill level interaction. Follow-up analyses suggested that
average fixation count per word differed significantly

across readings for both lower and higher performing


readers. Analyses of significant aggregate effects of rereading indicated large effect sizes for both higher performing readers (Cohens d=1.58) and lower performing
readers (Cohens d=1.08). Pairwise comparisons revealed
significant decreases in average fixation count per word
between the first and second readings for both groups, although the effect size for rereading was large for higher
performing readers (Cohens d = 1.05) and medium for
lower performing readers (Cohens d = 0.60). Pairwise
comparisons were also significant for both groups between the second and third readings, with small effects
indicated for higher performing readers (Cohens d=0.46)
in contrast with large effects for lower performing readers
(Cohens d = 0.83). Collectively, findings suggest that RR
facilitated processing sooner for higher performing readers than for lower performing readers. Furthermore, between the second and third readings, the lower performing
readers made large gains, whereas the amount of benefit
decreased for higher performing readers.
In sum, findings from all global analyses indicated
that RR was beneficial for groups of lower and higher performing readers. As expected, analyses for all measures
indicated significant differences between groups.
Additionally, RR facilitated improvement for both groups
between the first and second readings on all measures
with the exception of first fixation duration. Improvement
continued between the second and third readings for
number of interword regressions, total fixation time, average fixation count per word, and number of intraword regressions. Interestingly, analyses for average fixation
count per word suggest that higher performing readers
may experience larger improvements sooner than lower
performing readers, who required an additional reading
to experience larger effects.

Using Eye Tracking to Observe Differential Effects of Repeated Readings for Second-Grade Students as a Function of Achievement Level | 177

Target Word Analyses

First Fixation Duration

Target word analyses were conducted on first fixation


duration, gaze duration, total fixation time, and
average fixation count per target word. A 4 (Rereading)
2 (Skill level) 2 (Word frequency) mixed ANOVA,
with rereading and word frequency as within-subjects
factors, was conducted for each measure. Due to extreme outliers and violations of Levenes test for homogeneity of variance, log(x) transformations were
conducted for all variables. Following transformation,
data were successfully normalized, with the exception
of one data point in the lower performing group
for gaze duration and one data point in the higher
performing group for total fixation time. These outliers were removed, and data were normalized. After
transformation, all variables met assumptions of
homogeneity of variance. Detailed test statistics are
subsequently presented (see Figure 1 for a visual
comparison of means for target word measures separated by skill level).

The upper left panel of Figure 1 provides a graph of


means for lower and higher performing readers on lowand high-frequency target words across readings. Target
word analyses of first fixation duration revealed no significant between-group differences, F(1, 42) = 3.66,
p=.062, p2 = .08. The three-way interaction of Rereading
Skill level Word frequency was also not significant,
F(3, 126) = 1.40, p = .248, p2 = .03. Furthermore, twoway interactions were not significant for Rereading
Word frequency, F(3, 126) = 0.68, p = .567, p2 = .02;
Rereading Skill level, F(3, 126) = 0.73, p = .537, p2 =
.02; or Skill level Word frequency, F(1, 42) = 0.84,
p=.366, p2 = .02. Also, the main effect of rereading was
not significant, F(3, 126) = 2.19, p = .092, p2 = .05, although a significant main effect for word frequency was
observed, F(1, 42) = 18.39, p<.001, p2 = .31.
The lack of significant interaction and main effects
taken together with the lack of significant betweengroup effects suggests that with respect to first fixation

FIGURE 1
Averages for Low- and High-Frequency Target Words Across Readings by Skill Level

First Fixation Duration

350
300

1000

Time (ms)

250

Time (ms)

Gaze Duration

1200

200
150
LP-High Freq
100

800
600
400

LP-Low Freq
200

HP-High Freq

50

HP-Low Freq
0

2000

Readings

Total Fixation Time

Number of Fixations

1750

Time (ms)

1500
1250
1000
750

Readings

Average Fixation Count per Target Word

5
4
3
2

500
1

250
0

Readings

Readings

Note. Averages for analyses of high-frequency (High Freq) and low-frequency (Low Freq) target words across readings are separated into groups of
lower performing (LP) and higher performing (HP) readers.

178 | Reading Research Quarterly, 50(2)

duration, reading behavior was similar for lower and


higher performing readers. Specifically, RR did not facilitate significant improvement in first fixation duration, regardless of readers skill level. Given that the
target words were fixated more than once, it is likely
that lower and higher performing readers did not improve on this measure because early processing was not
yet complete. Additionally, low-frequency words required more time to process than high-frequency words
for both lower and higher performing readers.

Gaze Duration
As expected, significant between-group differences
were observed for target word analyses of gaze duration,
F(1, 41) = 45.31, p<.001, p2 = .53, a measure thought
torepresent early processing. The three-way interaction
of Rereading Skill level Word frequency was not
significant for gaze duration, F(3, 123) = 1.09, p = .358,
p2 = .03. However, there was a significant two-way interaction between rereading and word frequency, F(3,
123) = 8.73, p<.001, p2 = .18, suggesting that facilitative
effects of RR were greatest on low-frequency target
words as opposed to high-frequency words (see the upper right panel of Figure1).
Examination of two-way interactions involving
skill level indicated that effects were not significant for
Rereading Skill level, F(3, 123) = 2.04, p = .112, p2 =
.05, but were significant for Skill level Word frequency, F(1, 41) = 10.37, p = .003, p2 = .20. Pairwise
comparisons for the Skill level Word frequency interaction indicated significant effects of word frequency
for both groups but revealed that the magnitude of the
word frequency effect was nearly twice as large for lower
performing readers, F(1, 20) = 119.95, p<.001, p2 = .86,
than for higher performing readers, F(1, 21) = 49.46,
p<.001, p2 = .70.

Total Fixation Time


Consistent with expectations, significant betweengroup differences were observed for target word analyses of total fixation time, F(1, 41) = 53.44, p<.001, p2 =
.57, a measure thought to represent higher level processing. Target word analyses of total fixation time indicated a significant Rereading Skill level Word
frequency interaction, F(3, 123) = 2.771, p<.044, p2 =
.06. To further explore the three-way interaction and its
relevance to differences between lower and higher performing readers, data were split by skill level. For higher
performing readers, there was a significant Rereading
Word frequency interaction, F(3, 60) = 11.258, p<.001,
p2 = .36, suggesting that RR resulted in greater improvements on low-frequency target words as compared
with high-frequency target words (see the lower left
panel of Figure 1). In contrast, the Rereading Word

frequency interaction was not significant for lower performing readers, F(3, 63) = 2.478, p = .07, p2 = .11.
Significant main effects of rereading, F(3, 63) = 16.903,
p<.001, p2 = .45, and word frequency, F(1, 21) = 96.321,
p<.001, p2 = .821, were observed for lower performing
readers.
Overall, findings related to total fixation time indicated that effects of RR differed for lower and higher performing readers. Specifically, lower performing readers
demonstrated improvement across readings on target
words regardless of word frequency but still required
more time to process low-frequency target words, as compared with higher performing readers. For higher performing readers, effects of RR were dependent on word
frequency, such that RR resulted in larger improvements
in total fixation time on low-frequency target words.

Average Fixation Count per Target Word


Significant between-group differences were observed
for target word analyses of average fixation count per
target word, F(1, 42) = 47.38, p<.001, p2 = .53. As with
total fixation time, this measure is thought to represent
late processing. The three-way interaction for Rereading
Skill level Word frequency was not significant, F(3,
126) = 2.28, p = .083, p2 = .05. However, all two-way
interactions were significant. These effects can be observed in the lower right panel of Figure 1. Consistent
with findings related to gaze duration and total fixation
time, there was a significant Rereading Word frequency interaction, F(3, 126) = 8.88, p<.001, p2 =.18,
indicating that the facilitative effects of RR were greatest
for low-frequency target words as opposed to highfrequency target words.
Two-way interactions related to skill level were explored further. Analysis of the Rereading Skill level
interaction, F(3, 126) = 3.10, p=.029, p2 = .07, indicated
that effects of rereading were significant for higher performing readers between the first and second readings
(p < .001), whereas significant effects of rereading for
lower performing readers were not observed until between the second and third readings (p=.025). For both
groups, significant effects were not observed between
the third and fourth readings. The significant Skill
level Word frequency interaction, F(1, 42) = 6.26,
p = .016, p2 = .13, was further explored, revealing that
the magnitude of the word frequency effect on average
fixation count per target word was greater among lower
performing readers, F(1, 21) = 103.76, p<.001, p2 = .83,
compared with higher performing readers, F(1, 21) =
55.71, p<.001, p2 = .73.
Taken together, findings revealed that RR permitted
both lower and higher performing readers to make reductions in average fixation count per target word across
readings. Interestingly, higher performing readers made

Using Eye Tracking to Observe Differential Effects of Repeated Readings for Second-Grade Students as a Function of Achievement Level | 179

significant improvements sooner (i.e., between the first


and second readings) than the lower performing readers
(i.e., not until between the second and third readings).
Furthermore, lower performing readers were more sensitive to effects of word frequency than higher performing readers.

Summary
As expected, higher performing readers spent less time
on the target words compared with the lower achieving
students, and RR facilitated processing of the target
words. In addition, lower performing students were more
sensitive to word frequency effects than were higher performing readers. Unique results from total fixation time
revealed that for lower performing readers, RR reduced
total fixation time on all target words, whereas for higher
performing readers, RR primarily reduced total fixation
time on low-frequency target words. Finally, unique
findings for analyses of average fixation count per target
word revealed that effects of rereading on average fixation count per target word were significant for higher
performing readers between the first and second readings, whereas effects were significant for lower performing readers between the second and third readings.

Discussion
Findings from numerous applied research studies suggest that RR is an effective instructional technique
(Chard etal., 2002; Therrien, 2004) for readers through
fourth grade (NICHD, 2000). Despite evidence of effectiveness, methodological limitations of these studies
leave unanswered questions regarding how and why
readers of different skill levels benefit from the same intervention. Fortunately, eye-tracking technology provides a means for researchers to observe how RR
impacts behaviors resulting in differential reading outcomes. Past eye movement studies conducted with early
elementary readers (Ardoin, Binder, etal., 2013; Foster
etal., 2013) extend upon findings from applied research
by providing more specific information about childrens
eye movements during RR but involve data analyses
across skill levels. To better understand how secondgrade readers at different skill levels benefit from rereading, the current study used eye-tracking technology
to examine underlying changes in the reading behavior
of lower and higher performing students during RR.
Students reread passages silently and without adult error correction, which closely mirrors implementation
procedures for many popular computer- or group-based
reading intervention programs.
First, results suggested that students benefited from
RR regardless of their skill level, as evidenced by increased efficiency of eye movements across the four

180 | Reading Research Quarterly, 50(2)

reading trials. Particularly, there were significant decreases in fixation frequency, fixation duration, and
number of regressions made within words or between
portions of text. Furthermore, higher performing readers
made fewer and shorter fixations than lower performing
readers. Higher performing readers also made fewer inter- and intraword regressions, indicating that they did
not need to reread parts of individual words or previously
viewed text as often as lower performing readers did.

Global Analyses
Global analyses of eye movements during RR revealed
improvement among lower and higher performing readers on all measures between the first and fourth readings. Although higher performing readers consistently
outperformed lower performing readers, the pattern of
observed improvement across readings was similar between groups. Specifically, lower and higher performing
readers made improvements on all measures except for
first fixation duration between the first and second
readings and continued to make significant improvements on measures considered to reflect late processing
throughout the second and third readings (i.e., total fixation time, number of interword regressions, average
fixation count per word). This finding adds to the theory of automaticity (LaBerge & Samuels, 1974) and contributes to our understanding of how RR works because
it suggests that in earlier readings, RR facilitated both
early and late processing, and in later readings, RR allowed students to focus attention on late processing.
Given that participants in both groups made improvements in measures of late processing, findings
suggest that RR may help students achieve the major
goal of reading instruction, which is to gain meaning
from text at the passage level. Conclusions regarding
changes in students reading comprehension are not
made because the study was not designed to measure
this variable. However, changes in the eye movement
measures that are thought to reflect later stages of processing suggest that readers had to devote less effort to
this type of processing during the later readings.
Interestingly, findings indicated that neither group
improved significantly between the third and fourth
readings, suggesting that readers in both groups required only three readings to attain the full facilitative
effects of RR. Foster etal. (2013) also observed that students made optimal gains within three readings. Thus,
findings from the current study support and extend
these findings to lower and higher performing readers.
The significant interaction between rereading and
skill level in average fixation count per word revealed interesting differences between lower and higher performing readers pattern of reading. Specifically, the greatest
decrease in average fixation count per word for higher

performing readers occurred between the first and second readings, whereas for lower performing readers, it
occurred between the second and third readings. Given
that a decrease in fixation count during RR is thought to
represent facilitation in late processing (Hyn & Niemi,
1990), results imply that higher performing readers, who
were rereading the passage at a mastery level, engaged in
more late processing sooner than lower performing
readers. In contrast, lower performing readers, who
were rereading the passage at the instructional level, required an additional reading before they could achieve a
comparable degree of late processing. Failure to find significant interactions between rereading and skill level
on other measures may be due to low statistical power.

averaged across readings, the difference between gaze


duration on low- versus high-frequency target words
was 432 ms for lower performing readers but only
182ms for higher performing readers.
The significant three-way interaction observed in
target word analyses of total fixation time revealed important differences in how RR facilitated processing for
readers of different skill levels. Specifically, findings indicated that RR mainly helped the higher performing
readers process low-frequency target words, because
floor effects were observed on high-frequency target
words. In contrast, findings suggested that RR facilitated lower performing readers processing of both lowand high-frequency target words.

Target Word Analyses

Limitations and Future Directions

Target word analyses supported conclusions from previous research suggesting that RR primarily improves
the reading fluency of second-grade readers by reducing the amount of additional processing time required
to read low-frequency target words (Foster etal., 2013).
In comparison with high-frequency target words, lowfrequency target words required more processing time,
particularly for lower performing readers. Findings
suggested that total fixation time on high-frequency
target words across readings steadily decreased for
lower performing readers. However, for higher performing readers, effects leveled out across readings and
resembled a floor effect, suggesting that they may
not have required as much facilitation on the highfrequency target words as did the lower performing
readers. Evidence from target word analyses also supported findings from global analyses suggesting that
higher performing readers made greater improvements
sooner than lower performing readers.
Findings from target word analyses pertaining to
the impact of word frequency replicate previous research and extend implications to lower and higher performing readers. As observed by Foster etal. (2013) and
Raney and Rayner (1995), word frequency effects were
significant across target word analyses, revealing that
low-frequency target words required more processing
time than did high-frequency target words. However,
significant interactions between skill level and word
frequency were observed on measures of gaze duration
and average fixation count per target word. Results suggested that the magnitude of the word frequency effect
was greater for lower performing readers compared
with higher performing readers. This finding implies
that low-frequency target words had a greater negative
impact on the reading of lower performing readers than
they did on higher performing readers. Conversely, the
low-frequency target words were potentially not as challenging for higher performing readers. As an example,

Results from the current study should be interpreted


with consideration of multiple limitations. First, because the lower performing participants were not normatively low performing, additional research needs to
be conducted to examine the impact of RR on students
who are struggling to develop reading fluency. However,
given that for the lower performing readers, the gradelevel passage was at their instructional level, conclusions about how they improved from RR may apply to
struggling readers who are rereading a passage that is at
their own instructional level. Future researchers are encouraged to consider how assessment passages align
with students reading levels and examine withinstudent differences related to the impact of text difficulty on eye movements during reading.
A persistent limitation to eye movement research on
RR with children is that the current technology does
not allow for reliable data collection during oral reading
tasks. Although these data extend the RR research by
demonstrating that RR improves students silent reading, studies examining students eye movements during
oral reading would also contribute to the understanding of instructional techniques. For example, the lack of
reading accuracy data in the current study presents a
significant limitation because it is unclear whether participants read words correctly during silent reading.
Once technology allows for valid oral reading data collection, researchers examining eye movements during
oral reading could expand on the current study by examining skill-based differences in students reading accuracy. Researchers may also investigate the online
effects of error correction procedures.
A limitation pertaining to the low- and highfrequency target words is that the lengths and locations
of the words were not controlled well. Specifically, the
low-frequency target words were longer then the highfrequency target words, which may have contributed to
the increased amount of time and number of fixations

Using Eye Tracking to Observe Differential Effects of Repeated Readings for Second-Grade Students as a Function of Achievement Level | 181

required for students to read these words. However, our


purpose was not to demonstrate, as many others have,
that reading behavior differs based on word frequency
(Raney & Rayner, 1995; Rayner, 1998). Rather, the purpose of examining low- and high-frequency target
words in the current study was to gather information
about how RR facilitates improvement of words based
on frequency. Importantly, the target words, high and
low frequency, were the same in each reading, and what
is of most interest is how the processing changed for
those target words across readings.
A final limitation of this study is that generalization
effects were not assessed. Thus, results do not provide
additional support for the efficacy of RR across materials. Future researchers examining eye movements of
children during RR are encouraged to include a generalization passage and outcome measures in their research.

Summary and Implications


Overall, findings from the current study imply that RR
will improve second-grade students reading fluency,
regardless of their skill level. Results demonstrated that
for groups of both lower and higher performing readers,
RR facilitated early and late processing in initial readings and continued to facilitate late processing in later
readings. Interestingly, on average fixation count per
word, a measure of late processing, effect size for each
reading varied between groups, revealing differences in
eye movement behavior for lower and higher performing readers. Specifically, effects were greater for higher
performing readers in earlier repetitions and greater for
lower performing readers in later repetitions. Results
from target word analyses indicated that lower performing readers were more sensitive to effects of word
frequency than the higher performing readers were.
Together, findings have important implications for
improving eye movement research and measurement,
tailoring classroom instruction by skill level, and enhancing the effectiveness of RR.

Implications for Researchers


First, researchers interested in studying the eye movements of children during reading are encouraged to
consider how participants skill level may impact results. Given observed differences in the eye movements
of students within the same grade level, researchers are
cautioned against grouping all readers in one grade together for experimental analysis.
A second interesting implication for eye movement
research pertains to findings from target word analyses
of first fixation duration. The lack of a significant
between-group effect may relate to how lower and higher
performing early elementary readers initially approach
words. Unlike adult readers, early elementary readers

182 | Reading Research Quarterly, 50(2)

are more likely to refixate on words (Blythe, Hiki,


Bertam, Liversedge, & Hyn, 2011). Therefore, significant effects of RR and between-subject differences in
processing time are more likely to emerge on measures that include these additional fixations (e.g., gaze
duration, total fixation time). As a result, researchers
conducting future eye movement studies with early elementary readers may consider placing emphasis on interpretations of results related to gaze duration or total
fixation time. Additionally, similarities in the first fixation durations of lower and higher performing readers
suggest that even if higher performing elementary readers are reading at the mastery level, they are still learning to read and do not process text in the same manner
that adults would.

Classroom Implications
In addition to research implications, results have applied implications for modifying classroom instruction
to benefit lower and higher performing readers. Data
collected on low-frequency target words (see Figure 1)
suggests that after four readings of a text, lower performing readers performance was similar to that of
higher performing readers on their first reading. Thus,
RR alone was not sufficient to enable lower performing
readers to match the levels of reading efficiency exhibited by higher performing readers. RR merely allowed
lower performing readers to meet the levels of eye
movement efficiency that higher performing readers
had attained on their first reading. To facilitate faster
improvement, lower performing readers may require
RR with the addition of evidence-based antecedent
strategies such as listening passage preview (Daly,
Martens, Hamler, Dool, & Eckert, 1999) or contingent
reinforcement and performance feedback (Eckert,
Ardoin, Daly, & Martens, 2002).
Results may also be useful for improving the effectiveness and efficiency of RR. Because neither lower nor
higher performing readers improved significantly between the third and fourth readings, all students may
experience diminishing returns from reading a text four
or more times in one session. This finding adds clarity
to recommendations provided by Therriens (2004)
meta-analysis of RR, which state that students should
reread passages three or four times. Programming for
three readings of a text may maximize the benefits of
RR and minimize the amount of time students spend
outside of general classroom instruction. Of note, the
current study evaluated the benefits of RR within a single session, as opposed to the benefits of RR on the same
passage across time (e.g., hours, days) or the effects of
RR across multiple passages. Results of this study are
limited to the effects of RR on passages practiced during
a single session.

Findings also have implications for the selection of


the difficulty level of materials that teachers might have
students read during RR exercises. Past research indicates that students are likely to benefit more from reading instruction conducted with difficult text (Daly,
Martens, Kilmer, & Massie, 1996; Stahl, Heubach, &
Cramond, 1997; Therrien & Kubina, 2006). Our results
replicate these findings in that the absolute level of
change observed was greater for lower performing students for whom the text was at an instructional level
than for higher performing students for whom the text
was at a mastery level. However, the data also suggest
that higher performing students made more rapid improvements in average fixation count per word, a measure thought to reflect late processing. Lower performing
readers, who read at the instructional level, exhibited
significantly slower efficiency of improvement on late
processing measures, as compared with higher performing readers. Thus, if a teachers goal is to improve a
students late processing, such improvement may be
more likely to occur if the materials are at a mastery
level for the student because he or she will not need to
expend attention on decoding individual words and can
focus on reading the passage as a whole. However, if
faster word recognition, decoding, and comprehension
are the goals of instruction, then students can improve
these skills with repeated practice on instructional-level
passages, although it is likely that improvements in late
processing will not occur until later repetitions.
Certainly, questions remain regarding differences
in how RR impacts readers eye movements based on
their reading skill. However, the general conclusion that
RR is effective for early elementary readers of varying
skill levels is important and provides basic insight into
the development of reading. Ultimately, conclusions
confirm the National Reading Panel s (NICHD, 2000)
recommendation that RR will improve reading skills of
all students through fourth grade and support the panel s implication that RR may differentially benefit students based on their achievement level.
NOTE
The research reported here was supported by the Institute of
Education Sciences, U.S. Department of Education, through Grant
R305A100496 to the University of Georgia. The opinions expressed
are those of the authors and do not represent views of the Institute
of Education Sciences or the U.S. Department of Education.

REFERENCES
Ardoin, S.P., Binder, K.S., Zawoyski, A.M., Foster, T.E., & Blevins,
L.A. (2013). Using eye-tracking procedures to evaluate generalization effects: Practicing target words during repeated reading within
versus across texts. School Psychology Review, 42(4), 477495.
Ardoin, S.P., Morena, L.S., Binder, K.S., & Foster, T.E. (2013).
Examining the impact of feedback and repeated readings on oral
reading fluency: Lets not forget prosody. School Psychology
Quarterly, 28(4), 391 404. doi:10.1037/spq0000027

Ashby, J., Rayner, K., & Clifton, C. (2005). Eye movements of highly
skilled and average readers: Differential effects of frequency and
predictability. The Quarterly Journal of Experimental Psychology:
Section A: Human Experimental Psychology, 58(6), 1065 1086.
doi:10.1080/02724980443000476
Aud, S., Hussar, W., Johnson, F., Kena, G., Roth, E., Manning, E.,
Zhang, J. (2012). The condition of education 2012 (NCES
Publication No. 2012-045). Washington, DC: Institute of
Education Sciences, National Center for Education Statistics, U.S.
Department of Education.
Binder, K.S. (2003). Sentential and discourse topic effects on lexical
ambiguity processing: An eye movement examination. Memory
& Cognition, 31(5), 690 702. doi:10.3758/BF03196108
Binder, K.S., & Morris, R.K. (1995). Eye movements and lexical
ambiguity resolution: Effects of prior encounter and discourse
topic. Journal of Experimental Psychology: Learning, Memory,
and Cognition, 21(5), 1186 1196. doi:10.1037/0278-7393.21.5
.1186
Blythe, H.I., Hiki, T., Bertam, R., Liversedge, S.P., & Hyn, J.
(2011). Reading disappearing text: Why do children refixate words?
Vision Research, 51(1), 8492. doi:10.1016/j.visres.2010.10.003
Carroll, J.B., Davies, P., & Richman, B. (1971). The American
Heritage word frequency book. Boston, MA: Houghton Mifflin.
Chard, D.J., Vaughn, S., & Tyler, B.J. (2002). A synthesis of research on
effective interventions for building fluency with elementary students
with learning disabilities. Journal of Learning Disabilities, 35(5),
386406. doi:10.1177/00222194020350050101
Christ, T.J., Ardoin, S.P., Monaghen, B., Van Norman, E., & White,
M.J. (2013). CBMReading: Technical manual. Minneapolis:
Department of Educational Psychology, University of Minnesota.
Cohen, J. (1988). Statistical power analysis for the behavioral sciences
(2nd ed.). Hillsdale, NJ: Erlbaum.
Daly, E.J., III, Martens, B.K., Hamler, K.R., Dool, E.J., & Eckert, T.L.
(1999). A brief experimental analysis for identifying instructional
components needed to improve oral reading fluency. Journal of
Applied Behavior Analysis, 32(1), 8394. doi:10.1901/jaba.1999.32-83
Daly, E.J., III, Martens, B.K., Kilmer, A., & Massie, D.R. (1996). The
effects of instructional match and content overlap on generalized
reading performance. Journal of Applied Behavior Analysis, 29(4),
507518. doi:10.1901/jaba.1996.29-507
Eckert, T.L., Ardoin, S.P., Daly, E.J., III, & Martens, B.K. (2002).
Improving oral reading fluency: A brief experimental analysis of
combining an antecedent intervention with consequences.
Journal of Applied Behavior Analysis, 35(3), 271281. doi:10.1901/
jaba.2002.35-271
Faulkner, H.J., & Levy, B.A. (1994). How text difficulty and reader
skill interact to produce differential reliance on word and content
overlap in reading transfer. Journal of Experimental Child
Psychology, 58(1), 124. doi:10.1006/jecp.1994.1023
Faulkner, H.J., & Levy, B.A. (1999). Fluent and nonfluent forms of
transfer in reading: Words and their message. Psychonomic
Bulletin & Review, 6(1), 111116. doi:10.3758/BF03210817
Foster, T.E., Ardoin, S.P., & Binder, K.S. (2013). Underlying changes
in repeated reading: An eye movement study. School Psychology
Review, 42(2), 140 156.
Huey, E.B. (1968). The psychology and pedagogy of reading.
Cambridge, MA: MIT Press. (Original work published 1908)
Hyn, J., & Niemi, P. (1990). Eye movements during repeated reading of a text. Acta Psychologica, 73(3), 259280. doi:10.1016/
0001-6918(90)90026-C
Just, M.A., & Carpenter, P.A. (1980). A theory of reading: From eye
fixations to comprehension. Psychological Review, 87(4), 329354.
doi:10.1037/0033-295X.87.4.329
LaBerge, D., & Samuels, S.J. (1974). Toward a theory of automatic
information processing in reading. Cognitive Psychology, 6(2),
293323. doi:10.1016/0010-0285(74)90015-2

Using Eye Tracking to Observe Differential Effects of Repeated Readings for Second-Grade Students as a Function of Achievement Level | 183

Levy, B.A., & Burns, K.I. (1990). Reprocessing text: Contributions


from conceptually driven processes. Canadian Journal of
Experimental Psychology, 44(4), 465 482. doi:10.1037/h0084265
Levy, B.A., Nicholls, A., & Kohen, D. (1993). Repeated readings:
Process benefits for good and poor readers. Journal of Experimental
Child Psychology, 56(3), 303327. doi:10.1006/jecp.1993.1037
National Institute of Child Health and Human Development.
(2000). Report of the National Reading Panel. Teaching children to
read: An evidence-based assessment of the scientific research literature on reading and its implications for reading instruction:
Reports of the subgroups (NIH Publication No. 00-4754).
Washington, DC: U.S. Government Printing Office.
Raney, G.E., & Rayner, K. (1995). Word frequency effects and eye movements during two readings of a text. Canadian Journal ofExperimental
Psychology, 49(2), 151172. doi:10.1037/1196-1961.49.2.151
Rayner, K. (1998). Eye movements in reading and information processing: 20years of research. Psychological Bulletin, 124(3), 372
422. doi:10.1037/0033-2909.124.3.372
Rayner, K., Ardoin, S.P., & Binder, K.S. (2013). Childrens eye movements
in reading: A commentary. School Psychology Review, 42(2), 223233.
Rayner, K., Chace, K.H., Slattery, T.J., & Ashby, J. (2006). Eye movements as reflections of comprehension processes in reading. Scientific
Studies of Reading, 10(3), 241255. doi:10.1207/s1532799xssr1003_3
Samuels, S.J. (1979). The method of repeated readings. The Reading
Teacher, 32(4), 403 408.
Samuels, S.J. (2006). Looking backward: Reflections on a career in
reading. Journal of Literacy Research, 38(3), 327344. doi:10.1207/
s15548430jlr3803_3
Shebilske, W.L., & Fisher, D.F. (1980, December). Eye movements reveal
components of flexible reading strategies. Paper presented at the annual meeting of the National Reading Conference, San Diego, CA.
Sindelar, P.T., Monda, L.E., & OShea, L.J. (1990). Effects of repeated
readings on instructional- and mastery-level readers. The Journal
of Educational Research, 83(4), 220 226.

Spache, G. (1953). A new readability formula for primary-grade


reading materials. The Elementary School Journal, 53(7),
410 413. doi:10.1086/458513
Stahl, S.A., Heubach, K., & Cramond, B. (1997). Fluencyoriented reading instruction (Reading Research Report No.79).
Athens, GA & College Park, MD: National Reading Research
Center.
Therrien, W.J. (2004). Fluency and comprehension gains as a result
of repeated reading. Remedial and Special Education, 25(4), 252
261. doi:10.1177/07419325040250040801
Therrien, W.J., & Kubina, R.M. (2006). Developing reading fluency
with repeated reading. Intervention in School and Clinic, 41(3),
156 160. doi:10.1177/10534512060410030501
Woodcock, R.W., McGrew, K.S., & Mather, N. (2001). Woodcock
Johnson III Tests of Achievement. Itasca, IL: Riverside.
Submitted June 19, 2014
Final revision received September 10, 2014
Accepted September 11, 2014

ANDREA M. ZAWOYSKI is a doctoral student in the School


Psychology Program within the Department of Educational
Psychology at the University of Georgia, Athens, USA;
e-mail zawoyski@uga.edu.
SCOTT P. ARDOIN (corresponding author) is a professor in
the School Psychology Program within the Department of
Educational Psychology at the University of Georgia,
Athens, USA; e-mail spardoin@uga.edu.
KATHERINE S. BINDER is a professor in the Department of
Psychology and Education at Mount Holyoke College, South
Hadley, Massachusetts, USA; e-mail kbinder@mtholyoke.edu.

A PPENDIX

Stimulus Passage
In this presentation of the stimulus passage, highfrequency target words are italicized, and low-frequency
target words are underlined. Target words were not italicized or underlined in the version that participants
viewed.
Emma is the most colorful dragon you will ever see. She has
yellow flecks on her pink body. Her head is deep purple but
her tail is green. Her wings are fire red with ocean blue spots.
Emmas best buddy is Queen Kathy. They met a long time
ago when Queen Kathy helped Emma.

184 | Reading Research Quarterly, 50(2)

It began, years ago, when Emma flew to Dees. Dees is the


only supermarket that sells cartons of Emmas favorite food,
Dragon Pops. Dees new owner did not like dragons. He
would not order the Dragon Pops for her. This made her feel
sad. She went to the woods for refuge and cried.
When Queen Kathy saw Emma crying, she asked what
incident happened, and Emma told her. Queen Kathy told
Emma that because she is Queen, she is the boss of all the
stores. She told the new owner to order Dragon Pops and he
did. Ever since then, Queen Kathy and Emma have been
best buddies.

Copyright of Reading Research Quarterly is the property of Wiley-Blackwell and its content
may not be copied or emailed to multiple sites or posted to a listserv without the copyright
holder's express written permission. However, users may print, download, or email articles for
individual use.

También podría gustarte