Content uploaded by Mark W Noordenbos
Author content
All content in this area was uploaded by Mark W Noordenbos on Apr 21, 2016
Content may be subject to copyright.
Scientific Studies of Reading,00:1–20,2015
Copyright © 2015 Society for the Scientific Study of Reading
ISSN: 1088-8438 print/1532-799X online
DOI: 10.1080/10888438.2015.1052455
The Categorical Perception Deficit in Dyslexia:
AMeta-Analysis
Mark W. Noordenbos
Radboud University Nijmegen
Willy Serniclaes
Université Libre de Bruxelles, CNRS & Université Paris Descartes
Speech perception in dyslexia is characterized by a categorical perception (CP) deficit, demonstrated
by weaker discrimination of acoustic differences between phonemic categories in conjunction with
better discrimination of acoustic differences within phonemic categories. We performed a meta-anal-
ysis of studies that examined the reliability of the CP deficit in dyslexia. The results show a reliable
CP deficit in individuals with dyslexia compared to both chronological-age and reading-level con-
trols. The CP deficit is stronger for discrimination than for identification, suggesting that the latter
may only reveal between-category differences that do not fully reflect the CP deficit. The implications
of these findings for the allophonic theory of dyslexia are discussed.
Developmental dyslexia, a specific impairment of written language acquisition (Lyon, Shaywitz,
&Shaywitz,2003), has important implications for education. Many studies have attempted to
determine its nature and origin, producing evidence of many different correlates of dyslexia
(Sprenger-Charolles, Colé, & Serniclaes, 2013). Dyslexia is basically a failure to relate written
symbols to phonological segments, that is, to relate graphemes to phonemes in alphabetic writ-
ing systems. A review of different sources of evidence (behavioral, neural, and genetic) points
to three broad categories of core deficits in dyslexia (Serniclaes & Sprenger-Charolles, 2015):
aphonologicaldeficit,whichisclassicallyattributedtoalesserdegreeofphonemicawareness
(I. Y. Liberman, Shankweiler, Fischer, & Carter, 1974;Melby-Lervåg,Lyster,&Hulme,2012);
agraphemicdeficit,duetoafailuretocombinegraphemesintowordrepresentations(Dehaene,
2014;Vogel,Petersen,&Schlaggar,2014); and a grapho-phonemic deficit, due to a specific fail-
ure to establish grapheme–phoneme correspondences despite the absence of either phonemic or
graphemic impairments (Blomert, 2011).
The study of dyslexia is further complicated by the fact that there are several possible causes
for each of these core deficits. Here we focus on the perceptual determinants of the phonological
deficit in dyslexia. Although this deficit is often ascribed to an underlying deficit in phonemic
awareness, it may instead be caused by a more remote deficit in the perception of phoneme
Correspondence should be sent to Mark W. Noordenbos, Centre for Language Studies, Radboud University Nijmegen,
Nijmegen, P.O. Box 9103, 6500 HD Nijmegen, The Netherlands. E-mail: m.noordenbos@let.ru.nl
Downloaded by [Radboud Universiteit Nijmegen] at 07:14 10 July 2015
2 NOORDENBOS AND SERNICLAES
categories, which can either lead to a deficit in phonemic awareness and thus in turn to sound-
letter matching difficulties, or lead more directly to such difficulties.
THE CATEGORICAL PERCEPTION DEFICIT
Although dyslexia is now thought to be a multifactorial disorder, with a range of associated
behavioral symptoms that cannot be explained by a single cognitive deficit, a number of studies
have demonstrated that at least part of the dyslexic population has a deficit in the “categorical
perception” of phonemes (e.g., Godfrey, Syrdal-Lasky, Millay, & Knox, 1981;Werker&Tees,
1987). Categorical perception ( CP) of phonemes is an important phenomenon wherein segrega-
tion between speech sounds is constrained by their phonemic labels, that is, where only acoustic
differences between phonemic categories can be discriminated, whereas within-category differ-
ences cannot (A. M. Liberman, Harris, Hoffman, & Griffith, 1957). Consequently, a deficit in CP
means that discrimination responses are less constrained by category labels, so that in principle
not only discrimination data but also identification data are needed to evidence a CP deficit, with
the latter indicating whether two different speech sounds belong to different categories. However,
discrimination data are sufficient for stimuli varying along some acoustic continuum, and CP can
be assessed by comparing between-category discrimination (for acoustic contrasts straddling the
phonemic boundary) and within-category discrimination (for acoustic contrasts located within
each phonemic category).
The most common manifestation of the CP deficit in dyslexia is weaker discrimination of
acoustic differences between phonemic categories in conjunction with better discrimination
of acoustic differences within phonemic categories (Serniclaes, Sprenger-Charolles, Carré, &
Démonet, 2001;Werker&Tees,1987; see Figure 1A). Finally, more recent studies have shown
that the enhanced discrimination of acoustic differences within phoneme categories observed in
1 2 3 4 5 6 7
25
50
75
100
Pair
% Correct Discrimination
Controls
Dyslexics
1 2 3 4 5 6 7
25
50
75
100
Pair
% Correct Discrimination
Controls
Dyslexics
AB
FIGURE 1 Schematic representations of the categorical perception (CP)
deficit in dyslexia. Note. The left panel represents the CP deficit, which
is characterized by a smaller between-category discrimination peak,
and increased within-category discrimination. The right panel represents
allophonic perception, which is characterized by a smaller between-
category discrimination peak and enhanced sensitivity to allophonic
features.
Downloaded by [Radboud Universiteit Nijmegen] at 07:14 10 July 2015
CATEGORICAL PERCEPTION DEFICIT IN DYSLEXIA 3
individuals with dyslexia is in fact due to enhanced sensitivity to allophonic features, that is, the
different universal acoustic features that are normally combined to perceive a given phonemic
feature (for a review, see Serniclaes & Sprenger-Charolles, 2015;seeFigure 1B).
THEORETICAL ACCOUNTS OF THE CP DEFICIT
This enhanced sensitivity to allophonic features suggests that the CP deficit in dyslexia might be
due to another mode of speech perception that is characterized by the use of allophones instead
of phonemes (Serniclaes, Van Heghe, Mousty, Carré, & Sprenger-Charolles, 2004). Allophonic
perception is also found in prelinguistic children, but this ability is normally reorganized during
the 1st year of life according to the contrasts present in the ambient language (Hoonhorst et al.,
2009). In addition, typically developing preschool children’s sensitivity to allophonic contrasts
has been shown to partly depend on school experience (Horlyck, Reid, & Burnham, 2012), sug-
gesting that school experience might enhance the use of top-down strategies to focus on relevant
contrasts and ignore irrelevant ones. However, allophonic perception may persist until later in life
in some part of the population: particularly, this reorganization of phonological representations
may not occur to the same extent in individuals with dyslexia for genetic reasons and is prob-
ably one of the causes of dyslexia. In these cases, allophonic perception blurs the relationships
between phonemes and graphemes, even in perfectly transparent alphabetic writing systems, and
is thus highly disruptive of reading acquisition.
Interpretations of the CP deficit other than the allophonic theory have been advanced. Auditory
explanations are mainly based on data evidencing a shallower slope of the phonemic identification
curve for stimuli varying along some continuum (e.g., Vandermosten et al., 2010;seeFigure 2).
1 2 3 4 5 6 7 8
0
25
50
75
100
Stimulus
% Category Labelling
Controls
Dyslexics
A
FIGURE 2 Schematic representation of the identification deficit in
dyslexia, which is characterized by a shallower slope of the identification
function and lower asymptotic floors and ceilings.
Downloaded by [Radboud Universiteit Nijmegen] at 07:14 10 July 2015
4 NOORDENBOS AND SERNICLAES
Ashalloweridentificationslopehasmuchthesamemeaningasaweakerbetween-categorydis-
crimination peak: Both indicate lesser precision in the perception of a phonemic distinction.
However, as just explained, the observed CP deficit in dyslexia is characterized not only by
weaker sensitivity to phonemic contrasts but also by enhanced sensitivity to allophonic contrasts.
Therefore, it remains to be seen whether such differential sensitivity to different contrasts can be
explained solely on auditory grounds.
The CP deficit’s comorbidity with attention problems has also been questioned (Breier
et al., 2001). A CP deficit has been found in children both with and without attention
deficit/hyperactivity disorder (ADHD). This puts the attentional interpretation into question and
suggests that ADHD is unlikely to be a determinant of the CP deficit. Furthermore, correlations
between the CP deficit and reading performance were only found for children without ADHD.
In summary, allophonic perception provides a comprehensive explanation of the CP deficit
and has straightforward implications for learning to read. However, a possible contribution of
auditory factors cannot be entirely dismissed.
RELIABILITY OF THE CP DEFICIT
Despite the potential implications of the CP deficit for the understanding of dyslexia, its reliability
remains in question. Some studies claim to evidence a CP deficit (see above), whereas others con-
clude to the absence of such a deficit (e.g., Blomert & Mitterer, 2004;Hazan,Messaoud-Galusi,
Rosen, Nouwens, & Shakespeare, 2009;Messaoud-Galusi,Hazan,&Rosen,2011); the issue
thus remains uncertain. A problem for establishing the reliability of the CP deficit in dyslexia is
that most studies claiming to investigate the CP deficit have only collected either identification
or discrimination data. As just mentioned, it is possible to estimate a CP deficit with discrim-
ination data. However, doing so with identification data only is less straightforward. Typical
differences in the identification of speech sounds varying along some acoustic continuum are
presented in Figure 2. When compared to controls, individuals with dyslexia usually display
both a shallower slope around the phonemic boundary and less extreme asymptotic floors and
ceilings at the continuum endpoints. A shallower slope indicates less precise phonemic cate-
gorisation, a difference that is comparable to the lower between-category peak evidenced with
discrimination data. However, the difference in asymptotic floors and ceilings is more difficult to
interpret in a CP framework. Such asymptotic differences might simply be another manifestation
of lesser categorical precision, which thus bear no obvious relationship to enhanced sensivity to
within-category differences. However, it is also possible that the difference in asymptotic floors
and ceilings might somehow be due to the conflicting use of phonemic labels when reporting
within-category differences. In short, it is clear that identification data can reveal differences
in the perception of between-category differences, but they might not reveal differences in the
perception of within-category differences, which is also needed to demonstrate a CP deficit.
Finally, it is worth mentioning that if an identification function has weaker asymptotic floors
and ceilings at the endpoints of some acoustic continuum, this is roughly equivalent to a deficit in
the identification of pairs of stimuli differing in a single phonological feature (i.e., minimal pairs
for voicing, place-of-articulation, etc.). However, data collected with minimal pairs can reveal
only differences that are roughly equivalent to those present at the continuum endpoints, not
those that might also be present around the phonological boundary.
Downloaded by [Radboud Universiteit Nijmegen] at 07:14 10 July 2015
CATEGORICAL PERCEPTION DEFICIT IN DYSLEXIA 5
THE PRESENT STUDY
In the present meta-analysis, we examine the general reliability of the CP deficit in studies that
compared individuals with dyslexia with both chronological-age (CA) and reading-level (RL)
controls, and investigate whether the CP deficit depends on the type of task used. We also compare
the reliability of the CP deficit to that of classical indicators of dyslexia such as pseudoword read-
ing (e.g., van IJzendoorn & Bus, 1994), pseudoword repetition (Melby-Lervåg & Lervåg, 2012),
and the phonological short-term memory deficit (Swanson, Zheng, & Jerman, 2009). As previ-
ously explained, the CP deficit is more readily evidenced with a discrimination task than with
an identification task because the latter may reveal only between-category differences that do not
fully reflect the CP deficit. Therefore, we hypothesized that effect sizes based on discrimination
tasks would be larger than effect sizes based on identification tasks.
METHODS
This meta-analysis was conducted in accordance with the MOOSE guidelines for the meta-
analysis of observational studies in epidemiology (Stroup et al., 2000).
Data Collection, Inclusion, and Extraction
The data collection and inclusion process is summarized in Figure 3.Studieswereidentified
by searching electronic databases (Medline, PubMed, PsycINFO, Google Scholar) up to and
including 2014, using the following keywords: categorical perception and speech perception,
combined with terms related to dyslexia (dyslexia, reading disorder, specific reading disorder,
poor readers, poor reading, reading impairment and poor decoders). Our search was limited to
original research published as a full article in English, that is, not a review, abstract, editorial,
letter, or commentary. We also manually searched the reference lists of published studies in order
to locate other relevant studies. The authors independently screened the retrieved studies on the
basis of titles and abstracts, and where necessary the full article, to determine suitability for
inclusion or exclusion.
Studies were eligible for inclusion in the meta-analysis if they assessed CP in individuals
diagnosed with dyslexia for speech stimuli along a continuum using identification and/or dis-
crimination tasks. Studies that reported data based on minimal pairs were excluded, as there was
no stimulus continuum making the minimal pairs equivalent to continuum endpoints, making only
rough estimation of the CP deficit possible. Studies that reported data based on nonspeech stim-
uli or speech-in-noise perception were excluded. Multiple studies published by the same author
were carefully examined in order to prevent the inclusion of data from the same sample more
than once.
The authors independently extracted the data using an electronic form. The intercoder reliabil-
ity for the extracted data related to identification and discrimination ranged from 95% to 100%.
Any disagreements were resolved by discussion or consulting the original paper.
Downloaded by [Radboud Universiteit Nijmegen] at 07:14 10 July 2015
6 NOORDENBOS AND SERNICLAES
FIGURE 3 Flow diagram for the study selection process.
Data Synthesis
As not all studies reported both identification and discrimination data, effect sizes were calculated
from either identification or discrimination results, or both. Effect sizes were mainly estimated
Downloaded by [Radboud Universiteit Nijmegen] at 07:14 10 July 2015
CATEGORICAL PERCEPTION DEFICIT IN DYSLEXIA 7
from reported F-ort-test values. When statistical information was not reported, but mean and
standard deviations were available for each group, we used these values to estimate the effect size.
For identification data, effect sizes were calculated from group effects for slope when av ail-
able. Results based directly on identification scores were avoided as far as possible, because
Stimulus × Group interactions can be significant in an analysis of variance even with two strictly
parallel identification curves with different 50% crossover points as a result of the S-shaped form
of the curves. Nevertheless, for studies that did not report group effects for slope (or did not
report exact test values for group effects for slope) we based the effect size on Stimulus × Group
interactions (this was the case in four studies; see also Supplementary Table 1). This approach
was appropriate, because the 50% crossover points of the identification curves were fairly similar
between the groups in these studies (based on visual inspection of the curves). For discrimination
data, effect sizes were calculated from Between-/Within-Category Difference × Group interac-
tions whenever available, and otherwise on Stimulus Pair × Group interactions. When possible
we used test statistics based on observed discrimination scores, not those based on the differ-
ence between observed discrimination scores and those predicted from identification data (see
Supplementary Table 1). The precise references of the data that were selected to calculate the
effect sizes of the included studies are reported in Supplementary Table 1.
We made an a priori decision to include all reported speech continua, in order to draw conclu-
sions about the CP deficit in general rather than for a limited or selected set of stimulus continua.
The lexical tone continua of the studies under scope were characterized by syllables differing in
fundamental frequency (F0) patterns (a continuum of Cantonese /si/ syllables varying between
ahigh-levelF0andamid-levelF0:Cheungetal.,2009;acontinuumofMandarin/pa/ syllables
varying between a high rising F0 and a falling F0: Zhang et al., 2012)makingthemcompa-
rable to consonant and vowel features, such as voicing and place-of-articulation, which also
have acoustic correlates extending over several phonemic segments (Mattock, Molnar, Polka,
&Burnham,2008). If data were collected on several stimulus continua within the same study,
asingleweightedmeaneffectsizewascalculatedforthedifferentcontinuainordertoavoid
assigning more weight to studies with more than one stimulus continuum.
For studies reporting statistical tests that were not significant (e.g., F < 1) and where means
and standard deviations, p values, or any other measures needed to estimate an effect size were
not available, we contacted the authors to request further information. Of the five authors we
contacted, three responded and one provided numerical data to calculate the effect size. If the
information was still not available, we conservatively assumed an F value of 0.5 (halfway
between 0 and 1; this assumption was made for four studies). All analyses were repeated
excluding the assumed F values and yielded essentially identical results.
Meta-Analytic Methods
For each outcome measure, effect sizes were computed using Cohen’s d (Cohen, 1988)andcor-
rected for bias from small sample sizes (Hedges, 1981). Cohen’s d is calculated as the difference
between the means of the control and dyslexic groups divided by their pooled standard deviation
(d = [M
ctl
–M
dys
]/σ ,whereσ
2
= within-group variance). A positive d indicates that the control
group had a higher group mean on average, whereas a negative d indicates a group difference in
favor of individuals with dyslexia. According to Cohen’s classification, values between 0.2 and
0.5 can be interpreted as small effects, values between 0.5 and 0.8 as moderate effects, and values
greater than 0.8 as large effects.
Downloaded by [Radboud Universiteit Nijmegen] at 07:14 10 July 2015
TA B L E 1
Characteristics of Studies Comparing Individuals With Dyslexia and Controls on Categorical Perception for Identification and Discrimination Tasks
Sample Characteristics Effect Size (d)
Native Age (Years) N Discrimination Identification Continua
Study Language Dys CA RL Dys CA RL CA RL CA RL Place Voicing Tones
Godfrey et al. (1981) English 10.4 10.4 17 17 0.78 0.61 ba-da
0.78 0.63 da-ga
Werker and Tees (1987) English 10.3 10.5 14 14 1.14 0.52 ba-da
De Weirdt (1988) Dutch 6.6 6.4 18 18 0.63 p@-t@
Reed (1989) English 9.0 8.9 23 23 0.49 0.62 ba-da
Steffens, Eilers, Gross-Glenn, and
Jallad (1992)
English 39.6 33.3 18 18 0.24 0.71 sta-sa
0.63 0.61 a-e
0.24 0.66 b-d
Manis et al. (1997) English 13.3 12.0 8.5 25 25 24 0.64 0.20 bath-path
Adlard and Hazan (1998) English 10.3 10.1 8.2 13 12 12 0.51 0.28 sue-zoo
0.28 0.28 date-gate
de Gelder and Vroomen (1998)Dutch11.411.48.1141414 0.74 0.74 ba-da
Nittrouer (1999) English 9.0 9.0 17 93 0.43 da-ta
0.30 sei-stei
0.74 sa-sha
0.94 su-shu
Joanisse et al. (2000) English 8.6 8.4 6.9 61 52 37 0.21 0.17 spy-sky
0.15 0.09 dug-tug
Breier et al. (2001) English 10.5 10.6 47 48 0.77 ga-ka
Chiappe et al. (2001) English 7.0 6.8 26 36 1.86 bif-pif
1.34 bis-pis
Maassen, Groenen, Crul,
Assman-Hulsmans, and Gabreëls
(2001)
Dutch 8.8 8.8 7.0 8 12 8 1.03 1.13 0.16 0.16 bak-dak
1.07 1.17 1.01 1.11 bak-pak
Rosen and Manganari (2001)English13.013.0 88 0.69 ba-da
0.77 ab-ad
Serniclaes et al. (2001) French 12.9 13.0 19 17 0.57 ba-da
8
Downloaded by [Radboud Universiteit Nijmegen] at 07:14 10 July 2015
Blomert and Mitterer (2004) Dutch 9.0 9.3 10 12 0.88 ba-da
0.54 tart-kart
Blomert, Mitterer, and Paffen (2004) Dutch 9.1 8.5 14 18 0.38 tart-kart
Chiappe, Chiappe, and Gottardo
(2004)
English 7.8 7.6 13 49 0.66 bath-path
Serniclaes et al. (2004) French 9.1 9.2 18 23 0.65 ba-pa/ga-ka
a
van Beinum, Schwippert, Been, van
Leeuwen, and Kuijpers (2005)
Dutch 25.0 25.0 12 12 1.55 1.55 bak-dak
Boada and Pennington (2006)English12.312.18.4202020 0.92 0.94 sa-sha/su-shu
White et al. (2006) English 10.5 10.3 23 22 0.00 ba-da
0.40 coat-goat
Veu i l l et , M a gn a n , E c al l e , Th a i- Va n ,
and Collet (2007)
French 10.9 10.8 23 23 0.36 ba-pa
Bogliotti, Serniclaes,
Messaoud-Galusi, and
Sprenger-Charolles (2008)
French 9.6 9.8 7.6 10 11 10 0.93 0.69 1.12 0.26 do-to
Cheung et al. (2009) Chinese 10.4 10.4 8.9 24 25 24 0.65 0.02 si
55
-si
33
30 30 30 0.73 0.01 gi
55
-ki
55
Hazan et al. (2009) English 22.8 23.4 17 20 0.70 0.46 pea-bee
Liu, Shu, and Yang (2009) Chinese 10.4 10.2 25 25 0.45 pa-p
h
a
Robertson, Joanisse, Desroches, and
Ng (2009)
English 10.5 9.7 8.0 14 14 14 0.77 0.00 0.12 0.51 ball-doll
Van d e r mo s t e n e t a l . (2010) Dutch 21.4 21.5 31 31 1.04 bak-dak
0.18 u-y
Boets et al. (2011) Dutch 5.3 5.3 16 46 0.78 bak-dak
Johnson, Pennington, Lowenstein,
and Nittrouer (2011)
English 10.9 11.2 16 16 0.29 da-ta
0.78 sue-shoe
Messaoud-Galusi et al. (2011)English11.310.56251 2.03 0.64 bee-pea
Van d e r mo s t e n e t a l . (2011) Dutch 11.6 11.7 13 25 1.32 bak-dak
0.62 u-y
Baart, de Boer-Schellekens, and
Vroomen (2012)
Dutch 20.0 20.0 22 22 0.66 aba-ada
Zhang et al. (2012) Chinese 10.3 10.3 18 18 1.64 pa
2
-pa
4
Noordenbos et al. (2013) Dutch 20.9 21.4 19 19 0.27 0.68 b@-d@
Note. Dys = individuals with dyslexia; CA = chronological-age controls; RL = reading-level controls.
a
Results of continua were combined as no differences between continua were observed irrespective of the group.
9
Downloaded by [Radboud Universiteit Nijmegen] at 07:14 10 July 2015
10 NOORDENBOS AND SERNICLAES
Ameta-analysiswasperformedtoobtainaglobalassessmentoftheCPdeficitbycalculating
an inverse variance-weighted overall effect size. A normal distribution test was used to test the
null hypothesis of a zero mean effect size, as well as differences in overall effect size between
identification and discrimination tasks. As it is unlikely that any variables which might have an
effect on the outcome of a study are identical across all of the studies included in the analysis, a
random effects model, which takes both within-study and between-study variability into account,
was used to calculate the overall effect sizes with corresponding 95% confidence intervals (CIs).
Between-study variance was estimated using a restricted maximum likelihood (REML) model.
Amajoradvantageofarandomeffectsmodelisthattheresultsarenotrestrictedtotheidenti-
fied population but can be generalized to the whole population (Borenstein, Hedges, Higgins, &
Rothstein, 2009).
Heterogeneity—that is, whether variation in effect sizes between studies was greater than
would be expected by chance alone—was estimated using Cochran’s Q and quantified with the
Higgins I
2
statistic (Higgins & Thompson, 2002). The latter describes the percentage of true
heterogeneity across the sample of studies (Higgins, Thompson, Deeks, & Altman, 2003). Note
that I
2
is not directly affected by the number of studies in the analysis and that values of 25%,
50%, and 75% are considered as low, moderate, and high levels of heterogeneity, respectively.
To further explore heterogeneity, sensitivity analyses were performed to evaluate the stability of
the overall effect size and to determine the impact of possible outliers (Viechtbauer & Cheung,
2010). All statistical analyses were carried out in the statistical software R (version 3.1.2) using
the metafor package (Viechtbauer, 2010). All reported p values were two-sided, and p < .05 was
considered statistically significant.
Publication Bias
Potential publication bias was examined using a funnel plot (Egger, Smith, Schneider, & Minder,
1997). A funnel plot is obtained by plotting a sample size-related measure (y-axis) as a function of
effect size (x-axis). Publication bias is indicated by an imbalance of variance between small and
large sample results. Typically, effect sizes are more variable for small versus large samples, but
mean effect sizes should not be larger with larger samples if there is no selection bias. If there is
no publication bias, the scatter plot should be symmetrical, resembling an inverse funnel. When
multiplied by some precision coefficient (i.e., a value inversely related to sample size), effect
sizes for very small samples should tend toward zero, because small samples have almost zero
precision. This implies that the regression line between the Effect Size × Precision interaction
and Precision should run through the origin. The intercept of that regression line provides a test
of asymmetry (Egger et al., 1997): the larger the deviation from zero, the higher the proportion
of small sample results with small effect sizes that are missing from the survey.
RESULTS
Characteristics of the Included Studies
Of the 632 studies retrieved by the literature search, 36 studies were found to meet the inclusion
criteria. The main reasons for exclusion were the use of minimal pairs rather than a stimulus
Downloaded by [Radboud Universiteit Nijmegen] at 07:14 10 July 2015
CATEGORICAL PERCEPTION DEFICIT IN DYSLEXIA 11
continuum, or participants who were not (yet) diagnosed with dyslexia (e.g., at-risk children).
The seminal study of Brandt and Rosen (1980)wasnotincludedinthemeta-analysisbecause
it did not contain appropriate results to calculate an effect size. We contacted the authors for
further information, but records of this study were no longer available (J. Brandt, personal com-
munication, February 14, 2015). The included studies were mainly published in journals, most
frequently in the Journal of Experimental Child Psychology (10 publications out of 36); fol-
lowed by Journal of Speech, Language, and Hearing Research (7 publications); Research in
Developmental Disabilities; Brain and Language; Journal of Child Psychology and Psychiatry;
and Developmental Science (2 publications each). The remainder (11 publications) were pub-
lished in 11 different journals. Half of the studies (18 publications) were conducted with English-
speaking participants. The remainder were conducted with Dutch-speaking (11 publications),
French-speaking (4 publications), and Chinese-speaking (3 publications) participants.
Effect sizes as a function of stimulus characteristics (continua) and group characteristics (age
and language) are listed in Table 1.The36relevantstudiesincluded754individualswithdyslexia
(M = 20.94, SD = 12.42, range = 8–62) and 914 CA controls (M = 25.39, SD = 16.73,
range = 8–93). Most of these studies ensured that the individuals with dyslexia had average or
above-average nonverbal intelligence scores, showed difficulties on standardized reading and/or
spelling tasks and had no other neurological or psychological disorders. Three studies included
participants regardless of any comorbid ADHD diagnosis (Boada & Pennington, 2006;Breier
et al., 2001; White et al., 2006). Nine of the 36 studies also included RL controls, a total of
169 individuals (M = 18.78, SD = 9.82, range = 8–37).
Most studies (20 of 36) used only one stimulus continuum. Place-of-articulation continua were
more frequently used (62%) than voicing continua (35%), and two studies (4%) used a lexical
tone continuum. A large majority of the studies (34 of 36) collected identification data, about
one third of the studies (13 of 36) collected discrimination data, and only 11 of the 36 studies
collected both identification and discrimination data.
Comparisons of Individuals With Dyslexia and Chronological-Age Controls
As expected, CP was systematically weaker in individuals with dyslexia compared to CA con-
trols. For identification data, the overall mean effect size was d = 0.66, 95% CI [0.54, 0.77],
and was significantly different from zero (z = 10.89, p < .001). Heterogeneity was s mall and
not significant, Q(33) = 43.78, p = .10, I
2
= 27%, indicating no statistical evidence for differ-
ences between the studies. In the sensitivity analyses, two studies were identified as influential
(Chiappe, Chiappe, & Siegel, 2001;Joanisse,Manis,Keating,&Seidenberg,2000). The study by
Joanisse et al. (2000)hadthelargestweightintheanalysis,whereasthestudybyChiappeetal.
(2001)hadoneofthelargesteffectsizesintheanalysis.Weexaminedthesestudiestodetermine
whether they had any characteristics that set them apart from the other studies besides extreme
values, but no other complicating factors were clearly evident. Excluding these two studies did
not significantly change the overall mean effect size (p = .89).
For discrimination data, the overall mean effect size was d = 0.86, 95% CI [0.56, 1.16], and
was also significantly dif ferent from zero (z = 5.68, p < .001). Heterogeneity was moderate
and significant, Q(12) = 40.29, p < .001, I
2
= 65%. In the sensitivity analyses, one study was
identified as influential with respect to between-study variance (Messaoud-Galusi et al., 2011).
We examined this study to determine whether any characteristics set this study apart from the
Downloaded by [Radboud Universiteit Nijmegen] at 07:14 10 July 2015
12 NOORDENBOS AND SERNICLAES
other studies, but besides a large sample size (> 50 participants per group) no other factors were
clearly evident. Excluding this study did not significantly change the overall mean effect size
(p = .33), but the study had an effect on between-study variance, as between-study variance
among the remaining 12 studies was small, and did not significantly differ from chance. Effect
sizes with CIs for each study are displayed in Figures 4 and 5 for identification and discrimination
data, respectively.
To test whether the extent of evidence for a CP deficit depended on the type of task used, we
compared the effect sizes based on discrimination tasks with the effect sizes based on identifica-
tion tasks. The difference in overall mean effect size between identification and discrimination
data was significant in both an independent group comparison including all 36 studies (z = 2.52,
p = .01) and in a pairwise comparison including only the 11 studies in which both identification
and discrimination data were available (d = 0.64 vs. 0.97; z = 2.37, p = .02), with a larger effect
size for discrimination data.
To check for potential publication bias, we used funnel plots in which identification or dis-
crimination effect sizes were plotted on the x-axis and the weight of each study was plotted on
the y-axis. The resulting funnel plots are shown in Supplementary Figure 1. Examination of the
funnel plots revealed no indication of publication bias (i.e., no decrease in either identification
or discrimination effect size variability at higher weights). Egger’s asymmetry test (Egger et al.,
1997)fortheidentificationordiscriminationeffectsizesshowedthattheinterceptoftheregres-
sion lines between the Effect Size × Precision interaction and Precision did not significantly
differ from zero in either case (both p > .05). According to Egger’s asymmetry tests, there seems
to have been no publication bias in the studies within the scope of the present meta-analysis.
However, a closer visual inspection of the funnel plot with discrimination data reveals a slight
asymmetry, suggesting potential publication bias.
Comparisons of Individuals With Dyslexia and Reading-Level Controls
The CP of individuals with dyslexia was weaker than that of RL controls in all nine studies
examined here. The overall mean effect size for differences in identification between individuals
with dyslexia and RL controls was d = 0.32, 95% CI [0.10, 0.53], and was significantly different
from zero (z = 2.91, p < .01). Heterogeneity was small and not significant, Q(8) = 8.36, p = .40,
I
2
= 14%, indicating no statistical evidence for differences between the studies. In the sensitivity
analyses, one study was identified as an influential study (Boada & Pennington, 2006). Other
than the fact that this study had the largest effect size in the analysis, no other factors were clearly
evident, which set this study apart from the other studies in the analysis. Excluding this study did
not significantly change the overall mean effect size (p = .58). The number of studies reporting
discrimination results with RL controls was too small (only three) to draw any conclusions.
DISCUSSION
The results of the 36 studies that were eligible for the present meta-analysis give a fairly clear
picture of the CP deficit in dyslexia. The results of the meta-analysis show that the CP deficit is
indeed reliable overall, with large and significant effects for both identification and discrimination
tasks. Although no significant publication bias was identified statistically, visual inspection of the
discrimination data suggests possible publication bias.
Downloaded by [Radboud Universiteit Nijmegen] at 07:14 10 July 2015
CATEGORICAL PERCEPTION DEFICIT IN DYSLEXIA 13
FIGURE 4 Forest plot for group differences in categorical perception
between individuals with dyslexia and chronological-age controls, based
on identification data. Note. The effect sizes of the individual studies are
indicated by squares and the 95% confidence intervals by horizontal lines.
The size of each square is proportional to the weight of the corresponding
study. The diamond indicates the overall mean effect size, and the width
of the diamond represents the overall confidence interval.
Downloaded by [Radboud Universiteit Nijmegen] at 07:14 10 July 2015
14 NOORDENBOS AND SERNICLAES
FIGURE 5 Forest plot for group differences in categorical perception
between individuals with dyslexia and chronological-age controls, based
on discrimination data. Note. The effect sizes of the individual studies are
indicated by squares and the 95% confidence intervals by horizontal lines.
The size of each square is proportional to the weight of the corresponding
study. The diamond indicates the overall mean effect size and the width of
the diamond represents the overall confidence interval.
In comparisons between individuals with dyslexia and CA controls, the CP deficit was highly
significant for both tasks, identification and discrimination, although effect sizes were signifi-
cantly larger for the discrimination task. The overall mean effect size for the discrimination task
was d = 0.86 (13 studies, more than 250 participants per group), which is a large effect size
according to Cohen’s (1988) classification. The overall mean effect size for the identification
task was d = 0.66 (34 studies, more than 700 participants per group), an effect size that falls
between moderate (0.50) and large (0.80) in Cohen’s classification.
In comparisons of individuals with dyslexia to RL controls—that is, to younger individuals
of the same reading age—a significant difference in overall mean effect size was also found.
However, only the effect sizes for identification could be estimated here, an analysis that yielded
a d value of only 0.32 (nine studies, about 170 participants per group), which falls between small
(0.20) and moderate (0.50) in Cohen’s classification.
Comparing the CP Deficit With Other Deficits in Dyslexia
There is ample evidence that the majority of individuals with dyslexia have a phonological deficit
and that pseudoword reading is one of the best indicators of this phonological deficit. Therefore,
pseudoword reading can be used as a benchmark to assess the reliability of the CP deficit. Given
the importance of reading development, most studies on pseudoword reading have used a design
Downloaded by [Radboud Universiteit Nijmegen] at 07:14 10 July 2015
CATEGORICAL PERCEPTION DEFICIT IN DYSLEXIA 15
in which individuals with dyslexia are compared to RL controls rather than CA controls, to ensure
that observed group differences could not be explained by reading level differences between
groups. Meta-analytic studies comparing individuals with dyslexia to RL controls have shown
that the mean effect size for the deficit in pseudoword reading was about d = 0.48 (van IJzendoorn
&Bus,1994), or, in a more recent meta-analysis, d = 0.65 (Herrmann, Matyas, & Pratt, 2006).
For similar comparisons with RL controls, the mean effect size for the CP deficit (d = 0.32) was a
bit smaller than this pseudoword reading deficit. However, the latter effect size is based on identi-
fication data and is probably much smaller than what would be obtained with discrimination data.
Recall that the number of comparisons between individuals with dyslexia and RL controls with
discrimination data within the scope of the present meta-analysis was too small (three studies) to
allow reliable assessment. But based on the results obtained with CA controls, the discrimination
effect size should be much larger than the identification effect size (d = 0.86 vs. 0.66). An esti-
mate of the effect size of the CP deficit based on discrimination data would therefore probably
reach much the same level as in pseudoword reading. This is not to say that pseudoword reading
and CP stand in the same relationship to dyslexia. It may be that the pseudoword reading deficit
is related to different core deficits in dyslexia, whereas the CP deficit is specifically related to
the phonological deficit. The fact that, despite such limitations, the effect size of the CP deficit is
comparable to that of the deficit in pseudoword reading is impressive.
Another meta-analytic study investigated the deficit in pseudoword repetition that is also asso-
ciated with dyslexia (Melby-Lervåg & Lervåg, 2012). This study showed a mean effect size
of d = 0.92 in comparisons of pseudoword repetition performance between individuals with
dyslexia and CA controls, which is close to the mean effect size obtained here for the CP deficit
with discrimination data (d = 0.86). Also, the mean effect size in pseudoword repetition was
d = 0.36 for comparisons between individuals with dyslexia and RL controls, a value not far
above the one obtained here for the CP deficit in nonoptimal conditions (with identification data:
d = 0.32).
Other meta-analyses have investigated possible determinants of dyslexia and evidenced large
effect sizes for phonological skills, well above those obtained here for the CP deficit (d = 1.37 for
CA controls, d = 0.57 for RL controls in Melby-Lervåg et al., 2012; d = 1.20 for CA controls,
d = 0.52 for RL controls in Melby-Lervåg & Lervåg, 2012). However, the effect sizes found here
for the CP deficit are similar to those obtained elsewhere for oral language skills (d = 0.82 for
CA controls, d = 0.32 for RL controls in Melby-Lervåg & Lervåg, 2012)andforphonological
short-term memory (d = 0.83 for CA controls in Swanson et al., 2009). Apart from phonemic
awareness, which is assessed with tasks that are very close to the processes involved in reading,
the CP deficit is certainly not less reliable than the other possible determinants of dyslexia. This
is a noteworthy result, recalling that there are different possible sources of dyslexia, and that
therefore no deficit can be completely reliable.
Explaining the Variability of the CP Deficit
Although the overall CP deficit in dyslexia is reliable, the magnitude of this deficit is variable
across studies. Studies comparing individuals with dyslexia to typically developing readers dif-
fer in sample characteristics, experimental design, language, and selection criteria, among other
factors. Because it is unlikely that all factors which may account for v ariation in effect sizes of
the included studies are identical, we used a random effects model, which also takes into account
Downloaded by [Radboud Universiteit Nijmegen] at 07:14 10 July 2015
16 NOORDENBOS AND SERNICLAES
the factors that differ between studies. The advantage of this approach is that the results allow for
generalization beyond the particular set of studies included in the meta-analysis.
Our meta-analysis focused on the extent to which the CP deficit depends on the type of task
used. The mean effect size of the CP deficit was significantly larger for discrimination data
than for identification data (d = 0.86 vs. 0.66). As explained in the Introduction, identification
responses might be adequate only to establish sensitivity to differences between categories but
not to those within phonemic categories. As both (decreased) sensitivity to between-category dif-
ferences and (increased) sensitivity to within-category differences contribute to the CP deficit in
dyslexia, this might explain why a weaker CP deficit was obtained here with identification data.
Implications of the CP Deficit for Explaining Dyslexia
Concerning the causal role of the CP deficit in dyslexia, there is no doubt that reading instruction
exerts a strong influence on CP, and more specifically on the perception of allophonic contrasts.
This raises an important issue, because it bears on the direction of the causal relationship between
CP and reading difficulties. A study with Australian children suggested that the perception of
allophonic contrasts was negatively related to school experience but had no relationship with
age (Horlyck et al., 2012). In this study, a group of 5-year-old children with 6 months of school
experience were found to be less sensitive to an allophonic contrast than children of the same
age without school experience. Although the children in this Australian study were not at risk
for dyslexia, they exhibited some allophonic sensitivity, the degree of which was modulated by
school experience. The effect of school experience on allophonic sensitivity was confirmed in a
longitudinal study with Dutch children at risk for dyslexia. These children exhibited a strong CP
deficit when they were in kindergarten, characterized by both a weaker phonemic discrimination
peak and a stronger allophonic peak; but this CP deficit was completely absent when they were
in the first grade, after 6 months of formal reading instruction (Noordenbos, Segers, Serniclaes,
Mitterer, & Verhoeven, 2012a). These findings might suggest that allophonic perception is not
entirely specific to dyslexia and that it decreases with reading experience. However, these conclu-
sions have been called into question by neurophysiological data, which point to the persistence of
allophonic percepts in dyslexia, irrespective of reading experience. Even though allophonic sensi-
tivity was no longer apparent in the behavioral responses of the Dutch children at risk f or dyslexia
in first grade, it was still present in neurophysiological recordings performed on the same chil-
dren (Noordenbos, Segers, Serniclaes, Mitterer, & Verhoeven, 2012b). Similar results have been
obtained for adults with dyslexia (Noordenbos, Segers, Serniclaes, & Verhoeven, 2013). These
apparently conflicting results might be explained by differences in sensitivity between behav-
ioral and neural measures. Neural measures accurately reflect the temporal development of the
underlying processes, whereas overt behavioral responses only reflect the endpoint of process-
ing and may abstract away from subtle underlying processing deficits. In summary, although the
behavioral manifestations of allophonic perception and the related CP deficit may disappear with
reading experience, this sensitivity may still be present in the form of neural activation.
CONCLUSIONS
The results of this meta-analysis demonstrate that the finding of a CP deficit in individuals with
dyslexia is highly reliable, although the magnitude of the effect is quite variable across studies.
Downloaded by [Radboud Universiteit Nijmegen] at 07:14 10 July 2015
CATEGORICAL PERCEPTION DEFICIT IN DYSLEXIA 17
Part of this variability is due to random variations between groups, but it is also due to the type
of task used, the CP deficit being more easily evidenced with a discrimination task than with an
identification task.
SUPPLEMENTAL MATERIAL
Supplemental data for this article can be accessed at www.tandfonline.com/hssr.
ACKNOWLEDGMENTS
We thank Sunjing Ji for providing assistance on tone perception in Chinese. Many thanks to Paul
Reeve for proofreading the manuscript.
FUNDING
The research reported in this meta-analysis was supported by a public grant overseen by the
French National Research Agency (ANR) as part of the “Investissements d’Avenir” program
(reference: ANR-10-LABX-0083) to Willy Serniclaes.
REFERENCES
References marked with an asterisk indicate studies included in the meta-analysis.
∗
Adlard, A., & Hazan, V. (1998). Speech perception in children with specific reading difficulties (dyslexia). The Quarterly
Journal of Experimental Psychology Section A, 51,153–177.doi:10.1080/713755750
∗
Baart, M., de Boer-Schellekens, L., & Vroomen, J. (2012). Lipread-induced phonetic recalibration in dyslexia. Acta
Psychologica, 140, 91–95. doi:10.1016/j.actpsy.2012.03.003
Blomert, L. (2011). The neural signature of orthographic–phonological binding in successful and failing reading
development. Neuroimage, 57,695–703.doi:10.1016/j.neuroimage.2010.11.003
∗
Blomert, L., & Mitterer, H. (2004). The fragile nature of the speech-perception deficit in dyslexia: Natural vs synthetic
speech. Brain and Language, 89, 21–26. doi:10.1016/S0093-934X(03)00305-5
∗
Blomert, L., Mitterer, H., & Paffen, C. (2004). In search of the auditory, phonetic, and/or phonological problems in
dyslexia: Context effects in speech perception. Journal of Speech, Language, and Hearing Research, 47, 1030–1047.
doi:10.1044/1092-4388(2004/077)
∗
Boada, R., & Pennington, B. (2006). Deficient implicit phonological representations in children with dyslexia. Journal
of Experimental Child Psychology, 95,153–193.doi:10.1016/j.jecp.2006.04.003
∗
Boets, B., Vandermosten, M., Poelmans, H., Luts, H., Wouters, J., & Ghesquière, P. (2011). Preschool impairments
in auditory processing and speech perception uniquely predict future reading problems. Research in Developmental
Disabilities, 32, 560–570. doi:10.1016/j.ridd.2010.12.020
∗
Bogliotti, C., Serniclaes, W., Messaoud-Galusi, S., & Sprenger-Charolles, L. (2008). Discrimination of speech sounds
by children with dyslexia: Comparisons with chronological age and reading level controls. Journal of Experimental
Child Psychology, 101, 137–155. doi:10.1016/j.jecp.2008.03.006
Borenstein, M., Hedges, L. V., Higgins, J. P. T., & Rothstein, H. R. (2009). Introduction to meta-analysis.Chichester,
UK: Wiley.
Brandt, J., & Rosen, J. J. (1980). Auditory phonemic perception in dyslexia: Categorical identification and discrimination
of stop consonants. Brain and Language, 9,324–337.doi:10.1016/0093-934x(80)90152-2
Downloaded by [Radboud Universiteit Nijmegen] at 07:14 10 July 2015
18 NOORDENBOS AND SERNICLAES
∗
Breier, J. I., Gray, L., Fletcher, J. M., Diehl, R. L., Klaas, P., Foorman, B. R., & Molis, M. R. (2001). Perception of
voice and tone onset time continua in children with dyslexia with and without attention deficit/hyperactivity disorder.
Journal of Experimental Child Psychology, 80, 245–270. doi:10.1006/jecp.2001.2630
∗
Cheung, H., Chung, K. K. H., Wong, S. W. L., McBride-Chang, C., Penney, T. B., & Ho, C. S. H. (2009). Perception
of tone and aspiration contrasts in Chinese children with dyslexia. Journal of Child Psychology and Psychiatry, 50,
726–733. doi:10.1111/j.1469-7610.2008.02001.x
∗
Chiappe, P., Chiappe, D. L., & Gottardo, A. (2004). Vocabulary, context, and speech perception among good and poor
readers. Educational Psychology, 24(6), 825–843. doi:10.1080/0144341042000271755
∗
Chiappe, P., Chiappe, D. L., & Siegel, L. S. (2001). Speech perception, lexicality, and reading skill. Journal of
Experimental Child Psychology, 80, 58–74. doi:10.1006/jecp.2000.2624
Cohen, J. (1988). Statistical power analysis for the behavioral sciences.Hillsdale,NJ:Erlbaum.
∗
de Gelder, B., & Vroomen, J. (1998). Impaired speech perception in poor readers: Evidence from hearing and speech
reading. Brain and Language, 64,269–281.doi:10.1006/brln.1998.1973
∗
De Weirdt, W. (1988). Speech perception and frequency discrimination in good and poor readers. Applied
Psycholinguistics, 9, 163–183. doi:10.1017/S0142716400006792
Dehaene, S. (2014). Reading in the brain revised and extended: Response to comments. Mind & Language, 29, 320–335.
doi:10.1111/mila.12053
Egger, M., Smith, G. D., Schneider, M., & Minder, C. (1997). Bias in meta-analysis detected by a simple, graphical test.
BMJ: British Medical Journal, 315, 629–634. doi:10.1136/bmj.315.7109.629
∗
Godfrey, J. J., Syrdal-Lasky, K., Millay, K. K., & Knox, C. M. (1981). Performance of dyslexic children on speech
perception tests. Journal of Experimental Child Psychology, 32,401–424.doi:10.1016/0022-0965(81)90105-3
∗
Hazan, V., Messaoud-Galusi, S., Rosen, S., Nouwens, S., & Shakespeare, B. (2009). Speech perception abilities of
adults with dyslexia: Is there any evidence for a true deficit? Journal of Speech, Language, and Hearing Research,
52, 1510–1529. doi:10.1044/1092-4388(2009/08-0220)
Hedges, L. V. (1981). Distribution theory for Glass’s estimator of effect size and related estimators. Journal of Educational
and Behavioral Statistics, 6, 107–128. doi:10.3102/10769986006002107
Herrmann, J. A., Matyas, T., & Pratt, C. (2006). Meta-analysis of the nonword reading deficit in specific reading disorder.
Dyslexia, 12, 195–221. doi:10.1002/dys.324
Higgins, J. P. T., & Thompson, S. G. (2002). Quantifying heterogeneity in a meta-analysis. Statistics in Medicine, 21,
1539–1558. doi:10.1002/sim.1186
Higgins, J. P. T., Thompson, S. G., Deeks, J. J., & Altman, D. G. (2003). Measuring inconsistency in meta-analyses.
British Medical Journal, 327,557–560.doi:10.1136/bmj.327.7414.557
Hoonhorst, I., Colin, C., Markessis, E., Radeau, M., Deltenre, P., & Serniclaes, W. (2009). French native speakers in the
making: From language-general to language-specific voicing boundaries. Journal of Experimental Child Psychology,
104, 353–366. doi:10.1016/j.jecp.2009.07.005
Horlyck, S., Reid, A., & Burnham, D. (2012). The relationship between learning to read and language-
specific speech perception: Maturation versus experience. Scientific Studies of Reading, 16, 218–239.
doi:10.1080/10888438.2010.546460
∗
Joanisse, M. F., Manis, F. R., Keating, P., & Seidenberg, M. S. (2000). Language deficits in dyslexic chil-
dren: Speech perception, phonology, and morphology. Journal of Experimental Child Psychology, 77, 30–60.
doi:10.1006/jecp.1999.2553
∗
Johnson, E. P., Pennington, B. F., Lowenstein, J. H., & Nittrouer, S. (2011). Sensitivity to structure in the speech signal
by children with speech sound disorder and reading disability. Journal of Communication Disorders, 44, 294–314.
doi:10.1016/j.jcomdis.2011.01.001
Liberman, A. M., Harris, K. S., Hoffman, H. S., & Griffith, B. C. (1957). The discrimination of speech sounds within and
across phoneme boundaries. Journal of Experimental Psychology, 54,358–368.doi:10.1037/h0044417
Liberman, I. Y., Shankweiler, D., Fischer, F. W., & Carter, B. (1974). Explicit syllable and phoneme segmentation in the
young child. Journal of Experimental Child Psychology, 18, 201–212. doi:10.1016/0022-0965(74)90101-5
∗
Liu, W., Shu, H., & Yang, Y. (2009). Speech perception deficits by Chinese children with phonological dyslexia. Journal
of Experimental Child Psychology, 103,338–354.doi:10.1016/j.jecp.2009.03.005
Lyon, G. R., Shaywitz, S. E., & Shaywitz, B. A. (2003). A definition of dyslexia. Annals of Dyslexia, 53,1–14.
doi:10.1007/s11881-003-0001-9
Downloaded by [Radboud Universiteit Nijmegen] at 07:14 10 July 2015
CATEGORICAL PERCEPTION DEFICIT IN DYSLEXIA 19
∗
Maassen, B., Groenen, P., Crul, T., Assman-Hulsmans, C., & Gabreëls, F. (2001). Identification and discrimination
of voicing and place-of-articulation in developmental dyslexia. Clinical Linguistics & Phonetics, 15, 319–339.
doi:10.1080/02699200010026102
∗
Manis, F. R., McBride-Chang, C., Seidenberg, M. S., Keating, P., Doi, L. M., Munson, B., & Petersen, A. (1997). Are
speech perception deficits associated with developmental dyslexia? Journal of Experimental Child Psychology, 66,
211–235. doi:10.1006/jecp.1997.2383
Mattock, K., Molnar, M., Polka, L., & Burnham, D. (2008). The developmental course of lexical tone perception in the
first year of life. Cognition, 106, 1367–1381. doi:10.1016/j.cognition.2007.07.002
Melby-Lervåg, M., & Lervåg, A. (2012). Oral language skills moderate nonword repetition skills in children with
dyslexia: A meta-analysis of the role of nonword repetition skills in dyslexia. Scientific Studies of Reading, 16,1–34.
doi:10.1080/10888438.2010.537715
Melby-Lervåg, M., Lyster, S.-A. H., & Hulme, C. (2012). Phonological skills and their role in learning to read: A meta-
analytic review. Psychological Bulletin, 138,322–352.doi:10.1037/a0026744
∗
Messaoud-Galusi, S., Hazan, V., & Rosen, S. (2011). Investigating speech perception in children with dyslexia: Is there
evidence of a consistent deficit in individuals? Journal of Speech, Language, and Hearing Research, 54, 1682–1701.
doi:10.1044/1092-4388(2011/09-0261)
∗
Nittrouer, S. (1999). Do temporal processing deficits cause phonological processing problems? Journal of Speech,
Language, and Hearing Research, 42, 925–942. doi:10.1044/jslhr.4204.925
Noordenbos, M. W., Segers, E., Serniclaes, W., Mitterer, H., & Verhoeven, L. (2012a). Allophonic mode of speech
perception in Dutch children at risk for dyslexia: A longitudinal study. Research in Developmental Disabilities, 33,
1469–1483. doi:10.1016/j.ridd.2012.03.021
Noordenbos, M. W., Segers, E., Serniclaes, W., Mitterer, H., & Verhoeven, L. (2012b). Neural evi-
dence of allophonic perception in children at risk for dyslexia. Neuropsychologia, 50, 2010–2017.
doi:10.1016/j.neuropsychologia.2012.04.026
∗
Noordenbos, M. W., Segers, E., Serniclaes, W., & Verhoeven, L. (2013). Neural evidence of the allophonic
mode of speech perception in adults with dyslexia. Clinical Neurophysiology, 124, 1151–1162.
doi:10.1016/j.clinph.2012.12.044
∗
Reed, M. A. (1989). Speech-perception and the discrimination of brief auditory cues in reading disabled-children.
Journal of Experimental Child Psychology, 48(2), 270–292. doi:10.1016/0022-0965(89)90006-4
∗
Robertson, E. K., Joanisse, M. F., Desroches, A. S., & Ng, S. (2009). Categorical speech perception
deficits distinguish language and reading impairments in children. Developmental Science, 12, 753–767.
doi:10.1111/j.1467-7687.2009.00806.x
∗
Rosen, S., & Manganari, E. (2001). Is there a relationship between speech and nonspeech auditory processing in children
with dyslexia? Journal of Speech, Language, and Hearing Research, 44,720–736.doi:10.1044/1092-4388(2001/057)
Serniclaes, W., & Sprenger-Charolles, L. (2015). Reading impairment: From behavior to brain. In R. H. Bahr & E. R.
Silliman (Eds.), Routledge handbook of communication disorders (pp. 34–45). London, UK: Routledge.
∗
Serniclaes, W., Sprenger-Charolles, L., Carré, R., & Démonet, J. F. (2001). Perceptual discrimination of
speech sounds in developmental dyslexia. Journal of Speech, Language, and Hearing Research, 44, 384–399.
doi:10.1044/1092-4388(2001/032)
∗
Serniclaes, W., Van Heghe, S., Mousty, P., Carré, R., & Sprenger-Charolles, L. (2004). Allophonic mode of speech
perception in dyslexia. Journal of Experimental Child Psychology, 87, 336–361. doi:10.1016/j.jecp.2004.02.001
Sprenger-Charolles, L., Colé, P., & Serniclaes, W. (2013). Reading acquisition and developmental dyslexia (2nd ed.).
New York, NY: Psychology Press.
∗
Steffens, M. L., Eilers, R. E., Gross-Glenn, K., & Jallad, B. (1992). Speech perception in adult subjects with familial
dyslexia. Journal of Speech and Hearing Research, 35(1), 192–200. doi:10.1044/jshr.3501.192
Stroup, D. F., Berlin, J. A., Morton, S. C., Olkin, I., Williamson, D., Rennie, D., & Thacker, S. B. (2000). Meta-analysis
of observational studies in epidemiology: A proposal for reporting. Journal of the American Medical Association,
283, 2008–2012. doi:10.1001/jama.283.15.2008
Swanson, H. L., Zheng, X., & Jerman, O. (2009). Working memory, short-term memory, and reading disabilities: A selec-
tive meta-analysis of the literature. Journal of Learning Disabilities, 42, 260–287. doi:10.1177/0022219409331958
∗
van Beinum, F. J., Schwippert, C. E., Been, P. H., van Leeuwen, T. H., & Kuijpers, C. T . L. (2005). Development and
application of a /bAk/-/dAk/ continuum for testing auditory perception within the Dutch longitudinal dyslexia study.
Speech Communication, 47,124–142.doi:10.1016/j.specom.2005.04.003
Downloaded by [Radboud Universiteit Nijmegen] at 07:14 10 July 2015
20 NOORDENBOS AND SERNICLAES
van IJzendoorn, M. H., & Bus, A. G. (1994). Meta-analytic confirmation of the nonword reading deficit in developmental
dyslexia. Reading Research Quarterly, 29,267–275.doi:10.2307/747877
∗
Van d e r mo s t e n, M . , Bo e t s , B. , Lu t s , H. , P oe l m a ns , H. , Go le s t a ni , N. , Wo u t e rs , J. , & Ghe s q u iè r e , P. (2 01 0 ) . Ad ul t s wi t h
dyslexia are impaired in categorizing speech and nonspeech sounds on the basis of temporal cues. Proceedings of the
National Academy of Sciences of the United States of America , 107,10389–10394.doi:10.1073/pnas.0912858107
∗
Van d e r mo s t e n, M. , B o et s , B . , L u ts , H. , Po e l m an s , H. , Wo ut e r s , J . , & Gh e s q ui è re , P. ( 2 0 11 ) . Im p a i rm e n t s in sp e e c h an d
nonspeech sound categorization in children with dyslexia are driven by temporal processing difficulties. Research in
Developmental Disabilities, 32,593–603.doi:10.1016/j.ridd.2010.12.015
∗
Veu i l l et , E., Ma gn a n , A. , Ec a ll e , J. , Tha i - Va n, H. , & C o l le t , L. (2 00 7 ) . Au d i to r y pr o c e ss i n g di so r d e r i n ch il d r e n w i th
reading disabilities: Effect of audiovisual training. Brain, 130, 2915–2928. doi:10.1093/brain/awm235
Viechtbauer, W. (2010). Conducting meta-analyses in R with the metafor package. Journal of Statistical Software, 36(3),
1–48.
Viechtbauer, W ., & Cheung, M. W . L. (2010). Outlier and influence diagnostics for meta-analysis. Research Synthesis
Methods, 1, 112–125. doi:10.1002/jrsm.11
Vo g e l , A . C ., P e t er s e n , S. E. , & S c h la g g a r , B . L. (2 0 1 4 ) . T h e VW FA : It ’ s n o t ju s t fo r wo r d s a ny m o r e . Frontiers in Human
Neuroscience, 8(88). doi:10.3389/fnhum.2014.00088
∗
Werker, J. F., & Tees, R. C. (1987). Speech perception in severely disabled and average reading children. Canadian
Journal of Psychology/Revue Canadienne de Psychologie, 41, 48–61. doi:10.1037/h0084150
∗
White, S., Milne, E., Rosen, S., Hansen, P., Swettenham, J., Frith, U., & Ramus, F. (2006). The role of sensori-
motor impairments in dyslexia: A multiple case study of dyslexic children. Developmental Science, 9, 237–255.
doi:10.1111/j.1467-7687.2006.00483.x
∗
Zhang, Y., Zhang, L., Shu, H., Xi, J., Wu, H., Zhang, Y., & Li, P. (2012). Universality of categorical perception deficit
in developmental dyslexia: An investigation of Mandarin Chinese tones. Journal of Child Psychology and Psychiatry,
53,874–882.doi:10.1111/j.1469-7610.2012.02528.x
Downloaded by [Radboud Universiteit Nijmegen] at 07:14 10 July 2015