Developmental dyslexia in diﬀerent
languages: Language-speciﬁc or universal?
Johannes C. Ziegler,
and Gerd Schulte-K€
CNRS and Universit
ee de Provence, Marseille, France
Joint Laboratories for Language and Cognitive Neuroscience, The University of Hong Kong, Hong Kong
The Smith-Kettlewell Eye Research Institute, San Francisco, USA
Department of Child and Adolescent Psychiatry, Phillips University of Marburg, Marburg, Germany
Received 21 April 2003; revised 1 August 2003
Most of the research on developmental dyslexia comes from English-speaking countries.
However, there is accumulating evidence that learning to read English is harder than learning
to read other European orthographies (Seymour, Aro, & Erskine, 2003). These ﬁndings there-
fore suggest the need to determine whether the main English ﬁndings concerning dyslexia can
be generalized to other European orthographies, all of which have less irregular spelling-to-
sound correspondences than English. To do this, we conducted a study with German- and En-
glish-speaking children (n¼149) in which we investigated a number of theoretically important
marker eﬀects of the reading process. The results clearly show that the similarities between
dyslexic readers using diﬀerent orthographies are far bigger than their diﬀerences. That is, dys-
lexics in both countries exhibit a reading speed deﬁcit, a nonword reading deﬁcit that is greater
than their word reading deﬁcit, and an extremely slow and serial phonological decoding mech-
anism. These problems were of similar size across orthographies and persisted even with re-
spect to younger readers that were at the same reading level. Both groups showed that they
could process larger orthographic units. However, the use of this information to supplement
grapheme–phoneme decoding was not fully eﬃcient for the English dyslexics.
Ó2003 Elsevier Inc. All rights reserved.
Keywords: Developmental dyslexia; Phonological deﬁcit; Reading; Length eﬀect; Cross-language research
Corresponding author. Fax: +33-4-91-10-62-55.
E-mail address: firstname.lastname@example.org (J.C. Ziegler).
0022-0965/$ - see front matter Ó2003 Elsevier Inc. All rights reserved.
J. Experimental Child Psychology 86 (2003) 169–193
Children aﬀected with developmental dyslexia have diﬃculty learning to read and
spell despite adequate intelligence and educational opportunity, and in the absence
of any profound sensory or neurological impairment. Dyslexia is the most common
of the childhood learning disorders.
Much of what we know about the nature and the origin of developmental dys-
lexia comes from studies that were conducted in English-speaking countries. A
simple check in the Medline database (http://www.ncbi.nlm.nih.gov) shows that
about two-thirds of all publications on developmental dyslexia since 1998 have
came out of English-speaking countries (US, UK, Australia, Canada, and New
Zealand). Given the dominance of research using the English language, it is of
great importance to know whether dyslexia is the same in countries that use dif-
ferent languages. The ideal research design would be to give dyslexic children who
are learning to read diﬀerent languages the same reading tests with highly compa-
rable linguistic material while also ensuring that participants had been matched for
other potentially important variables such as age, vocabulary development, and
general intellectual ability. Unfortunately, only very few studies have met these
strict criteria (e.g., Landerl, Wimmer, & Frith, 1997a; Paulesu et al., 2001) and
they produced slightly conﬂicting results. The neuroimaging studies suggest a uni-
versal basis for dyslexia (Paulesu et al., 2001), whereas the behavioral studies sug-
gest that the nature and prevalence of dyslexia might diﬀer between orthographies
(Landerl et al., 1997a).
As concerns the origin of developmental dyslexia, the most unifying hypothesis
suggests that dyslexic children have speciﬁc impairments in representing, storing,
and retrieving phonological information (i.e., the phonological deﬁcit hypothesis,
see Ramus, 2003; Snowling, 2000; Wagner & Torgeson, 1987). Because reading ac-
quisition requires the child to learn the mapping between orthography and phonol-
ogy (Jorm, Share, MacLean, & Matthews, 1984; Share, 1995), problems in the
representation and use of phonological information inevitably lead to problems
in reading acquisition (e.g., Bradley & Bryant, 1983; Bryant & Bradley, 1985;
Goswami & Bryant, 1990). The phonological deﬁcit hypothesis is supported by a
number of studies showing that dyslexics have diﬃculties in tasks requiring verbal
short-term memory, phonological awareness, phonological recoding, and rapid
automatized naming (e.g., Brady & Shankweiler, 1991; Denckla & Rudel, 1976;
Rack, Snowling, & Olson, 1992). Although it is still a matter of debate whether
the phonological deﬁcit can be reduced to a more low-level sensory deﬁcit (e.g.,
Goswami et al., 2002; Helenius, Uutela, & Hari, 1999; Manis et al., 1997; Marshall,
Snowling, & Bailey, 2001; Schulte-K€
oorne, Deimel, Bartling, & Remschmidt, 1998a,
1998b; Serniclaes, Sprenger-Charolles, Carr
ee, & Demonet, 2001; Tallal, 1980), it
seems quite clear that any theory of dyslexia needs to account for the robust pho-
nological deﬁcits that are present in most dyslexics (Ramus, 2001, 2003; Ramus
et al., 2003). A universal theory of dyslexia predicts that phonological deﬁcits are
very similar for dyslexics in diﬀerent countries, an argument that we discuss further
170 J.C. Ziegler et al. / Journal of Experimental Child Psychology 86 (2003) 169–193
Reading acquisition in diﬀerent countries
An accumulating number of studies suggest that learning to read English is hard-
er, and possibly qualitatively diﬀerent, from learning to read other European orthog-
raphies (Goswami, Gombert, & Fraca de Barrera, 1998; Goswami, Porpodas, &
Weelwright, 1997; Goswami, Ziegler, Dalton, & Schneider, 2001, 2003; Oney &
Goldman, 1984; Porpodas, 1999; Thorstad, 1991). Most strikingly, in a collaborative
eﬀort, a group of European scientists
compared word and nonword reading perfor-
mance after the ﬁrst year of reading instruction across 13 European countries (for a
description of the results, see Seymour et al., 2003). They showed that word and non-
word reading accuracy was only about 40% for English children at the end of grade
1. In contrast, word reading accuracy in most other European orthographies was
close to ceiling (about 95%, with the exception of France and Denmark, which were
about 75%). Similarly, nonword reading accuracy was quite high in all European or-
thographies (about 90%, with the exception of Denmark, 53%, and Portugal, 76%).
These results suggest that English children have speciﬁc reading diﬃculties, a ﬁnding
that raises doubts about whether studies of dyslexia predominantly conducted in En-
glish can be easily generalized to other orthographies.
It is commonly accepted that the main reason for the delay of the English-speak-
ing children lies in the irregularity (inconsistency) of the writing system (e.g., Frith,
Wimmer, & Landerl, 1998). English, in comparison to all other European orthogra-
phies, such as Italian, Spanish, German, Greek, or Turkish, has a much more incon-
sistent mapping between spelling and sound. That is, in English the same spelling
patterns can often be pronounced in multiple ways (e.g., compare the pronunciation
of -ough in cough,bough,tough,through, and dough, see Ziegler, Stone, & Jacobs,
One of the strongest cross-language comparisons is between German and English.
Because of their common Germanic origin, both languages have a very similar or-
thography and phonology but they diﬀer with respect to how consistently units of
spelling map onto units of sound (Ziegler, Perry, & Coltheart, 2000). The Ger-
man–English comparison has been used in a number of studies to investigate reading
development and skilled reading comparing regular with less regular writing systems
(Frith et al., 1998; Goswami et al., 2001, 2003; Wimmer & Goswami, 1994; Ziegler,
Perry, Jacobs, & Braun, 2001). One of the key ﬁndings of this cross-language re-
search is that children learning to read a regular orthography rely to a greater extent
on grapheme–phoneme decoding whereas children in English-speaking countries
supplement grapheme–phoneme decoding by rhyme and whole word strategies
National representatives of the COST Action were: H. Wimmer, T. Reinelt (Austria), J. Alegria, J.
Morais (Belgium), C. Elbro, E. Arnbak (Denmark), H. Lyytinen, P. Niemi (Finland), J.-E. Gombert, M.-T.
Le Normand, L. Sprenger-Charolles, S. Valdois (France), A. Warnke, W. Schneider (Germany), C. Porpodas
(Greece), V. Csepe (Hungary), H. Ragnarsdottir (Iceland), C. Cornoldi, P. Giovanardi Rossi, C. Vio, A.
Parmeggiani (Italy), C. Firman (Malta), R. Licht, A.M.B. De Groot (Netherlands), F.-E. Tonnessen
(Norway), L. Castro, L. Cary (Portugal), S. Deﬁor, F. Martos, J. Sainz, X. Angerri(Spain), S. Stromqvist, A.
Olofsson (Sweden), P. Seymour, P. Bryant, U. Goswami (UK).
J.C. Ziegler et al. / Journal of Experimental Child Psychology 86 (2003) 169–193 171
(e.g., Goswami et al., 2001, 2003). Thus, there are not only quantitative but also
qualitative diﬀerences in the way English children and children using other languages
learn to read.
Developmental dyslexia in diﬀerent countries
A number of studies have investigated dyslexia in diﬀerent countries. However,
most of them used a monolingual research design; that is, they did not compare per-
formance of dyslexics from diﬀerent countries on identical tasks and stimulus mate-
rial. Taking the English orthography ﬁrst, virtually all the monolingual research
studies have found a nonword reading deﬁcit in dyslexic children compared to both
reading level and control age (CA) matched children (for a review, see Rack et al.,
1992). This nonword reading deﬁcit is particularly marked when the nonwords were
constructed to be dissimilar to real words (e.g., Ôtegwop,ÕÔtwamket,ÕSnowling, 1980).
Rack et al. (1992) noted that English studies that did not ﬁnd a nonword reading
deﬁcit in dyslexia in comparison to a reading level match typically used young read-
ers (7 years) as controls, and used nonwords with relatively familiar orthographic
patterns. However, a quantitative meta-analysis using the same database showed
that diﬀerences in age, intelligence, and word recognition ability were more impor-
tant than the type of nonwords used (Van Ijzendoorn & Bus, 1994).
A number of monolingual studies from diﬀerent countries seem to support the ex-
istence of a phonological decoding deﬁcit in consistent orthographies. For example,
Porpodas (1999) found that Greek dyslexic children read 93% of the nonwords cor-
rectly, compared to 97% for the CA controls (a signiﬁcant diﬀerence). Dyslexic chil-
dren in France read about 75% of nonwords correctly compared to 90% for the CA
controls (Sprenger-Charolles, Cole, Lacert, & Serniclaes, 2000). Finally, Dutch dys-
lexic fourth-graders showed marked nonword reading deﬁcits both in comparison to
CA and reading level controls (Van der Leij, Van Daal, & De Jong, 2002).
The problem with monolingual studies is that they can tell us only very little about
the degree of comparability between the diﬀerent orthographies. For this purpose,
cross-language studies on developmental dyslexia are particularly important. One
of the rare cross-language studies was carried out by Landerl et al. (1997a). They em-
ployed the German–English comparison to investigate whether the nature and prev-
alence of developmental dyslexia diﬀered between consistent and less consistent
orthographies. In their study, Landerl et al. (1997a) asked German and English dys-
lexics and controls to read aloud words and nonwords. Both words and nonwords
were very similar in terms of orthography, phonology, and meaning across the
two orthographies (e.g., words: bear-B€
aar,yacht-Jacht,ﬁsh-Fisch, etc.; nonwords:
aar,cacht-Kacht, and bish-Bisch). Their main ﬁnding was that English dyslex-
ics suﬀered from much more severe reading impairments in both word and nonword
reading than the German dyslexics, with their nonword reading deﬁcit being much
greater than their word reading deﬁcit. For example, despite a lenient error scoring
method, the English dyslexics had an error rate of around 70% on three-syllable non-
words. In contrast, the error rate of the German dyslexics was only 20% on the same
stimulus group. The enormous diﬃculties the English dyslexics had with word and
172 J.C. Ziegler et al. / Journal of Experimental Child Psychology 86 (2003) 169–193
nonword reading were also reﬂected in a very slow reading speed. This was true even
for short monosyllabic nonwords, which the English dyslexics read twice as slowly as
the German dyslexics.
Although the conclusion that dyslexia is more severe in English-speaking coun-
tries ﬁts the developmental data suggesting that reading acquisition is harder in En-
glish-speaking countries, there are three serious limitations of the Landerl et al.
(1997a) study. First, the poor reading performance of the English children might
have been ampliﬁed in their study by the inclusion of an exceptionally large number
of words that were irregular in English but not in German (e.g., bear,wolf,yacht,
sword,comb). Given that these irregular words produce regularization errors and
strong latency costs even for skilled adult readers (e.g., Baron & Strawson, 1976; Se-
idenberg, Waters, Barnes, & Tanenhaus, 1984; Ziegler, Perry, & Coltheart, in press),
it could be problematic to include too many of these items in a cross-language com-
parison. Similarly, given that the nonwords were derived from the real words, some
of the nonwords contained particularly inconsistent correspondences, which again
might contribute to the more severe impairment of the English children.
Second, many authors have suggested that reading speed is more critical than ac-
curacy when regular and irregular orthographies are compared (e.g. De Jong, 2003;
De Jong & van der Leij, 2003; Sprenger-Charolles et al., 2000; Wimmer, 1993). It
should be noted that Landerl et al. (1997a) did in fact measure reading speed of
the children. However, they did so by having the children press a mouse button as
soon as they knew the pronunciation. This is only a coarse estimation of reading
speed. For example, such a procedure would not be sensitive enough to detect a
speed deﬁcit in rapid automatized naming. Furthermore, this mode of responding
might be contaminated by decision and response criterion eﬀects.
Third, and most importantly, although absolute diﬀerences in accuracy and speed
between languages are certainly intriguing, they give little insight into the processes
underlying reading deﬁcits. To gain further insight into the processes underlying
reading, the strategy of the present study was to go beyond absolute levels of accu-
racy or speed by considering marker eﬀects of the underlying reading process.
Goal of the present study
The goal of the present study was to investigate reading characteristics of dyslexic
children in regular and less regular orthographies. In particular, we wanted to test
whether the English orthography ampliﬁes phonological decoding diﬃculties, as sug-
gested by Landerl et al. (1997a). More generally, because two-thirds of all studies on
dyslexia have been done in English, it is important to know whether these results can
be generalized to more regular orthographies. To study whether reading-related def-
icits are comparable across orthographies, it is necessary to consider theoretically
critical marker eﬀects of the reading process, such as eﬀects of lexicality, length,
and large orthographic units.
The lexicality eﬀect (diﬀerence in word and nonword reading) allowed us to inves-
tigate whether the phonological decoding deﬁcit was similar across languages. A
phonological decoding deﬁcit is indicated when the diﬀerence between nonword
J.C. Ziegler et al. / Journal of Experimental Child Psychology 86 (2003) 169–193 173
and word reading is relatively bigger in dyslexics than in controls. The length eﬀect
allowed us to investigate to what extent phonological decoding processes operate in a
serial manner. That is, the greater the length eﬀect, the more people rely on serial
decoding strategies (Coltheart, Rastle, Perry, Langdon, & Ziegler, 2001; Weekes,
1997). Finally, large-unit eﬀects were used to investigate whether dyslexics were able
to eﬃciently process larger orthographic units. This is an important test because it
could be that dyslexics must rely on low-level phonological decoding because they
cannot process larger orthographic units. As a marker eﬀect for large-unit process-
ing, we used body neighborhood eﬀects (Ziegler & Perry, 1998). Body neighbors are
words that share the same orthographic rime (i.e., body), such as street,meet,feet.
It has been found that large body neighborhoods facilitate word and nonword read-
ing, especially for English-speaking readers (Ziegler et al., 2001).
In the present study, the relative size of these marker eﬀects was investigated in a
German and an English population of dyslexic children. Performance of dyslexic
children was compared to that of two control groups, one matched on chronological
age and one matched on reading age. The reading-age match is important to exclude
the possibility that a given deﬁcit is simply a consequence of the poorer reading ex-
perience of dyslexic children (developmental delay). Alternatively, a deﬁcit that per-
sists in comparison to a group of younger children of the same reading age as the
dyslexic children is likely to reﬂect a fundamental deviation from normal reading
processes. As in the Landerl et al. study (1997a), we used the German/English com-
parison because identical stimulus material can be used across diﬀerent orthogra-
phies. In addition, we measured reaction times (RTs) for each item and each
participant using a voice key system that allowed us to digitally record all responses.
Forty-nine dyslexic children participated in this study, 30 from an English-speaking
population and 19 from a German speaking population. The English dyslexic children
were recruited from a dyslexia center in Sydney, Australia. About half of them came to
the dyslexia center for a one-day assessment. The other half was tested at the beginning
of a short-term remediation program. The German dyslexic children were recruited
from the Department of Child and Adolescent Psychiatry of the University Hospital,
Marburg, Germany. Testing of these children was part of a one-day assessment. The
English and German children were native speakers of their respective languages.
To qualify for the study, the children in both countries had to fall within the age
range 9–13 years, attain an IQ score of 85 or above on the Wechsler Intelligence
Scale for Children (WISC-III), and a reading score at or below the 25th percentile
on standardized reading tests. For the assessment of the reading level in Australia,
we used the Woodcock Word Identiﬁcation Subtest (Woodcock, 1989). Reading level
in Germany was assessed with the word reading subtest of the Salzburger Reading
and Spelling Test (SLRT, Landerl, Wimmer, & Moser, 1997b). The Woodcock
174 J.C. Ziegler et al. / Journal of Experimental Child Psychology 86 (2003) 169–193
reading test is based on accuracy data. The SLRT takes accuracy and reading speed
into account. Reading performance of dyslexic children in both countries was on av-
erage three years behind that of their peers. Reading age rather than standard scores
was used to match the dyslexic children in the two countries because the SLRT does
not provide standard scores.
In both countries, two control groups were tested, one that was matched on chro-
nological age (CA) and one that was matched on reading level. The reading age of
the German CA group was slightly lower than that of the English CA group. How-
ever, it is likely that we underestimated the reading age of the CA controls because
the SLRT provides reading ages only up to fourth grade. Thus, the reading age
scores for some of the German high achievers were truncated to a reading age that
would correspond to ﬁfth grade. Because of time constraints imposed by the school,
no IQ data could be obtained for these controls. All these children were native speak-
ers of their respective languages and had no history of reading problems. Character-
istics of the participants are shown in Table 1.
Design and material
The experimental conditions were created from a factorial combination of Lan-
guage (English vs. German), Stimulus Length (three, four, ﬁve, and six letters),
Body-N (large vs. small), and Lexicality (words vs. nonwords). Stimulus selection
was based on the CELEX database, which exists for both German and English (Baa-
yen, Piepenbrock, & van Rijn, 1993). In a ﬁrst run, all German/English cognates were
selected from the CELEX database. Cognates are words with identical meaning and
similar orthography and phonology in two languages (e.g., zoo in English vs. Zoo in
German). Pairs were then considered for further selection if: (1) both words were
monosyllabic; (2) both words had regular grapheme–phoneme correspondences;
(3) both words had the same number of letters; and (4) both words belonged to
the same body-N class (either large or small).
These constraints resulted in the selection of 80 monosyllabic German/English word
pairs. All items are listed in Appendix. Half (n¼40) had a small body neighborhood
Characteristics of dyslexics, chronological-age controls and reading-level controls
NAge Range SD RA Range SD
Dyslexics 30 (22 male) 10;9 9;0–12;1 0.85 7;6 6;7–8;7 0.49
CA-controls 35 (17 male) 10;9 9;0–12;3 0.92 11;9 9;0–14;5 1.82
RL-controls 16 (6 male) 8;4 7;3–9;3 0.45 8;2 7;1–8;9 0.61
Dyslexics 19 (13 male) 10;4 9;5–11;6 0.43 7;6 6;6–8;6 0.53
CA-controls 29 (15 male) 10;4 9:2–11;1 0.43 10;6 9;6–11;0 0.55
RL-controls 20 (11 male) 7;9 6;4–8;6 0.61 7;6 6;6–8;6 0.71
Note.N, Number of participants; SD, standard deviation; RA, reading age; RL, reading level.
J.C. Ziegler et al. / Journal of Experimental Child Psychology 86 (2003) 169–193 175
(3.3 and 4.3 body-Ns for German and English, respectively), and half had a large
body neighborhood (9.8 and 13.2 body-Ns for German and English, respectively).
Each word pair belonged to one of four orthographic length groups (3, 4, 5, and
6 letters) with an equal number of items per group (n¼20). The stimuli in the
two language groups were matched in terms of number of phonemes, and hence
number of grapheme–phoneme correspondences (3.7 vs. 3.6 phonemes). For the
six letter words, it was not possible to ﬁnd monosyllabic German/English cognates.
We therefore used words with comparable word frequencies that were orthographi-
cally and phonologically similar in the two languages (e.g., ﬂight/Frucht). Mean word
frequency was 192 per million for the German items and 191 per million for the En-
glish items. Word frequencies were taken from the CELEX database. Note that word
frequencies greater than 1000 per million were truncated to this cut-oﬀ value. Word
frequencies were matched in each cell of the experimental design.
The nonword manipulation paralleled the word manipulation. That is, 80 non-
words were used in each language. Nonwords were mainly identical across the
two languages (fot/Fot,lank/Lank,plock/Plock, etc.). When this was not possible be-
cause of body-N constraints, we made sure that they were orthographically and pho-
nologically as similar as possible (e.g., ler/Lir,sibe/Seib,meast/Miest, etc). Half of the
nonwords (n¼40) had a small body neighborhood (2.5 and 2.9 body-Ns for Ger-
man and English, respectively); and half had a large body neighborhood (8.5 and
10.9 body-Ns for German and English, respectively). Each cell in the design con-
tained an identical number of items.
In both countries, dyslexic children and control children were tested individually
in a quiet room. Each session started with the reading aloud task and ﬁnished with
the administration of the standardized reading tests. For the dyslexic children, the
administration of reading and IQ tests was not part of the experimental session
but had been completed shortly before the experiment by staﬀ of the dyslexia centers.
For the reading aloud task, items were presented visually in the center of a com-
puter monitor. Words and nonwords were mixed within a single block. The order of
presentation was randomized for each participant. Inter-stimulus interval was six
seconds for the dyslexics and reading level controls, and three seconds for the CA
controls. The same typography and font size was used in the two countries. Partic-
ipants were asked to read aloud the stimuli as quickly as possible. In both countries,
the same software was used to control the experiment and to collect naming re-
sponses. This software recorded each response as a digital sound ﬁle.
For each response of each participant, RTs were calculated from the onset of the
stimulus until the onset of the response. This was initially done via a computer algo-
rithm, but each response was also checked oﬀ-line using digital sound ﬁles that were
176 J.C. Ziegler et al. / Journal of Experimental Child Psychology 86 (2003) 169–193
recorded for each response of each child. Only RTs of correct responses were used to
calculate the RT means. Errors were scored oﬀ-line. For error coding of nonwords,
all phonologically plausible responses were considered correct, even if they did not
respect the most frequent grapheme–phoneme correspondences (e.g., voop would
be considered correct if the ‘‘oo’’-correspondence was pronounced either long, /u/,
or short, /f/). Thus, as in previous cross-language studies (e.g., Frith et al., 1998;
Landerl et al., 1997a), a lenient error coding criterion was adopted.
The presentation of the results is organized into ﬁve sections, which correspond to
the main questions of the study: (1) Are there global diﬀerences in reading speed and
accuracy between dyslexics and controls across languages? (2) Is the phonological
deﬁcit speciﬁc to English children? (3) Do dyslexics show stronger length eﬀects than
controls, and if so, does the size of this eﬀect diﬀer across languages? (4) Do dyslexics
have problems in the processing of larger orthographic units? (5) Can large-unit pro-
cessing compensate for small-unit decoding deﬁcits?
The results were analyzed using analyses of variance (ANOVAs) with Group
(Dyslexics vs. CA controls vs. reading level controls), Language (German vs. En-
glish), Length (three, four, ﬁve, and six), Body-N (high vs. low), and Lexicality
(words vs. nonwords) as factors. Two additional sets of ANOVAs were performed,
one for the comparison between dyslexics and CA controls, and one for the compar-
ison between dyslexics and reading level controls. Instead of presenting a long list of
main eﬀects and interactions, we organize the presentation of the statistical results
according to the ﬁve sections described above.
Global reading speed and accuracy
Means and eﬀect sizes for reading speed and accuracy of dyslexics and controls in
Germany and Australia are given in Table 2. The global ANOVA showed a signif-
icant main eﬀect of Language both for RTs, Fð1;141Þ¼13:0;p<:001, and errors,
Reading speed and accuracy for dyslexics, chronological age, and reading level controls in Australia and
RTs (ms) Accuracy (% errors)
Australia Germany Australia Germany
Groups Dyslexic 2209 1829 38.3 14.3
RL 1823 1431 36.9 16.7
CA 911 821 10.5 6.4
Eﬀect sizes Dyslexic-RL 3863981.4 )2.4
Eﬀect sizes Dyslexic-RL 0.4 0.6 0.0 )0.1
(zscoresa) Dyslexic-CA 2.1 2.3 1.2 0.6
Note. Asterisks indicate signiﬁcant group diﬀerences (p<:05).
zscores were derived from the group diﬀerences expressed in standard deviation units using pooled
J.C. Ziegler et al. / Journal of Experimental Child Psychology 86 (2003) 169–193 177
Fð1;141Þ¼51:0;p<:0001, indicating that the German children were generally fas-
ter and more accurate than the English children.
Dyslexics vs. CA controls
In terms of reading speed, there was a signiﬁcant diﬀerence between dyslexics and
CA controls, Fð1;105Þ¼198:3;p<:0001. The interaction between Language and
Group, however, failed to reach signiﬁcance. In terms of accuracy, dyslexics made
signiﬁcantly more errors than CA controls, Fð1;107Þ¼53:1;p<:0001. The interac-
tion between Language and Group was signiﬁcant, Fð1;107Þ¼16:5;p<:0001, in-
dicating that English dyslexics made relatively more errors than German dyslexics in
comparison with their respective controls.
Dyslexics vs. reading level controls
In terms of reading speed, there was a signiﬁcant diﬀerence between dyslexics and
reading level controls, Fð1;77Þ¼8:7;p<:005. The interaction between Language
and Group was not signiﬁcant, F<1. In terms of accuracy, dyslexics were not sig-
niﬁcantly diﬀerent from reading level controls, F<1. The interaction between Lan-
guage and Group was not signiﬁcant, F<1.
The results showed signiﬁcant speed diﬀerences between dyslexics and controls.
The speed impairment was not signiﬁcantly diﬀerent across languages as indicated
by the absence of a signiﬁcant Language by Group interaction. If eﬀect sizes are ex-
pressed in zscores (see Table 2), the speed impairment was numerically even stronger
in German than in English. In terms of error rates, English dyslexics made more er-
rors than German dyslexics, as indicated by the signiﬁcant Language by Group in-
teraction. Because dyslexics were matched to reading level controls, we did not
expect major diﬀerences between these two groups with regard to overall perfor-
mance. Indeed, dyslexics in both countries were not diﬀerent from reading level con-
trols in terms of error rates. However, signiﬁcant diﬀerences were obtained for
reading speed. Note, however, that dyslexics and reading level controls were matched
on the basis of word reading. To the extent that overall performance comprises both
word and nonword reading, it is likely that the signiﬁcant speed diﬀerences reﬂect
diﬃculties in nonword reading rather than inadequate selection procedures of the
reading level controls.
Nonword reading deﬁcit
Average reading speed and accuracy for word and nonword reading is presented
in Table 3. The size of the phonological decoding deﬁcit can be estimated by com-
paring the diﬀerence between word and nonword reading (i.e., the lexicality eﬀect)
across diﬀerent groups of readers. The global ANOVA showed signiﬁcant main ef-
fects of Lexicality on both RTs, Fð1;141Þ¼384:3;p<:0001, and errors,
Fð1;141Þ¼288:6;p<:0001. This main eﬀect indicates that words were read faster
and more accurately than nonwords.
178 J.C. Ziegler et al. / Journal of Experimental Child Psychology 86 (2003) 169–193
Dyslexics vs. CA controls
The latency data exhibited a strong eﬀect of Lexicality, Fð1;107Þ¼269:7;
p<:0001, an interaction between Lexicality and Group, Fð1;107Þ¼110:1;p<
:0001, and an interaction between Lexicality and Language, Fð1;107Þ¼6:8;p<
:01. Most importantly, the triple interaction between the eﬀects of Lexicality, Group,
and Language was not signiﬁcant, p>:20. The main eﬀect of Lexicality reﬂects the fact
that words were read faster than nonwords by both groups of readers. The interaction
between Lexicality and Group conﬁrms that dyslexics were generally slower in non-
word reading than CA controls. The interaction of Lexicality and Language suggests
that the word-nonword diﬀerence for both groups of readers was stronger in English
than in German. However, the absence of the triple interaction suggests that the non-
word reading deﬁcit of the dyslexic children was not diﬀerent across languages. In fact,
if the corresponding zscores are considered (see Table 3), the nonword reading deﬁcit
of the German dyslexics was very similar to that of the English dyslexics (1.35 vs. 1.27,
The error data mirrored the RT data with signiﬁcant eﬀects of Lexicality,
Fð1;107Þ¼196:2;p<:0001, signiﬁcant interactions between Lexicality and
Language, Fð1;107Þ¼27:1;p<:0001, and signiﬁcant interactions between Lexi-
cality and Group, Fð1;107Þ¼41:1;p<:0001. Most importantly, there was no
hint of a triple interaction between the eﬀects of Group, Lexicality, and Lan-
guage, F<1, suggesting again that the nonword reading deﬁcit of the English
dyslexics was not quantitatively diﬀerent from that of the German dyslexics
(13.9% vs. 11.8%).
Latencies and accuracy for word and nonword reading
Length Australia Germany
Dys RL CA Dys RL CAs
Words 1746 1513 806 1490 1277 749
Nonwords 2673 2133 1015 2168 1585 894
Dyslexic-RL 307(0.35) — — 370(0.59) — —
Dyslexic-CA 719(1.27) — — 533(1.35) — —
Words 25.2 23.7 4.4 6.9 11.7 4.9
Nonwords 51.4 50.1 16.7 21.6 21.7 7.8
Dyslexic-RL )0.3 ()0.01) — — 4.7 (0.28) — —
Dyslexic-CA 13.9(0.65) — — 11.8(0.93) — —
Note. Asterisks indicate signiﬁcant diﬀerences (p<:05).
zscores (in parentheses) were derived from the group diﬀerences in eﬀect sizes expressed in SD units
using pooled SDs.
Eﬀect size ¼nonword/word diﬀerence.
J.C. Ziegler et al. / Journal of Experimental Child Psychology 86 (2003) 169–193 179
Dyslexics vs. reading level controls
The latency data exhibited a main eﬀect of Lexicality, Fð1;79Þ¼212:0;
p<:0001, an interaction between Lexicality and Group, Fð1;79Þ¼15:1;p<:001,
and an interaction between Lexicality and Language, Fð1;79Þ¼10:4;p<:01. As
with analyses involving CA controls, there was no signiﬁcant interaction between
the eﬀects of Lexicality, Group, and Language, F<1. That is, the nonword reading
deﬁcit was present for dyslexics in both countries (as indicated by the Lexicality by
Group interaction), but the deﬁcit was of similar size across orthographies.
In terms of error rates, we obtained a diﬀerent pattern, with the critical Lexicality by
Group interaction not being signiﬁcant, F<1. This suggests that, in terms of reading
accuracy, the nonword decoding deﬁcit was not greater for dyslexics than for reading
level controls. This is also reﬂected in the eﬀect sizes. The nonword reading deﬁcit of the
English dyslexics was even smaller than that of the controls ()0.3%). The triple inter-
action between Lexicality, Group, and Language also failed to reach signiﬁcance,
F<1:2, conﬁrming the absence of a nonword decoding deﬁcit on error rates in both
orthographies (zscores: 0.28 vs. )0.01, for German and English, respectively).
With regard to reading speed, the data showed that dyslexic children in both
countries exhibited a robust nonword reading deﬁcit. The deﬁcit persisted in both
countries even in the comparison with the reading level children. The size of the non-
word reading deﬁcit did not diﬀer across orthographies. The error data showed the
same pattern but only in the CA comparison. In the reading level comparison, the
nonword reading deﬁcit was not apparent. The absence of a nonword reading deﬁcit
for accuracy seems to stand in contrast to most English studies showing reliable non-
word reading deﬁcits even in comparison with a reading level group. However, this
does not mean that that our dyslexics were not seriously reading disabled. It is more
likely due to the fact that we employed very simple monosyllabic words and non-
words. In contrast, many of the English studies that found a nonword reading deﬁcit
in the reading level comparison used nonwords that were constructed to be dissimilar
to real words (e.g., Ôtegwop,ÕÔtwamket,ÕSnowling, 1980). Also, a number of English
studies that failed to ﬁnd a nonword reading deﬁcit in the reading level comparison
used young readers (7 years) as controls (see Rack et al., 1992), and this applies to
the present study as well.
The global ANOVA showed a strong length eﬀect both on RTs, Fð3;423Þ
¼204:3;p<:0001, and errors, Fð3;423Þ¼159:3;p<:0001. More importantly,
there was a Length by Group interaction on RTs, Fð6;423Þ¼48:9;p<:0001,
and errors, Fð6;423Þ¼15:3;p<:0001, suggesting that there were signiﬁcant diﬀer-
ences in the size of the length eﬀect between diﬀerent groups of readers. Length ef-
fects as a function of language and group are presented in Table 4. Word and
nonword data were collapsed in Table 4, because the ANOVA showed no signiﬁcant
interaction between Lexicality and Length, F<1, or between Lexicality, Length,
180 J.C. Ziegler et al. / Journal of Experimental Child Psychology 86 (2003) 169–193
and Group, F<1. In other words, the length eﬀect itself was not modulated by
word/nonword status. Although the length eﬀect diﬀered in size across groups, it
is worth noting that separate ANOVAs for each group showed that the length eﬀect
was nevertheless signiﬁcant in each group, all p’s<:001.
Dyslexics vs. CA controls
The ANOVAs exhibited the critical Group by Length interaction both on RTs,
Fð3;321Þ¼110:0;p<:0001, and errors, Fð3;321Þ¼16:9;p<:0001, conﬁrming
that dyslexic children showed greater length eﬀects in reading than normal readers.
On average, the costs per additional letter (as estimated by the slopes of a linear re-
gression) were between 300 and 350 ms for dyslexics compared to 30–50 ms for CA
controls. As mentioned above, this pattern was similar for words and nonwords, as
indicated by the absence of an interaction with Lexicality ( F<1). With regard to
reading speed, the length impairment of the English dyslexics was not signiﬁcantly
diﬀerent from that of the German dyslexics, as indicated by the absence of an inter-
action between the eﬀects of Length, Group, and Language, F<1. However, En-
glish dyslexics showed a greater impairment than German dyslexics with regard to
Dyslexics vs. reading level controls
With regard to reading speed, the critical Group by Length interaction was signif-
icant, Fð3;237Þ¼11:9;p<:0001, which indicates that dyslexic children showed
Length eﬀects in reading (words and nonwords) for diﬀerent groups of readers in Australia and Germany
Dys RL CA Dys RL CA
3 1679 1503 834 1341 1208 770
4 2050 1747 899 1728 1373 822
5 2358 1878 908 1988 1463 835
6 2751 2164 1002 2261 1680 857
Dyslexic-RL 141— — 151——
Dyslexic-CA 301— — 275——
Error rates (%)
3 24.0 20.1 6.6 7.7 5.2 1.1
4 37.2 33.9 10.9 13.5 14.1 3.8
5 37.8 40.8 10.9 13.5 21.1 8.6
6 54.1 52.7 13.7 22.5 26.4 12.0
Dyslexic-RL )1.4 — — )2.6——
Dyslexic-CA 7.0—— 0.7——
Note. Asterisks indicate signiﬁcant diﬀerences (p<:05).
Eﬀect size ¼slope of the linear regression. i.e., costs per additional letter.
J.C. Ziegler et al. / Journal of Experimental Child Psychology 86 (2003) 169–193 181
greater length eﬀects even in comparison to younger readers of the same reading le-
vel. The costs per additional letter were still twice as large for dyslexics compared to
reading level controls (see Table 4 for exact values). However, in terms of accuracy,
dyslexics did not show greater length eﬀects than controls. In fact, the opposite was
the case. Dyslexics in both countries showed slightly smaller length eﬀects than read-
ing level controls, which is the opposite of what one would expect. With regard to the
cross-language comparison, the presence of a length impairment for speed but also
the absence of such an eﬀect for accuracy was similar across orthographies, as indi-
cated by the absence of a triple interaction between Group, Length and Language,
In terms of reading speed, dyslexic children in both countries showed a strong
length eﬀect. The length eﬀect was up to 11 times greater for the dyslexics than
for CA controls, and the length eﬀect was still twice as large for dyslexics than for
reading level controls. Although longer words generally produced more errors than
shorter words, there was no length-speciﬁc deﬁcit for dyslexics with regard to error
rates when performance was compared to reading level controls. The English dyslex-
ics but not the German dyslexics showed a length-speciﬁc deﬁcit on errors but only
when compared to CA controls.
A pattern of particular interest in the results was the absence of an interaction
between Length and Lexicality, which reﬂects the fact that length eﬀects were of a
similar size for both words and nonwords. This is diﬀerent from what is found
with adults, where length eﬀects are typically stronger for nonwords than for words
(Weekes, 1997). The adult data pattern is consistent with the idea that words as
opposed to nonwords beneﬁt from direct parallel orthographic access, whereas
nonwords need to be phonologically decoded in a serial fashion (Coltheart et
al., 2001). The absence of such an interaction in our study suggests that, in chil-
drenÕs reading, words cannot yet beneﬁt from direct parallel orthographic access.
Instead, both words and nonwords are probably submitted to a similar serial de-
Before accepting this interpretation, however, we need to make sure that the
length eﬀects were not simply due to the fact that the longer words were more likely
to contain complex onsets (e.g., prince,strict) than the shorter words. Indeed, be-
cause we only employed monosyllabic words, it was not possible to avoid having
the longer words start with more complex onsets than the shorter words. To address
the possibility that length eﬀects could have been due to onset complexity, we ran a
post-hoc analysis excluding all items with complex onsets. The results showed that
the length eﬀect for dyslexics was still highly signiﬁcant once items with complex on-
sets were excluded (p<:0001). However, the resulting length eﬀect was smaller in
size. The slope for the English dyslexics was reduced from 352 ms per letter to
207 ms; the slope for the German dyslexics was reduced from 302 ms per letter to
202 ms. Thus, although onset complexity cannot fully account for the greater length
eﬀects found with dyslexics, it is certainly a factor that ampliﬁes diﬃculties with
182 J.C. Ziegler et al. / Journal of Experimental Child Psychology 86 (2003) 169–193
Body N eﬀects
Reading latencies and errors are presented in Table 5 as a function of body N. The
global ANOVA showed signiﬁcant main eﬀects of Body N on both RTs,
Fð1;141Þ¼36:1;p<:0001, and errors, Fð1;141Þ¼106:3;p<:0001. This main ef-
fect indicates that items with many body neighbors were named faster and more accu-
rately than items with few body neighbors. In general, body N eﬀects were stronger for
nonwords than for words, Fð1;141Þ¼4:25;p<:05, and Fð1;141Þ¼15:0;p<:001,
for RTs and errors, respectively. However, because this word/nonword eﬀect did
not interact with Group, Language, or Language and Group (all F’s<1), we collapsed
word/nonword data in order to simplify data presentation. The Body N by Language
interaction was signiﬁcant on RTs, Fð1;141Þ¼8:7;p<:01, and errors,
Fð1;141Þ¼10:9;p<:001, reﬂecting the fact that body N eﬀects were stronger in
English than in German. Contrast analyses showed that body N eﬀects were still
signiﬁcant in each of the German groups (see Table 5). Finally, there was no signiﬁ-
cant interaction between Body N and Group and between Body N, Group, and
Language (p>:10), suggesting that diﬀerent groups of readers showed similar body
Dyslexics vs. CA controls
The main eﬀect of body N was signiﬁcant on both RTs and errors, all p’s<:0001.
The interaction between Body N and Group was marginally signiﬁcant in the RT
analysis, Fð1;107Þ¼3:0;p<:10, and signiﬁcant in the error analysis,
Fð1;107Þ¼19:8;p<:001. This interaction reﬂects the fact that the dyslexics
showed stronger facilitatory body N eﬀects than the CA controls. The triple interac-
tion between Body N, Group, and Language failed to reach signiﬁcance, which sug-
gests that the pattern of body N facilitation for dyslexics vs. CA controls was fairly
similar across languages.
Latencies and accuracy for items with many body neighbors vs. items with few body neighbors
Dys RL CA Dys RL CA
Few body N 2266 1925 939 1858 1454 834
Many body N 2153 1721 882 1800 1409 809
zscores 0.11 0.22 0.16 0.07 0.08 0.13
Error rates (%)
Few body N 44.2 40.7 12.3 16.8 19.3 7.3
Many body N 32.3 33.0 8.7 11.8 14.1 5.5
zscores 0.39 0.24 0.20 0.28 0.30 0.16
Note. Asterisks indicate signiﬁcant diﬀerences (p<:05).
Eﬀect size ¼‘‘Few body N’’ minus ‘‘many body N’’.
J.C. Ziegler et al. / Journal of Experimental Child Psychology 86 (2003) 169–193 183
Dyslexics vs. reading level controls
This comparison again showed that there were signiﬁcant body N eﬀects on both
RTs and errors, all p’s<:0001. More importantly, the Body N by Group interaction
was not signiﬁcant, all F’s<1:3, suggesting that facilitatory body N eﬀects were sim-
ilar for dyslexics and controls. Furthermore, this pattern was similar across lan-
guages as indicated by the absence of the triple interaction between the eﬀects of
Body N, Group, and Language, all F’s<1:7.
The data showed that the facilitatory body N eﬀects were as large for dyslexics as
they were for reading level controls. Dyslexics showed even greater levels of facilita-
tion than CA controls. Body N eﬀects were numerically larger in English than in
German, and they were larger in nonwords than in words. However, none of these
eﬀects interacted with Group, suggesting that the facilitatory pattern in each lan-
guage was stable across diﬀerent groups of readers.
Body N by length interaction
The previous section showed normal body N eﬀects in dyslexics compared to their
controls. A crucial question is whether dyslexics can engage in large-unit processing
to supplement grapheme–phoneme decoding, as would be expected for normal read-
ers. If they were able to do so, an interaction between Body N and Length indicating
that the length eﬀects are reduced for words with many body NÕs should be found.
The slopes of the length eﬀect are presented in Table 6 as a function of body N.
The overall ANOVA showed that length eﬀects were signiﬁcantly reduced for
items with many body NÕs, Fð3;423Þ¼12:96;p<:0001. The biggest reduction
was obtained for the English CA controls who showed length eﬀect reductions in
the order of 50%. German children showed no signiﬁcant length eﬀect reductions,
F<1, which is consistent with the fact that body N eﬀects were much weaker in
German than in English children (see Discussion). Most importantly, however, the
English dyslexic children showed a smaller reduction than both reading level and
CA controls, Fð6;255Þ¼3:41;p<:005. For the Germans, this interaction was
not signiﬁcant, p>:20.
Length eﬀect slopes (in ms per letter) for words with many body neighbors as compared to words with few
Dys RL CA Dys RL CA
Many body N 152 283 33 313 148 21
Few body N 271 420 69 290 153 33
% Reduction 3344520036
Note. Asterisks indicate whether the reduction of the length eﬀect was signiﬁcant (i.e., Length By Body
184 J.C. Ziegler et al. / Journal of Experimental Child Psychology 86 (2003) 169–193
The data clearly showed that normally-progressing readers of English eﬃciently
use large-grain size information to supplement grapheme phoneme decoding (i.e.,
all of them showed reductions of the length eﬀect for items with many body NÕs).
Dyslexic children were able to do so but their beneﬁts were smaller than those of con-
trols. For the German children, large-grain size processing did not seem to play a
major role. Length eﬀect reductions were minimal for older children and absent
for both younger children and dyslexics.
Among European orthographies, English seems to be the hardest to learn (Sey-
mour et al., 2003). Yet, most of the research on reading development and dyslexia
comes out of English-speaking countries. In the present research, an attempt was
made to investigate whether the English dyslexia results can be generalized to a more
regular orthography, like German. Of particular interest was a comparison of basic
reading processes across orthographies. This comparison was designed to include
three theoretically important psycholinguistic marker eﬀects that are relevant for un-
derstanding the basic processes underlying reading. The lexicality eﬀect was chosen
to investigate whether the phonological decoding deﬁcit was similar across lan-
guages. The length eﬀect allowed us to quantify serial processes in word and non-
word reading. Finally, the body N eﬀect allowed us to investigate to what extent
dyslexics were sensitive to larger orthographic units. Before considering each of these
marker eﬀects, we will brieﬂy discuss similarities and diﬀerences in terms of the ab-
solute speed and accuracy level of performance.
Reading speed and accuracy
The global speed analysis showed that the English dyslexics did not diﬀer from the
German dyslexics. That is, in both countries, dyslexics exhibited a marked speed def-
icit in comparison to both CA and reading level controls. The speed deﬁcit was of
similar size across orthographies. English dyslexics made more errors than German
dyslexics. Note, however, that the English orthography also provides more opportu-
nities for reading errors to be made because of its relatively high degree of inconsis-
tency. As described earlier, even normally developing children make far more
decoding errors in English-speaking countries than in German-speaking countries
(e.g., Frith et al., 1998; Goswami et al., 2001).
Another problem with the interpretation of reading errors in our study is the fact
that dyslexics in both countries did not make signiﬁcantly more errors than reading
level controls. Had we considered only reading errors, as is often done in dyslexia
studies, we would have come to the conclusion that dyslexics in both countries ex-
hibit a simple developmental delay (i.e., they are at the same level as the reading level
controls). Such a developmental delay could have been the result of insuﬃcient or
inappropriate exposure to print. In contrast, the latency data clearly showed that
J.C. Ziegler et al. / Journal of Experimental Child Psychology 86 (2003) 169–193 185
the speed deﬁcit of the dyslexics persists even in comparison with the reading level
controls. This suggests that the speed deﬁcit is not simply a consequence of poor ex-
posure to text but rather a fundamental problem underlying dyslexia. Again, the
speed deﬁcit aﬀects dyslexic children in both countries in a very similar way, as sug-
gested by almost identical eﬀect sizes. Altogether, the present data support previous
claims that reading latencies rather than errors are the more sensitive variable when
comparing reading performance across languages (De Jong, 2003; Sprenger-Charol-
les et al., 2000; Wimmer, 1993).
Does the English orthography amplify the phonological decoding deﬁcit?
There is now overwhelming evidence in favor of a speciﬁc phonological deﬁcit in
dyslexia (e.g., Ramus, 2003; Snowling, 2000). One of the hallmarks of the phonolog-
ical deﬁcit is a nonword reading deﬁcit. Given the nonword reading diﬃculties of
normally developing English-speaking children (e.g., Frith et al., 1998; Goswami
et al., 2001; Seymour et al., 2003), it was of particular interest to examine the possi-
bility that the English orthography exaggerates the nonword reading deﬁcit of dys-
lexic children. That is, it could be that the relatively irregular English orthography
ampliﬁes the decoding diﬃculties of dyslexic children, especially with respect to non-
The results presented here are extremely clear. The nonword reading deﬁcit was
present in both languages. With regard to reading speed, the deﬁcit persisted even
in comparison with reading level controls, which points to a fundamental deﬁcit.
The size of the nonword reading deﬁcit was similar across orthographies with almost
identical eﬀect sizes and zscores. The error data were less revealing because they
failed to distinguish between dyslexics and reading level controls in either country.
Together then, the present study replicates the importance of a nonword decoding
speed deﬁcit that aﬀects dyslexics in both regular and irregular orthographies in sim-
ilar ways (see also De Jong, 2003; Wimmer, 1993). That is, dyslexics in both coun-
tries show remarkably slower decoding of nonwords than of words. Their
decoding was even slower than that of children in primary school who were on av-
erage three years younger than they (i.e., reading level controls). The present result
allows us to reject the possibility that research coming out of predominantly English-
speaking countries has ampliﬁed the phonological decoding deﬁcit. Given the funda-
mental role of phonological decoding for the process of learning-to-read (Jorm et al.,
1984; Share, 1995), it is not surprising that problems in phonological decoding lead
to similar reading delays in both countries.
Serial reading processes in dyslexics
The systematic investigation of stimulus length in words and nonwords allowed us
to estimate processing costs that were caused by each additional letter. The results
revealed a striking length eﬀect for dyslexics in both countries. That is, with each ad-
ditional letter, processing times dramatically increased in a linear fashion. For the
German dyslexics, the processing costs per letter were up to 11 times greater than
186 J.C. Ziegler et al. / Journal of Experimental Child Psychology 86 (2003) 169–193
those of CA controls; for the English dyslexics, the costs were seven times greater.
Such length eﬀects suggest that the reading process of dyslexic children is extremely
serial and letter-by-letter based. In contrast, the reading process of the CA controls is
much more parallel. The length eﬀect of dyslexics was also signiﬁcantly greater than
that of reading level controls, which points again to a fundamental deﬁcit, at least
with regard to speed.
Two points are worth noting. First, serial eﬀects were not only present in nonword
reading, as might have been predicted by models in which nonlexical processing op-
erates in a serial fashion (e.g., Coltheart et al., 2001), but they were also present in
word reading. In fact, we did not ﬁnd interactions with lexicality (word/nonword sta-
tus), which suggests that words and nonwords were processed by dyslexics in a serial
letter-by-letter fashion. In dual route terminology (Coltheart et al., 2001), it is not
only the nonlexical route that is deﬁcient but also the direct lexical route. Second,
we employed fairly ‘‘easy’’ (i.e., regular) monosyllabic words, for which the length
manipulation was not as extreme as it could have been (i.e., stimuli varied between
three and six letters). Still, for such ‘‘easy’’ material, dyslexics in both countries ex-
hibited length eﬀects that were up to 11 times greater than those of normal readers.
Given that in normal text reading it is not unusual to encounter words with more
than six letters, it becomes quite clear why the reading process of dyslexics is ex-
tremely eﬀortful and time consuming. In addition, a serial decoding strategy would
increase short-term memory demands because the serially assembled phonemes need
to be blended into whole word representations. If short-term memory resources are
used up during low-level decoding and blending, they might be lacking for higher-le-
vel syntactic processing and text comprehension.
Is the processing of larger orthographic units deﬁcient?
One possible reason for a serial decoding strategy could be that dyslexics are not
as eﬃcient in the integration of large-unit information as normally developing read-
ers are. Previous research with skilled readers and children has shown that words and
nonwords with many body neighbors produce faster and more accurate reading re-
sponses than words with few body neighbors (e.g., Treiman, Goswami, & Bruck,
1990; Ziegler & Perry, 1998). In the developmental literature, this facilitatory eﬀect
is often taken to suggest that readers are able to use lexical analogies at the rime level
to supplement phonological decoding (Goswami & Bryant, 1990). Alternatively, the
eﬀect could be explained by postulating that lexical feedback is particularly sensitive
to body-size information because bodies allow the system to disambiguate inconsis-
tencies at smaller grain sizes (Treiman, Mullennix, Bijeljac-Babic, & Richmond-
In our study, dyslexics showed normal facilitatory body N eﬀects compared to their
respective controls. That is, having a large number of body neighbors was as beneﬁcial
to dyslexics as it was to normally developing readers. German readers showed smaller
body N eﬀects than English readers, which is consistent with previous ﬁndings showing
that German readers rely to a much smaller extent on large-unit processing than do
English readers (Goswami et al., 2001, 2003; Ziegler et al., 2001).
J.C. Ziegler et al. / Journal of Experimental Child Psychology 86 (2003) 169–193 187
If the integration of large-unit information were fully functional in dyslexics, one
would, however, expect to ﬁnd reductions of the length eﬀect whenever dyslexics have
the opportunity to extract larger orthographic units (i.e., when words have many body
neighbors). Indeed, the English dyslexics exhibited a reduction of the length eﬀect by
about 33%. However, the reduction was smaller than that obtained by the controls
(52 and 44%, for CA and reading level controls, respectively). German dyslexics
showed no reduction of the length eﬀect for large body-N items. However, their respec-
tive controls did not show signiﬁcant reductions either. The German pattern is fully
consistent with previous data showing that German readers rely to a great extent on
small-unit processing even if large-unit information is available to them (see Goswami
et al., 2001, 2003).
The goal of the present study was to obtain a more precise description of dyslexia
in diﬀerent orthographies. The results clearly showed that the similarities between or-
thographies were far bigger than their diﬀerences. That is, dyslexics in both countries
exhibited a reading speed deﬁcit, a speciﬁc nonword reading deﬁcit, and a phonolog-
ical decoding mechanism that operates extremely slowly and serially. These problems
were of similar size across orthographies and persisted even with respect to younger
readers (i.e., reading level controls). The dyslexics in both countries did not show a
major deﬁcit in using body-size information, that is, both groups showed normal fa-
cilitatory body N eﬀect compared to their controls. The English dyslexics, however,
showed that the integration of body-size information during the phonological decod-
ing process was not fully eﬃcient. German dyslexics showed no integration of body-
size information during phonological decoding either. However, this process seems
of little use for German readers because the controls did not show this eﬀect either.
Thus, the extremely slow and serial grapheme–phoneme decoding process remains
the main problem for dyslexic readers in both countries.
English and German dyslexics clearly diﬀered in terms of overall accuracy. How-
ever, accuracy diﬀerences should be treated with caution because overall accuracy
failed to distinguish dyslexics from reading level controls and might therefore be sim-
ply related to the fact that the English orthography provides more opportunities for
making reading errors than the German orthography. Moreover, when eﬀect sizes
rather than absolute levels of accuracy were compared, the dyslexics in the two coun-
tries exhibited very similar patterns.
There is one aspect about the present conclusion that seems puzzling at ﬁrst. Why
is the normal reading development of the English-speaking children delayed com-
pared to children from other European countries (i.e., Seymour et al., 2003), and
yet dyslexia looks quite similar in diﬀerent orthographies? One possible answer to
this puzzle is that the problems of the dyslexic children in both countries occur be-
fore regularity comes into play. That is, the bottleneck of the dyslexic children in
both countries seems to lie in the establishment of basic phonological recoding pro-
cedures. In theory, regularity might only make a noticeable diﬀerence once the basic
recoding skills are acquired. It is possible that dyslexic children never acquire the
188 J.C. Ziegler et al. / Journal of Experimental Child Psychology 86 (2003) 169–193
kind of eﬃcient phonological decoding procedures that would be sensitive to the sta-
tistical regularities of the writing system.
In sum, the present study provides behavioral evidence that is perfectly in line
with recent brain imaging work that compared dyslexia in England, France, and It-
aly (Paulesu et al., 2001). These authors found strong evidence for a common biolog-
ical origin of dyslexia in diﬀerent orthographies. If the results of the present study are
taken together with those of Paulesu et al. (2001), it seems quite clear that the causes
and consequences of dyslexia are extremely similar across regular and less regular or-
thographies. On the practical side, the present study allows us to conﬁrm that dys-
lexia results from studies using the English language can be generalized to more
regular orthographies, provided one uses ‘‘culturally fair’’ marker eﬀects of the read-
ing process in diﬀerent orthographies.
This research was conducted while J.Z. and C.P. were postdoctoral fellows at the
Macquarie Centre for Cognitive Science, Sydney, Australia. This research was spon-
sored by a Macquarie Research Grant (MURG) to J.Z. and C.P. We wish to thank
the children and staﬀ from the Department of Child and Adolescent Psychiatry of
the Phillips University Marburg and the Dalwood dyslexia centre in Sydney for their
participation in this study. Also, we would like to thank the children and teachers from
the Dietrich-Bonheoﬀer-Schule, Dietzenbach, Germany and the Lawson Public
School, Lawson, Australia, for their participation. Special thanks to Debbie Knight
for letting us test at the Dalwood centre, J€
uurgen Bartling for technical assistance,
and Kathy Borner, Vijay Narayanasamy, Snehasis Mukhopadhyay and Mathew Pal-
akal for their help with the Medline analysis. Thanks are extended to three anonymous
reviewers and a number of colleagues who provided helpful comments and suggestions,
most notably Laurie Cestnick, Max Coltheart, Usha Goswami, and Heinz Wimmer.
Low Body N High Body N Low Body N High Body N
German English German English German English German English
Tee tea Hut hat Lir ler Fot fot
Zeh toe Heu hay Rie rea Lat lat
Bus bus Eis ice Foo foo Dee dee
per per Fan fan sil sil lan lan
Zoo zoo rot red Tis tis pei bry
See sea pro pro Nuf nup meu moy
Arm arm Tag day Tof tof Sar sar
Gas gas Bar bar Nal nal Sut sut
J.C. Ziegler et al. / Journal of Experimental Child Psychology 86 (2003) 169–193 189
Low Body N High Body N Low Body N High Body N
German English German English German English German English
Box box Kuh cow Heg heg Gat gat
Akt act roh raw Fas fas nan nop
Laut loud Post post Naul naul Gack gack
plus plus Sand sand saun saun nist nist
Form form Reis rice Trub trup Tord tord
Text text Bank bank Seib sibe Torn tain
Haar hair blau blue Bohm boam pund pamp
Boot boat Nest nest Goft goft Lank lunk
Film ﬁlm vier four Noof noof nuck nuck
Verb verb Mahl meal Furg furk Plar plar
Floh ﬂea Wein wine Parn parn Bick bick
Golf golf Bier beer Perd perd Tind tump
Sturm storm Nacht night Glirt glird Steck steck
Punkt point Stein stone Brohl broal Plock plock
Kleid cloth Start start Plohr ploar Grost grost
Kreuz cross Stahl steel Spaut spond Dratz drace
Sport sport Sicht sight Miest meast Bruck bruck
Stuhl chair Traum dream Silch spilk Praum proom
steif stiﬀ Stock stick pisch prish dreil drail
Biest beast Block block Glief glief Gruck gruck
Schal scarf steil steep Drohn droan brist brist
Front front Pfund pound Stork sterk Krost crost
leicht length Frucht ﬂight strond strond Sprand sprand
Tausch taught Strand strong Friest freast Straum stroom
Wunsch weight Schein school Blorst blorce Drecht dright
prompt prince Brauch bright klackt gladge Spreil sprail
Schlaf change gleich ground Breist Baint Flisch ﬂitch
Storch course Strich stream Parsch plarch Grecht gratch
frisch phrase schwer spring ﬂurst ﬂurst pleich plench
feucht ﬁerce Pracht please strinz stince schast scrast
Frosch freeze bleich bought Straul straul sprank sprank
strikt strict Tracht thrill balsch bralse Fatsch fratch
Baayen, R. H., Piepenbrock, R., & van Rijn, H. (1993). The CELEX lexical database (CD-ROM).
Philadelphia, PA: Linguistic Data Consortium, University of Pennsylvania.
190 J.C. Ziegler et al. / Journal of Experimental Child Psychology 86 (2003) 169–193
Baron, J., & Strawson, C. (1976). Use of orthographic and word-speciﬁc knowledge in reading words
aloud. Journal of Experimental Psychology: Human Perception and Performance, 2, 383–393.
Bradley, L., & Bryant, P. E. (1983). Categorising sounds and learning to read: A causal connection.
Nature, 310, 419–421.
Brady, S., & Shankweiler, D. (1991). Phonological processes in literacy. Hillsdale, NJ: Erlbaum.
Bryant, P. E., & Bradley, L. (1985). ChildrenÕs reading problems. Oxford: Basil Blackwell.
Coltheart, M., Rastle, K., Perry, C., Langdon, R., & Ziegler, J. C. (2001). DRC: A Dual Route Cascaded
model of visual word recognition and reading aloud. Psychological Review, 108, 204–256.
De Jong, P. F. (2003). Problems in the acquisition of ﬂuent word decoding in Dutch children. In N.
Goulandris (Ed.), Dyslexia in diﬀerent languages: Cross-linguistic comparisons (pp. 33–52). London:
Whurr Publishers Ltd.
De Jong, P. F., & van der Leij, A. (2003). Developmental changes in the manifestation of a phonological
deﬁcit in dyslexic children learning to read a regular orthography. Journal of Educational Psychology,
Denckla, M. B., & Rudel, R. G. (1976). Rapid automatized naming (R.A.N.): Dyslexia diﬀerentiated from
other learning disabilities. Neuropsychologia, 14, 471–479.
Frith, U., Wimmer, H., & Landerl, K. (1998). Diﬀerences in phonological recoding in German- and
English-speaking children. Scientiﬁc Studies of Reading, 2, 31–54.
Goswami, U., & Bryant, P. E. (1990). Phonological skills and learning to read. Hillsdale, NJ: Erlbaum.
Goswami, U., Gombert, J. E., & Fraca de Barrera, L. (1998). ChildrenÕs orthographic representations and
linguistic transparency: Nonsense word reading in English, French, and Spanish. Applied Psycholin-
guistics, 19, 19–52.
Goswami, U., Porpodas, C., & Weelwright, S. (1997). ChildrenÕs orthographic representations in English
and Greek. European Journal of Psychology of Education, 3, 273–292.
Goswami, U., Thomson, J., Richardson, U., Stainthorp, R., Hughes, D., Rosen, S., et al. (2002).
Amplitude envelope onsets and developmental dyslexia: A new hypothesis. Proceedings of the National
Academy of Sciences of the United States of America, 99, 10911–10916.
Goswami, U., Ziegler, J. C., Dalton, L., & Schneider, W. (2001). Pseudohomophone eﬀects and
phonological recoding procedures in reading development in English and German. Journal of Memory
and Language, 45, 648–664.
Goswami, U., Ziegler, J. C., Dalton, L., & Schneider, W. (2003). Nonword reading across orthographies:
How ﬂexible is the choice of reading units? Applied Psycholinguistics, 24, 235–247.
Helenius, P., Uutela, K., & Hari, R. (1999). Auditory stream segregation in dyslexic adults. Brain, 122,
Jorm, A. F., Share, D. L., MacLean, R., & Matthews, R. G. (1984). Phonological recoding skills and
learning to read—a longitudinal study. Applied Psycholinguistics, 5, 201–207.
Landerl, K., Wimmer, H., & Frith, U. (1997a). The impact of orthographic consistency on dyslexia: A
German–English comparison. Cognition, 63, 315–334.
Landerl, K., Wimmer, H., & Moser, E. (1997b). Salzburger Lese- und Rechtschreibtest: Verfahren zur
Diﬀerentialdiagnose von St€
oorungen des Lesens und Schreibens f€
uur die 1. bis 4. Schulstufe [Salzburg
reading and spelling test]. Bern: Hans Huber.
Manis, F. R., McBride-Chang, C., Seidenberg, S. M., 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.
Marshall, C. M., Snowling, M. J., & Bailey, P. J. (2001). Rapid auditory processing and phonological
ability in normal readers and readers with dyslexia. Journal of Speech, Language and Hearing Research,
Oney, B., & Goldman, S. R. (1984). Decoding and comprehension skills in Turkish and English—eﬀects of
the regularity of grapheme phoneme correspondences. Journal of Educational Psychology, 76, 556–568.
Paulesu, E., Demonet, J., Fazio, F., McCrory, E., Chanoine, V., Brunswick, N., et al. (2001). Dyslexia:
cultural diversity and biological unity. Science, 291, 2165–2167.
Porpodas, C. D. (1999). Patterns of phonological and memory processing in beginning readers and spellers
of Greek. Journal of Learning Disabilities, 32, 406–416.
J.C. Ziegler et al. / Journal of Experimental Child Psychology 86 (2003) 169–193 191
Rack, J. P., Snowling, M. K., & Olson, R. K. (1992). The nonword reading deﬁcit in developmental
dyslexia: A review. Reading Research Quarterly, 27, 29–53.
Ramus, F. (2001). Dyslexia—talk of two theories. Nature, 412, 393–395.
Ramus, F. (2003). Developmental dyslexia: Speciﬁc phonological deﬁcit or general sensorimotor
dysfunction? Current Opinion in Neurobiology, 13, 212–218.
Ramus, F., Rosen, S., Dakin, S. C., Day, B. L., Castellote, J. M., White, S., et al. (2003). Theories of
developmental dyslexia: Insights from a multiple case study of dyslexic adults. Brain, 126, 841–865.
oorne, G., Deimel, W., Bartling, J., & Remschmidt, H. (1998a). Auditory processing and dyslexia:
Evidence for a speciﬁc speech processing deﬁcit. Neuroreport, 9, 337–340.
oorne, G., Deimel, W., Bartling, J., & Remschmidt, H. (1998b). Role of auditory temporal
processing for reading and spelling disability. Perceptual and Motor Skills, 86, 1043–1047.
Seidenberg, M. S., Waters, G. S., Barnes, M. A., & Tanenhaus, M. K. (1984). When does irregular spelling
or pronunciation inﬂuence word recognition? Journal of Verbal Learning and Verbal Behavior, 23, 383–
Serniclaes, W., Sprenger-Charolles, L., Carr
ee, R., & Demonet, J.-F. (2001). Perceptual discrimination of
speech sounds in developmental dyslexia. Journal of Speech, Language, and Hearing Research, 44, 384–
Seymour, P. H. K., Aro, M., & Erskine, J. M. (2003). Foundation literacy acquisition in European
orthographies. British Journal of Psychology, 94, 143–174.
Share, D. L. (1995). Phonological recoding and self-teaching: Sine qua non of reading acquisition.
Cognition, 55, 151–218.
Snowling, M. J. (1980). The development of grapheme–phoneme correspondence in normal and dyslexic
readers. Journal of Experimental Child Psychology, 29, 294–305.
Snowling, M. J. (2000). Dyslexia. Oxford: Blackwell.
Sprenger-Charolles, L., Cole, P., Lacert, P., & Serniclaes, W. (2000). On subtypes of developmental
dyslexia: Evidence from processing time and accuracy scores. Canadian Journal of Experimental
Psychology, 54, 87–103.
Tallal, P. (1980). Auditory temporal perception, phonic, and reading disabilities in children. Brain and
Language, 9, 182–198.
Thorstad, G. (1991). The eﬀect of orthography on the acquisition of literacy skills. British Journal of
Psychology, 82, 527–537.
Treiman, R., Goswami, U., & Bruck, M. (1990). Not all nonwords are alike—implications for reading
development and theory. Memory & Cognition, 18, 559–567.
Treiman, R., Mullennix, J., Bijeljac-Babic, R., & Richmond-Welty, E. D. (1995). The special role of times
in the description, use, and acquisition of English orthography. Journal of Experimental Psychology:
General, 124, 107–136.
Van der Leij, A., Van Daal, V., & De Jong, P. (2002). Task related factors in reading eﬃciency of dyslexic
children. In L. Verhoeven, C. Elbro, & P. Reitsma (Eds.), Precursors of functional literacy (pp. 229–
245). Amsterdam: John Benjamins.
Van Ijzendoorn, M. H., & Bus, A. G. (1994). Meta-analytic conﬁrmation of the nonword reading deﬁcit in
developmental dyslexia. Reading Research Quarterly, 29, 266–275.
Wagner, R. K., & Torgeson, J. (1987). The nature of phonological processing and its causal role in the
acquisition of reading skills. Psychological Bulletin, 101, 192–212.
Weekes, B. S. (1997). Diﬀerential eﬀects of number of letters on words and nonword naming latency.
Quarterly Journal of Experimental Psychology, 50A, 439–456.
Wimmer, H. (1993). Characteristics of developmental dyslexia in a regular writing system. Applied
Psycholinguistics, 14, 1–33.
Wimmer, H., & Goswami, U. (1994). The inﬂuence of orthographic consistency on reading development—
word recognition in English and German children. Cognition, 51, 91–103.
Woodcock, R. W. (1989). Woodcock reading mastery tests-revised. Circle Pines, MN: American Guidance
Ziegler, J. C., & Perry, C. (1998). No more problems in ColtheartÕs neighborhood: resolving neighborhood
conﬂicts in the lexical decision task. Cognition, 68, 53–62.
192 J.C. Ziegler et al. / Journal of Experimental Child Psychology 86 (2003) 169–193
Ziegler, J. C., Perry, C., & Coltheart, M. (2000). The DRC model of visual word recognition and reading
aloud: An extension to German. European Journal of Cognitive Psychology, 12, 413–430.
Ziegler, J. C., Perry, C., & Coltheart, M. (in press). Speed of lexical and nonlexical processing in French:
The case of the regularity eﬀect. Psychonomic Bulletin &Review.
Ziegler, J. C., Perry, C., Jacobs, A. M., & Braun, M. (2001). Identical words are read diﬀerently in diﬀerent
languages. Psychological Science, 12, 379–384.
Ziegler, J. C., Stone, G. O., & Jacobs, A. M. (1997). WhatÕs the pronunciation for OUGH and the spelling
for /u/? A database for computing feedforward and feedback inconsistency in English. Behavior
Research Methods, Instruments, & Computers, 29, 600–618.
J.C. Ziegler et al. / Journal of Experimental Child Psychology 86 (2003) 169–193 193