Content uploaded by Beate Sabisch
Author content
All content in this area was uploaded by Beate Sabisch on Jul 21, 2014
Content may be subject to copyright.
Content uploaded by Beate Sabisch
Author content
All content in this area was uploaded by Beate Sabisch on Jul 21, 2014
Content may be subject to copyright.
Content uploaded by Beate Sabisch
Author content
All content in this area was uploaded by Beate Sabisch on Jul 21, 2014
Content may be subject to copyright.
Auditory Language Comprehension in Children
with Developmental Dyslexia: Evidence from
Event-related Brain Potentials
B. Sabisch
1,2
, A. Hahne
1
, E. Glass
2
, W. von Suchodoletz
2
,
and A. D. Friederici
1
Abstract
& In the present study, event-related brain potentials (ERPs)
were used to compare auditory sentence comprehension in
16 children with developmental dyslexia (age 9–12 years) and
unimpaired controls matched on age, sex, and nonverbal intel-
ligence. Passive sentences were presented, which were either
correct or contained a syntactic violation (phrase structure)
or a semantic violation (selectional restriction). In an overall
sentence correctness judgment task, both control and dyslexic
children performed well. In the ERPs, control children and
dyslexic children demonstrated a similar N400 component for
the semantic violation. For the syntactic violation, control chil-
dren demonstrated a combined pattern, consisting of an early
starting bilaterally distributed anterior negativity and a late
centro-parietal positivity (P600). Dyslexic children showed a
different pattern that is characterized by a delayed left later-
alized anterior negativity, followed by a P600. These data in-
dicate that dyslexic children do not differ from unimpaired
controls with respect to semantic integration processes (N400)
or controlled processes of syntactic reanalyses (P600) during
auditory sentence comprehension. However, early and pre-
sumably highly automatic processes of phrase structure build-
ing reflected in the anterior negativity are delayed in dyslexic
children. Moreover, the differences in hemispheric distribu-
tion of the syntactic negativity indicate different underlying
processes in dyslexic children and controls. The bilateral dis-
tribution in controls suggests an involvement of right hemi-
spherically established prosodic processes in addition to the
left hemispherically localized syntactic processes, supporting
the view that prosodic information may be used to facilitate
syntactic processing during normal comprehension. The left
hemispheric distribution observed for dyslexic children, in con-
trast, suggests that these children do not rely on information
about the prosodic contour during auditory sentence com-
prehension as much as controls do. This finding points toward
a phonological impairment in dyslexic children that might
hamper the development of syntactic processes. &
INTRODUCTION
The majority of children learns to master reading and
writing with a sufficient level of proficiency, but a sub-
stantial portion (4–9%) of them has severe problems
(Shaywitz, Shaywitz, Fletcher, & Escobar, 1990). Devel-
opmental dyslexia is a developmental disorder and re-
fers to a failure in the acquisition of reading and/or
spelling skills despite adequate intelligence, education,
and social background. An additional criterion that is
used for defining developmental dyslexia is a significant
difference between nonverbal intelligence and spelling
and/or reading skills (Dilling, Mombour, & Schmidt,
1993). In a number of cases reading skills improve with
age (Georgiewa et al., 2004; Pennington, Van Orden,
Smith, Green, & Haith, 1990), whereas spelling prob-
lems have a lifelong persistence. As a consequence,
dyslexia leads very often to problems in school and
subsequently to limited opportunities in realizing occu-
pational objectives (Esser, Wyschkon, & Schmidt, 2002).
Developmental dyslexia is one of the most frequently
diagnosed developmental disorders in childhood, and
recent evidence suggests that it has a genetic origin.
This genetic approach of dyslexia is mainly supported
by twin studies (Davis et al., 2001) and linkage studies
(for an overview, see Fisher & DeFries, 2002; Fisher
et al., 2002).
During the last decade, research on developmental
dyslexia has accumulated, and yet the theories about
mechanisms underlying this deficit are still controversial
(Ramus et al., 2003; Witruk, Friederici, & Lachmann,
2002; Habib, 2000; von Suchodoletz, 1999). We will start
by briefly reviewing two of the most prominent theo-
ries in current research on developmental dyslexia, the
phonological deficit hypothesis and the auditory deficit
hypothesis.
1
Max Planck Institute for Human Cognitive and Brain Sciences,
Leipzig, Germany,
2
Ludwig-Maximilians-University, Munich,
Germany
D 2006 Massachusetts Institute of Technology Journal of Cognitive Neuroscience 18:10, pp. 1676–1695
According to the phonological deficit hypothesis (e.g.,
Goswami, 2003) dyslexia results from a specific impair-
ment of the phonological representations and related
processes one has to perform to read and to write.
Learning to read requires the understanding that a sin-
gle word can be segmented into small units of sound,
which ultimately correspond to single graphemes. This
process, called phonological awareness, is divided into
three different linguistic levels, syllables, onset rhymes,
and phonemes. The awareness of syllables and onset
rhymes develops before literacy acquisition, whereas
awareness of single phonemes emerges as reading and
writing are taught (Goswami, 2003). Even for children
with developmental dyslexia, the awareness of syllables
and onset rhymes improves very rapidly during the ear-
liest phase of literacy acquisition (Landerl, Wimmer, &
Frith, 1997). However, their awareness of phonemes is
still impaired even after reading and writing have been
taught, and this impairment seems to persist into adult-
hood (Snowling, Nation, Moxham, Gallagher, & Frith,
1997).
The auditory temporal processing deficit hypothesis,
proposed by Tallal, Miller, and Fitch (1993) and Tallal
and Piercy (1973, 1974), goes one step further and
claims that a deficit in the processing of transient (within
milliseconds) or rapidly changing acoustic tones leads to
difficulties in the perception of phonemes. In this view,
developmental dyslexia is caused by a primary deficit in
the auditory temporal analysis, which yields as a conse-
quence a phonological deficit and subsequently leads
to difficulties in the acquisition of reading and spelling
(Tallal & Stark, 1982; Tallal, 1980).
Despite the difference in the origin of the deficit un-
derlying developmental dyslexia, both theories assume
that the development of phonological representations is
deficient in children with developmental dyslexia. Up to
now, most studies on developmental dyslexia focused
on single phonemes or syllables, that is, segmental in-
formation (Ramus, 2001). Phonological representations,
however, encode different levels of segmental informa-
tion (e.g., individual phonemes), but also suprasegmen-
tal information such as word stress, intonation, and
rhythm. Based on this observation, we ask the question
whether, in addition to the deficiency in processing
phonological representations and processes concerning
segmental information, the processing of suprasegmen-
tal information is deficient in children with developmen-
tal dyslexia.
Studies with young unimpaired language learners
( Jusczyk, 2002) and children with developmental lan-
guage impairment (Weinert, 1992) suggest that supraseg-
mental phonological information, in particular prosodic
information provide an important cue in discovering
the syntactic structure of a language. The assumption
is that the speaker’s task during understanding a
sentence is to segment the speech input into major
syntactic constituents such as phrases and clauses. In
general, boundaries are reliably marked by prosodic
cues (e.g., preboundary lengthening, falling pitch at
the end of the clause; Snow & Balog, 2002). According
to previous research, infants are sensitive to such cues
and prefer pauses at syntactic boundaries compared
with pauses within syntactic constituents (Hirsh-Pasek
et al., 1987).
Breaking down the speech stream into phrases or
clauses might further help to discover the underlying
syntactic structure within prosodic phrases. For exam-
ple, function words (e.g., determiner, the) are likely to
occur at certain edges (i.e., boundaries) with respect
to the prosodic phrase. Because the function word the
signals the beginning of a noun phrase, once the
location of such a phrase is detected, it also helps to
find the location of other elements within the same
syntactic constituent ( Jusczyk, 2002). Further evidence
for the connection between syntax and prosody comes,
for instance, from a study with 7- to 9-month-old Ger-
man infants who recognized the internal structure of a
noun phrase (the + content word) in continuous speech
(Ho¨hle & Weissenborn, 2003). As for the importance of
prosody in acquiring syntactic rules, Weinert (1992)
showed that normal control children were able to learn
the syntactic rules from prosodically enriched sentences,
whereas children with developmental language impair-
ment could not make use of the available prosodic
information for the acquisition of syntactic rules.
Following from these observations, we hypothesize
that children with developmental dyslexia might also be
deficient in processing syntactic information. This hy-
pothesis is guided by the following suggestions from
previous research: (1) children with developmental dys-
lexia are impaired in the processing of suprasegmen-
tal information (Ramus, 2001); (2) prosodic information
provide an important cue in the acquisition of syntactic
rules for normal children; and (3) children with devel-
opmental dyslexia are deficient in the processing of
gender and number agreement, which represent infor-
mation encoded in the syntactic domain ( Jime´nez et al.,
2004). In other words, our proposal is that if children
with developmental dyslexia are deficient in processing
prosodic information, they might as well be deficient in
processing syntactic information, because prosodic and
syntactic information are closely connected as described
above.
The critical question then is, ‘‘Are dyslexic children
with problems in reading and writing impaired in syn-
tactic processing as well as phonological processing or
either syntactic or phonological processing alone?’’ This
is an important question that we address in the research
reported here, as our hypothesis about dyslexia ties
syntactic processes with prosodic processes, and it is
therefore essential to determine the exact nature of the
relation between these two processes. In addition,
independent from our hypothesis, there has been a
long-lasting debate whether children with developmen-
Sabisch et al. 1677
tal dyslexia have an impairment in syntactic processing
at all. It has been proposed that better developed syn-
tactic abilities can facilitate the development of reading
and writing (for the study on normally developing chil-
dren, see the work of Muter & Snowling, 1997; for the
study on dyslexia, see the work of Nation & Snowling,
2000). Moreover, it was reported that normally develop-
ing children showed a higher reading proficiency when
they could use syntactic and semantic information (e.g.,
for syntax, word order rules; for semantic, word knowl-
edge) as context information to predict the upcoming
word (Rego & Bryant, 1993).
The present study was conducted to investigate
whether our assumption holds that children with de-
velopmental dyslexia differ from control children in
syntactic processing, and if so, whether this is because
of differences in prosodic processes that have an impact
on the syntactic processing. In addition, we investigated
the use of syntactic and semantic contextual informa-
tion during sentence comprehension. We utilized event-
related brain potentials (ERPs), which for many years
have been proven to be a very useful time-sensitive
on-line measurement for language processing. In addi-
tion, this method has been used successfully to investi-
gate the developmental changes in normally developing
children in sentence comprehension processes (Hahne,
Eckstein, & Friederici, 2004) as well as the on-line use
of prosodic information in sentence comprehension
(Steinhauer, Alter, & Friederici, 1999).
Before moving on to the next section, we will brief-
ly summarize the major findings in the previous ERP
research regarding adult sentence comprehension, be-
cause these will guide us in the interpretation of the
data on children reported in the present article. In
relation to the processing of syntactic, prosodic, and
lexical–semantic information, different ERP components
have been reported by using violation paradigms (for an
overview, see Friederici, 2002) that contrast correct sen-
tences with sentences containing a specific linguistic vio-
lation. In relation to the processing of lexical–semantic
information, an ERP component called N400 (centro-
parietal negativity occurring between 300 and 500 msec
poststimulus onset) was found (for an overview, see the
work of Kutas & Federmeier, 2000). The N400 is elic-
ited by semantic anomalies such as selectional restric-
tion violations, in which a word cannot be integrated
in the preceding context. It is assumed that the N400
reflects processes of lexical–semantic integration. With
respect to syntactic information, two ERP components,
the so-called ELAN (early left anterior negativity regis-
tered between 100 and 300 msec) and the P600 (late
parietal positivity elicited between 500 and 1000 msec)
were reported. Syntactic violations, realized by phrase
structure violations, elicit an ELAN that is thought to
reflect early processes of phrase structure building
(Hahne & Friederici, 1999). The P600 is commonly as-
sumed to represent syntactic processes of reanalysis
and repair (for an overview, see the work of Hagoort,
Brown, & Osterhout, 1999). In a recent study, Eckstein
and Friederici (2005) reported a right anterior negativity
(RAN) for prosodic incongruities, resulting from the
mismatch between the syntactic structure and the pros-
ody expected for that particular syntactic structure in a
sentence. In a further study, Eckstein and Friederici (in
press) showed that a prosodic incongruity could influ-
ence early syntactic phrase structure building processes
(reflected in the ELAN) by additional involvement of
right anterior regions.
Present Study
The major question that the present study focuses on is
whether children with dyslexia are different from nor-
mally developing control children in the processing of
prosodic information and, as a consequence, also in the
processing of syntactic information. To control whether
possible differences observed between unimpaired con-
trol children and children with developmental dyslexia
are specifically related to the domains of prosody and
syntax, we also tested the ERP response to lexical–
semantic anomalies. This study also investigates segmen-
tal and suprasegmental information because we want to
examine whether children with developmental dyslexia
are impaired in the processing of segmental phonol-
ogy, as frequently reported in the literature and consid-
ered to be the origin of developmental dyslexia by the
phonological deficit hypothesis. In addition, this study
tries to disentangle whether children with developmen-
tal dyslexia and unimpaired control children differ in the
processing of auditory information, as postulated by the
auditory temporal deficit hypothesis.
In the ERP experiment, we used a well-established
paradigm in which correct conditions and two violation
conditions, namely a syntactic and for control purposes,
a semantic violation condition were presented. The syn-
tactic violation condition contained two different as-
pects, a phrase structure violation (syntactic level) and
a prosodic incongruence (prosodic level). In more de-
tail, our test sentences contained a noun phrase fol-
lowed by an auxiliary, a prepositional phrase, and the
past participle. On the syntactic level, the prepositional
phrase requires a noun phrase (e.g., noun or adjective
plus noun) immediately after the preposition. In our test
sentences for the syntactic violation condition, however,
the noun phrase was spliced out, and a past participle
immediately appeared after the preposition, yielding
an ungrammatical construction. On the prosodic level,
a preposition signals that additional constituents should
follow. Specifically, the word immediately following
the preposition was not expected to convey the sen-
tence or phrase final prosody. Because the noun phrase
was removed by the splicing procedure, the constituent
following the preposition contains sentence final pros-
1678 Journal of Cognitive Neuroscience Volume 18, Number 10
ody and therefore resulted in an incongruent continua-
tion of the sentence melody. For the semantic violation
condition, the selectional restriction of the verb did
not match with the noun phrase at the subject position
(e.g., in ‘‘The volcano was eaten,’’ the noun phrase ‘‘the
volcano’’ does not satisfy the selectional restriction of
the verb eaten and cannot be integrated).
The predictions are as follows. For the syntactic
violation condition, we hypothesize that unimpaired
control children show a left anterior negativity (repre-
senting early processes of phrase structure building) and
a RAN (representing the detection of the prosodic
incongruent sentence continuation). Using the iden-
tical violations in sentences with passive constructions,
an early left anterior adultlike negativity was found for
13-year-olds but not for younger children who dem-
onstrated a sustained bilaterally distributed anterior
negativity (Hahne et al., 2004). The more bilateral dis-
tribution of the negativity was attributed to a possible
involvement of the right hemispherically localized pro-
sodic processes (Friederici & Alter, 2004) that may sup-
port syntactic processes during development. Based on
the assumption that children with developmental dys-
lexia might differ in processing prosodic information,
we hypothesize that this should result in a modulation
in latency or amplitude of the anterior negativity over
the right electrode sites. Assuming that prosody sup-
ports the acquisition of syntactic rules, early processes
of phrase structure building should be particularly af-
fected in dyslexic children, and this in turn should also
be reflected in latency or amplitude modulations for the
ELAN. For the syntactic violation condition, we predict
a P600 component in addition to the anterior negativity
for the control children. From developmental studies,
we know that the P600 is present in young children (for
active sentences, Oberecker, Friedrich, & Friederici,
2005) and in normally developing children from 7 years
on (for sentences of passive voice, Hahne et al., 2004).
Because the processes of reanalyses and repair, re-
flected in the P600, are established early during de-
velopment, we expect children with developmental
dyslexia to show a similar P600 component as the con-
trol children.
In the semantic violation condition, the past participle
cannot be integrated into the preceding context, and
in line with previous findings, an N400 component is
expected (cf. Kutas & Federmeier, 2000). Two studies
that used the identical or similar semantic violation con-
dition reported an N400 component for normally devel-
oping children from the age of 5 years onward (Hahne
et al., 2004; Holcomb, Coffey, & Neville, 1992). Neville,
Coffey, Holcomb, and Tallal (1993) found for a group
of children with language impairment an increased am-
plitude of the N400, whereas Bonte and Blomert (2004)
reported similar N400 components for children with
dyslexia and control children. We expect a similar
N400 component for the unimpaired control children
and children with developmental dyslexia.
METHODS
Participants
Sixteen children with developmental dyslexia and 16 con-
trol children (ranging in age between 9 years 5 months
and 12 years 10 months; mean age = 11 years 1 month;
SD = 1 year 1 month) were tested. The children in both
groups were matched pairwise on age, sex, and non-
verbal IQ. The paired t test revealed no significant dif-
ference between the two groups on their age and
nonverbal IQ (see Table 1). The children with develop-
mental dyslexia were selected from a sample of par-
ticipants either referred to a special outpatient unit
for developmental disorders at the Child and Adoles-
cent Psychiatry of the Ludwig-Maximilians-University in
Munich or from schools for children and adolescents
with specific problems in literacy and language. Control
children attended local schools of the urban area of
Munich and were recruited via letters.
All participants, both control children and children
with developmental dyslexia, were German-speaking
monolinguals, and none of them had any hearing deficit
Table 1. Descriptive Data for the Control and Dyslexic Children (Mean ± SD)
Controls Dyslexic Children Differences (t Value) p Value
Age (years; months) 11;1 ± 1;0 11;1 ± 1;1 0.99 .34
Sex (male/female) 9/7 9/7
Nonverbal IQ
a
104.9 ± 8.4 104.9 ± 11.1 0.00 1.0
Spelling test
b
53.7 ± 11.1 29.4 ± 4.0 7.63 <.001
Reading test
b
(speed) 52.2 ± 6.2 37.5 ± 8.0 5.61 <.001
Reading test
b
(error rate) 54.0 ± 7.5 33.5 ± 9.3 6.36 <.001
a
IQ scores (M = 100, SD = 15).
b
t scores (M = 50, SD = 10).
Sabisch et al. 1679
(confirmed by an audiometric screening procedure) or
reported a history of neurological disorders (assessed by
parent questionnaire and spontaneous electroencephalo-
gram [EEG]). Furthermore, all of them had a nonverbal
IQ within the normal range. The nonverbal IQ of children
older than 12 years 5 month was measured by the Ger-
man version of the Wechsler Intelligence Scale for Chil-
dren (Wechsler, 2000) and the nonverbal IQ of younger
children was assessed by the Kaufman-Assessment-
Battery for Children (Kaufman & Kaufman, 1991). In
addition, we assessed all participants’ reading abilities
by using the Zu¨rcher Lesetest (Linder & Grissemann,
1980). Although this test had no updated norms, it is
the only reading test in German that provides norms
from grades 2 to 6. The test was particularly appropriate
for the present purpose, allowing the assessment of
reading errors and reading speed.
Children with developmental dyslexia had been diag-
nosed based on the criteria of the International Classi-
fication of Diseases, 10th Revision (ICD-10; Dilling et al.,
1993). Their spelling abilities were at least 1.5 standard
deviations below the mean on a standardized score
in the spelling test according to their attended school
grade (Weingartener Grundwortschatz Rechtschreib-
Test, WRT 3+, Birkel, 1994a; WRT 2+, Birkel, 1994b;
Diagnostischer Rechtschreibtest, DRT 5, Grund, Haug,
& Naumann, 1995; DRT 4, Grund, Haug, & Naumann,
1998; DRT 3, Mu¨ller, 1997; Westermann Rechtschreib-
Test, WRT 4/5, Rathenow, 1979; WRT 6+, Rathenow,
Vo¨ge, & Laupenmu¨hlen, 1980). Furthermore, they had
to show a discrepancy of at least one standard deviation
between the nonverbal IQ score and the score in the
spelling test.
The control children had to have no signs of devel-
opmental or psychiatric disorders. This was confirmed
by the Child Behavior Checklist (CBCL/4–18; Achenbach,
1998) and a parent questionnaire. Control children who
scored above 60 t scores on the main scale or any
subscales of the CBCL (externalizing and/or internalizing
disorders) were excluded from the study. In addition
children were excluded when their parents reported any
signs of a developmental disorder in the parent ques-
tionnaire (e.g., hyperactivity). All participants’ parents
were given informed consent. The study was approved
by the Ethical Committee of the Ludwig-Maximilians-
University.
To assess the phonological awareness, a phoneme
substitution task (Wimmer, 1993; Landerl, Linortner, &
Wimmer, 1992) was applied. Moreover, an oddity task
(Berwanger, 2002), similar to the task developed by
Bradeley and Bryant (1985), was used to measure rhyme
awareness. The phoneme substitution task, and the odd-
ity task had two different versions each, one for children
in Grade 1 and the other for children in Grade 2.
The version for children in Grade 2 contained test
items that were mono-, bi-, tri-, or quatrosyllabic words.
They were presented as single words or embedded in
sentences. Participants had the task to replace the vow-
els such as a, e, u, and o by the vowel i. After a training
session, the first two items had to be answered cor-
rectly, otherwise the assessment test continued with the
phoneme substitution task of Grade 1. The version for
Grade 1 only contained mono- and/or bisyllabic words,
having only the vowel a, which again had to be replaced
by i.
In the oddity task, 10 groups of four words were
presented, and the children had to identify a phonotac-
tically odd word, which differed from the rest of the
words in either its medial or final sound. From Grade 2
onward, 90% of all items were bisyllabic words. In
more detail, among the 10 groups of words provided,
half of the groups contained odd words with either a
voiceless or voiced sound (e.g., Grube, Stube, Hupe,
Tube), and the remaining five groups contained odd
words with either a labial or alveolar sound (e.g., Rappe,
Kappe, Ratte, Mappe). Again, after a training session,
three of the first four items had to be answered cor-
rectly, otherwise testing continued with the oddity task
for Grade 1.
Stimuli and Event-related Brain
Potential Experiment
The experiment consisted of 192 sentences in the pas-
sive voice that were identical to those used in previ-
ous studies with adults, children, and second-language
learners (Hahne et al., 2004; Hahne & Friederici, 2002).
Altogether, four experimental conditions, two correct
conditions and two incorrect conditions, each including
48 sentences were auditorily presented. One of the cor-
rect conditions contained sentences that consisted of
a noun, an auxiliary, and a past participle (‘‘Das Brot
wurde gegessen.’’ [‘‘The bread was eaten.’’]). The other
correct condition additionally included a prepositional
phrase (‘‘Die Pizza wurde im Restaurant gegessen.’’
[‘‘The pizza was eaten in-the restaurant.’’]). In con-
trast, the two incorrect conditions contained sentences
that were either semantically or syntactically incorrect.
In the semantic violation condition, the selectional re-
strictions of the verb were violated (‘‘Der Vulkan wurde
gegessen.’’ [‘‘The volcano was eaten.’’]). In the syntactic
violation condition, sentences consisted of a noun, an
auxiliary, and a preposition. In this condition, a noun
or an adjective was expected to follow the preposition;
however, a past participle was presented immediately
after the preposition. This resulted in a violation of
the phrase structure (‘‘Das Eis wurde im gegessen.’’
[‘‘The ice cream was in-the eaten.’’]). In all conditions,
the participles represented the target words, beginning
with the morpheme ge-, which marks the regular form of
German participle forms.
All sentences were spoken by a female native speaker
of German and recorded on a digital audio tape. The
1680 Journal of Cognitive Neuroscience Volume 18, Number 10
taped sound files had a resolution of 16-bit and were
sampled at 20 kHz. To establish a precisely time-locked
presentation of all items, the auditory signal and the
oscillogram were compared, and the onset of each word
was marked on the time scale. Sentences of both cor-
rect conditions and the semantic violation condition
were naturally spoken. Syntactically incorrect sentences
were created in the following way: First, a correct sen-
tence with a prepositional phrase was produced by the
speaker (e.g., ‘‘Das Eis wurde im Gebrumm gegessen.’’
[‘‘The ice cream was in-the hum eaten.’’]). Afterward,
the noun (e.g., Gebrumm [hum]) was spliced out from
the recorded sentence (see Figure 1). To prevent the
participants from recognizing the splicing point and to
control for coarticulation of the phonemes at the critical
transition between the preposition and the past partici-
ple, the following constraints were set up: (1) the noun
(e.g., Gebrumm [hum]) and the past participle (e.g.,
gegessen [eaten]) had an identical onset phoneme; (2)
the last phoneme of the preposition (e.g., im [in-the])
and the noun (e.g., Gebrumm [hum]) were identical. A
behavioral experiment (Hahne & Friederici, 2002)
showed that the splicing procedure alone did not lead
to unnaturalness in prosody as long as the procedure is
carefully adapted in test sentences. In this experiment,
participants listened to the violation conditions (without
a splicing) and their counterparts created by using a
splicing procedure, which either controlled coarticula-
tion of the phonemes in the splicing position or lacked
such an experimental manipulation. Participants were
asked to judge whether sentences contained any splic-
ing. The results indicated that the participants could not
reliably detect splicing points when the sentences were
created by controlling for coarticulation. In contrast,
when coarticulation was not controlled in the spliced
sentences, the participants classified these sentences as
spliced.
Although the syntactically incorrect sentences were
created by controlling for coarticulation at the local
transition between the preposition and the past par-
ticiple, they also induced a manipulation of the pro-
sodic contour, resulting in an incongruent continuation
of the sentence melody that does not satisfy the parsers
expectations. Note that German is a verb-final language,
which in a passive construction requires the participle
to be in sentence final position. In more detail, by
encountering the preposition of this passive construc-
tion, the parser expects ongoing speech, and therefore
at least two more words should appear to have a
congruous continuation. The first word (e.g., noun or
adjective) that immediately follows the preposition
cannot contain the sentence final intonation in this
passive construction, whereas the second and at the
same time last word (e.g., past participle) should do
so. In the syntactically incorrect sentences, in which the
noun of the prepositional phrase was spliced out, the
parser now encounters a word that contains the sen-
tence final prosody immediately after the preposition
Figure 1. The splicing
procedure is illustrated by
the oscillogram and the
fundamental frequency ( f
0
)
for a sentence with a complete
prepositional phrase (upper)
and the spliced version of the
identical sentence (lower). The
scissors
C
indicate the splicing
points. The past participle
gegessen contains the sentence
final prosody that is marked by
the following four prosodic
features: (1) starting point of
the past participle ( gegessen)is
higher than that of the noun
(Gebrumm); (2) f
0
maximum
of the past participle is higher
than that of the noun; (3)
sharp fall in the f
0
contour at
the end of the past participle;
(4) small f
0
rise at the end of
the past participle following
the sharp fall described in the
third prosodic feature.
Sabisch et al. 1681
(see Figure 1, lower), which leads to an incongruent
prosodic contour.
Procedure
The diagnostic procedure and the ERP recording were
carried out in two sessions. Each of them took approx-
imately 2 hours and was conducted at the research
center of the Child and Adolescent Psychiatry of the
Ludwig-Maximilians-University. At the end of the two
sessions, participants received an age-appropriate re-
ward for participating in the study (e.g., toys for youn-
ger children and movie tickets for teenagers). During
the first session, all diagnostic data were collected. The
second session included the audiometric screening, the
hand preference test, the ERP experiment, and a spon-
taneous EEG.
Pure tone audiometry was used to measure hearing at
frequencies varying from low pitch (250 Hz) to high
pitch (6000 Hz). Tones were presented via headphones,
either to the left or the right ear. Participants had to
signal to the experimenter whenever they heard a tone
by raising their hand corresponding to the ear where the
tone had been detected. Participants that scored the
hearing threshold level with a score of 25 dB or higher
were excluded from the study.
In the ERP experiment, participants were seated in a
comfortable chair that was placed approximately 1.5 m
away from a computer screen. They were instructed to
listen carefully to the sentences presented via loud-
speakers. The experiment was divided into four blocks,
each containing 48 trials, with breaks in between that
lasted as long as the participants wanted. At the begin-
ning of each block, a wizard appeared in the middle of
the computer screen, which gradually became smaller
and finally disappeared by 1200 msec. Each trial start-
ed with three exclamation marks that were shown for
500 msec before the presentation of a sentence. When
the sentence was played, the small wizard appeared
again on the screen and remained until 3000 msec after
the offset of the sentence. Participants were requested
to fixate on the wizard and to avoid eye blinks and
movements. After the sentence offset, there was a break
of 3000 msec to prevent any movement elicited by the
button press interfering with EEG recordings. After the
break, participants were instructed to judge the correct-
ness of the sentence by pressing one of two buttons on a
reaction panel, which was held with their hands during
the experiment. A sticker of a smiling face and a frown-
ing face, each corresponding to correctness and incor-
rectness, were attached to the buttons on the reaction
panel. The smiling and frowning faces were also shown
on the computer screen. They remained for 2000 msec.
This was also the maximum response time that was
given. The next trial started after a break of 1500 msec.
The sentences were presented in pseudorandomized
order whereby only three sentences of the same condi-
tion could appear in succession. Furthermore, the same
type of sentences (i.e., correct or incorrect sentences)
did not appear more than four times in a sequence.
Before the experiment started, a short introduction was
given, which explained that a wizard played a magic trick
and produced both correct and incorrect sentences.
Afterward, 15 practice trials were applied, to ensure that
the participant had understood the task before the
actual experiment took place.
Electroencephalogram Recording
ERP data were recorded from 22 Ag/AgCl electrodes
following the international 10-20 system ( Jasper, 1958),
placed in the positions FP1/2, F7/8, F3/4, Fz, FT9/10,
T3/4, C3/4, Cz, T5/6, P3/4, Pz, O1/2, and Oz. Blinks and
vertical eye movements were monitored by a bipolar
montage, using two electrodes that were placed above
and below the right eye. Similarly, by using a bipolar
montage again, the horizontal eye movements were
recorded from two electrodes, located at the outer left
and right canthus. The EEG signals were recorded with a
right mastoid reference. The activity over the left mas-
toid was recorded as well. All EEG recordings were re-
referenced off-line to the average of the EEG signals
recorded at the right and the left mastoid. The electrode
placed at AFz was used as ground. All electrode imped-
ances were kept below 5 k. The electrophysiological
signals were filtered with a bandpass filter from 0.16 to
30 Hz and digitized at a rate of 256 Hz.
Data Analyses
Separate analyses were conducted for the behavioral and
ERP data. The performance, reflecting the percentage of
correct responses per condition, was calculated separate-
ly for each participant. A selection criterion used here
was calculated based on the binomial distribution. There-
fore, the data with only the participants who performed
above a chance level, with at least 60% of correct trials,
were analyzed and are discussed in the following.
For the ERP analysis, only the trials with correct
responses were analyzed. As the first step, the standard
deviation of the mean amplitude within a 200-msec
sliding window was calculated. Whenever the standard
deviation exceeded a threshold of 40 AV, this part of the
EEG signal was automatically marked as artifact. After-
ward, the trials with artifact markings were inspected
and were rejected manually if in fact those trials had real
artifacts (e.g., jaw or head movements, disturbance of
EEG due to technical problems). As the second step,
trials with typical eye movements were marked and
corrected by applying an electrooculogram correction
tool (xeog, EEP software 3.1 for Unix; Nowak & Pfeifer,
Leipzig, Germany).
Table 2 provides the mean number of retained trials
for each condition and each group (dyslexic and control
1682 Journal of Cognitive Neuroscience Volume 18, Number 10
children). These mean numbers of trials were submitted
to a repeated measure analysis with the variables condi-
tion (serving as the within-subject factor) and group
(serving as the between-subject factor). There was no
reliable main effect of condition (F < 1) and no signif-
icant interaction (Condition Group interaction, F < 1).
The between-subject factor revealed a significant main
effect, F(1,30) = 5.16, MSE = 8.57, p < .05, due to a
better overall performance of the control participants.
Epochs of 1500 msec, beginning from the onset of the
critical past participle, were extracted from the contin-
uous EEG data. In addition, a baseline correction with
a 100-msec poststimulus onset baseline was applied
(Hahne & Friederici, 1999). Statistical evaluation of the
ERPs was performed on the mean amplitude for spe-
cific time windows (TWs) relative to the onset of the
past participle. The TWs were defined based on the re-
sults of previous studies concerning normally develop-
ing children (Hahne et al., 2004; Byrne et al., 1999;
Holcomb & Neville, 1991) and the visual inspection of
the grand averages from the present study with the re-
striction that the selected TWs do not cover more than
one component.
Four TWs (300–400, 400–650, 650–800, and 1000–
1500 msec) were chosen to compare semantically incor-
rect with correct sentences, and three different TWs
(100–300, 300–600, and 600–1300 msec) were chosen to
compare the sentences containing the syntactic violation
with those of the correct condition. Repeated measure
analyses of variance (ANOVAs) were performed for each
TW, following a hierarchical schema for the analysis. The
between-subject factor group (dyslexic children vs. con-
trol children) and three within-subject factors (condi-
tion: correct vs. incorrect; hemisphere: left vs. right;
region: anterior vs. posterior) were defined for analyses
of 16 lateral electrodes. The variables hemisphere and
region were completely crossed, yielding four regions of
interest, each of which had four electrodes: left anterior
(FP1, F7, F3, and FT9), right anterior (FP2, F8, F4, and
FT10), left posterior (C3, T5, P3, and O1), and right
posterior (C4, T6, P4, and O2). The analysis of the
midline electrodes included the between-subject factor
group (dyslexic children vs. control children) and the
within-subject variables condition (correct vs. incorrect)
and the four electrodes (Fz, Cz, Pz, and Oz).
1
Whenever
a significant interaction between the factors group and
condition and one of the topographical variables was
reached ( p < .10), a follow-up analysis for each group
on the topographical level (left anterior, right anterior,
left posterior, and right posterior) was performed.
Greenhouse-Geisser-corrected p values were reported
whenever a variable with more than two levels was
included in the statistical test.
To obtain further information about early auditory
perception, independent from prosodic, syntactic, and
lexical–semantic information, auditory evoked potentials
elicited at the onset of the sentence were analyzed. For
the auditory sentence onset potentials, latencies of the
peak amplitude were measured for each participant,
using an automatic peak detection within three distinct
TWs (P1 positive peak between 30 and 100 msec, N1 neg-
ative peak between 80 and 140 msec, and P2 positive
peak between 150 and 250 msec). Auditory sentence on-
set potentials were averaged over a 500-msec epoch with
a 100-msec prestimulus baseline. Ponton, Eggermont,
Khosla, Kwong, and Don (2002) showed that the P1–
N1–P2 complex is most prominent over central elec-
trodes. Therefore, the peak latencies at C3, C4, and Cz
were measured for each component. Paired t tests were
used to compare the peak latencies of the P1–N1–P2
complex for both groups of participants.
RESULTS
Behavioral Data
Phonological Awareness
Table 3 lists the means and standard deviations of the
scores (correct answers) obtained in the phoneme sub-
stitution task and the oddity task, separately for each
group.
2
The nonparametric Wilcoxon test (one-tailed) showed
a significant effect for the phoneme substitution task,
showing a better performance of the control children
over the dyslexic children (z = 2.94, p < .001). For the
oddity task, again a significant difference was observed
Table 2. Mean Number (±SD) of Remained Trials for the Control and the Dyslexic Children
Correct Semantically Incorrect Syntactically Incorrect All Conditions
Control children (n = 16) 27.9 ± 7.4 26.9 ± 7.0 27.4 ± 8.3 27.4 ± 7.2
Dyslexic children (n = 16) 22.3 ± 3.9 23.2 ± 4.8 22.7 ± 5.2 22.7 ± 4.1
Table 3. Means (±SD) of Performance of Correct Answers
for the Control and the Dyslexic Children
Performance of Correct Answers (%)
Phoneme Substitution Oddity Task
Control children
(n = 15)
98.5 ± 2.3 85.3 ± 18.6
Dyslexic children
(n = 15)
70.7 ± 32.3 65.3 ± 28.2
Sabisch et al. 1683
between the two groups (z = 1.92, p < .05), suggest-
ing that dyslexic children performed worse in the rhyme
detection task than the unimpaired control children.
The results suggest that children with developmental
dyslexia are deficient in the processing of segmental
phonological information.
Performance in the Grammaticality Judgment Task
Control children as well as children with developmental
dyslexia performed above chance level in the gram-
maticality judgment task. The results are presented in
Table 4. The ANOVA yielded a main effect of condition,
F(2,29) = 10.35, MSE = 23.7, p < .01, and a main effect
of group, F(1,30) = 11.55, MSE = 38.6, p < .01, but no
other significant differences. The main effect of group
reflects that the control children had a better overall
performance than the dyslexic participants. For the main
effect of condition, a follow-up analysis in a pairwise
comparison of the conditions independent of the group
was conducted, and it revealed a significant difference
between the semantic violation condition and the cor-
rect condition (correct condition: 91.9%; semantic vio-
lation condition: 96.0%, p < .001). This implies that
semantically incorrect sentences were more often cor-
rectly classified than sentences of the correct condition.
This might be because of a higher expectedness of the
sentence final element that is not fulfilled by the up-
coming element.
The analysis of the reaction times revealed no main
effect of condition (F < 1) or group (F < 1) and no
significant interaction (F < 1). This is not a surprising
result because the judgment was required 3000 msec
after the offset of the sentences, and it therefore does
not represent on-line sentence comprehension.
Event-related Brain Potential Data
Sentence Comprehension
The ERPs of the control children and the dyslexic chil-
dren evoked by the syntactic violation and the semantic
violation are presented in Figures 2 and 3, respectively.
The corresponding difference waves are shown in Fig-
ure 4. For each TW, an analysis with the between-subject
variable group (dyslexic children vs. control children)
and three within-subject variables (condition: correct
vs. incorrect; hemisphere: left vs. right; region: anterior
vs. posterior) was conducted. Whenever there was a sig-
nificant two- or three-way interaction between the var-
iable group and condition and one of the topographical
variables, a subsequent analysis was performed for each
group of children.
For the syntactic violation condition, a biphasic pat-
tern with an early anterior negativity commonly referred
to as ELAN followed by a P600 component was observed
for the control children. The anterior negativity showed
an early beginning, as reported for 13-year-old teenag-
ers and adults (Hahne et al., 2004), and sustained until
600 msec poststimulus onset. The anterior negativity for
the control children was bilaterally distributed. The P600
had its onset at 600 msec and remained until 1300 msec,
which is also in agreement with earlier findings (Hahne
et al., 2004). The results for the dyslexic children ex-
hibited some similarities with respect to the P600 but
also remarkable differences with respect to the anterior
negativity observed for the control children. The anteri-
or negativity for the dyslexic children was delayed start-
ing about 300 msec. The P600 displayed a similar pattern
for the dyslexic children as well as for the control
children.
For TW 1 (100–300 msec) and TW 2 (300–600 msec),
the analyses revealed a significant main effect of condi-
tion and a reliable three-way interaction of the variables
group, condition, and region for the lateral electrodes,
but no significant effect for the midline electrodes (see
Table 5). For TW 3 (600–1300 msec), a significant main
effect of condition was observed for the midline elec-
trodes, but no other main effect or interaction was
found for the lateral electrodes.
Following the three-way interaction between the vari-
ables group, condition, and region in TWs 1 and 2, analy-
ses were calculated with the variables condition (correct
vs. syntactically incorrect) and electrode for each region
(anterior: FP1/2, F7/8, F3/4, and FT9/10; posterior: C3/4,
T5/6, P3/4, and O1/2) and group (see Table 6). In TW 1,
for the control children, a reliable main effect of con-
dition was observed in the anterior region, reflecting
an early anterior negativity (see Table 6). There was
also a significant two-way interaction between condi-
tion and electrode. Follow-up analyses on the electrode
level revealed a reliable negativity for all anterior elec-
trodes. The interaction between condition and electrode
observed here is because of larger amplitudes of the
Table 4. Means (±SD) of Performance of Correct Answers for the Control and the Dyslexic Children in the EEG Experiment
Performance of Correct Answers (%)
Correct Sentences Semantically Incorrect Sentences Syntactically Incorrect Sentences
Control children (n = 16) 93.8 ± 4.3 97.6 ± 2.8 96.2 ± 3.2
Dyslexic children (n = 16) 89.8 ± 5.4 94.3 ± 4.0 90.6 ± 8.9
1684 Journal of Cognitive Neuroscience Volume 18, Number 10
negativity at frontal electrodes and smaller amplitudes at
the fronto-temporal electrodes for the violation condi-
tion. For the dyslexic children, no effect reached signif-
icance (see Table 6).
In TW 2, for the control children, a main effect of
condition for the anterior region was found, supporting
the presence of an anterior negativity. No such effect
was observed for the posterior region. A reliable two-
way interaction of condition and electrode was observed
for the anterior region for the same group of children. In
a subsequent follow-up analysis, a significant effect of
condition was found at all electrodes. This negativity
showed slight differences in amplitude across the differ-
ent electrodes and is more pronounced over frontal
electrodes compared with fronto-temporal electrodes.
For the dyslexic children, a reliable effect of condition
was observed for the anterior region. For the posterior
region, a marginally significant interaction between con-
dition and electrode was found. Further analyses on the
electrode level revealed a reliable effect of condition
only at the right occipital electrode (O2) and showed a
marginally significant difference in the conditions at the
right temporal electrode (T6).
In summary, for the syntactic violation condition, con-
trol children showed a reliable anterior negativity for
TW 1 (100–300 msec) and TW 2 (300–600 msec). In con-
Figure 2. Grand average ERPs of the control (left) and dyslexic children (right) in the syntactic violation condition. The syntactically
incorrect condition (dotted line) is plotted against the correct condition (solid line). The axis of the ordinates indicates the onset of the
critical word (past participle). Negative voltage is plotted upward. The gray hatched sections refer to TWs that were statistically tested
and revealed significance.
Sabisch et al. 1685
trast, the dyslexic children demonstrated a rather differ-
ent pattern by showing an anterior negativity only for
TW 2 (300–600 msec). A reliable late positivity for TW 3
(600–1300 msec) was observed for both the control and
the dyslexic children.
ERP studies on the development of language compre-
hension using similar syntactic constructions showed
that early syntactic processes, in particular, as reflected
by the early anterior negativity, develop gradually (Hahne
et al., 2004), such that an adultlike ELAN emerges only
in 13-year-olds, whereas 10-year-olds show a left ante-
rior negativity for the later TW. These results suggest
that substantial developmental changes occur within
an age range of 7–13 years. The participants of the pres-
ent study fall within this age range, and we therefore
sought to gain further insight into the differences ob-
served for the anterior negativity in the early and the
late TWs by conducting additional comparisons, analyz-
ing the left and right anterior regions for each group
and for each TW, separately. The following variables
were included in the analyses conducted below: con-
dition (correct vs. incorrect) and electrode (left ante-
rior: FP1, F7, F3, and FT9; right anterior: FP2, F8, F4, and
FT10).
The analyses of the left and the right anterior regions
in TW 1 for the control children revealed a significant
main effect of condition [left anterior: F(1,15) = 15.48,
MSE = 12.42, p < .01; right anterior: F(1,15) = 9.80,
MSE = 8.83, p < .01] and a reliable interaction between
electrode and condition for the left side only [left an-
terior: F(3,45) = 4.72, MSE = 1.28, p < .05, right an-
terior: F(3,45) = 2.11, MSE = 1.56, p = .15]. As already
described above, the interaction reflects amplitude dif-
ferences due to a larger negativity at frontal electrodes
compared with the fronto-temporal electrode. The re-
sults show a reliable early anterior negativity with a
bilateral distribution. In contrast, no significant effect
of condition, neither for the left nor the right anterior
region, was observed for the dyslexic children (left and
right anterior: F < 1). For the control children, in TW 2,
a similar pattern was observed as for the early TW,
yielding a negativity over left, F(1,15) = 28.80, MSE =
Figure 3. Grand average ERPs of the control children (left) and the dyslexic children (right) in the semantic violation condition. The
semantically incorrect condition (dotted line) is plotted against the correct condition (solid line). The axis of the ordinates indicates
the onset of the critical word (past participle). Negative voltage is plotted upward. The gray hatched sections refer to TWs that were
statistically tested and revealed significance.
1686 Journal of Cognitive Neuroscience Volume 18, Number 10
15.26, p < .001, and right frontal electrodes, F(1,15) =
27.07, MSE = 13.46, p < .001, and additionally an in-
teraction between condition and electrode for both re-
gions [left: F(3,45) = 6.21, MSE = 4.42, p < .01, right:
F(3,45) = 4.48, MSE =2.47,p < .05]. Again, the inter-
action refers to amplitude differences, with a larger neg-
ativity at frontal electrodes than at the fronto-temporal
electrode. A different pattern of results emerged for the
dyslexic children, showing a main effect of condition
only for the left anterior region, F(1,15) = 4.48, MSE =
23.10, p = .05, but not for the right anterior region,
F(1,15) = 2.79, MSE = 21.32, p = .12.
In summary, control children showed a bilaterally
distributed early anterior negativity (100–300 msec) as
well as a bilateral negativity between 300 and 600 msec.
Dyslexic children demonstrated a slightly different ERP
pattern, without any reliable effect for the early TW, and
an anterior negativity (300–600 msec) restricted only to
the left region.
For the semantic violation condition, descriptively,
both the control children and the dyslexic children
demonstrated an N400 effect. The N400 of both groups
showed a similar centro-parietal distribution. The con-
trol children additionally showed a broadly distributed
late positivity, which was not observed for the dyslexic
children.
For TW 1 (300–400 msec), no significant main effects
or interactions were observed (all p > .12). For TW 2
(400–650 msec) and TW 3 (650–800 msec), for both the
lateral and the midline electrodes, a significant main
effect of condition was found. These results reflect a
reliable N400 effect between 400 and 800 msec post-
stimulus onset. Crucially, there was no main effect of
group or any significant interaction (see Table 7) reflect-
ing a comparable N400 for the control children and the
dyslexic children. For TW 4 (1000–1500 msec), a signif-
icant main effect of condition and a Group Condition
interaction were found for the midline electrodes. For
Figure 4. Difference waves for the control children (left) and the dyslexic children (right) in the syntactic violation condition (upper row)
and the semantic violation condition (lower row). The difference waves were calculated by subtracting the grand average ERPs of the correct
condition from the incorrect conditions. The axis of the ordinates indicates the onset of the critical word (past participle). Negative voltage is
plotted upward. The gray hatched sections refer to TWs that were statistically tested and revealed significance.
Sabisch et al. 1687
the lateral electrodes, a marginally significant interac-
tion of group and condition was observed (see Table 7).
Subsequent analyses for each group, containing the
variables condition (correct vs. semantically incorrect)
and electrode (lateral electrodes: FP1/2, F7/8, F3/4,
FT9/10, C3/4, T5/6, P3/4, and O1/2; midline electrodes:
Fz, Cz, Pz, and Oz) were conducted (see Table 8). For
the control children, there was a main effect of condi-
tion for the midline electrodes, whereas a marginally sig-
nificant effect of condition and an interaction between
condition and electrode were observed for the lateral
electrodes. Follow-up analyses for the control children
on the electrode level revealed a reliable positivity for
the lateral electrodes F3, C3, C4, and P4 (all p .05)
and a marginally significant positivity for FP1, FP2, F4,
FT10, and O2 (all p .09). In contrast, there was no
Table 5. Global Analyses of ERP Data for Children of the Control and the Dyslexic Group in the Syntactic Violation
Condition Compared with the Correct Condition
Syntactic Violation Condition vs. Correct Condition
TW 1 (100–300 msec) TW 2 (300–600 msec) TW 3 (600–1300 msec)
Source df F MSE p F MSE p F MSE p
Lateral
Group 1,30 <1 <1 <1
Cond 1,30 5.41 6.56 <.05 6.15 10.48 <.05 <1
Group Cond 1,30 <1 1.39 14.61 .25 <1
Group Cond Reg 1,30 8.46 18.84
<
.01 5.83 16.12
<
.05 <1
Group Cond Hem 1,30 <1 <1 <1
Group Cond Reg Hem 1,30 1.69 .66 .20 1.60 1.49 .22 1.59 2.41 .22
Midline
Group 1,30 <1 1.40 64.80 .25 <1
Cond 1,30 1.62 13.39 .21 <1 11.20 25.46 <.01
Group Cond 1,30 <1 <1 <1
Group Cond Elec 3,90 1.74 4.63 .19 1.77 7.47 .18 <1
Significant effects are marked in boldface. Additional analyses were conducted for italicized results and reported in the text. Cond = condition;
Reg = region; Hem = hemisphere; Elec = electrode.
Table 6. Follow-up Analyses of ERP Data for Children of the Control Group and the Dyslexic Group Separately in the
Syntactic Violation Condition Compared with the Correct Condition
Syntactic Violation Condition vs. Correct Condition
TW 1 (100–300 msec) TW 2 (300–600 msec)
Group
Source df F MSE p F MSE p
Control
children
Anterior
Cond 1,15 14.22 18.87 <.01 36.13 22.20 <.001
Cond Elec 7,105 3.53 1.55
<
.05 4.32 3.89
<
.001
Posterior
Cond 1,15 <1 <1
Cond Elec 7,105 1.22 1.83 .31 1.67 4.61 .17
Dyslexic
children
Anterior
Cond 1,15 <1 4.74 33.74 <.05
Cond Elec 7,105 <1 <1
Posterior
Cond 1,15 <1 <1
Cond Elec 7,105 1.02 3.05 .39 2.43 6.19 .08
Significant effects are marked in boldface. Additional analyses were conducted for italicized results and reported in the text. Cond = condition;
Elec = electrode.
1688 Journal of Cognitive Neuroscience Volume 18, Number 10
significant main effect of condition or a significant inter-
action between condition and electrode for the dyslexic
children.
In summary, for the semantic violation condition, a
comparable N400 (400–800 msec) was observed for the
control and dyslexic children. For the control children, in
addition to the N400, a late positivity (1000–1500 msec)
was found over lateral and midline electrodes, whereas
such a positivity was absent for the dyslexic children.
Early Auditory Processing
The auditory potentials evoked by the sentence onset
showed a similar P1–N1–P2 complex with respect to
Table 7. Global Analyses of ERP Data for Children of the Control Group and the Dyslexic Group in the Semantic Violation
Condition Compared with the Correct Condition
Semantic Violation Condition vs. Correct Condition
TW 1
(300–400 msec)
TW 2
(400–650 msec)
TW 3
(650–800 msec)
TW 4
(1000–1500 msec)
Source df F MSE p F MSE p F MSE p F MSE p
Lateral
Group 1,30 1.53 25.93 .23 1.04 39.05 .32 <1 <1
Cond 1,30 2.58 18.82 .12 10.98 15.96 <.01 17.59 11.52 <.001 <1
Group Cond 1,30 <1 <1 <1 3.21 18.53 .08
Group Cond Reg 1,30 1.68 6.28 .20 <1 <1 <1
Group Cond Hem 1,30 <1 <1 <1 <1
Group Cond Reg Hem 1,30 <1 <1 <1 <1
Midline
Group 1,30 1.24 42.34 .27 1.57 53.51 .22 <1 <1
Cond 1,30 1.86 32.94 .18 18.97 27.50 <.001 24.17 21.90 <.001 4.50 22.04 .04
Group Cond 1,30 <1 <1 <1 5.39 22.04 .03
Group Cond Elec 3,90 1.12 5.95 .33 <1 <1 <1
Significant effects are marked in boldface. Additional analyses were conducted for italicized results and reported in the text. Cond = condition;
Reg = region; Hem = hemisphere; Elec = electrode.
Table 8. Follow-up Analyses of the ERP Data for Children of the Control Group and the Dyslexic Group Separately in the
Semantic Violation Condition Compared with the Correct Condition
Semantic Violation Condition vs. Correct Condition
TW 4 (1000–1500 msec)
Group
Source df F MSE p
Control
children
Lateral
Cond 1,15 3.10 61.87 .10
Cond Elec 15,225 2.44 5.94 .07
Midline
Cond 1,15 12.76 17.06 <.01
Cond Elec 3,45 1.25 7.61 .30
Dyslexic
children
Lateral
Cond 1,15 <1
Cond Elec 15,225 <1
Midline
Cond 1,15 <1
Cond Elec 3,45 1.11 6.48 .34
Significant effects are marked in boldface. Additional analyses were conducted for italicized results and reported in the text. Cond = condition;
Elec = electrode.
Sabisch et al. 1689
latency for the control and dyslexic children. The peak
latencies for the P1, N1, and P2 at the electrodes C3, C4,
and Cz are reported in Table 9. Pairwise t tests did not
reveal any significant difference between the groups (all
p .09).
DISCUSSION
Utilizing ERP measures, the present study investigated
auditory sentence comprehension in children with de-
velopmental dyslexia and control children. We posed
the question whether children with developmental dys-
lexia differ in the syntactic processing and if so, whether
this is because of differences in prosodic processes that
have an impact on the processing of syntactic informa-
tion. We further tested whether the differences between
control children and children with developmental dys-
lexia are specific to the domains of prosody and syntax
or hold also for the lexical–semantic domain. In addi-
tion, we examined both the level of phonological encod-
ing (i.e., segmental phonological information) and the
level of prosodic encoding (i.e., suprasegmental phono-
logical information), as well as early auditory processing
and its input on phonological processing. The compar-
ison between the phonological processes and processes
of early auditory perception should help to disentangle
whether a phonological deficit or an auditory temporal
deficit can be observed in children with developmental
dyslexia. In the following, we first discuss the findings
(behavioral results and the ERP results) from sentence
comprehension. Second, we discuss the results for the
processing of segmental and prosodic phonological in-
formation. Third, we discuss the findings for the early
auditory processing.
Sentence Comprehension
Behavioral Data
In the grammaticality judgment task, children with devel-
opmental dyslexia and control children both performed
well, the control children however showing a better
performance, independently of the conditions. The fact
that children with developmental dyslexia demonstrated
poorer language comprehension abilities points toward
a language comprehension deficit for at least some of
these children. This conclusion is consistent with previ-
ous findings that focused on different aspects of lan-
guage comprehension in children with developmental
dyslexia (Gallagher, Frith, & Snowling, 2000; McArthur,
Hogben, Edwards, Heath, & Mengler, 2000).
Both control children and children with developmen-
tal dyslexia detected the semantic violation more often
than the syntactic violation. This suggests that the pro-
cessing of semantic information is relatively intact. More-
over, correct sentences were less often recognized as
correct than sentences with the semantic violation were
identified as correct, which could be because of the fact
that sentences with a violation are very rare in every-
day conversation, therefore, catching children’s atten-
tion more often than correct sentences. Being in the
experimental setting, children might expect the presen-
tation of sentences with a violation more often than the
presentation of sentences without a violation.
Syntactic Violation Condition
For the syntactic violation condition, a combined pattern,
comprising a bilaterally distributed early anterior negativ-
ity and a P600, was observed only for the control children.
The children with developmental dyslexia showed a
slightly different pattern containing a delayed anterior
negativity restricted to the left hemisphere and a P600
component. In the following, these findings will be dis-
cussed in detail for each of the selected TW separately.
Converging with findings from Hahne et al. (2004),
a bilaterally distributed early anterior negativity (100–
300 msec) was observed for the control children. In
contrast, for children with developmental dyslexia, no
early anterior negativity was observed within this TW. The
ELAN is assumed to reflect early processes of phrase
structure building based on word category information,
and Hahne and Friederici (1999) showed that these pro-
cesses are highly automatic. Critically, the ELAN has also
been reported for healthy children aged 2 years 8 months
in simple active sentences (Oberecker et al., 2005). This
suggests that in these healthy children, a neural process-
ing system, as observed in adults, has been established for
the processing of default structures in active sentences,
but that it takes some years before it is applied in an
adultlike manner to more complex syntactic structures,
such as passive sentences (Hahne et al., 2004).
Table 9. Averages (±SD) of Peak Latencies (Milliseconds) of the P1, N1, and P2 for Control and Dyslexic Children
Control Children (n
=
16) Dyslexic Children (n
=
16)
C3 C4 Cz C3 C4 Cz
P1 70.7 (17.9) 76.0 (18.8) 71.2 (20.0) 69.7 (16.8) 67.7 (16.8) 69.0 (16.5)
N1 116.7 (16.1) 121.0 (9.7) 114.7 (11.6) 117.7 (12.5) 118.2 (13.1) 112.7 (12.1)
P2 203.7 (18.8) 202.0 (13.9) 198.2 (11.0) 203.7 (21.2) 206.0 (15.9) 206.7 (16.6)
1690 Journal of Cognitive Neuroscience Volume 18, Number 10
Functionally, a right fronto-temporal network has
been associated with the processing of prosodic infor-
mation (Friederici & Alter, 2004). This idea of a right
inferior frontal involvement in the detection of prosodic
anomalies received further support by a recent ERP
study (Eckstein & Friederici, 2005), where a RAN was
elicited for a prosodic incongruity, in which the sen-
tence final word possessed the prosody of the penulti-
mate word. The RAN was observed for sentences either
containing a pure prosodic violation or a combined
syntactic and prosodic violation, leading to the conclu-
sion that the RAN reflects the processing of prosody
(Eckstein & Friederici, 2005). Furthermore, Eckstein and
Friederici (in press) report that prosodic processes can
influence early syntactic processes as evidenced by an
involvement of the right anterior region.
The notion that processes reflecting syntactic phrase
structure building are linked to left anterior regions and
that prosodic processes are linked to right anterior
regions is further supported by studies investigating
phrase structure violations either using naturally pro-
duced sentences or sentences in which a constituent has
been spliced out. A bilaterally distributed early anterior
negativity has been reported for studies with auditorily
presented sentences in which a constituent had been
spliced out. Although a careful splicing procedure was
applied, which controlled, for example, for local co-
articulation, a prosodic incongruity was created as evi-
denced by the elicited ERP responses (Kubota, Ferrari, &
Roberts, 2004; Hahne & Friederici, 1999, 2002; Kno¨sche,
Maess, & Friederici, 1999). These results are in agree-
ment with findings of the current study. Children of the
control group showed also a bilaterally distributed early
anterior negativity suggesting that both prosodic infor-
mation and syntactic phrase structure information are
processed.
In contrast, studies that presented naturally produced
sentences without a prosodic incongruity reported an
early anterior negativity that was more left-lateralized
(Friederici, Gunter, Hahne, & Mauth, 2004; Friederici,
Hahne, & Mecklinger, 1996, Experiment 1). Similarly,
a more left-lateralized early negativity was observed
for fast visually presented stimuli (Gunter, Friederici,
& Hahne, 1999; Friederici et al., 1996, Experiment 2).
These findings suggest that there is less involvement of
the right hemisphere when no prosodic incongruity is
presented and when the sentences are not auditorily
presented, so that prosodic information is indirectly
available. Brain imaging experiments reported a right
hemispheric activation for the processing of supraseg-
mental information (Gandour et al., 2004; Meyer, Alter,
Friederici, Lohmann, & von Cramon, 2002). These find-
ings imply that the right anterior hemisphere supports
the processing of prosodic information and that the
left hemisphere supports the processing of syntac-
tic information. Overall, the findings from the present
study indicate that in the early TW, the control chil-
dren activated left and right anterior regions during sen-
tence comprehension, which we attribute to processes
of phrase structure building and prosodic encoding
respectively. Crucially, however, children with develop-
mental dyslexia did not pattern with the control children
and showed no effects over anterior electrode sites.
Within the later TW (300–600 msec), a bilaterally distrib-
uted late anterior negativity was found for the control
children. In contrast, for the children with developmental
dyslexia, a late anterior negativity that was restricted to
the left anterior region was observed. The appearance
of the late anterior negativity observed for the control
group might suggest that enhanced processing re-
sources are still required during the comprehension of
the passive constructions (but see the work of Hahne
et al., 2004). More crucial for the present research
question is the observation that the left anterior nega-
tivity that is found within the later TW for the children
with developmental dyslexia might reflect syntactic
phrase structure building processes that are not devel-
oped to the same degree as generally expected for their
age. This idea is based on findings from younger nor-
mally developing children (7–10 years old) who showed,
similar to the children with developmental dyslexia, a
left anterior negativity for the later TW but not for the
early TW (Hahne et al., 2004). Consistent with this idea
is that children with developmental dyslexia performed
relatively well in detecting the syntactic violation, which
means that they process syntactic phrase structure in-
formation, but not automatically and as early as control
children of the same age. Finally, the absence of the right
anterior negativity (RAN) in children with developmental
dyslexia suggests that prosodic information might not
be used in the same manner as by control children.
Turning to TW 3, as hypothesized, for both the con-
trol children and children with developmental dyslexia,
a P600 component (600–1300 msec) was observed in
addition to the anterior negativity. These P600 com-
ponents had a similar distribution in the two groups
and were mostly prominent over parietal and occipital
regions. With respect to the phrase structure violation
condition tested in this study, the P600 component gen-
erally reflects processes of syntactic reanalysis and re-
pair that emerge during a late processing phase after
the parser has detected a syntactic violation (Friederici,
2002). These processes are assumed to be highly con-
trolled (Hahne & Friederici, 1999). The presence of the
P600 thus indicates that both children with developmen-
tal dyslexia and control children are sensitive to the
syntactic violation. This is in line with recent develop-
mental findings that showed that the P600 is present in
normally developing 2-year-old children, when simple
active sentences were used with a syntactic violation
(Oberecker et al., 2005; Silva-Pereyra, Klarman, Lin, &
Kuhl, 2005). Furthermore, for the processing of complex
passive structures, as presented in our study, a P600 has
been reported in normally developing children from age
Sabisch et al. 1691
7 years onward (Hahne et al., 2004). The emergence of
the P600 further reveals that late controlled processes
of revision were initiated and carried out in a similar
manner in the two groups and therefore suggests that
these processes are also present in children with devel-
opmental dyslexia.
Semantic Violation Condition
The ERP data for the semantic processing domain reveal
that both children with developmental dyslexia and
control children registered similar N400 components
(between 400 and 800 msec) with comparable distribu-
tions. Our results indicate that the processing of seman-
tic anomalies is carried out in a similar manner by both
groups during auditory sentence comprehension. Com-
parable results, showing a similar N400 component for
both the dyslexic and the control children, were ob-
tained in an alliteration priming task that utilized words
and nonwords (Bonte & Blomert, 2004) and a categori-
zation task using figures and words (Silva-Pereyra et al.,
2003). The present results, however, differ from findings
of a study conducted by Neville et al. (1993), which used
a visual sentence comprehension paradigm and tested
the semantic integration of the final word of the sen-
tence into the preceding context in language-impaired
children and unimpaired control children. In this study,
the N400 of the language-impaired children had a larg-
er amplitude compared with the N400 of the control
children. The following is our explanation for the dif-
ference between the study of Neville et al. and the
study presented in the current article. First, the children
participating in the present study were on average
2 years older than the children involved in the study
by Neville et al. Second, the present study was con-
ducted using the auditory modality, testing semantic
processes independent of reading, whereas Neville
et al. utilized the visual modality. As Neville et al. pointed
out, a considerable portion of the language-impaired
children also develop signs of reading impairment, and
as it turned out, ‘‘only 12 of the language-impaired
subjects could read well enough to do the visual sen-
tence task’’ (p. 250). Therefore, the increased am-
plitude of the N400 could be a consequence of the
reading requirement, which is more demanding for the
language-impaired children.
Interestingly, the semantic violation condition elicited
in addition to the N400 a late, broadly distributed
positivity between 1000 and 1500 msec for the control
children but not for the children with developmental
dyslexia. The late positivity could be caused by the ex-
perimental setting and the task requirement. Performing
a similar judgment task yielded an N400 followed by
a P600 (650–850 msec) for adults (Kolk, Chwilla, van
Herten, & Oor, 2003). In the first experiment of this
study among other conditions, sentences with a selec-
tional restriction violation and acceptable sentences
were compared. Participants had to judge the overall
correctness of the sentences. To control the impact of
the task, a second experiment was conducted in which
participants had to read the sentences and to perform
a probe verification task to ensure that they paid atten-
tion to the test sentences. In contrast to the first ex-
periment, the P600 was absent, and only an N400 was
observed. Kolk et al. (2003) interpreted the presence of
the P600 in their first experiment as a consequence of
the judgment task. However, the positivity observed in
our study appeared later between 1000 and 1500 msec,
which could be due to latency differences in the ERP
components between children and adults. Similarly, a
biphasic pattern containing an N400 and a late positivity
was reported by Ors et al. (2001) and Holcomb et al.
(1992). Again, in both studies, adults and children had to
judge the congruency of a word with respect to its
preceding context.
Segmental and Suprasegmental Information
As hypothesized, children with developmental dyslexia
performed worse in the phoneme substitution task and
also in the rhyme detection task. These findings are
in agreement with the phonological deficit hypothesis
(Goswami, 2003), which predicts an impairment in the
processing of segmental phonological information. Fur-
thermore, the absence of the RAN in the sentence com-
prehension task is here interpreted as an indication that
children with developmental dyslexia do not rely on su-
prasegmental information in the way that normal chil-
dren do. These findings reveal that the processing of
both segmental and suprasegmental information differs
in the children with developmental dyslexia and the con-
trol children. The present results therefore confirm the
phonological deficit hypothesis and critically extend it to
the assumption that the processing of suprasegmental
information is also affected in children with developmen-
tal dyslexia.
Early Auditory Processing
For the auditory sentence onset potentials (P1–N1–P2
complex), similar peak latencies were found for both
children with developmental dyslexia and control chil-
dren. These results suggest that early auditory percep-
tion might not be impaired in our group of children
with developmental dyslexia. These results are consist-
ent with the findings of Neville et al. (1993). They found
that the auditory ERP component (N140) was delayed
in latency and reduced in its amplitude in the sub-
group of language-impaired/reading-disabled children
who scored poorly on the Tallal Auditory Repetition
Test (Tallal, 1978), but not in the subgroup of language-
impaired/reading-disabled children that scored within
the normal range on the same test. In their study, the
1692 Journal of Cognitive Neuroscience Volume 18, Number 10
same group of language-impaired/reading-disabled chil-
dren was also divided into two groups, based on their
performance on the Cycle Test of Syntax, to investigate
whether there is any correlation in their impairment
between auditory processing and the syntactic pro-
cessing. Interestingly, they found that the language-
impaired/reading-disabled children who performed
poorly on the auditory repetition task and exhibited
a delayed and reduced N140 did not necessarily dem-
onstrate impaired syntactic processes. Thus our find-
ings, just as those of Neville et al. suggest that there is
not obligatorily a functional link between the syntactic
processing and the early auditory processing. Rather,
the present results indicate an impairment in dyslexic
children for the processing of suprasegmental auditory
information.
Conclusion
The present study investigated ERPs to compare audi-
tory sentence comprehension in 9- to 12-year-old chil-
dren with developmental dyslexia and their controls,
which were matched on age, sex, and nonverbal intelli-
gence. Semantic processes as indicated by the N400
appeared to be very similar in both groups. Syntactic
processes appeared to be different with respect to the
early anterior negativity but similar with regard to the
P600 component. The results for the P600 suggest com-
parable controlled processes of syntactic reanalysis. For
the control children, automatic processes of syntactic
structure building, as represented by the early left ante-
rior negativity (ELAN), and the formation of prosodic
structure, reflected in the RAN, were observed. In con-
trast, the dyslexic children showed a delayed anterior
negativity (300–600 msec) compared with their age-
matched controls. It could be assumed that the left
anterior negativity serves as a precursor for the ELAN,
because dyslexic children were able to detect the syn-
tactic violation in the behavioral task and performed on
a high level. The absence of the RAN suggests that
suprasegmental (prosodic) information, which provides
an important cue in the acquisition of syntactic informa-
tion, is not processed in the same manner as in un-
impaired control children. Thus, the present findings
strengthen the view that developmental dyslexia is
associated with a phonological deficit that might hamper
the acquisition of automatic syntactic processes.
Acknowledgments
We thank Masako Hirotani, Petra Burkhardt, and three anon-
ymous reviewers, for their constructive comments on earlier
versions of this article, and Monika Hage, Elisabeth Greiner,
and Ulrike Barth, for their help with data collection. We also
thank all children and their parents for participating in this
study. This study was supported by the Deutsche Forschungs-
gemeinschaft (German Research Foundation, DFG) grant SU
135/3-1 awarded to W. v. S.
Reprint requests should be sent to Angela D. Friederici, Max
Planck Institute for Human Cognitive and Brain Sciences, P.O.
Box 500 355, 04303 Leipzig, Germany, or via e-mail: angelafr@
cbs.mpg.de.
Notes
1. Note that the present study was done in a clinical setting.
Therefore, the two lateral electrodes (T3 and T4) were used to
check for signs of epileptic seizure, and the EEG recorded by
those electrodes were not included in the ERP analyses.
2. Note that one pair of children (i.e., one dyslexic child and
his matched control) was excluded from the analyses of the
phonological awareness task due to a technical error.
REFERENCES
Achenbach, T. M. (1998). CBCL/4–18 Elternfragebogen u¨ber
das Verhalten von Kindern und Jugendlichen [Child
Behavior Checklist (CBCL/4–18)]. Cologne: Arbeitsgruppe
Kinder-, Jugend- und Familiendiagnostik (KJFD).
(Original work published 1991).
Berwanger, D. (2002). Untersuchung der zeitlichen
Diskriminationsfa¨higkeit bei Kindern mit einer
Sprachentwicklungssto¨rung und—oder Lese-
Rechtschreibsto¨rung. Munich: Verlag Dr. Hut.
Birkel, P. (1994a). Weingartener Grundwortschatz
Rechtschreib-Test fu¨r dritte und vierte Klassen
(WRT 3
+
) [Weingarten basic vocabulary orthography
test for 3rd- and 4th-graders]. Go¨ttingen: Hogrefe.
Birkel, P. (1994b). Weingartener Grundwortschatz
Rechtschreib-Test fu¨r zweite und dritte Klassen (WRT 2
+
)
[Weingarten orthography test for 2nd- and 3rd-graders].
Go¨ttingen: Hogrefe.
Bonte, M. L., & Blomert, L. (2004). Developmental dyslexia:
ERP correlates of anomalous phonological processing
during spoken word recognition. Brain Research,
Cognitive Brain Research, 21, 360–376.
Bradley, L., & Bryant, P. (1985). Rhyme and reason in
reading and spelling. Ann Arbor, MI: University of
Michigan Press.
Byrne, J. M., Connolly, J. F., MacLean, S., Dooley, J. M.,
Gordon, K. E., & Beattie, T. L. (1999). Brain activity
and language assessment using event-related potentials:
Development of a clinical protocol. Developmental
Medicine and Child Neurology, 41, 740–747.
Davis, C. J., Gayan, J., Knopik, V. S., Smith, S. D., Cardon,
L. R., Pennington, B. F., Olson, R. K., & DeFries, J. C.
(2001). Etiology of reading difficulties and rapid naming:
The Colorado twin study of reading disability. Behavior
Genetics, 31, 625–635.
Dilling, H., Mombour, W., & Schmidt, M. H. (1993).
International classification of mental diseases, ICD-10
(German edition). Bern: Huber.
Eckstein, K., & Friederici, A. (2005). Late interaction of
syntactic and prosodic processes in sentence
comprehension as revealed by ERPs. Brain Research,
Cognitive Brain Research, 25, 130–143.
Eckstein, K., & Friederici, A. D. (2006). It’s early: ERP
evidence for initial interaction of syntax and prosody
in speech comprehension. Journal of Cognitive
Neuroscience, 18, 1696–1711.
Esser, G., Wyschkon, A., & Schmidt, M. H. (2002). Long-term
outcome in 8-year-old children with specific reading
retardation: Results at age 25 years. Zeitschrift fu¨r
Klinische Psychologie und Psychotherapie, 31, 235–242.
Sabisch et al. 1693
Fisher, S. E., & DeFries, J. C. (2002). Developmental
dyslexia: Genetic dissection of a complex cognitive trait.
Nature Reviews Neuroscience, 3, 767–780.
Fisher, S. E., Francks, C., Marlow, A. J., MacPhie, I. L.,
Newbury, D. F., Cardon, L. R., Ishikawa-Brush, Y.,
Richardson, A. J., Talcott, J. B., Gayan, J., Olson, R. K.,
Pennington, B. F., Smith, S. D., DeFries, J. C., Stein, J. F.,
& Monaco, A. P. (2002). Independent genome-wide
scans identify a chromosome 18 quantitative-trait locus
influencing dyslexia. Nature Genetics, 30, 86–91.
Friederici, A. D. (2002). Towards a neural basis of auditory
sentence processing. Trends in Cognitive Sciences, 6,
78–84.
Friederici, A. D., & Alter, K. (2004). Lateralization of auditory
language functions: A dynamic dual pathway model.
Brain and Language, 89, 267–276.
Friederici, A. D., Gunter, T. C., Hahne, A., & Mauth, K. (2004).
The relative timing of syntactic and semantic processes
in sentence comprehension. NeuroReport, 15, 165–169.
Friederici, A. D., Hahne, A., & Mecklinger, A. (1996).
Temporal structure of syntactic parsing: Early and late
event-related brain potential effects. Journal of
Experimental Psychology: Learning, Memory, and
Cognition, 22, 1219–1248.
Gallagher, A., Frith, U., & Snowling, M. J. (2000). Precursors
of literacy delay among children at genetic risk of dyslexia.
Journal of Child Psychology and Psychiatry, and Allied
Disciplines, 41, 203–213.
Gandour, J., Tong, Y., Wong, D., Talavage, T., Dzemidzic, M.,
Xu, Y., Li, X., & Lowe, M. (2004). Hemispheric roles in
the perception of speech prosody. Neuroimage, 23,
344–357.
Georgiewa, P., Gru¨nling, C., Ligges, M., Filz, C., Moller, U., &
Blanz, B. (2004). Age dependent changes in phonological
reading in dyslexia. Zeitschrift fu¨r Klinische Psychologie
und Psychotherapie, 33, 281–289.
Goswami, U. (2003). Phonology, learning to read and dyslexia:
A cross-linguistic analysis. In V. Csepe (Ed.), Dyslexia:
Different brain, different behavior (Vol. 23, pp. 1–40).
New York: Kluwer Academic Publishers.
Grund, M., Haug, G., & Naumann, C. L. (1995). DRT 5
Diagnostischer Rechtschreibtest fu¨r 5. Klassen [Diagnostic
orthography test for 5th-graders]. Weinheim: Beltz.
Grund, M., Haug, G., & Naumann, C. L. (1998). DRT 4
Diagnostischer Rechtschreibtest fu¨r 4. Klassen [Diagnostic
orthography test for 4th graders]. Go¨ttingen: Beltz.
Gunter, T. C., Friederici, A. D., & Hahne, A. (1999). Brain
responses during sentence reading: Visual input affects
central processes. NeuroReport, 10, 3175–3178.
Habib, M. (2000). The neurological basis of developmental
dyslexia: An overview and working hypothesis. Brain,
123, 2373–2399.
Hagoort, P., Brown, C. M., & Osterhout, L. (1999). The
neurocognition of syntactic processing. In C. M. Brown
& P. Hagoort (Eds.), The neurocognition of language
(pp. 273–316). New York: Oxford University Press.
Hahne, A., Eckstein, K., & Friederici, A. D. (2004). Brain
signatures of syntactic and semantic processes during
children’s language development. Journal of Cognitive
Neuroscience, 16, 1302–1318.
Hahne, A., & Friederici, A. D. (1999). Electrophysiological
evidence for two steps in syntactic analysis: Early
automatic and late controlled processes. Journal of
Cognitive Neuroscience, 11, 194–205.
Hahne, A., & Friederici, A. D. (2002). Differential task
effects on semantic and syntactic processes as revealed
by ERPs. Brain Research, Cognitive Brain Research, 13,
339–356.
Hirsh-Pasek, K., Kemler Nelson, D. G., Jusczyk, P. W., Cassidy,
K. W., Druss, B., & Kennedy, L. (1987). Clauses are
perceptual units for young infants. Cognition, 26, 269–286.
Ho¨hle, B., & Weissenborn, J. (2003). German-learning infants’
ability to detect unstressed closed-class elements in
continuous speech. Developmental Science, 6, 122–127.
Holcomb, P. J., Coffey, S. A., & Neville, H. J. (1992). Visual
and auditory sentence processing: A developmental analysis
using event-related brain potentials. Developmental
Neuropsychology, 8, 203–241.
Holcomb, P. J., & Neville, H. J. (1991). Natural speech
processing: An analysis using event-related brain
potentials. Psychobiology, 19, 286–300.
Jasper, H. H. (1958). The ten-twenty electrode system
of the International Federation. Journal of
Electroencephalography and Clinical Neurophysiology,
10, 371–375.
Jime´nez, J. E., Garcı´a, E., Este´vez, A., Dı´az, A., Ortiz, R., Rodrigo,
M., Guzma´n, R., Herna´ndez-Valle, I., & Herna´ndez, S.
(2004). An evaluation of syntactic-semantic processing in
developmental dyslexia. Electronic Journal of Research
in Educational Psychology, 2, 127–142.
Jusczyk, P. W. (2002). How infants adapt speech-processing
capacities to native language structure. Current
Directions in Psychological Science, 11, 15–18.
Kaufman, A. S., & Kaufman, N. L. (1991). K-ABC Kaufman—
Assessment Battery for Children [Kaufman Assessment
Battery for Children (K-ABC)] (P. Melchers & U. Preuss,
German version). Amsterdam: Swets & Zeitlinger.
(Original work published 1983).
Kno¨sche, T. R., Maess, B., & Friederici, A. D. (1999). Processing
of syntactic information monitored by brain surface
current density mapping based on MEG. Brain
Topography, 12, 75–87.
Kolk, H. H. J., Chwilla, D. J., van Herten, M., & Oor, P. J. W.
(2003). Structure and limited capacity in verbal working
memory: A study with event-related potentials. Brain and
Language, 85, 1–36.
Kubota, M., Ferrari, P., & Roberts, T. P. L. (2004). Human
neuronal encoding of English syntactic violations as
revealed by both L1 and L2 speakers. Neuroscience
Letters, 368, 235–240.
Kutas, M., & Federmeier, K. D. (2000). Electrophysiology
reveals semantic memory use in language comprehension.
Trends in Cognitive Sciences, 4, 463–470.
Landerl, K., Linortner, R., & Wimmer, H. (1992). Phonological
awareness and spelling acquisition in German language.
Zeitschrift fu¨r Pa¨dagogische Psychologie, 6, 17–33.
Landerl, K., Wimmer, H., & Frith, U. (1997). The impact of
orthographic consistency on dyslexia: A German-English
comparison. Cognition, 63, 315–334.
Linder, M., & Grissemann, H. (1980). Zu¨rcher Lesetest
(ZLT) [Zu¨rcher reading test]. Bern: Huber.
McArthur, G. M., Hogben, J. H., Edwards, V. T., Heath, S. M.,
& Mengler, E. D. (2000). On the ‘‘specifics’’ of specific
reading disability and specific language impairment.
Journal of Child Psychology and Psychiatry, and Allied
Disciplines, 41, 869–874.
Meyer, M., Alter, K., Friederici, A. D., Lohmann, G., &
von Cramon, D. Y. (2002). FMRI reveals brain regions
mediating slow prosodic modulations in spoken
sentences. Human Brain Mapping, 17, 73–88.
Mu¨ller, R. (1997). DRT 3 Diagnostischer Rechtschreibtest 3
[Diagnostic orthography test for 3rd graders].
Go¨ttingen: Beltz.
Muter, V., & Snowling, M. (1997). Grammar and phonology
predict spelling in middle childhood. Reading & Writing,
9, 407–425.
1694 Journal of Cognitive Neuroscience Volume 18, Number 10
Nation, K., & Snowling, M. J. (2000). Factors influencing
syntactic awareness skills in normal readers and poor
comprehenders. Applied Psycholinguistics, 21, 229–241.
Neville, H. J., Coffey, S. A., Holcomb, P. J., & Tallal, P. (1993).
The neurobiology of sensory and language processing in
language-impaired children. Journal of Cognitive
Neuroscience, 5, 235–253.
Oberecker, R., Friedrich, M., & Friederici, A. D. (2005).
Neural correlates of syntactic processing in two-year-olds.
Journal of Cognitive Neuroscience, 17, 1667–1678.
Ors, M., Lindgren, M., Berglund, C., Hagglund, K., Rosen, I.,
& Blennow, G. (2001). The N400 component in parents
of children with specific language impairment. Brain and
Language, 77, 60–71.
Pennington, B. F., Van Orden, G. C., Smith, S. D., Green,
P. A., & Haith, M. M. (1990). Phonological processing
skills and deficits in adult dyslexics. Child Development,
61, 1753–1778.
Ponton, C., Eggermont, J. J., Khosla, D., Kwong, B., &
Don, M. (2002). Maturation of human central auditory
system activity: Separating auditory evoked potentials by
dipole source modeling. Clinical Neurophysiology, 113,
407–420.
Ramus, F. (2001). Outstanding questions about phonological
processing in dyslexia. Dyslexia: The Journal of the
British Dyslexia Association, 7, 197–216.
Ramus, F., Rosen, S., Dakin, S. C., Day, B. L., Castellote, J. M.,
White, S., & Frith, U. (2003). Theories of developmental
dyslexia: Insights from a multiple case study of dyslexic
adults. Brain, 126, 841–865.
Rathenow, P. (1979). WRT 4/5 Westermann Rechtschreibtest
4/5 [Westermann orthography test 4/5]. Braunschweig:
Westermann.
Rathenow, P., Vo¨ge, J., & Laupenmu¨hlen, D. (1980).
WRT 6
+
Westermann Rechtschreibtest 6
+
[Westermann
Orthography Test 6+]. Braunschweig: Westermann.
Rego, L. L. B., & Bryant, P. E. (1993). The connection between
phonological, syntactic and semantic skills and children’s
reading and spelling. European Journal of Psychology
of Education, 8, 235–246.
Shaywitz, S. E., Shaywitz, B. A., Fletcher, J. M., & Escobar, M. D.
(1990). Prevalence of reading disability in boys and girls.
Results of the Connecticut Longitudinal Study. Journal
of the American Medical Association, 264, 998–1002.
Silva-Pereyra, J., Rivera-Gaxiola, M., Fernandez, T., Diaz-Comas,
L., Harmony, T., Fernandez-Bouzas, A., Rodriguez, M.,
Bernal, J., & Marosi, E. (2003). Are poor readers
semantically challenged? An event-related brain potential
assessment. International Journal of Psychophysiology,
49, 187–199.
Silva-Pereyra, J. F., Klarman, L., Lin, L. J. F., & Kuhl, P. K.
(2005). Sentence processing in 30-month-old children:
An event-related potential study. NeuroReport, 16,
645–648.
Snow, D., & Balog, H. L. (2002). Do children produce the
melody before the words? A review of developmental
intonation research. Lingua, 112, 1025–1058.
Snowling, M., Nation, K., Moxham, P., Gallagher, A., &
Frith, U. (1997). Phonological processing skills of dyslexic
students in higher education: A preliminary report.
Journal of Research in Reading, 20, 31–41.
Steinhauer, K., Alter, K., & Friederici, A. D. (1999). Brain
potentials indicate immediate use of prosodic cues in natural
speech processing. Nature Neuroscience, 2, 191–196.
Tallal, P. (1978). An experimental investigation of the role of
auditory temporal processing in normal and disordered
language development. In A. Caramazza & E. Zurif (Eds.),
Language acquisition and language breakdown: Parallels
and divergences (pp. 25–61). Baltimore, MD: The John
Hopkins University Press.
Tallal, P. (1980). Auditory temporal perception, phonics,
and reading disabilities in children. Brain and Language,
9, 182–198.
Tallal, P., Miller, S., & Fitch, R. H. (1993). Neurobiological
basis of speech: A case for the preeminence of temporal
processing. Annals of the New York Academy of Sciences,
682, 27–47.
Tallal, P., & Piercy, M. (1973). Developmental aphasia:
Impaired rate of non-verbal processing as a function
of sensory modality. Neuropsychologia, 11, 389–398.
Tallal, P., & Piercy, M. (1974). Developmental aphasia: Rate
of auditory processing and selective impairment of
consonant perception. Neuropsychologia, 12, 83–93.
Tallal, P., & Stark, R. E. (1982). Perceptual/motor profiles of
reading impaired children with or without concomitant
oral language deficits. Annals of Dyslexia, 32, 163–176.
von Suchodoletz, W. (1999). 100 years of research on specific
reading and spelling disorder—What is our knowledge
today? Zeitschrift fu¨r Kinder- und Jugendpsychiatrie, 27,
199–206.
Wechsler, D. (2000). HAWIK-III Hamburg-Wechsler-
Intelligenztest fu¨r Kinder [Wechsler Intelligence Scale
for Children (WISC-III)] (U. Tewes, P. Rossmann, &
U. Schallberger, German version). Bern: Huber. (Original
work published 1991).
Weinert, S. (1992). Deficits in acquiring language structure:
The importance of using prosodic cues. Applied Cognitive
Psychology, 6, 545–571.
Wimmer, H. (1993). Characteristics of developmental dyslexia
in a regular writing system. Applied Psycholinguistics,
14, 1–33.
Witruk, E., Friederici, A. D., & Lachmann, T. (2002). Basic
functions of language, reading and reading disability.
Dordrecht: Kluwer Academic Publishers.
Sabisch et al. 1695
A preview of this full-text is provided by The MIT Press.
Content available from Journal of Cognitive Neuroscience
This content is subject to copyright.
