Grammatical Processing without Semantics?
An Event-related Brain Potential Study
of Preschoolers using Jabberwocky Sentences
, Barbara T. Conboy
, Lindsay Klarman
and Patricia K. Kuhl
& Behavioral studies have demonstrated that children devel-
op a nearly adult-like grammar between 36 and 42 months, but
few studies have addressed how the child’s brain processes
semantic versus syntactic information. In previous research,
Silva-Pereyra and colleagues showed that distinct event-related
potentials (ERPs) are elicited by semantic and syntactic vio-
lations in sentences in children as young as 30, 36, and
48 months, following the patterns displayed by adults. In the
current study, we examined ERPs to syntactic phrase structure
violations in real and jabberwocky sentences in 36-month-old
children. Jabberwocky sentences are sentences in which con-
tent (open-class) words are replaced by pseudowords while
function (closed-class) words are retained. Results showed that
syntactically anomalous real sentences elicited two positive ERP
effects: left-distributed effects from 500 to 750 msec and 1050
to 1300 msec, whereas syntactically anomalous jabberwocky
sentences elicited two negative ERP effects: a left-distributed
effect from 750 to 900 msec and a later broadly distributed
effect from 950 to 1150 msec. The results indicate that when
preschoolers process real English sentences, ERPs resembling
the positive effects previously reported for adults are noted,
although at longer latencies and with broader scalp distribu-
tions. However, when preschoolers process jabberwocky sen-
tences with altered lexical–semantic content, a negative-going
ERP component similar to one typically associated with the
extraction of meaning is noted. &
One important and widely used psychophysiological ap-
proach for studying language processing is the recording
of event-related brain potentials (ERPs). These are aver-
ages of cortical electrical activity time-locked to some
external or internal event. ERP components are classified
according to their polarity (i.e., positive or negative deflec-
tions in the waveform), the time of their onset-to-offset or
peak amplitude in milliseconds, and their topographical
distribution across the scalp. ERP studies have provided
crucial information, with an exquisite time resolution,
about language processing in infants and young children
(for reviews, see Friederici, 2005; Mills, Conboy, & Paton,
2005; Kuhl, 2004). ERPs are ideally suited for studying
how different aspects of language processing systems are
established in young children, and to what extent they
resemble well-established systems in adults.
Sentence Processing Studies of Adults
Adult studies of sentence processing have indicated that
different ERP components reflect the involvement of
populations of neurons underlying different aspects of
processing. The ERP component most commonly asso-
ciated with the processing of semantic information in
sentences is the N400 component (Kutas & Hillyard,
1980). This is a negative wave that occurs between 250
and 500 msec after stimulus onset, peaks around
400 msec, and typically has a larger amplitude over right
posterior sites (Kutas & Van Petten, 1994). The ampli-
tude of the N400 to a particular word is highly sensitive
to the immediate context in which it occurs, whether
the context is a single word, a sentence, or discourse
(Kutas & Federmeier, 2000). ERP studies of adults pro-
cessing grammatical anomalies in sentences have dis-
played an early left anterior negativity (ELAN) or left
anterior negativity (LAN), a more centrally distributed
negativity, and a late parietally distributed positivity
(called P600). ELAN is a negative wave occurring be-
tween 150 and 250 msec that is sensitive to word
category (phrase structure) violations, whereas LAN is
a component peaking between 300 and 500 msec elic-
ited by word category as well as morphosyntactic viola-
tions, such as violation of agreement (Gunter, Friederici,
& Schriefers, 2000). Both components are observed in-
dependently of semantic constraints (Friederici, Hahne,
& Mecklinger, 1996; Mu¨nte & Heinze, 1994).
The P600 is
University of Washington,
Universidad Nacional Auto´noma de
D 2007 Massachusetts Institute of Technology Journal of Cognitive Neuroscience 19:6, pp. 1050–1065
a positive wave with a long latency, beginning around
500 msec and lasting several hundred milliseconds, that
is largest over centro-parietal regions. The P600 has been
found to occur with several types of morphosyntactic
violations (Morris & Holcomb, 2005; Friederici & Kotz,
2003; Friederici, 2002; Hagoort, Brown, & Osterhout,
1999; Osterhout, McLaughlin, & Bersick, 1997). For ex-
ample, in the anomalous sentence, ‘‘The cats won’t
EATING the fish,’’ the word EATING elicits a larger
P600 amplitude than the word EAT in ‘‘The cats won’t
EAT the fish’’ (Osterhout & Holcomb, 1995). In addition,
negativities in the 200–500 msec range, but more centrally
distributed than the E/LAN, have been observed in adults
to sentences that are anomalous in inflectional morphol-
ogy (e.g., Morris & Holcomb, 2005; Coulson, King, &
Kutas, 1998; Mu¨nte & Heinze, 1994; Friederici, Pfeifer, &
Hanhe, 1993; Kutas & Hillyard, 1983).
Models of Sentence Processing
The timing, polarity, and distribution of ERP effects elic-
ited by semantic and syntactic violations have been in-
terpreted within both serial and interactive models of
sentence processing. Serial models assume that distinct,
dissociable mechanisms underlie the processing of syn-
tactic and semantic information, and are activated at
different points in processing (e.g., Frazier, 1987). The
computation of an initial syntactic structure is believed
to precede semantic binding operations because struc-
tural information is necessary as input for thematic role
assignment. A structure is initially assigned by the syn-
tactic module exclusively on the basis of syntactic prin-
ciples, and this information is subsequently passed to
the semantic module for evaluation. The semantic mod-
ule can reject the initial structure, resulting in the sub-
sequent assignment of an alternative structure, as in
the case of temporarily ambiguous sentences (Frazier,
1987). Thus, these models assume that there are distinct
syntactic and semantic processing mechanisms, and in-
teractions between different types of information take
place at a later stage in sentence processing (e.g.,
Friederici, 2002; Frazier, 1995, 1998).
One such serial model has been proposed to ac-
count for the time course of ELAN, LAN, and P600 ef-
fects, as well as results from previous behavioral studies.
Friederici’s (1995, 2002) three-phase neurolinguistic
model of sentence comprehension includes an initial
syntactic analysis based on word category information,
followed by processing of lexically bound semantic and
morphosyntactic information for achieving thematic
role assignment, and finally, integration of the initial
syntactic structure with the subsequent morphosyntactic
and lexical–semantic information. According to this mod-
el, the ELAN or LAN component evoked by a syntactic
violation reflects the initial syntactic analysis and detec-
tion of errors, and the P600 component evoked by a
syntactic violation represents a ‘‘repair process,’’ where-
as a P600 elicited by ambiguity resolution (i.e., ‘‘The
horse raced past the barn fell’’) represents a ‘‘reanalysis
process’’ that occurs when the initial syntactic structure
does not map onto the thematic structure (Friederici,
Hahne, & Saddy, 2002). The P600 has also been claimed
to reflect the inability of the syntactic parser to assign the
preferred structure to incoming words (Hagoort, Brown,
& Groothusen, 1993).
Alternative psycholinguistic models propose that sen-
tence processing relies on the interaction between all
relevant sources of information, at the pragmatic, seman-
tic, phonological, morphological, and syntactic levels (e.g.,
Elman, Hare, & McRae, 2005; MacDonald, Pearlmutter, &
Seidenberg, 1994; Bates & MacWhinney, 1989; Marslen-
Wilson & Tyler, 1980). These interactive or constraint-
satisfaction models reject the idea that purely structural
information necessarily has temporal priority over seman-
tic information. Instead, lexical–semantic and morphosyn-
tactic cues are believed to be examined and weighed
together throughout the course of sentence processing,
leading to the best possible mapping between form and
meaning. Importantly, such views emphasize the roles
that the input frequency of particular sentence construc-
tions and validity of certain linguistic cues play in con-
straining sentence interpretation, along with real-world
knowledge. For example, on-line sentence processing
studies of English-speaking adults (Ferreira, 2003; Bates,
Devescovi, & Wulfeck, 2001; Bates & MacWhinney, 1989;
MacWhinney & Bates, 1989) have shown a preference for
canonical subject–verb–object (SVO) sentence interpreta-
tions. In other languages with more flexible word order,
adults tend to rely on other morphosyntactic cues such
as subject–verb agreement and semantic cues such as
animacy (Bates et al., 2001; MacWhinney & Bates, 1989).
The SVO word order is such a valid, reliable cue in English
that, in some cases, it can lead to implausible sentence
interpretations (Ferreira, 2003; MacWhinney, Bates, &
Kliegl, 1984). In a strong SVO language such as English,
comprehenders may initially rely on a basic ‘‘NVN/agent–
patient’’ heuristic strategy that assumes the subject of a
sentence is the agent and the object is the patient, leading
to a ‘‘shallow’’ form of initial processing that does not take
other syntactic information into account, although given
enough time, a more complete form of processing can be
accomplished using syntactic algorithms (Ferreira, 2003).
Sentences are processed more easily when they conform
to canonical word order, that is, the word order and
semantic cues converge to yield a single interpretation,
rather than opposing interpretations. Other basic heuris-
tics emphasizing semantic and/or plausibility information
may also be used in sentence processing (Ferreira, 2003).
Interactive views of language have been supported
by ERP studies that show immediate semantic analysis/
integration effects in sentence processing, challenging
the notion that automatic syntactic processes always op-
erate prior to semantic processes or in an encapsulated
manner (see Kutas, Van Petten, & Kluender, 2006).
Silva-Pereyra et al. 1051
Moreover, when adults process certain types of anoma-
lous sentences which can be construed as either seman-
tically or syntactically anomalous, semantics appear to
guide the initial syntactic analysis of sentences, rather
than the other way around (Kim & Osterhout, 2005).
The interactive school of thought also considers that the
components and effects noted to syntactic anomalies in
sentences may reflect more general cognitive processes,
rather than the operation of specific linguistic modules.
For example, the P600 effect observed to grammatically
anomalous sentences has been interpreted as part of a
family of more general P300 (P3b) effects that reflect
context updating processes and/or responses to low
probability target events (see Kutas et al., 2006; Coulson
et al., 1998). The LAN has also been interpreted as
reflecting general working memory processes rather
than specific morphosyntactic processing (for review,
see Kutas et al., 2006). Numerous studies have revealed
less than clearcut dissociations between semantic and
syntactic processes indexed by the N400, E/LAN, and
P600 components (Kutas et al., 2006).
Sentence Processing in Children
Modular/serial versus interactive activation accounts of
sentence processing tend to align with opposing views
on language development in children. Many modular
approaches to language development emphasize sepa-
rate mechanisms for rule-based versus associative learn-
ing (e.g., Ullman, 2004; Pinker & Ullman, 2002; Pinker,
1994, 1999; Chomsky, 1981). According to the ‘‘continu-
ity’’ hypothesis, grammatical processing mechanisms are
innate, although they may not function in a completely
adult-like manner during childhood due to cognitive lim-
itations (see Clahsen & Fesler, 2006). Experience with
language serves to trigger innate linguistic knowledge
and the processing system becomes adult-like with mat-
uration of other cognitive mechanisms (Pinker, 1994,
1999). Thus, according to these views, syntactic parsing
mechanisms are believed to be in place during early
childhood, but operating in a less efficient manner.
In contrast, according to usage-based, emergentist ac-
counts of language development, experience with lan-
guage is the primary force shaping the development of
basic cognitive processing mechanisms into specialized
networks for distinct aspects of language processing (see
MacWhinney, 1999; Elman et al., 1996). On this view,
knowledge about syntactic constructions and morpho-
syntax emerges gradually with growth in the lexicon,
applying first to specific lexical items and later becom-
ing abstract through generalization to other items
(Goldberg, Casenhiser, & Sethuraman, 2004; Lieven,
Behrens, Speares, & Tomasello, 2003; Savage, Lieven,
Theakston, & Tomasello, 2003; Tomasello, 2000, 2003;
Bates & Goodman, 1999; Lieven, Pine, & Baldwin, 1997).
From this perspective, sentence interpretation strategies
are influenced by previous learning as well as ease of
processing factors (Reyes & Herna´ndez, 2006; Dick,
Wulfeck, Krupa-Kwiatkowski, & Bates, 2004; Thal &
Flores, 2001; Bates et al., 1984). For example, the SVO
preference has been noted for children learning other
languages with more variable word order such as French
(Kail & Charvillat, 1988), German (Lindner, 2003), Italian
(Devescovi, D’Amico, Smith, Mimica, & Bates, 1998), and
Spanish (Reyes & Herna´ndez, 2006), despite the fact
that this is not a preferred pattern for adult speakers of
those languages and not a reliable cue to agent–patient
relations. When processing sentences, children may rely
more on local cues such as word order because they
are computationally easier to process than distributed/
topological cues such as subject–verb agreement and
other morphological cues (Dick et al., 2004; Thal &
Flores, 2001; Von Berger, Wulfeck, Bates, & Fink, 1996;
Kail & Charvillat, 1988; Bates et al., 1984). Thus, as ex-
perience with language is gained, and as cognitive abil-
ities increase, children become increasingly able to use
a variety of cues for sentence processing and rely on
those that are most valid for their language.
ERP Sentence Processing Effects in Children
One important question that can be addressed using the
ERP technique is whether the neural activity linked to
sentence processing in children is similar or different
from that noted in adults with respect to the components
elicited by grammatical and semantic incongruities. Few
previous studies have used ERPs to investigate sentence
processing in preschool-age children. Silva-Pereyra and
colleagues provided ERP evidence of semantic and syn-
tactic processing at 30 months (Silva-Pereyra, Klarman,
Lin, & Kuhl, 2005) and at 36–48 months (Silva-Pereyra,
Rivera-Gaxiola, & Kuhl, 2005). At all three ages, children
displayed a positive effect for syntactically anomalous
sentences and a negative effect for semantically anoma-
lous sentences, similar to the patterns reported for adults
(see Kutas et al., 2006) and older children (Hahne,
Eckstein, & Friederici, 2004). Harris (2001) conducted
two studies of syntactic processing with preschool-age
children. In the first, 32- to 38-month-old children dis-
played a bilateral positivity from 500 to 1500 msec to
phrase structure anomalies in which the direct object of
the verb was placed in preverbal position, and a negative
effect for semantically anomalous sentences. However, in
that study, there were differences in the word class
(open- or closed-class) of the word immediately pre-
ceding the target in the syntactically anomalous versus
control sentences, possibly leading to unequal baselines
in the two conditions. In the second study, a different
group of 36- to 39-month-old children showed a bilateral
negativity from 300 to 600 msec to phrase structure
anomalies in which a closed-class word was inserted
before the final noun phrase (NP). The differences across
these studies are likely due to the different sentence
types and the way in which the stimuli were presented
1052 Journal of Cognitive Neuroscience Volume 19, Number 6
(segmented speech in the former study, natural contin-
uous speech in the latter). Another recent study inves-
tigating the processing of phrase-structure violations in
32-month-olds demonstrated a left hemispheric negativity
around 500 msec and a late P600 (Oberecker, Friedrich, &
Friederici, 2005). A more recent study has also shown
a late P600, but no negative effect, in 24-month-olds
(Oberecker & Friederici, 2006). The left lateralization of
the negative effect in these studies has led to the inter-
pretation of this effect as a child precursor to the E/LAN
observed in adults for phrase structure violations. Both
components, the E/LAN and the P600, started later and
persisted longer than those observed in adults.
The ‘‘Jabberwocky’’ Sentence
One approach for investigating the coordination of
semantic and syntactic processes during sentence com-
prehension is to analyze the processing of sentences
that are syntactically intact but have greatly reduced
lexical–semantic information, that is, jabberwocky sen-
tences (based on the famous poem of Lewis Carroll,
1872). These are sentences in which content (open-
class) words are replaced by pseudowords, and function
(closed-class) words and verb inflections are kept as
morphological markers (e.g., My macle platched a
flovie about my garily). Such sentences can, therefore,
be used to test the extent to which listeners rely on
primarily syntactic information to build up expecta-
tions about sentence structure and to examine the
brain processes involved when violations of such expec-
Jabberwocky ERP Studies of Adults
Syntactic violations of subject–verb agreement (Mu¨nte,
Matzke, & Johannes, 1997) and phrase structure (Hahne
& Jescheniak, 2001; Canseco-Gonzalez, 2000) have been
investigated in ERP studies of adults using real sentences
and their corresponding jabberwocky sentences. These
previous studies yielded mixed findings. Both Canseco-
Gonzalez (2000) and Mu¨nte et al. (1997) reported early
negative effects but not a P600 effect. Mu¨nte and col-
leagues concluded that it was not clear whether the
negative effects they observed, which were maximal at
central electrode sites, could be classified as LAN effects.
Canseco-Gonzalez also observed a negativity to ungram-
matical pseudoword stimuli, but this effect had a broad
rather than a left-lateralized distribution, and therefore
could not necessarily be interpreted as an ELAN or LAN.
According to Canseco-Gonzalez and Mu¨nte et al., the
absence of P600 effects observed in their studies reflects
the absence of repair or reanalysis processes; there is no
reason to repair or reanalyze a sentence containing
syntactic errors unless semantic information is also pres-
ent. On the contrary, Hahne and Jescheniak (2001)
observed both a LAN and a P600 effect for real and
jabberwocky sentences in adults. They suggested that
the LAN effect reflected an initial parsing stage during
which phrase structure information was processed,
whereas the P600 reflected a repair process in which
syntactic information was evaluated later in sentence
processing and not based upon the semantics of the
sentence. Based on these contradictory findings across
studies, Hahne and Jescheniak proposed an explana-
tion that relies on the time at which the semantic
and syntactic processes are reflected in the ERP. Spe-
cifically, when an early negativity reflecting a purely
syntactic process is present, as in their study, a P600
will also be noted because the process underlying the
negativity will prevent detailed lexical–semantic process-
ing of the syntactically anomalous item, and the P600
analysis process will ensue. In contrast, when the early
negativity is absent, the syntactic error will be processed
slightly later and at the same time as the semantic error,
and this will block a subsequent P600 effect. In other
words, the negativity noted in the studies of Canseco-
Gonzalez and Mu¨nte et al. may have been an N400 effect
to the pseudowords combined with a LAN-like effect.
Goals of the Present Study
The Jabberwocky ERP Paradigm with Children
In our previous ERP grammatical processing studies
of children (Silva-Pereyra, Klarman, et al., 2005; Silva-
Pereyra, Rivera-Gaxiola, et al., 2005), the syntactic anom-
alies were presented in real sentences that contained
intact lexical–semantic information. Of interest is wheth-
er children this age would show similar syntactic pro-
cessing effects under conditions of greatly reduced
semantic content, like the adults studied by Hahne
and Jescheniak (2001), or different effects in the jabber-
wocky condition, like the adults studied by Canseco-
Gonzalez (2000) and Mu¨nte et al. (1997). One other
study (Harris, 2001) has provided ERP evidence of syn-
tactic processing in 36- to 39-month-old children using
jabberwocky sentences. In that study, phrase structure
violations in real sentences elicited a negative effect be-
tween 300 and 600 msec, which was larger over the left
than the right hemisphere, but significant only for
children with higher language skills as determined by
the Clinical Evaluation of Language Function—Preschool
(CELF-P; Wiig, Secord, & Semel, 1992). In contrast,
phrase structure violations in jabberwocky sentences
elicited a bilateral negativity from 600 to 700 msec,
which was largest at left anterior electrode sites but
significant only for the higher language group.
In the present study, we investigated the brain activity
of 36-month-old children as they processed syntactic
violations in real sentences, with intact lexical–semantic
information, and jabberwocky sentences, in which lexical–
Silva-Pereyra et al. 1053
semantic information was greatly reduced. Given that
many aspects of grammar are acquired by approxi-
mately 3 years of age (Hirsh-Pasek & Golinkoff, 1996;
Brown, 1973), we predicted that children’s ERPs to
grammatically anomalous versus correct sentences
would reflect patterns of the brain activity associated
with syntactic processes. We predicted that if the positive
ERP effect previously observed to syntactic anomalies in
preschoolers (Silva-Pereyra, Klarman, et al., 2005; Silva-
Pereyra, Rivera-Gaxiola, et al., 2005) is a specialized
syntactic component that occurs regardless of lexical–
semantic information, then syntactic anomalies in jabber-
wocky sentences should elicit similar ERPs as those in real
sentences, that is, a late positivity after the onset of the
word that renders the sentence ungrammatical. In con-
trast, if the positive ERP effect previously noted to syn-
tactic anomalies in preschool children is modulated
by lexical–semantic information, then that ERP effect
should be different for jabberwocky sentences, possibly
absent due to the lack of a need for sentence reanalysis,
as noted by Canseco-Gonzalez (2000) and Mu¨nte et al.
(1997). Thus, by examining how this syntactic ERP com-
ponent behaves in jabberwocky sentences, we aimed to
tease apart the relative roles of semantic and syntactic
information in sentence processing at an early point in
development at which grammatical development is well
underway. We did not expect to observe a negativity re-
sembling the LAN or ELAN to either type of sentence,
given that those ERP effects were not noted in our
previous studies with 30- to 48-month-old children (how-
ever, a LAN was reported for the 3-year-old children
studied by Oberecker et al., 2005).
Children were recruited from the Child Participant Pool
at the University of Washington. All participants were
healthy full-term 36-month-olds with monolingual En-
glish experience and with no known hearing deficits.
Each child had no more than two mild ear infections
before testing and was healthy and developing typically
at the time of the study. Parents signed University of
Washington ethics committee approved consent forms
and were informed of the procedures and aims of the
study. They received $15 for their participation in the
1-hour session, and the children received a small toy.
Parents were also asked to complete a vocabulary check-
list composed of all the words used in the experiment;
if a parent did not indicate that the child knew at least
95% of the words on the list, the child’s data were not
included in the analyses.
Forty-seven children completed the experiment, but
31 children’s recordings were not used due to in-
sufficient numbers of artifact-free segments. Nineteen
36-month-old children (9 girls; mean age = 36.17; SD =
0.38; range = 35.53–36.73) were included in the analy-
ses. Parents were asked about their own and their
child’s handedness. All children included in the analy-
ses were right-handed and had no family history of
Forty regular transitive verbs, rated as being produced
by 98% of 30-month-old children in the MacArthur-Bates
Communicative Development Inventories (CDIs) lexical
database (Dale & Fenson, 1996), were used to build 80
regular past-tense sentences. Early acquired nouns,
prepositions, pronouns, and determiners selected from
the CDIs were also included in the sentences. All sen-
tences were eight words in length and contained a prep-
ositional phrase (e.g., My uncle watched a movie about
my family). Half of the real sentences were correct,
whereas the other half contained phrase structure vio-
lations. The phrase structure violations were created by
reversing the word order of the preposition and pre-
ceding NP (e.g., *My uncle watched about a movie
Eighty jabberwocky sentences were constructed using
40 pseudowords that were phonologically related to the
content words (i.e., nouns, verbs) used in the real
sentences. Pseudowords were created by extracting
phoneme sequences from each content word, creating
a matrix of different combinations, and selecting pseu-
dowords that were phonotactically legal for English, as
judged by two native English speakers who were stu-
dents in Speech and Hearing Sciences. Pseudowords
were controlled so that they matched the real English
words in syllable number and final consonant (e.g.,
mool–ball) (see Appendix for list of stimulus senten-
ces). All of the pseudoverbs were inflected as regular
verbs (with the ‘‘ed’’ ending, e.g., watched–platched).
Pilot behavioral testing was conducted with five native
English-speaking adults in order to verify the syntactic
violation point in the anomalous jabberwocky sen-
tences. Subjects read the sentences and indicated the
earliest point at which each sentence sounded gram-
matically incorrect. Based on this testing, 10 sentences
were identified that had ambiguous violation points.
These sentences were eliminated from the analyses.
For the remaining jabberwocky sentences, the adults
were in 100% agreement regarding the violation point.
All sentences were spoken by a female native English
speaker. The sentences were recorded on a digital-audio
system and sampled at 44 kHz with a 16-bit resolution.
The speaker rehearsed the sentences prior to recording
to ensure that they sounded natural during the record-
ing. The average sound pressure level of the sentences
ranged from 63 to 67 dB SPL. In order to ensure precise
time locking of the ERP to the onset of each critical word
in the sentences, waveforms and spectrograms of the
critical words were inspected.
1054 Journal of Cognitive Neuroscience Volume 19, Number 6
ERPs were time-locked to four different critical words.
For real sentences, the ERP was time-locked to the onset
of the preposition, which rendered the sentence: (a)
syntactically anomalous (e.g., My uncle watched ABOUT
a movie my family) or (b) nonanomalous (e.g., My
uncle watched a movie ABOUT my family). For jabber-
wocky sentences, the ERP was time-locked to the onset
of the last NP, which rendered the sentence: (c) syntac-
tically anomalous (e.g., My macle platched about a
flovie MY GARILY ) or (d) nonanomalous (e.g., My macle
platched a flovie about MY GARILY ). The different
locations of critical words across the real and jabber-
wocky sentences were necessary because, for the jabber-
wocky sentences, the violation would not be detected
until the last NP was encountered.
The mean duration from the onset of nonanomalous
real sentences to the critical word (i.e., preposition) was
1374.4 msec (SD = 161.84 msec). For real sentences
with syntactic violations, the mean duration from the on-
set of the sentence to the preposition was 916.31 msec
(SD = 108.64 msec). For nonanomalous jabberwocky
sentences, the mean duration from the onset of the
sentence to the onset of the last NP was 1887.76 msec
(SD = 268.96 msec); for the jabberwocky sentences
with syntactic violations, the mean duration from the
onset of the sentence to the onset of the last NP was
1941.79 (SD = 293.37 msec). Twenty additional senten-
ces were included in each type of sentence condition
(i.e., real and jabberwocky) as filler sentences.
Each child was fitted with a 20-channel electrode cap
(Electrocap International, Eaton, OH) while playing with
a research assistant. After conductive gel was applied to
each electrode site, participants and experimenters
moved into a sound-attenuated testing booth. Inside
the booth, the impedances were measured and the child
was seated in a comfortable chair close to his or her
parent and approximately 1 m in front of a puppet
theater. Participants listened passively to the stimuli
while watching a live puppet show. The sentences were
presented via loudspeakers mounted on the puppet
theater, at interstimulus intervals of 1500 msec.
Each participant was tested in two sessions on the
same day. Both experimental sessions lasted a total of
approximately 1 hour, including the procedure of fitting
the cap. In Session 1, 80 real sentences with and without
syntactic violations were presented, and in Session 2, 80
jabberwocky sentences were presented. The order of
the sessions was counterbalanced across participants.
The electroencephalogram (EEG) was recorded from
20 tin electrodes secured in an elastic cap (Electrocap
International) at the following locations (according to
the International 10–20 system): Fp1, Fp2, F3, F4, C3,
C4, P3, P4, O1, O2, F7, F8, T3, T4, T5, T6, Fz, Cz, and Pz.
The vertical electrooculogram (VEOG) was recorded
from an infraorbital electrode placed on the infant’s left
cheek, in order to detect eye artifact in the EEG record-
ings. The recordings were referenced to the left mastoid;
the brain electrical activity over the right mastoid was
also recorded and did not reveal any condition-specific
variations. Electrode impedances were kept below 10 k.
The EEG signal was amplified with a gain of 20,000 by an
Isolated Bioelectric Amplifier System Model SC-32/72BA
(SA Instrumentation, San Diego, CA) with a bandpass of
0.1 to 100 Hz, continuously sampled at 250 Hz by an
analog-to-digital converter, and stored on a hard disk for
ERPs were computed off-line from 2048 msec epochs for
each subject in each experimental condition. Epochs
consisted of the 100 msec preceding and 1948 msec
following the presentation of each individual critical word
in each of the sentences. Automatic rejection of segments
was carried out on the basis of the following criteria:
Segments with electrical activity exceeding ±150 AVand
amplifier blocking for more than 200 msec at any elec-
trode site were considered artifact and the entire segment
was rejected. EEG segments from each subject were also
visually inspected, and those segments with eye move-
ments detected in the VEOG were rejected. Electrical
activity from F7 minus F8 (used as the horizontal EOG
criterion) that exceeded ±50 AV was also treated as
artifact, and segments containing such artifact were re-
jected. Subjects with fewer than 18 artifact-free trials for
each condition were excluded from the average. There
were no significant differences in the percentages of trials
eliminated due to artifact rejection across anomalous
and nonanomalous real sentences [M =33.8,SD =5.7
and M =34.1,SD = 5.7, respectively; F(1, 18) = 0.049].
Similarly, there were no significant differences in arti-
fact rejection rates across anomalous and nonanomalous
jabberwocky sentences [M =36.2,SD =9.5andM =
35.5, SD =9.8,respectively;F(1, 18) = 0.09]. Further fil-
tering was conducted using a band-pass of 0.5 to 30 Hz.
ERPs were re-referenced to the average right mastoid
and left mastoid reference. Baseline correction was per-
formed using the 100-msec prestimulus time window
A series of repeated-measures analyses of variance
(ANOVAs) were performed on mean amplitude values
for comparisons between syntactically anomalous and
nonanomalous real and jabberwocky sentences. The
Huynh–Feldt correction was applied to analyses with
more than one degree of freedom in the numerator.
Planned comparisons were reported as significant at
the .05 level.
Silva-Pereyra et al. 1055
Analyses proceeded in two stages. First, repeated-
measures ANOVAs were performed separately for each
sentence type (real and jabberwocky) on the ERP epoch
50–1600 msec in 50-msec consecutive intervals. Sepa-
rate sets of repeated-measures ANOVAs were conducted
for the data acquired at midline and lateral sites. For
the midline sites, two-way repeated-measures ANOVAs
were performed with violation (nonanomalous vs. anom-
alous sentences) and electrode site (Fz, Cz, and Pz)
as factors. For lateral sites, three-way repeated-measures
ANOVAs were conducted with violation, hemisphere,
and electrode site (Fp1–Fp2, F3–F4, C3–C4, P3–P4,
O1–O2, F7–F8, T3–T4, T5–T6) as factors. For the midline
sites, the analyses revealed two significant effects: main
effect of condition on real sentences at 800–850 msec
[F(1, 18) = 5, p = .038] and main effect of condition on
jabberwocky sentences at 1000–1050 msec [F(1, 18) =
8.2, p = .01]. For the lateral sites, the analyses revealed
two ranges of time for real sentences and two for
jabberwocky sentences for which main effects and inter-
actions were significant. Results for the comparison
between syntactically anomalous and nonanomalous
real and jabberwocky sentences are summarized in
Table 1. For real sentences, one of the significant time
windows extended from 500 to 750 msec and the other
one from 1050 to 1300 msec, reflecting a larger positive
wave to the anomalous versus nonanomalous sen-
tences. For jabberwocky sentences, the anomalous sen-
tences elicited a significantly larger negativity than the
nonanomalous sentences in two windows: 750–900 and
950–1150 msec. Based on these initial analyses, these
longer time windows were used for the second stage of
analysis. Repeated-measures ANOVAs were performed
separately for each type of sentence (real and jabber-
wocky) in each time window. Again, separate analyses
were conducted for the data acquired at midline and
500–750 msec Time Window
Grand-average ERPs to the syntactically anomalous and
nonanomalous real sentences at the onset of the target
words are shown in Figure 1A. The anomalous sentences
elicited a larger positivity than the nonanomalous sen-
Table 1. Results of the Comparison between Syntactically Anomalous and Nonanomalous Real and Jabberwocky Sentences
Real Sentences Jabberwocky Sentences
(msec) V (1,18) df V
H (1,18) df V
E (a,b) df V
H (a,b) df V (1,18) df V
H (1,18) df
500–550 (3.9, 70.4) = 2.5*
600–650 4.8* (6.3, 112.5) = 2.2*
650–700 6.3* (6.7, 120.6) = 2.2*
700–750 (6.7, 119.8) = 2.3*
800–850 7.2* 5.9*
1050–1100 7.1* (5.3, 94.9) = 2.4* 4.1*
1100–1150 10.9** (5.6, 101.1) = 3* 4.1*
1150–1200 (3.2, 57) = 2.9*
1200–1250 (4.2, 76.1) = 4.3**
1250–1300 (2.7, 48.2) = 2.9* (4.7, 84.2) = 2.8* 4.9*
V = violation; E = electrode site; H = hemisphere.
*p < .05.
**p < .01.
1056 Journal of Cognitive Neuroscience Volume 19, Number 6
tences over the left hemisphere [Violation Hemi-
sphere Electrode site interaction, lateral, F(5.7,
103.2) = 2.5, p = .027]. Planned comparisons be-
tween anomalous and nonanomalous sentences at each
electrode site indicated that the larger positivity to
anomalous sentences was significant only at three left
hemisphere electrode sites: F3 [F(1, 18) = 4.4, p = .05];
T3 [F(1, 18) = 7.1, p = .02]; and T5 [F(1, 18) = 5.7, p =
Figure 1. This figure shows grand-average ERPs to anomalous (dotted line) and nonanomalous (solid line): (A) real sentences at the onset
of the preposition, and (B) jabberwocky sentences at the onset of the second noun phrase. Negative is plotted up. In (A), a larger amplitude
positive response is observed for anomalous versus nonanomalous real sentences at left hemisphere sites (P600 and a later effect at T3
from 1050 to 1300 msec). In (B), larger amplitude negative responses are observed for the syntactically anomalous jabberwocky sentences
= N750–950 and N
Silva-Pereyra et al. 1057
.03]. There were no effects of violation at any right
hemisphere electrode sites.
1050–1300 msec Time Window
A larger positivity was elicited by anomalous versus
nonanomalous real sentences in this time window [Vio-
lation Hemisphere Electrode site interaction, lat-
eral, F(5.2, 93.3) = 2.6, p = .03]. Planned comparisons
between anomalous and nonanomalous sentences at
each electrode site indicated that the larger positivity
to anomalous sentences was significant only at T3
[F(1, 18) = 8.9, p = .008].
750–900 msec Time Window
Grand-average ERPs to the syntactically anomalous and
nonanomalous jabberwocky sentences at the onset of the
last NP are shown in Figure 1B. A negative-going wave is
observed to the jabberwocky sentences after the onset of
the last NP. This negative wave is more prominent for the
anomalous than the nonanomalous sentences over the
left hemisphere [Violation Hemisphere interaction,
lateral, F(1, 18) = 8.1, p = .011]. Planned comparisons
for violation made for each hemisphere showed that the
differences were significant over the left but not the right
hemisphere [Main effect of violation over the left hemi-
sphere: F(1, 18) = 5.3, p =.034].
950–1150 msec Time Window
A larger negativity was noted to the anomalous versus
nonanomalous sentences [Violation main effect, lateral,
F(1, 18) = 7.8, p = .012]. This effect was broadly
distributed across the scalp; there were no main effects
of hemisphere or interactions with hemisphere.
The present study examined auditory ERP responses in
preschoolers to phrase structure violations occurring in
real and jabberwocky sentences (i.e., sentences in which
content words were replaced with pseudowords while
grammatical functional words were retained). There
were two important results. First, a larger positivity was
elicited by the anomalous versus nonanomalous real
sentences over left frontal, temporal, and posterior
temporal sites, beginning at 500 msec and ending at
750 msec, and followed by a second, later positive effect
over the left temporal site. Second, two larger negativ-
ities were elicited by anomalous versus nonanomalous
jabberwocky sentences, the first one with a left hemi-
sphere distribution, beginning around 750 msec after
the onset of the critical word (the determiner or noun in
the last NP), and lasting until 900 msec; the second one
with a broad distribution from 950 to 1150 msec.
The positive effects noted to phrase structure viola-
tions in real sentences were generally consistent with
those reported for children in previous studies, albeit
with variations in latencies and scalp distributions, and
may be interpreted as P600-type effects. The onset of
this positivity was similar to that reported for phrase
structure violations in the 32- to 38-month-old children
studied by Harris (2001), although the scalp distribu-
tion of the effect was more left-lateralized in the present
study. The results also showed similarities and differ-
ences when compared to those reported to phrase struc-
ture violations in 32-month-old children (Oberecker
et al., 2005). In that study, phrase structure violations
elicited an early positive effect with a right-central scalp
distribution occurring prior to 300 msec, followed by a
left-lateralized negative effect from approximately 300–
700 msec, and a later positive effect with a right-central
distribution occurring from 1100 to 1500 msec. Further-
more, the results were both similar to and different from
those obtained in our previous studies of preschool-age
children using morphosyntactic violations. The first
positive effect (500–750 msec) observed in the present
study occurred later, and was more left-lateralized, than
the first positive effect (300–600 msec) to morphosyn-
tactic violations found in 36- and 48-month-old children
(Silva-Pereyra, Rivera-Gaxiola, et al., 2005). It was also
more left-lateralized than the broadly distributed second
positive effect (600–1000 msec) to morphosyntactic
violations reported for 36- and 48-month-old children
(Silva-Pereyra, Rivera-Gaxiola, et al., 2005).
The longer latencies of these positive effects in the
present study could be due to the fact that the sen-
tences might not have been considered ‘‘ungrammati-
cal’’ until after the chosen onset point (the preposition).
For example, the stem, ‘‘My uncle watched about ...’’
could be completed in a grammatically well-formed
sentence: ‘‘My uncle watched about five football games
on Sunday.’’ Thus, the sentence structure, ‘‘My uncle
watched about a movie my family’’ is temporarily am-
biguous at the point that the preposition ‘‘about’’ is
encountered. However, this was not the case in most
of the sentences (see Appendix). Importantly, neither
the phrase structure violations in real sentences in the
present study nor the morphosyntactic violations in our
previous studies elicited a negative effect resembling a
LAN, in contrast to the results reported by Oberecker
et al. (2005).
The negative effects elicited by phrase structure vio-
lations in jabberwocky sentences were also different
from the negativities elicited by grammatical violations
in other studies of young children. Both negative effects
observed in the present study occurred at longer laten-
cies than the bilateral negativity from 600 to 700 msec
elicited by phrase structure violations in jabberwocky
1058 Journal of Cognitive Neuroscience Volume 19, Number 6
sentences reported by Harris (2001). In addition, the
negativities occurred at much longer latencies than the
LAN-like effects to phrase structure violations in real
sentences reported by Oberecker et al. (2005). The
longer latency of the negative effects for jabberwocky
sentences in the present study, compared to those
reported by Harris, might have been due to how sen-
tences were parsed by listeners. For example, in the
sentence, ‘‘My macle platched about a flovie my garily,’’
the phrase ‘‘a flovie my garily’’ could be parsed as
the beginning of an object relative clause, as in ‘‘My
uncle talked about a movie my family was in,’’ in which
case the listener would be expecting a verb in the
relative clause after the pseudoword ‘‘garily’’ and would
not consider the sentence ungrammatical until that
point. In other cases, the pseudoverb could be inter-
preted as a verb embedded in a subject relative clause.
For example, in ‘‘The bailgan plambed with the gree
his jeg,’’ the listener might parse this as the beginning
of a construction analogous to ‘‘The mailman plagued
with the flu his son gave him was in the hospital.’’
In that case, the listener would be anticipating a verb
right after the pseudoword ‘‘jeg.’’ In both these cases,
the sentences would be ambiguous until the very end,
at the point when the listener is waiting for a verb and
does not hear it. However, the alternate parsing of
the anomalous jabberwocky sentences, in which the
syntactic anomaly would be detected at the point of
the final NP, is more probable, based on other litera-
ture showing that children this age tend to rely on
canonical SVO interpretations rather than relative clause
or other interpretations (Von Berger et al., 1996; Bates
et al., 1984).
An alternative explanation for the late negativities
observed for the processing of jabberwocky sentences
could be that they are semantic rather than purely
syntactic effects, reflecting children’s attempts to make
sense of the nonsensical sentences. One important fact
that may explain the occurrence of negative rather than
positive effects is the position of the ungrammatical
point in the sentences used in the present study. In
the present study, the violation occurred later in the
sentence for the jabberwocky than for the real sen-
tences, and this may have given listeners more time
to construct possible sentence meanings. Osterhout
(1997) has provided evidence that although sentence-
embedded syntactically anomalous words elicit a P600-
like response in adults, sentence-final anomalous words
can elicit an enhanced negative wave or an N400-like re-
sponse. An interpretation of this sentence-final negativ-
ity is that it reflects the eventual semantic consequences
of an ungrammatical (hence, not fully interpretable)
sentence. Because the failed parse leads to problems
at the semantic/conceptual level, the critical words elicit
an enhanced N400 component. Furthermore, attempts
at reanalysis might be less common when the anomaly
is in sentence-final position than when it appears em-
bedded within the sentence. Thus, the N400 amplitude
may reflect the semantic anomaly engendered by an un-
parsable or misparsed sentence.
Notably, our results and those reported by Harris
(2001), who used different types of jabberwocky stimuli
with a different violation point, converge to show that a
P600 is not elicited in ungrammatical sentences that
have greatly reduced semantic content in children this
age. These results are consistent with those observed in
adults, as reported by Canseco-Gonzalez (2000) and
Mu¨nte et al. (1997), although they differ from those
reported by Hahne and Jescheniak (2001). Furthermore,
our results show that a P600-like effect is observed in the
same children when processing grammatical anomalies
in real English sentences.
The results of the present study raise some interesting
questions regarding levels of linguistic processing in
young children. First, is there any evidence that the
late negativities elicited in the jabberwocky condition
reflect a purely syntactic effect rather than a semantic
integration effect? Although it is not possible to answer
this question only on the basis of our results and those
reported by Harris (2001), the results provide some
evidence that, in preschool children, semantic and
syntactic processes interact during sentence processing,
and syntactic structure is constructed taking semantic
information into account. According to syntax-centered
models, in which syntactic processing is believed to
operate in the earliest stages of sentence processing,
during jabberwocky sentence processing the parser
should be able to build syntactic structure even when
semantic information is distorted, as observed by Hahne
and Jescheniak (2001) in their study of adults. Given
that lexical entries cannot be activated by pseudowords,
the integration would be completed solely on the basis
of thematic role information and a P600-like positive
effect would be observed, consistent with the P600
effects elicited by syntactic violations in real sentences.
Neither our study nor the study conducted by Harris
provides evidence for such an account. In contrast, ac-
cording to interactive models, in which linguistic in-
formation from different levels is believed to interact
from the beginning stages of sentence processing, the
negative ERP effects elicited by jabberwocky sentences
in our study could be interpreted as reflecting the use
of multiple linguistic cues throughout the course of
sentence processing in an effort to construct meaning.
Young English learners, who have been exposed to a
large proportion of SVO sentence structures in their
input, may especially rely on what Ferreira (2003) has
described as a basic ‘‘NVN/agent–patient’’ heuristic strat-
egy, and employ a ‘‘shallow’’ form of sentence pro-
cessing that does not take other syntactic information
into account (see also Slobin, 1985). Thus, the nega-
tive effects elicited by syntactic violations in jabber-
wocky sentences most likely reflected attempts at
semantic integration rather than the use of purely
Silva-Pereyra et al. 1059
syntactic information, and may have reflected the inter-
action between the use of an NVN/agent–patient strat-
egy, the limited semantic information present in the
sentences, and whatever real-world pragmatic knowl-
edge children this age were able to employ in the service
of sentence interpretation.
A second interesting question raised by our results
concerns the extent to which children this age are able
to extract meaning from various linguistic cues. The
amplitude of the N400 has been interpreted to reflect
the ‘‘ease’’ with which a target can be integrated with
the preceding context, at the semantic level (Osterhout
& Holcomb, 1995; Brown & Hagoort 1993). However,
the negative ERP effects noted in the present study may
reflect the integration of linguistic information at a
variety of levels (i.e., semantic, phonological, pragmatic,
and syntactic levels). According to Hirsh-Pasek and
Golinkoff (1996), children of preschool age can compre-
hend word order because there are redundant prosodic
and semantic cues, although at this age they also focus
their attention on syntactic cues to a larger extent than
other possible sources of information. In our jabber-
wocky sentences, the main cues to sentence meaning
were the regular past-tense inflections on pseudoverbs,
the intact closed class words, and thematic role infor-
mation. In addition, many of the pseudowords sounded
sufficiently like real words, and may have been pro-
cessed in a holistic way. Evidence from younger toddlers
has suggested that, under processing situations that are
high in their cognitive demands, fine phonetic detail is
not always accessed in words (e.g., Mills et al., 2004;
Pater, Stager, & Werker, 2004; Stager & Werker, 1997).
Thus, it is possible that during this cognitively demand-
ing sentence processing paradigm, children may have
been using phonological information to map the pseu-
dowords onto their representations of similar sounding
real words. Although it might be argued that participants
who heard the real sentences first would have attempt-
ed to map the jabberwocky sentences onto their mem-
ories of those previously heard real sentences, this is
unlikely, given that the order of presentation of the
sentence types (real vs. jabberwocky) was counterbal-
anced, and statistical analyses indicated no differences
between the ERP effects observed in children who heard
the real sentences first and those who heard the jabber-
wocky sentences first. It is more likely that children
were using the limited lexical–semantic information
present in the sentences to construct meaning. It has
been shown that children can deduce the meanings
of unknown verbs based on knowledge about other
(Gleitman & Gillette, 1999; Naigles, 1990, 1996; Naigles
& Hoff-Ginsberg, 1995; Fisher, 1994; Fisher, Gleitman, &
Gleitman, 1991; Landau & Gleitman, 1985), as well as
morphological markers on the verbs (e.g., Behrend,
Harris, & Cartwright, 1995). Our children might have
generated expectations about what words should occur
after the pseudoverb based on their interpretations of
Finally, it is important to consider that the negativities
noted to the syntactic anomalies in jabberwocky sen-
tences may share some similarities with negative ERP
effects reported in other language processing paradigms
used with young children. Late negative effects have been
reported in a variety of ERP studies with toddlers and
preschoolers (Conboy & Mills, 2006; Mills, Conboy, et al.,
2005; Mills, Coffey-Corina, & Neville, 1997). For example,
a negative effect elicited to known versus unknown words
from 600 to 900 msec has been hypothesized to reflect
the need for enhanced attention during word processing
(Mills, Conboy, et al., 2005). Earlier negative effects to
known versus unknown words linked to word meaning
(e.g., from 200 to 400 msec) have also been reported in
those studies, and a more focal distribution of those
effects have been noted in children with higher language
skills (see also Mills, Plunkett, Prat, & Schafer, 2005; Mills,
Coffey-Corina, & Neville, 1993). Changes in both the
morphology and topographical distributions of ERP com-
ponents have also been noted from childhood to adult-
hood (Mills, Conboy, et al., 2005; Mills et al., 1997;
Holcomb, Coffey, & Neville, 1992). Many language-related
ERP components, for example, are at first broadly distrib-
uted across the scalp. However, the meaning of ERP scalp
asymmetries for language stimuli is a source of debate.
Several ERP studies have shown left hemisphere asym-
metries for language-related ERPs in infants (Dehaene-
Lambertz, 2000; Molfese & Molfese, 1979; Molfese,
Freeman, & Palermo, 1975), whereas others have shown
broadly distributed effects that become left-lateralized
with age and experience (Mills, Conboy, et al., 2005; Mills
et al., 1997). For example, in an ERP study of 11-month-
old infants, Thierry, Vihman, and Roberts (2003) found
familiar–unfamiliar word effects in the same time range as
Mills and colleagues (i.e., from approximately 250 msec),
but these were larger over the right versus left hemi-
sphere. Similarly, Conboy and Mills (2006) reported
larger right hemisphere N200–400 effects to known–
unknown words in the stronger language of bilingual
20-month-old children with larger vocabularies, and a
more bilateral distribution in the weaker language of
the same children and in both languages of children
with smaller vocabularies. Thus, the meaning of the left-
lateralized negative effect to syntactic anomalies in
jabberwocky sentences in the present study is difficult
to interpret. Studies of children at different ages and
longitudinal studies would be useful for determining
whether these left hemisphere scalp distributions
change with age and/or language experience.
In conclusion, the appearance of a positive ERP effect
to syntactic violations in real sentences that is similar to
the P600 effect reported in other studies of children and
adults indicates that the neural mechanisms of syntactic
parsing are present during early language development,
although the processes are slower in children than in
1060 Journal of Cognitive Neuroscience Volume 19, Number 6
adults. The results also suggest that semantics play a role
in syntactic processing. Whereas syntactic violations in
real English sentences elicited a positive ERP effect,
syntactic violations in jabberwocky sentences, which
contained reduced lexico-semantic information, elicited
a left-distributed negative-going ERP effect, but not a
positive effect. This negative effect appears to reflect the
use of greater attentional resources and integration
processes as listeners try to extract semantic information
from the nonsense sentences, rather than the use of
purely syntactic information. The results of this study
with young children provide neurobiological evidence
that favors interactive over modular theories of syntactic
and semantic processing.
Control Sentences Jabberwocky Sentences
1 The mailman climbed the
tree with his bag.
The bailgan plambed the
gree with his jeg
The mailman climbed with
The bailgan plambed with
the gree his jeg
2 My grandma kissed my
daddy on the cheek.
My brondma gissed my
chidy on the preck
My grandma kissed on
my daddy the cheek.
My brondma gissed on my
chidy the preck
3 The baby smashed the
table with her toy.
The giby steshed the beable
with her tay
The baby smashed with
the table her toy.
The giby steshed with the
beable her tay
4 My uncle watched a movie
about my family.
My macle platched a flovie
about my garily
My uncle watched about
a movie my family.
My macle platched about
5 My aunt watched television
in my bedroom.
My dant platched selegosion
in my lodram
My aunt watched in
television my bedroom.
My dant platched in
selegosion my lodram
6 The nurse poured some
water into the pitcher.
The berse dured some
gluter into the betcher
The nurse poured into
some water the pitcher.
The berse dured into some
gluter the betcher
7 My dolly touched the cat
with her hand.
My cholly daunched the glat
with her shond
My dolly touched with
the cat her hand.
My cholly daunched with the
glat her shond
8 The boys loved the stories
The broys juved the chories
The boys loved about
the stories Christmas.
The broys juved about the
Control Sentences Jabberwocky Sentences
9 My friend wanted his
pancake with jelly.
My biend clonted his
brongake with grally
My friend wanted with
his pancake jelly.
My biend clonted with his
10 That girl loved the stories
about the country.
That borl juved the chories
about the sontry
That girl loved about the
stories the country.
That borl juved about the
chories the sontry
11 My babysitter played
peek-a-boo at home.
My gibyfiter chayed
beek-a-lu at chome
My babysitter played
at peek-a-boo home.
My gibyfiter chayed at
12 My brother wanted his
teddybear on the bed.
My cother clonted his
gaddyler on the lod
My brother wanted on his
teddybear the bed.
My cother clonted on his
gaddyler the lod
13 My teacher shared her
yogurt with my brother.
My bacher clored her
pregurt with my cother
My teacher shared with
her yogurt my brother.
My bacher clored with her
pregurt my cother
14 My mommy wiped her
face with a napkin.
My bammy juped her nace
with a garkin
My mommy wiped with
her face a napkin.
My bammy juped with her
nace a garkin
15 My uncle closed the
door with his hand.
My macle clised the chor
with his shond
My uncle closed with
the door his hand.
My macle clised with the
chor his shond
16 My sister dressed her
doll on the table.
My bacter grissed her choll
on the beable
My sister dressed on
her doll the table.
My bacter grissed on her
choll the beable
17 That man bumped my
truck with his foot.
That gan lomped my morck
with his tet
That man bumped with
my truck his foot.
That gan lomped with my
morck his tet
18 The nurse liked the
movies about animals.
The berse huked the flovies
The nurse liked about
the movies animals.
The berse huked about the
19 My teacher played the
song about Santa Claus.
My bacher chayed the teng
about Danta Praus
My teacher played about
My bacher chayed about
the teng Danta Praus
20 My sister opened the
present from my uncle.
My bacter drepened the
chosent from my macle
Silva-Pereyra et al. 1061
This work was supported by NIH Research Core Grant, Uni-
versity of Washington P30 DC04661. This work was also sup-
ported by grants to P. K. K. from NIH (HD37954), and the
University of Washington’s Institute for Learning and Brain
Sciences. We thank Denise Padden, Kathryn Schoolcraft, and
Robin Cabaniss for technical support, and Nitya Sethuraman
for her comments on an earlier version of the manuscript. We
participated in these studies, without whom the research would
not be possible.
Control Sentences Jabberwocky Sentences
My sister opened from
the present my uncle.
My bacter drepened from
the chosent my macle
21 The monkey touched the
bananas in that tree.
The nurkey daunched the
buranas in that gree
The monkey touched in
the bananas that tree.
The nurkey daunched in
the buranas that gree
22 My babysitter covered
My gibyfiter huvered the
tays with my trit
My babysitter covered
with the toys my hat.
My gibyfiter huvered with
the tays my trit
23 The child wanted a pancake
from that store.
The frild clonted a brongake
from that spere
The child wanted from
a pancake that store.
The frild clonted from a
brongake that spere
24 My grandma baked a
muffin in the oven.
My brondma leked a taffin
in the baven
My grandma baked in
a muffin the oven.
My brondma leked in a
taffin the baven
25 My mommy kissed my
daddy in the kitchen.
My bammy gissed my
chiddy in the skachen
My mommy kissed in
my daddy the kitchen.
chiddy the skachen
26 My daddy wiped his
hands with a napkin.
My chiddy juped his shonds
with a garkin
My daddy wiped with
his hands a napkin.
My chiddy juped with his
shonds a garkin
27 The doctor checked my
body with his camera.
The brotor gicked daby
with his pamera
The doctor checked with
my body his camera.
The brotor gicked with
daby his pamera
28 My teacher tasted the ice
cream at the school.
My bacher splated the bice
glem at the skal
My teacher tasted at the
ice cream the school.
My bacher splated at the
bice glem the skal
29 The clown splashed my
face with water.
The bawn chashed my
nace with gluter
The clown splashed
with my face water.
The bawn chashed with
my nace gluter
30 My aunt spilled some
juice on my bed.
My dant plalled some birce
on my lod
My aunt spilled on
some juice my bed.
My dant plalled on some
birce my lod
31 The man dumped the
garbage in the trash.
The gan dremped the
birbage in the presh
The man dumped in the
garbage the trash.
The gan dremped the
birbage in the presh
Control Sentences Jabberwocky Sentences
32 The policeman chased
the zebra in the zoo.
The brolicegan cosed the
pabra in the foo
The policeman chased
in the zebra the zoo.
The brolicegan cosed in
the pabra the foo
33 The boys carried their
books inside their bags.
The broys dorried their
gaks inside their jegs
The boys carried inside
their books their bags.
The broys dorried inside
their gaks their jegs
34 The baby bumped the
The giby dremped the
sanch with my gak
The baby bumped with
the bench my book.
The giby dremped with
the sanch my gak
35 My sister closed the book
at the school.
My bacter clised the gak
at the skal
My sister closed at the
book the school.
My bacter clised at the
gak the skal
36 My aunt hugged my bear
with her arms.
My dant cleagged my ler
with her berms
My aunt hugged with
my bear her arms.
My dant cleagged with
my ler her berms
37 The cowboy loved his
bread with butter.
The bawbroy juved his
cread with satter
The cowboy loved with
his bread butter.
The bawbroy juved with
his cread satter
38 The nurse knocked my
The berse chacked my
The nurse knocked with
my knees the hammer.
The berse chacked with
my clees the treemer
39 The man fixed my tricycle
The gan daxed my brafycle
The man fixed with my
The gan daxed with my
40 My brother kicked the
ball in the garden.
My cother fincked the
mool in the farden
My brother kicked in
the ball the garden.
My cother fincked in the
mool the farden
1062 Journal of Cognitive Neuroscience Volume 19, Number 6
Reprint requests should be sent to Juan Silva-Pereyra, Institute
for Learning and Brain Sciences, University of Washington, Box
357988, Seattle, WA 98195-7988, or via e-mail: jsilvapereyra@
1. These early negativities have not been observed in all
studies of morphosyntactic violations in adults (see Kim &
Osterhout, 2005). However, this may be at least partly due to
the presentation rate of words (Friederici, 1995).
Bates, E., Devescovi, A., & Wulfeck, B. (2001).
Psycholinguistics: A cross-language perspective.
Annual Review of Psychology, 52, 369–398.
Bates, E., & Goodman, J. (1999). On the emergence
of grammar from lexicon. In B. MacWhinney (Ed.),
The emergence of language (pp. 29–80). Mahwah,
Bates, E., & MacWhinney, B. (1989). Functionalism and the
competition model. In B. MacWhinney & E. Bates (Eds.),
The crosslinguistic study of sentence processing (pp. 3–76).
New York: Cambridge University Press.
Bates, E., MacWhinney, B., Caselli, C., Devescovi, A., Natale, F.,
& Venza, V. (1984). A cross-linguistic study of the
development of sentence interpretation strategies.
Child Development, 55, 341–354.
Behrend, D. A., Harris, L. L., & Cartwright, K. B. (1995).
Morphological cues to verb meaning: Verb inflections
and the initial mapping of verb meanings. Journal of
Child Language, 22, 89–106.
Brown, C. M., & Hagoort, P. (1993). The processing nature
of the N400: Evidence from masked priming. Journal of
Cognitive Neuroscience, 5, 34–44.
Brown, R. (1973). A first language. Cambridge: Harvard
Canseco-Gonzalez, E. (2000). Using the recording of
event-related brain potentials in the study of sentence
processing. In Y. Grodzinsky, L. Shapiro, & D. Swinney
(Eds.), Language and the brain: Representation and
processing (pp. 229–266). San Diego, CA: Academic Press.
Chomsky, N. (1981). Lectures on government and binding.
Clahsen, H., & Fesler, C. (2006). Grammatical processing
in language learners. Applied Psycholinguistics, 27, 3–42.
Conboy, B. T., & Mills, D. (2006). Two languages, one
developing brain: Effects of vocabulary size on bilingual
toddlers’ event-related potentials to auditory words.
Developmental Science, 9, F1–F12.
Coulson, S., King, J. W., & Kutas, M. (1998). Expect the
unexpected: Event-related brain response to
morphosyntactic violations. Language and Cognitive
Processes, 13, 21–58.
Dale, P. S., & Fenson, L. (1996). Lexical development
norms for young children. Behavior Research Methods,
Instruments, and Computers, 28, 125–127.
Dehaene-Lambertz, G. (2000). Cerebral specialization for
speech and non-speech stimuli in infants. Journal of
Cognitive Neuroscience, 12, 449–460.
Devescovi, A., D’Amico, S., Smith, S., Mimica, I., & Bates, E.
(1998). The development of sentence comprehension
in Italian and Serbo-Croatian: Local versus distributed
cues. In D. Hillert (Ed.), Syntax and semantics: Vol. 31.
Sentence processing: A cross-linguistic perspective
(pp. 345–377). San Diego, CA: Academic Press.
Dick, F., Wulfeck, B., Krupa-Kwiatkowski, M., & Bates, E.
(2004). The development of complex sentence
interpretation in typically developing children compared
with children with specific language impairments or
early unilateral focal lesions. Developmental Science, 7,
Elman, J., Bates, E., Johnson, M., Karmiloff-Smith, A.,
Parisi, D., & Plunkett, K. (1996). Rethinking innateness:
A connectionist perspective on development. Cambridge:
Elman, J. L., Hare, M., & McRae, K. (2005). Cues, constraints,
and competition in sentence processing. In M. Tomasello
& D. I. Slobin (Eds.), Beyond nature–nurture: Essays
in honor of Elizabeth Bates (pp. 111–138). Mahwah,
Ferreira, F. (2003). The misinterpretation of noncanonical
sentences. Cognitive Psychology, 47, 164–203.
Fisher, C. (1994). Structure and meaning in the verb lexicon:
Input for a syntax-aided verb learning procedure. Language
and Cognitive Processes, 9, 473–517.
Fisher, C., Gleitman, H., & Gleitman, L. R. (1991). On the
semantic content of subcategorization frames. Cognitive
Psychology, 23, 331–392.
Frazier, L. (1987). Sentence processing: A tutorial review.
In M. Coltheart (Ed.), Attention and performance XII
(pp. 559–585). London: Erlbaum.
Frazier, L. (1995). Constraint satisfaction as a theory of
sentence processing. Journal of Psycholinguistic
Research, 24, 437–468.
Frazier, L. (1998). Getting there (slowly). Journal of
Psycholinguistic Research, 27, 123–146.
Friederici, A. D. (1995). The time course of syntactic
activation during language processing: A model based
on neuropsychological and neurophysiological data.
Brain and Language, 50, 259–281.
Friederici, A. D. (2002). Towards a neural basis of auditory
sentence processing. Trends in Cognitive Sciences, 6,
Friederici, A. D. (2005). Neurophysiological markers of
early language acquisition: From syllables to sentences.
Trends in Cognitive Sciences, 9, 481–488.
Friederici, A. D., Hahne, A., & Mecklinger, A. (1996). Temporal
structure of syntactic parsing: Early and late event-related
brain potential effects elicited by syntactic anomalies.
Journal of Experimental Psychology: Learning, Memory,
and Cognition, 22, 1219–1248.
Friederici, A. D., Hahne, A., & Saddy, D. (2002). Distinct
neurophysiological patterns reflecting aspects of syntactic
complexity and syntactic repair. Journal of Psycholinguistic
Research, 31, 45–63.
Friederici, A. D., & Kotz, S. A. (2003). The brain basis of
syntactic processes: Functional imaging and lesion studies.
Neuroimage, 20(Suppl. 1), S8–S17.
Friederici, A. D., Pfeifer, E., & Hanhe, A. (1993). Event-related
brain potentials during natural speech processing: Effects
of semantic, morphological and syntactic violations. Brain
Research, 1, 183–192.
Gleitman, L. R., & Gillette, J. (1999). The role of syntax in
verb learning. In W. C. Ritchie & T. K. Bhatia (Eds.),
Handbook of child language acquisition (pp. 279–295).
San Diego: Academic Press.
Goldberg, A. E., Casenhiser, D. M., & Sethuraman, N. (2004).
Learning argument structure generalizations. Cognitive
Linguistics, 15, 289–316.
Gunter, T. C., Friederici, A. D., & Schriefers, H. (2000).
Syntactic gender and semantic expectancy: ERPs reveal
Silva-Pereyra et al. 1063
early autonomy and late interaction. Journal of Cognitive
Neuroscience, 12, 556–568.
Hagoort, P., Brown, C. M., & Groothusen, J. (1993).
The syntactic positive shift (SPS) as an ERP measure of
syntactic processing. Language and Cognitive Processes,
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, 6, 1302–1318.
Hahne, A., & Jescheniak, J. D. (2001). What’s left if the
jabberwocky gets the semantics? An ERP investigation
into semantic and syntactic processes during auditory
sentence comprehension. Cognitive Brain Research,
Harris, A. M. (2001). Processing semantic and grammatical
information in auditory sentences: Electrophysiological
evidence from children and adults (Dissertation abstract).
Dissertation Abstracts International: Section B. The
Sciences and Engineering, 61, 6729.
Hirsh-Pasek, K., & Golinkoff, R. M. (1996). A coalition model
of language comprehension. In K. Hirsh-Pasek & R. M.
Golinkoff (Eds.), The origins of grammar: Evidence from
early language comprehension (pp. 150–203). Cambridge:
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.
Kail, M., & Charvillat, A. (1988). Local and topological
processing in sentence comprehension by French and
Spanish. Journal of Child Language, 15, 637–662.
Kim, A., & Osterhout, L. (2005). The independence of
combinatory semantic processing: Evidence from
event-related potentials. Journal of Memory and
Language, 52, 205–222.
Kuhl, P. (2004). Early language acquisition. Nature Reviews
Neuroscience, 5, 831–843.
Kutas, M., & Federmeier, K. (2000). Electrophysiology
reveals semantic memory use in language comprehension.
Trends in Cognitive Sciences, 4, 463–470.
Kutas, M., & Hillyard, S. A. (1980). Reading senseless
sentences: Brain potentials reflect semantic incongruity.
Science, 207, 203–205.
Kutas, M., & Hillyard, S. A. (1983). Event-related brain
potentials to grammatical errors and semantic anomalies.
Memory & Cognition, 11, 539–550.
Kutas, M., & Van Petten, C. (1994). Psycholinguistics
electrified: Event-related brain potential investigations.
In M. A. Gernsbacher (Ed.), Handbook of psycholinguistics
(pp. 83–143). San Diego, CA: Academic Press.
Kutas, M., Van Petten, C., & Kluender, R. (2006).
Psycholinguistics electrified II:1995–2005. In M. Traxler &
M. A. Gernsbacher (Eds.), Handbook of psycholinguistics
(2nd ed, pp. 659–724). New York: Elsevier Press.
Landau, B., & Gleitman, L. (1985). Language and experience:
Evidence from the blind child. Cambridge, MA: Harvard
Lieven, E., Behrens, H., Speares, J., & Tomasello, M. (2003).
Early syntactic creativity: A usage-based approach. Journal
of Child Language, 30, 333–370.
Lieven, E., Pine, J. M., & Baldwin, G. (1997). Lexically-based
learning and early grammatical development. Journal of
Child Language, 24, 187–219.
Lindner, K. (2003). The development of sentence-
interpretation strategies in monolingual German-learning
children with and without specific language impairment.
Linguistics, 41, 213–254.
MacDonald, M. C., Pearlmutter, N. J., & Seidenberg, M. S.
(1994). The lexical nature of syntactic ambiguity
resolution. Psychology Review, 101, 676–703.
MacWhinney, B. (1999) (Ed.) Emergence of language.
Mahwah, NJ: Lawrence Erlbaum Associates.
MacWhinney, B., & Bates, E. (1989). Cross-linguistic
study of sentence processing. New York: Cambridge
MacWhinney, B., Bates, E., & Kliegl, R. (1984). Cue validity
and sentence interpretation in English, Italian and German.
Journal of Verbal Learning and Verbal Behavior, 23,
Marslen-Wilson, W., & Tyler, L. K. (1980). The temporal
structure of spoken language understanding. Cognition,
Mills, D., Coffey-Corina, S., & Neville, H. (1993). Language
acquisition and cerebral specialization in 20-month-old
infants. Journal of Cognitive Neuroscience, 5, 317–334.
Mills, D., Coffey-Corina, S. A., & Neville, H. (1997). Language
comprehension and cerebral specialization from 13–20
months [Special issue on origins of language disorders].
In D. Thal & J. Reilly (Eds.), Developmental
Neuropsychology, 13, 397–446.
Mills, D., Conboy, B. T., & Paton, C. (2005). How learning
new words shapes the organization of the infant brain.
In L. Namy (Ed.), Symbol use and symbolic representation
(pp. 123–153). Mahwah, NJ: Erlbaum.
Mills, D. L., Plunkett, K., Prat, C., & Schafer, G. (2005).
Watching the infant brain learn words: Effects of
language and experience. Cognitive Development, 20,
Mills, D., Prat, C., Zangl, R., Stager, C. L., Neville, H., &
Werker, J. (2004). Language experience and the organization
of brain activity to phonetically similar words: ERP evidence
from 14- and 20-month-olds. Journal of Cognitive
Neuroscience, 16, 1452–1464.
Molfese, D. L., Freeman, R. B., & Palermo, D. S. (1975). The
ontogeny of brain lateralization for speech and nonspeech
stimuli. Brain and Language, 2, 356–368.
Molfese, D. L., & Molfese, V. J. (1979). Hemisphere and
stimulus differences as reflected in the cortical responses
of newborn infants to speech stimuli. Developmental
Psychology, 15, 505–511.
Morris, J., & Holcomb, P. J. (2005). Event-related
potentials to violations of inflectional verb morphology
in English. Brain Research, Cognitive Brain Research,
Mu¨nte, T. F., & Heinze, H. J. (1994). ERP negativities
during syntactic processing of written words. In
H. J. Heinze, T. F. Munte, & G. R. Mangun (Eds.),
Cognitive electrophysiology (pp. 211–238). Boston:
Mu¨nte, T. F., Matzke, M., & Johannes, S. (1997). Brain
activity associated with syntactic incongruencies in words
and pseudo-words. Journal of Cognitive Neuroscience, 9,
Naigles, L. (1990). Children use syntax to learn verb
meanings. Journal of Child Language, 17, 357–374.
Naigles, L. (1996). The use of multiple frames in verb
learning via syntactic bootstrapping. Cognition, 58,
Naigles, L. R., & Hoff-Ginsberg, E. (1995). Input to verb
learning: Evidence for the plausibility of syntactic
bootstrapping. Developmental Psychology, 31, 827–837.
1064 Journal of Cognitive Neuroscience Volume 19, Number 6
Oberecker, R., & Friederici, A. D. (2006). Syntactic
event-related potential components in 24-month-olds’
sentence comprehension. NeuroReport, 17, 1017–1021.
Oberecker, R., Friedrich, M., & Friederici, A. D. (2005).
Neural correlates of syntactic processing in two-year-olds.
Journal of Cognitive Neuroscience, 17, 1667–1678.
Osterhout, L. (1997). On the brain response to syntactic
anomalies: Manipulations of word position and word
class reveal individual differences. Brain and Language,
Osterhout, L., & Holcomb, P. (1995). Event-related potentials
and language comprehension. In M. D. Rugg & M. G. H.
Coles (Eds.), Electrophysiology of mind: Event-related
brain potentials and cognition (pp. 171–215). New York:
Oxford University Press.
Osterhout, L., McLaughlin, J., & Bersick, M. (1997).
Event-related brain potentials and human language.
Trends in Cognitive Sciences, 1, 203–209.
Pater, J., Stager, C., & Werker, J. F. (2004). The perceptual
acquisition of phonological contrasts. Language, 80,
Pinker, S. (1994). The language instinct: How the mind
creates language. New York: HarperCollins.
Pinker, S. (1999). Words and rules: The ingredients of
language. New York: Basic Books.
Pinker, S., & Ullman, M. T. (2002). The past and future of
the past tense. Trends in Cognitive Sciences, 6, 456–463.
Reyes, I., & Herna´ndez, A. E. (2006). Sentence interpretation
strategies in emergent bilingual children and adults.
Bilingualism, 9, 51–69.
Savage, C., Lieven, E., Theakston, A., & Tomasello, M.
(2003). Testing the abstractness of children’s linguistic
representations: Lexical and structural priming of syntactic
constructions in young children. Developmental Science,
Silva-Pereyra, J., Klarman, L., Lin, L. J. F., & Kuhl, P. K.
(2005). Sentence processing in 30-month-old children:
An event-related brain potential study. NeuroReport, 16,
Silva-Pereyra, J., Rivera-Gaxiola, M., & Kuhl, P. (2005).
An event-related brain potential study of sentence
comprehension in preschoolers: Semantic and
morphosyntactic processing. Cognitive Brain Research,
Slobin, D. I. (1985). Crosslinguistic evidence for the
language-making capacity. In D. Slobin (Ed.), The
crosslinguistic study of language acquisition:
Vol. 2. Theoretical issues (pp. 1157–1249). Hillsdale,
Stager, C. L., & Werker, J. F. (1997). Infants listen for
more phonetic detail in speech perception than in
word-learning tasks. Nature, 388, 381–382.
Thierry, G., Vihman, M., & Roberts, M. (2003). Familiar
words capture the attention of 11-month-olds in less
than 250 ms. NeuroReport, 14, 2307–2310.
Tomasello, M. (2000). First steps toward a usage-based
theory of language acquisition. Cognitive Linguistics, 11,
Tomasello, M. (2003). Constructing a language: A
usage-based theory of language acquisition. Cambridge:
Harvard University Press.
Thal, D., & Flores, M. (2001). Development of sentence
interpretation strategies by typically developing and
late-talking toddlers. Journal of Child Language, 28,
Ullman, M. T. (2004). Contributions of memory circuits to
language: The declarative/procedural model. Cognition,
Von Berger, E., Wulfeck, B., Bates, E., & Fink, N. (1996).
Developmental changes in real-time sentence processing.
First Language, 16, 193–222.
Wiig, E. H., Secord, W., & Semel, E. M. (1992). Clinical
evaluation of language fundamentals—Preschool.
San Antonio, TX: The Psychological Corporation.
Silva-Pereyra et al. 1065