Electrophysiological insights into language processing
TATIANA SITNIKOVA,aDEAN F. SALISBURY,b,cGINA KUPERBERG,b,d,e
and PHILLIP J. HOLCOMBa
aDepartment of Psychology, Tufts University, Medford, Massachusetts, USA
bDepartment of Psychiatry, Harvard Medical School, Boston, Massachusetts, USA
cMcLean Hospital, Belmont, Massachusetts, USA
dDepartment of Psychiatry, Massachusetts General Hospital, Charlestown, Massachusetts, USA
eDepartment of Psychological Medicine, Institute of Psychiatry, London, Great Britain
Deficits in language comprehension in schizophrenia were examined using event-related potentials ~ERPs!. Schizo-
phrenic and healthy participants read sentences in which the first clause ended with a homograph, and the second clause
started with a target word that was semantically related to the homograph’s dominant meaning ~e.g., 1. Diving was
forbidden from the bridge because the river had rocks in it. or 2. The guests played bridge because the river had rocks
in it.!. Processing of the targets ~e.g., “river”! was expected to be primarily influenced by the preceding overall sentence
context ~congruent in 1; incongruent in 2! in healthy participants, but to be inappropriately affected by the dominant
meaning of homographs ~e.g., the “structure” meaning of “bridge”! in sentences like 2 in schizophrenic patients. The
N400 ERP component that is known to be sensitive to contextual effects during language processing confirmed these
predictions. This showed that language abnormalities in schizophrenia may be related to deficient processing of
context-irrelevant semantic representations of words from the discourse.
Descriptors: Schizophrenia, Language, Context-irrelevant semantic associations, Event-related potentials, N400
For over a century, positive thought disorder has been considered
one of the core symptoms of schizophrenia. The phenomenology
of this symptom has been extensively documented and includes,
primarily, a pattern of incoherent speech, but may also extend to
deficits in language comprehension ~American Psychiatric Asso-
ciation, 1994; Kuperberg, McGuire, & David, 1998; Maher, Man-
schreck, & Rucklos, 1980; Speed, Toner, Shugar, & Di Gasbarro,
1991!. Unfortunately, the cognitive and neural processes underly-
ing schizophrenic positive thought disorder are not well under-
stood, and there have been a number of competing accounts that
attempt to explain this deficit. Below, we review evidence for two
proposals that attempt to explain positive thought disorder: defects
in context sensitivity and problems in semantic memory.
According to one hypothesis, schizophrenic patients have a
reduced sensitivity to the information provided by earlier elements
in a conversation or text ~so-called discourse context!. For exam-
ple, in one group of studies, schizophrenic patients have been
shown to be impaired in using the context of sentences to deter-
mine an appropriate meaning of a homographic word ~i.e., a word
with multiple meanings; Benjamin & Watt, 1969; Blaney, 1974;
Chapman, Chapman, & Miller, 1964; Strauss, 1975!. Additional
evidence for the context-insensitivity hypothesis comes from a
recent report that found that contextual facilitation of word recog-
nition was reduced in thought-disordered schizophrenic patients in
comparison to healthy volunteers and schizophrenic patients with
no thought disorder ~Kuperberg et al., 1998!.
There have also been attempts to pinpoint the specific problems
that patients with schizophrenia have in processing contextual
information. For example, Cohen and Servan-Schreiber ~1992!
demonstrated that schizophrenic patients have difficulty using sen-
tence context to interpret correct meanings of multimeaning words,
but only when the disambiguating context precedes the target
homograph and not when it follows the homograph. These authors
suggested that utilizing contextual information may be difficult for
patients with schizophrenia because of their inability to maintain
this information in memory. Another group of researchers ~Titone,
Levy, & Holzman, 2000! contrasted the ability of patients with
schizophrenia to use context to facilitate processing of upcoming
information and to suppress irrelevant information. Their study
found that when a sentence context moderately biased a less
This research was supported in part by a grant from the National
Alliance for Research in Schizophrenia and Depression to DFS.
We thank Deirdre C. Farrell, Iris A. Fischer, and Courtney Brown for
their assistance in collecting the data, and Brendan A. Maher and David
Harder for informative discussions on the design and data as well as
comments on drafts of this article.
Address requests for reprints to: Tatiana Sitnikova, Department of
Psychology, Tufts University, Medford, MA 02155, USA. E-mail:
Psychophysiology, 39 ~2002!, 851–860. Cambridge University Press. Printed in the USA.
Copyright © 2002 Society for Psychophysiological Research
frequently used meaning of a sentence final homographic word,
postsentence target words that were semantically related to this
meaning were primed in control subjects and schizophrenic pa-
tients alike. However, postsentence targets that were related to the
alternative, more frequently used meaning of the homographs were
primed only in patients with schizophrenia. This suggested that
patients with schizophrenia may have a specific deficit in inhibit-
ing context-inappropriate information, but are rather competent in
detecting the relevant context. These authors suggested that the
observed deficit in suppression might be related to general prob-
lems that schizophrenic patients have with inhibitory processing
that also are observed in such tasks as the Stroop Test ~e.g., Cohen,
Barch, Carter, & Servan-Schreiber, 1999! and the Trail Making
Task ~e.g., Flashman, Flaum, Gupta, & Andreasen, 1996!.
Asecond major account of positive thought disorder focuses on
abnormalities in semantic memory. One version of this theory was
generated to explain the greater facilitation in word-recognition
speed that occurs in schizophrenic patients with positive thought
disorder when target words are preceded by a semantically related
rather than unrelated word ~i.e., greater semantic priming; e.g.,
Kwapil, Hegley, Chapman, & Chapman, 1990; Manschreck et al.,
1988; Spitzer, Braun, Hermle, & Maier, 1993!. This so-called
hyperpriming has been proposed to result from increased auto-
matic activity within semantic memory in thought-disordered pa-
tients and might account for the context-inappropriate semantic
associations that frequently intrude into patients’discourse ~Maher,
1983!. Unfortunately, a number of studies have failed to support
this hypothesis, finding normal or reduced semantic priming
in schizophrenia ~e.g., Barch et al., 1996; Vinogradov, Ober, &
Asecond semantic-memory account holds that positive thought
disorder results from an abnormal pattern of connections between
to lead to atypical organization of concepts in patients’speech. The
evidence for this view comes from studies that have found an
association between thought disorder and an impaired ability to
use semantic-relatedness information for word generation ~Gold-
berg et al., 1998; Kerns, Berenbaum, Barch, Banich, & Stolar,
1999!. Moreover, schizophrenic patients with positive thought
disorder, unlike healthy and non-thought-disordered patient con-
trols, have been demonstrated to show less semantic priming with
strongly related than with mildly related word pairs ~Aloia et al.,
1998!, and word recall of schizophrenic patients cued by semantic
associates is influenced less by the number of connections that
each word has in semantic memory ~Nestor et al., 1998!.
Thus, a number of explanations ranging from general deficits in
memory and inhibitory mechanisms ~e.g., Cohen and Servan-
Schreiber, 1992; Titone et al., 2000! to more specific impairments
in semantic memory ~e.g., Goldberg et al., 1998; Manschreck
et al., 1988! have been proposed to account for the positive thought
disorder in schizophrenia. However, it is still an open question how
or if these seemingly somewhat contradictory hypotheses relate to
The dependent variables in all of the studies cited above in-
volved overt measures of behavior that are notorious for being
sensitive to cognitive activity over and above the contextual and
semantic processes of interest ~e.g., activity related to the demands
of an experimental task; see Kounios & Holcomb, 1992!. One
technique that offers considerable promise in providing a more
pure measure of semantic processing is the recording of event-
related potentials ~ERPs!. Of particular relevance are previous
ERPstudies of language comprehension in healthy individuals that
have identified a negative-going deflection in the ERP wave peak-
ing around 400 ms poststimulus onset ~the N400 component; e.g.,
Kutas & Hillyard, 1980!.According to one view, the N400 reflects
cognitive processes whereby a word’s meaning is integrated into a
larger contextual framework ~the context can be a single word,
sentence, or larger discourse!, the size of the N400 being propor-
tional to mental effort required for such integration ~Brown &
Hagoort, 1993; Holcomb, 1993!. These findings in healthy com-
prehenders suggest that ERPs might offer a more direct means to
assess language processing in schizophrenia.
A number of studies have examined language comprehension
in schizophrenia using ERPs. In many of these investigations,
similar to healthy participants, patients with schizophrenia elicited
larger N400s to words preceded by the incongruent than congruent
context ~i.e., a relatively intact N400 congruity effect; Andrews
et al., 1993; Grillon, Ameli, & Glazer, 1991; Koyama et al., 1994;
Nestor et al., 1997; Niznikiewicz et al., 1997; Olichney, Iragui,
Kutas, Nowacki, & Jeste, 1997; Spitzer, 1997!. However, three
studies have reported a lack of the N400 contextual modulation in
sentences ~Adams et al., 1993; Ohta, Uchiyama, Matsushima, &
Toru, 1999; Salisbury, O’Donnell, McCarley, Nestor, & Shenton,
2000!, and two additional studies have observed a reduced N400
congruity effect in word pairs ~Condray, Steinhauer, Cohen, van
Kammen, & Kasparek, 1999; Strandburg et al., 1997!. Taken
together, these results suggest that although, in general, patients
with schizophrenia may be able to process and utilize linguistic
context, under certain experimental conditions, this ability may be
The goal of the present study was to use ERPs to investigate
one set of the experimental variables that have been previously
suggested to result in processing problems in patients with schizo-
phrenia. That is, we examined whether ERPs would reveal abnor-
malities in the processing of sentences that included words that can
potentially activate context-inappropriate semantic associations.
Based on earlier research, we predicted that in schizophrenic
patients, such associations may not be sufficiently inhibited due to
their excessive activation ~Maher, 1983! and0or due to a deficient
suppression mechanism ~Titone et al., 2000!, and therefore would
interfere with semantic integration of the stimuli.
More specifically, the present study attempted to advance our
knowledge about the effects that such unsuppressed associations
may have on natural language processing. Much of the earlier
work reviewed above was restricted to word pairs ~e.g., Man-
schreck et al., 1988! or measured responses to items outside of the
sentences of interest ~e.g., Titone et al., 2000!. It is therefore
unclear which effects would occur in more natural language tasks.
Furthermore, the effects in these earlier studies were observed only
over very brief time periods ~e.g., semantic hyperpriming was
detectable only with SOAs shorter than 500 ms; and in the sup-
pression experiment, the targets immediately followed the critical
words activating irrelevant associations!. Nevertheless, some mod-
els of language processing suggest that these effects may have a
longer time course during processing of natural language ~e.g.,
Gernsbacher, Varner, & Faust, 1990!. Thus, the goal of the present
study was to determine if inadequate suppression of context-
inappropriate associations has an effect on comprehension of words
in sentences, and whether these effects have a prolonged time
To achieve this goal, we designed sentences that elicited context-
inappropriate associations. Within these sentences, we embedded
target words that could potentially be affected by these associa-
tions. Furthermore, we asked participants to read the sentences for
T. Sitnikova et al.
comprehension.1To show that the effects of irrelevant associations
may extend over longer time intervals, target words followed the
critical words activating the context inappropriate associations
with a time lag of at least 1,500 ms.
The design was based on earlier findings in healthy participants
reading sentences containing homographs. In normal readers, when
the preceding context biases a particular meaning of a homograph,
activation of the alternative ~context-inappropriate! meaning~s!
has been shown to be suppressed2~e.g., Gernsbacher & Faust,
1990!. However, in the absence of context, a more frequently used
~dominant! meaning has been shown to be activated more strongly
than less frequently used ~subordinate! meaning~s! ~Simpson &
Burgess, 1985!. The logic of the current study was based on these
previous results. It was reasoned that because the activation of
context-inappropriate meanings is suppressed when the homo-
graph is integrated into the sentence context, these meanings nor-
mally should have no effect on the processing of later words in a
sentence. However, when the homograph is not successfully inte-
grated with the context, the effects of the context-inappropriate
dominant meaning should be noticeable.
In the current experiment we used sentences such as the ones
shown in @1# and @2#:
@1# The book must have great stories because the author won an
award0escape for it.
@2# The skyscraper had ninety stories because the author won an
award0escape for it.
In the experimental comparison, ERPs were recorded to target
words at the beginning of the second clause ~e.g., “author”!. In
50% of the sentences, target words were congruent with the overall
context of the first clause ~see @1# above!, and in the other 50%,
target words were incongruent with the overall context of the first
clause ~see @2# above!. However, in both cases, the target words
were semantically related to the dominant meaning of the homo-
graph that ended the first clause ~e.g., “stories”!.
Healthy participants were predicted to produce the typical N400
congruity effect when the target words that were congruent with
the context set up by the first clause were compared to the target
words that were semantically incongruent with first-clause mean-
ing ~e.g., when words like “author” in @1# were compared to words
like “author” in @2#!. In contrast, this N400 congruity effect was
predicted to be attenuated or absent in patients with schizophrenia.
This could be the case due to either of two deficits hypothesized in
schizophrenia: ~a! because of an inadequate contextual suppression
mechanism; and0or ~b! because of increased levels of word-
meaning activation ~semantic hyperactivity!. In either case, the
context-inappropriate dominant meaning of the homograph in sen-
tences like @2# was expected to be insufficiently inhibited and to
affect processing of target words to which it was semantically
related ~e.g., the “tales” meaning of “stories” in @2# was expected
to affect processing of “author”!. More specifically, the context-
inappropriate dominant meaning of the homograph in sentences
like @2# was hypothesized to provide a congruent context for the
target word, and therefore, semantic integration of the incongruent
target words was expected to be relatively easy. As a result, the
N400 was predicted to be small not only in the congruent but also
in the incongruent condition, which would serve to reduce the
N400 congruity effect at the target words.
The present study also recorded ERPs to words at the end of
sentences like @1#. In half of the sentences, these items were
congruent with the preceding context ~e.g., “award” in “The book
must have great stories because the author won an award for it.”!
and in half they were incongruent ~e.g., “escape” in “The book
must have great stories because the author won an escape for it.”!.3
In this control condition, the incongruent target words were se-
mantically unrelated to all words preceding them in the sentence,
and therefore were less likely to be semantically primed by any
item in the prior context. The rationale for including this compar-
ison was as a control for contextual effects in the absence of
context-irrelevant associations. If, in general, contextual effects are
preserved in schizophrenic patients, then there should be no dif-
ference in the size of the control N400 effect between patients and
Twelve patients with schizophrenia recruited at McLean psychiat-
ric hospital and 12 healthy participants took part in this study. The
diagnosis of schizophrenia was established according to clinical
charts and the Structured Clinical Interview for the DSM-IIIR
~SCID; Spitzer, Williams, Gibbon & First, 1990a!. Healthy par-
ticipants were recruited with newspaper advertisements and were
invited to participate if they had no lifetime and family history of
psychopathology based on the Structured Clinical Interview for
DSM-IIIR—Non-Patient Edition ~SCID-NP; Spitzer,Williams, Gib-
bon & First., 1990b!. Additional inclusion criteria applied to all
participants were: native English speaker, normal or corrected-to-
normal vision, and no lifetime history of head trauma, neurological
disorder, or substance abuse.
Demographic, cognitive, and clinical characteristics of partici-
pants are shown in Table 1. The cognitive measures that included
the Mini-Mental State Examination ~MMSE; Folstein, Folstein, &
1After reading each sentence, participants were required to decide
whether a test word presented after the offset of the sentence referred to one
of the ideas in the sentence. This task was chosen because arguably it
encourages reading for comprehension but does not engage strategic pro-
cesses other than those involved during normal reading.
2Three major types of models of word-meaning ambiguity resolution
have been proposed. Selective-access models hold that only a context-
appropriate meaning of a word gets activated ~e.g., Perfetti and Goodman,
1970!; exhaustive-access models maintain that all meanings of a word are
initially activated, but thereafter context-inappropriate meanings are inhib-
ited ~e.g., Gernsbacher and Faust, 1990!; and ordered-access models posit
that the dominant meaning is generally retrieved more readily than the
subordinate meaning ~e.g., Hogaboam and Perfetti, 1975!. The hypotheses
proposed in this article are compatible with all three views. The target
words in the present study were presented at least 1,500 ms after the onset
of homographs: all three kinds of models hold that in healthy participants,
the context-appropriate meaning of a visually presented word is selected
within 750 ms. Furthermore, selective-access models generally maintain
that contextual influence simply prevents irrelevant meanings from acti-
vation, whereas exhaustive-access models posit that contextual effects
inhibit irrelevant meanings that initially are automatically activated.Again,
both conceptions are compatible with the present hypotheses: We assume
that the contextual inhibitory mechanism is responsible for keeping irrel-
evant meanings from interfering with language processing, should they
be prevented from being activated, or suppressed after being activated
3To achieve a more balanced design, we included incongruent words
like “escape” at the end of the sentences like @2# as well as in sentences like
@1#. However, we collected ERPs to words at the end of the second clause
only in sentences like @1#. Thus, these control target words were preceded
by a coherent sentence stem.
Language in schizophrenia and ERPs
McHugh, 1975! and Information and Vocabulary subscales of the
Wechsler Adult Intelligence Scale–Revised ~WAIS-R; Wechsler,
1981! were administered to all participants. The clinical state of
patients with schizophrenia was evaluated using the Brief Psychi-
atric Rating Scale ~BPRS; Overall & Gorham, 1962!, the Scale for
the Assessment of Positive Symptoms ~SAPS; Andreasen, 1984a!,
and the Scale for the Assessment of Negative Symptoms ~SANS;
Andreasen, 1984b!. The total BPRS score was used to assess
general level of psychopathology. Positive thought disorder was
determined as a composite of SAPS and SANS scales measuring
derailment, poverty of content of speech, circumstantiality, tan-
gentiality, and incoherence ~a similar procedure for assessment of
positive thought disorder was used in Goldberg et al., 1998!.
Materials included 120 foursomes of sentences that were 11 to 14
words long and consisted of two clauses interconnected with the
conjunction “because.” Examples of sentences are given in sen-
tences @3# to @6#.
@3# Diving was forbidden from the bridge because the river had
rocks in it.
@4# Diving was forbidden from the bridge because the river had
cracks in it.
@5# Guests played bridge because the river had rocks in it.
@6# Guests played bridge because the river had cracks in it.
In all sentences, the first clause ended with a homograph ~e.g.,
bridge!, and the second clause started with a strong semantic
associate of the homograph’s dominant meaning ~e.g., river!. The
latter served as a target word in the experimental condition. Ho-
mographs and target words were selected based on published
normative studies where participants were asked to produce a word
association for each homograph ~Geis & Winograd, 1974; Gorfein,
Viviani, & Leddo, 1982; Kausler & Kollasch, 1970; Nelson,
McEvoy, Walling, & Wheeler, 1980; Twilley, Dixon, Taylor, &
Clark, 1994!. Homographs were chosen if they had two alternative
noun meanings, and if they prompted an uneven distribution of
responses across these meanings, that is, more than 70% of par-
ticipants responded with a word semantically related to one of the
homograph’s meanings. This meaning was considered dominant,
whereas an alternative meaning that received less than 30% of
responses was considered subordinate. The most frequent noun
response produced for the dominant meaning was used as a target
Although the sentences in each foursome had the same target
word, their first clauses were different. In two sentences, the first
clause biased the homograph’s dominant meaning and was con-
gruent with the target word ~e.g., sentences like @3# and @4#!,
whereas in the other two sentences, it biased the subordinate
meaning and was incongruent with the target word ~e.g., sentences
like @5# and @6#!. The dominant-meaning biasing clauses and the
subordinate-meaning biasing clauses did not differ in their length
~p ? 1; on average they were 6.5 words long!, or cloze probability
of the homograph within the clause ~p ? 1; on average the cloze
probability was .25!. The cloze probability was determined in a
separate sample of healthy readers ~N ? 15!, who were given the
first clause of the sentences with a blank in place of the homo-
graph, and were asked to fill in the blank with the first word fitting
the context that came to their mind.
First, we designed sentences in which the second clause was in
itself coherent ~e.g., sentences like @3# and @5#!. Then, in these
sentences, we chose one noun towards the end of the second clause
~e.g., “rocks”!, and created sentences like @4# and @6# by rearrang-
ing the selected words across the sentences so that they were
incongruent with the prior context and were not preceded by any
words that were semantically related to them ~e.g., “rocks” was
replaced by “cracks” in sentences like @3# and @5# to form sen-
tences like @4# and @6#!. The words in the end of the second clause
in sentences like @3# and @4# served as control target words ~i.e., in
the control condition words like “rock” in @3# were compared to
words like “cracks” in @4#!. However, ERPs were not recorded
to words like “rocks” and “cracks” in sentences like @5# and @6#. To
confirm that no semantically related words preceded incongruent
words like “cracks” in the sentences, we collected normative data
in a sample of healthy readers ~N ? 10!. Participants were pre-
sented with content words ~nouns, verbs, and adjectives! preceding
the incongruent word in each sentence. For a given sentence, these
content words were arranged in a random order, and were followed
by the incongruent word itself that was underlined. Participants
were instructed to circle the words that were related to the under-
lined word ~on average 0.5 words were circled—both in sentences
like @4# and sentences like @6#!.
The sentences were arranged into four lists, each consisting of
30 sentences of each type ~120 sentences in each list; for a full set
of materials visit http:00neurocog.psy.tufts.edu0papers!. The as-
signment of sentences to lists was such that only one sentence from
each sentence foursome was included in each list. Across each
group of participants, each list was presented to the same number
of people. Twelve additional sentences were constructed to be used
in a practice session.
Finally, test words were selected for each sentence that either
referred to one of the ides in a sentence or were inconsistent with
the sentence ~e.g., “prohibition” was chosen as a consistent test
word for the sample sentence @3#, and “thief” was used as an
inconsistent one!. Each participant was given consistent test words
with one half of each type of sentences. Further, one half of
consistent test words for sentences like @5# and @6# referred to the
Table 1. Demographic, Cognitive, and Clinical Characteristics
Positive thought disorder
Neuroleptic dose ~mg CPZE daily!
Length of illness ~years!
Note: Values for continuous variables represent means ~with SD!.
aPatients significantly different from healthy controls, p ? .05 ~t test!.
bOldfield ~1971!. ?1: left-handed; 1: right-handed.
cSocioeconomic status; Hollingshead ~1965!. 5: lowest; 1: highest.
T. Sitnikova et al.
first clause and the other half referred to the second clause.Aword
that was used as an inconsistent test word with one sentence was
used as a consistent test word with another sentence.
Participants were seated in a comfortable chair in a sound-
attenuated chamber approximately 1 m from the computer monitor
used for a sentence presentation. Viewing of each sentence started
with a fixation point ~a cross! presented for 650 ms and followed
by a blank screen for 100 ms. Then, each word of the sentence was
presented for 650 ms followed by a blank screen for 100 ms. After
the sentence final word, the screen was blank for 800 ms, and then
the test word appeared in capital letters for 3,000 ms followed by
a blank screen for 1,100 ms, after which the next sentence was
Participants were instructed to read each sentence naturally as
they would a newspaper or a novel, and decide whether the test
word presented shortly after the sentence referred to one of the
main ideas in the sentence ~semantic judgment task!. Participants
were asked to signal their decision by pressing a “YES” or “NO”
button at the time they saw the test word. Before the experiment,
participants were familiarized with the task in a practice session.
An elastic cap ~Electrode-Cap International! with 28 tin electrodes
was fitted to the participant’s head. Six additional electrodes were
attached to the left and right linked earlobes ~serving as a refer-
ence!, above and below the right eye ~to monitor vertical eye
movements and eye blinks!, and at the outer canthi of the eyes ~to
monitor horizontal eye movements!. The impedance of all elec-
trodes was below 3 kV, and the earlobe electrodes were matches
within 1 kV.The recorded electroencephalogram ~EEG! and electro-
oculogram ~EOG! was amplified with a bandpass of 0.15 to 40 Hz
and was digitized at a sampling rate of 3.5 ms0sample for a
sampling epoch of 1,000 ms duration ~100 ms prestimulus and
900 ms poststimulus!.
The EEG data were corrected for eye-movement artifacts using
regression-based weighting coefficients ~Semlitsch, Anderer,
Schuster, & Presslich, 1986! and were averaged time locked to the
onset of stimuli of interest. For each participant, separate average
ERPs were acquired at the congruent and incongruent target words
in the experimental and control conditions ~i.e., in the experimen-
tal comparison, average ERPs to the words like “river” were ob-
tained across sentences like @3# and @4# in the congruent condition,
and across sentences like @5# and @6# in the incongruent condition;
in the control comparison, average ERPs to the words like “rocks”
in sentences like @3# were obtained in the congruent condition, and
to the words like “cracks” in sentences like @4# in the incongruent
Mixed design analyses of variance ~ANOVAs! having within-
subjects factor of Congruity ~congruent and incongruent! and a
between-subjects factor of Group ~patients with schizophrenia and
healthy participants! were used to analyze group differences in the
N400. The N400 magnitude was determined as a mean voltage
value ~relative to a 100-ms prestimulus baseline! for the ERP
points in the time window between 300 and 550 ms poststimulus
onset ~the N400 window!. Mean ERPs in 0–300-ms and 550–
800-ms windows were also measured. The analyses in the exper-
imental and control conditions included four separate ANOVAs
each. These ANOVAs were conducted to examine parasagittal
columns of scalp electrodes along the anterior–posterior axis of the
head. The midline analysis included four levels of electrode site
~Fz, Cz, Pz, Oz!. The medial-parietal analysis had two levels of
electrode site and two levels of hemisphere ~C30C4, CP10CP2!.
The lateral analysis had five levels of electrode site and two levels
of hemisphere ~F30F4, FC50FC6, CP50CP6, P30P4, PO10PO2!.
The peripheral analysis had five levels of electrode site and two
levels of hemisphere ~FP10FP2, F70F8, T30T4, T50T6, O10O2!.
The Geisser–Greenhouse correction was applied when repeated
measures analyses had more than one degree of freedom ~Geisser
& Greenhouse, 1959!.
Response accuracy on the semantic judgment task was deter-
mined for each participant. Further, because schizophrenia and
healthy groups were not matched on a number of demographic and
cognitive variables ~see Table 1!, the potential confounding effects
of these variables on the N400 group differences were assessed by
computing Pearson’s correlations between these variables and the
N400 congruity effect at the target words in the experimental
condition in the overall sample of participants ~i.e., healthy and
schizophrenic participants combined!. For this analysis, the N400
was computed as an average voltage across Fz, Cz, Pz, Oz, F7, F8,
T5, and T6 electrode sites.
Event-Related Potential Data
Target words—experimental condition. The average ERPs time-
locked to the target words in the experimental condition are shown
for healthy participants in Figure 1A and for patients with schizo-
phrenia in Figure 1B. The waveforms were similar between con-
gruent and incongruent target words and the participant groups
until about 300 ms poststimulus onset. No differences in the
average ERPs between 0 and 300 ms were significant. Note that
robust sensory potentials ~N1 and P2! could be seen both in
healthy participants and patients with schizophrenia.
In the 300–550-ms time window, the ERP pattern was different
between healthy and schizophrenia groups ~Congruity ? Group
interaction: midline F~1,22! ? 8.596, p ? .01, effect size ? .281;
medial-parietal F~1,22! ? 9.641, p ? .01, effect size ? .305;
lateral F~1,22! ? 10.324, p ? .01, effect size ? .319!. Follow-up
analyses demonstrated that in healthy participants, a negative-
going deflection that occurred during this time interval ~the N400!
was larger at incongruent than congruent target words ~main effect
of Congruity: midline F~1,11! ? 10.704, p ? .01, effect size ?
.493; medial-parietal F~1,11! ? 12.625, p ? .01, effect size ?
.534; lateral F~1,11! ? 9.826, p ? .01, effect size ? .472; periph-
eral F~1,11! ? 4.912, p ?.049, effect size ? .309!. In contrast, in
patients with schizophrenia, there was no significant N400 differ-
ence between the two types of words ~main effect of Congruity:
midline F~1,11! ? 0.462, n.s., effect size ? .040; medial-parietal
F~1,11! ? 0.241, n.s., effect size ? .021; lateral F~1,11! ? 0.948,
n.s., effect size ? .079; peripheral F~1,11! ? 0.000, n.s., effect
size ? .000!. Furthermore, whereas there were no significant
difference in the congruent targets between patients with schizo-
phrenia and healthy participants ~main effect of Group: midline:
F~1,22!?0.051, n.s., effect size?.002; medial-parietal F~1,22!?
0.223, n.s., effect size ? .010; lateral F~1,22! ? 0.000, n.s., effect
size ? .000; peripheral F~1,22! ? 0.046, n.s., effect size ? .002!,
the N400 to the incongruent words was significantly smaller in
4.557, p ?.044, effect size ? .172; medial-parietal F~1,22! ?
Language in schizophrenia and ERPs
5.819, p ? .025, effect size ? .209; lateral F~1,22! ? 5.380, p ?
.030, effect size ? .197!.
In the 550–800-ms time window, the waveforms were similar
between healthy and schizophrenia groups. Further, no differences
were obtained between the congruent and incongruent conditions.
Target words—control condition. Figure 2 shows the average
ERPs elicited by the targets in the control condition in healthy and
schizophrenia groups. Again, no significant differences were
obtained for the average ERPs between 0 and 300 ms after stim-
In the 300–550-ms window, a negative-going wave ~the N400!
was larger to the incongruent than congruent words ~main effect of
Congruity: midline F~1,22! ? 26.008, p ? .01, effect size ? .542;
medial-parietal F~1,22! ? 25.509, p ? .01, effect size ? .537;
lateral F~1,22! ? 26.844, p ? .01, effect size ? .550; peripheral
F~1,22! ? 17.000, p ? .01, effect size ? .436!. Importantly, this
congruity effect did not interact with the Group factor ~midline:
F~1,22!?1.468, n.s., effect size?.063; medial-parietal F~1,22!?
1.290, n.s., effect size ? .055; lateral F~1,22! ? 1.137, n.s., effect
size ? .049; peripheral F~1,22! ? 1.044, n.s., effect size ? .045!,
Figure 1. A: Average ERPs elicited by target words in the experimental condition in healthy participants. B: Average ERPs elicited
by target words in the experimental condition in patients with schizophrenia.
T. Sitnikova et al.
3.351, n.s., effect size ? .132; medial-parietal F~1,22! ? 3.424,
n.s., effect size ? .135; lateral F~1,22! ? 1.947, n.s., effect size ?
.081; peripheral F~1,22! ? 0.201, n.s., effect size ? .009!. In both
participant groups, the N400 effect at Cz and Pz electrodes peaked
around 410 ms after the target-word onset ~the peak latency was
defined as the point where the difference in the ERPs between
congruent and incongruent conditions was maximal!. A mixed-
design ANOVA with a within-subjects factor of Electrode ~Cz and
Pz! and a between-subjects factor of Group revealed no latency
differences—no main effects ~Electrode: F~1,22! ? 2.481, n.s.,
effect size?.101; Group: F~1,22!?0.083, n.s., effect size?.004!
and no interaction between the factors, F~1,22! ? 0.870, n.s.,
effect size ? .038.
The average ERPs between 550 and 800 ms after stimulus
onset were more positive at the incongruent than congruent words
at the posterior electrode sites in both participant groups. However,
this difference did not reach significance.
Patients with schizophrenia made more errors in semantic judg-
ments about the sentences than healthy participants. The accuracy
rate was 92.7 6 3.5% in healthy participants and 78.3 6 12.9% in
Figure 2. A:Average ERPs elicited by target words in the control condition in healthy participants. B:Average ERPs elicited by target
words in the control condition in patients with schizophrenia.
Language in schizophrenia and ERPs
patients with schizophrenia. However, both healthy and schizo-
phrenic participants were better in their judgments about sentences
in which the first clause biased the dominant meaning of the
homograph ~sentences like @3# and @4#; accuracy rate: healthy
controls 95.2 6 4.6%; schizophrenic patients 80.9 6 13.5%! than
about sentences in which the first clause biased the subordinate
meaning ~sentences like @5# and @6#; accuracy rate: healthy controls
90.3 6 4.6%; schizophrenic patients 75.8 6 13.5%!. This was
determined in a mixed design ANOVA with a within-subjects
factor of Sentence Type ~dominant vs. subordinate meaning bias-
ing first clause! and a between-subjects factor of Group. This
analysis revealed significant main effects of Sentence Type,
F~1,22!?11.990, p ? .01, effect size?.353, and Group, F~1,22!?
14.000, p ? .01, effect size ? .389, but no significant interaction
between these factors, F~1,22! ? 0.009, n.s., effect size ? .000.
Potential Confounding Variables
The analysis of Pearson correlations between the potential con-
founding demographic0cognitive variables and the N400 congru-
ity effect in the experimental condition conducted in the overall
sample of participants revealed no significant results. The corre-
lations were: age, ?.179, p ? 1; SES, ?.338, p ? 1; mental state,
.328, p ? 1; WAIS-R Information score, .336, p ? 1; and WAIS-R
Vocabulary score, .321, p ? 1.
The present study focused on the effects of inadequate suppression
of context-irrelevant information on language comprehension in
schizophrenia. In the primary experimental comparison, ERPs
were recorded to target words that were semantically related to the
dominant meaning of a preceding homographic word and were
either congruent or incongruent with prior sentence context. In a
control comparison, ERPs were also recorded to target words that
either were congruent or incongruent with the context of the prior
sentence stem. However, in this comparison, the incongruent tar-
gets were not preceded by semantically related words. In the
experimental condition, the N400 component discriminated be-
tween target words from congruent and incongruent sentences in
healthy but not schizophrenic participants. Conversely, the N400
to target words in the control condition revealed N400 congruity
effects in both participant groups. This suggested that sentence
processing was impaired in schizophrenia only when it was inter-
fered with by the effects of the context-inappropriate meanings of
homographs. This finding is important because it shows that de-
ficient processing of words that activate multiple semantic repre-
More specifically, the absence of the N400 congruity effect for
the schizophrenic group in the experimental condition suggests
representations of homographs. The experimental comparison was
designed so that, if not suppressed, context-inappropriate dominant
meanings of homographs would provide a congruent context for
the target words that were otherwise incongruent with the overall
preceding sentence context. Therefore, the attenuated N400 to
these incongruent targets can account for the lack of the N400
effect due to the overall-context congruity. This interpretation was
supported by the results of a comparison of the amplitude of the
N400 component in the experimental condition between schizo-
phrenic and healthy participant groups. Although the N400 at the
congruent targets was similar between the groups, the N400 at the
incongruent targets was significantly smaller in patients than in
The present finding of an inadequate suppression of context-
irrelevant information is in line with prior research suggesting that
linguistic deficits of schizophrenic patients may be caused by their
less potent contextual suppression mechanism ~Titone et al., 2000!.
One possibility is that such weakened effects of linguistic context
may be caused by a more general impairment of inhibitory neural
processes that, in earlier studies, has been argued to underlie
abnormal performance of patients with schizophrenia on a number
of nonverbal tasks ~e.g., Flashman et al., 1996!. In addition, a lack
of suppression of context-inappropriate associations is consistent
with the hypothesis about hyperactivity in semantic memory of
schizophrenic patients. When the presentation of homographs trig-
gered retrieval of their meanings from semantic memory, all of the
meanings could be abnormally strongly activated. This would
account for the inability of the sentence context to suppress irrel-
evant dominant meanings of the homographs.
An important aspect of the present study is that it demonstrates
the effects of unsuppressed context-irrelevant semantic represen-
tations on processing of words embedded within a sentence con-
text during normal reading ~i.e., under conditions typical of natural
language processing!. Furthermore, these effects were observed
even though there was a time-lag of over 1,500 ms between the
presentation of words that activated context-irrelevant associations
and the target words that were influenced by these associations.
One potential mechanism that could mediate such prolonged prim-
ing effects is the “structure building” process proposed by Gerns-
bacher et al. ~1990!. In their formulation, language comprehension
generally involves building semantic representations with multiple
substructures, with shifting to a new substructure occurring every
time the incoming information does not cohere with the current
substructure. According to these authors, uninhibited context-
inappropriate associations may be integrated into the created rep-
resentation of a discourse. These associations, however, are
incoherent with the preceding context, and therefore usually tend
to trigger shifting to a new substructure. In the present experiment,
the uninhibited contextually irrelevant dominant meaning of homo-
graphs could become a basis for a new representational substruc-
ture, which then would be held in working memory available for
the incoming information to be mapped on it. Thus, the target
words, even though they arrived after a considerable time-lag,
could be integrated into this alternative substructure. Note that a
structure building account also explains why the context-appropriate
dominant meanings, even though they might have been hyperac-
tivated, did not exert excessive influence on the processing of
target words ~i.e., the N400s to congruent targets in the experi-
mental condition were similar between the patients and healthy
participants!. These meanings were likely integrated into the rep-
resentational structure of the first clause, and it was this structure
rather than the dominant meanings themselves that the target
words were integrated with.
It is noteworthy that whereas homographs represent one end of
the spectrum of words with multiple meanings, most content words
are polysemous ~e.g., the word “apple” may activate representa-
tions including either “gardening” or “baking” attributes of an
apple concept; Gernsbacher & Faust, 1990!. The present results
suggest that in patients with schizophrenia, all representations of a
word are active and may have effects on processing of subsequent
words independently of the discourse context.
Further studies are required to examine the relationship be-
tween the present findings obtained during language com-
T. Sitnikova et al.
prehension and positive thought disorder that usually is diagnosed
based on abnormal language production. Nevertheless, some mod-
els of normal language processing hold that semantic processes0
~e.g., Ellis & Young, 1988; MacKay, 1987!. Thus, the deficient
processing of multimeaning words observed in the present study
may also be at the root of the problem of speech production in
patients with schizophrenia. In language production, where the
words used are under the control of the speaker, such effects might
be expected to result in seemingly unpredictable changes in the
theme of an utterance.
A potential criticism of the present study could be that the
schizophrenic and healthy groups were not matched for a number
of demographic and cognitive variables. Of particular concern
could be that, because of their inferior verbal skills ~suggested by
the scores on the WAIS-R!, schizophrenic patients might have
mental representations of homographs that were of poorer quality
than was characteristic of healthy controls ~Perfetti & Hart, 1999!.
Therefore, patients might activate only dominant meanings of
homographs in all sentences, which, on its own, might disrupt
semantic integration of the clauses that biased the subordinate
meanings. Nevertheless, the results of several previous studies
demonstrating that schizophrenic patients who had comparable
verbal skills to the present sample could access subordinate mean-
ings of homographs ~e.g., Chapman et al., 1964; Titone et al.,
2000! argue against this possibility. Moreover, in the study by
Titone and colleagues, schizophrenic patients activated subordi-
nate meanings rather quickly ~these meanings were found to prime
processing of visual targets presented immediately after the offset
of the spoken homograph!. Further, in the present study, patients
with schizophrenia elicited the N400 effect in the control condition
that did not differ in size or latency from that of healthy partici-
pants. If representations of words in semantic memory of patients
were of poorer quality, then patients would be expected to take
more time to access the word meanings, and their N400 effect
should have been delayed. Finally, correlations conducted on the
overall sample of participants in the present study revealed no
relationships between the N400 in the experimental condition and
available demographic0cognitive variables. This also suggested
that the imperfect match between the study groups was unlikely to
account for the observed group differences in ERPs.
Finally, some concern about the effects of general cognitive
abnormalities on the study results may be relevant as the patient
group made significantly less accurate semantic judgments than
did healthy participants. The inferior performance of patients could
be a reflection of such deficits as attention or motivation impeding
sentence processing at a deep semantic level. However, this pos-
sibility seems unlikely given that patients produced a statistically
normal N400 effect to target words in the control condition. The
latter finding also rules out any explanation of differences between
groups in the experimental condition being due to generally smaller
N400s in patients with schizophrenia.
In summary, the most important finding of the present study is
that context-irrelevant semantic representations of the words from
the discourse inappropriately affected processing of the upcoming
words during language comprehension in schizophrenia. This re-
sult could be a consequence of inadequate contextual suppression
of such representations and0or excessive activation of such repre-
sentations as they were retrieved from semantic memory. Impor-
tantly, patients with schizophrenia appeared to use sentence context
normally when materials included no words contextual effects of
which could disrupt sentence integration.
Adams, J., Faux, S. F., Nestor, P. G., Shenton, M., Marcy, B., Smith, S., &
McCarley, R. W. ~1993!. ERP abnormalities during semantic process-
ing in schizophrenia. Schizophrenia Research, 10, 247–257.
Aloia, M. S., Gourovitch, M. L., Missar, D., Pickar, D., Weinberger, D. R.,
& Goldberg, T. E. ~1998!. Cognitive substrates of thought disorder, II:
Specifying a candidate cognitive mechanism. American Journal of
Psychiatry, 155, 1677–1684.
American Psychiatric Association. ~1994!. Diagnostic and statistical man-
ual of mental disorders ~4th ed.!. Washington, DC: Author.
Andreasen, N. C. ~1984a!. The scale for the assessment of positive symp-
toms. Iowa City, Iowa: The University of Iowa.
Andreasen, N. C. ~1984b!. The scale for the assessment of negative symp-
toms. Iowa City, Iowa: The University of Iowa.
Andrews, S., Shelley, A. M., Ward, P. B., Fox, A., Catts, S. V., & Mc-
Conaghy, N. ~1993!. Event-related potential indices of semantic pro-
cessing in schizophrenia. Biological Psychiatry, 34, 443–458.
Barch, D. M., Cohen, J. D., Servan-Schreiber, D., Steingard, S., Cohen,
J. D., Steinhauer, S. S., & van Kammen, D. P. ~1996!. Semantic priming
in schizophrenia: An examination of spreading activation using word
pronunciation and multiple SOAs. Journal of Abnormal Psychology,
Benjamin, T. B., & Watt, N. F. ~1969!. Psychopathology and semantic
interpretation of ambiguous words. Journal of Abnormal Psychology,
Blaney, P. H. ~1974!. Two studies of the language behavior of schizophren-
ics. Journal of Abnormal Psychology, 83, 23–31.
Brown, C., & Hagoort, P. ~1993!. The processing nature of the N400:
Evidence from masked priming. Journal of Cognitive Neuroscience, 5,
Chapman, L. J., Chapman, J. P., & Miller, G. A. ~1964!. A theory of verbal
behavior in schizophrenia. In B. A. Maher ~Ed.!, Progress in experi-
mental personality research ~Vol. 1, pp. 49–77!. New York: Academic
Cohen, J. D., Barch, D. M., Carter, C., & Servan-Schreiber, D. ~1999!.
Context-processing deficits in schizophrenia: Converging evidence from
three theoretically motivated cognitive tasks. Journal of Abnormal
Psychology, 108, 120–133.
Cohen, J. D., Servan-Schreiber, D. ~1992!. Context, cortex, and dopamine:
A connectionist approach to behavior and biology in schizophrenia.
Psychological Review, 99, 45–77.
Condray, R., Steinhauer, S. R., Cohen, J. D., van Kammen, D. P., &
Kasparek,A. ~1999!. Modulation of language processing in schizophre-
nia: Effects of context and haloperidol on the event-related potential.
Biological Psychiatry, 45, 1336–1355.
Ellis, A. W., & Young A. W. ~1988!. Human cognitive neuropsychology.
Hove, UK; Hillsdale, USA: Erlbaum.
Flashman, L. A., Flaum, M., Gupta, S., & Andreasen, N. C. ~1996!. Soft
signs and neuropsychological performance in schizophrenia. American
Journal of Psychiatry, 153, 526–532.
Folstein, M. F., Folstein, S. E., & McHugh, P. R. ~1975!. Mini-mental state:
A practical method for grading the state of patients for the clinician.
Journal of Psychiatric Research, 12, 189–198.
Geis, M. F., & Winograd, E. ~1974!. Norms of semantic encoding vari-
ability for fifty homographs. Bulletin of the Psychonomic Society, 3,
Geisser, S., & Greenhouse, S. ~1959!. On methods in the analysis of profile
data. Psychometrica, 24, 95–112.
Gernsbacher, M. A., & Faust, M. ~1990!. The role of suppression in
sentence comprehension. In G. B. Simpson ~Ed.!, Understanding word
and sentence ~pp. 97–128!. Amsterdam: North Holland.
Gernsbacher, M. A., Varner, K. R., & Faust, M. E. ~1990!. Investigating
differences in general comprehension skill. Journal of Experimental
Psychology: Learning, Memory, & Cognition, 16, 430–445.
Goldberg, T. E., Aloia, M. S., Gourovitch, M. L., Missar, D., Pickar, D., &
Weinberger, D. R. ~1998!. Cognitive substrates of thought disorder, I:
The semantic system. American Journal of Psychiatry, 155, 1671–1676.
Language in schizophrenia and ERPs
Gorfein, D. S., Viviani, J. M., & Leddo, J. ~1982!. Norms as a tool for the Download full-text
study of homography. Memory & Cognition, 10, 503–509.
Grillon, C., Ameli, R., & Glazer, W. M. ~1991!. N400 and semantic
categorization in schizophrenia. Biological Psychiatry, 29, 467–480.
Hogaboam, T. W., & Perfetti, C.A. ~1975!. Lexical ambiguity and sentence
comprehension. Journal of Verbal Learning & Verbal Behavior, 14,
Holcomb, P. J. ~1993!. Semantic priming and stimulus degradation: Impli-
cations for the role of the N400 in language processing. Psychophysi-
ology, 30, 47–61.
Hollingshead, A. B. ~1965!. Two-factor index of social position. New
Haven, CT: Yale Station.
Kausler, D. H., & Kollasch, S. F. ~1970!. Word associations to homographs.
Journal of Verbal Learning & Verbal Behavior, 9, 444–449.
Kerns, J. G., Berenbaum, H., Barch, D. M., Banich, M. T., & Stolar, N.
~1999!. Word production in schizophrenia and its relationship to pos-
itive symptoms. Psychiatry Research, 87, 29–37.
Kounios, J., & Holcomb, P. J. ~1992!. Structure and process in semantic
memory: Evidence from event-related brain potentials and reaction
times. Journal of Experimental Psychology: General, 121, 459–479.
Koyama, S., Hokama, H., Miyatani, M., Ogura, C., Nageishi,Y., & Shimoko-
chi, M. ~1994!. ERPs in schizophrenic patients during word recognition
task and reaction times. Electroencephalography and Clinical Neuro-
physiology, 92, 546–554.
Kuperberg, G. R., McGuire, P. K., & David, A. S. ~1998!. Reduced
sensitivity to linguistic context in schizophrenic thought disorder: Evi-
dence from on-line monitoring for words in linguistically anomalous
sentences. Abnormal Psychology, 107, 423–434.
Kutas, M., & Hillyard, S. A. ~1980!. Event-related brain potentials to
semantically inappropriate and surprisingly large words. Biological
Psychiatry, 11, 99–116.
Kwapil, T. R., Hegley, D. C., Chapman, L. J., & Chapman, J. P. ~1990!.
Facilitation of word recognition by semantic priming in schizophrenia.
Journal of Abnormal Psychology, 99, 215–221.
MacKay, D. G. ~1987!. Asymmetries in the relationship between speech
perception and production. In H. Heuer & A. F. Sanders ~Eds.!, Per-
spectives on perception and action ~pp. 301–333!. Hillsdale, NJ: Law-
rence Erlbaum Associates.
Maher, B.A. ~1983!.Atentative theory of schizophrenic utterance. In B.A.
Maher & W. B. Maher ~Eds.!, Progress in Experimental Personality
Research ~Vol. 11, pp. 1–51!. San Diego, CA: Academic Press.
Maher, B. A., Manschreck, T. C., & Rucklos, M. E. ~1980!. Contextual
constraint and the recall of verbal material in schizophrenia: The effect
of thought disorder. British Journal of Psychiatry, 137, 69–73.
Manschreck, T. C., Maher, B.A., Milavetz, J. J.,Ames, D., Weisstein, C. C.,
& Schneyer, M. L. ~1988!. Semantic priming in thought disordered
schizophrenic patients. Schizophrenia Research, 1, 61–66.
Nelson, D. L., McEvoy, C. L., Walling, J. R., & Wheeler, J. W. ~1980!. The
university of South Florida homograph norms. Behavior Research
Methods & Instrumentation, 12, 16–37.
Nestor, P. G., Akdag, S. J., O’Donnell, B. F., Niznikiewicz, M., Law, S.,
Shenton, M. E., & McCarley, R. W. ~1998!. Word recall in schizophre-
nia: A connectionist model. American Journal of Psychiatry, 155,
Nestor, P. G., Kimble, M. O., O’Donnell, B. F., Smith, L., Niznikiewicz,
M., Shenton, M. E., & McCarley, R. W. ~1997!. Aberrant semantic
activation in schizophrenia:Aneurophysiological study. American Jour-
nal of Psychiatry, 154, 640–646.
Niznikiewicz, M. A., O’Donnell, B. F., Nestor, P. G., Smith, L., Law, S.,
Karapelou, M., Shenton, M. E., & McCarley, R. W. ~1997!. ERP
assessment of visual and auditory language processing in schizophre-
nia. Journal of Abnormal Psychology, 106, 85–94.
Ohta, K., Uchiyama, M., Matsushima, E., & Toru, M. ~1999!. An event-
related potential study in schizophrenia using Japanese sentences. Schizo-
phrenia Research, 40, 159–170.
Oldfield, R. C. ~1971!. The assessment and analysis of handedness: The
Edinburg inventory. Neuropsychologia, 9, 97–113.
Olichney, J. M., Iragui, V. J., Kutas, M., Nowacki, R., & Jeste, D. V. ~1997!.
N400 abnormalities in late life schizophrenia and related psychoses.
Biological Psychiatry, 42, 13–23.
Overall, J. E., & Gorham, D. R. ~1962!. The brief psychiatric rating scale.
Psychological Reports, 10, 799–812.
Perfetti, C. A., & Goodman, D. ~1970!. Semantic constraint on the decod-
ing of ambiguous words. Journal of Experimental Psychology, 86,
Perfetti, C. A., & Hart, L. A. ~1999!. Quality lexical representations ~not
suppression! are central to reading skill. Poster presented at the 40th
Annual Meeting of the Psychonomic Society, Los Angeles, CA, USA.
Salisbury, D. F., O’Donnell, B. F., McCarley, R. W., Nestor, P. G., &
Shenton, M. E. ~2000!. Event-related potentials elicited during a context-
free homograph task in normal versus schizophrenic subjects. Psycho-
physiology, 37, 456–463.
Semlitsch, H. V., Anderer, P., Schuster, P., & Presslich, O. ~1986!. A
solution for reliable and valid reduction of ocular artifacts, applied to
the P300 ERP. Psychophysiology, 23, 695–703.
Simpson, G. B., & Burgess, C. ~1985!. Activation and selection processes
in the recognition of ambiguous words. Journal of Experimental Psy-
chology: Human Perception & Performance, 11, 28–39.
Speed, M., Toner, B. B., Shugar, G., & Di Gasbarro, I. ~1991!. Thought
disorder and verbal recall in acutely psychotic patients. Journal of
Clinical Psychology, 47, 735–744.
Spitzer, M. ~1997!.Acognitive neuroscience view of schizophrenic thought
disorder. Schizophrenia Bulletin, 23, 29–50.
Spitzer, M., Braun, U., Hermle, L., & Maier, S. ~1993!. Associative se-
mantic network dysfunction in thought-disordered schizophrenic pa-
tients: Direct evidence from indirect semantic priming. Biological
Psychiatry, 34, 864–877.
Spitzer, R., Williams, J., Gibbon, M., & First, M. ~1990a!. The structured
clinical interview for DSM-IIIR (SCID). Washington, DC: American
Spitzer, R., Williams, J., Gibbon, M., & First, M. ~1990b!. The structured
clinical interview for DSM-IIIR—Non-patient edition (SCID-NP).Wash-
ington, DC: American Psychiatric Association.
Strandburg, R. J., Marsh, J. T., Brown, W. S., Asarnow, R. F., Guthrie, D.,
Harper, R., Yee, C. M., & Nuechterlein, K. H. ~1997!. Event-related
potential correlates of linguistic information processing in schizophren-
ics. Biological Psychiatry, 42, 596–608.
Strauss, M. E. ~1975!. Strong meaning-response bias in schizophrenia.
Journal of Abnormal Psychology, 84, 295–298.
Titone, D., Levy, D. L., Holzman, P. S. ~2000!. Contextual insensitivity in
schizophrenic language processing: Evidence from lexical ambiguity.
Journal of Abnormal Psychology, 109, 761–767.
Twilley, L. C., Dixon, P., Taylor, D., & Clark, K. ~1994!. University of
Alberta norms of relative meaning frequency for 566 homographs.
Memory & Cognition, 22, 111–126.
Vinogradov, S., Ober, B.A., & Shenaut, G. K. ~1992!. Semantic priming of
word pronunciation and lexical decision in schizophrenia. Schizophre-
nia Research, 8, 171–181.
Wechsler, D. ~1981!. Manual for the Wechsler adult intelligence scale—
Revised. New York: Psychological Corporation.
~Received April 1, 2000; Accepted June 21, 2002!
T. Sitnikova et al.