The Language System in Schizophrenia: Effects
of Capacity and Linguistic Structure
by Ruth Condray, Stuart R. Steinhauer, Daniel P . van Kammen,
and Annette Kasparek
Abstract
Dysfunction in receptive language processes has been
reliably observed in individuals diagnosed with schizophrenia and their first degree family members. The present study addressed the unresolved issue of whether
receptive syntax is intact in schizophrenia. The principal
question concerned whether comprehension dysfunction
in schizophrenia involves a disturbance in the processing
of syntactic structure, a susceptibility to demands placed
on temporal auditory processing, or some combination
of these two general factors. Comprehension accuracy
was compared between 32 males diagnosed with schizophrenia and 22 males with no lifetime diagnosis of psychiatric disorder. Accuracy was examined for responses
to Who questions ("Who did X?" and "Who was done
X?") about information in the sentential clauses (main
vs. relative) of two types of relative sentences (subjectrelatives vs. object-relatives) that were presented aurally
at conversational and accelerated rates. The relationship
between cognitive functions and comprehension accuracy was also tested. Results showed highly significant
effects of diagnosis, syntactic structure, and temporal
demand. Patients were characterized by reduced overall
comprehension accuracy compared to controls. More
important, patients and controls differed in their patterns of accuracy across the different types of syntactic
structure. Finally, cognitive functions predicted but did
not completely account for comprehension accuracy.
Findings suggest the hypothesis that receptive syntax is
disrupted in schizophrenia, and this dysfunction may
not be entirely explained by compromised general cognitive ability.
Keywords: Schizophrenia, receptive syntax, relativization, word order.
Schizophrenia Bulletin, 28(3):475-490, 2002.
(Bleuler 1911; Kraepelin 1919; American Psychiatric
Association 2000) of schizophrenia. Dysfunction in both
receptive language and speech production has been
described for schizophrenia patients and their biological
relatives. Although speech production abnormalities are
more obvious as clinical signs and have been more commonly targeted for empirical study (for recent reviews, see
Docherty et al. 1999; Docherty et al. 2000), dysfunction in
receptive language has also been consistently observed for
patients and their first degree family members, including
disturbances in the perception of words presented aurally
at various levels of intensity (Bull and Venables 1974) and
background noise (DeLisi et al. 1997; Shedlack et al.
1997); and reduced comprehension accuracy for information in sentences (Thomas and Huff 1971; Faber and
Reichstein 1981; Morice and McNicol 1985; Condray et
al. 1992, 1995, 1996; Landre et al. 1992; Goldberg et al.
1998). The collective findings to date are therefore consistent with the hypothesis that integrity of the language system may reflect biological risk for the disorder. However,
detailed analyses of complex behavioral phenotypes are
required to provide a solid empirical basis for selecting
variables for genetic studies of schizophrenia (for discussion about the importance of this issue for behavior genetics in general, see Pennington and Smith 1997). If receptive language disturbance is to serve as an informative
phenotype in behavior genetics studies of schizophrenia,
detailed analyses of its underlying functional components
are essential.
An important unresolved question is whether receptive syntax is intact in schizophrenia. Attempts to address
this issue have involved a variety of strategies, including
comparisons of schizophrenia patients and control groups
on aphasia and neuropsychology test batteries, examinations of patients' sensitivity to syntactic structure, and
Send reprint requests to Dr. R. Condray, Western Psychiatric Institute
and Clinic, Department of Psychiatry, University of Pittsburgh School of
Medicine, 3811 O'Hara Street, Pittsburgh, PA 15213; e-mail: [email protected].
The language system has been assigned a central role in
the etiology (Crow 1997) and clinical manifestation
475
Schizophrenia Bulletin, Vol. 28, No. 3, 2002
R. Condray et al.
evaluations of their accuracy in responding to complex
syntactic constructions. The findings from these empirical
efforts are not conclusive, although clear trends are present. Schizophrenia patients appear sensitive to syntactic
boundaries in sentences (embedded click paradigm developed by Fodor and Bever 1965: Rochester et al. 1973;
Carpenter 1976; Grove and Andreasen 1985). In contrast,
reduced accuracy has been documented for patients across
a range of tasks involving complex syntactic constructions: carrying out instructions described in syntactically
complex sentences (revised Token Test: Morice and
McNicol 1985), ordering direct and indirect objects within
sentences (Rausch et al. 1980), and understanding logical
relationships specified in relative-clause sentences (Condray et al. 1992, 1996). One linguistic feature common to
these latter studies is a transformational operation, movement of phrase constituents (Chomsky 1986; for a discussion regarding the relevance of movement to receptive
syntax in Broca's aphasia, see Grodzinsky 2000). The
appearance in schizophrenia patients of intact sensitivity
to syntactic boundaries but disrupted syntactic movement
requires empirical refinement.
An additional unresolved question concerns the role
of extralinguistic capacities in schizophrenia patients' syntactic processing (for discussions relevant to language disorders in schizophrenia in general, see Schwartz 1982;
Morice 1986; Frith and Allen 1988; Maher 1991). In particular, results from several lines of investigation are suggestive about the importance of memory capacity in
patients' receptive language function. Knight and SimsKnight (1979) reported an association between severity of
schizophrenia illness and the ability to organize and integrate in memory the ideas described in serial sentences.
Grove and Andreasen (1985) observed that the performance of patients and controls differed significantly for a
short-term memory task, but that those same individuals
did not differ for sensitivity to syntax boundaries in sentences or memory for gist in serial sentences. The results
of Grove and Andreasen are consistent with the patterns
reported earlier by Carpenter (1976) and Rochester et al.
(1973). Moreover, memory span for sentence-length materials has been found to correlate with overall comprehension accuracy in patients and controls (Condray et al.
1996). It has not been established whether patients' memory capacity is associated with their processing of linguistic structures requiring transformational operations.
Theoretical Framework for the Present
Study
The present study was designed to address the issue of
whether receptive syntax is intact in schizophrenia. The
476
methodology involved examination of the influence of
syntactic structure and psychological function on comprehension accuracy in individuals with schizophrenia and
normal controls. The research was guided by a general
capacity hypothesis that is broadly consistent with the
capacity theory of Just and Carpenter and their colleagues
(Just and Carpenter 1992; Miyake et al. 1994; Haarmann
et al. 1997) as well as earlier formulations in which the
interaction of syntactic parsing and semantic integration
was emphasized (Perfetti and Lesgold 1977; Bock 1982;
Kintsch 1988). The primary assumption was that dysfunction of receptive syntax in schizophrenia is due, in part, to
capacity or activation disturbances that compromise the
storage and computation functions required to create representations of logical relationships that are specified by
the syntax of linguistic constructions. In the following discussion, a limited review is provided for the two classes of
variables of interest.
Linguistic Structure. The structure of language influences understanding. Comprehension processes in nonclinical populations are affected by syntactic complexity,
with systematic declines in comprehension accuracy
accompanying increases in complexity (for reviews, see
Just and Carpenter 1992; MacDonald et al. 1992). Insight
into receptive syntax in schizophrenia can therefore be
gained through examinations of the effects of syntactic
structure on understanding accuracy in this patient population, an approach that has led to increased understanding about receptive syntax in Broca's aphasia (for review
and commentary, see Grodzinsky 2000). Structural factors
believed to influence comprehension include word order,
transitivity, and relativization.
Word order. Word order specifies the logical relationships that hold for a linguistic expression. Although
determining the dominant or basic typology for a given
language is not a straightforward issue (see Comrie 1981),
English is typically considered a fixed-order language
(Quirk et al. 1985), with the SVO (subject verb object)
order representing the most consistent scheme (Siewierska
1999). Most of the worlds' languages (95%) employ an
SO ordering (subjects preceding objects) (for a review, see
Siewierska 1999), and it has been suggested that processing demand may influence the principle(s) that govern this
ordering in some nonrandom manner (Bock 1982). One
implication of this assumption is that an OS ordering
(objects preceding subjects) increases the demand on processing resources.
Transitivity. Transitive clauses communicate ideas
about "who did what to whom." Traditional definitions of
transitivity include the presence of a direct object and a
verb denoting that an action has been transferred from a
subject to an object. More recently, the interface of seman-
Effects of Capacity and Linguistic Structure
Schizophrenia Bulletin, Vol. 28, No. 3, 2002
tic and syntactic knowledge has been emphasized, with a
distinction made between semantic and syntactic aspects
and an emphasis placed on transitivity as a matter of
degree (for a discussion, see Hopper and Thompson 1980).
According to this perspective, the semantic aspects of
transitive clauses include the following: two participants,
at a minimum, who fulfill the semantic roles of subject and
object; a subject who carries out a completed action in a
volitional and controlling manner; and an object who is
changed by that action. The degree to which the completed
action creates a change in the object produces the continuum of transitivity.
Relativization. Most languages include relative
clauses (Fabb 1999; for an exception, see Comrie 1981),
and these constructions have been used in theoretical considerations of transformational operations (e.g., movement: Chomsky 1986). Relative clauses commonly modify nouns and are typically introduced with wh- pronouns
(e.g., who, whom, which) and that (for discussions regarding the status of that as a relative pronoun, see Sag 1997;
Van Der Auwera 1985; and for etymology, see Fabb 1999).
In one method of relative clause formation, the noun-head
to be modified can differ with respect to grammatical function (subject vs. object) and location (before, after, or
within the clause) (Keenan and Comrie 1977). Relative
clauses that are formed in this manner involve a modification of either the clause subject (subject-relative: The
detective that shot the murderer . . .) or clause object
(object-relative: The burglar that the fireman rescued...).
tion that human beings can comprehend accurately per
unit of time. Comprehension accuracy declines systematically as presentation rates increase above the conversational rate of 250 to 300 words per minute (Fairbanks et al.
1957; Carver 1982; for a review of the early literature, see
Foulke and Sticht 1969; Wingfield 1975; Hausfeld 1981).
Training with accelerated speech does not appear to eliminate these comprehension decrements entirely (Wallace
1983). Deficits in temporal auditory processing may be
integral to certain developmental learning disorders, such
as dyslexia (for discussions, see Llinas 1993; Tallal et al.
1993). Moreover, schizophrenia is associated with disturbances in the processing of sequential, rapidly presented
auditory stimuli (suppression of sensory gating: Freedman
et al. 1987; Braff and Geyer 1990). Though the rates that
are relevant for natural language and the sensory gating
phenomenon obviously differ, it is possible that some type
of temporal sequencing disturbance may also influence
receptive language in schizophrenia.
Memory capacity. Individuals differ in the efficacy
of their comprehension processes (Perfetti and Lesgold
1977; Daneman and Carpenter 1980; King and Just 1991;
Just and Carpenter 1992). A perspective that has proven
useful in accounting for this fact is the capacity theory, as
developed and refined by PA. Carpenter and M.A. Just
and their colleagues. In a recent formulation, capacity represents a composite of activation that is finite in extent,
and mediates the storage and computations necessary for
understanding linguistic constructions (see Miyake et al.
1994; Haarmann et al. 1997). A distinction was also made
between the capacity construct and traditional working
memory and short-term storage concepts (e.g., Baddeley
1986), with capacity viewed more generally as a resource
that enables storage and computations of sentence representations at multiple levels (lexical, syntactic, semantic,
thematic, referential). The general assumption underlying
the present study design is broadly consistent with this
view. Supportive evidence includes findings that the
capacity construct, as denned by reading span, predicted
comprehension performance for college students (Daneman and Carpenter 1983; Just and Carpenter 1992; MacDonald et al. 1992) and patients with schizophrenia (Condray et al. 1996). Moreover, central nervous system
function appears modulated by syntactic complexity, with
manipulations of syntactic structure producing systematic
changes in the amplitude of event-related potentials (King
and Kutas 1995) and the pattern of neural activation in
functional magnetic resonance imaging (fMRI) (Just et al.
1996).
Psychological Functions. An assumption common to the
various theories of language comprehension is that meaning is formed through multiple layers of linguistic structures that are mutually constraining. According to this
general framework, comprehension accuracy depends on
some combination of the processing efficacy that occurs
at the micro and macro levels: minimally, in the decoding
of perceptual input (prerecognition lexical access models:
McClelland and Elman 1986; Marslen-Wilson 1987;
Marslen-Wilson et al. 1996; Wallace et al. 1998) and in
the integration of lexical items to create a grammatical
representation (postrecognition computational models:
Kintsch 1988; Just and Carpenter 1992; Haarmann et al.
1997). In the comprehension of speech, this process must
also be accomplished for lexical items that occur in a temporal sequence (Marslen-Wilson and Tyler 1980;
Marslen-Wilson et al. 1996).
Two psychological functions have consistently
emerged as important to comprehension processes in nonclinical populations: temporal auditory processing and
memory capacity.
Temporal auditory processing. There does appear
to be an upper limit on the amount of linguistic informa-
Present Study. In an initial report (Condray et al. 1996),
comprehension accuracy was compared between schizophrenia patients and normal controls as a function of syn-
477
Schizophrenia Bulletin, Vol. 28, No. 3, 2002
R. Condray et al.
tactic complexity and presentation rate. The focus of that
initial report was on the global sentential level: simple
versus subject-relative versus object-relative sentences.
The present report represents a refinement and extension
of that earlier work and includes additional participants.
In the present report, patients' receptive syntax was examined through analyses of their understanding about "Who
did what to whom" in clauses with variable word orders,
which presumably influence transformational operations.
The association was also tested between psychological
functions, including memory capacity, and comprehension
accuracy about information in sentential clauses requiring
transformational operations.
Methods
Participants. Study participants included 32 males diagnosed with DSM-HI-R schizophrenia (APA 1987), and 22
males diagnosed as no lifetime DSM-III—R Axis I or Axis
II disorder (APA 1987). All study participants were evaluated with the Structured Clinical Interview for
DSM-III-R (Spitzer et al. 1989). Normal control subjects
were additionally evaluated with the Personality Disorder
Examination (Loranger 1988). Psychiatric diagnoses were
assigned during case conferences. General exclusion criteria were the following: first language other than
American English, history of major medical or neurological disorders, Wechsler Adult Intelligence Scale-Revised
(WAIS-R, Wechsler 1981) Verbal IQ < 85, and reading
level < eighth grade as measured by the Wide Range
Achievement Test (WRAT). Schizophrenia patients were
excluded if they met criteria for current substance use disorder. Audiometric screening was conducted (both ears)
using 500, 1,000, 2,000, and 4,000 Hz at 40 dBA SL. All
individuals passed the screening except for one individual
who was able to detect further screening tones of 800 and
1,500 Hz at 65 dBA SL.
Following explanation of procedures and prior to testing, all subjects provided written, informed consent to participate. Control subjects and outpatients were reimbursed
for study participation.
Evaluation of Sentence Processing. Sentence processing was evaluated using tests of sentence span memory,
intelligibility of accelerated presentation rate, and sentence comprehension.
Sentence span memory. Memory span for sentencelength materials was evaluated with a modified aura! version of the Sentence Span Test developed by Leahy (1987)
for use with clinical and general population individuals.
The Sentence Span Test was based on the Reading Span
task developed earlier by Daneman and Carpenter (1980)
for use with college students. The Sentence Span Test
478
measures memory for the final words in serial sentences
that are presented in sets of sentences that gradually
increase in size: 2 sets of 2 sentences each; 3 sets of 3 sentences each; 3 sets of 4 sentences each; 3 sets of 5 sentences each; 3 sets of 6 sentences each; 3 sets of 7 sentences each; and 1 set of 8 sentences. Sentence Span is the
largest set size for which all final words are recalled correctly. Subjects were advised that they would hear a series
of sentences after which they would be asked to recall the
final word of each sentence in each set. They were also
advised to expect that the number of sentences would
increase as the test progressed.
Intelligibility of accelerated presentation rate. An
estimate of intelligibility for the accelerated presentation
rate was obtained from the responses recorded to a separate set of 16 warm-up sentences presented at the accelerated rate (proportion of words repeated correctly). As
described below (in Procedure section), the task was to
repeat each warm-up sentence verbatim. Warm-up sentences were simple sentence constructions used in previous work on the intelligibility of accelerated speech (Wallace and Koury 1981).
Sentence comprehension task. Two types of sentence processing demand were varied: syntactic and temporal. Syntactic demand was varied by contrasting sentence constructions that differed in grammatical structure,
while holding the item load constant (i.e., all sentences
contained an equal number of words). Temporal demand
was varied using a conversational and an accelerated rate.
Sentences and questions were presented aurally.
Syntactic structure. The focus of the present report
concerns accuracy of understanding about the roles of
actor versus object as they were specified by the syntax
of the main and embedded relative clauses in subjectrelative and object-relative sentences (i.e., "Who did
what to whom?"). Three types of sentence constructions
were presented to all study participants: 16 object-relative sentences, 16 subject-relative sentences, and 16
simple declarative sentences. All sentences were 9
words in length. All relative sentences contained two
past tense verbs, the majority (n = 53/64 = 83%) of
which were classified as transitive in their first or dominant meaning (Webster's 1979). Word order differs
between the two types of relative sentences: the basic
word order in English (SVO) is preserved in subject-relative sentences, while the uncommon word order involving objects preceding subjects (object subject verb
[OSV]) occurs in object-relative sentences. The 48 sentences included items used by King and Just (1991, p.
600) and items that were developed for the present
study. Examples of the relative sentences are
Object-relative: The reporter that the senator attacked
admitted the error.
Effects of Capacity and Linguistic Structure
Schizophrenia Bulletin, Vol. 28, No. 3, 2002
Subject-relative: The accountant that sued the lawyer
read the paper.
Simple declarative sentences were SVO constructions
that served as control items in earlier comparisons of
global sentence processing (Condray et al. 1996); for
example,
Simple declarative: The curator purchased the antique
book for the museum.
language stimuli through AKG Acoustics K340 stereo
headphones. Prior to the test sentences, a separate set of 16
warm-up sentences was presented at the accelerated rate.
The task was to repeat each warm-up sentence verbatim.
The purposes of this warm-up set were to accommodate
subjects to the rapid rate, and to obtain an estimate of
intelligibility for this presentation rate. Previous work
showed that college students accommodated to frequencycontrolled speech after similar exposure (Wallace and
Koury 1981). The 48 test sentences and comprehension
questions immediately followed the warm-up sentences.
Each subject was presented with all 48 test sentences.
Each of the three sentence types was presented under each
of the two presentation rates, with 8 sentences in each of
the 6 combinations of sentence type and presentation rate.
Sentences were assigned randomly to the two presentation
rate conditions. The 6 sentence X rate combinations were
distributed evenly across the testing session in 8 blocks
that consisted of 6 sentences per block. Each block thus
included one presentation of each sentence type X presentation rate condition. Assignment of sentences to blocks
and presentation order of conditions within each block
were also randomized. Comprehension questions were
presented at the conversational rate, and responses were
recorded by a research technician.
Sentence comprehension questions. Relative sentences: Three Who questions immediately followed each relative sentence to assess understanding of the facts and logical relationships that were specified by the syntax. For
example, recognizing that the initial noun plays dual roles
(actor and object) is necessary for an accurate understanding
of object-relative sentences. Thus, in the sentence "The
reporter that the senator attacked admitted the error," it is
necessary to understand that the reporter serves as the actor
in the main clause ("The reporter . . . admitted the error"),
and as the object of the action in the relative clause (". . .
(that) the senator attacked"). The Who questions used to
assess understanding of these relations were (1) "Who
admitted the error?"—that is, Who did X? in the main
clause; (2) "Who was attacked?"—or, Who was done X? in
the relative clause; and (3) "Who did the attacking?"—or,
Who did X? in the relative clause. Scoring was straightforward, with a total possible of 96 correct: 32 sentences (16
object-relative and 16 subject-relative) X 3 questions per
sentence. Simple declarative sentences: Three Wh- questions
(Who? What? Where? and/or Whose?) followed each simple
declarative sentence to assess understanding of the facts and
relationships described in the sentence. Scoring was straightforward, with a total possible of 48 correct: 16 sentences X 3
questions per sentence.
Presentation rate. Sentences were presented aurally
at two rates: conversational and accelerated. All sentences
and questions were recorded by a female speaker (native
of western Pennsylvania) using her typical conversational
rate. The accelerated presentation rate was accomplished
using a frequency-controlled technique. Frequency-controlled techniques of accelerating speech records have
been demonstrated to produce superior intelligibility when
compared with frequency-shifted techniques (Wallace and
Koury 1981). In the present study, the frequency-controlled technique increased the playback speed to approximately twice that of the conversational rate without producing a linear increase in the frequency or pitch of the
speech record. A Media Vision sound board was used to
produce this accelerated rate. The sound board recorded
16-bit samples at 44.1 kHz. The software application randomly sampled the original speech record digitally and
deleted half of the samples (50% compression).
Procedure. A PC-based computer was used to present
the sentence and question stimuli. Subjects listened to the
Data Analyses
Primary Measures. The measures of primary interest
included response accuracy in the Sentence Span Test, the
Intelligibility task, and the Sentence Comprehension
task—Relative Sentences. Response accuracy (number
correct) was converted to the proportion correct and then
transformed by the arcsine transformation prior to conducting the statistical tests (Fleiss 1986). Data values that
are presented in tables and graphs are proportions (exception: Sentence Span data are presented as Span Set Size to
facilitate interpretation). However, all statistical analyses
are based on the arcsine transformation of the proportion
of correct responses. This strategy was used to adjust for
distributions associated with high accuracy rates that were
produced by some of the tasks and experimental conditions, and to provide more conservative tests of the data.
The experimental design for the Sentence Comprehension
task was a mixed factorial that included both betweenand within-subjects factors. Results were tested using
analyses of variance (BMDP2V ANOVAs), with diagnosis (two levels) as the between-subjects factor, and sentence complexity (two levels), presentation rate (two levels), and within-sentence clause (three levels) as the
within-subjects factors. The Greenhouse-Geisser correction was applied to analyses of the within-sentence clause
variable, and the epsilon factors for these adjustments are
479
R. Condray et al.
Schizophrenia Bulletin, Vol. 28, No. 3, 2002
reported. The standard significance level of a = 0.05 was
used for the omnibus ANOVAs. Theoretically motivated
analytical comparisons followed findings of significant
results in the omnibus ANOVAs (see Keppel 1991). These
analytical comparisons were evaluated at the more stringent significance level of a = 0.01 to reduce the risk of
type I errors that may accompany such comparisons
(Keppel 1991). Trend significance for analytical comparisons was set at a = 0.05.
dard errors of controls' observed relative clause scores;
(3) each individual's standard residualized scores (embedded—object and embedded—actor) were then entered
into an ANOVA, with diagnosis (two levels) as the
between-subjects factor and relative clause condition (two
levels) as the within-subjects factor.
Results
Characteristics of Sample. Table 1 presents the characteristics of the sample. Patients and controls did not differ
for age, education, or single-word reading ability
(WRAT). General verbal ability was assessed using the
subtests from the WAIS-R Verbal IQ. Compared to controls, patients were characterized by an uneven pattern of
performance on these measures of verbal ability, with significant group differences observed only for the
Arithmetic and Comprehension subtests. (Data were missing for one control on the Vocabulary subtest and one
patient on the Similarities subtest, which is reflected in the
degrees of freedom.) Handedness (Kutas and Van Petten
1994) was assessed using items from the Edinburgh
Handedness Inventory (Oldfield 1971), and patients and
controls showed a similar proportion of individuals with a
right-hand preference.
Tests of Association. The association between language
comprehension and other psychological functions was
examined using linear regression analysis with simultaneous entry of variables (SPSS 10.0). Unstandardized betas
are reported to facilitate comparisons across samples
(Pedhazur 1982). The association between language comprehension and general intellectual ability (WAIS-R) was
examined using standard tests of correlation (Pearson r).
Standard Residualized Scores. To remove extraneous
variance measured by the conditions of theoretical interest
(relative clauses), standard residualized scores were computed in the manner recommended by Chapman and
Chapman (1989, p. 363) and entered as the measure of
individual differences in an additional ANOVA: (1) for
each individual, the predicted scores on the relative clause
conditions (embedded—object and embedded—actor)
were computed based on their observed main clause
score, using the regression equations for controls; (2) each
residualized score was then standardized using the stan-
Clinical Characteristics of Patients. This sample of 32
schizophrenia patients included 17 inpatients and 15 outpatients who were receiving treatment in the Veterans
Affairs Pittsburgh Healthcare System at the time of test-
Table 1. Characteristics of the sample
Age, mean (SD)
Education (yrs), mean (SD)
Single-word reading ability (WRAT),
mean (SD)
Verbal ability (WAIS-R), mean (SD)
Information
Digit span
Vocabulary
Arithmetic
Comprehension
Similarities
Handedness, f (%)
Right
Left
Both
Schizophrenia patients
(n = 32)
Normal controls
(/7=22)
ftest
df
42.4 (7.9)
12.8(1.9)
76.7 (7.4)
38.9 (8.4)
13.6(1.2)
73.0 (9.2)
1.55
1.85
1.65
52
52
52
0.13
0.07
0.11
10.6(2.1)
9.6 (2.4)
9.9 (1.9)
8.3 (2.4)
8.2(2.1)
9.2 (2.8)
11.0(1.9)
10.7(3.0)
10.8(1.9)
10.7(3.0)
10.6 (2.3)
10.2(2.4)
<1
1.49
1.62
3.16
3.97
1.30
52
52
51
52
52
51
0.42
0.14
0.11
0.003
0.0002
0.20
29 (90.6)
2 (6.3)
1 (3.1)
20 (90.9)
1 (4.5)
1 (4.5)
P
Note.—SD = standard deviation; WAIS-R = Wechsler Adult Intelligence Scale-Revised; WRAT = Wide Range Achievement
Test.
480
Effects of Capacity and Linguistic Structure
Schizophrenia Bulletin, Vol. 28, No. 3, 2002
ing. These patients were assigned the following
DSM-III-R schizophrenia disorder subtypes: undifferentiated (n = 12), paranoid (n = 6), and residual (/? = 11).
Three patients were assigned the diagnosis of schizoaffective disorder. Their clinical course of illness was characterized by a mean age of 24.6 years (standard deviation
[SD] = 4.7; range: 15-34) at the time of first psychiatric
hospitalization, an average illness duration of 18.2 years
(SD = 6.7; range: 2-34), and a mean of 9.8 (SD = 6.9;
range: 3-28) psychiatric hospitalizations.
Medication status of patients. Of the 32 patients, 18
were receiving haloperidol and 14 were receiving other
antipsychotic medications (chlorpromazine, fluphenazine,
loxapine, perphenazine, risperidone, thioridazine, thiothixene). Of these 32 patients, 29 were receiving oral antipsychotic medication at the time of study participation, and
their mean daily dose in chlorpromazine equivalents
(Appleton 1988) was 504.02 mg/day (n = 29; median =
400; SD = 298.57; range: 50-1127.8); 1 patient was
receiving oral risperidone (8 mg/day); and 2 patients
received injections of haloperidol decanoate (100
mg/month and 125 mg/month, respectively) during the
month prior to study participation. Of the 32 patients, 20
were additionally receiving adjunct anticholinergic medication (benztropine, mg/day: n = 20; mean = 2.75; median
= 2; SD = 1.29; range: 1-6). (Seven of these 20 patients
receiving anticholinergic medication were participating in
a double-blind haloperidol maintenance and placebo
replacement protocol, in which adjunct anticholinergic
medication was discontinued 2 weeks prior to testing [see
van Kammen et al. 1996]). Finally, of the 32 patients, 13
were receiving only antipsychotic and adjunct anticholinergic medications; the pharmacology treatment regimens
of the remaining 19 patients involved a variety of additional medications, including albuterol (n = 2); amitriptyline (n = 1); clonazepam (n = 2); desipramine (n = 2);
diphenhydramine (n = 2); fluoxetine (n = 2); imipramine
(n = 1); lithium carbonate (n = 6); lorazepam (n = 2); nizatidine (n = 1); sertraline (n = 6); theophylline (n = 1); trazodone (n = 1); trihexyphenidyl (n = 1).
Sentence Processing. The measures of primary interest
included response accuracy in the Sentence Span Test
(largest set size possible = 8), the Intelligibility task (total
possible =100 words), and the Sentence Comprehension
task—Relative Sentences (total correct possible = 96).
Response accuracy (number correct) for these primary
measures was converted to the proportion correct and then
transformed by the arcsine transformation prior to conducting the statistical tests (Fleiss 1986). Data values that
are presented in tables and graphs are proportions (exception: Sentence Span data are presented as Span Set Size to
facilitate interpretation); however, all statistical analyses
are based on the arcsine transformation of the proportion
of correct responses.
Sentence span memory. Performance on the Sentence Span Test (set size) differed (Fj 52 = 11.45, p =
0.001) between patients (mean/SD = 2.7/0.94) and controls (mean/SD = 3.5/0.80).
Intelligibility of accelerated presentation rate.
Both diagnostic groups exhibited a high rate of intelligibility for the individual words in sentences that were presented at the accelerated rate used in the sentence comprehension task: mean proportion of words repeated verbatim
in the intelligibility task for patients (mean/SD =
0.97/0.09) and controls (mean/SD = 0.99/0.01) did not differ (F 1 5 2 = 1.68, p = 0.20).
Sentence Comprehension Performance
Simple sentences. The mean proportion of correct
responses to questions about the simple declarative sentences for each diagnostic group under each presentation
rate condition were as follows: conversational rate
(mean/SD), patients = 0.77/0.12, controls = 0.91/0.07;
accelerated rate (mean/SD), patients = 0.75/0.14, controls
= 0.87/0.06. Diagnosis and presentation rate influenced
overall understanding accuracy for simple sentences, and
patients and controls exhibited different patterns of accuracy as a function of presentation rate. These results were
obtained as follows: Groups differed in their overall proportion of correct responses to questions about simple sentences (Fj 52 = 21.88, p < 0.001), with patients showing
reduced accuracy compared to controls (76% vs. 89%,
respectively). Presentation rate also influenced comprehension accuracy (F t 52 = 7.75, p < 0.01), with the conversational rate producing greater accuracy compared to the
accelerated rate (84% vs. 81%, respectively). The diagnosis X rate interaction effect (Fl 52 = 4.71, p < 0.05) indicated that the two groups were characterized by different
patterns of accuracy for simple sentences under the two
presentation rates. Analyses of the simple effects of presentation rate on accuracy in each diagnostic group
showed that rate influenced accuracy in controls (F( 21 =
10.87, p < 0.01) but not in patients {p = 0.64).
Relative sentences. Table 2 presents the mean proportion of correct responses to Who questions for each
diagnostic group under each relative sentence and withinsentence clause condition at each presentation rate.
Effects of linguistic structure and presentation rate
on comprehension accuracy. Each of the factors of theoretical interest exerted significant main effects on comprehension accuracy. First, sentence complexity influenced
comprehension (F, 52 = 43.92, p < 0.001), with greater
response accuracy occurring for subject-relative sentences
(84%) compared to object-relative sentences (73%).
481
Schizophrenia Bulletin, Vol. 28, No. 3, 2002
R. Condray et al.
Table 2. Comprehension of relative sentences: Proportion correct responses to Who questions,
mean (SD)
Schizophrenia patients (n = 32)
Object-relative sentences
Presentation rate
Conversational
Accelerated
Subject-relative sentences
Presentation rate
Conversational
Accelerated
Normal controls (n = 22)
Object-relative sentences
Presentation rate
Conversational
Accelerated
Subject-relative sentences
Presentation rate
Conversational
Accelerated
MainActor
Embedded—
Object
Embedded—
Actor
0.91 (0.13)
0.69 (0.22)
0.69 (0.25)
0.53 (0.24)
0.57 (0.27)
0.52 (0.24)
0.88(0.16)
0.92 (0.16)
0.75 (0.21)
0.66 (0.23)
0.71 (0.19)
0.73 (0.20)
0.95 (0.07)
0.84(0.10)
0.83(0.19)
0.70 (0.25)
0.77(0.19)
0.73 (0.23)
0.95 (0.08)
0.95 (0.09)
0.86(0.16)
0.84(0.17)
0.89(0.15)
0.93(0.11)
Note.—SD = standard deviation.
Second, type of within-sentence clause information also
affected comprehension (F 2 104 = 73.60, p < 0.001,
Greenhouse-Geisser e = 0.71). The highest rate of accuracy occurred in response to questions concerning main
clause—actor information (89%), while a reduced and
comparable accuracy rate was associated with questions
about the two types of embedded relative clauses: embedded—object (73%) and embedded—actor (73%). Pairwise
comparisons confirmed the statistical reliability of these
differences: main clause—actor versus embedded
clause—object (F, 53 = 84.82, p < 0.001); main clause—
actor versus embedded clause—actor (Fj 53 = 87.45, p <
0.001); embedded clause—object versus embedded
clause—actor (p = 0.66). Third, presentation rate additionally influenced comprehension (Fj 52 = 19.06, p =
0.0001), with the normal or conversational rate producing
a higher rate of accuracy compared to the accelerated rate
(81% and 76%, respectively).
To provide a more detailed understanding of the interaction effect involving sentence complexity, presentation
rate, and within-sentence clause (F 2 104 = 10.93, p =
0.0001, Greenhouse-Geisser e = 0.96), a set of theoretically motivated simple interactions was conducted in
which the joint effects of sentence complexity and presentation rate were examined under each level of within-sentence clause. Results indicate that the combined effects of
sentence complexity and presentation rate were influential
as a function of the type of clause information queried.
Specifically, sentence complexity and presentation rate
exerted joint effects on understanding accuracy about
"Who did X?" but not on accuracy about "Who was done
X?" These results were obtained as follows: The sentence
complexity X presentation rate interaction was significant
for accuracy about main clause—actor information (Fj 53
= 52.72, p < 0.001) and embedded clause—actor information (Fj 53 = 10.33, p = 0.0022) but not for accuracy about
embedded clause—object information (p = 0.13).
Effects of linguistic structure and presentation rate
on comprehension accuracy in schizophrenia. Patients
and controls differed in their overall proportion of correct
responses (Fj 5 2 = 14.45, p = 0.0004), with patients
exhibiting reduced comprehension accuracy compared to
controls (71% vs. 85%, respectively). More important, the
diagnosis X within-sentence clause interaction effect
(F2,i04 = 7-27> P = 0 0 0 4 ' Greenhouse-Geisser e = 0.71)
indicates that patients and controls were characterized by
different patterns of accuracy across the various types of
within-sentence clause structures. This two-way interaction is presented in figure 1. None of the other interactions
involving the diagnostic group factor reached statistical
significance.
Two sets of simple effects were tested to locate the
source of the significant diagnosis X within-sentence
clause interaction reported above. These analyses identified a group difference in the comparative levels of accuracy for information contained in main versus relative
clauses. In both groups, the highest response accuracy was
observed for information described in main clauses, com-
482
Effects of Capacity and Linguistic Structure
Schizophrenia Bulletin, Vol. 28, No. 3, 2002
Figure 1. Mean percent correct responses to Who questions about the information described in the
sentential clauses of relative sentences in schizophrenia patients (n = 32) and normal controls (n = 22)
(± standard error of the mean)
• Controls
• Patients
Main-Actor
Embedded-Object
pared to information contained in embedded relative
clauses, while similar rates of accuracy occurred for the
two types of relative clauses. However, patients showed an
average decrement of 20.5 percent in their accuracy about
embedded clause information, compared to their accuracy
about main clause information. In contrast, controls
showed an average decrement of only 10 percent in their
accuracy about embedded clause information, compared
to their accuracy about main clause information. These
results were obtained as follows: First, analyses of the
simple effects of within-sentence clause on response accuracy in each diagnostic group showed a main effect of
clause in both patient and control groups: patients (F 2 62 =
87.27, p < 0.001, Greenhouse-Geisser e = 0.69); controls
(F 2 4 2 = 13.87, p = 0.0004, Greenhouse-Geisser e = 0.64).
Moreover, pairwise comparisons of the various clause
types showed a similar pattern in each diagnostic group:
main clause—actor versus embedded clause—object
(patients: F, 31 = 99.26, p < 0.001; controls: Fj 21 = 14.98,
p = 0.0009); main clause—actor versus embedded
clause—actor (patients: F, 31 = 98.59, p < 0.001; controls:
F[ 2i = 16.01, p - 0.0006); and embedded clause—object
versus embedded clause—actor (patients: p = 0.03; controls: p = 0.09). Second, analyses of the simple effects of
diagnosis on the response accuracy associated with each
Embedded-Actor
clause type showed that accuracy for embedded clause
information differed between the two groups: embedded
clause—object in patients versus controls: Fj 52 = 10.09, p
= 0.0025; embedded clause—actor in patients versus controls: F, 52 = 21.86, p < 0.001. In contrast, the effect of
diagnosis on the accuracy rate for main clause—actor
information reached only trend significance (p = 0.04).
Standard residualized scores. To remove extraneous
variance measured by the within-sentence clause conditions of theoretical interest (relative clauses), standard
residualized scores (Chapman and Chapman 1989) were
used as the measures of individual differences (see Data
Analyses section): patients: M embedded _ object = -0.58, SD
= 0.64; M e m b e d d e d _ a c t o r = - 0 . 8 9 , SD = 0.92; controls:
Membedded-^bject = 0.0000, SD = 0.98; M embedded _ actor =
0.0008, SD = 0.98. ANOVA results were consistent with
findings based on the conventional analyses: patients and
controls differed in overall accuracy (F, 52 = 10.02, p =
0.003); type of relative clause information (embedded—
actor vs. embedded—object) influenced performance
accuracy (F, 52 = 5.49, p = 0.023); and the diagnosis X
within-sentence clause interaction effect (Fj 52 = 5.56, p =
0.022), which indicated that patients and controls were
characterized by different patterns of accuracy for the two
types of relative clause information.
483
R. Condray et al.
Schizophrenia Bulletin, Vol. 28, No. 3, 2002
Association Between Language Comprehension
Accuracy and Capacity. The association between comprehension accuracy and the psychological functions of
theoretical interest was tested using linear regression
analysis. The two diagnostic groups were combined for
these analyses (n = 54). This relationship was examined
separately for three dependent variables that involved
comprehension accuracy for each type of sentential clause
(main and embedded) and word order (SVO and OSV).
Main clauses from the two types of relative sentences
(subject- and object-relatives) represented an SVO word
order, and they were combined for the analyses.
Embedded clauses were examined separately based on the
type of relative-clause word order: the SVO order of subject-relative sentences, and the OSV order of object-relative sentences. Thus, comprehension accuracy was
denned as the number of correct responses to Who questions about the information contained in main clauses of
both subject-relative and object-relative sentences (total
possible correct = 32); embedded relative clauses of subject-relative sentences (total possible correct = 32); and
embedded relative clauses of object-relative sentences
(total possible correct = 32). Two models were tested that
involved different sets of psychological functions as the
independent variables, which were selected and organized
based on (1) the theoretical rationale outlined in the
Psychological Functions section; and (2) collinearity
issues (absence of correlations between independent variables in the present sample). Model 1: The independent
variables for Model 1 were age (control variable), digit
span (WAIS-R subtest score), and memory span for sentences (Sentence Span Test score). Results in table 3 indi-
cate an association between comprehension accuracy and
these psychological functions in this sample. Model 1
accounted for 20 percent of the variance associated with
response accuracy for information in the embedded
clauses of subject-relative sentences (SVO word order),
and 23 percent of the variance associated with response
accuracy for information in the embedded clauses of
object-relative sentences (OSV word order). The variables
in Model 1 were not successful in predicting accuracy
about main clause—information. Model 2: Because of
collinearity issues, the control variable, age, was replaced
by education in Model 2. The independent variables for
Model 2 were education, vocabulary (WAIS-R subtest
score), and intelligibility. (Note that analyses for Model 2
are based on 53 subjects because of missing data for 1
normal control for the vocabulary measure.) Results in
table 4 show an association between comprehension accuracy and this set of psychological functions in this sample.
Model 2 accounted for 32 percent of the variance associated with response accuracy for information contained in
the main clauses of relative sentences (SVO word order);
32 percent of the variance associated with response accuracy for information contained in the embedded clauses of
subject-relative sentences (SVO word order); and 39 percent of the variance associated with response accuracy for
information contained in the embedded clauses of objectrelative sentences (OSV word order).
The association between language comprehension
and general intellectual ability (WAIS-R IQ) was examined using standard tests of correlation (Pearson r) conducted separately for the two diagnostic groups. Results in
table 5 show that general intellectual functioning was not
Table 3. Association between psychological functions and comprehension accuracy for information in
sentential clauses (n = 54)
Main clauses: SVO
Predictor variables
Age
Digit span
Sentence span memory
Embedded relative clauses: SVO
Predictor variables
Age
Digit span
Sentence span memory
Embedded relative clauses: OSV
Predictor variables
Age
Digit span
Sentence span memory
R2
Adj. R2
F ratio
df
0.11
0.05
1.99
3,50
0.25
0.28
0.2
0.23
5.51
6.37
3,50
3,50
beta
SE beta
-0.0011
0.0179
0.338
0.004
0.012
0.234
<—1
-O.0045
0.017
0.968
0.005
0.015
0.296
<-1
1.16
3.28
0.35
0.25
0.002
-0.0087
0.0328
0.719
0.005
0.015
0.299
-1.81
2.21
0.08
0.032
0.02
0.13
1.54
1.45
0.77
0.13
0.15
0.002
0.001
Note.—adj. = adjusted; OSV = object subject verb; SE = standard error; SVO = subject verb object.
484
f test
2.4
Effects of Capacity and Linguistic Structure
Schizophrenia Bulletin, Vol. 28, No. 3, 2002
Table 4. Association between psychological functions and comprehension accuracy for information in
sentential clauses (n = 53)
Main clauses: SVO
Predictor variables
Education (yrs)
Vocabulary
Intelligibility
Embedded relative clauses: SVO
Predictor variables
Education (yrs)
Vocabulary
Intelligibility
Embedded relative clauses: OSV
Predictor variables
Education (yrs)
Vocabulary
Intelligibility
0.36
0.36
0.42
Adj. R2
F ratio
df
0.32
9.09
3,49
9.26
0.32
0.39
3,49
11.92
3,49
beta
SE beta
t test
0.0383
0.0228
0.594
0.015
0.014
0.158
2.52
1.69
3.76
0.02
0.1
0.001
0.0518
0.0563
0.563
0.021
0.018
0.216
2.49
3.04
2.6
0.02
0.004
0.012
0.0443
0.0592
0.794
0.02
0.018
0.212
2.17
3.26
3.74
0.04
0.002
0.001
0.001
0.001
0.001
Note.—adj. = adjusted; OSV = object subject verb; SE = standard error; SVO = subject verb object.
Table 5. Correlations between WAIS-R scores and language comprehension accuracy
Schizophrenia patients (n = 32)
Relative sentences
Main clauses
Embedded—object
Embedded—actor
Total correct
Simple and relative sentences
Normal controls (n = 22)
Relative sentences
Main clauses
Embedded—object
Embedded—actor
Total correct
Simple and relative sentences
Verbal
IQ
Performance
IQ
0.24
0.29
0.27
-0.26
-0.05
0.002
0.04
0.17
0.18
0.28
-0.09
0.15
0.30
0.52*
0.57"
0.14
0.56**
0.38
0.26
0.58
0.53
0.60**
0.47*
0.59*
Full-scale
IQ
Note.—WAIS-R = Wechsler Adult Intelligence Scale-Revised.
* p < 0.05, 2-tailed; ** p < 0.01, 2-tailed
significantly associated with language comprehension
accuracy for schizophrenia patients. In contrast, general
intellectual functioning was strongly and significantly correlated with language comprehension accuracy for normal
controls.
their overall comprehension accuracy about relative sentences, with patients exhibiting a mean reduction of 14
percent in their overall accuracy, compared to controls.
More important, patients and controls differed in their pattern of accuracy, and this pattern was a function of syntactic structure. Patients showed an average decrement of 20
percent in their accuracy about "Who did X?" and "Who
was done X?" in relative clauses, compared to their accuracy about actors in main clauses. In contrast, controls
exhibited an average decrement of only 10 percent
between relative and main clause information. Patients
Discussion
The present study addresses an unresolved question in the
literature: whether receptive syntax is intact in schizophrenia. Results showed that patients and controls differed in
485
R. Condray et al.
Schizophrenia Bulletin, Vol. 28, No. 3, 2002
and controls did not show different patterns of accuracy
for information in relative sentences as a function of presentation rate, although this finding may be due to ceiling
effects for the rate factor. For example, it is possible that
even further increases in temporal demand (e.g., deletion
of > 50% of the speech sample) would result in a different
pattern of effects of presentation rate on understanding
accuracy about relative sentences for patients versus controls. Indeed, the effects of presentation rate on language
understanding in schizophrenia disorder may be complex,
as seen in the significant diagnostic group X presentation
rate interaction obtained for simple sentence constructions, with controls showing a greater reduction in accuracy between the slow versus fast rates, compared to
patients. There are at least two possible interpretations of
the diagnostic group X presentation rate interaction for
simple sentences: (1) the faster presentation rate may have
assisted patients in maintaining context for the simple sentence constructions; or (2) the accelerated rate may have
been more disruptive to the understanding of simple sentences at higher levels of performance accuracy; that is,
controls' greater reduction in accuracy for simple sentences at the accelerated rate may have been due to their
higher overall accuracy about simple sentences (89%)
compared to patients' overall accuracy about simple sentences (76%). Finally, comprehension accuracy was correlated with functioning in the domains of memory, semantic knowledge, and temporal auditory processing.
The present findings provide a refinement of the previous literature, in which it had been shown that patients
are sensitive to syntactic structure (Rochester et al. 1973;
Carpenter 1976; Grove and Andreasen 1985) but are characterized by reduced comprehension for grammatically
complex sentences (Morice and McNicol 1985; Condray
et al. 1992; Condray et al. 1996). Consideration of the
present data, however, must include recognition of certain
interpretive challenges. An important qualification pertains to how receptive syntax function is formulated, with
a crucial question concerning which functional components are entered into the model. It is generally assumed
that receptive syntax involves multiple functional components. It is therefore possible that an increase in syntactic
complexity may merely burden further cognitive functions
that are commonly deployed across a broad range of information processing tasks (e.g., the holding online of incoming information in short-term storage, the accessing of
long-term memory and semantic knowledge networks, the
rapid processing of sequential stimuli). Thus, one interpretation is that a generalized cognitive deficit (Chapman and
Chapman 1989)—in combination with variations in task
structure, complexity, or both—produced the more precipitous decline in patients' receptive syntax function in the
present study. It is also possible, however, that increased
486
syntactic complexity requires the involvement of one or
more additional functional components (e.g., transformation operations such as movement). An alternative explanation, therefore, is that disturbances in specific components of receptive syntax function may be responsible for
patients' reduced understanding accuracy about grammatically complex constructions; for example, a disturbance in
transformational operations, such as syntactic movement,
produced difficulty in connecting "Who" with "did X"
versus "was done X." It is important to emphasize that
these two accounts do not represent mutually exclusive
alternatives, with combinations of the logic yielding additional, equally viable explanations; for example, a unique,
specific deficit in function(s) used in linguistic analyses
that is superimposed on the generalized cognitive deficiency. The two accounts are also not exhaustive. As an
additional consideration, performance accuracy may have
reflected a methodology confound whereby accuracy was
influenced by an interaction of memory function and
within-sentence word order. Specifically, the ordering of
main and embedded relative clauses in both types of relative sentences is such that the pattern of accuracy could
merely reflect a primacy-recency effect (Underwood et al.
1978), with accuracy a function of the position held by the
clause within the sentence. As a result, greater accuracy
would be expected for information in initial and last positions (main clauses), compared to the level of accuracy for
information occupying the middle positions (relative
clauses). All of these possibilities require empirical testing.
Of interest, the proportion of variance in comprehension accuracy that was accounted for by selected cognitive
functions ranged from 20 percent to 39 percent. This pattern suggests that receptive syntax function includes, but is
not fully explained by, an online interface with cognitive
functions such as memory, semantic knowledge, and temporal auditory processing. Such an interpretation is consistent with a modularity formulation (Fodor 1983), as well
as with recent formulations of cognitive function as a
dynamic and distributed process (see Carpenter et al.
2000). It is important to recognize, however, that the correlation data from the present study must be qualified by
certain limitations: (1) the cognitive measures employed
are not absolute and (2) the model is not exhaustive (e.g.,
potential specification errors, see Pedhazur 1982). Finally,
general intellectual ability (WAIS-R scores) was associated with language comprehension accuracy in normal
controls. In contrast, general intellectual ability (WAIS-R
scores) was not significantly correlated with language
comprehension accuracy in schizophrenia patients. This
latter pattern may indicate that receptive syntax dysfunction in schizophrenia is not associated with compromised
general cognitive ability.
Effects of Capacity and Linguistic Structure
Schizophrenia Bulletin, Vol. 28, No. 3, 2002
Carver, R.P. Optimal rate of reading prose. Reading
A final consideration concerns the effects on memory
of the pharmacological agents used to treat the schizophrenia patients in the present study. Findings of recent clinical
and nonhuman primate studies suggest that working memory is influenced by agents associated with antagonism of
dopamine D2 receptors; specifically, verbal working memory in schizophrenia patients was influenced by such medications (Green et al. 1997; for a review, see Meltzer and
McGurk 1999), and impairment of working memory function in adult monkeys was produced by chronic administration of haloperidol (Castner et al. 2000). If memory
capacity does influence receptive syntax function, then
administration of neuroleptic medications may have influenced patients' syntax processing.
In summary, findings from the present study suggest
the hypothesis that receptive syntax operations are disrupted in schizophrenia, and this disturbance may not be
completely explained by compromised general cognitive
ability.
Research Quarterly, 18:56-88, 1982.
Castner, S.A.; Williams, G.V.; and Goldman-Rakic, P.S.
Reversal of antipsychotic-induced working memory
deficits by short-term dopamine Dl receptor stimulation.
Science, 287:2020-2022, 2000.
Chapman, L.J., and Chapman, J.P. Strategies for resolving
the heterogeneity of schizophrenics and their relatives
using cognitive measures. Journal of Abnormal Psychology, 98:357-366, 1989.
Chomsky, N. Knowledge of Language: Its Nature, Origin,
and Use. Westport, CT: Praeger, 1986.
Comrie, B. Language Universals and Linguistic Typology.
Chicago, EL: University of Chicago Press, 1981.
Condray, R.; Steinhauer, S.R.; and Goldstein, G. Language
comprehension in schizophrenics and their brothers. Biological Psychiatry, 32:790-802,1992.
Condray, R.; Steinhauer, S.R.; van Kammen, D.P.; and
Kasparek, A. Working memory capacity predicts language
comprehension in schizophrenic patients. Schizophrenia
Research, 20:1-13, 1996.
References
Condray, R.; van Kammen, D.P.; Steinhauer, S.R.; Kasparek, A.; and Yao, J.K. Language comprehension in
schizophrenia: Trait or state indicator? Biological Psychiatry, 38:287-296, 1995.
Crow, T.J. Is schizophrenia the price that Homo sapiens
pays for language? Schizophrenia Research, 28:127-141,
1997.
Daneman, M., and Carpenter, P.A. Individual differences
in working memory and reading. Journal of Verbal Learning and Verbal Behavior, 19:450-466, 1980.
Daneman, M., and Carpenter, P.A. Individual differences
in integrating information between and within sentences.
Journal of Experimental Psychology: Learning, Memory,
and Cognition, 9:561-584,1983.
DeLisi, L.E.; Sakuma, M.; Kushner, M.; Finer, D.L.; Hoff,
A.L.; and Crow, T.J. Anomalous cerebral asymmetry and
language processing in schizophrenia. Schizophrenia Bulletin, 23(2):255-271, 1997.
Docherty, N.M.; Gordinier, S.W.; Hall, M.J.; and Cutting,
L.P. Communication disturbances in relatives beyond the
age of risk for schizophrenia and their associations with
American Psychiatric Association. DSM-III-R: Diagnostic and Statistical Manual of Mental Disorders. 3rd ed.,
revised. Washington, DC: APA, 1987.
American Psychiatric Association. DSM-IV-TR: Diagnostic and Statistical Manual of Mental Disorders. 4th ed.,
text revision. Washington, DC: APA, 2000.
Appleton, W.S. Practical Clinical Psychopharmacology.
3rd ed. Baltimore, MD: Williams & Wilkins, 1988.
Baddeley, A. Working Memory. Oxford, U.K.: Clarendon
Press, 1986.
Bleuler, E. Dementia Praecox or the Group of Schizophrenias. (1911) Translated by J. Zinkin. New York, NY: International Universities Press, 1950.
Bock, J.K. Toward a cognitive psychology of syntax:
Information processing contributions to sentence formulation. Psychological Review, 89:1-47, 1982.
Braff, D.L., and Geyer, M.A. Sensorimotor gating and
schizophrenia. Archives of General
Psychiatry,
47:181-188, 1990.
Bull, H.C., and Venables, PH. Speech perception in schizophrenia. British Journal of Psychiatry, 125:350-354,
1974.
symptoms in patients. Schizophrenia
Bulletin,
25(4):851-862, 1999.
Carpenter, M.D. Sensitivity to syntactic structure: Good
versus poor premorbid schizophrenics. Journal of Abnormal Psychology, 85:41-50, 1976.
Docherty, N.M.; Hall, M.J.; Gordinier, S.W.; and Cutting,
L.P. Conceptual sequencing and disordered speech in
schizophrenia. Schizophrenia Bulletin, 26(3):723—735,
2000.
Fabb, N. Relative clauses. In: Brown, K., and Miller, J.,
eds. Concise Encyclopedia of Grammatical Categories.
Carpenter, P.A.; Just, M.A.; and Reichle, E.D. Working
memory and executive function: Evidence from neuroimaging. Current Opinion in Neurobiology, 10:195-199,2000.
487
Schizophrenia Bulletin, Vol. 28, No. 3, 2002
R. Condray et al.
Amsterdam, The Netherlands: Elsevier, 1999. pp.
319-324.
Hopper, P.J., and Thompson, S.A. Transitivity in grammar
and discourse. Language, 56:251-299, 1980.
Faber, R., and Reichstein, M.B. Language dysfunction in
Jastak, S., and Wilkinson, G.S. Wide Range Achievement
Test-Revised. Los Angeles, CA: Western Psychological
Services, 1984.
schizophrenia.
British
Journal
of
Psychiatry,
139:519-522, 1981.
Fairbanks, G.; Guttman, N.; and Miron, M.S. Effects of
time compression upon the comprehension of connected
speech. Journal of Speech and Hearing Disorders,
22:10-19, 1957.
Just, M.A., and Carpenter, PA. A capacity theory of comprehension: Individual differences in working memory.
Psychological Review, 99:122-149, 1992.
Fleiss, J.L. The Design and Analysis of Clinical Experiments. New York, NY: John Wiley and Sons, 1986.
Just, M.A.; Carpenter, P.A.; Keller, T.A.; Eddy, W.F.; and
Thulborn, K.R. Brain activation modulated by sentence
comprehension. Science, 274:114-116, 1996.
Fodor, J.A. Modularity of Mind. Cambridge, MA: MIT
Press, 1983.
Keenan, E.L., and Comrie, B. Noun phrase accessibility
and universal grammar. Linguistic Inquiry, 8:63-99, 1977.
Fodor, J.A., and Bever, T.G. The psychological reality of
linguistic segments. Journal of Verbal Learning and Verbal Behavior, 4:414-420, 1965.
Keppel, G. Design and Analysis: A Researcher's Handbook. Englewood Cliffs, NJ: Prentice Hall, 1991.
Foulke, E., and Sticht, T.G. Review of research on the
intelligibility and comprehension of accelerated speech.
Psychological Bulletin, 72:50-62, 1969.
Freedman, R.; Adler, L.E.; Gerhardt, G.A.; Waldo, M.;
Baker, N.; Rose, G.M.; Drebing, C ; Nagamoto, H.; Bickford-Wimer, P.; and Franks, R. Neurobiological studies of
sensory gating in schizophrenia. Schizophrenia Bulletin,
13(4):669-678, 1987.
Frith, CD., and Allen, H.A. Language disorders in schizophrenia and their implications for neuropsychology. In:
Bebbington, P., and McGuffin, P., eds. Schizophrenia: The
Major Issues. Oxford, U.K.: Heinemann Medical Books,
1988. pp. 172-200.
King, J., and Just, M.A. Individual differences in syntactic
processing: The role of working memory. Journal of Memory and Language, 30:580-602, 1991.
King, J.W., and Kutas, M. Who did what and when? Using
word- and clause-level ERPs to monitor working memory
usage in reading. Journal of Cognitive Neuroscience,
7:376-395, 1995.
Kintsch, W. The role of knowledge in discourse comprehension: A construction-integration model. Psychological
Review, 95:163-182, 1988.
Knight, R.A., and Sims-Knight, J.E. Integration of linguistic ideas in schizophrenia. Journal of Abnormal Psychology, 88:191-202, 1979.
Kraepelin, E. Dementia Praecox and Paraphrenia. (1919)
Translated by R.M. Barclay. Huntington, NY: Robert E.
Krieger, 1971.
Goldberg, T.E.; Aloia, M.S.; Gourovitch, M.L.; Missar, D.;
Pickar, D.; and Weinberger, D.R. Cognitive substrates of
thought disorder: I. The semantic system. American Journal of Psychiatry, 155:1671-1676, 1998.
Kutas, M., and Van Petten, C.K. Psycholinguistics electrified: Event-related brain potential investigations. In:
Gernsbacher, M.A., ed. Handbook of Psycholinguistics.
San Diego, CA: Academic Press, 1994. pp. 83-143.
Green, M.F.; Marshall, Jr., B.D.; Wirshing, W.C.; Ames,
D.; Marder, S.R.; McGurk, S.; Kern, R.S.; and Mintz, J.
Does risperidone improve verbal working memory in
treatment-resistant schizophrenia? American Journal of
Psychiatry, 154:799-844, 1997.
Landre, N.A.; Taylor, M.A.; and Kearns, K.P. Language
functioning in schizophrenic and aphasic patients. Neuropsychiatry, Neuropsychology, and Behavioral Neurology, 5:7-14, 1992.
Grodzinsky, Y. The neurology of syntax: Language use
without Broca's area. Behavioral and Brain Sciences,
23:1-71,2000.
Leahy, L. "Sentence Span Test." Unpublished test, 1987.
Llinas, R. Is dyslexia a dyschronia? Annals of New York
Academy of Sciences, 682:48-56, 1993.
Grove, W.M., and Andreasen, N.C. Language and thinking
in psychosis: Is there an input abnormality? Archives of
General Psychiatry, 42:26-32, 1985.
Loranger, A.W. Personality Disorder Examination (PDE)
Manual. New York, NY: DV Communications, 1988.
Haarmann, H.J.; Just, M.A.; and Carpenter, P.A. Aphasic
sentence comprehension as a resource deficit: A computational approach. Brain and Language, 59:76-120, 1997.
MacDonald, M.C.; Just, M.A.; and Carpenter, P.A. Working memory constraints on the processing of syntactic
ambiguity. Cognitive Psychology, 24:56-98, 1992.
Hausfeld, S. Speeded reading and listening comprehension
for easy and difficult materials. Journal of Educational
Psychology, 73:312-319, 1981.
Maher, B.A. Language and schizophrenia. In: Steinhauer,
S.R.; Gruzelier, J.H.; and Zubin, J., eds. Handbook of
488
Effects of Capacity and Linguistic Structure
Schizophrenia Bulletin, Vol. 28, No. 3, 2002
Schizophrenia, vol. 5, Neuropsychology, Psychophysiology, and Information Processing. Amsterdam, The
Netherlands: Elsevier, 1991. pp. 437-464.
Rochester, S.R.; Harris, J.; and Seeman, M.V. Sentence
processing in schizophrenic listeners. Journal of Abnormal Psychology, 82:350-356, 1973.
Marslen-Wilson, W.D. Functional parallelism in spoken
word-recognition. Cognition, 25:71-102, 1987.
Sag, LA. English relative clause constructions. Journal of
Linguistics, 33:431-^83, 1997.
Marslen-Wilson, W.D., and Tyler, L.K. The temporal
structure of spoken language understanding. Cognition,
8:1-71, 1980.
Schwartz, S. Is there a schizophrenic language? Behavioral and Brain Sciences, 5:579-626, 1982.
Shedlack, K.; Lee, G.; Sakuma, M.; Xie, S-H.; Kushner,
M.; Pepple, J.; Finer, D.L.; Hoff, A.L.; and DeLisi, L.E.
Language processing and memory in ill and well siblings
from multiplex families affected with schizophrenia.
Schizophrenia Research, 25:43-52, 1997.
Marslen-Wilson, W.D.; Moss, H.E.; and van Halen, S. Perceptual distance and competition in lexical access. Journal
of Experimental Psychology: Human Perception and Performance, 22:1376-1392, 1996.
McClelland, J.L., and Elman, J.L. Interactive processes in
speech recognition: The TRACE model. In: McClelland,
J.L., and Rumelhart, D.E., eds. Parallel Distributed Processing: Explorations in the Microstructure of Cognition.
Cambridge, MA: Bradford, 1986. pp. 58-121.
Siewierska, A. Word order and linearization. In: Brown,
K., and Miller, J., eds. Concise Encyclopedia of Grammatical Categories. Amsterdam, The Netherlands: Elsevier,
1999. pp. 412-418.
Meltzer, H.Y., and McGurk, S.R. The effects of clozapine,
risperidone, and olanzapine on cognitive function in schizophrenia. Schizophrenia Bulletin, 25(2):233-255, 1999.
Spitzer, R.L.; Williams, J.B.W.; Gibbon, M.; and First,
M.B. Structured Clinical Interview for DSM-III-R. New
York, NY: Biometrics Research Department, New York
State Psychiatric Institute, 1989.
Miyake, A.; Carpenter, P.A.; and Just, M.A. A capacity
approach to syntactic comprehension disorders: Making
normal adults perform like aphasic patients. Cognitive
Neuropsychology, 11:671-717, 1994.
Tallal, P.; Miller, S.; and Fitch, R.H. Neurobiological basis
of speech: A case for the preeminence of temporal processing. Annals of New York Academy of Sciences,
682:27^7, 1993.
Morice, R. Beyond language: Speculations on the prefrontal cortex and schizophrenia. Australian and New
Zealand Journal of Psychiatry, 20:7-10, 1986.
Thomas, H.B.G., and Huff, F.J. Comprehension and
recoding-time of transformed sentences. Language and
Speech, 14:352-372, 1971.
Underwood, B.J.; Boruch, R.F.; and Malmi, R.A. Composition of episodic memory. Journal of Experimental Psychology: General, 107:393-419, 1978.
Morice, R., and McNicol, D. The comprehension and production of complex syntax in schizophrenia. Cortex,
21:567-580,1985.
Oldfield, R.C. The assessment and analysis of handedness:
The Edinburgh Inventory. Neuropsychologia, 9:97-113,
1971.
Van Der Auwera, J. Relative that—a centennial dispute.
Journal of Linguistics, 21:149-179, 1985.
van Kammen, D.P.; Kelley, M.E.; Gurklis, J.A.; Gilbertson, M.W.; Yao, J.K.; Condray, R.; and Peters, J.L. Predicting duration of clinical stability following haloperidol
withdrawal in schizophrenic patients. Neuropsychopharmacology, 14:275-283, 1996.
Pedhazur, E.J. Multiple Regression in Behavioral
Research: Explanation and Prediction. 2nd ed. New York,
NY: Holt, Rinehart, and Winston, 1982.
Pennington, B.F., and Smith, S.D. Genetic analysis of
dyslexia and other complex behavioral phenotypes. Current Opinion in Pediatrics, 9:636-641, 1997.
Wallace, W.P. Speed Listening: Exploring an Analogue
of Speed Reading. Technical Report NIE-G-81-0012,
1983.
Wallace, W.P, and Koury, G. Transfer effects from listening to frequency-controlled and frequency-shifted accelerated speech. Journal of Speech and Hearing Research,
24:185-191, 1981.
Perfetti, C.A., and Lesgold, A.M. Discourse comprehension and sources of individual differences. In: Just, M.A.,
and Carpenter, P.A., eds. Cognitive Processes in Comprehension. Hillsdale, NJ: Erlbaum, 1977. pp. 141-183.
Quirk, R.; Greenbaum, S.; Leech, G.; and Svartvik, J. A
Comprehensive Grammar of the English Language. New
York, NY: Longman Group, 1985.
Wallace, W.P; Stewart, M.T.; Shaffer, T.R.; and Wilson,
J.A. Are false recognitions influenced by prerecognition
processing? Journal of Experimental Psychology: Learning, Memory, and Cognition, 24:299-315, 1998.
Rausch, N.A.; Prescott, T.E.; and Dewolfe, A.S. Schizophrenic and aphasic language: Discriminable or not? Journal of Clinical and Consulting Psychology, 48:63-70,
1980.
Wechsler, D. Wechsler Adult Intelligence Scale—Revised.
New York, NY: The Psychological Corporation, 1981.
489
Schizophrenia Bulletin, Vol. 28, No. 3, 2002
R. Condray et al.
Wingfield, A. Acoustic redundancy and the perception of
time-compressed speech. Journal of Speech and Hearing
Research, 18:96-104, 1975.
Medical Center. Valuable comments were provided about
the manuscript by Angela G. Glasgow, D.V.M., Mary E.
Kelley, M.S., and four anonymous reviewers.
Acknowledgments
The Authors
This research was supported by the National Institute of
Mental Health (MH50631) and the Medical Research Service of the Department of Veterans Affairs. The authors
thank the patients for their participation in this study.
William H. Broadley IV developed the computer software
for the presentation of the language stimuli and the production of the accelerated rate in the sentence processing experiment. Consultation regarding the psychiatric diagnoses of
study participants was provided by George G. Dougherty,
Jr., M.D., John A. Gurklis, M.D., and the clinical staff of the
Schizophrenia Research Unit at the Highland Drive VA
Ruth Condray, Ph.D., is Assistant Professor of Psychiatry,
and Stuart R. Steinhauer, Ph.D., is Research Associate
Professor, University of Pittsburgh School of Medicine,
Pittsburgh, PA. Daniel P. van Kammen, M.D., Ph.D., is
Director, Global Clinical Research and Development,
Robert Wood Johnson Pharmaceutical Research Institute,
Raritan, NJ; Professor Emeritus, University of Pittsburgh
School of Medicine; and Adjunct Clinical Professor, University of Pennsylvania, Philadelphia, PA. Annette Kasparek, B.S., is Senior Research Associate, University of
Pittsburgh Medical Center, Pittsburgh, PA.
490