Psychiatry Research 216 (2014) 168–176

Contents lists available at ScienceDirect

Psychiatry Research journal homepage: www.elsevier.com/locate/psychres

Higher-order semantic processing in formal thought disorder in schizophrenia Karen Dwyer a, Anthony David b,c, Rosaleen McCarthy d,e, Peter McKenna f,g, Emmanuelle Peters a,c,n a

Kings College London, Institute of Psychiatry, Department of Psychology, London, UK Kings College London, Institute of Psychiatry, Department of Psychosis Studies, London, UK c National Institute for Health Research (NIHR), Biomedical Research Centre for Mental Health, South London and Maudsley NHS Foundation Trust, UK d University of Southampton, Department of Psychology, Southampton, UK e Wessex Neurological Centre, Southampton General Hospital, Southampton University Hospital Trust,Tremona Road, Southampton, UK f FIDMAG Research Foundation, Germanes Hospitalàries, Barcelona, Spain g CIBERSAM, Spain b

art ic l e i nf o

a b s t r a c t

Article history: Received 24 April 2013 Received in revised form 5 February 2014 Accepted 9 February 2014 Available online 17 February 2014

Higher-order semantic impairments and lack of sensitivity to linguistic context have both been implicated in formal thought disorder (FTD) in schizophrenia. Most investigations have focused on comprehension. We investigated the processing of higher-order semantic relations and the role of emotional arousal in FTD patients' linguistic comprehension and production. We compared FTD schizophrenia patients (n ¼14) with non-FTD schizophrenia patients (n ¼18) and healthy controls (n ¼15) on sense-judgment and repetition tasks, in emotionally negative and neutral conditions. We predicted that the FTD group would display poor sensitivity compared to the other two groups in comprehension and production, and that this would be exacerbated by sentence complexity and negative emotional arousal. The emotional manipulation was not robustly successful, and did not affect task performance in the patient groups. FTD patients made significantly more errors on sense-judgments and repetition than the other two groups. Complexity affected all groups to a similar extent in sensejudgments, but affected FTD patients disproportionately in the repetition task. These results support the view that a lack of sensitivity to context underlies FTD in comprehension and production stages of processing. Patients fail to utilise linguistic context to integrate lexical forms into a global whole and guide their access to lexical targets. & 2014 Elsevier Ireland Ltd. All rights reserved.

Keywords: Schizophrenia Contextual processing Comprehension Production

1. Introduction 1.1. Semantic processing and formal thought disorder Formal thought disorder (FTD) is a cardinal symptom of schizophrenia observed in approximately 16% of patients (Andreasen, 1979). It is characterised by disorganised and often incomprehensible speech, and can fluctuate in severity with emotional arousal. There is considerable evidence that a purely linguistic abnormality contributes to the speech pattern seen in FTD (see Covington n Correspondence to: Institute of Psychiatry, King's College London, Henry Welcome Building, De Crespigny Park, London SE5 8AF, UK. Tel.: þ 44 207 848 0347; fax: þ 44 207 848 5006. E-mail address: [email protected] (E. Peters).

http://dx.doi.org/10.1016/j.psychres.2014.02.011 0165-1781 & 2014 Elsevier Ireland Ltd. All rights reserved.

et al., 2005 for a review). Impaired performance on semantic processing tasks distinguishes FTD patients from non-FTD patients, in the presence of preserved syntactic functioning in both groups (Rodriguez-Ferrera et al., 2001; Oh et al., 2002). However, not all aspects of semantics are compromised. FTD patients do not necessarily have difficulty with the meaning of single isolated words but are impaired in processing higher-order semantic relations (Barrera et al., 2005). FTD patients also lack sensitivity to contextualised words (Spitzer et al., 1994), to semantic, syntactic and pragmatic anomalies (Kuperberg et al., 1998, 2000, 2006a) and to semantic relations (Besche et al., 1997). The ability to process higher-order semantic relations is partially dependent on monitoring of linguistic context. Yet it has been proposed that, “a degradation in the ability to construct and maintain internal representations of context” is a primary cognitive deficit underlying schizophrenia (Cohen and Servan-Schreiber,

K. Dwyer et al. / Psychiatry Research 216 (2014) 168–176

1992). Within the domain of language processing, a growing body of research reporting impairment in the processing of linguistic context in schizophrenia, and FTD specifically (see Kuperberg et al., 2010 for a review), has led to the proposal that FTD patients are impaired at using linguistic context to encode meaning required for the integration of information in single words in sentence construction, and that this inability underlies disorganised speech characterising FTD (Kuperberg et al., 1998). Event related potential (ERP) studies in schizophrenia using the N400 paradigm have also demonstrated that, as a group, individuals with a diagnosis of schizophrenia show abnormalities in processing linguistic context. In healthy individuals, words preceded by a semantically incongruous context (words or sentences) elicit N400s of larger amplitude than words preceded by a semantically congruous context (Kutas and Federmeier, 2011). Findings from studies investigating the N400 effect in schizophrenia are mixed. However, in a recent meta-analysis, Wang et al. (2011) concluded that patients with schizophrenia show an abnormally large N400 amplitude for contextually congruous material while they are not different to healthy controls when encountering contextually incongruous material. In other words, patients' language processes seem to be less constrained by semantic context in that it does not facilitate the generation of expectancies, while inhibition of contextually incongruent information seems to be relatively preserved. ERP studies in schizophrenia have also found abnormalities of modulation of the P600, a late positivity thought to reflect a later attempt to integrate semantic and syntactic information in order to interpret the overall message. Kuperberg et al. (2006b) found that the amplitude of the P600 in patients relative to controls was significantly reduced in morphosyntactic violations (e.g. For breakfast the boys would only eats....) and in animacy–semantic violations (e.g. For breakfast the eggs would only eat...). These findings seem to suggest that in schizophrenia an abnormality in combining semantic and syntactic information online to build up propositional meaning leaves sentence processing to be primarily driven by semantic relationships between individual words. Together these findings have led to the proposal of an imbalance in schizophrenia between locally based semantic analysis and the extraction of syntactic, pragmatic and other cues, resulting in sentence level processing anomalies (Kuperberg et al., 2010). This proposal of an imbalance in processing streams biasing semantic association analysis provides an account of a pattern of language abnormalities which seems to neatly capture the main characteristic of FTD, namely a preoccupation with the relationship between individual words. Indeed many electrophysiological studies have found correlations between reduced N400 and FTD (e.g. Andrews et al., 1993; Debruille, 2007; Kuperberg et al., 2006b Ditman and Kuperberg, 2007; Kostova et al., 2005; Kreher et al., 2008; Sitnikova et al., 2009), which seems to suggest a specific link between FTD and the inefficient use of linguistic context. It has alternatively been proposed that impaired contextual processing is not specific to FTD, but characteristic of schizophrenia as a whole, and that FTD only manifests when integration processes break down completely, a proposal yet to be rigorously tested. Moreover, contextual processing studies have primarily assessed patients' performance on comprehension-based tasks, assuming the same failure of mechanisms to be involved in the processing context in speech. There are good theoretical reasons for expecting the same factors to influence patients' speech production. Comprehension and production processes are interwoven (Pickering and Garrod, 2007, 2013): production is understood as a mapping of information from higher to lower linguistic levels while comprehension operates in reverse. Production relies on internal self-monitoring (of the utterance under construction) and external self-monitoring (of the output), both using comprehension processes that convert sound to

169

meaning (Levelt, 1989). Furthermore, comprehension and production processes demonstrate the recruitment of overlapping neural circuits (Scott and Johnsrude, 2003). Thus, the combining of semantic and syntactic information from context to create higherorder semantic connections serves both speech input and output systems (Levelt, 1989; Monsell, 1987). A disruption to these connections would thus manifest in both comprehension and production. However, the crucial experiments comparing comprehension and production have not been conducted. 1.2. Emotional arousal and formal thought disorder FTD can vary over time and severity may fluctuate. It has been widely observed that, in a clinical setting, patients show more thought disorder when they are emotionally aroused, for example, when discussing their symptoms. FTD in schizophrenia has been shown to increase in interviews with emotionally salient material (Haddock et al., 1995), and similar results have been found in thought disordered manic patients (Tai et al., 2004). Furthermore, certain types of communication disturbances in schizophrenia have been found to be more reactive to affect, namely poor reference performance (Docherty et al., 1994a, 1994b), vague references, confused references and ambiguous word meanings (Docherty and Hebert, 1997). While there is considerable evidence that affect modulates language processing in schizophrenia, this has not been examined specifically in relation to the emerging evidence that an abnormality of contextual processing underlies FTD. 1.3. Aims The first aim of the present study was to extend the findings of Kuperberg et al. (1998, 2000) by investigating FTD patients' sensitivity to semantic context in both comprehension and production in an off-line task using the same linguistic material for both modalities, in comparison to non-FTD patients and matched healthy controls. A verbatim sentence repetition task was used for the production task since this type of immediate recall is believed to employ the same mechanisms as those used in normal speech production (Potter and Lombardi, 1992), in that recall is believed to be reconstructed by drawing on activated conceptual and lexicalsemantic information (e.g. Levelt, 1989). This allows investigation of participants' ability to construct the global whole within a visible context. The second aim of this study was to examine whether emotional arousal would have a detrimental impact specifically on FTD patients' ability to integrate contextual information in sentence comprehension and production. It was predicted that the FTD group would produce more errors in both comprehension and production than non-FTD patients and healthy controls. While the rate of errors made by all groups was expected to increase with sentence complexity due to the increased working memory demand, FTD patients were predicted to be more affected by increasing complexity. Similarly, it was predicted that negative emotional arousal would result in poorer performance in all groups, but that the greatest decrement would be observed in the FTD patients.

2. Method 2.1. Participants All patients met Diagnostic and Statistical Manual of Mental Disorders (4th ed., DSM-IV, American Psychiatric Association, 1994) criteria for schizophrenia, were aged 18–65, and were native British speakers of English. All patients had ongoing chronic symptoms despite having been relatively stable on neuroleptic medication between 4 months and 5 years (average length of treatment, 3.9 years). They were

170

K. Dwyer et al. / Psychiatry Research 216 (2014) 168–176

recruited from inpatient wards in the South London and Maudsley NHS Foundation Trust (SLaM), and from rehabilitation community teams in Fulbourn Hospital, Cambridge. The patients were subdivided into FTD and non-FTD groups on the basis of ratings on the Thought, Language and Communication Scale (TLC; Andreasen, 1979). Interviews using the Scale for the Assessment of Positive Symptom (SAPS; Andreasen, 1984) were carried out by the 1st author at the time of testing and audio recorded. These interviews were used to determine the presence of FTD for purposes of assigning patients to FTD and non-FTD groups subsequent to participation in the study. Positive FTD (derailment, tangentiality, incoherence, neologisms, illogicality, circumstantiality, pressure of speech, distractible speech, clanging) and poverty of speech were assessed with the version of Andreasen's TLC scale in the Comprehensive Assessment of Symptoms and History (CASH) (Andreasen et al., 1992). All recorded interviews were rated by a separate author (P.J.M), a senior clinician with extensive experience of schizophrenia. A summative score of positive FTD was established by adding ratings for the nine positive FTD measures and poverty of speech. Those with a summative CASH score of 3, 4, or 5 (moderate, marked or severe FTD) were assigned to the FTD group (N ¼14). Participants receiving a global score of r 2 were allocated to the non-FTD group (N¼ 18). Healthy control participants were recruited from the Camberwell job centre in London and from administrative positions at University College London. Participants were included in the healthy control group only if they reported no history of contact with mental health services during a phone screen and no history of neurological disorder, or of drug or alcohol abuse. 2.2. Materials This study involved two tasks designed to measure how well participants could (i) determine whether or not sentences made sense (judgment task), and (ii) accurately repeat sentences verbatim (repetition task). The same sentences, with two levels of varying complexity (single sentences and sentence pairs) were used in both tasks. Where participants failed to recall sentences verbatim, participants' responses were rated for intactness of meaning in spite of any structural or lexical changes in their responses. This was scored as a ‘gist' score. Materials were adapted from Almor et al. (1999). 2.2.1. Language tasks – comprehension and production 2.2.1.1. Single sentences task. This task consisted of 104 sentences, half of which were emotionally salient and described negative emotional states (e.g. scared, tormented). Neutral verbs (e.g. influenced, featured) were used in the neutral condition. Emotionally salient and neutral sentences were matched for sentence length, linguistic complexity, and verb frequency (Kucera, and Francis, 1967) (all p's 4 0.4). Sentences appeared in four forms, varying in plausibility (plausible vs. implausible) and length (one clause sentences vs. three clause sentences). The four constructions are illustrated in Table 1. The different sentence lengths were used to prevent participants from predicting the structure of the sentences. Presentation of the different forms was randomised. 2.2.1.2. Sentence pairs task. This task consisted of 64 sentence pairs, half of which were emotionally salient. They were matched with an emotionally neutral pair for length (t¼0.948, df¼63, p¼0.38), and main verb frequency (Kucera and Francis, 1967; (t¼ 0.887, df¼ 63 p¼ 0.35). In each of the first sentences, two animate entities appeared in the grammatical roles of subject and object. In the second sentence, the same entities appeared either as a pronominal form or a noun phrase (NP) synonym. In half of the sentences, the subject (NP or pronoun) of the second clause referred to the subject in the first clause; in the other half, the subject (NP or pronoun) in the second clause referred to the object of the first clause. These two sentence forms were used in order to prevent participants from predicting the order of the subject and object in the second clause. Half the sentence pairs were plausible and half implausible. In the implausible sentences there were two forms. In the agreement implausible stimuli, the pronoun in the second clause did not match the antecedent in number (singular vs. plural). In the second type of implausible construction, the meaning of the second clause was semantically incongruous with the first. Example sentences for both the Single Sentences and Sentences Pairs Tasks are shown in Table 1. Sentence pairs were randomized and appeared in their appropriate forms either in the first set of 16 sentence pairs or in the second set. Instructions were the same for both the Single Sentences Task and Sentence Pairs Task. Participants were informed that the experimenter would read out a list of sentences, only some of which would make sense. Sentences were read aloud to participants rather than using a recording as this provided a more naturalistic presentation and aimed to help participants maintain focus over the 40–60 min session. Participants were asked to carry out two tasks after each of the sentences was read out: they were instructed to (1) decide whether the sentence made sense or not, (by stating yes or no) (judgment); (2) repeat the sentence as accurately as possible (repetition). All responses were recorded on an Olympus DM-20 Dictaphone and later transcribed.

2.2.2. Emotional arousal The role of emotional arousal was addressed in two ways: (i) by manipulating the emotional salience of the material (see above), and (ii) by conducting an emotionally arousing interview prior to task presentation. The SAPS (Andreasen, 1984) was used to induce negative emotional arousal in the clinical groups by discussing an emotionally significant topic, i.e. their symptoms. Healthy control participants were presented with a written cue card containing a negative personal characteristic (e.g. scared, worried, miserable), and were asked to think of a specific time when they exhibited or displayed the characteristic. Words were taken from the Autobiographical Memory Test (AMT; Williams and Broadbent, 1986). Emotionally neutral conditions were established for both patients and healthy control participants by using the positive cue words adapted from the AMT (positive words rather than purely neutral words were used because most people have a slightly positive baseline mood).

2.2.3. Tests of cognitive functioning Estimated pre-morbid IQ was measured using the National Adult Reading Test (NART; Nelson, 1982). Current IQ scores for the patient groups were calculated on the basis of seven subtests from the WAIS-R (Wechsler, 1981); four verbal tests: Digit Span, Arithmetic, Similarities, and Vocabulary, and three performance tests: Picture Arrangement, Block Design, and Picture Completion. Working memory function was assessed using the WAIS-III (Wechsler, 1997) forward Digit Span and Letter-to-Number Span tasks. Working memory was assessed in both emotionally salient and neutral conditions in case it was also affected by the emotional manipulation.

2.3. Procedures Participants were interviewed in two to three sessions each lasting approximately 40–60 min. Sessions were between 2 days and 2 weeks apart. The order of administration of the tasks (WAIS-R, NART, Language tasks, and letter-to-number working memory) was counterbalanced across participants and across the neutral and emotionally salient conditions. Emotional salience was manipulated by presenting the SAPS and AMT interviews immediately prior to the language tasks. For half of the participants, emotionally salient conditions were established on the first day with the emotionally neutral conditions being established on the second. For the other participants the order was reversed. Only sentences containing emotionally negative items were presented in emotionally negative conditions (following the SAPS or negative condition of the AMT interview), and only sentences containing emotionally neutral items were presented in the neutral condition following the positive word items from the AMT. The interviews were conducted to create arousal while the presentation of either negative or neutral verbs in the experimental items was included in order to maintain it for the duration of the task. The State version of the State and Trait Anxiety Inventory (STAI) Self Evaluation Questionnaire (Spielberger et al., 1970) was administered both prior to and following testing on the language tasks in both conditions to assess changes in emotional arousal.

3. Results 3.1. Clinical and demographic group differences Demographic details of the patients and healthy controls, and statistical test results of any differences between groups are shown in Table 2. The three participant groups did not differ in age or gender distribution. There were no differences between the patient groups on the NART (Nelson, 1982), but both patient groups scored higher than healthy controls. The three groups did not differ on working memory, as assessed by summed scores of the Digit Span and Letter-to-Number Span scores. However, the effect was borderline significant for the Letter-to-Number Span, and pairwise contrasts showed that the FTD group scored significantly lower than the healthy controls. For this reason, the score on Letter-toNumber Span was entered as a covariate in all main analyses. The two patient groups did not differ on WAIS-R or on scores for delusions or hallucinations in the SAPS, but they differed significantly on global scores for FTD as expected.

Implausible semantic forms

The second sentence is semantically implausible when taking into account the clausal relationship between the subject and the object in the first sentence.

The tourists listened to the amusing safari guide on the long trip. The elderly gentleman listened to the loud young women During the dull parts, they really tried to entertain the travellers on the bus. After a while, he really began to annoy the old man The MP begged the press not to release the story. But The secretary asked the salesmen not to smoke in reception. Reluctantly, she obeyed the reps typically, he exposed the media Either the synonym or the pronoun in the 2nd clause is incongruous with the semantic features of its antecedent, so that it cannot be mapped logically.

A plausible sentence in which the synonym and pronoun in the 2nd clause refer back to the The tourists listened to the amusing safari guide on the long trip. The elderly gentleman listened to the loud young women object and subject of the 1st clause respectively. During the dull parts, the rep really tried to entertain them on the bus. After a while, the girls really began to annoy him. The government uncovered three secret agents in the A plausible sentence in which the synonym and pronoun in the 2nd clause refer back to the The careers advisor questioned the job seekers about work. subject and object of the 1st clause respectively. Mostly, the employment officer found them something department. Naturally, the state punished them

The man who overheard it amused the comment that was made The man who overheard it troubled the comment that was made Three clause sentence with the subject and object clauses reversed, creating an animacysemantic violation

The comment that was made amused the man who overheard it The comment that was made troubled the man who overheard it

The comment was troubled by the man The comment was amused by the man

One clause passive sentence with the object and subject reversed, creating an animacysemantic violation Plausible three-clause sentence consisting of main clause and two relative clauses

The man was troubled by the comment The man was amused by the comment One clause plausible passive construction

Single sentences Passive plausible Passive implausible Relative clauses plausible Relative clauses implausible Sentence pairs Plausible (object synonym) Plausible (subject synonym) Implausible agreement

Explanation

Table 1 Example sentences from the single sentence and the sentence pairs task.

Emotionally neutral condition

Emotionally salient condition

K. Dwyer et al. / Psychiatry Research 216 (2014) 168–176

171

3.2. Scoring of language tasks 3.2.1. Comprehension (‘judgment’) Correct judgments were scored as 1, with a range of scores of 0–52 for both emotional conditions in the Single Sentences Task. In the Sentence Pairs Task scores ranged between 0 and 32 for each condition. 3.2.2. Production (‘repetition’) On the Single Sentences Task, repetition responses were scored 1 for verbatim recall, and 0 if the response included any deviation from the target. The potential range of scores for this task was therefore 0–52 for each condition. The Sentence Pairs Task consisted of two sentences in the target. Verbatim repetition was scored as 0, 1 or 2 for a correct response. The potential range of scores was therefore 0–64 for each condition. A more lenient scoring system was adopted for the sentence pairs task, in which errors involving articles, prepositions, coordinators and inflectional morphemes, such as plural markers, were accepted. 3.2.3. Gist scores Two linguists rated each of the responses for accuracy of gist on a scale of 0 (poor accuracy) to 5 (high degree of accuracy). Each rater scored gist of the target independently and their scores were averaged. The potential range of scores was 0–160 for each condition. In order to calibrate inter-rater reliability, the two raters compared their scores on 15 scripts totalling 120 sample sentence pairs. A Spearmans' correlation of the two linguists' sets of scores on the same sentences showed a highly significant relationship between the sets of scores (rs¼ 0.75, p o0.01), demonstrating good agreement. 3.3. Emotional salience manipulation In order to establish whether the salience manipulation was successful, the STAI (Spielberger et al., 1970) scores measuring emotional arousal prior to and after performing the language tasks were entered as a 3 (group)  2 (salience: emotional negative vs. neutral)  2 (time: pre language task vs. post language task) ANOVA. Results showed that there was a main effect of group (F(2,44) ¼4.93, p ¼0.01), with Tukey's Least Significant Difference (LSD) post hoc tests revealing that the FTD group were more emotionally aroused than the non-FTD group (p ¼0.01) and the healthy controls (p ¼0.01), while there was no significant difference between the non-FTD group and healthy controls (p ¼ 0.97). The main effect of emotional salience did not reach significance (F(1,44) ¼ 3.91, p ¼0.054), and there was no main effect of time (F(1,44) ¼ 0.32, p ¼0.75), indicating that the manipulation in increasing emotional arousal was not robustly successful. There was no interaction between salience and group (F(2,44) ¼1.03, p¼ 0.37), thus the FTD group scored higher on emotional arousal regardless of the manipulation. There was no significant interaction between time, salience and group (F(2,44) ¼11.81 p ¼0.28). Descriptive scores for the STAI are shown in Table 3. 3.4. Main analyses The correct judgment responses and correct repetitions scores were summed for each task to produce total scores. Since the maximum total scores on the two tasks differed, the total scores were transformed into percentages. Repetition scores were not normally distributed, and logit transformations were carried out to normalize the data. In order to control for response bias, a signal detection index of sensitivity was used for judgment (d' prime), based on a division of responses into true positives, true negatives,

172

K. Dwyer et al. / Psychiatry Research 216 (2014) 168–176

Table 2 Demographic characteristics of the three groups (means; standard deviations (SDs) in parentheses).

Gender (M/F) Age in years NART estimated IQ WAIS-R Letter-to-number Forward digit span SAPS positive symptoms without FTD) Thought disorder

FTD patients (n ¼14)

Non-FTD patients (n¼ 18)

Healthy controls (n ¼15)

Tests of significance

10/4 41 (12.6) 112.93(11.2) 94.36 (8.9) 8.5 (2.6) 7.00 (1.7) 2.43 (1.04) 3.57(0.9)

14/ 4 37.5 (10.2) 113.72 (9.1) 98. (12.9) 9.67 (2.9) 8.5 (2.0) 1.71(1.34) 0.06 (0.24)

9/6 35.9 (14.8) 103.51 (9.4) — 11.47 (3.9) 9.17 (3.4) — —

χ2 ¼ 2.63, df¼ (2), p ¼ 0.304 F(2,45) ¼ 0.63, p ¼0.54 F(2,44) ¼ 5.2, p¼ 0.01b*c* t ¼  .90, df¼ (30), p ¼ 0.22 F(2,44) ¼3.22, p ¼ 0.05 b* F(2,44) ¼2.9, p ¼ 0.07 t¼ 1.7, df¼ (31), p ¼ 0.27 t¼ 18.8, df¼ (31), p o 0.001

†

Tukey's LSD Least significant difference test. FTD patients versus non-FTD patients. FTD patients versus healthy controls. c Non-FTD patients versus healthy controls. a

b

n

po 0.05.

false positives and false negatives. Scores for judgment and repetition were entered into separate 3 (group: FTD vs. non-FTD vs. healthy control)  2 (salience vs. neutral)  2 (complexity: single vs. pairs) ANCOVAs for each dependent variable (judgment or repetition) and letter-to-number scores representing working memory were entered as a covariate. (See Fig. 1.) Descriptive statistics are displayed in Table 4. We predicted main effects of group and complexity; significant interactions of group and complexity; and of group and emotional salience, for both judgement and repetitions tasks. Results for repetition, judgment and gist are shown in Table 5. 3.4.1. Judgment and repetition 3.4.1.1. Judgement. As displayed in Tables 5, a 3 (group)  2 (complexity)  2 (salience) ANCOVA revealed that, as expected, there was a main effect of group, with LSD post hoc tests showing that the FTD group was significantly less sensitive to sentence acceptability than either the non-FTD group (po0.01) or the healthy controls (po0.001). There was no difference between the non-FTD group and healthy controls (p¼0.23). A significant effect of complexity emerged, but there was no interaction between complexity and group, showing that complexity had the same deleterious effect on performance in all groups. There was no main effect of emotional salience or interaction with group. There was an unexpected interaction between salience, complexity and group which just reached significance. This interaction was driven by differences between the FTD patients and healthy controls (but not between the two patient groups), and was due to both groups performing slightly better in the emotionally salient condition for single sentences than in the neutral condition, while the opposite was the case for complex sentences in the healthy control group only, with FTD patients performing ‘at floor' for complex forms in both conditions. There was a significant effect of working memory, with low working memory being associated with decreased performance in all groups, but no interaction with complexity or salience. 3.4.1.2. Repetition. A 3 (group)  2 (complexity)  2 (salience) ANCOVA revealed a main effect of group. LSD post-hoc tests showed that the FTD group made significantly more errors than both non-FTD patients and the healthy controls (po0.001). The difference between non-FTD patients and healthy controls was at trend (p¼0.07). There was a main effect of complexity, with performance overall being poorer on the sentence pairs. There was no main effect of salience or interaction with group. There was a main effect of working memory (again with low working memory being associated with decreased performance), but no interaction with complexity or salience.

Table 3 Means and SDs (in parentheses) for the state measure of the State and Trait Anxiety Inventory (STAI) Self Evaluation Questionnaire in the three groups pre- and postlanguage tasks. Higher scores represent higher levels of anxiety.

Negative condition Pre task Post task Positive condition Pre task Post task

FTD patients (n¼ 14)

Non-FTD patients (n¼ 18)

Healthy controls (n¼ 15)

22.07 (8.47) 21.14 (7.2)

16.0 (5.39)

15.73 (4.2)

16.06 (5.5)

16.2 (3.49)

20.0 (7.58) 20.86 (6.2)

16.56 (6.17) 14.94 (4.5)

15.67 (4.2) 15.73 (5.12)

Fig. 1. Medians for relative percentage scores on judgment, repetition and gist tasks (averaged across the two emotional conditions).

There was also an interaction between complexity and group. Follow-up ANCOVAs for two groups at a time revealed significant or near-significant interactions between all group comparisons. Pairwise comparisons showed that although the FTD patients were significantly worse than healthy controls on single and paired sentences (p o0.01), they were equivalent to the non-FTD patients on single sentences (p¼ 0.29), only making more errors on the sentence pairs task (po 0.01). The non-FTD patients were only worse than healthy controls on the sentence pairs task (p ¼0.01).

3.4.1.3. Gist repetition. Since pairwise comparisons showed that the FTD patients performed comparably to the non-FTD patients

K. Dwyer et al. / Psychiatry Research 216 (2014) 168–176

173

Table 4 Medians and interquartile ranges (25 and 75 percentiles) of the relative percentage scores for judgment, repetition and gist, and of d-prime scores, for the language task performance of patients with formal thought disorder (FTD), non-formal thought disorder (non-FTD), and healthy control participants. High scores represent more accurate responses. FTD patients (n¼ 14)

Single sentence task Judgment d' prime Repetition: Sentence pairs task Judgment d' prime Repetition Gist

Non-FTD patients (n¼ 18)

Healthy controls (n¼ 15)

Neutral

E.S.

Neutral

E.S.

Neutral

E.S.

73 (56–85) 1.7 (.78–2.4) 72 (45–87)

77 (56–92) 2.1 (1.1–2.89) 80 (43–95)

87 (77–91) 2.3 (1.58–3.57) 87 (66–93)

94 (83–98) 3.4 (2.28–4.18) 83 (72–95)

90 (87–94) 2.79 (2.64–3.86) 88 (88–96)

96 (94–98) 4.09 (3.22–5.33) 94 (84–98)

56 (47–72) 0.72 (  0.13–1.12) 23 (05–48) 44 (25–57)

53 (49–63) 0.17 (0.00–1.2) 18 (07–35) 39 (21–55)

75 (64–84) 1.76 (0.85–2.44) 48 (34–68) 71 (61–81)

77 (52–91) 2.1 (0.29–2.72) 50 (32–75) 72 (53–87)

88 (75–91) 2.42 (1.38–3.07) 77 (73–88) 88 (83–91)

75 (66–88) 1.77 (0.83–2.42) 72 (59–86) 88 (81–90)

Legend: ES ¼ emotionally salient.

Table 5 Results of 3-Way ANCOVAs for judgement, repetition and gist scores. ANCOVAs Judgement Group Complexity Salience Complexity and group Salience and group Complexity, salience and group FTD vs. non-FTD FTD vs. healthy controls Non-FTD vs. healthy controls Working memory (covariate) Repetition Group Complexity Salience Complexity and group FTD vs. non-FTD FTD vs. healthy controls Non-FTD vs. healthy controls Salience and group Complexity, salience and group Working memory (covariate) Gist Group Salience Salience and group Working memory (covariate)

(F(2,43) ¼ 9.64, po 0.01)a**b*** (F(1,43) ¼ 13.21, p ¼0.01) (F(1,43) ¼ 0.04, p ¼0.84) (F(2,43) ¼0.54, p ¼ 0.6) (F(2,43) ¼1.53, p¼ 0.23) (F(2,43) ¼ 3.12, p ¼0.05) (F(1,29) ¼ 0.22, p¼ 0.64) (F(1,26) ¼ 6.94, po 0.05) (F(1,30) ¼ 3.33, p ¼ 0.08) (F(1,43)¼ 36.29, p ¼0.001) (F(2,43) ¼ 8.79, p o0.01)a** (F(1,43) ¼ 8.75, po 0.01) (F(1,43) ¼ 0.29, p¼ 0.59) (F(2,43) ¼ 7.12, po 0.01) (F(1,29) ¼ 4.02, p¼ 0.054) (F(1,26) ¼ 11.75, p o 0.01) (F(1,30) ¼ 5.24, p¼ 0.03) (F(2,43) ¼0.33, p ¼ 0.72) (F(2,43) ¼0.15, p ¼ 0.86) (F(1,43) ¼ 8.65, po 0.01)

b***c

(F(2,43) ¼ 28.90, po 0.001) (F(1,43) ¼ 0.49, p ¼0.49) (F(2,43) ¼0.5, p ¼ 0.61) (F(1,43) ¼ 12.47, p o0.01)

a***b***c*

(at trend)

Significant differences in bold. Tukey's LSD least significant difference test. a

FTD patients versus non-FTD patients. b FTD patients versus healthy controls. c Non-FTD patients versus healthy controls. n po 0.05. nn p o0.01. nnn p o 0.001.

on single sentences, gist scores were calculated only for the 2-sentence scenarios. A 3 (group)  2 (salience) ANCOVA showed a main effect of group. LSD post hoc tests showed that the FTD group produced more errors than the non-FTD group (p o0.001) and the healthy controls (p o0.001), with the non-FTD group also producing more errors than the healthy controls (p ¼0.01). There was no main effect of emotional salience or interaction with group. There was a significant main effect of working memory, but no interaction with salience.

3.4.1.4. Relationship between judgment and repetition. While verbatim recall is believed to reflect speech production in that it draws on activated conceptual and lexical-semantic information

and engages full semantic and syntactic integration in order to produce accurately (Potter and Lombardi, 1990, 1992; Bock, 1982; Garrett, 1982; Levelt, 1989), the production repetition processes in recall here may not necessarily capture those processes that are engaged in the de novo production of utterances in patients. In particular, repetition is partially dependent on comprehension in order to facilitate accurate recall. In order to determine whether FTD patients' poorer performance on production was purely the influence of impaired comprehension, the relationship between correct judgement and accuracy of repetition was further investigated. Relative percentage scores were entered into a bivariate correlation. Separate Spearman's correlations for the three groups showed that the positive relationship between judgement and repetition was weaker and non-significant for the FTD group on both the single sentences rs ¼0.46, (p¼ 0.1) and at trend for paired sentences rs ¼0.42, (p ¼0.09), while the non-FTD groups showed a positive relationship with judgement and repetition on the single sentences rs ¼0.78, (p o0.001) and sentences pairs rs ¼0.61, (p ¼0.01), and the healthy control group showed a positive relationship on both single sentences rs ¼ 0.71, (po 0.001) and a trend on the paired sentences task rs ¼ 0.47, (p ¼0.07).

4. Discussion 4.1. Judgment and repetition This study compared FTD patients, non-FTD patients, and a healthy control groups' sensitivity to linguistic context in both comprehension and production of the same linguistic material, using plausibility judgment and verbatim repetition tasks. Our findings of FTD patients' poor performance on sense–judgments, compared to the other two groups, indicates a difficulty in sentence comprehension, a finding consistent with both behavioural and electrophysiological studies of poor contextual processing in FTD (see Kuperberg et al., 2010 for a review). FTD patients' poor sensitivity to higher-order semantic relations in comprehension was also observed in their poor reconstruction of higherorder relations in production. This is an important finding because a problem of processing higher-order semantic relations in comprehension is assumed to reflect a deficit in using linguistic context on-line, which could account for the disorganised speech observed in FTD. In light of the interaction between comprehension and production processes (Pickering and Garrod, 2013), which share word meaning connections and conceptual representation that guides lexical selection (Levelt, 1989), such an assumption predicts a similarly poor performance on the reconstruction of the same structures in production.

174

K. Dwyer et al. / Psychiatry Research 216 (2014) 168–176

One interpretation of these findings is that verbatim recall may have a greater dependency on comprehension than de novo production, and thus poor performance on recall reflects compromised comprehension. However, the FTD patients in fact showed greater difficulty in the reconstruction aspect of the task, demonstrating that performance on production was not simply an artefact of poor comprehension. Furthermore, the FTD group was also the only group that did not actually show a correlation between production and comprehension. What this does indicate is that FTD as defined by a detailed analysis of de novo production abnormalities (TLC) is a good indicator of heterogeneous disruption to comprehension and repetition processing. Plausibly the poor performance on both comprehension and production in FTD patients is indicative of a disruption to a level shared by these processes, yet this cannot be determined by the data in the present study. Further research may benefit from investigating the possibility that the disruption observed in both comprehension and repetition processing may entail a level shared by these processes. There are a number of confounding cognitive factors which may have impacted on performance. In particular, the detrimental impact on both FTD patients' comprehension and production with increasing complexity is suggestive of problems of working memory, which seems to contribute to some aspects of language processing in schizophrenia. Considerable evidence seems to indicate that WM is a core feature of schizophrenia (see Lee and Park, 2005 for a meta-analysis) and WM deficits have been found to be strongly correlated with both language comprehension (Bagner et al., 2003) and language production deficits (Cohen et al., 1999; Docherty et al., 1996). There was indeed an effect of working memory in both repetition and judgment, in both cases associated with decreased performance in all groups. However, these results are difficult to account for entirely on the basis of differences on a measure of WM that taps into the central executive component of WM between the groups, since all effects were significant despite WM scores being co-varied out. It is also important to note that the comprehension tasks used an off-line measure to assess language in schizophrenia, which may capture quite different processes from the types of mechanisms engaged as people comprehend or produce language as it unfolds on-line. Off-line comprehension may also be more vulnerable to general cognitive deficits than the on-line procedures. Referential communication impairments have been demonstrated to be correlated with poor performance on a number of neuropsychological measures in addition to WM, namely immediate auditory memory and distractibility, sustained attention and sequencing (Docherty and Gordinier., 1999; Docherty, 2005). It is possible that the additional difficulty in speech production, drawing on resources not required for comprehension, for example selfmonitoring and planning, crucial in generating a discourse plan (Docherty et al., 1996) could have impacted to some degree on repetition. Yet, a generalised cognitive deficit is unlikely to provide an explanation for the pattern of results we have found since there was no significant difference between the two patient groups on the WAIS-R or between the three groups on the NART. Findings here are also consistent with previous studies that have found impaired performance of schizophrenia (Kuperberg et al., 2006b) and FTD patients (Kuperberg et al., 1998; Tamlyn et al., 1992) on sense judgments in off-line tasks testing patients' ability to reflect on the meaning of anomalous sentences. It is possible that FTD patients' poorer performance in repetition as complexity increases may in part be the result of a failure to use lexical items after they have been recognised. In contrast, the gist analysis was not based on accuracy of lexical retrieval, but simply relied on recall of the general meaning, regardless of the structure of the sentence or of the lexical items. In healthy participants, sentence recognition tends to improve as the number

of propositions increases. This is believed to reflect the tendency to integrate semantic information and to store the message as a whole, rather than storing the structural form verbatim. Again, the FTD patients' poor gist production seems to suggest a failure to integrate semantic information from context to produce an accurate conceptual representation. We interpret these findings as supporting evidence for the view that patients are impaired at combining semantic and syntactic information effectively to build up linguistic context. An imbalance in semantic association analysis and syntactic algorithm processing strategies in schizophrenia has recently been proposed (Kuperberg et al., 2010). Over-reliance on semantic associations, in place of employing a complete combinatorial parse to a sentence structure, i.e., a global analysis of the intended meaning, could account for the FTD group's poor sensitivity to violations and failure to spot errors. It might also explain the reasons for their failure to monitor sentences construction in their spontaneous speech. 4.2. Emotional salience and FTD language performance It was predicted that with increased negative emotional arousal, FTD patients would show a decrement in processing linguistic context, which was not supported by our findings. However the emotional manipulation was not robustly successful, and thus we cannot make firm conclusions on the potential impact of emotional arousal on language performance in the patient groups. We did find a salience, complexity, and group interaction between the FTD patients and healthy controls (but not between the two patient groups), but this interaction seemed to be due to both groups performing slightly better in the emotionally salient conditions for single sentences than in the neutral condition, while the opposite was true for complex sentences in the healthy control group only, with FTD patients performing poorly with complex forms in both conditions. Since the manipulation of emotional arousal was not robust across the three groups, it is difficult to know how meaningful this interaction was. Of most theoretical interest and predicted in our hypotheses were the two-way interactions that involved complexity/group and salience/group, which were not significant. There were a number of limitations in the emotional arousal manipulation used in this study. First, the use of a state anxiety measure may not have been the best way to assess the negative emotional arousal we were attempting to induce, and more sophisticated measures such as skin conductance would have been preferable. Second, the selection of positive rather than neutral words in the neutral condition may have created a positive mood induction in the neutral condition, which may also have led to increased arousal. Further research is therefore needed to ascertain the potential role of emotional arousal in FTD language performance. In conclusion, this study is the first to establish that FTD patients display impairment of both judgment and repetition compared to non-FTD patients and healthy controls. We propose these findings can be accounted for by a problem of contextual processing whereby patients are impaired at combining semantic and syntactic information effectively to build up linguistic context. These finding were specific to FTD. We have shown that the most basic form of meaning judgments were intact in non-FTD patients, supporting a categorical distinction between thought disordered and non-thought disordered schizophrenia patients. However, the other tasks tended to expose weaknesses in the non-FTD patients as well supporting the notion of a continuum of context processing deficits in schizophrenia. There may be more than one level of linguistic impairment underlying the communication abnormalities found in FTD. Schizophrenia is a heterogeneous disorder, and some symptoms may also be heterogeneous in their origin and underlying processes, including FTD (Docherty et al., 2003). An imbalance in the two

K. Dwyer et al. / Psychiatry Research 216 (2014) 168–176

streams of processing (semantic and syntactic) would not necessarily exclude the possibility of an additional problem within semantic memory in this group (although this might be expected to impact on measures such as WAIS-R Vocabulary scores). An additional problem with semantic memory could itself contribute to the neglect of a more syntactic analysis, in that failure to activate lexical-semantic information could trigger the reliance on the use of a semantic stream of analysis strategy, in the absence of full information to perform an algorithm. Further research may benefit from combining more in-depth probes of semantic memory alongside psycholinguistic measures.

Funding Support for this study was provided by the Senate House Central Research Fund and by the British Academy Small Research Grants Fund. P.J. McKenna is supported by CIBERSAM, Spain.

Acknowledgements We are grateful to the patients who participated in the study, and to the anonymous reviewers for their helpful advice on the reporting of the statistical analyses.

Appendix A. Supporting information Supplementary data associated with this article can be found in the online version at http://dx.doi.org/10.1016/j.psychres.2014.02.011. References Almor, A., Kempler, D., MacDonald, M.C., Andersen, E.S., Tyler, L.K., 1999. Why do Alzheimer patients have difficulty with pronouns? Working memory, semantics and reference in comprehension and production in Alzheimer's disease. Brain and Language 67, 202–227. American Psychiatric Association, 1994. Diagnostic and Statistical Manual of Mental Disorders, 4th ed. American Psychiatric Association, Washington, DC. Andreasen, N.C., 1979. Thought, language and communication disorders: I. Clinical assessment, definition of terms and evaluation of their reliability. Archives of General Psychiatry 26, 1315–1321. Andreasen, N.C., 1984. The Scale for the Assessment of positive Symptoms (SAPS). University of Iowa, Iowa City. Andreasen, N.C., Flaum, M., Arndt, S., 1992. The Comprehensive Assessment of Symptoms and History (CASH): an instrument for assessing psychopathology and diagnosis. Archives of General Psychiatry 49, 615–623. Andrews, S., Shelley, A.M., Ward, P.B., Fox, A., Catts, S.V., McConaghy, N., 1993. Event-related potential indices of semantic processing in schizophrenia. Biological Psychiatry 34, 443–458. Bagner, D.M., Melinder, M.R.D., Barch, D.M., 2003. Language comprehension, working memory and working memory deficits in patients with schizophrenia. Schizophrenia Research 60, 299–309. Barrera, A., McKenna, P.J., Berrios, G.E., 2005. Formal thought disorder in schizophrenia: an executive or a semantic deficit? Psychological Medicine 35, 121–132. Besche, C., Passerieux, C., Segui, J., Chevalier, J.F., Widlocher, D., Hardy-Bayle, M.C., 1997. Heterogeneity in cognitive functioning of schizophrenic patients evaluated by a lexical decision task. Psychological Medicine 27, 1295–1302. Bock, J.K., 1982. Toward a cognitive psychology of syntax: Information processing contributions to sentence formulation. Psychological Review 89, 1–47. Cohen, J.D., Barch, D.M., Carter, C., Serven-Schreiber, D., 1999. Context-processing deficits in schizophrenia: Converging evidence from three theoretically motivated cognitive tasks. Journal of Abnormal Psychology 108 (1), 120–133. Cohen, J.D., Servan-Schreiber, D., 1992. Context, cortex, and dopamine: a connectionist approach to behaviour and biology in schizophrenia. Psychological Review 99 (1), 45. Covington, M.A., Congzhou, H., Brown, B., Naci, L., McClaina, J.T., Sirmon Fjordbaka, B., Semple, J., Brown, J., 2005. Schizophrenia and the structure of language: the linguist's view. Schizophrenia Research 77, 85–98. Debruille, J.B., 2007. The N400 potential could index a semantic inhibition. Brain Research Reviews 56, 472–477. Ditman, T., Kuperberg, G.R., 2007. The time course of building discourse coherence in schizophrenia: An ERP investigation. Psychophysiology 44, 991–1001.

175

Docherty, N.M., 2005. Cognitive impairments and disordered speech in schizophrenia: thought disorder, disorganization, and communication failure perspectives. Journal of Abnormal Psychology 114, 269–278. Docherty, N.M., Cohen, A.S., Nienow, T.M., Dinzeo, T.J., Dangelmaier, R.E., 2003. Stability of formal thought disorder and referential communication disturbances in schizophrenia. Journal of Abnormal Psychology 112 (3), 469–475. Docherty, N.M., Evans, I.M., Sledge, W.H., Seibyl, J.P., Krystal, J.H., 1994a. Affective reactivity of language in schizophrenia. Journal of Nervous and Mental Disease 182, 98–102. Docherty, N.M., Gordinier, S.W., 1999. Immediate memory, attention and communication disturbances in schizophrenia patients and their relatives. Psychological Medicine 29, 189–197. Docherty, N.M., Hebert, A.S., 1997. Comparative affective reactivity of different types of communication disturbances in schizophrenia. Journal of Abnormal Psychology 106 (2), 325. Docherty, N.M., Rakfeldt, J., Hawkins, K.A., Hoffman, R.E., Quinlan, D.M., Sledge, W. H., 1996. Working memory, attention, and communication disturbances in schizophrenia. Journal of Abnormal Psychology 105 (2), 212–219. Docherty, N.M., Sledge, W.H., Wexler, B.E., 1994b. Affective reactivity of language in stable schizophrenic outpatients and their parents. Journal of Nervous and Mental Disease 182, 313–318. Garrett, M.F., 1982. Production of speech: observations from normal and pathological language use. In: Ellis, A.W. (Ed.), Normality and Pathology in Cognitive Functions. Academic Press, London. Haddock, G., Wolfenden, M., Lowens, I., Tarrier, N., Bentall, R.P., 1995. Effect of emotional salience on thought disorder in patients with schizophrenia. British Journal of Psychiatry 167, 618–620. Kostova, M., Passerieux, C., Laurent, J.P., Hardy-Bayle, M.C., 2005. N400 anomalies in schizophrenia are correlated with the severity of formal thought disorder. Schizophrenia Research 78, 285–291. Kreher, D.A., Holcomb, P.J., Goff, D., Kuperberg, G.R., 2008. Neural evidence for faster and further automatic spreading activation in schizophrenic thought disorder. Schizophrenia Bulletin 34, 473–482. Kucera, H., Francis, W.N., 1967. Computational Analysis of Present-day American English. Brown University Press, Providence. Kuperberg, G.R., Krecher, D.A., Goff, D., McGuire, P.K., David, A.S., 2006a. Building up linguistic context in schizophrenia: evidence from self-paced reading. Neuropsychology 20 (4), 442–452. Kuperberg, G.R., Kreher, D.A., Ditman, T., 2010. What can event-related potentials tell us about language, and perhaps even thought, in schizophrenia? International Journal of Psychophysiology 75, 66–76. Kuperberg, G.R., McGuire, P.K., David, A.S., 1998. Reduced sensitivity to linguistic context in schizophrenic thought disorder. Evidence from on-line monitoring for words in linguistically anomalous sentences. Journal of Abnormal Psychology 107 (3), 423–434. Kuperberg, G.R., McGuire, P.K., David, A.S., 2000. Sensitivity to linguistic anomalies in spoken sentences: a case study approach to understanding thought disorder in schizophrenia. Psychological Medicine 30, 345–357. Kuperberg, G.R., Sitnikova, T., Goff, D.C., Holcomb, P.J., 2006b. Making sense of sentences in schizophrenia: electrophysiological evidence for abnormal interactions between semantic and syntactic processes. Journal of Abnormal Psychology 115 (2), 251–265. Kutas, M., Federmeier, K.D., 2011. Thirty years and counting: finding meaning in the N400 component of the event-related brain potential (ERP). Annual Review of Psychology 62, 621–647. Levelt, W.J.M., 1989. From Speaking to Articulation. MIT Press, London. Lee, J., Park, S., 2005. Working memory impairments in schizophrenia: A metaanalysis. Journal of Abnormal Psychology 114, 599–611. Monsell, S., 1987. On the Relation Between Lexical Input and Output Pathways for Speech. In: Allport, A., Mackay, D., Prinz, W., Sheerer, E. (Eds.), Language perception and production: Shared mechanisms in listening, speaking, reading and writing, Academic Press, London, pp. 273–311. Nelson, H.E., 1982. The National Adult Reading Test (NART). NFER-Nelson, Windsor. Oh, T.M., McCarthy, R.A., McKenna, P.J., 2002. Is there a schizophasia? A study applying the single case approach to formal thought disorder in schizophrenia. Neurocase 8, 233–244. Pickering, M.J., Garrod, S., 2007. Do people use language production to make predictions during comprehension? Trends in Cognitive Science 11, 105–110. Pickering, M.J., Garrod, S., 2013. An integrated theory of language production and comprehension. Behavioural Brain Science 36, 329–392. Potter, M.C., Lombardi, L., 1990. Regeneration in the short-term recall of sentences. Journal of Memory and Language 29 (6), 633–654. Potter, M.C., Lombardi, L., 1992. The regeneration of syntax in short term memory. Journal of Memory and Language 31, 713–733. Rodriguez-Ferrera, S., McCarthy, R.A., McKenna, P.J., 2001. Language in schizophrenia and its relationship to formal thought disorder. Psychological Medicine 31, 197–205. Scott, S., Johnsrude, I.S., 2003. The neuroanatomical and functional organisation of speech perception. Trends in Neuroscience 26, 100–107. Sitnikova, T., Goff, D., Kuperberg, G.R., 2009. Neurocognitive abnormalities during comprehension of real-world goal-directed behaviors in schizophrenia. Journal of Abnormal Psychology 118, 256–277. Spielberger, C.D., Gorsuch, R.L., Lushene, R.E., 1970. Manual for the State-Trait Anxiety Inventory. Consulting Psychologists Press, Palo Alto.

176

K. Dwyer et al. / Psychiatry Research 216 (2014) 168–176

Spitzer, M., Beuckers, J., Maier, S., Hermle, L., 1994. Contextual insensitivity in schizophrenic patients is due to semantic network pathology: evidence from pauses in spontaneous speech. Language and Speech 37, 171–185. Tai, S., Haddock, G., Bentall, R., 2004. The effects of emotional salience on thought disorder in patients with bipolar disorders. Psychological Medicine 34 (5), 803–809. Tamlyn, D., McKenna, P.J., Mortimer, A., Lund, C.E., Hammond, S., Baddeley, A.D., 1992. Memory impairment in schizophrenia: its extent, affiliations and neuropsychological character. Psychological Medicine 22 (1), 101–115.

Wang, K., Cheung, E.F., Gong, Q.Y., Chan, R.C., 2011. Semantic processing disturbance in patients with schizophrenia: a meta-analysis of the N400 component. PloS ONE 6 (10), e25435. Wechsler, D., 1981. WAIS-R manual; Wechsler Adult Intelligence Scale-Revised. The Psychological Corporation, New York. Wechsler, D., 1997. WAIS-III manual; Wechsler Adult Intelligence Scale, 3rd ed. The Psychological Corporation, San Antonio. Williams, J.M.G., Broadbent, K., 1986. Autobiographical memory in suicide attempters. Journal of Abnormal Psychology 95, 144–149.