Referential Failures and Affective Reactivity of Language in Schizophrenia and Unipolar Depression

Abstract

Reference failures, and their increase in affectively negative conditions (known as affective reactivity of speech), are more frequently observed in schizophrenia patients than in normal controls, but no information is available comparing schizophrenia with depression, ie, a mental disorder closely linked to the concept of affective reactivity. To address this gap in the literature, the present study compared 24 schizophrenia inpatients, 21 unipolar depression inpatients and 21 normal controls. Two 10-minute conversational speech samples (1 on negative and 1 on positive memories) were collected from each patient. The transcripts of the audiotaped interviews were analyzed blindly for frequencies of 6 types of referential failures, employing the Communication Disturbances Index, adapted for use with Italian. The schizophrenia patients made more frequent total reference failures and, specifically, more missing information references than the depression patients. The depression patients made more frequent reference failures than the normal controls, overall, and on most of the specific types of failures. Affective reactivity of speech was observed only for the schizophrenia sample and was greatest for missing information references. This study supports the viability of reference failure analysis as a measure of communication disturbance in a language other than English. The findings indicate that schizophrenia and depression both are associated with high levels of referential failures but that affective reactivity of speech is present only in schizophrenia and not in depression.

Introduction

A series of studies has shown that natural speech contains a number of subtle communication disturbances, called referential failures, which occur more frequently in the speech of schizophrenia patients than in the speech of nonschizophrenic individuals.^1–3 A referential failure involves the use of a reference in speech for which the referent is unclear, ambiguous, or not previously introduced, and it impairs the understandability of the whole message. Referential failures correlate moderately highly with measures of classic positive formal thought disorder in schizophrenia, are quite stable across clinical state in schizophrenia patients, ⁴ and occur with higher than normal frequency in the speech of nonschizophrenic first-degree relatives of patients, ⁵ suggesting that they may reflect vulnerability to schizophrenia to some extent. However, in contrast to the relatives, schizophrenia patients have shown significant increases in frequency of referential failures when asked to speak about affectively negative topics such as stressful memories.⁶ This so-called “affective reactivity of speech” has been found to occur more often in positive than in negative schizophrenia,¹ and patients with deficit syndrome schizophrenia do not differ from healthy controls.⁷

Although referential failures and affective reactivity of speech have been found in a number of studies on schizophrenia patients, little is known about the clinical specificity of referential failures or of affective reactivity of speech. Most studies have compared schizophrenia patients with normal controls or relatives of schizophrenia patients, rather than with patients with other severe mental disorders. In 2 previous studies,^4,8 frequency of referential failures was compared in the speech of patients with acute mania vs acute schizophrenia: During the acute episode, no significant differences on total scores were found, but proportions of the types of unclarity differed among groups,⁸ and only schizophrenia showed stability of referential failures at follow-up.⁴ Both clinical groups had much higher total scores compared with normal controls.⁸ Affective reactivity of speech was not tested in these studies.

Major depression is closely linked conceptually with the idea of affective reactivity. Recent contributions have supported the independent role of daily affective reactivity as a predictor of symptom increases in depression patients.⁹ No information is available on affective reactivity of speech in major depression. Therefore, depression patients were examined as a psychiatric comparison group in the present study of the diagnostic specificity of affective reactivity of speech to schizophrenia. We hypothesized that, notwithstanding the affective reactivity of mood of depression patients, their affective reactivity of speech, if present, would be less pronounced than that of schizophrenia patients. This prediction was based on 2 considerations about affective reactivity of speech: (1) it has been found to be marked in schizophrenia, and (2) it is associated with severity of the positive syndrome, an association which suggests that affective reactivity of speech is a product of overactivation rather than underactivation. As regards the comparison of depressed and healthy subjects, no prediction could be made.

The present study examined whether reference failures and affective reactivity of speech are more highly associated with schizophrenia than with depression. A sample of schizophrenia inpatients was compared with a sample of severely ill inpatients diagnosed with unipolar depression, employing the standard experimental paradigm of Docherty et al.⁸ Speech samples were analyzed with the Communication Disturbances Index (CDI) and each type of disturbance was compared in the positive vs negative conditions. The goals of the study were (a) to test the hypothesis that the speech of schizophrenia patients contains more frequent referential failures than the speech of depression patients and of normal controls, (b) to test the hypothesis that affective reactivity of speech is associated with schizophrenia more than with depression and with normality, (c) to explore the differences between depressed and healthy subjects relating to reference failures and to affective reactivity of speech, (d) to ascertain whether some types of referential failures are more reactive to affective condition than others, and whether the diagnostic groups differ in this regard.

The study also had a goal of independent replication—in a different language—of previous research on affective reactivity of communication disturbances in schizophrenia, which has been done until now only in the US, and primarily by one research team (directed by N.M.D.).

Method

Participants

The following exclusion criteria were adopted: (1) primary language other than Italian; (2) history of severe head trauma; (3) clinical or laboratory indices of central nervous system impairment; (4) substance abuse or dependence in the last year; (5) lack of adequate information about the previous course of the disorder; (6) report of past episodes of particular well-being, possibly suggestive of hypomania; (7) age outside the 18–60 years range; (8) education below 8 years; (9) a Mini-Mental Examination Scale (MMSE) score below 24; and (10) unwillingness to collaborate. Descriptive information is presented in table 1.

Table 1.

Demographic Information for Schizophrenia, Unipolar Depression, and Normal Control Groups

	Schizophrenia	Unipolar Depression	Normal Control
N	24	21	21
Mean age	36.3 ± 11.2	41.9 ± 14.1	38.2 ± 11.9
Mean education	11.4 ± 3.3	10.8 ± 3.8	11.6 ± 3.1
% Male	83.3	71.4	71.4
% Caucasian	100	100	100
GAF scores	41 ± 8	54 ± 10	NA
BPRS	40.6 ± 11.8	30.4 ± 9.2	NA
SAPS	37.9 ± 21.7	NA	NA
SANS	35.9 ± 20.8	NA	NA
HRSD	NA	21.5 ± 6.1	NA
BDI	NA	31.7 ± 10.6	NA
Typical antipsychotics (%)	50	0	NA
Atypical antipsychotics (%)	71	14	NA
Antidepressants (%)	29	100	NA
Mood stabilizers (%)	38	29	NA
Anticholinergic medications (%)	58	0	NA

Note: GAF, Global Assessment of Functioning scale; BPRS, Brief Psychiatric Rating Scale; SAPS, Scale for The Assessment of Positive Symptoms; SANS, Scale for the Assessment of Negative Symptoms; HRSD, Hamilton Rating Scale for Depression; BDI, Beck Depression Inventory.

All patients were tested while on an open psychiatric ward of a university hospital in Rome, Italy; hospitalizations on the ward could not exceed 1 month and could not be made against the will of the patients. Depression patients were hospitalized either because of psychotic features (n = 4, all with delusions of poverty), or severe suicidal risk, or marked psychomotor disturbances. To ensure that all participants were still actively psychotic or depressed but also that they were able to participate, testing took place between the 7th and the 10th day of stay on the ward. As the study was performed on a ward for male patients that occasionally also admitted female patients, most of the participants were men.

Two groups of inpatients diagnosed as meeting Diagnostic and Statistical Manual of Mental Disorder (Fourth Edition) criteria for schizophrenia (n = 27) or major depressive disorder (n = 22), were included in the study. Clinical diagnosis was confirmed by a structured interview (Structured Clinical Interview for DSM-IV Axis I Disorders. Patient Edition). Three schizophrenia and 1 depression patients were subsequently excluded from comparisons because they were not able to comply with the instructions (either they talked only of bad memories, or continued to query about some current disturbances, or were too incoherent to be understood). Therefore, the final clinical groups consisted of 24 schizophrenia and 21 unipolar depression patients. A sample of 21 nonclinical volunteers with no history of psychiatric disorders during adulthood (as checked with the SCID-P) was also included, matched as closely as possible to the demographic characteristics of the depressed group.

Procedure

Speech Samples.

Each participant provided two 10-minute conversational speech samples, which were collected on separate days. On one occasion, participants were requested to describe some affectively positive past events from their own lives, ie, “good memories” or “pleasant, nonstressful times” (the affectively positive condition), and on the other occasion to describe some affectively negative events, that is “bad memories” or “stressful times” (the affectively negative condition). The order of these 2 speech samples was counterbalanced. The interviewer asked relevant, open-ended questions and made comments as necessary to keep the participants talking or to return them to the topics of good or bad memories (for further information on the paradigm, cf. Docherty et al.¹). Speech samples were audiotaped and transcribed for rating. To ensure a high level of accuracy, transcripts were carefully proofread by undergraduate research assistants blind to diagnosis and condition.

Communication Disturbance Ratings.

The transcribed speech samples were rated according to the CDI.⁸ The CDI is related to measures of formal thought disorder,¹⁰ but it focuses explicitly on failures in the communication of meaning in speech rather on underlying thought disorder as such. The total number of words is tallied, instances of communication failures are counted, and the frequency of each type of communication failure per 100 words of speech is calculated. Instances of disturbance are counted only if they render an utterance ambiguous or obscure in meaning. Four types of failures are structural, in the sense that they reflect errors that are interclausal. They are utterances that are unclear in meaning because of problems in the relationships between different segments of the speech. These include (1) the confused reference—unclear because more than one possible referent is present, and the correct choice is not clear (eg, “This (female) sergeant keeps trying to mess with her, and I try to keep her away from her”); (2) the missing information reference, a reference for which the referent has not been provided, ie, is absent (eg, “Sometimes I get mad at the little bullshit game they play”—without mention of anything that might be considered a game); (3) the ambiguous word meaning, a word or phrase with more than one possible definitional meaning, and the correct choice is not clear (eg, “My search for life is neither good nor bad”); and (4) the structural unclarity—a phrase or clause that is unclear in meaning because of grammatical breakdown (eg, “I seen her cry with on the telephone”). The two nonstructural failures are unclear in themselves, without reference to other segments of the speech. They are (5) the vague reference, a word or phrase with impaired meaning due to overgenerality (eg, “I stick to myself till things get too bad, then I leave the room”) and (6) the wrong word reference, an incorrect word choice that renders the intended meaning unclear, ie, a word that is simply incorrect in the context is which it is used (eg, “… . seeing one another express what we got for Christmas”) (for further instances in English, cf. Nienow and Docherty¹¹).

All speech samples from schizophrenia and depression patients were rated blindly by a resident trained in the method. Interrater reliability with a second blind rater was acceptable for all 6 types of communication failures. Intraclass correlations (ICC) were as follows: confused references, ICC = 0.90; missing information references, ICC = 0.88; ambiguous word meaning, ICC = 0.81; structural unclarities, ICC = 0.82; vague references, ICC = 0.76; wrong word references, ICC = 0.90; and total communication failures, ICC = 0.92. The same experimenter rated blindly the normal control protocols; ICC values with a second blind rater were similar. In the following text, the terms referential failures, reference failures, and communication disturbances are used interchangeably.

Analyses

The analysis was done in two parts. First, the 3 groups were compared on total CDI ratings in the affectively positive vs affectively negative conditions to determine whether there were intergroup differences on overall frequencies of reference failures and affective reactivity of speech. Second, affective reactivity of specific types of disturbances was tested in each group, and patterns of reactivity were compared between the groups. All results reported are 2-tailed. In the following, partial and total ratings indicate the ratings of the specific types of referential failures and their sum, respectively.

Results

Total CDI Ratings in the Affectively Positive Vs Negative Condition

Means and standard deviations of total CDI ratings for the 3 groups are presented at the bottom of table 2. The ratings of the 3 groups were compared by means of a 2-way (group × affective condition) repeated measures analysis of variance (ANOVA), with affective condition as the repeated factor. This analysis yielded a significant main effect of group, F_2,63 = 23.88 , P < .001, indicating that the groups differed on total CDI ratings across conditions. The main effect of condition was significant, F_1,63 = 5.55, P =.022, indicating that for the whole sample the speech in the affectively negative condition contained more communication disturbances than in the affectively positive condition. Most importantly, there was a significant group × condition interaction effect, F_2,63 = 6.60, P =.002, indicating significant differences among the groups in degree of affective reactivity of speech. Paired samples t tests showed that only the schizophrenia sample displayed a significant increase of total referential failures (t = −3.23; df = 23; P = .003). The depression (t = 0.73; df = 20; P = .47) and normal (t = −0.75; df = 20; P = .46) groups had stable total scores in the 2 affective conditions. Therefore, the group × condition interaction was attributable to affective reactivity of speech in the schizophrenia group.

Table 2.

Mean Scores of Communication Failures in the Speech of 24 Schizophrenia, 21 Unipolar Depression Inpatients, and 21 Normal Control Subjects in Affectively Positive Vs Affectively Negative Condition

Condition and Group	Mean ± SD for condition
	Positive	Negative
Vague references
Schizophrenia	0.51 ± 0.37	0.62 ± 0.54
Unipolar depression	0.62 ± 0.50	0.53 ± 0.41
Normal control	0.28 ± 0.23	0.37 ± 0.29
Confused references
Schizophrenia	0.05 ± 0.10	0.15 ± 0.20
Unipolar depression	0.06 ± 0.15	0.07 ± 0.12
Normal control	0.08 ± 0.11	0.07 ± 0.10
Missing information references
Schizophrenia	0.36 ± 0.28	0.66 ± 0.53
Unipolar depression	0.30 ± 0.34	0.21 ± 0.24
Normal control	0.10 ± 0.11	0.09 ± 0.09
Ambiguous word meaning
Schizophrenia	0.29 ± 0.30	0.36 ± 0.33
Unipolar depression	0.20 ± 0.21	0.32 ± 0.22
Normal control	0.12 ± 0.16	0.12 ± 0.14
Wrong word references
Schizophrenia	0.18 ± 0.17	0.19 ± 0.19
Unipolar depression	0.27 ± 0.29	0.17 ± 0.22
Normal Control	0.18 ± 0.04	0.01 ± 0.04
Structural unclarities
Schizophrenia	0.40 ± 0.62	0.54 ± 0.50
Unipolar depression	0.22 ± 0.23	0.26 ± 0.21
Normal control	0.01 ± 0.03	0.03 ± 0.08
Total communication disturbance
Schizophrenia	1.78 ± 0.99	2.52 ± 1.12
Unipolar depression	1.66 ± 0.91	1.56 ± 0.91
Normal control	0.61 ± 0.35	0.69 ± 0.35

Note: Communication disturbance scores are presented as instances per 100 words of speech.

The significant interaction was followed up with 2-group comparisons. As the main goal of the study was to compare schizophrenia and depression patients, these 2 groups were compared first. Results were as follows: main effect for group, F_1,43 = 4.27, P = .045; main effect for affective condition, F_1,43 = 5.16, P = .028; interaction effect, F_1,43 = 9.12, P = .004. These results support the hypothesis that the speech of schizophrenia patients is more reactive than that of depression patients. However, there were several potentially confounding variables on which these 2 groups differed in this sample. In an effort to control statistically for these variables to the extent possible (but see Miller ¹¹), the analysis was also run with the following covariates: psychopathological severity assessed by the Brief Psychiatric Rating Scale (BPRS), time elapsed from first hospitalization, number of hospitalizations, cognitive levels as measured by the MMSE, current use of anticholinergic drugs, and diagnosis of current major depression in the schizophrenia patients. With all the covariates included simultaneously in the analysis, the interaction effect remained significant, F_1,37 = 9.93, P = .003. When entered as single covariates, none of the above variables led to a nonsignificant interaction effect.

The ratings of the depression and normal control groups were compared next. This analysis yielded a significant main effect of group, F_1,40 = 25.11, P < .001, with no significant main effect of condition, F_1,40 = .014, ns, or interaction effect, F_1,40= 1.08, ns. The significant main effect for group supports the hypothesis that the speech of inpatients with severe depression contains more frequent reference failures than that of normal controls. As regards the comparison of schizophrenia and normal subjects, the analysis yielded a significant main effect of group, F1,43 = 55.87, P < .001, with a significant main effect of condition, F_1,43=9.54, P = .004, and a significant interaction effect, F_1,43 = 6.06, P = .018. Thus, compared with normal subjects, schizophrenia patients had more reference failures and this difference became more marked in the negative affective condition.

Reactivity of Specific Types of Communication Disturbances

Means and standard deviations for each type of disturbance in the speech of each group in the affectively positive and affectively negative conditions are presented in Table 2. To examine differences between the 3 groups in affective reactivity of the 6 types of communication disturbances measured by the CDI, a 3-group repeated measure multivariate analysis of variance (MANOVA) was computed. Affective condition was the repeated measure and the 6 types of disturbances were the multiple dependent variables. The main effect of group was significant, Wilk’s λ = .43, F_12,116 = 5.05, P < .001. Although the main effect of condition was not significant, Wilk’s λ = .85, F_6,58 = 1.74, P = .13, the overall interaction effect was significant, Wilk’s λ = .65, F_12,116 = 2.28 , P = .012. Univariate F tests found that the main effect of group was significant for all types of reference failures, with the exception of confused references; a significant group × condition interaction was observed only for missing references, F_2,63 = 8.78, P < .001.

As regards the comparison of schizophrenia and depression patients, the main effect of group was significant, Wilk’s λ = .72, F_6,38 = 2.42 , P = .044 . Although the main effect of condition was not significant, Wilk’s λ = .76, F_6,38 = 2.01, P = .088 , the overall interaction effect was significant, Wilk’s λ= .70, F_6,38 = 2.71 , P = .027. Univariate F tests found that the main effect of group was significant only for missing information references (P = .009), indicating that only this language variable was more represented in the schizophrenia group, and the group × condition interaction effect was significant only for missing information references, F_1,43 = 11.16, P = .002.

As regards the comparison of depression and normal control subjects, a similar statistical analysis yielded a significant main effect of group, Wilk’s λ = .48, F_6,35 = 6.45, P < .001, but no effect of condition, Wilk’s λ = .77, F_6,35 = 1.76, ns, or overall interaction effect, Wilk’s λ = .77, F_6,35 = 1.78, ns. Univariate F tests (1, 40) found that the effect of group was significant for vague references (P = .012), Missing information references (P = .008), ambiguous word meanings (P = .002), wrong word references (P < .001), and structural unclarities (P < .001), but not for confused references. Neither the effect of affective condition nor group × condition interaction effects were significant for any of the 6 variables.

Finally, concerning the comparison of schizophrenia and normal subjects, the main effect of group was significant, Wilk’s λ = .41, F_6,38 = 9.16, P < .001. The main effect of condition was not significant, Wilk’s λ =.75, F_6,38 = 2.13, P = .07, nor was the interaction effect, Wilk’s λ = .75, F_6,38 = 2.09, P = .078. Univariate F tests found that the main effect of group was significant for all types of reference failure, with the exception of confused references, and the group × condition interaction effect was significant only for missing information references, F_1,43 = 9.71, P = .003.

Within-Group Analyses

To examine patterns of affective reactivity of speech in each group, we computed single-group ANOVAs of total CDI ratings and of each specific type of disturbance. We used a significance criterion of P < .01 for these analyses because there were many comparisons.

Schizophrenia.

A single-group repeated measure ANOVA examining total CDI ratings indicated that the speech of schizophrenia patients contained more frequent instances of communication failures in the affectively negative condition than in the affectively positive condition, F_1,22 = 10.40, P = .004. In single-group analyses of specific types of disturbances in schizophrenia patients’ speech, frequency of missing information references increased significantly in response to negative affect, F_1,22 = 11.37, P = .003. Frequencies of the other types of communication disturbances did not increase significantly.

Unipolar Depression.

In a single-group analysis, the speech of depression patients did not differ on total CDI ratings between affective conditions, F_1,19 = .53, P = .47. None of the 6 types of communication disturbance increased significantly in response to negative affect.

Normal Control Group.

In a single-group analysis, the speech of normal control subjects did not differ on total CDI ratings between affective conditions, F_1,19 = .57 , P = .46. No significant increase of frequencies in response to the negative condition was observed for any of the referential failures.

Discussion

The speech of schizophrenia patients showed higher levels of referential disturbance compared with that of depression patients, and the depression patients showed higher levels than the normal controls. The schizophrenia patients made more frequent reference failures in the affectively negative condition than in the affectively positive condition, whereas neither the depression patients nor the normal controls showed any effect of affective condition. These findings suggest that affective reactivity of speech is associated with schizophrenia and is not only a correlate of a severe mental disorder or of high baseline levels of reference failures. They also indicate that affective reactivity of mood, a fundamental psychopathological feature of depression, is not conducive to affective reactivity of speech. The differences between the schizophrenia and depression groups in frequencies and affective reactivity of reference failures were not diminished by the inclusion of several clinical covariates in the statistical analysis, indicating that variables other than diagnosis were not likely to have been responsible for the effects. Unfortunately, we did not measure levels of depression in the schizophrenia sample and could therefore only covary out the current presence/absence of major depression; a covariation using ratings of level of depression might have been more informative. The same limitation holds true for the levels of adaptation; as we lacked a measure of adaptation independent from symptomatology, it was meaningless to enter adaptation as a covariate, as both clinical groups were composed of severe inpatients.

Previous work⁶ found that total ratings of referential failures in the 2 affective conditions were higher in nonschizophrenic first-degree relatives of schizophrenia patients compared with normal controls, but that affective reactivity of speech was not present in the relatives. This suggests that a general elevation of referential failures may be a component of vulnerability to schizophrenia, but that a high level of affective reactivity of speech may be a marker of manifest schizophrenia. The present results are congruent with this idea.⁷

Statistical analyses further pointed to a second relevant issue: among the 6 variants of referential failures considered, the missing information reference was the only one that was significantly more highly associated with schizophrenia than depression across affective conditions. It also was the only one that showed affective reactivity in the schizophrenia patients. The primacy of this variant is at the same time incongruent and congruent with previous data. Two earlier studies with the CDI^6,13 found that 3 of the other types of communication disturbances increased significantly in the negative condition, ie, vague references, confused references, and ambiguous word meaning. Missing information references showed only slight and nonsignificant reactivity in these studies; linguistic peculiarities of the Italian language or subtle differences of scoring between research groups may explain this difference between studies.

On the other hand, the missing information reference has received growing interest in recent years, in a direction that fully converges with the present data indicating its particular association with schizophrenia. A factor-analytic study suggested that the neurocognitive correlates of the missing information reference may be different from those of the other types of referential failure.¹⁴ Most of the other types loaded on a factor associated with weaknesses in conceptual sequencing and abstraction, whereas the missing information reference loaded on a factor associated with impaired memory processes. Missing information references have been hypothesized to occur when the subject does not recall what information he has and has not actually provided to the listener, ie, cannot distinguish in memory between the material he has previously presented in speech vs only thought about. This hypothesis was supported by a study that found significant and specific correlations of missing information references with poor source discrimination on an internal/external, say-think, reality monitoring task, and specifically, with more frequent think-report-say errors on the task.¹¹ None of the other referential variables showed significant correlations with this type of error. Furthermore, 2 other types of memory—immediate auditory and working memory—were not related to missing information references, suggesting a rather specific relationship between missing information references and deficits of source monitoring. Deficits in say–think source discrimination are known to be substantial in schizophrenia, in both acute patients¹⁵ and relatively asymptomatic outpatients,¹⁶ and have been associated with verbal hallucinations, one of the key positive symptoms of schizophrenia.¹⁷ A unique role of missing information references in schizophrenia is also suggested by the results of a twin study, in which the missing information reference was the only type of communication disturbance made more frequently by the monozygotic nonschizophrenic co-twins of schizophrenia patients compared with the dizygotic nonschizophrenic co-twins of patients.¹⁸ When the findings of the present study are integrated with those of these neurocognitive and twin investigations, it becomes possible to suggest that the missing information reference is the communication disturbance most closely associated both with the vulnerability and with the illness periods of schizophrenia.

Regarding limitations of the study, instruments such as the CDI are not easily employed by researchers who are not native speakers of the language of the experimental probands. Therefore, we did not try to compare our scorings with those of American experts; although we translated the CDI manual, checked the translation with a bilingual expert, closely followed the instructions, and reached good interrater reliabilities within our group, small differences from the American groups could not be wholly avoided. Furthermore, languages with different grammars, syntaxes, modes of expression, and cultural contexts may engender different types of communication failures: we suspect that, compared with English, Italian is more conducive to vague references and less to confused references. Another limitation of the study is the possibility that bad memories, being mood congruent with depression, may be less disruptive for depression patients and therefore lead to a lack of affective reactivity of speech. Further study on reference failures in remitted unipolar patients could clarify the topic. Finally, we did not rate how bad the bad memories were and thus we could not control whether the groups with higher scores of reference disturbances were simply those with the worst memories; a quantification of the valence of memories is certainly advisable for future research with the paradigm we employed.

Conclusions

The present study found that communication disturbances, as rated with the CDI, are more frequent in schizophrenia than in unipolar depression (and more frequent in the latter compared with a normal control group) and that affective reactivity of speech characterizes schizophrenia patients but not depression patients or normal controls. The missing information reference was the only type of referential failure that significantly differentiated between clinical groups and, in contrast to previous findings, was the only one that was reactive to affect in the schizophrenia group. Further research is needed to determine the clinical and neurocognitive correlates of the missing information reference, as it seems to be a manifestation of a core, specifically schizophrenic, cognitive deficit.