Grammatical processing in schizophrenia: evidence from morphology

Semantic coherence markers: The contribution of perplexity metrics

Devising automatic tools to assist specialists in the early detection of mental disturbances and psychotic disorders is to date a challenging scientific problem and a practically relevant activity. In this work we explore how language models (that are probability distributions over text sequences) can be employed to analyze language and discriminate between mentally impaired and healthy subjects. We have preliminarily explored whether perplexity can be considered a reliable metrics to characterize an individual's language. Perplexity was originally conceived as an information-theoretic measure to assess how much a given language model is suited to predict a text sequence or, equivalently, how much a word sequence fits into a specific language model. We carried out an extensive experimentation with healthy subjects, and employed language models as diverse as N-grams - from 2-grams to 5-grams - and GPT-2, a transformer-based language model. Our experiments show that irrespective of the complexity of the employed language model, perplexity scores are stable and sufficiently consistent for analyzing the language of individual subjects, and at the same time sensitive enough to capture differences due to linguistic registers adopted by the same speaker, e.g., in interviews and political rallies. A second array of experiments was designed to investigate whether perplexity scores may be used to discriminate between the transcripts of healthy subjects and subjects suffering from Alzheimer Disease (AD). Our best performing models achieved full accuracy and F-score (1.00 in both precision/specificity and recall/sensitivity) in categorizing subjects from both the AD class, and control subjects. These results suggest that perplexity can be a valuable analytical metrics with potential application to supporting early diagnosis of symptoms of mental disorders.

Processing Argument Structure and Syntactic Complexity in People with Schizophrenia Spectrum Disorders

INTRODUCTION

Deficits in language comprehension and production have been repeatedly observed in Schizophrenia Spectrum Disorders (SSD). However, the characterization of the language profile of this population is far from complete, and the relationship between language deficits, impaired thinking and cognitive functions is widely debated.

OBJECTIVE

The aims of the present study were to assess production and comprehension of verbs with different argument structures, as well as production and comprehension of sentences with canonical and non-canonical word order in people with SSD. In addition, the study investigated the relationship between language deficits and cognitive functions.

METHODS

Thirty-four participants with a diagnosis of SSD and a group of healthy control participants (HC) were recruited and evaluated using the Italian version of the Northwestern Assessment of Verbs and Sentences (NAVS, Cho-Reyes & Thompson, 2012; Barbieri et al., 2019).

RESULTS

Results showed that participants with SSD were impaired - compared to HC - on both verb and sentence production, as well as on comprehension of syntactically complex (but not simple) sentences. While verb production was equally affected by verb-argument structure complexity in both SSD and HC, sentence comprehension was disproportionately more affected by syntactic complexity in SSD than in HC. In addition, in the SSD group, verb production deficits were predicted by performance on a measure of visual attention, while sentence production and comprehension deficits were explained by performance on measures of executive functions and working memory, respectively.

DISCUSSION

Our findings support the hypothesis that language deficits in SSD may be one aspect of a more generalized, multi-domain, cognitive impairment, and are consistent with previous findings pointing to reduced inter- and intra-hemispheric connectivity as a possible substrate for such deficits. The study provides a systematic characterization of lexical and syntactic deficits in SSD and demonstrates that psycholinguistically-based assessment tools may be able to capture language deficits in this population.

Differences in fractal patterns and characteristic periodicities between word salads and normal sentences: Interference of meaning and sound

Fractal dimensions and characteristic periodicities were evaluated in normal sentences, computer-generated word salads, and word salads from schizophrenia patients, in both Japanese and English, using the random walk patterns of vowels.　In normal sentences, the walking curves were smooth with gentle undulations, whereas computer-generated word salads were rugged with mechanical repetitions, and word salads from patients with schizophrenia were unreasonably winding with meaningless repetitive patterns or even artistic cohesion. These tendencies were similar in both languages. Fractal dimensions between normal sentences and word salads of schizophrenia were significantly different in Japanese [1.19 ± 0.09 (n = 90) and 1.15 ± 0.08 (n = 45), respectively] and English [1.20 ± 0.08 (n = 91), and 1.16 ± 0.08 (n = 42)] (p < 0.05 for both). Differences in long-range (>10) periodicities between normal sentences and word salads from schizophrenia patients were predominantly observed at 25.6 (p < 0.01) in Japanese and 10.7 (p < 0.01) in English. The differences in fractal dimension and characteristic periodicities of relatively long-range (>10) presented here are sensitive to discriminate between schizophrenia and healthy mental state, and could be implemented in social robots to assess the mental state of people in care.

Anomalies in language as a biomarker for schizophrenia

PURPOSE OF REVIEW

After more than a century of neuroscience research, reproducible, clinically relevant biomarkers for schizophrenia have not yet been established. This article reviews current advances in evaluating the use of language as a diagnostic or prognostic tool in schizophrenia.

RECENT FINDINGS

The development of computational linguistic tools to quantify language disturbances is rapidly gaining ground in the field of schizophrenia research. Current applications are the use of semantic space models and acoustic analyses focused on phonetic markers. These features are used in machine learning models to distinguish patients with schizophrenia from healthy controls or to predict conversion to psychosis in high-risk groups, reaching accuracy scores (generally ranging from 80 to 90%) that exceed clinical raters. Other potential applications for a language biomarker in schizophrenia are monitoring of side effects, differential diagnostics and relapse prevention.

SUMMARY

Language disturbances are a key feature of schizophrenia. Although in its early stages, the emerging field of research focused on computational linguistics suggests an important role for language analyses in the diagnosis and prognosis of schizophrenia. Spoken language as a biomarker for schizophrenia has important advantages because it can be objectively and reproducibly quantified. Furthermore, language analyses are low-cost, time efficient and noninvasive in nature.

The Neurocognition of Developmental Disorders of Language

Developmental disorders of language include developmental language disorder, dyslexia, and motor-speech disorders such as articulation disorder and stuttering. These disorders have generally been explained by accounts that focus on their behavioral rather than neural characteristics; their processing rather than learning impairments; and each disorder separately rather than together, despite their commonalities and comorbidities. Here we update and review a unifying neurocognitive account—the Procedural circuit Deficit Hypothesis (PDH). The PDH posits that abnormalities of brain structures underlying procedural memory (learning and memory that rely on the basal ganglia and associated circuitry) can explain numerous brain and behavioral characteristics across learning and processing, in multiple disorders, including both commonalities and differences. We describe procedural memory, examine its role in various aspects of language, and then present the PDH and relevant evidence across language-related disorders. The PDH has substantial explanatory power, and both basic research and translational implications.

Modeling Incoherent Discourse in Non-Affective Psychosis

BACKGROUND

Computational linguistic methodology allows quantification of speech abnormalities in non-affective psychosis. For this patient group, incoherent speech has long been described as a symptom of formal thought disorder. Our study is an interdisciplinary attempt at developing a model of incoherence in non-affective psychosis, informed by computational linguistic methodology as well as psychiatric research, which both conceptualize incoherence as associative loosening. The primary aim of this pilot study was methodological: to validate the model against clinical data and reduce bias in automated coherence analysis.

METHODS

Speech samples were obtained from patients with a diagnosis of schizophrenia or schizoaffective disorder, who were divided into two groups of n = 20 subjects each, based on different clinical ratings of positive formal thought disorder, and n = 20 healthy control subjects.

RESULTS

Coherence metrics that were automatically derived from interview transcripts significantly predicted clinical ratings of thought disorder. Significant results from multinomial regression analysis revealed that group membership (controls vs. patients with vs. without formal thought disorder) could be predicted based on automated coherence analysis when bias was considered. Further improvement of the regression model was reached by including variables that psychiatric research has shown to inform clinical diagnostics of positive formal thought disorder.

CONCLUSIONS

Automated coherence analysis may capture different features of incoherent speech than clinical ratings of formal thought disorder. Models of incoherence in non-affective psychosis should include automatically derived coherence metrics as well as lexical and syntactic features that influence the comprehensibility of speech.

A Comprehensive Review of Computational Methods for Automatic Prediction of Schizophrenia With Insight Into Indigenous Populations

Psychiatrists rely on language and speech behavior as one of the main clues in psychiatric diagnosis. Descriptive psychopathology and phenomenology form the basis of a common language used by psychiatrists to describe abnormal mental states. This conventional technique of clinical observation informed early studies on disturbances of thought form, speech, and language observed in psychosis and schizophrenia. These findings resulted in language models that were used as tools in psychosis research that concerned itself with the links between formal thought disorder and language disturbances observed in schizophrenia. The end result was the development of clinical rating scales measuring severity of disturbances in speech, language, and thought form. However, these linguistic measures do not fully capture the richness of human discourse and are time-consuming and subjective when measured against psychometric rating scales. These linguistic measures have not considered the influence of culture on psychopathology. With recent advances in computational sciences, we have seen a re-emergence of novel research using computing methods to analyze free speech for improving prediction and diagnosis of psychosis. Current studies on automated speech analysis examining for semantic incoherence are carried out based on natural language processing and acoustic analysis, which, in some studies, have been combined with machine learning approaches for classification and prediction purposes.

The language profile of formal thought disorder

Formal thought disorder (FTD) is clinically manifested as disorganized speech, but there have been only few investigations of its linguistic properties. We examined how disturbance of thought may relate to the referential function of language as expressed in the use of noun phrases (NPs) and the complexity of sentence structures. We used a comic strip description task to elicit language samples from 30 participants with schizophrenia (SZ), 15 with moderate or severe FTD (SZ + FTD), and 15 minimal or no FTD (SZ−FTD), as well as 15 first-degree relatives of people with SZ (FDRs) and 15 neurotypical controls (NC). We predicted that anomalies in the normal referential use of NPs, sub-divided into definite and indefinite NPs, would identify FTD; and also that FTD would also be linked to reduced linguistic complexity as specifically measured by the number of embedded clauses and of grammatical dependents. Participants with SZ + FTD produced more referential anomalies than NC and produced the fewest definite NPs, while FDRs produced the most and thus also differed from NC. When referential anomalies were classed according to the NP type in which they occurred, the SZ + FTD group produced more anomalies in definite NPs than NC. Syntactic errors did not distinguish groups, but the SZ + FTD group exhibited significantly less syntactic complexity than non-SZ groups. Exploratory regression analyses suggested that production of definite NPs distinguished the two SZ groups. These results demonstrate that FTD can be identified in specific grammatical patterns which provide new targets for detection, intervention, and neurobiological studies.

Deficits in nominal reference identify thought disordered speech in a narrative production task

Formal thought disorder (TD) is a neuropathology manifest in formal language dysfunction, but few behavioural linguistic studies exist. These have highlighted problems in the domain of semantics and more specifically of reference. Here we aimed for a more complete and systematic linguistic model of TD, focused on (i) a more in-depth analysis of anomalies of reference as depending on the grammatical construction type in which they occur, and (ii) measures of formal grammatical complexity and errors. Narrative speech obtained from 40 patients with schizophrenia, 20 with TD and 20 without, and from 14 healthy controls matched on pre-morbid IQ, was rated blindly. Results showed that of 10 linguistic variables annotated, 4 showed significant differences between groups, including the two patient groups. These all concerned mis-uses of noun phrases (NPs) for purposes of reference, but showed sensitivity to how NPs were classed: definite and pronominal forms of reference were more affected than indefinite and non-pronominal (lexical) NPs. None of the measures of formal grammatical complexity and errors distinguished groups. We conclude that TD exhibits a specific and differentiated linguistic profile, which can illuminate TD neuro-cognitively and inform future neuroimaging studies, and can have clinical utility as a linguistic biomarker.

Language abnormality in deaf people with schizophrenia: a problem with classifiers

INTRODUCTION

Although there is evidence for language abnormality in schizophrenia, few studies have examined sign language in deaf patients with the disorder. This is of potential interest because a hallmark of sign languages is their use of classifiers (semantic or entity classifiers), a reference-tracking device with few if any parallels in spoken languages. This study aimed to examine classifier production and comprehension in deaf signing adults with schizophrenia.

METHOD

Fourteen profoundly deaf signing adults with schizophrenia and 35 age- and IQ-matched deaf healthy controls completed a battery of tests assessing classifier and noun comprehension and production.

RESULTS

The patients showed poorer performance than the healthy controls on comprehension and production of both nouns and entity classifiers, with the deficit being most marked in the production of classifiers. Classifier production errors affected handshape rather than other parameters such as movement and location.

CONCLUSIONS

The findings suggest that schizophrenia affects language production in deaf patients with schizophrenia in a unique way not seen in hearing patients.

Verbal working memory in schizophrenia: The role of syntax in facilitating serial recall

BACKGROUND

Deficits in verbal working memory (VWM) have consistently been observed in schizophrenia, ranging from impairments in capacity, encoding, to irregular semantic organisation. However, syntactic deficits are less well-characterised, despite its crucial role in language construction. This study examines the role of simple syntactic structure (basis of the "sentence superiority effect") in VWM of patients with psychotic disorders.

METHODS

Patients with schizophrenia-spectrum disorders (n=40) and healthy controls matched on age, sex and education (n=40) were administered an auditory serial recall task containing word lists with low semantic coherence and either syntactically familiar structure (noun-verb-noun sequence) or syntactically unfamiliar structure. Other neurocognitive measures, symptoms and social functioning of patients were also assessed.

RESULTS

A 4-way analysis of variance (group×version×list type×serial position) indicated that patients had significantly worse performance overall, suggesting a generalised verbal memory impairment. In addition, a significant interaction was found for list type and Group, demonstrating that healthy controls, but not patients, had superior performance in syntactically familiar word lists. A subgroup analysis of high-performing patients revealed that the interaction was not an artefact of poor verbal memory, but a selective deficit in syntactic facilitation.

CONCLUSIONS

Our findings may suggest segregated mechanisms for maintenance and computational aspects of VWM, and show that even simple syntactic structure facilitates recall of syntactically unfamiliar words lists. Additionally, schizophrenic patients show difficulty utilising syntactic information, which highlights the need to understand the neuropsychological basis of working memory and linguistic impairments in psychosis.

Action (verb) fluency deficits in schizophrenia spectrum disorders: linking language, cognition and interpersonal functioning

Deficits in action (verb) fluency have previously been reported in schizophrenia spectrum disorders. The degree to which this reflects difficulties generating verbs in different semantic categories is unknown. Here, action fluency responses of 46 patients with schizophrenia spectrum disorders and 76 healthy controls were classified as action or mental state verbs, using well-established taxonomies. The word length, frequency, age of acquisition, valence and concreteness of the verbs produced were also examined. Participants also completed measures of cognitive function, and clinical symptoms. Independent inter-rater agreement of semantic categorization was high. The percentage of action verbs produced was significantly lower in patients than controls, whilst the percentage of mental state verbs produced did not differ. Patients' action verbs were: significantly less concrete; positively correlated with memory and intelligence; and negatively correlated with interpersonal symptoms. Impaired action verb, but intact mental state verb generation is consistent with the neural separability of these processes.

Examining lexical processing with two word tasks using the schizotypy analogue

Semantic processing is impaired in schizophrenia, and is often assessed using word-based semantic tasks that additionally require lexical processing for accurate performance. This study aimed to examine lexical processing in relation to psychosis proneness. 61 individuals (mean age=22.92, SD=2.67) completed a schizotypy questionnaire and two lexical processing tasks (recognition and production). Results revealed no relationship between increasing schizotypy scores and overall performance on either lexical processing task. Lexical processing performance is not related to psychosis-proneness. This is discussed in relation to schizophrenia semantic processing research.

Characterising receptive language processing in schizophrenia using word and sentence tasks

INTRODUCTION

Language dysfunction is proposed to relate to the speech disturbances in schizophrenia, which are more commonly referred to as formal thought disorder (FTD). Presently, language production deficits in schizophrenia are better characterised than language comprehension difficulties. This study thus aimed to examine three aspects of language comprehension in schizophrenia: (1) the role of lexical processing, (2) meaning attribution for words and sentences, and (3) the relationship between comprehension and production.

METHODS

Fifty-seven schizophrenia/schizoaffective disorder patients and 48 healthy controls completed a clinical assessment and three language tasks assessing word recognition, synonym identification, and sentence comprehension. Poorer patient performance was expected on the latter two tasks.

RESULTS

Recognition of word form was not impaired in schizophrenia, indicating intact lexical processing. Whereas single-word synonym identification was not significantly impaired, there was a tendency to attribute word meanings based on phonological similarity with increasing FTD severity. Importantly, there was a significant sentence comprehension deficit for processing deep structure, which correlated with FTD severity.

CONCLUSIONS

These findings established a receptive language deficit in schizophrenia at the syntactic level. There was also evidence for a relationship between some aspects of language comprehension and speech production/FTD. Apart from indicating language as another mechanism in FTD aetiology, the data also suggest that remediating language comprehension problems may be an avenue to pursue in alleviating FTD symptomatology.

Bilingualism and schizophrenia

Although a bilingual advantage has been described for neurodegenerative disease in general, it is not known whether such an advantage could accrue to individuals suffering from schizophrenia, since language networks are known to be disrupted in this condition. The aim of this minireview was to scan the existing literature to determine: (1) whether individuals with schizophrenia are able to learn a second language as adults; (2) whether clinical assessment, both for the purpose of accurate diagnosis and for the prediction of treatment response, should be carried out in both languages in bilinguals with schizophrenia; (3) whether psychotherapy in schizophrenia is affected by bilingualism; and (4) whether speaking a second language improves outcome in schizophrenia. The literature to date is too sparse to make definitive statements, but: (1) individuals with schizophrenia appear to be capable of learning a new languages as adults; and (2) it is possible that teaching a foreign language may serve as a form of cognitive rehabilitation for this condition. This literature review recommends research into the effects of bilingualism on the outcome of schizophrenia. Included in this review is a retrospective pilot study conducted in Canada, which suggests that employment opportunities for patients with schizophrenia are improved when they speak more than one language. This is important to note because employment is generally problematic in the context of schizophrenia while, at the same time, the ability to obtain work contributes significantly to quality of life.

The linguistics of schizophrenia: thought disturbance as language pathology across positive symptoms

We hypothesize that linguistic (dis-)organization in the schizophrenic brain plays a more central role in the pathogenesis of this disease than commonly supposed. Against the standard view, that schizophrenia is a disturbance of thought or selfhood, we argue that the origins of the relevant forms of thought and selfhood at least partially depend on language. The view that they do not is premised by a theoretical conception of language that we here identify as 'Cartesian' and contrast with a recent 'un-Cartesian' model. This linguistic model empirically argues for both (i) a one-to-one correlation between human-specific thought or meaning and forms of grammatical organization, and (ii) an integrative and co-dependent view of linguistic cognition and its sensory-motor dimensions. Core dimensions of meaning mediated by grammar on this model specifically concern forms of referential and propositional meaning. A breakdown of these is virtually definitional of core symptoms. Within this model the three main positive symptoms of schizophrenia fall into place as failures in language-mediated forms of meaning, manifest either as a disorder of speech perception (Auditory Verbal Hallucinations), abnormal speech production running without feedback control (Formal Thought Disorder), or production of abnormal linguistic content (Delusions). Our hypothesis makes testable predictions for the language profile of schizophrenia across symptoms; it simplifies the cognitive neuropsychology of schizophrenia while not being inconsistent with a pattern of neurocognitive deficits and their correlations with symptoms; and it predicts persistent findings on disturbances of language-related circuitry in the schizophrenic brain.

Abnormal dynamics of language in schizophrenia

Language could be conceptualized as a dynamic system that includes multiple interactive levels (sub-lexical, lexical, sentence, and discourse) and components (phonology, semantics, and syntax). In schizophrenia, abnormalities are observed at all language elements (levels and components) but the dynamic between these elements remains unclear. We hypothesize that the dynamics between language elements in schizophrenia is abnormal and explore how this dynamic is altered. We, first, investigated language elements with comparable procedures in patients and healthy controls. Second, using measures of reaction time, we performed multiple linear regression analyses to evaluate the inter-relationships among language elements and the effect of group on these relationships. Patients significantly differed from controls with respect to sub-lexical/lexical, lexical/sentence, and sentence/discourse regression coefficients. The intercepts of the regression slopes increased in the same order above (from lower to higher levels) in patients but not in controls. Regression coefficients between syntax and both sentence level and discourse level semantics did not differentiate patients from controls. This study indicates that the dynamics between language elements is abnormal in schizophrenia. In patients, top-down flow of linguistic information might be reduced, and the relationship between phonology and semantics but not between syntax and semantics appears to be altered.

An examination of the associations among multiple memory systems, past tense, and vocabulary in typically developing 5-year-old children

PURPOSE

Considerable research has investigated the role of verbal working memory in language development in children with and without language problems. Much less is currently known about the relationship between language and the declarative and procedural memory systems. This study examined whether these 2 memory systems were related to typically developing children's past tense and lexical knowledge.

METHOD

Fifty-eight typically developing children approximately 5 years of age completed a battery of linguistic and nonlinguistic tasks, including tests of vocabulary, past tense production, and procedural and declarative memory.

RESULTS

The results showed that declarative and procedural memory were not correlated with either regular or irregular past tense use. A significant correlation was observed between declarative memory and vocabulary.

CONCLUSIONS

The results of the study were not consistent with the view that the declarative and procedural memory systems support children's use of the regular and irregular past tense. However, evidence was found suggesting that declarative memory supports vocabulary in this age group.

Action (verb) fluency in schizophrenia: getting a grip on odd speech

BACKGROUND

Formal thought disorder (TD) is a key symptom of schizophrenia with a significant impact on interpersonal relationships. Current cognitive models emphasize disordered language functioning and abnormalities accessing semantic representations. The cortical mechanisms for language and motor function are closely linked, hence action-related language may be impaired in TD, yet existing studies have focussed exclusively on object (noun) rather than action (verb) semantics.

METHOD

In order to examine this issue both action (verb) and traditional semantic (tools, fruits, musical instruments) and phonological (FAS) fluency tasks were completed by individuals with schizophrenia (N=53) and healthy controls (N=69). Fluency performance was measured as the total number of correct words generated in 60s. The Schizotypal Personality Questionnaire (SPQ) was used to index odd and disorganized speech, as well as positive and negative symptoms.

RESULTS

Fluency on all tasks was impaired in schizophrenia, compared to controls, with a similar effect size. Within the schizophrenia group Odd Speech was correlated with poor fluency for actions, tools and musical instruments but not fruit or phonological fluency. These action-related fluency deficits were also correlated with Constricted Affect and Social Anxiety but not with Unusual Perceptions/Odd Beliefs.

CONCLUSION

These results point to a unique connection and possible common aetiology between action fluency and odd speech in schizophrenia rather than a general impairment in language/executive functions common to fluency tasks. The findings provide the first evidence of a specific role of action-based language production deficits in TD together with a joint effect on social interaction skills.