Language disturbances in schizophrenia: the relation with antipsychotic medication

Language disturbance is a hallmark of schizophrenia, affecting semantics, syntax, phonology, and prosody, and is linked to poorer social functioning and quality of life. Antipsychotics, while effective for positive symptoms, may influence language via dopamine receptor blockade in prefrontal (cognitive) and striatal (motor) systems. High D2 receptor occupancy drugs (e.g., haloperidol, risperidone, aripiprazole) might exacerbate negative language features such as alogia, slower articulation, and increased pausing more than low D2R occupancy drugs (e.g., clozapine, quetiapine, olanzapine). The study’s hypothesis was that patients using high D2R occupancy antipsychotics would show more severe language disturbances than those using low D2R occupancy drugs, beyond illness effects observed relative to healthy controls.

Prior work shows schizophrenia is associated with reduced speech rate, decreased proportion of time speaking, increased pausing, reduced syntactic complexity, shorter sentences with fewer embeddings, limited lexical diversity, and word-finding difficulties leading to disfluencies. Antipsychotics can induce extrapyramidal side effects that slow and soften speech and may worsen cognitive aspects of language via prefrontal hypodopaminergia. Not all antipsychotics have equal D2R binding; tighter-binding agents may have stronger effects on language production. There is mixed evidence on whether antipsychotics improve coherence or induce poverty of speech, and speech impacts have been noted across disorders and treatments (e.g., Parkinson’s dopamine therapy improves speech tempo; antipsychotics can reduce intelligibility). These observations motivated examining medication class (high vs low D2R occupancy) in relation to language features known to be disturbed in schizophrenia.

Design: Cross-sectional observational study comparing language measures among three groups: schizophrenia spectrum patients on high D2R occupancy antipsychotics, patients on low D2R occupancy antipsychotics, and healthy controls (HC). Participants: 81 adults (41 patients; 40 HCs), native Dutch speakers; patients met DSM-IV 295.x or 298.9 diagnoses confirmed via CASH or MINI-Plus. Exclusions included uncorrected hearing disabilities or speech impediments; HCs had no psychiatric history or family history of psychosis. PANSS assessed symptom severity. Recruitment and data collection occurred Dec 2015–Mar 2018 under ethics approval with consent. Antipsychotic classification: Low D2R occupancy drugs included quetiapine, paliperidone, olanzapine, clozapine; high D2R occupancy drugs included aripiprazole, risperidone, flupentixol, amisulpride, haloperidol. Dosages converted to chlorpromazine equivalents. Speech elicitation: Semi-structured, neutral-topic interviews (5–30 min; average ~14 min) conducted by trained staff. Recording: Head-worn cardioid microphones; digital recording at 44.1 kHz/16-bit. Speaker separation into participant vs interviewer tiers by trained raters; inter-rater ICC 97.7%. Audio normalized to 60 dB. Acoustic and linguistic processing: Praat and Syllable Nuclei script extracted total syllables, pauses (silences >200 ms), and derived articulation/speech rates and pausing metrics. Transcriptions via CLAN (CHAT); EVAL and FLUCALC provided linguistic measures including total words, type-token ratio (TTR), open-closed ratio, noun-verb ratio, mean length of utterance (MLU), disfluencies, clauses per utterance, percentage of time speaking, speaking turn duration, pause-to-word ratio. Variables are detailed in Table 1 of the paper. Statistical analysis: IBM SPSS 25. Group differences in demographics by ANOVA/χ²; language comparisons by MANCOVA with covariates age, sex, education; Bonferroni corrections. Backward binary logistic regression predicted medication group (high vs low D2R) from language variables with dosage as covariate; model fit via Nagelkerke R² and Hosmer-Lemeshow. Multivariate linear regression models assessed effects of PANSS (positive, negative, general), D2R occupancy category, and dosage on each language variable; multiple comparisons controlled with FDR.

• Sample: 23 high D2R patients, 18 low D2R patients, 40 HCs; groups matched on age and sex; patients had fewer years of education than HCs; PANSS totals and chlorpromazine equivalents did not differ between patient medication groups.
• MANCOVA: Significant main effect of group on language (Pillai’s trace = 0.526, F (2,80) = 1.843, p = 0.016, partial η² = 0.263). Significant effects of age and male sex; no education effect; no interactions.
• Post-hoc comparisons: High vs low D2R patients differed on total number of words (fewer in high D2R) and TTR. High D2R patients differed from HCs on multiple parameters including articulation rate, TTR, MLU. Both patient groups differed from HCs on percentage of time speaking and clauses per utterance, indicating illness effects. Low D2R patients were more similar to HCs on most language measures.
• Direction of effects: High D2R associated with more severe negative language features: slower articulation rate, increased and prolonged pauses, shorter utterances with fewer clauses, and fewer total words (consistent with alogia).
• Logistic regression predicting medication class: Final model included mean pause duration, MLU, TTR, and speaking turn duration; good fit (Nagelkerke R² = 0.560; Hosmer-Lemeshow p = 0.932); classification performance sensitivity 80.0%, specificity 76.5%.
• Multivariate regressions: Some language measures were primarily related to medication variables rather than symptoms. Pause duration and clauses per utterance were predicted by D2R occupancy (and pause duration also by dosage). Speaking turn duration was predicted by PANSS positive. Articulation rate, percentage of time speaking, MLU, TTR, and pause-to-word ratio were influenced by both symptom measures and D2R occupancy. Noun-verb ratio and percentage of disfluencies were not significantly predicted by the models.

Findings support the hypothesis that high D2R occupancy antipsychotics are associated with more severe language disturbances in schizophrenia than low D2R occupancy drugs. The pattern suggests two mechanisms: (1) cognitive-linguistic effects likely mediated by prefrontal hypodopaminergia and potential impacts on language-relevant white matter connectivity, reflected in increased pausing and reduced complexity (MLU, clauses per utterance, TTR); and (2) motor speech slowing due to extrapyramidal system effects, reflected in reduced articulation rate. The observed increase in pausing with high D2R drugs argues against simple sedation as the cause, since many low D2R drugs are more sedating. Results imply clinicians should consider medication effects when evaluating language symptoms (e.g., alogia may be medication-related), and researchers should account for antipsychotic class in language analyses.

Schizophrenia patients using high D2R occupancy antipsychotics exhibit more pronounced language disturbances than those on low D2R occupancy agents and healthy controls, independent of psychotic symptom severity. Language features can help discriminate medication class. The findings suggest that language disturbances may be mitigated under low D2R occupancy drugs or exacerbated by high D2R occupancy drugs, with important implications for functional outcomes and quality of life. Future work should use larger, randomized or longitudinal designs, include medication-naïve and non-psychotic antipsychotic groups, and disentangle contributions of different neurotransmitter systems and specific drugs.

• No medication-naïve or medication-free patient group and no non-psychotic antipsychotic-treated control group, limiting causal inference about medication effects.
• Cross-sectional design; no randomization; language changes over time could not be assessed and prescribing bias cannot be excluded.
• Small sample; a majority of high D2R patients used aripiprazole (65%), and clozapine grouped with low D2R despite its use in treatment-refractory patients; insufficient power to parse drug-specific effects.
• Potential confounders were not fully assessed (e.g., neuropsychological deficits, premorbid functioning, rapport, sleep, depression, paranoia).
• Bilingualism/language fluency may confound measures; groups did not differ statistically, but the study was underpowered to explore this factor.
• Analyses focused on dopaminergic mechanisms; many antipsychotics also act on serotonergic and anticholinergic systems which may affect speech (e.g., mucosal dryness).
• TTR differences may reflect smaller speech samples in patients (TTR inflates with fewer total words).