Computational Linguistic Alignment in Psychosis from Semi-Structured Clinical Interviews
- 1McGill University
Abstract
BackgroundSomething in discourse with a person experiencing psychosis often "feels off" before formal assessment is completed, yet this phenomenon has not been quantified at the level of ongoing dyadic conversation. Prior work has largely treated patient speech in isolation, limiting our capacity to measure how communicative disruption emerges within clinical exchange.
MethodsWe applied a three-level decomposition of conversational alignment in 109 patients with psychotic disorders (26 female) and 60 healthy controls (22 female) at baseline and 12 months (n = 115). Register divergence captured lexical distance between interviewer and participant; embedding-based synchrony measured semantic trajectory coupling; within-speaker coherence was computed separately for each speaker. We used linear mixed-effects models adjusted for timepoint and participant clustering.
ResultsPatients showed significantly greater lexical-semantic divergence from the interviewer (d = 0.48, p < .001) and reduced embedding-based synchrony (d = -0.59, p < .001), both effects observed at each timepoint, consistent with a partially stable profile. Critically, the interviewers within-speaker coherence was reduced during conversations with patients (d = -0.33, p = .016), indicating that the linguistic phenomenon of interest extended beyond patient speech to the interaction itself. Register divergence tracked impoverished thinking, while synchrony tracked disorganized thinking (both FDR-corrected q = .038).
ConclusionsConversational alignment in psychosis reveals a reconfiguration of dyadic semantic coordination, marked by reduced interviewer coherence despite preserved patient narrative continuity. These transcript-derived alignment metrics offer a scalable approach to quantifying interpersonal communicative function in routine clinical encounters.
Related
- Large language model linguistic perplexity in childhood onset psychosis: unique features and developmental trends
- Evaluating Open-Weight Large Language Models for Structured Depression Assessment from Clinical Interviews
- Violence prevention and incident documentation in elderly and special needs care institutions in southern Germany: results from semi-structured interviews
- Theory of Artificial Intelligence Adaptation for Linguistic Concordance: Bridging Gaps in Patient-Centered Communication
- How to Elicit Thought Disorder During Clinical Interviews: A Practical Guide
BibTeX
@misc{olarewaju2026computational,
title = {{Computational Linguistic Alignment in Psychosis from Semi-Structured Clinical Interviews}},
author = {E. Olarewaju and A. E. Voppel and F. Meister and C. El Mouslih and P. Dzialoszynski and L. PALANIYAPPAN},
howpublished = {medRxiv},
year = {2026},
month = may,
doi = {10.64898/2026.05.24.26353973},
url = {https://www.medrxiv.org/content/10.64898/2026.05.24.26353973v2},
note = {CC BY-NC-ND},
}
RIS
TY - UNPB
TI - Computational Linguistic Alignment in Psychosis from Semi-Structured Clinical Interviews
AU - E. Olarewaju
AU - A. E. Voppel
AU - F. Meister
AU - C. El Mouslih
AU - P. Dzialoszynski
AU - L. PALANIYAPPAN
PY - 2026
DA - 2026/05/24
PB - medRxiv
DO - 10.64898/2026.05.24.26353973
UR - https://www.medrxiv.org/content/10.64898/2026.05.24.26353973v2
AB - BackgroundSomething in discourse with a person experiencing psychosis often "feels off" before formal assessment is completed, yet this phenomenon has not been quantified at the level of ongoing dyadic conversation. Prior work has largely treated patient speech in isolation, limiting our capacity to measure how communicative disruption emerges within clinical exchange. MethodsWe applied a three-level decomposition of conversational alignment in 109 patients with psychotic disorders (26 female) and 60 healthy controls (22 female) at baseline and 12 months (n = 115). Register divergence captured lexical distance between interviewer and participant; embedding-based synchrony measured semantic trajectory coupling; within-speaker coherence was computed separately for each speaker. We used linear mixed-effects models adjusted for timepoint and participant clustering. ResultsPatients showed significantly greater lexical-semantic divergence from the interviewer (d = 0.48, p < .001) and reduced embedding-based synchrony (d = -0.59, p < .001), both effects observed at each timepoint, consistent with a partially stable profile. Critically, the interviewers within-speaker coherence was reduced during conversations with patients (d = -0.33, p = .016), indicating that the linguistic phenomenon of interest extended beyond patient speech to the interaction itself. Register divergence tracked impoverished thinking, while synchrony tracked disorganized thinking (both FDR-corrected q = .038). ConclusionsConversational alignment in psychosis reveals a reconfiguration of dyadic semantic coordination, marked by reduced interviewer coherence despite preserved patient narrative continuity. These transcript-derived alignment metrics offer a scalable approach to quantifying interpersonal communicative function in routine clinical encounters.
ER -