The CIPP-TRS Corpus: Corpus Construction and Preliminary Analyses

Abstract

Schizophrenia, a neurodevelopmental disorder, significantly affects cognitive and linguistic functions, often resulting in disorganized speech, reduced syntactic complexity, and impaired discourse cohesion. While previous corpora have explored linguistic disruptions in schizophrenia, no dataset has systematically distinguished between treatment-resistant schizophrenia (TRS) and non-treatment-resistant (non-TRS) speech patterns. This study presents the CIPP-TRS Corpus, an annotated collection of transcribed speech from 20 individuals with schizophrenia (10 TRS, 10 non-TRS), alongside a control group of 10 neurotypical speakers. By analyzing peri-linguistic (e.g., interjections, pauses) and paralinguistic (e.g., breath patterns, output modalities) features, we investigate the linguistic manifestations of schizophrenia across these subgroups. Our preliminary findings suggest that TRS patients exhibit richer peri-linguistic markers, and increased hesitation phenomena, while non-TRS patients demonstrate greater lexical retrieval difficulties. Moreover, TRS individuals struggle more with temporal processing, particularly when recalling past events or engaging with past retellings, reinforcing theories on Theory of Mind (ToM) impairments and lived-time disturbances in schizophrenia. The CIPP-TRS Corpus represents a crucial step toward identifying linguistic biomarkers of schizophrenia and its treatment-resistant subtype. Future research will expand the dataset and incorporate prosodic, syntactic, and pragmatic analyses to refine our understanding of speech pathology in schizophrenia, with potential applications in clinical diagnostics and therapeutic interventions.