Detecting Features of Interpersonal Difficulties in First-Person Accounts of Schizophrenia; Automated Linguistic and Network Analyses

Abstract Background First-person accounts (FPAs) provide multi-faceted insights into the lived social experiences of schizophrenia. We examined FPAs and third-person accounts (TPAs) using automated linguistic and network analyses to extract narrative patterns in relation to interpersonal difficulties. Study Design Narratives from 279 articles published in Schizophrenia Bulletin (1979-2025) were screened to identify 133 FPAs and 44 TPAs. Schizotypal Personality Questionnaire (SPQ) was used to assess syndromes of schizotypy expressed in the narratives. At least one cognitive-perceptual syndrome (ideas of reference, odd beliefs, magical thinking) was detected in every FPA (n=133). Interpersonal difficulties (social anxiety, no close friends, constricted affect) were detected in 56 FPAs and designated as the “interpersonal difficulties” group. We used the Linguistic Inquiry and Word Count software (LIWC) to extract interpersonally relevant features: authenticity, emotional tone, pronoun use, social behavior, and confidence. Linguistic features of narratives in FPAs with and without interpersonal difficulties were compared using network analysis. Results Greater authenticity and lower confidence were detected in FPAs compared to TPAs. FPAs contained fewer social words and more “I” words than TPAs. FPAs with interpersonal difficulties were longer, more negative, and less linguistically formal than FPAs without interpersonal difficulties. Network analysis indicated a link between interpersonal difficulties and "voices.” Conclusions Narratives categorized by interpersonal difficulties exhibit distinct linguistic signatures revealed through automated linguistic tools and network analyses. Despite the vast range of topics in FPAs, these computational methods provide a robust quantitative framework for validating the lived experience of negative syndrome in schizotypy.