Linguistic markers of demoralization improvement in schizophrenia: A pilot study

• F. Folesani ^[1] ; Martino Belvederi Murri ^[1] ; C. Puggioni ^[1] ; E. Tiberto ^[1] ; M. Marella ^[1] ; T. Toffanin ^[1] ; L. Zerbinati ^[1] ; M.G. Nanni ^[1] ; R. Caruso ^[1] ; D. Brunato ^[2] ; A.A. Ravelli ^[2] ; F. Dell Orletta ^[2] ; H.M. Chochinov ^[3] ; L. Grassi ^[1]
  1. [1] University of Ferrara

     University of Ferrara

     Ferrara, Italia

     
  2. [2] ItaliaNLP Lab, Institute for Computational Linguistics “A. Zampolli” (ILC-CNR), Pisa, Italy
  3. [3] Max Rady College of Medicine, Department of Psychiatry, University of Manitoba and Cancer Care Manitoba Research, Winnipeg, Canada

  Mostrar afiliaciones +
• Localización: European journal of psychiatry, ISSN 0213-6163, Vol. 37, Nº 3, 2023, págs. 149-159
• Idioma: inglés
• DOI: 10.1016/j.ejpsy.2023.03.001
• Enlaces
  • Texto completo
• Resumen

  • Background and objectives Individuals with schizophrenia display language impairments involving pragmatics, semantics and syntax. Language impairments may show diagnostic specificity and could relate to the ability of engaging in psychotherapy. This pilot study sought to: (1) identify linguistic features that might differentiate individuals with schizophrenia from distressed controls without psychotic symptoms; and (2) examine the association between linguistic abilities and clinical changes during psychotherapy.

    Methods We recruited patients with schizophrenia and a comparison group of individuals with demoralization and distress due to cancer. Participants underwent Dignity Therapy (DT), an existentially-oriented brief psychotherapy focused on legacy and subjective dignity. Verbatim transcripts of the DT sessions were analysed using Natural Language Processing (NLP). In addition, we measured changes in levels of demoralization and dignity-related distress before and after DT, exploring the association with linguistic variables with network analysis.

    Results Patients with schizophrenia could be differentiated from those with cancer-related distress using only three out of 141 linguistic variables: total number of words, number of prepositional chains and conversational elements. Across groups, better levels of discourse coherence and higher number of arguments controlled by a predicate (verb “arity”) were associated with larger improvements in demoralization and, indirectly, dignity-related distress.

    Conclusions Reproducible linguistic markers may be able to differentiate individuals with schizophrenia from those with less severe psychopathology, and to predict better uptake of psychotherapy independent from diagnosis. Future studies should explore whether linguistic features derived from NLP may be exploited as accessible diagnostic or prognostic markers to tailor psychotherapy and other interventions in schizophrenia.