[Linguistic markers in improving the predictive model of the transition to schizophrenia]

Over the past twenty years, research in psychiatry has focused primarily on the early detection of schizophrenia. The objective has been to engage the patient with prodromal symptoms in a trajectory of care. It has also been a question of being able to offer treatment as soon as the patient "at risk" of schizophrenia triggered a possible first psychotic episode. Standardized clinical tools were developed and now allow identification of subjects at risk of developing psychotic disorders. However, the reliability of predictions of the psychotic transition, which is between 15 and 25%, remains insufficient. In order to improve care, it is now necessary to highlight markers to refine the prediction of the risk of developing schizophrenia. Some teams are trying to identify linguistic anomalies in UHR subjects (disorganized speech, illogical thoughts, poor speech, altered semantic verbal fluencies…). Some of these abnormalities could be specific to the transition to psychosis. The severity of these markers could be proportional to the progressive stage of the disorder, consistent with the hypothesis of a continuum from normal to pathological in schizophrenia. In addition, automated speech analysis techniques in UHR subjects allow identification of subtle semantic and syntactic anomalies (a decrease in semantic coherence, but also the use of possessive pronouns and a poverty of speech) predictive in 79% of cases of psychotic transition. Some authors demonstrate the value of using linguistic markers and automated speech analysis methods to improve the predictive model of the transition to schizophrenia. However, from reification of language to desubjectification of the individual, this transformation in clinical practice raises ethical and epistemological challenges.