24. ANALYSIS OF LANGUAGE IN SCHIZOPHRENIA

Abstract Language is an essential anamnestic source of information in psychiatry. More specifically, disorganized language is a core criterion for schizophrenia and analyzing this language may hold clues as to what it is that goes astray in the schizophrenic brain. However, language, in part due to its wide range of applications and nuances, is difficult to analyze and quantify. Through the advent of computational language processing, it has become possible to study a patient’s language functioning in an objective, quantifiable way. This creates the opportunity to use language as a marker for diagnosis, prognosis and perhaps treatment response of a patients with a schizophrenia spectrum disorder. Since language involves so many complex cognitive functions, a wide variety of aspects of verbal communication can be assessed by means of automated language analysis. The speakers in this symposium will use a wide range of approaches to assess different aspects of language in schizophrenia subjects, covering early detection and diagnosis (Natalia Mota), classification based on specific language markers (Guillermo Cecchi), symptom assessment and side-effects of medication (Alban Voppel) and the transformation of research into clinically useful tools (Terje Holmlund). First, Alban Voppel will present research investigating language disturbances in schizophrenia-spectrum disorders, both innate and those stemming from medication side-effects. Making use of among other measures a novel word2vec semantic and linguistic approach, disturbances were found between schizophrenia-spectrum patients and healthy controls, as well as between patients depending on dopaminergic characteristics of antipsychotics. Quantified markers derived from spontaneous language can be used both in assessing symptoms as well as side-effects that present possible confounders. Secondly, Dr. Guillermo Cecchi (IBM) will present research involving metaphor usage in ultra-high-risk as well as schizophrenia-spectrum disorder. His research shows that automated linguistic analyses of schizophrenia and its risk states can be extended from semantic coherence metrics at the phrase or sentence level to the level of single words. Since this method is also sensitive in regards to ultra-high-risk participants, it might be used as a screen to identify individuals with attenuated psychosis syndrome. Thirdly, Dr. Natalia Mota will present her research concerning early markers of thought disorganization found in sp