Automated Analysis of Discourse Coherence in Schizophrenia: Approximation of Manual Measures
Disorganized, or incoherent, speech is one of the important criteria for diagnosing schizophrenia. However, there is still a lack of a rather quick objective method of measuring speech coherence. Automated discourse analysis is a possible solution to this problem. We analyzed discourse coherence in a set of spoken narratives by people with schizophrenia and neurotypical speakers of Russian. All narratives were manually rated for violations of completeness, local, global and dimensional coherence. A number of automated vector semantics methods were used for approximation of the manual rating scores. The metrics used proved to be a good approximation for manual scoring, and a combination of them was efficient for classification narratives in schizophrenia and neurotypical groups.