Discrimination of Speech Content in Unipolar Depression and Bipolar Mania: A Computer-Based Analysis with "General Inquirer"

Objective: Speech disorders in mental illnesses are usually chronic and associated with poorer outcome. Recently, different types of speech features in mental illnesses can be examined by computer technology. The aim of our study is to examine the content of speech in depression and mania and to investigate the themes that differentiate the diagnostic groups. Method: 30 patients diagnosed with depression, 30 patients diagnosed with bipolar disorder manic episode and 30 healthy control were included in the study. All participants were performed with the Structured Clinical Interview for DSM-IV Axis I Disorders. The participants were asked to speak free for ten minutes and then their speech content was analyzed with the "General Inquirer" computer program. This program analyzes the participants' use of a total of 4919 words in the Harvard Psychosocial Dictionary, which are categorized in 83 themes on topics related to psychosocial, emotion, behavior, thought, natural and cultural environment. Results: The diagnostic groups were identified by speech content categories with an accuracy rate of 81%. Patients in mania and depression groups were clustered in the same direction in discriminant analysis by the themes of speech content. ''self" and ''academic" themes were the most discriminative categories between the patient and control groups. Conclusion: The content of speech in mania and depression is different from individuals without mental disorders and that computer-assisted analysis tools can distinguish diagnostic groups from each other and from healthy group. Future studies in which structural, vocal and content features of speech are evaluated together and used more advanced computer technologies will contribute to the literature.