Speaking behavior and voice sound characteristics associated with negative schizophrenia.

Based on a sample of 42 chronic schizophrenic patients and 42 carefully matched controls, we investigated potential relationships between acoustic variables on the one hand, and negative syndromes, positive syndromes and affective disturbances, on the other. A set of 12 acoustic variables automatically assessed in a standardized experimental setting allowed an almost perfect discrimination between schizophrenic patients and normal subjects. Acute side-effects of medication did not explain this finding. However, the question of whether the observed changes in speaking behavior and voice sound characteristics were caused by long-term neuroleptic treatment, for example, as a consequence of tardive dyskinesia, could not be answered by our investigation. In view of a biological validation of the negative-positive model of schizophrenia, the reliability of various psychopathological subscales was tested through repeated assessments at 14 day intervals. We found most psychopathology scores to be sufficiently stable and reproducible over time, thus representing a suitable basis for the estimation of severity with respect to the negative and positive component of schizophrenia. Using the first measurements as training samples and the second measurements of 14 days later as test samples, discriminant analysis yielded conclusive proof of a close relationship between acoustic variables and the severity of the negative and positive component of schizophrenia. In particular, by means of "objective" acoustic variables and under the constraint of reproducibility, 75.9% of patients were correctly classified as low or high scorers with respect to the negative syndrome, 71.9% of patients with respect to the positive syndrome, and 79.4% of patients with respect to their depressive symptomatology.