Computerized content analysis of conversational interactions

The emotional impact of illness sustained by patients, their families, and friends is increasingly recognized as an important dimension of clinical evaluation and intervention. This study first attempted to extend a computerized technique of measuring the magnitude of mental events from a content analysis of speech initially designed for use on single verbal samples or verbal texts to the analysis of conversational interactions. The study then applied this method of extended measurement to conversations concerning seriously ill surgical patients. The authors' objective was to develop a graphic method that illustrates the neuropsychological interchanges occurring in these conversations. Speech samples were recorded during discussions among nurses, surgeons, intensivists, and family members involved in the care and treatment of seriously ill patients hospitalized in the surgical intensive care unit at Barnes-Jewish Hospitals associated with Washington University School of Medicine, St. Louis, Missouri. These samples then were transcribed into computer-readable text files and divided into utterances: uninterrupted speech by a single individual. The utterances were assigned numeric scores by a computer program on each of 14 validated content analysis scales measuring various neuropsychological dimensions. The scores are presented graphically in chronological sequence to illustrate the emotional interactions in these conversations. The graphic representations depicting sequences of computerized content analysis-derived scores for the mental reactions of the speakers in such conversations provide insights into the psychodynamics occurring in such stressful situations.