Client and Therapist Psychotherapy Sentiment Interaction Throughout Therapy

Abstract

Examining language used in psychotherapy can provide benefits for therapy outcomes and understanding therapeutic processes. Sentiment analysis is a type of content analysis using Natural Language Processing that can be applied to analyze the degree of positive or negative sentiment of therapist and client language in actual therapy sessions. Therapy transcripts were coded for degree of positive and negative sentiment at each exchange between therapist and client. Hierarchical linear models were constructed to evaluate change in sentiment within and across therapy sessions and the relationship between therapist and client sentiment. Results indicate that there was significant interaction effect, with increases in positive sentiment across therapy sessions, while positive sentiment tended to decrease within sessions. There was also found to be no difference, and instead significant correlation, between therapist and client sentiment in therapeutic dialogue over time, suggesting that client and therapist sentiment seemed to correspond over time.