A computational approach to measure the linguistic characteristics of psychotherapy timing, responsiveness, and consistency

Abstract

Although individual psychotherapy is generally effective for a range of mental health conditions, little is known about the moment-to-moment language use of effective therapists. Increased access to computational power, coupled with a rise in computer-mediated communication (telehealth), makes feasible the large-scale analyses of language use during psychotherapy. Transparent methodological approaches are lacking, however. Here we present novel methods to increase the efficiency of efforts to examine language use in psychotherapy. We evaluate three important aspects of therapist language use - timing, responsiveness, and consistency - across five clinically relevant language domains: pronouns, time orientation, emotional polarity, therapist tactics, and paralinguistic style. We find therapist language is dynamic within sessions, responds to patient language, and relates to patient symptom diagnosis but not symptom severity. Our results demonstrate that analyzing therapist language at scale is feasible and may help answer longstanding questions about specific behaviors of effective therapists.

Natural language processing for mental health interventions: a systematic review and research framework

“…Conversational skills. Five manuscripts examined the linguistic ability of providers [ 33 , 50 , 65 , 125 , 126 ]. One study [ 33 ] generated a model from 80,885 counseling interventions to extract therapist conversational factors, and showed how differences in content and timing of these factors predicted outcome (patient-reported helpfulness; AUC = 0.72).…”

“…One study [ 33 ] generated a model from 80,885 counseling interventions to extract therapist conversational factors, and showed how differences in content and timing of these factors predicted outcome (patient-reported helpfulness; AUC = 0.72). Importantly, conversational markers not only captured between-provider differences, but also found within-provider differences related to patients’ diagnoses [ 50 ] and as they gained clinical experience over time [ 65 ].…”

“…To successfully differentiate and recombine these clinical factors in an integrated model, however, each phenomenon within a clinical category must be operationalized at the level of utterances and separable from the rest. The reviewed studies have demonstrated that this level of definition is attainable for a wide range of clinical tasks [ 34 , 50 , 52 , 54 , 73 ]. Utterance-level operationalization exists for some therapy frameworks [ 153 , 154 ], which can serve as exemplars to inform the specification process for other treatment approaches that have yet to tie aspects of speech to their proposed mechanisms of change and intervention.…”

Assessing the Reproducibility of the Structured Abstracts Generated by ChatGPT and Bard Compared to Human-Written Abstracts in the Field of Spine Surgery: Comparative Analysis

“…Our data suggest that Bard is less capable of generating scientific abstracts compared with ChatGPT in response to prompts referring to rare, poorly known, or new data. Nevertheless, given Bard's characteristics, the program is likely to be used to provide scientific abstracts from unlimited sources connected to Google, and a reassessment is essential [16][17][18].…”