Examining the relationship between therapists' communication and common factors in psychotherapy

Abstract

Background: Empathy is considered a significant factor in facilitating client improvements and has been found to be closely related to therapeutic alliance. There is currently a lack of research examining the behavioural markers of empathy. Active listening has been identified as an important factor in demonstrating empathy. Non-verbal active listening communication, otherwise known as backchannels, can be communicated through non-lexical items, for example “ah” and “hmm”, as well as encouraging gestures such as smiling or head nods. The study of behaviour in social interactions is a resource intensive process but affective computing (the study or mimicry of human emotion using computing techniques) may be able to alleviate this burden. Method: Non-verbal active listening behaviours displayed in therapeutic roleplay recordings were annotated by hand and using affective computing software. Valence and arousal were also analysed using affective computing software. Observers provided continuous ratings of empathy for each roleplay. T-tests and linear mixed effects models were employed to analyse the difference in empathy ratings and levels of arousal and valence, in the therapist speaking versus listening periods. Results: Head nods and non-lexical backchannels were the main non-verbal active listening behaviours identified. Observer ratings of empathy were significantly higher in the speaking periods. Analyses of arousal and valence levels demonstrated that therapists displayed lower levels of arousal and moderately more negative valence when listening compared to speaking. Discussion: Findings support the application of affective computing techniques in the analysis of social interaction. Potential implications, limitations and recommendations for future research are discussed. Keywords: Empathy, active listening, affective computing Thesis is embargo 31 December 2026.