Modeling Empathetic Alignment in Conversation

TL;DR

This paper presents a novel method for modeling empathetic alignment in conversation using Appraisal Theory, supported by a new dataset and experiments showing accurate recognition of appraisals and alignments, with implications for mental health communication.

Contribution

Introduces a new dataset with span-level annotations of appraisals and empathetic alignments, and demonstrates that these can be accurately recognized in conversational data.

Findings

Appraisals group behaviors meaningfully

Most responses show minimal alignment

Mental health professionals exhibit higher empathetic alignment

Abstract

Empathy requires perspective-taking: empathetic responses require a person to reason about what another has experienced and communicate that understanding in language. However, most NLP approaches to empathy do not explicitly model this alignment process. Here, we introduce a new approach to recognizing alignment in empathetic speech, grounded in Appraisal Theory. We introduce a new dataset of over 9.2K span-level annotations of different types of appraisals of a person's experience and over 3K empathetic alignments between a speaker's and observer's speech. Through computational experiments, we show that these appraisals and alignments can be accurately recognized. In experiments in over 9.2M Reddit conversations, we find that appraisals capture meaningful groupings of behavior but that most responses have minimal alignment. However, we find that mental health professionals engage with substantially more empathetic alignment.