Modeling Empathetic Alignment in Conversation

Jiamin Yang,David Jurgens

2024-05-02

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.

Computation and Language

What problem does this paper attempt to address?

### The Problem the Paper Aims to Solve

This paper aims to address the issue of empathetic dialogue modeling in the field of Natural Language Processing (NLP), particularly understanding empathy from a cognitive perspective rather than just an emotional one. Specifically:

1. **Empathetic Alignment**: The paper introduces a new method to identify empathetic alignment in dialogues, based on Appraisal Theory. This method not only focuses on emotional resonance but also emphasizes cognitive understanding and perspective-taking.

2. **Dataset Construction**: The paper proposes a new dataset called ALOE (Alignment of Empathy), which contains over 9200 different types of appraisal annotations and more than 3000 empathetic alignment annotations between speakers and observers. These data are sourced from therapeutic dialogues on Reddit.

3. **Model Performance Evaluation**: Through experiments, the paper validates the accuracy of appraisal annotations and empathetic alignment recognition, and demonstrates the application effects on a large-scale Reddit dialogue dataset.

4. **Comparison with Professionals**: The paper compares the level of empathetic alignment between mental health professionals and laypeople in Reddit dialogues, finding that professionals exhibit a higher degree of empathetic alignment.

In summary, the paper primarily addresses how to better understand and model empathetic alignment in NLP, especially in clinical and mental health support scenarios.