Using Adaptive Empathetic Responses for Teaching English
Li Siyan, Teresa Shao, Zhou Yu, Julia Hirschberg
TL;DR
This work addresses integrating empathetic feedback into English-language tutoring by detecting negative affect from learner audio to trigger adaptive support in a sensor-free spoken chatbot. It couples a grammar-correction pipeline with a prompt-optimized empathetic feedback generator, using a DSPy-based workflow to tune GPT-4 outputs for colloquial delivery and a Rewrite stage to improve naturalness. A Mandarin-accented English speech dataset with emotion annotations is released, and a preliminary user study shows learners perceiving affective support and engaging more in conversations, suggesting potential improvements in L2 grit. The approach demonstrates a pragmatic path to embedding affective pedagogy in real-time spoken tutoring and identifies concrete directions for scaling, evaluation, and robustness in future work.
Abstract
Existing English-teaching chatbots rarely incorporate empathy explicitly in their feedback, but empathetic feedback could help keep students engaged and reduce learner anxiety. Toward this end, we propose the task of negative emotion detection via audio, for recognizing empathetic feedback opportunities in language learning. We then build the first spoken English-teaching chatbot with adaptive, empathetic feedback. This feedback is synthesized through automatic prompt optimization of ChatGPT and is evaluated with English learners. We demonstrate the effectiveness of our system through a preliminary user study.