Dynamicity of EFL learners’ emotions and WTC in Human-AI interactions

Abstract

This study investigates dynamic fluctuations in willingness to communicate (WTC) and emotional states among six Chinese English as a Foreign Language (EFL) learners during interactions with Replika, an AI-powered chatbot. Using an idiodynamic method and stimulated recall interviews, the study provides a granular understanding of WTC and emotional dynamics. Participants exhibited a range of emotional states, including epistemic and retrospective emotions, which were intermittently elicited by moment-to-moment interactional events. Fluctuations were induced by various factors, including anthropomorphism, conversational responsiveness, perceived naturalness of interaction, and learners’ perceptions of Replika as either a conversational companion or a tool for language practice. Finally, this study proposes expectation appraisal as a key mediating mechanism between interactional factors, emotional states, and WTC. Learners’ expectations constituted a significant factor influencing emotional outcomes and ongoing WTC. These findings are discussed with reference to the L2 WTC pyramid model, emphasizing the interplay between emotions and communicative behaviors in AI-mediated language-learning environments.