“Can you hear me, ChatGPT?”: interaction patterns, challenges, and learner perceptions in out-of-class EFL speaking practice

Abstract

Abstract As Generative AI (GenAI) expands voice interaction capabilities, it offers new possibilities for foreign language speaking practice in more realistic interactional environments. This study investigates EFL learners’ interaction patterns with ChatGPT as an out-of-class language partner and examines how these patterns and learner perceptions evolve over time. Over a 15-week semester, thirty EFL students at a Japanese university used their smartphones to converse with ChatGPT on six designated topics. Data were collected through recordings, transcripts, and pre- and post-surveys. Content analysis revealed that students relied mainly on reactive strategies, mostly responding to ChatGPT’s questions, while demonstrating fewer proactive behaviors, such as steering the conversation or asking questions. Task design also influenced interaction patterns. Over time, increases in conversation length and proactive output suggested greater interactional engagement and growing familiarity with AI-mediated speaking practice. However, recurring technical issues, particularly interruptions, often disrupted conversational flow, leading to disengagement, as also reflected in open-ended survey responses. The tension between potential benefits and practical limitations challenges uncritical technological optimism. Although GenAI offers valuable practice opportunities, its limitations mean it complements rather than replaces human interaction. Rather than assuming technology alone transforms learning, educators must develop frameworks that address technical constraints, scaffold learner autonomy, and cultivate critical AI literacy.