Speakers
Description
The rapid advancement of generative AI, particularly ChatGPT, has enabled ESL learners to engage in self-directed pronunciation practice with immediate, individualized feedback. While existing research has explored automated feedback tools, limited attention has been paid to how learners interact with AI systems that both generate learning materials and provide feedback autonomously. Addressing this gap, this qualitative case study examines how 15 ESL learners interact with ChatGPT-generated pronunciation materials and automated feedback over a 4–6 week period. Data from reflective journals, interviews, and practice recordings are analyzed thematically to investigate learner engagement, feedback interpretation, and self-correction strategies. Findings suggest that learners generally report positive experiences, valuing the flexibility and personalization of AI-supported practice. However, perceptions of feedback clarity and consistency are mixed: while some learners benefit from detailed explanations (e.g., IPA and examples), others experience cognitive overload and ambiguity. Immediate feedback appears to support noticing and self-monitoring, yet learning depth varies, with some learners developing effective strategies and others remaining feedback-dependent. This study provides empirical insights into the affordances and constraints of generative AI in supporting pronunciation development in self-directed ESL learning contexts.
