Understanding development of learner agency in GenAI-assisted oral argumentation through the critical GenAI literacy-based dialogic teacher feedback

Learner agency is central to effective human-GenAI collaboration, and understanding its dynamic and context-dependent nature requires attention to how learners’ agentic actions evolve through interactions with GenAI and instructional support. This study examined the development of learner agency in GenAI-assisted EFL oral argumentation within a seven-week critical GenAI literacy/CGAIL-informed dialogic teacher feedback intervention. Adopting a longitudinal multiple-case design, the study drew on multiple rounds of screencast recordings of learner-GenAI interactions, semi-structured interviews and stimulated recalls, together with three rounds of dialogic teacher feedback. Six learners participated in the qualitative substudy, with three focal cases selected to represent distinct developmental trajectories. The findings revealed transformative, unbalanced, and fluctuating patterns of agentic enactment. Across cases, learners generally became more capable of using specific, goal-oriented prompts and regulating the linguistic features of GenAI output, whereas their ability to critically evaluate and regulate its content and logical coherence developed less consistently. Learners also responded differently to dialogic feedback, with their developmental trajectories shaped by the interplay between instructional support and individual factors, including language proficiency, goal clarity, and self-efficacy. These findings demonstrate that learner agency in GenAI-assisted speaking develops in differentiated and nonlinear ways and highlight the value of sustained, responsive CGAIL-informed support that accommodates learners’ evolving needs.