From AI use to AI appropriation: how EFL learners develop interaction trajectories in AI-mediated speaking practice

Recent advances in artificial intelligence (AI), including generative AI, have expanded opportunities for second language speaking beyond the classroom. However, research has focused largely on AI affordances, learner perceptions, and learning outcomes, with less attention to how learners gradually appropriate AI into sustainable speaking practices. This longitudinal qualitative case study examines how EFL learners move from AI use to AI appropriation within an AI-mediated out-of-class speaking ecology. Drawing on sociocultural theory and an ecological perspective, with appropriation as the primary analytical lens, the study followed 30 first-year university students in China through an eight-week oral English programme integrating an AI-supported practice platform, teacher mediation, peer interaction, and reflection. Data comprised AI interaction logs, learner reflections, group interaction records, and semi-structured interviews. Analysis identified three engagement trajectories – strategic, negotiated, and limited – and selected three information-rich cases for within- and cross-case analysis. Findings show that learners with access to similar AI affordances differed substantially in how they interpreted feedback, judged its relevance, and incorporated it into subsequent speaking practice. Productive appropriation emerged through the interaction of learner agency, teacher mediation, and reflective engagement rather than technological affordances alone. By shifting attention from adoption to appropriation, the study conceptualises AI-mediated innovation as a learning process rather than simple technology use. It highlights AI feedback literacy, reflective scaffolding, and teacher mediation as key pedagogical conditions for sustaining meaningful learner engagement with GenAI.