Answers are grounded in this paper and cite their sources.
Qualitative study reveals multidimensional learner agency in generative AI-assisted speech learning among language students, indicating how human oversight refines automated text.
Generative artificial intelligence (GenAI) is increasingly integrated into second language education, reshaping the dynamics between teachers, learners and GenAI. However, little is known about learner agency in GenAI-assisted English speech learning. Drawing on Cultural-Historical Activity Theory, this qualitative study examines the manifestation and mediating role of learner agency among 43 Chinese EFL learners engaged in a GenAI-assisted speech project. Data were collected from multiple sources including human–GenAI chat logs, group discussions, speech manuscripts, reflective journals, interviews and questionnaires. Findings revealed three patterns of GenAI use: as a research assistant, a co-constructor and a consultant. Four types of agency–thematic framing, structure planning, curatorial, and contextual agency–were found to shape these patterns of use by mediating relationships among the speech learning activity system. Findings demonstrated that agency operates as a socially distributed, contextually embedded force enacted through collaborative deliberation, transforming generic GenAI outputs into thematically rich, logically coherent, and audience-adapted speeches. This study contributes to understanding how learner agency dynamically reshapes technology-mediated learning systems, offering pedagogical implications for integrating GenAI into L2 speech instruction.
No takes yet. Share an insight, caveat, or question.
Huiying Diao (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: