AI-Assisted Self-Paced Grammar Instruction in Secondary Education: Balancing Individualized and Collaborative EFL Learning

Abstract

This study explores the implementation of a self-paced grammar instruction system supported by generative AI tools (ChatGPT and Google Gemini) in an under-resourced Japanese high school. The program integrated level-based worksheets, AI-generated feedback, and teacher scaffolding over six weeks, targeting students with diverse academic backgrounds and interrupted schooling histories. Data were collected through surveys, progress records, and classroom observations. Results indicated that students appreciated the flexibility and autonomy of the self-paced approach. While some desired more direct teacher input, the overall findings suggest that the AI-supported learning environment contributed to increased motivation, iterative learning, and differentiated support. Rather than focusing on the effects of AI alone, this study highlights how learner-centered design, when supplemented by AI tools, can foster engagement and autonomy in linguistically diverse classrooms. This practitioner- research contributes to the growing field of AI-supported education by presenting a scalable model designed for under-resourced or alternative secondary education contexts. The study underscores the relevance of learner-centered pedagogy and self-paced systems enhanced by AI tools, especially in schools serving students with diverse needs. Implications for using generative AI to support autonomy, inclusion, and engagement in EFL classrooms are discussed.