Prompt-Based Adaptive English Learning: Leveraging AI for Personalized Practice

Abstract

The integration of generative artificial intelligence (AI) into language education has enabled scalable and personalized English learning experiences. However, the pedagogical effectiveness of such systems largely depends on the quality of prompt design that guides AI response generation. Most current AI-assisted English learning platforms rely on static or pre-set prompts, limiting adaptability to learners' varying proficiency levels and emotional needs. The absence of a systematic framework connecting prompt construction with adaptive learning theory constrains instructional precision and learner engagement. This study proposes an Adaptive Prompt Learning Model (APLM) that integrates prompt engineering with educational psychology. Using mixed methods, comparative case studies of ChatGPT, Duolingo Max, and iFLYTEK AI Tutor, combined with a six-week classroom experiment involving 120 learners, the research evaluates how structured prompts shape personalized exercises and feedback. Adaptive prompting improved vocabulary retention by 23.6%, grammar accuracy by 17.8%, and reduced learner anxiety through emotionally supportive feedback. The findings demonstrate that pedagogically informed prompt design can align AI behavior with cognitive and affective learning needs, providing a practical framework for scalable, interpretable, and equitable AI-assisted English education.