AI-BASED ADAPTIVE LEARNING: A NEW PARADIGM FOR PERSONALIZED FOREIGN LANGUAGE TEACHING

Abstract

Foreign language classrooms are linguistically and affectively heterogeneous, yet most instruction is still paced for an imagined average learner. This article examines whether artificial intelligence (AI)-based adaptive learning constitutes a genuinely new paradigm for personalized foreign language teaching or merely automates older ideas. Using a focused analytical review of ten scholarly sources from American and Kazakhstani research traditions, supplemented by a comparative analysis of effect sizes reported in the intelligent tutoring literature, the study asks which theoretical principles justify adaptivity, what evidence supports its effectiveness, and which conditions constrain it in Kazakhstani higher education. The analysis shows that adaptive systems reproduce, at scale, mechanisms long associated with effective tutoring, namely diagnosis, calibrated challenge, and immediate feedback, and that reported effects are substantial but vary widely with design. Kazakhstani evidence is promising but thin; it identifies infrastructure, teacher digital literacy, data privacy, and shallow personalization in current tools as recurring constraints. The article proposes a four-layer model (diagnosis, adaptation, feedback, and human oversight) and derives implementation principles for universities. It concludes that adaptivity is a paradigm shift only when pedagogy, rather than technology, governs the learning loop. Limitations include the small corpus and reliance on secondary data.