THE ROLE OF LARGE LANGUAGE MODELS (LLMS) IN PERSONALIZED ENGLISH LANGUAGE INSTRUCTION

Abstract

This study addresses a practical gap in English Language Teaching research, namely the lack of quantitative, case-grounded evidence on how large language models deliver personalized instruction at scale. The purpose is to model how LLM personalization and feedback translate into learner outcomes within real classrooms using enterprise cloud deployments. Adopting a quantitative, cross-sectional, case-based design, we sampled 312 learners across three cloud or enterprise cases that had embedded LLM-mediated tasks for 6 to 8 weeks. Key variables were personalization features, feedback quality, perceived usefulness, perceived ease of use, interaction frequency, engagement, self-efficacy, and perceived learning outcomes, each measured with five-point Likert scales. The analysis plan combined descriptive statistics and reliability checks with correlation matrices, hierarchical regressions using case fixed effects and HC3 standard errors, and bootstrapped mediation and moderation tests. Headline findings show that personalization features and ease of use jointly predict perceived usefulness, feedback quality is the strongest predictor of self-efficacy, and engagement and self-efficacy are the primary proximal predictors of outcomes after controls. Mediation results indicate that personalization influences outcomes serially through usefulness and engagement, while feedback quality acts via self-efficacy; moderation shows stronger returns for lower-proficiency learners and a steeper usefulness–outcome link for learners with higher AI familiarity. Implications include prioritizing actionable feedback design, pacing to reduce extraneous load, providing prompt-fluency scaffolds for novices, and governing prompt libraries and data flows as curricular assets so that value perceptions convert into frequent, purposeful use and measurable gains. These results offer a replicable blueprint for enterprise cloud implementations of LLM-enabled English instruction that centers learner engagement and efficacy as the engines of improvement.