Memory-Augmented Generative Intelligence Framework for Student-Centered Adult Language Teaching

Abstract

The systems designed for language learning using artificial intelligence (AI) have been well recognized for offering adaptive and personalized learning experiences; however, often lack long-term memory, contextual understanding, and adaptability for learners. To address these gaps, the current study proposes a new memory-augmented generative intelligence framework for student-centered adult language teaching that includes learner profiling, memory-enhanced knowledge management, semantic information retrieval, and generative AI for personalized language services. The proposed framework is based on five major modules, including learner information profiling, memory-augmented knowledge repository, semantic retrieval module, generative intelligence engine, and adaptive recommendations with feedback-based memory updating. For this research, a mixed-methods empirical approach consisting of framework development, learner-based evaluation, and statistical validation has been adopted. Data have been collected from 250 adult language learners through questionnaires and an AI-aided learning experience with respect to learner proficiency, interaction behavior, preferences, feedback, and learning outcomes. Construct validity was established since there were high Cronbach’s alpha values, with the minimum being 0.847 and the maximum 0.891, and there were positive correlations between AI personalization, memory capability, information quality, user satisfaction, and learning effectiveness. The framework had a total performance score of 4.26, with learning effectiveness and memory capability receiving highly evaluated performance values of 4.35 and 4.31, respectively.