AI-Assisted Language Education, Learner Agency, and Multilingual Classroom Discourse: A Comparative Corpus-Based Study of Secondary English Learning in Finland and South Korea

Abstract

The increasing institutional adoption of artificial intelligence in language education has transformed classroom discourse, learner agency, and multilingual pedagogical interaction. This article comparatively investigates how AI-assisted English language learning reshapes communicative practices in secondary education contexts in Finland and South Korea. Drawing on corpus-based discourse analysis, educational sociolinguistics, and critical applied linguistics, the study analyzes 1.6 million words of classroom interaction, AI-generated learning feedback, teacher policy documents, and learner writing samples collected between 2023 and 2025. The findings indicate that Finnish classrooms used AI tools primarily for dialogic scaffolding, multilingual reflection, and learner autonomy, whereas South Korean classrooms employed AI more strongly for performance optimization, corrective feedback, and examination-oriented linguistic accuracy. The comparative evidence demonstrates that AI-mediated language education does not produce uniform pedagogical outcomes; rather, it is shaped by institutional assessment cultures, language ideologies, teacher roles, and national education policy traditions. This study argues that AI-assisted language learning transforms classroom discourse through a causal chain linking educational policy, technological mediation, discourse participation, learner identity, and communicative agency. The article contributes to applied linguistics by proposing a sociotechnical model of AI-mediated learner agency in multilingual language education.