2025/06/26 by Eun‐Jung Kim · 1 voice
Computer Science · #Online Learning and Analytics #Natural Language Processing Techniques
paper · pdf · doi:10.64152/10125/73633
openalex publication_date 2025/06/26 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/21
This study investigated English learner profiles and challenges among 884 Korean 5th-grade students, with a focus on the role of AI-assisted language learning in shaping proficiency outcomes. While AI-based interventions have gained popularity, their effectiveness across diverse learner populations remains insufficiently explored. The study aimed to (a) empirically identify distinct learner profiles based on motivational, contextual, and socio-affective characteristics through latent class analysis (LCA), and (b) examine the predictive effects of AI participation and learner-specific factors on English proficiency levels using multinomial logistic regression. Results indicated that AI-assisted learning positively influenced class membership among students with strong affective traits, such as high motivation and confidence, but demonstrated limited effectiveness for learners facing multiple vulnerabilities. Although AI-supported instruction contributed to proficiency growth for certain groups, its independent predictive power was modest overall. These findings suggest that AI-based tools should be integrated thoughtfully within broader educational frameworks that include teacher mediation, structured curricula, and targeted support for underperforming learners. Future research should examine the long-term impacts of AI learning on diverse learner populations.