Speaker
Abstract section 2: Contribution/research questions
This study asks: How has the relationship between English proficiency and AI engagement in classroom learning changed between 2023 and 2025, and what patterns emerge across diffusion stages for any observed differences? Most existing research treats the proficiency–AI relationship as stable across time; this presentation challenges that assumption through interpretive cross-cohort comparison. By comparing cohorts across different stages of AI normalization, this presentation advances a temporally sensitive perspective on learner differences in AI-mediated language education.
Abstract section 3: Content/method
Data were drawn from two EMI language courses conducted at different diffusion stages. The 2023 study (N = 142) examined AI integration depth using a SAMR-based framework and ANCOVA analysis (Kikuchi, 2024). The 2025 study (N = 78) employed exploratory factor analysis and Spearman correlations to examine relationships among TOEIC scores, AI-supported engagement, and academic performance. Interpretive cross-cohort comparison was then conducted to examine shifts in the proficiency–AI relationship.
Abstract section 4: Outcomes/results
In 2023, English proficiency significantly predicted AI integration depth (η² = .071). In contrast, 2025 findings revealed no significant association between TOEIC scores and perceived AI-supported engagement (rs (68) = −.00, p = .972), although proficiency remained significantly associated with examination performance. These results suggest that AI engagement may have shifted from proficiency-dependent to proficiency-neutral as generative tools became normalized. The findings invite reconsideration of stable learner-difference models in AI-integrated language classrooms and have implications for task design and AI policy.
Abstract section 1: Relevance
Research in English-medium instruction (EMI) and language education has consistently identified English proficiency as a predictor of academic performance (Rose et al., 2026). At the same time, research on technology use in EMI has only recently begun to emerge as a distinct area of inquiry (High et al., 2025). As generative AI tools become embedded in classrooms, a common assumption is that higher proficiency enables more effective AI use. Early empirical evidence appeared to support this view. However, rapid technological diffusion and increasing student familiarity may be reshaping how proficiency interacts with AI engagement.
Abstract section 5: References
High, M. D., McIntosh, A., Li, S., & Ji, Y. (2025). Student machine translation use in a transnational English-medium instruction university: Navigating development and expedience. Journal of English-Medium Instruction, 4(2), 238–258.
Kikuchi, H. (2024). Transforming English Medium Instruction (EMI): The role of generative AI in overcoming EMI challenges and enhancing learning environments. In T. Bastiaens (Ed.), Proceedings of EdMedia + Innovate Learning (pp. 1046–1051). Association for the Advancement of Computing in Education (AACE). https://www.learntechlib.org/p/224625
Rose, H., Sahan, K., Wei, M., Aizawa, I., Zhou, S., & Shepard, C. (2026). A systematic review of English medium instruction in higher education: An update of Macaro et al. (2018). System, 136, 103892. https://doi.org/10.1016/j.system.2025.103892
| Title | Has AI Changed the Role of English Proficiency in Classroom |
|---|---|
| Teaching Context | College and university education |