Generative AI is rapidly being integrated into English education, marking a new paradigm of innovation. Having experienced the pandemic’s abrupt shift in learning environments without adequate preparation, the present moment calls for proactive readiness. South Korea, where enthusiasm for English learning and an advanced ICT infrastructure create a unique context, could offer valuable insights. This study examines thematic patterns in Korean tertiary-level English education research during the COVID-19 pandemic. Text mining was applied to 190 Korea Citation Index (KCI)-indexed journal articles using three analytic methods: frequency analysis, semantic network analysis, and Latent Dirichlet Allocation (LDA) topic modeling. The findings indicate that learner-centered themes emerged as major areas of focus. Frequency analysis showed “learner” in 98.4% of documents, “online” in 91.6%, and “learning” in 76.8%. Semantic network analysis identified “learner” as the central node, most strongly linked with “online” (91.1%) and “learning” (75.8%) and highlighted low-proficiency learners, as “beginner-level” and “beginner-proficiency” pairs ranked among the top five associations. LDA topic modeling identified eleven thematic clusters, with learner satisfaction, engagement, and autonomy emerging alongside instructional design issues such as delivery formats, digital tools, and blended learning. Overall, the results suggest that the central pedagogical concern was not technology but the learner. In this new paradigm, as generative AI transforms educational practice, learners may become active participants who design, regulate, and critically engage with their own learning. Future research should explore how these evolving roles can be effectively supported through pedagogical design.
Sojeong Kim (2026) studied this question.