Structuring human–AI interaction in Korean as a Foreign language translation classes: the HAIT framework

AI-based translation tools have become an inescapable presence in language classrooms, yet the field of Korean as a Foreign Language pedagogy lacks a systematic framework for integrating them educationally. This paper proposes the HAIT (Human–AI Interaction in Translation Learning) framework, a five-phase instructional structure designed to pedagogically organize human–AI interaction in Korean as a Foreign Language translation classes. In HAIT, learners back-translate an English text – prepared by the teacher from a Korean source – into Korean without AI assistance, then engage in small-group tripartite comparisons of their translations, the Korean original and AI output, before a teacher-led whole-class discussion consolidates the learning. Drawing on translation competence models (PACTE Group, Citation2003; EMT Expert Group, Citation2017), Vygotsky’s sociocultural theory, and machine translation post-editing research, HAIT repositions AI output as a critical comparative resource rather than a prescriptive answer. The framework addresses three structural challenges unique to the Korean–English language pair: the SOV/SVO word order contrast, the sociopragmatic complexity of Korean honorifics, and the translation of culturally specific Korean lexical concepts. HAIT is presented as a theoretical framework; it has not yet been implemented or empirically tested in Korean as a Foreign Language classrooms. Empirical support for each phase is reviewed, existing research gaps are identified, and pedagogical implications for Korean as a Foreign Language teacher education and curriculum design are discussed.