Mediation in AI-Assisted Experiential Language Learning: Extending TPACK Through the TPAEK Framework for Vietnamese as a Foreign Language Education

Abstract

Generative artificial intelligence (GenAI) is reshaping language education through its capacity to generate linguistic and multimodal learning materials. In low-resource contexts such as Vietnamese as a Foreign Language (VFL), however, AI outputs may produce culturally generalized or inaccurate representations because culturally grounded training data remain uneven. This mixed-methods questionnaire study examines how 54 VFL teachers perceive the pedagogical usefulness, applicability, and ethical implications of GenAI in experiential language learning. Building on the limitations of TPACK, the study proposes the Technological–Pedagogical–AI Ethical Knowledge (TPAEK) framework as a context-sensitive analytical lens that treats ethical knowledge as a mediating dimension of pedagogical decision-making. Quantitative findings indicate positive perceptions of GenAI’s pedagogical value and applicability, whereas ethical awareness emerged as a distinct but not directly predictive dimension in this exploratory model. Open-ended responses identified recurrent teacher-reported forms of AI-induced cultural distortion, particularly cross-cultural blending, visual inaccuracy, cultural misinformation, and symbolic stereotyping. Overall, the study suggests that teachers act as interpretive mediators who evaluate, contextualize, and regulate AI-generated content to preserve cultural validity in experiential language learning.