From attitudes to actions: A critical interaction-based study of teacher mediation of AI-generated materials through an ELF lens

Abstract

This study investigates how 71 Iranian EFL teachers critically engage with AI-generated materials through an ELF-informed perspective. While prior research has largely focused on perceptions or technology acceptance, little is known about how teachers actively mediate AI content in real time. To address this gap, the study employed a Critical Interaction-Based Mixed-Methods Design (CIMMD) consisting of three phases: (1) quantitative diagnostic assessment using TAAI and ELF-AIMQ, (2) a prompt-engineering and think-aloud protocol with a subset of 21 teachers, and (3) thematic and document analysis of interview data and revised AI outputs.

Quantitative results showed stable attitudinal trends across participants, with only modest correlations between ELF awareness and AI acceptance. However, interaction-based evidence demonstrated clear critical AI mediation, as teachers identified native-speaker bias, recontextualized cultural content, and integrated plurilingual elements. Four recurrent strategies were observed: language norm editing, cultural recontextualization, plurilingual integration, and ideological evaluation. The study introduces the AI–ELF Mediated Engagement Cycle, a process model conceptualizing teachers’ engagement with AI as a dynamic sequence of awareness, evaluation, revision, and ideological negotiation. This methodological shift suggests that attitudes alone may not predict AI use; rather, teacher agency emerges most clearly through interaction. The findings advocate for ELF-responsive AI design, teacher training in critical digital literacy, and pedagogical control over AI-driven language content. The study contributes to a growing trajectory of critical, socio-cognitive AI pedagogy in ELT.