SCAFFOLD OR CRUTCH? EFL LEARNERS' MULTIMODAL GEN AI EXPERIENCES IN INDONESIA

Abstract

Despite the rapid integration of generative AI (GenAI) into EFL learning, most existing research has relied on single-method survey designs that capture student attitudes without explaining how learners actually engage with these tools in practice. This study employed an explanatory sequential mixed-methods design to investigate EFL students’ perceptions of the pedagogical value of multimodal GenAI and to explain those perceptions through in-depth exploration of their learning experiences. Guided by Sociocultural Theory, Multimodal Learning Theory, and Self-Determination Theory, a five-point Likert-scale survey was administered to 183 undergraduate EFL students across five Indonesian higher education institutions, followed by semi-structured interviews with 13 purposefully selected participants representing three distinct perception profiles. Quantitative data were analysed using descriptive statistics and exploratory factor analysis in R, while qualitative data were analysed through reflexive thematic analysis. Findings showed moderately positive perceptions of GenAI’s pedagogical value (M = 3.62, SD = 0.75), with critical thinking concern emerging as the most salient risk (M = 3.90, SD = 1.15) and a strong demand for lecturer guidance on ethical AI use (M = 3.72, SD = 1.20). Four themes were identified: GenAI as a language empowerment tool, navigating the balance between support and overreliance, enhancing engagement through multimodality, and shaping teacher roles. These findings point to the need for AI literacy curricula that develop students’ critical, ethical, and self-regulatory capacities in Indonesian higher education contexts.