Exploring AI-Assisted Translation and Cognitive Load in L2 Learning: Evidence From Indonesian University Students

Abstract

This study examines the role of AI-assisted translation in second language learning among Indonesian university students, with particular attention to cognitive load, engagement, feedback literacy, and perceived translation quality. Adopting a mixed-methods design, the research combined quantitative data from 217 students with qualitative insights from 12 semi-structured interviews involving learners from six language programs, including English, German, Arabic, Mandarin, Korean, and Japanese. The quantitative findings showed that all three measured constructs were rated at high levels, and the questionnaire subscales demonstrated satisfactory reliability, with Cronbach’s alpha values above 0.80. Students generally perceived AI-assisted translation as beneficial for reducing surface-level difficulties, especially in vocabulary selection, grammatical accuracy, and task completion speed. At the same time, the findings revealed that the tool did not eliminate mental effort, as students still needed to verify, evaluate, and revise AI-generated outputs carefully. The interview data added further depth by showing that AI-assisted translation was valued for improving efficiency and confidence, yet remained limited in handling idiomatic expressions, contextual appropriateness, and cultural nuance. Importantly, repeated use of the tool encouraged students to become more reflective and strategic, treating AI outputs as provisional drafts rather than final answers. The study suggests that AI-assisted translation functions not only as a practical aid for accuracy and efficiency, but also as a pedagogical resource that can promote critical, reflective, and strategy-driven learning in second language contexts.