Effectiveness of AI translation tools on EFL translation skills in higher education: a systematic review

This systematic review investigates how AI translation tools support EFL learners' translation learning by synthesizing empirical evidence on learner strategies, effectiveness, and pedagogical implications. Following PRISMA-informed procedures, 26 peer-reviewed studies published between 2008 and 2025 were screened and analyzed. To enable systematic cross-study comparison, the evidence was organized using an Environment-Task-Learner-Strategy (ETLS) framework. The synthesis shows that AI-assisted translation environments are predominantly shaped by readily accessible tools, with neural machine translation systems (e.g., Google Translate) dominating earlier studies and increasing attention to large language model tools (e.g., ChatGPT) in recent work. Across contexts, learners demonstrate a recurring repertoire of strategies, including cross-tool comparison and verification, reverse translation checks, selective tool use for difficult segments, iterative refinement (including prompt adjustment in LLM settings), and systematic post-editing. With respect to effectiveness, the evidence converges on short-term performance benefits such as improved linguistic accuracy and fluency, while also indicating boundary conditions. Learning benefits are more consistently observed when learners critically evaluate and revise AI-generated output rather than adopt it uncritically, and when outcomes are assessed using translation-relevant criteria. The findings emphasize the importance of structured AI translation literacy, post-editing-oriented instruction, proficiency-sensitive support, and more rigorous research designs to assess sustained translation competence.

Text 126369.pdf - Published Version Download (475kB) Official URL or Download Paper: https://ajue.uitm.edu.my/wp-content/uploads/2026/0...

Actions (login required)

View Item