Empowering GenAI with a guidance-based approach in MTPE learning: effect on student translators’ cognitive process, final translation quality and learning motivation

Generative AI (GenAI) offers significant potential for machine translation and post-editing (MTPE), yet concerns remain about students’ over-reliance, which may hinder independent thinking and genuine learning. To address this, the study introduces a guidance-based GenAI-assisted MTPE approach informed by previous research. This approach encourages students to articulate their initial thoughts before consulting GenAI and to iteratively refine their translation based on GenAI feedback. A quasi-experimental study involving 30 first-year translation postgraduates was conducted over a seven-week practicum to compare the effect of the guidance-based and traditional GenAI-assisted MTPE on cognitive process, final translation quality, and learning motivation. Data were collected using eye-tracking, think-aloud protocols, quality assessment, and a questionnaire. The results indicate that the guidance-based approach reduces overall cognitive effort while increasing engagement with GenAI and higher-order thinking, leading to a more globally oriented allocation of cognitive resources. In contrast, traditional GenAI-assisted MTPE follows a locally oriented pattern, indicating more passive GenAI use. Compared to traditional methods, the guidance-based approach yields more accurate and fluent translation with lexical richness, syntactic conciseness, coherence, and a wider range of translation techniques, and it significantly enhances students’ learning motivation. These findings provide empirical support for the effective integration of GenAI in translation education.