A Structured AI–Human Framework for Machine Translation Post-editing (MTPE) Training

Abstract

This research explores AI-human collaboration within a professional translation course for senior Applied English students (N=43), introducing a hands-on MTPE framework designed to foster translator autonomy. Throughout the course, students engaged in small-group weekly Chinese–English and English–Chinese tasks, documenting their interactions with large language models (LLMs) on Moodle. The pedagogical core consists of a structured Standard Operating Procedure (SOP) including source-text analysis, AI-assisted drafting, critical evaluation, iterative refinement, and justification which encourages systematic post-editing practices. Findings reveal that active engagement with AI, facilitated through precise prompting and iterative error detection, produced translations that were both semantically accurate and stylistically appropriate. In contrast, passive reliance on AI outputs led to weaker linguistic awareness and logical flaws. The study suggests that structured human-AI collaboration, anchored in professional standards, significantly enhances metalinguistic awareness and higher-order translation competence. By emphasizing process-oriented assessment and critical interaction with LLM outputs, this research provides a replicable model for modern AI-integrated translator training. This framework ensures that students develop the necessary skills to navigate the evolving landscape of professional translation while maintaining their role as critical decision-makers in the human-machine loop.