Web, GenAI, or hybrid? Effects of information-seeking conditions and text types on MTPE process, product, and perception

In translation contexts, information seeking (IS) refers to translators’ use of external resources to retrieve, verify, or evaluate information. IS has commonly relied on web-based resources, while GenAI has recently emerged as a novel IS tool. This study examines the effects of web-based, GenAI-based, and hybrid IS conditions on the machine translation post-editing (MTPE) processes, products, and perceptions of 78 trainees across general and domain-specific texts. The hybrid IS condition combined web-based and GenAI-based IS, reflecting real-world translation workflows in which both resource types may be used together. Results showed that GenAI-based support offered no speed advantage over web-based IS, with time-on-task shaped primarily by text type. IS behaviours varied across conditions: web-based IS prioritised lexical searches, GenAI-based IS involved more translation- or refinement-oriented requests, and hybrid IS featured verifying of GenAI outputs through web-based sources. IS conditions significantly affected translation quality, with the GenAI-based and hybrid groups outperforming the web-based group, while text type influenced accuracy only. Perceptions reflected mixed challenges, including revision uncertainty with web-based IS and difficulties interpreting inconsistent GenAI feedback. These findings suggest that incorporating a hybrid IS condition may better prepare students for GenAI-assisted workflows by fostering critical thinking and strategic tool use.