Effects of Interactive Machine Translation on Post-Editors’ Cognitive Resource Distribution: Evidence from Eye-Tracking, Key-Logging, and Retrospective Interviews

With real-time AI predictions, interactive machine translation (IMT) offers a promising approach to post-editing (PE). While previous studies have focused on speed and user experience, this study investigates how IMT affects post-editors’ cognitive resource distribution (CRD) across task sub-phases (initial, main, check) and activities (ST processing, MT processing, TT processing, consultation, typing). Data were collected via eye-tracking, key-logging, and retrospective interviews from twelve professional and twelve student translators with sufficient IMT experience, who performed both conventional PE (CPE) and interactive PE (IPE) on Chinese-English creative texts. Results indicate that IMT’s effects are skill-dependent: professionals show reduced CRD only in TT processing, whereas students exhibit lower CRD in all sub-phases and activities except the initial phase. IPE also narrows the cognitive-efficiency gap between professionals and students. The underlying factors contributing to these differences are analyzed. The findings have important implications for translation process research, the translation industry, translation pedagogy, and the design of IMT.