Using Machine Translation Error Identification to Improve Translation Students’ Post-Editing Skills

Abstract

Attempts to investigate challenges facing translation students in post-editing remain limited in literature. This research investigates a common challenge observed during post-editing training using SDL Trados 2021 which is deciding whether to post-edit MT output or translate from scratch. Making such decisions by translation students may need more focused procedures and criteria to improve this decision-making capability. The current research suggests integrating MT error identification proposed by Daems J, Vandepitte S, Hartsuiker RJ and Macken L (2016) as a basic criterion to undertake post-editing process and tested its effects on improving their performance in post-editing process regarding decision making and time consumed. A quasi- experimental approach was adopted, and the experiment was applied on thirty (30) translation students at university level divided into a control group and an experimental group. A pre-test was given to both groups to identify both performance and time consumed in the post-editing process. MT error identification techniques and Statistical MT error types were introduced in a five-day workshop using a task-based approach to instruct the experimental group. The control group received traditional post-editing training with no focus on MT error types. A post-test was administered to both groups to test the improvement in post-editing skills. The findings showed the significant improvement in students’ post-editing skills. This research would contribute to enhancing translation training courses provided to cope with the continuous advancement in translation technologies and market needs.