Instrumental Competence in Legal Translation: Post-editing Behaviours and Quality of Law and Translation Students

Abstract

With the development of AI technologies, the boundary between professional and non-professional translators has become increasingly blurred. However, the machine-assisted post-editing behaviours of law and translation students in legal translation remain underexplored. Drawing on Prieto Ramos’ Framework of Legal Translation Competence, this study examines the post-editing behaviours of five law students and five translation students who completed unrestricted post-editing of Google-translated legal materials. The analysis first compares the tool-use strategies adopted by the two groups during AI-assisted legal post-editing. Thematic analysis of retrospective interview transcripts identified four recurrent behavioural tendencies: authority validation, context-building, knowledge-driven judgement, and AI-augmented workflow integration. Within this exploratory sample, translation students more often prioritised authoritative resources for contextual verification, whereas law students more often combined domain knowledge with AI tools in comparatively efficiency-oriented workflows. The study then analyses participants’ revisions of five legal terms to examine how instrumental competence interacted with thematic, linguistic, and strategic sub-competences. This term-level analysis is framed primarily through task-specific terminological accuracy, while also distinguishing outright terminological error from contextually misaligned renderings that may remain valid in another legal setting. The findings suggest that problems in post-editing were less closely related to tool choice itself than to difficulties in coordinating multiple sub-competences when revising flawed machine output. These exploratory results highlight the continuing pedagogical importance of integrating language, legal, and contextual knowledge in legal translation education.