Abstract
This study explored the effects of three features of translator-computer interaction modes (translation resources, text segmentation and task roles), together with student educational levels, on textual coherence quality in Chinese to English translations. Fifty-two students at bachelor’s and master’s levels were divided into 26 pairs to complete four translation assignments in a classroom setting. Of the total, 35 submitted individual reflections on the challenges and strategies adopted for textual problems in the assignments. The translations were coded for five aspects of translation qualityccuracy, Completeness, Coherence, Normativity, and Appropriateness. After we discussed the overall translation quality and error counts along the five dimensions, Generalised Logistic Regression Models of textual coherence quality indicated that only translation resources had consistent main effects on the three textual coherence quality variables: coherence error presence, frequency, and severity. A significant interaction effect between translation resources and task roles was observed on textual coherence error presence. Educational levels had significant interaction effects with translation resources and text segmentation, respectively, on textual coherence error severity. The follow-up qualitative results supported these findings that students’ ability to detect coherence errors varied due to their different strategies to address textual problems. The study concludes by discussing the methodological implications for controlling source text coherence features and theory-informed manipulation of educational levels in the research agenda on TCI modes.

This work is licensed under a Creative Commons Attribution 4.0 International License.
Copyright (c) 2026 Xinyuan Liu, Weiqing Xiao, Sandra Louise Halverson

