Moderate evidence
More
Students taught to edit AI translations scored higher, a 82-student study finds
Study title: Research on the Transformation of College English Translation Teaching Driven by Artificial Intelligence
- Evidence type
- Controlled or validated comparison
- Sample
- n = 82
- Sources
- 1 sourcePeer-reviewed
- Published
- 2 min read
- Truvace record
- TRV-2026-2941cite or check it
Study title: Research on the Transformation of College English Translation Teaching Driven by Artificial Intelligence
A quasi-experimental study ran a 12-week teaching intervention with 82 non-English major undergraduates at one university. The experimental class followed a four-stage human-machine model: AI pre-translation, comparative diagnosis, post-translation editing, and reflection and summary.
Read the study abstract
In response to the current predicaments faced by university English translation teaching in the era of artificial intelligence, such as narrowed goals, fixed content, monotonous models, lagging evaluation, and the absence of AI literacy among teachers and students, this study has constructed a transformation analysis framework covering five dimensions: goal positioning, content organization, teaching model, evaluation method, and teacher-student roles. Based on this framework, a four-stage process-oriented teach…
The gain
In a 12-week quasi-experiment with 82 non-English major undergraduates, the class taught a four-stage AI pre-translation and editing method scored significantly higher on post-test translation and post-translation editing than the control class.
In brief
The experimental class beat the control class on post-test translation scores and post-translation editing ability. It also improved on AI usage strategies, accuracy of critical judgment, intrinsic motivation and perception of task value.
The authors built the model from a five-dimension framework covering goals, content, teaching model, evaluation and teacher-student roles. The evidence comes from one university and one course, so it does not show whether the results hold elsewhere.
Main points
- The study used a quasi-experimental design with 82 non-English major undergraduates over 12 weeks.
- The four stages were AI pre-translation, comparative diagnosis, post-translation editing, and reflection and summary.
- Post-test translation scores of the experimental class were significantly better than the control class.
- Post-translation editing ability was also significantly better in the experimental class.
- The experimental class improved on AI usage strategies, critical judgment accuracy, intrinsic motivation and task value perception.
- Qualitative data showed students became more aware of strategies, repositioned human and machine roles, and monitored their own thinking more closely.
The full story
The study responds to problems the authors identify in university English translation teaching: narrowed goals, fixed content, monotonous models, lagging evaluation, and teachers and students lacking AI literacy. It builds a framework across goal positioning, content organization, teaching model, evaluation method and teacher-student roles.
Based on that framework, it tested a process-oriented model for human-machine collaboration. The 12-week intervention involved 82 non-English major undergraduates, compared against a control class.
Beyond test scores, the experimental class improved on AI usage strategies, critical judgment accuracy, intrinsic motivation and task value perception. Qualitative analysis supported these shifts, and the authors conclude that structured collaborative training can lift translation quality, process strategy and learning motivation together.
Sources
1 on file- Journal of Education and Culture Studiesdoi:10.22158/jecs.v10n4p46
Talk it over
The experimental class followed a four-stage model: AI pre-translation, comparative diagnosis, post-translation editing and reflection and summary. Which stage do you think did the most to raise scores, and why?
Students taught to edit AI translations scored higher, a 82-student study findshttps://truvace.com/item/research-on-the-transformation-of-college-english-translation-teaching-driven-by
The experimental class followed a four-stage model: AI pre-translation, comparative diagnosis, post-translation editing and reflection and summary. Which stage do you think did the most to raise scores, and why?
Answer in the threadThe study ran for 12 weeks with 82 non-English major undergraduates at one university, comparing an experimental class with a control class. What could differ between those classes besides the method, and how might that change the result?
Answer in the threadWho benefits most from learning to edit AI translations, and which students, beyond these non-English major undergraduates, might gain less from this model?
Answer in the thread
If you have more time
Students became more aware of strategies and repositioned the roles of human and machine. What would you ask students themselves to learn whether that shift lasts after the 12 weeks?
Answer in the threadWhat result from a different university or a different course would settle whether this four-stage model works beyond the one course the authors tested?
Answer in the thread
ace
What readers say