Yen-Yu Lin
ABSTRACT
This study investigated the impact of using machine translation (MT) and post-editing training on the revision process of thirty intermediate-level EFL learners through a single-group pre-post design. The participants identified specific grammar and semantic errors in texts generated by Google Translate (GT) during the post-editing training. Before and after the training, the students translated their L1 writing into L2 without using GT and then edited their L2 writings by comparing them with the GT translations. Data were collected from various sources, including writing outcomes, screen recordings, perception surveys, and interviews. The results showed that there was a significant difference in word length and word diversity between students’ revised texts and L2 texts in both the pretest and posttest. In addition, the error rates in the posttest were much lower than those in the pretest. Moreover, the content similarity rate was found to negatively correlate with the error rates in students’ revised texts. The survey revealed that students expressed moderate to high satisfaction with the overall quality of texts generated by GT. The study presents implications for utilizing MT as a support for EFL students’ writing along with discussing ways for EFL teachers to incorporate MT into the classroom given its increasing demand.
Key words: machine translation, L2 writing, post-editing, revision
DOI: 10.30397/TJTESOL.202510_22(2).0002
 02_A Multifaceted Exploration of AI-Powered Google.pdf | [ ] | 499 kB |