Exploring EFL Students' Writing Performance and Their Acceptance of AI-based Automated Writing Feedback

This study examined the EFL writing performance of using AI-based writing feedback called Grammarly and students’ acceptance of this new technology. Fifty-three EFL students from two convenience sample classes in China participated in this quasi-experimental design study. They were randomly assigned to the experimental and control groups. The students in the experimental group (EG) applied Grammarly to edit and revise their essays, while those in the control group (CG) received traditional instruction without Grammarly intervention. A total of five essays were implemented during the 16-week intervention period. EG students' perceptions of Grammarly via a survey were additionally collected by the researchers at the end of the intervention. Results from an independent t-test showed that students in the EG significantly outperformed those in the CG in respect of posttest writing performance. The effect size value of the Cohen's d = 0.603, indicating a medium effect. Results from a survey with open-ended questions showed that students appreciated the AI-based instant grammar correction given by Grammarly. The disadvantages of Grammarly and future teaching strategies in EFL writing classrooms are provided.