The Effect of AI-Powered Automated Writing Assessment on EFL Students’ Writing Performance

Abstract

This study aims to evaluate the effectiveness of Grammarly as an Artificial Intelligence (AI)-powered Automated Writing Assessment (AWA) tool in improving EFL students' writing performance. A quasi-experimental design was employed, involving 61 ninth-grade students at MTs Al Islah, divided into an experimental group (n=32) receiving Grammarly-assisted feedback and a control group (n=29) receiving traditional teacher feedback. Data were collected through pre-tests and post-tests using an argumentative essay task and analyzed using descriptive statistics and t-tests in SPSS 25. The results showed significant improvement in both groups; however, the experimental group achieved a notably higher mean score (85.50) than the control group (75.00). Statistical analysis revealed a significant difference between the two groups (Sig. 2-tailed = 0.000), suggesting that AI-powered feedback serves as a more potent catalyst for writing improvement than conventional methods. The findings suggest that Grammarly reduces students' cognitive load by providing instant linguistic corrections, allowing them to focus more on content and organization. This study recommends integrating AI tools as a formative assessment in EFL pedagogy to foster self-regulated learning in the digital era.