The Use of AI-Powered Grammarly in Enhancing EFL Prepositions Usage

Abstract

Despite growing evidence supporting the use of Artificial Intelligence (AI)-powered automated written corrective feedback (AWCF) in English as a Foreign Language (EFL) writing, limited research has examined its effectiveness in learning context-dependent grammatical structures, especially English prepositions. This longitudinal quasi-experimental study investigated whether AI-powered Grammarly, supported by teacher mediation, improves EFL pre-service teachers' accuracy in using the English prepositions in, on, and at. Fifty first-year EFL pre-service teachers from a private university in southern Ecuador participated in this study. Two pre-existing groups received either Grammarly-supported instruction with teacher mediation or conventional grammar instruction over an academic term. Data were collected through pre- and post-test assessments, instructional performance tasks, a learner perception questionnaire, and classroom observations. The experimental group achieved significantly greater learning gains than the control group (t (31.9) = 2.91, p .007, d =0.823). A linear mixed effects model further revealed a significant effect of the instructional condition across the intervention (F (1, 216) = 136.30, p < .001). Learners also reported highly positive perceptions of Grammarly, particularly regarding its usability, feedback clarity, and support for learning English prepositions. The study concludes that teacher-mediated Grammarly instruction was associated with greater accuracy in using in, on, and at and was positively perceived in terms of usability and feedback clarity.