EFL Writing using AI Feedback: A Study of Effects on Syntax and Lexis

Abstract

This study examines the effectiveness of AI-powered adaptive feedback in learning and its impact on improving the syntactic and lexical sophistication of tertiary English as a Foreign Language (EFL) learners. Although technology-enhanced language learning (TELL) classrooms have flourished immensely in recent times, there is little empirical evidence on the differential effects of computer-based intelligent feedback on language production. Drawing on the principles of Sociocultural Theory and Interaction Hypothesis, this research adopted a quasi-experimental approach with a sample of 90 undergraduate EFL students who were randomly allocated to AI-assisted and traditional feedback groups for a 12-week teaching cycle. Computational linguistic measures (Coh-Metrix and the Tool for the Automatic Analysis of Lexical Sophistication (TAALES)) were employed to collect data on syntactic and lexical indicators from the pre- and post-test phases. Results indicated statistically significant improvements among participants in the AI-assisted feedback condition in clause density, subordination index, and type-token ratio over those who received conventional feedback. The study concluded with the prediction that AI-induced metalinguistic awareness is a key mediator in the writing process and should be adopted in EFL classrooms. The study results have profound implications for curriculum designers, EFL teachers, and educational technologists exploring the use of intelligent tutoring systems (ITS) in EFL writing analysis and pedagogies, especially in low-resource educational institutions in Saudi Arabia.