The Impact of Ideal L۲ Writing Self, L۲ Writing Self-Efficacy, AI Literacy, and AI-Assisted Feedback on EFL Learners’ L۲ Writing Achievement

The rapid advancement of generative artificial intelligence (AI) has transformed second language (L۲) writing instruction by enabling learners to receive immediate, individualized, and interactive feedback. While a growing body of research has examined AI-assisted writing from pedagogical and technological perspectives, comparatively little attention has been paid to the interplay between learners’ motivational dispositions, self-beliefs, AI literacy, and AI-supported feedback in shaping L۲ writing achievement. Addressing this gap, the present study investigated the combined effects of Ideal L۲ Writing Self, L۲ Writing Self-Efficacy, AI Literacy, and AI-assisted feedback on Iranian EFL learners’ writing performance. A quasi-experimental pretest–posttest design was employed with ۶۰ upper-intermediate-level Iranian EFL learners in the field of TEFL enrolled in an essay writing course at a higher education institute in Iran. Participants were randomly assigned to either an experimental group receiving AI-assisted written feedback generated through ChatGPT under teacher guidance or a comparison group receiving conventional teacher-written corrective feedback throughout a ten-session writing intervention. Prior to the treatment, participants completed three validated instruments: the L۲ Writing Selves Questionnaire (Tahmouresi & Papi, ۲۰۲۱) to assess Ideal L۲ Writing Self, the Second Language Writer Self-Efficacy Scale (Teng, Sun, & Xu, ۲۰۱۸) to measure learners’ confidence in performing L۲ writing tasks, and the Artificial Intelligence Literacy Scale (AILS) (Carolus et al., ۲۰۲۳) to evaluate learners’ knowledge, competencies, and critical understanding of AI technologies. L۲ writing achievement was measured through IELTS argumentative writing tasks administered before and after the intervention and scored independently by two trained raters using the official IELTS analytic writing descriptors. Independent-samples t-tests were conducted to compare the effectiveness of the two feedback conditions, whereas Pearson product–moment correlations and multiple regression analyses were performed to examine the predictive contributions of the learner-related variables. The findings indicated that learners receiving AI-assisted feedback significantly outperformed those receiving conventional teacher feedback on the posttest. Furthermore, Ideal L۲ Writing Self, L۲ Writing Self-Efficacy, and AI Literacy each demonstrated significant positive relationships with L۲ writing achievement. Multiple regression analyses revealed that L۲ Writing Self-Efficacy