Exploring AI-Generated Feedback for Formative Assessment in EFL Academic Writing in a Digital Context

This dataset was collected as part of a study examining the use of AI-generated feedback as a support for formative language assessment in English as a Foreign Language (EFL) academic writing within a formal digital learning context. The participants were 142 EFL students enrolled in an English for Academic Purposes (EAP) course. The study employed a mixed-methods approach using a concurrent embedded design. Data were collected from four primary sources. First, pre- and post-test academic writing samples were obtained to examine changes in students’ written performance following engagement with AI-generated feedback. Second, weekly reflective comments were collected over a ten-week instructional period to capture learners’ ongoing engagement with, and responses to, AI feedback. Third, survey data were gathered to document students’ participation in engagement-oriented classroom activities, including collaborative work, blog-based participation, and reflective commenting. Finally, responses to open-ended questions were collected to explore learners’ perceptions of the affordances and limitations of AI-generated feedback in an online instructional setting. Together, these data sources provide quantitative and qualitative evidence relating to students’ writing development, engagement patterns, and evolving attitudes towards AI-mediated feedback. The dataset supports analyses of changes in specific aspects of academic writing, including task fulfilment, lexical range, cohesion, and overall organization, as well as learners’ perceptions of both surface-level and higher-order dimensions of writing. The data are intended to inform research on the pedagogical integration of generative AI tools in formal digital EFL learning environments.