Generative AI-enhanced self-study materials for TOEFL-style in-class integrated writing test: student perceptions of skill development and material quality in an EMI university

Abstract This study investigates the effectiveness of generative AI (GenAI)-enhanced reading and listening materials, complemented by AI-generated writing samples, in developing students’ summarising and synthesising skills for an integrated writing test in an English for Academic Purposes (EAP) context. Findings revealed a statistically significant improvement in students’ writing performance from pre- to post-tests, with students who engaged more consistently showing higher scores than their peers. Questionnaire and interview data confirmed perceived skill gains in areas such as summarising, synthesising, and overall writing structure, alongside increased motivation and confidence. Despite positive impacts, concerns about unnatural language, vocabulary complexity and grammatical homogenisation underscore the need for careful adaptation by teachers. This research highlights the potential of well-implemented GenAI-enhanced tasks to support academic writing development when thoughtfully integrated into EAP instruction.