Integrating generative AI in EFL classrooms: A mixed-methods study on argumentative writing performance, writing...

Writing in English as a Foreign Language (EFL) is widely recognized as one of the most challenging skills for language learners, and recent advances in generative artificial intelligence (GAI) tools offer new possibilities for supporting EFL writing instruction. This study investigates how GAI-assisted process-genre instruction affects Chinese EFL learners' argumentative writing performance, motivation, and enjoyment. By comparing an experimental group using ERNIE Bot for brainstorming, feedback, and revision to a control group receiving traditional product-based instruction, the study aims to determine whether GAI support leads to greater improvements in writing performance and critical thinking. Additionally, it explores how GAI influences students' enjoyment and motivation, providing insight into how these affective factors contribute to writing success. Quantitative analyses of pre-, post-, and delayed post-tests were conducted using independent writing tasks, without the assistance of ERNIE Bot, to ensure that the effect sizes measured internalized writing skills rather than performance aided by AI. Alongside the tests, questionnaires and semi-structured interviews reveal that the experimental group achieved far greater gains across all measures, with medium to large effect sizes. Qualitative findings further highlight students' increased confidence, critical engagement, and positive perceptions of AI, although some concerns about overreliance and accuracy were raised. The results suggest that GAI meaningfully enhances both cognitive and affective dimensions of EFL writing when embedded within a process-genre framework. These findings, interpreted through the Technology Acceptance Model, the L2 Motivational Self System, and Paul-Elder's Critical Thinking Model, offer theoretical insights and pedagogical implications, demonstrating how GAI can support brainstorming, revision, and motivation while fostering AI literacy and encouraging critical use.