Cognitive Scaffolding Through AI Writing Assistants: Mixed Methods Evidence from Arabic Language Education Students

Abstract

This study examines whether AI writing assistants function as cognitive scaffolds that enhance students’ academic journal writing in an Arabic Language Education program in STAI Masjid Syuhada Yogyakarta. Using a mixed methods quasi experimental design, the research involved 90 final year students divided into an AI assisted writing group and a comparison group using conventional support. The intervention group employed tools such as Chat GPT, Grammarly, and SciSpace throughout drafting and revision, while the comparison group relied on instructor guidance and standard writing resources. Data were collected through pre-test and post-test writing assessments using an analytic rubric, alongside reflective journals and structured questionnaires. Quantitative results indicated a statistically significant improvement in writing quality for the AI assisted group with large effect sizes (Cohen’s d > 1.0). Qualitative evidence suggests that students used AI output to generate ideas, refine outlines, and troubleshoot language, but still required human feedback to ensure rhetorical fit, disciplinary accuracy, and ethical authorship. Overall, the findings position AI writing assistants as cognitive scaffolds that can expand access to timely feedback and promote learner autonomy, while underscoring the need for explicit integrity guidance and instructor oversight.