Beyond Proficiency-Matched Support: A Multidimensional Framework for Calibrating Generative AI Scaffolding in Arabic L2 Writing

Abstract

This article proposes a multi-dimensional instructional-design framework for scaffolding Arabic L2 writing using a structured integrative conceptual synthesis of scaffolding research, GenAI/L2 writing studies, evidence on Arabic L2 writing challenges, and Arabic writing-technology precedents. The framework suggests three diagnostic inputs: Learner Proficiency (P), Arabic-Specific Linguistic Demand (L), and Task/Stage Demand (T). Their joint consideration informs five distinct support decisions: explicitness, scope, initiative, verification, and fading. The framework separates language proficiency from GenAI literacy, linguistic demand from the overall Arabic difficulty, task stage from task complexity, and considers engagement, self-regulation, prior knowledge, and GenAI literacy as boundary conditions instead of further core axes. The suggested relationships are conceptual design propositions rather than empirically proven interactions or dosage rules. Further studies are needed to validate P–L–T relationships and support configurations across different Arabic learner groups, linguistic targets, and writing tasks.