<![CDATA[Generative artificial intelligence in Arabic language education requires learning designs that preserve student agency rather than merely accelerate answer production. This article formulates a pedagogical model for using Claude as a bounded cognitive scaffold in university-level Arabic learning, with emphasis on critical AI literacy, self-regulated learning, and process-oriented assessment. The study applies an integrative literature review combined with an affordance analysis of Claude. Sources were purposively selected from 2023-August 2026 publications on generative AI, language learning, Arabic education, assessment, and human-AI collaboration, supplemented by foundational Arabic linguistics and pedagogy and official Anthropic documentation. The synthesis identifies five principles: source-grounded reading, layered grammatical verification, reason-based writing revision, a PLAN-ASK-TEST-REVISE-REFLECT self-regulation cycle, and assessment of process traces. Claude is most useful when it supports text interrogation, comparison of linguistic alternatives, and feedback that students must verify. Its risks include fluent but inaccurate output, cognitive offloading, unequal paid access, and opaque AI assistance. The resulting framework positions human verification, task design, and accountable linguistic reasoning as the conditions for pedagogically responsible use.]]>
Copyrights © 2026
