Auteurs : قديري جميلة . Bennaceur Hanifi .
Abstract: The rapid integration of large language models (LLMs) into higher education is reshaping academic writing instruction in ways still poorly understood. In Arabic-speaking universities, where writing instruction has traditionally rested on a corrective paradigm centered on lexical accuracy and stylistic conformity, tools such as ChatGPT raise questions that go beyond plagiarism. When students can generate texts whose fluency exceeds their own production, what remains of the teacher's authority as corrector? And what becomes of the argumentative competencies academic writing is supposed to cultivate? Drawing on Hyland's work on academic discourse, Vygotsky's sociocultural theory of mediation, the Arabic-Islamic tradition of ḥijāj reformulated by Taha Abderrahmane, and recent empirical work on LLMs in second-language writing, this article argues that the teacher's role must shift from a corrective-normative pole toward a mediating-critical one. It proposes a four-posture model of the mediating teacher and outlines four transferable pedagogical sequences.
generative AI ; academic Arabic writing ; pedagogical mediation ; ḥijāj ; second-language