Generative AI in Language Instruction: Personalization, Practice, and Proficiency. A Pedagogy-First Framework

الذكاء الاصطناعي التوليدي في تدريس اللغات: التخصيص والممارسة وإتقان اللغة. إطار عمل يركز على التربية

Abstract

Generative artificial intelligence has entered language education with unusual speed, but its pedagogical meaning remains unsettled. This article argues that the educational value of generative AI in language instruction is best understood through three linked commitments: personalization, practice, and proficiency. Personalization refers to the capacity to generate differentiated input, scaffolds, and feedback at a scale that teacher labor alone cannot usually sustain. Practice refers to the expansion of low-stakes opportunities for conversation, writing, revision, and feedback beyond the time limits of the classroom. Proficiency refers to the need to anchor AI-supported tasks to recognized standards of communicative ability rather than to the surface fluency of machine output. The article develops these commitments through a pedagogy-first lens grounded in distributed agency, where learning emerges from the interaction among learner, teacher, and machine, but where the teacher remains the central designer, curator, and evaluator of the learning ecology. Arabic serves throughout as a stress test rather than a side example. As a less commonly taught but strategically important language, Arabic exposes the limitations of current models: uneven dialectal coverage, cultural bias, diglossic complexity, and thin evidence for long-term classroom effects. The article concludes that generative AI offers real affordances for language learning, but only under teacher-mediated conditions that protect authenticity, learner agency, assessment validity, and cultural integrity.