L’usage de l’IAG pour l’enseignement/apprentissage de la production écrite en FLE/FLS : une revue de portée

Abstract

This scoping review examines the uses of generative artificial intelligence (GAI)—notably ChatGPT and related tools—in the teaching and learning of written production in FLE/FLS. In accordance with PRISMA-ScR, we conducted a “all-source/all-type” search and then applied inclusion criteria based on the PCC (Population-Concept-Context) framework, requiring access to the full text. Recent studies point to several applications: text revision with AI-generated feedback, aimed at improving linguistic aspects (vocabulary, syntax) and coherence; typologies of prompts and responses useful for supporting writing (error detection/explanation, correction suggestions); training programs for student teachers and teachers that use AI-generated texts to design writing instruction (task design, criteria); and case studies documenting autonomy, motivation, and reasoned mastery in French as a Foreign Language (FLE). The reported benefits include linguistic correction, rapid feedback, and support for revision. Recurring limitations include reliance on prompts, variability in the quality of responses, and ethical issues (integrity, disclosure/traceability of use). Gaps remain: a scarcity of rigorous comparative studies by CEFR levels, few longitudinal protocols, and a need for alignment between instruction and assessment. Consequently, we propose approaches to scenario design (prompt engineering, metacognitive instructions, transparency of use) and a research agenda aimed at the responsible integration of IAG into the teaching of written production in FLE/FLS.