AI-Assisted Scaffolding for Inclusive CLIL in Higher Education

Résumé

The CLIL courses in universities frequently combine students of varying levels of academic and English literacy. Consequently, not all learners engage in reading and are unable to communicate complicated ideas when studying disciplinary content using English. This paper examines whether AI applied in a narrow and teacher-directed form can serve more inclusive CLIL by scaffolding both content and academic language differentially. The research will be carried out in a six-week CLIL module of an undergraduate course in which the instruction will be in English. Two intact groups with around 50-70 students will be subjected to the same syllabus, activities and the same standards of assessment. The support package of the experiment group will be a curated AI support package monitored by the instructor. It will offer adaptive glossary of important terms, concept-check prompts, levelled language frames in discussion and writing and feedback prompts which are based on a CLIL-oriented rubric. The conventional scaffolding of the comparison group will be the use of instructor-prepared materials without AI. It will consist of pre- and post-measures of content comprehension, academic vocabulary task which is discipline specific, and rubric based assessment of student writing. The indicators of participation will also be analyzed, i.e. frequency of contributions and the use of academic language in the classroom. To clarify the experience of using the scaffolds and what assistance proves the most valuable, short student and teacher interviews will be used. It is assumed that AI scaffolding through the teacher will enhance conceptual learning, reinforce academic language in student writing, and engage more learners with lower proficiency. The research provides useful suggestions on the responsible introduction of AI in the university CLIL with focus on academic integrity and quality control.