Comparing Arabic and English instruction in generative AI-supported teacher education: A quasi-experimental study
DOI:
https://doi.org/10.18488/61.v15i1.5359Keywords:
Arabic-medium instruction, English-medium instruction, Generative artificial intelligence, Language of instruction, Pedagogical competencies, Professional performance, Quasi-experimental design, Teacher preparation.Abstract
This study examined the effect of the language of instruction in generative artificial intelligence (AI)-supported learning environments on pre-service teachers’ pedagogical competencies and professional performance. A quasi-experimental design with two equivalent groups and pre- and post-measurements was used. The sample consisted of 64 professional teaching diploma students divided equally into an Arabic-medium group (n = 32) and an English-medium group (n = 32). Over eight weeks, both groups studied the same pedagogical content and completed comparable generative AI-supported activities; the language of instruction represented the experimental variable. Data were collected using a pedagogical competencies scale and a professional performance observation checklist. Pre-measurement results confirmed equivalence between the groups. Post-measurement findings showed statistically significant improvement in both groups, indicating the usefulness of structured generative AI-supported instruction in teacher preparation. However, the Arabic-medium group achieved higher post-test scores in pedagogical competencies and professional performance, whereas the English-medium group performed better in pedagogical terminology and AI prompt interaction. The findings suggest that Arabic may support deeper contextualized pedagogical understanding and professional application, while English may strengthen access to technical terminology and prompt design. The study recommends a moderated bilingual model that uses Arabic for conceptual and professional application and English for terminology, international resources, and generative AI interaction.
