Memuat jurnal…
Memuat jurnal…
This study examines the use of Large Language Models (LLMs) in Arabic language learning in Indonesia, addressing the gap between their theoretical potential and actual classroom practices. The study aims to analyze usage patterns, identify implementation challenges, and explore pedagogical implications. A hybrid design was employed, integrating a Systematic Literature Review (PRISMA-informed) of Scopus-indexed studies (2025–2026) with exploratory empirical data from a survey of Indonesian Arabic language students. The findings reveal that LLMs are predominantly used for translation, writing assistance, and grammar support, offering benefits such as instant feedback, learning efficiency, and personalization. However, significant challenges emerge, including output inaccuracies, linguistic bias, over-reliance, and limited verification practices. Empirical results indicate that students tend to use LLMs pragmatically for task completion rather than deep learning, leading to potential reductions in critical thinking. The study concludes that while LLMs hold transformative potential, their pedagogical integration remains underdeveloped. Effective implementation requires structured instructional design, AI literacy development, and guided usage to balance efficiency with meaningful learning outcomes.
India
United States
Singapore
France
Hong Kong
Italy
South Africa
Oman
Poland
CO