A scoping review of empirical studies on generative artificial intelligence in language education

Purpose:Artificial Intelligence (AI) has long played a crucial role in language education, and the emergence of Generative AI (GAI) is set to further enhance this impact. However, comprehensive reviews of high-quality empirical research on GAI in language education are lacking.Design: To fill this gap, the current study analyzed 43 empirical studies published in SSCI journals from 2022 to 2024, focusing on contextual, technological, theoretical characteristics as well as research objectives.Findings and Originality: The findings reveal both progress and ongoing challenges in GAI research. Contextually, most of the selected studies concentrated on English language teaching and learning, particularly among Chinese learners and teachers in higher education, with a strong emphasis on writing skills. Technologically, previous studies exhibited a notable overreliance on ChatGPT, insufficient details on GAI versions, and inadequate reporting of GAI prompts. Theoretically, 46.5% of the selected studies did not specify a clear framework, while those that did mainly drew from psychological, technological, and social theories. As for research objectives, 41.9% of the selected studies examined users' perceptions, while the rest evaluated AI impact, explored AI practice, and compared AI performance against other tools.This review concludes with six key recommendations to advance GAI research in language education: (1) Expand participant diversity; (2) Explore AI models beyond ChatGPT; (3) Examine prompt patterns to enhance AI literacy; (4) Integrate theoretical frameworks in GAI research; (5) Investigate cross-validation skills for enhanced critical thinking; (6) Leverage GAI's multimodal capabilities to explore language skills beyond writing.