1. Introduction

This section discusses the close relationship between education and technology, particularly the impact of artificial intelligence on education. In 2022, AI technology achieved significant breakthroughs, especially with the emergence of generative large language models like ChatGPT, which garnered attention in the education sector due to their text comprehension and generation capabilities. On CNKI, the number of related publications surged to 12,420. Foreign language teaching, as a field highly reliant on text and interaction, benefits from the new technical support provided by large language models. This study employs bibliometric methods to analyze literature on the application of large language models in university foreign language teaching from November 2022 to January 2025, aiming to outline current trends, predict research directions, and explore their potential in foreign language teaching to provide insights for educational reform.

2. Data Sources and Research Methods

This chapter introduces the data sources, which include CSSCI and Peking University Core Journals from CNKI, with search keywords such as "ChatGPT," "large language models," and "generative AI," limited to the fields of education and language. After screening, 52 valid articles related to foreign language teaching in higher education were obtained. Research methods involve using Excel to summarize literature information and conducting bibliometric analysis from perspectives such as publication trends, journals, authors, institutional and regional distribution, and research hotspots. The goal is to identify key applications of large language models in foreign language teaching and provide references for future research.

3. Research Results and Analysis

The application of large language models in foreign language teaching has attracted widespread attention, as evidenced by high download and citation counts, indicating strong research interest. Since the release of ChatGPT, 52 related studies have been published in the field of foreign language teaching, demonstrating active research activity. Publication trends show a minor peak in 2023, followed by fluctuations, reflecting sustained interest in large language model applications. The Chinese government's emphasis on educational digitization has further driven related research. Analysis of core journals reveals that 21 journals published the 52 articles, with Foreign Language Learning Technology and Instruction and Foreign Language World being the most prolific. Author analysis indicates that a high-output author group has yet to form, with collaborative research being the dominant approach. Regional and institutional analysis highlights the leading role of universities in Beijing and Shanghai. Research hotspots include the impact of large language models on foreign language teaching, applications in writing instruction, and translation teaching. Specific topics cover ChatGPT's influence on foreign language education, innovative teaching models, practical applications of large language models, academic English writing instruction, college English writing teaching, and translation pedagogy. While existing studies discuss the advantages and challenges of large language models, experimental teaching research remains limited, and research dimensions need further expansion. Future studies should explore technological features and their practical applications in teaching to meet the demands of the digital-intelligent era and talent cultivation.

4. Conclusion

This chapter summarizes the bibliometric study based on CSSCI and Peking University Core Journals from CNKI, finding that the application of large language models in foreign language teaching has garnered significant attention, with research primarily published in 13 core foreign language journals. Studies are relatively concentrated in terms of institutions and regions, with themes focusing on the impact of large language models on foreign language teaching, writing instruction, and translation pedagogy. The research is characterized by rich theoretical discussions but insufficient empirical studies, limited dimensions, and an overall early-stage status, requiring further exploration. Data source limitations mean the findings may not fully represent all scenarios.