Scientometric Analysis of AI in Applied Linguistics: A CiteSpace Tutorial with GPT-Based Cluster Labeling

Scientometrics is the study of research trends using quantitative methods to map how knowledge develops over time. It offers robust techniques for examining the intellectual structure and growth of academic fields. However, procedural guidance remains scarce in applied linguistics. This study addresses this gap by providing step-by-step guidance for conducting scientometric analysis using CiteSpace. We analyze 4,819 Scopus-indexed publications in applied linguistics research from 2015 to 2025 that used artificial intelligence (AI) in general. We demonstrate the use of key scientometric metrics, software functions, and analytical settings, and guide readers through the workflow. We then showcase each step using document co-citation analysis (DCA), which identified 15 major thematic research clusters. After the clustering stage, we demonstrate how to call OpenAI’s GPT through the CiteSpace interface to mine and label the clusters produced by the DCA algorithm on a reduced sample. We compare GPT-based and log-likelihood ratio (LLR)-based labeling methods and show that the former method offers more interpretable cluster labels, although it can encounter token-capacity limitations with large datasets. The study provides practical recommendations for metric selection, parameter configuration, and result interpretation. We conclude by recommending several potential research topics for future scientometric analyses of AI in applied linguistics research.