Generative AI and large language models (LLMs), such as ChatGPT, have transformed many working practices, including scientific writing. However, writing styles between LLMs and sci-entists have been found to differ, particularly in terms of word frequencies. Using a list of 16 stylistic words that are associated with AI use, we examine k = 26,010 published abstracts in the top 100 journals in linguistics research from 2020 to 2024. A significant rise of 28% in the relative frequency of 12 target words was found exclusively in 2024, suggesting a recent in-crease in LLM use. In particular, the words delve, enhancing, and pivotal saw significant in-creased use in 2024. Furthermore, higher-prestige journals exhibited slightly greater AI-associated word frequency. Country-level differences indicated particularly higher AI-word us-age in abstracts from China, South Korea, and Iran. While relative word frequencies serve only as a proxy for LLM use, the findings raise crucial questions about transparency, equity, and ethics in academic publishing.
Botes et al. (2025) studied this question.