M. N. Kozhina’s Functional Stylistics of Scientific Text and Current Corpus-Based Studies in Detecting Artificially Generated Content

The paper emphasizes the significance of scientific works by the outstanding Russian scientist, Professor at Perm State University M. N. Kozhina, as well as her contribution to linguistics in general and contemporary stylistics in particular. The paper shows the connections between current computer- and corpus-based statistical studies and those of Perm stylistics school of thought, founded by M. N. Kozhina. It also presents the latest corpus-based statistical research results in detecting artificially generated texts and demonstrates the continuity and effectiveness of quantitative studies in language, speech, and communication, as well as their significance for the development of science in general. Statistical data show a steady increase in the use of the open-source Artificial Intelligence tool ChatGPT-4 in scientific publications that started almost immediately after its release on November 30, 2022. The largest and fastest growth in artificially generated content has been noted in publications on computer sciences – up to 17.5%. This fact was established as a result of a systemic large-scale statistical comparison across more than 950,900 papers published in English in leading scientific journals from various academic fields between January 2020 and February 2024, i.e. before and after the release of ChatGPT-4. In abstracts published in computer science journals over the last 14 years (2010-2024), four words demonstrate disproportionately high frequency of use – realm, intricate, showcasing, pivotal. Indicatively, a sudden surge in the use of these words occurred in 2023, about five months after ChatGPT-4 became freely available, while in the period from 2010 to 2022 their use was consistently low. The paper highlights the challenges posed by the use of ChatGPT-4, especially in scientific communication, and the principles to be used for addressing them.