LINGUISTIC MARKERS OF DISTINCTION BETWEEN “MACHINE” AND “HUMAN” ACADEMIC STYLE IN RUSSIAN-LANGUAGE SCIENTIFIC TEXTS

Abstract

The widespread use of generative language models in academia has raised a pressing practical question: what linguistic features can distinguish machine-generated scientific text from human-written text? This article systematizes the markers of “machine-generated” academic style in Russian-language scientific texts based on an analysis of research conducted between 2020 and 2025. It is shown that machine-generated texts in Russian-language academic discourse more often use similar syntactic constructions, repeat keywords, and less frequently include particles and introductory words that mark the author’s position. Classifiers based on these features have so far been developed primarily for English, while Russian-language data are described fragmentarily. The article discusses possible steps for developing a corpus of Russian-language scientific texts and refining criteria for editorial and authorship assessment.