Background: The introduction of a research article plays a central role in shaping scientific argumentation. However, this section is often especially prone to stylistic overload, which can obscure the clarity of the author's position. While the issues of redundancy and wordiness have been broadly acknowledged in applied linguistics, there is still limited understanding of how these features are distributed in relation to rhetorical structure, particularly within Russian-language academic texts. Purpose: To identify rhetorically sensitive areas of stylistic overload in the introductions of Russian-language research articles in the field of education. Method: The analysis is based on a corpus of 40 introductions from empirical articles published in 2024 in leading Russian peer-reviewed journals in education. The rhetorical Move-Step model developed by Swales was used as the framework for annotation. Each fragment was manually coded for two dimensions: the type of deviation (wordiness or redundancy) and its communicative impact (according to the IMPACT scale). Pearson's chi-squared test was used to assess statistical significance. Results: Stylistic overload was found to cluster in specific rhetorical steps, especially those related to establishing the importance of the topic (M1_S2), identifying gaps in the literature (M2_S1), and stating research objectives (M3_S2). The most frequent features included syntactic overcomplexity, vague abstract nouns, and overused credibility markers. A high level of negative communicative impact (IMPACT = HIGH) was observed in 60 fragments, most of which were located in the mentioned segments. Statistical testing (χ², p < 0.0001) confirmed a significant relationship between rhetorical function and the type of deviation. Conclusion: The results confirm that stylistic overload in introductions is not accidental but structurally motivated. This supports the need for rhetorically informed strategies in teaching academic writing. The annotation scheme developed in the study may be applied in future corpus-based analyses of academic Russian.