UDC

[81’42:004.8](045)

DOI

https://doi.org/10.37482/2687-1505-V529

Authors

Natal’ya S. Avdonina^1* – Cand. Sci. (Polit.), Assoc. Prof., Assoc. Prof. at the Department of Journalism, Advertising and Public Relations
Anna A. Postnikova² – Specialist at the Centre for Cultural and Educational Programmes of N.A. Dobrolyubov Arkhangelsk Regional Scientific Library, Master’s Degree Student at the Department of Journalism, Advertising and Public Relations
^{1, 2}Northern (Arctic) Federal University named after M.V. Lomonosov, Arkhangelsk, Russia,
¹ n.avdonina@narfu.ru*, https://orcid.org/0000-0001-9871-3452
² s1611280@edu.narfu.ru, https://orcid.org/0009-0004-2087-6066

Abstract

The active integration of generative neural networks into media practice raises a fundamental question for linguistics regarding the nature of “technological authorship” and the mechanisms of text stylisation. This article examines the linguistic aspects of creating journalistic texts using artificial intelligence (AI). The subject of the study is the linguistic mechanisms and stylistic devices employed by neural networks to imitate various types of media discourse. The material consists of texts generated by the models GigaChat, DeepSeek, Perplexity AI, and QwenChat in Russian based on prompts requiring stylisation in the idiostyles of Joan Didion, Antoine de Saint-Exupéry, and Vasily Grossman. The research employs methods of generative linguistic experimentation, as well as discursive, comparative, and stylistic analysis. The study characterizes the linguistic mechanisms of style imitation at the lexical, syntactic, and compositional levels. Significant differences between the models have been revealed: DeepSeek demonstrated the most accurate reproduction of idiostyles, while GigaChat tended toward clichéd structures, and Perplexity AI and QwenChat showed instability in conveying rhythmic and syntactic patterns. It has been established that the strengths of AI lie in the successful imitation of general stylistic registers and genre forms, whereas its limitations include the occurrence of hallucinations, the averaging of idiostyle, the lack of a “charismatic style”, and the inability to convey existential depth and unique biographical experience. Using the author’s essay “Island of Childhood” as an example, the study demonstrates the areas where generative models fall short: a tendency toward semantic closure, levelling of ambivalent experiences, substitution of unique details with statistically probable clichés, and chronotopic errors. The conclusion has been drawn that contemporary AI stylization represents an imitation of form without intentional “meaningful choice”, which calls for rethinking of the categories of author and style in the digital age. A key imperative becomes human revision of generated text, whose functions include fact verification, restoration of biographical specificity, and semantic correction.

For citation: Avdonina N.S., Postnikova A.A. Linguistic Aspects of Stylised Texts Generation Using Artificial Intelligence Systems. Vestnik Severnogo (Arkticheskogo) federal’nogo universiteta. Ser.: Gumanitarnye i sotsial’nye nauki, 2026, vol. 26, no. 4, pp. 74–85. https://doi.org/10.37482/2687-1505-V529

Keywords

stylisation, journalistic text, artificial intelligence, generative models, idiostyle, prompt engineering, technological author, media linguistics

References

1. Klushina N.I. Idiostyle in a Generative Text. Kommunikativnye issledovaniya, 2025, vol. 12, no. 4, pp. 774–788 (in Russ.). https://doi.org/10.24147/2413-6182.2025.12(4).774-788

2. Kopus’ T.L. Means of Communicating Human and AI-Generated Subjectivity in Telegram Discourse. Filologicheskie nauki. Voprosy teorii i praktiki, 2025, vol. 18, no. 10, pp. 4273–4282 (in Russ.).

3. Vinogradov V.V. Plot and Style: A Comparative Historical Study. Moscow, 1963. 195 p. (in Russ.).

4. Bakhtin M.M. Aesthetics of Verbal Creativity. Moscow, 1979. 423 p. (in Russ.).

5. Kazak M.Yu. Specifics of Contemporary Media Text. Sovremennyy diskurs-analiz, 2012, no. 1, pp. 30–41. (in Russ.).

6. Raspopova S.S. Author as a Real Man and the Author’s Image in Media Texts. Voprosy teorii i praktiki zhurnalistiki, 2015, vol. 4, no. 2, pp. 149–158 (in Russ.).

7. Trofimov R.V. Linguistic Mechanisms for Generating Fake News Using Artificial Intelligence. Vestnik Severnogo (Arkticheskogo) federal’nogo universiteta. Seriya «Gumanitarnye i sotsial’nye nauki», 2025, vol. 25, no. 3, pp. 87–97 (in Russ.). https://doi.org/10.37482/2687-1505-V437

8. Olomskaya N.N., Yurova E.A. Linguopragmatic Features of Generated Text in Social Media Discourse (on the Example of Texts about the 2023 Gubernatorial Election in the Nizhny Novgorod Region, Russia). Rossiyskiy sotsial’no-gumanitarnyy zhurnal, 2025, no. 2, pp. 302–324 (in Russ.).

9. Ng L.H.X., Carley K.M. A Global Comparison of Social Media Bot and Human Characteristics. Sci. Reports, 2025, vol. 15, art. no. 10973. https://doi.org/10.1038/s41598-025-96372-1

10. Kayda L.G. Intermediality as a Philological Continent: The Voltaic Arc of the Author’s Subtext. Moscow, 2019. 112 p. (in Russ.).

11. Maracheva A.V. Creation of Individual Author’s Style as a Result of the Formation of a Future Journalist’s
Methodological Culture in Professional Education. Pedagogicheskiy vestnik, 2021, no. 18, pp. 36–38 (in Russ.).

12. Vyrovtseva E.V., Isakova T.B. Formation of a Journalist’s Author’s Style in the Learning Process. Vestnik
Volzhskogo universiteta im. V.N. Tatishcheva, 2017, vol. 1, no. 2, pp. 125–132 (in Russ.).

13. Dawkins H., Fraser K., Kiritchenko S. When Detection Fails: The Power of Fine-Tuned Models to Generate Human-Like Social Media Text. Arxiv, 2025, 16 Jun. https://doi.org/10.48550/arXiv.2506.09975

14. Bukovskaya A. Saint-Exupéry, or the Paradoxes of Humanism. Moscow, 1983. 208 p. (in Russ.).

15. Bocharov A.G. Vasiliy Grossman. Moscow, 1970. 303 p. (in Russ.).

16. Avdonina N.S., Malakhova A.M., Fokina I.A. Literary Journalism: A Contemporary Reading. Arkhangelsk, 2020. 153 p. (in Russ.).

17. González-Arias C., Chatzikoumi E., López-García X. The Anthropomorphic Pursuit of AI-Generated Journalistic Texts: Limits to Expressing Subjectivity. Front. Communication, 2024, vol. 9, art. no. 1456509. https://doi.org/10.3389/fcomm.2024.1456509

18. Macko D., Kopal J., Moro R., Srba I. MultiSocial: Multilingual Benchmark of Machine-Generated Text Detection of Social-Media Texts. 2024. https://doi.org/10.48550/arXiv.2406.12549