Egorshkina V.A. Lexico‑syntactic and stylistic specificity of generative media text (on material of audio guides)

The article deals with the analysis of generative media texts created by the DeepSeek neural network for a company that develops audio guides. A brief overview is presented, and a systematization of theoretical studies on the problem of introducing artificial intelligence into the media is carried out. It is clarified that modern foreign and domestic works have not yet developed a methodology for studying generative texts, especially from the perspective of linguistic and stylistic specifics. The work takes into account the achievements of neo‑stylistics, within the framework of which a new direction in the study of generative media texts is being formed. The case study method, which includes the analysis of the prompt and the linguistic and stylistic features of the generated text, has been tested. The material for the study consisted of 117 texts created for audio guides for such cities as Yaroslavl, Sergiev Posad, and Kaliningrad. Conclusions have been drawn regarding the compliance of the obtained texts with the requirements specified in the prompt and with the media materials represented in traditional audio tours. Lexical, syntactic, and stylistic features of generative texts have been identified. It has been found out that they are characterized by schematicity, simplification, standardization, and the presence of unreliable information, while not being devoid of emotionality, and expressiveness.