Characteristics of the generated text: linguistic and social-communicative analysis
Автор: E.V. Osetrova, A.V. Sedova
Журнал: Сибирский филологический форум @sibfil
Рубрика: Языкознание: новая реальность
Статья в выпуске: 2 (31), 2025 года.
Бесплатный доступ
Statement of the problem. In the modern information space, a huge array of texts is distributed and circulated in the Internet environment. This process has significantly changed the ways in which people interact and, with the development of scientific and technological progress, has led to the emergence of artificial intelligence (AI). The latter’s ‘abilities’ for self-learning and dialogue, especially in the field of verbal communication, have raised the question of identifying generated text against the background of natural texts created by humans, and in comparison with them. The study of the virtual ‘field’ of language application in the text generation mode is becoming relevant, correlated with the demands of society. K.A. Studenikina, N.A. Bagrova, R.E. Telpov and S.V. Lartsina, A.R. Aidagulova and others study them in order to identify a set of specific features, and study the neural network for its linguistic potential. The purpose of the study is to calculate the linguistic characteristics of texts generated by artificial intelligence, expanding their list, and to determine the prospects for analyzing the communicative framework – the circumstances and process of their production. Research methodology. The authors of the article analyzed artificial texts generated on one of the topics presented in the educational and methodical complex edited by M.M. Razumovskaya (8th grade). The material for observations included 12 texts in the essay genre created by ChatGPT4 bots, the analysis was provided by the method of factor analysis of the speech genre, the methods of topic- comment analysis and semantic analysis of the text / utterance. Research results. Thus, the list of currently known features of the generated text has been expanded by its five characteristics of grammatical, topic-comment and genre nature. The analysis of the organization of communication in the process of text generation, in turn, revealed the heterogeneity of the composition of its participants (human and ‘AI communicant’); the transformation of the user’s initiative remark into the prompt genre; the presence of a mode of assumption as a signal about the potential possibility of alternative answers in the AI communicator’s remarks; the prospects of applying the theory of communicative failures to the object under consideration. Conclusions. The list of currently known features of the generated text has been expanded with five characteristics of grammatical, thematic, and genre nature, the presence of which is illustrated by linguistic material.
Artificial intelligence, Neural network, Chatbot, ChatGPT, Generated text, Language, Communication, Communication failures
Короткий адрес: https://sciup.org/144163360
IDR: 144163360