Компьютерное моделирование как инструмент анализа художественного текста

Филологический класс. 2023. Т. 28. № 2. С. 22–33.

Kolmogorova A., Залевская Е. Д.

The article investigates the issue of heuristic productivity of using the method of computer-assisted topic modeling for philological analysis of fiction text. The study analyzes the results of applying the algorithm of Latent Placement Dirichlet (LDA) for searching intertextual connections of motifs in two sub-corpora of fiction texts: 62 texts of different genres (stories, essays, novels, critical articles) belonging to S. Dovlatov, on the one hand, and 35 fiction works, which the writer listed in one of the letters to T. Urzhumova as the works that had deeply influenced him and should be read by everybody. The algorithm has revealed 20 themes (topics), into which all the texts were distributed. Each topic obtained was a chain of words with weights of significance for the realization of that topic. As a result of the comparison of the texts and the topics, three “text - topic” correspondences were discovered. The texts in each of the following three groups belong to one common topic: 1) B. Pilyniak’s novel “The Bare Year” and Dovlatov’s story “By the River”; 2) G. Wells’s novel “The Time Machine”, E. Hemingway’s story “The Old Man and the Sea” and Dovlatov’s story “Emigrants”; 3) A. Grin’s story “The Commandant of the Port” and Dovlatov’s essay “We Speak Different Languages”. Further philological analysis demonstrated the intersection of motifs in these groups of works of fiction. The pilot study under consideration has shown that methods of computer-assisted text analysis, including those based on machine learning, can become a philologist’s tool for experimental search, guiding the expert intuition along the path outlined by the algorithm via processing large corpus arrays.

Keywords: intertextuality, интертекстуальность, topic modeling, тематическое моделирование, S. Dovlatov, С. Довлатов

Publication based on the results of: Текст как Big Data: моделирование конвергентных процессов в языке и речи цифровыми методами (2023)