Multidimensional Model of Sebastian Unger’s Idiostyle in Poetic Creativity: Corpus Analysis and NLP Methods

Abstract

The article presents a multidimensional model of Sebastian Unger’s idiostyle based on corpus analysis and natural language processing (NLP) methods. The study is based on a structured approach to the analysis of authorial style, comprising text certification, thematic modeling, and stylometric evaluation. A subcorpus of Unger’s texts (SPU) was created and subjected to automated processing using such methods as word vectorization (Word2Vec, TF-IDF), topic modeling (LDA, BERT), syntactic and morphological analysis, and emotional modeling (Sentiment Analysis). The results of the analysis show the presence of clear stylistic markers in Unger’s work, including metaphorical structures, fragmentary composition, dominance of expressive vocabulary, and specific syntactic models. It is found that the author’s poetry tends to the categories of “nature”, “myth”, “philosophy”, which is confirmed by thematic clustering and analysis of key concepts. The proposed methodology of corpus research allows automating the identification of the author’s style, providing a quantitative assessment of his linguistic features and opening up new perspectives for digital stylometry and authorial attribution.