Structural unity and ethnocultural distinctness of Russian and Altai fairy tales: An analysis based on the application of generative models

This article employs an interdisciplinary approach to examine both the commonalities and distinctive features of Altai and Russian fairy tales. It combines a method of philological structural analysis, aimed at revealing the invariant deep structures of narrative texts, with the application of generative models, which for the correlation between the invariance of deep structures and the variability of their surface realizations within the two fairy-tale traditions. The implemented methodological synthesis aligns with a pertinent contemporary direction in humanities research: defining the applications and limitations of artificial intelligence in addressing socially significant scholarly problems. One such problem is assessing the degree of cultural commonality and difference among communities interacting within the multiethnic space of the Russian Federation.Given the complexity of this issue,we argue that folklore offers a particularly productive Lens for its examination. As the product of centuries of collective creativity, folklore both accumulates unique cultural characteristics and transmits them across generations. Simultaneously, as demonstrated by scholars such as Vladimir Propp, Claude Levi-Strauss, and their successors, folklore exhibits profound structural similarities at a deep level, pointing to universal patterns of human imagination and experience. Consequently, this study has a dual aim. First, by employing Propp's methodology for analyzing the structures of magic tales, it compares the deep and surface structures of Altai and Russian fairy tales. Second, it tests the utility of text-generation technologies as an analytical tool to complement traditional philological methods. A large language model (YandexGPT 5 Lite) was used to generate Russian and Altai fairy tales based on structural elements identified by Propp. Through prompt manipulations - beginning with basic instructions like “Generate a Russian fairy tale”/ “Generate an Altai fairy tale” and progressively incorporating Propp's formal structures and ethnoculturally specific surface elements - the study achieved two outcomes. It generated texts that closely approximate canonical fairy tales, and it facilitated the identification of both invariant narrative elements and multi-level markers of cultural specificity. These markers include character attributes, the linguistic realization of fairy-tale formulas, and other culturally embedded features.

Keywords

Authors

References