The “Demobilization” of Dagestan Nationalities: A Discourse Analysis of Dagestanskaya Pravda Using Natural Language Processing

Abstract

This article examines the discursive mobilization of ethnic categories in the regional state press of a multi-ethnic republic, drawing on materials from the newspaper Dagestanskaya Pravda. The theoretical framework is Rogers Brubaker’s cognitive theory of ethnicity, according to which ethnicity constitutes not a property of groups but a variably activated cognitive schema realized through concrete communicative practices. The empirical basis consists of two corpora of the newspaper separated by more than seven decades: a Soviet corpus (1939–1945, 1,914 issues) and a contemporary corpus (2018–2024, 1,349 issues), totaling approximately 26 million tokens. The methodology combines distributional semantic analysis based on word vector representations (Word2Vec), collocational frequency analysis, the Word Embedding Association Test (WEAT), and qualitative analysis of extracted contexts. The study finds that in the Soviet subcorpus, ethnonyms of Dagestan nationalities and the generalizing category dagestanets (‘Dagestani’) functioned as interchangeable units of mobilization, activated across military, labor, and cultural narratives. In the post-Soviet subcorpus, a functional differentiation of categories is observed: ethnonyms shift in a statistically significant manner toward the domain of cultural-traditional representation (the “demobilization” of nationalities), while the mobilization function becomes concentrated in the regional category dagestanets, whose frequency increased approximately eightfold. These findings empirically corroborate Brubaker’s thesis of the variable activation of ethnicity and identify a discursive correlate of Dagestan’s institutional recentralization: the culturalization of ethnonyms proceeds alongside the construction of a regional civic identity as the dominant frame of loyalty.