Russian-Speaking Minorities in the Estonian Media: Topic Modeling of Newspaper Content 1996-2019

Building on the recognition that ethnic, national and racial groups are not essential, durable, and internally homogeneous entities but contingent products of particular group-making processes and practices, this paper analyzes Russian-speaking minorities in Estonia as objects of various projects and discourses aimed at constructing groupness and belonging. As a demographic legacy of the Soviet period, Russian-speakers of Estonia defy easy categorization, as they are neither an indigenous minority nor immigrants in the conventional sense. Subject to competing narratives, political claims and norms advanced by the host state, the external ethnocultural homeland, and European and international institutions, the position of Russian-speakers in the Estonian society remains a contested and evolving issue. The paper traces discursive representations of Russian-speakers in the Estonian-language media over a period of thirty years, from 1988 to 2018. Our interest lies with the variety of discursive labels applied to the Russian-speakers, including ‘occupants’, ‘settlers’, ‘immigrants’, ‘aliens,’ ‘national minorities’, ‘Russia’s compatriots, ’ and ‘Russian-speaking Estonians.’ The paper seeks to ascertain the relative prevalence of these constructions across the media space, the discursive contexts in which they are used, as well as changes in the discursive representations of the country’s Russian-speakers over time. The paper will link the empirical findings to the political agendas of actors producing particular discourses, and will discuss the implications of these discursive representations for the rights and status of the collectivities concerned. The paper relies on an extensive corpus of texts consisting of articles that have appeared in Estonian-language newspapers from 1988 to 2018. The digitized articles are available in the DIGAR Estonian Articles collection (digar.ee). We apply an analysis of word embeddings (using models like word2vec) and structural topic modelling to analyse the various and evolving contexts of how Russian-speakers in Estonia have been depicted in media discourse. An analysis of word embeddings allows us to understand which words are most similar to the various terms that are used to refer to Russian-speakers and thus constitute their context of use, while structural topic modelling enables us to extract topics that are related to Russian-speakers in Estonia and to model how these topics change and evolve over time. The results contribute to our understanding of nation-building processes in segmented societies.