This study investigates linguistic differences between Russian state-controlled media and social media reporting on the Russo-Ukrainian war, with a focus on propaganda frames and strategies. Using the WarMM-2022 corpus, we apply Kullback-Leibler Divergence (KLD) to identify distinctive linguistic features across media types. The use of KLD ensures the interpretability of the analysis, offering clear insights into the varying propaganda techniques, such as demobilization efforts on state media and more mobilizational rhetoric on social media. This method aligns with the principles of computational reproducibility, providing a transparent framework for linguistic analysis in political discourse. Our findings confirm previous research on Russian propaganda, underscoring the role of language in shaping propaganda strategies and highlighting the value of KLD as a method promoting accessibility and reusability in computational social science.

PDF