A computational analysis of emotionally manipulative content in media coverage of the Russia-Ukraine war

Abstract

This paper comprehensively examines emotional patterns and manipulative tactics in English-language digital news coverage of the Russia-Ukraine war. The research examines the use of emotions across English-language media outlets, explaining their rhetorical functions and their potential for ideological influence. Using a purpose-built corpus of 488 full-length news articles published between February 2022 and early 2025, we utilise the Emotion English DistilRoBERTa-base model, fine-tuned for effective classification. This model assigns Ekman’s (1992) six basic emotions (anger, disgust, fear, enjoyment, sadness, surprise), plus a neutral class, and enables analysis of their distribution across 14 thematic categories and four media domains: the US, the UK, the EU, and global. We investigate the relationship between dominant emotions and 18 manually coded emotionally manipulative tactics. The main findings of the research indicate that negative emotions, most notably fear and anger, predominate in the corpus, functioning as discursive tools for mobilisation, blame, and perception shaping. Sadness and disgust are primarily associated with humanitarian reporting, while enjoyment and surprise remain marginal. Although neutral tone is less emotionally charged, it plays a rhetorical role in diplomatic and strategic reporting, framing neutrality as a deliberate perspective rather than emotional engagement. The research reveals that emotionally manipulative tactics, such as fear-based mobilisation, emphasis on the scale of tragedy, and victim-aggressor contrast, are widely employed across all media outlets, yet differ in frequency and function depending on media origin. The findings obtained emphasise the pivotal role of emotional framing in shaping audience engagement and moral alignment. This paper deepens understanding of digital war reporting, offering insights into how automated emotion detection, alongside discourse analysis, can expose the latent ideological functions of emotion in English-language news coverage. The study contributes to media discourse analysis and highlights the methodological value of computational methods in detecting emotional manipulation in news coverage.