Dehumanising Language as a Tool of Ingroup Cohesion - Case of Wartime Dehumanisation in Russian and Ukrainian Telegram

Intergroup dehumanisation is a pressing concern for contemporary society, known to hinder empathy, undermine prosocial behaviour, and contribute to between-group aggression. In wartime contexts, dehumanisation can escalate conflict by fostering support for war and enabling violence and extreme outgroup-targeted policies, and genocide. Social media narratives, in particular, both reflect and shape intergroup dynamics, making them a powerful data source for examining dehumanising language. This study investigates dehumanising language in online discussions during the Russian war in Ukraine, with a particular focus on its potential function in reinforcing ingroup cohesion. Utilising Large Language Models we construct a labelled dataset of dehumanising and non-dehumanising posts from leading Telegram channels in Ukraine and Russia. Our analysis shows that dehumanising rhetoric is prevalent in both countries and rose sharply following the full-scale invasion on 24 February 2022. While prior computational approaches primarily focus of measurement techniques and conceptualise dehumanisation as a manifestation of negative outgroup perception, our work explores underlying motivations behind dehumanisation of the enemy, and empirically examines whether such language serves to strengthen ingroup cohesion and solidarity in hostile settings.