Polarized Digital Connective Narratives on the COVID-19 Vaccination Campaign: Dynamics of Conflict and (Dis)Engagement in South Africa

Digital polarization is often examined through identity-based frameworks that presuppose stable groups. We instead conceptualize Digital Connective Narratives (DCN) as a process of interactional co-construction that yields an emergent connective narrative field around shared issues and oppositional framings. Within this framework, Polarized DCN (PDCN) arise when these connective fields consolidate along five structural dimensions, social relevance, conflict scope, historical depth, power asymmetries, and emotional intensity. To illustrate the approach, we analyze debates on X (formerly Twitter) during South Africa’s COVID-19 vaccination campaign using topic modeling, stance classification, and adversarial-language detection tracked across rollout phases. Findings show that polarization did not diffuse across all topics: it concentrated in conflict-relevant debates, while institutional/informational topics remained largely non-polarized and faded over time. Within polarized debates, adversarial language intensified and engagement persisted, with challenger (non-institutional) streams gaining relative prominence as the campaign unfolded. As the pandemic’s immediate salience waned, public justification lost urgency and conflict retreated to a narrower circle of highly engaged users.