Conflicting narratives and polarization on social media

2025-07-21

AI Summary

This study investigates how conflictual narratives—divergent interpretations of the same political reality across social groups—drive issue-level polarization and cross-issue discursive alignment in social media. Drawing on ideologically stratified Twitter corpora from Germany concerning the Ukraine war, the COVID-19 pandemic, and climate change, the research employs text signal extraction, narrative analysis, and discourse modeling to systematically identify narrative divergence between opposing groups along two dimensions: “role assignment” and “subjective positioning.” It proposes and empirically validates the “narrative alignment” pattern, demonstrating how political actors achieve cross-issue discursive integration through shared narrative structures. Results confirm the empirical prevalence of narrative alignment, establishing narrative mechanisms as a critical analytical lens for understanding discursive polarization in the public sphere. The study contributes a novel, operationalizable framework for polarization research grounded in narrative theory and computational discourse analysis.

Technology Category

• Natural Language Processing: Discourse, Pragmatics & Argument Mining
• Application Domains: Humanities & Computational Social Science
• Knowledge Representation and Reasoning: Argumentation

Application Category

• Social Networks and Social Media: Media and governance, opinion dynamics, filter bubbles, polarization
• Web Mining and Content Analysis: Topic discovery and tracking
• Security and Privacy: Data transparency and provenance

Abstract

Narratives are key interpretative devices by which humans make sense of political reality. In this work, we show how the analysis of conflicting narratives, i.e. conflicting interpretive lenses through which political reality is experienced and told, provides insight into the discursive mechanisms of polarization and issue alignment in the public sphere. Building upon previous work that has identified ideologically polarized issues in the German Twittersphere between 2021 and 2023, we analyze the discursive dimension of polarization by extracting textual signals of conflicting narratives from tweets of opposing opinion groups. Focusing on a selection of salient issues and events (the war in Ukraine, Covid, climate change), we show evidence for conflicting narratives along two dimensions: (i) different attributions of actantial roles to the same set of actants (e.g. diverging interpretations of the role of NATO in the war in Ukraine), and (ii) emplotment of different actants for the same event (e.g. Bill Gates in the right-leaning Covid narrative). Furthermore, we provide first evidence for patterns of narrative alignment, a discursive strategy that political actors employ to align opinions across issues. These findings demonstrate the use of narratives as an analytical lens into the discursive mechanisms of polarization.

Problem

• Analyzing conflicting narratives in social media polarization
• Identifying discursive mechanisms of issue alignment
• Extracting textual signals from opposing opinion groups

Innovation

• Extracting textual signals from opposing tweets
• Analyzing conflicting narratives in political issues
• Identifying narrative alignment patterns across topics