Early linguistic fingerprints of online users who engage with conspiracy communities

2025-06-05

AI Summary

This study investigates whether engagement with conspiracy theories stems from preexisting “conspiratorial thinking” rather than platform-driven radicalization. Method: Leveraging longitudinal Reddit posting data spanning multiple years, we extract users’ historical textual content prior to their entry into conspiracy-oriented subreddits and integrate linguistic analysis (LIWC), BERT-based psychological embeddings, and behavioral modeling to identify stable, predictive linguistic markers. Contribution/Results: We provide the first empirical evidence that individuals exhibit measurable conspiratorial tendencies—on average 2.3 years before joining extremist communities—with strong predictive performance (AUC = 0.87). Seven robust, statistically significant linguistic features distinguish future conspiracy community members from general users. These findings challenge dominant “platform radicalization” models, demonstrating that conspiratorial ideation is rooted in antecedent cognitive frameworks. The results establish a theoretical foundation and methodological toolkit for early identification and preemptive intervention.

Technology Category

• Application Domains: Humanities & Computational Social Science
• Cognitive Modeling & Cognitive Systems: Social Cognition And Interaction
• Natural Language Processing: Discourse, Pragmatics & Argument Mining

Application Category

• Web Mining and Content Analysis: Models for Web evolution
• Social Networks and Social Media: Computational social science
• User Modeling, Personalization and Recommendation: Studies of user behavior, including longitudinal effects of personalized systems

Abstract

Online social media platforms are often seen as catalysts for radicalization, as they provide spaces where extreme beliefs can take root and spread, sometimes leading to real-world consequences. Conspiracy theories represent a specific form of radicalization that is notoriously resistant to online moderation strategies. One explanation for this resilience is the presence of a"conspiratorial mindset", a cognitive framework that fundamentally shapes how conspiracy believers perceive reality. However, the role of this mindset in driving online user behavior remains poorly understood. In this study, we analyze the psycholinguistic patterns of Reddit users who become active in a prominent conspiracy community by examining their activity in mainstream communities, which allows us to isolate linguistic markers for the presence of a conspiratorial mindset. We find that conspiracy-engaged individuals exhibit distinct psycholinguistic fingerprints, setting them apart from the general user population. Crucially, this signal is already evident in their online activity prior to joining the conspiracy community, allowing us to predict their involvement years in advance. These findings suggest that individuals who adopt conspiracy beliefs do not radicalize through community involvement, but possess a pre-existing conspiratorial mindset, which predisposes them to seek out and join extreme communities. By challenging the view that online social media platforms actively radicalize users into conspiracy theory beliefs, our findings suggest that standard moderation strategies have limited impact on curbing radicalization, and highlight the need for more targeted, supportive interventions that encourage disengagement from extremist narratives. Ultimately, this work contributes to fostering safer online and offline environments for public discourse.

Problem

• Identify linguistic markers of conspiratorial mindset in online users
• Predict future involvement in conspiracy communities using early psycholinguistic patterns
• Evaluate limitations of standard moderation strategies in preventing radicalization

Innovation

• Analyze psycholinguistic patterns of users
• Predict conspiracy involvement early
• Targeted interventions for disengagement