Language Mutations Sustain the Persistences of Conspiracy Theories on Social Media

Problem
This preprint addresses the gap in understanding how linguistic variations, or “mutations,” influence the persistence of conspiracy theories on social media platforms. While prior research has explored the spread of misinformation, there is limited empirical evidence on the role of language evolution in sustaining these narratives over time. The authors aim to elucidate the mechanisms by which semantic mutations contribute to the longevity of conspiracy claims, thereby informing content moderation strategies.

Method
The authors utilize a three-year dataset of conspiracy-related posts from the social media platform X. They employ computational linguistic analysis to quantify language mutations, focusing on psycholinguistic properties such as pronouns, social reference words, cognitive process terms, and risk- and health-related vocabularies. The study applies survival modeling techniques to assess the relationship between these mutations and the lifespan of conspiracy claims. Two primary mutation patterns are identified: simplification (reducing complexity) and assimilation (adapting to existing narratives). The analysis also categorizes mutations based on actor, action, and target (AAT) frameworks, providing a structured approach to understanding linguistic changes.

Results
The findings reveal that conspiracy claims exhibiting greater semantic mutations have significantly longer lifespans. Specifically, claims with high levels of mutation in psycholinguistic properties and AAT categories are associated with extended persistence. The authors report effect sizes indicating that claims with substantial simplification and assimilation mutations can last up to 50% longer than their less mutated counterparts. The study benchmarks these results against traditional models of misinformation diffusion, demonstrating a clear advantage for claims that undergo linguistic evolution.

Limitations
The authors acknowledge several limitations, including the potential biases inherent in the dataset, which is confined to posts from a single platform (X) and may not generalize to other social media environments. Additionally, the study does not account for the influence of external factors such as media coverage or political events that may also affect the lifespan of conspiracy theories. The reliance on computational linguistic analysis may overlook nuanced contextual factors that contribute to the mutation process.

Why it matters
This research has significant implications for the design of content moderation strategies on social media. By highlighting the mutability of conspiracy claims, the authors suggest that moderation efforts should focus on core claims rather than attempting to suppress all variations. This approach could enhance the effectiveness of interventions aimed at curbing the spread of misinformation. Furthermore, the insights gained from this study could inform future research on the dynamics of language in digital communication, particularly in the context of misinformation and public discourse.

Authors: Calvin Yixiang Cheng, Dorian Quelle, Scott A. Hale
Source: arXiv:2605.20050
URL: https://arxiv.org/abs/2605.20050v1