The Language of Conspiracy Theories: Negative Emotions and Themes Facilitate Diffusion Online

Abstract

Global polling has seen an increase in endorsement of conspiratorial beliefs in recent years, with social media platforms receiving criticism for their potential role in facilitating this spread. The current study analyzed the presence of emotions and themes in conspiratorial content (revolving around conspiracy theories or conspiratorial ideas) on Twitter (now “X”) and Reddit when compared to scientific and general discussion, and the ability of emotions and themes to predict engagement (likes, shares etc.) in conspiratorial spheres. Approximately four million words worth of content were collected from Twitter and Reddit and analyzed using the Linguistic Inquiry and Word Count (LIWC) tool to determine concentration of different emotions and themes. Conspiratorial discussions were found to show significantly higher rates of anger, anxiety, power, and death compared to both scientific and general discussions across both platforms. Engagement was successfully predicted within conspiracy discussions; higher levels of anger, anxiety, power, and death predicted higher levels of engagement in conspiracy discussions. Contrary to previous findings, positive emotions were predictive of lower levels of engagement in conspiracy-related discussions. The findings suggest that conspiratorial content relies on powerful negative emotional responses to gain traction and increase audience engagement. The findings are discussed in the context of cognitive processes involved in the adoption of conspiracy theories, and we offer suggestions for researchers developing educational interventions and social media platforms to reduce the spread of conspiracy theories online.