Cross-cutting interaction, inter-party hostility, and partisan identity: Analysis of offensive speech in social media

This study utilizes social media data and deep-learning-based text classification methods to investigate cross-cutting interactions on social media. Our findings reveal that people are more likely to use offensive speech in response to content published by opposing partisans. Furthermore, we demonstrate how inter-party hostility is associated with the partisan identity of both the message sender and the target in the interaction. On one hand, the findings indicate that strong partisans and people who publicly assert their partisan identities tend to attack opposing partisans, suggesting a relationship between the salience of partisan identity and value defense mechanisms. On the other hand, strong partisans, especially politicians, are more likely to be the target of offensive speech from opposing partisans. The disparity in the extent of received offensive speech is argued to result from individuals’ tendency to maintain their partisan identification by expressing hostility toward representative individuals of opposing partisans.