The present study aimed to expand our understanding of trolling interactions by examining 10,025 community-reported trolling incidents in the online game League of Legends to determine what characterizes messages sent by trolls, their teammates, and their opponents. To do this, we used a novel method blending content analysis and topic modelling. Contrary to extant literature, our study of complete trolling interactions found striking similarities between teammates’ and trolls’ chats, with both displaying the negative traits (e.g., exclusionary language) typically attributed to trolls. Findings also suggest that the transition from victim to perpetrator can occur extremely rapidly. This has important implications for the labelling of actors in trolling interactions, for future studies into the trolling cycle, and for theories of computer-mediated communication.
|Number of pages||26|
|Journal||Journal of Computer Mediated Communication|
|Publication status||Published - Nov 2019|
- Content Analysis
- Machine Learning
- Online Games
- Trolling Interactions