Research

Dissertation Project

Despite the initial optimism concerning the democratic potential of social media such as Facebook and Twitter, recent years have seen increasing concerns about the aggressive nature of political communication on those platforms. Although much scholarly attention has been paid to uncivil and hateful political discussion between political opponents, little is known about, perhaps, the most worrisome type of political communication: promotion of political violence. Social media posts containing promotion of political violence draw public attention not only because it is normatively undesirable in itself but also because such posts can spread via multiple chains of communication networks, thereby exposing many users to promotion of political violence. The spread of and exposure to such content is discomforting due to its potential for creating a combative climate of political discussion on social media. Given the gravity of the problem, I set out to explore the spread, causes, and consequences of promotion of political violence on social media and proceed to forecast its future occurrence.

In the first paper, I introduce an automated method to detect Tweets containing promotion of political violence on a large scale relying on machine learning and natural language processing approaches and investigate their spread on the communication network on Twitter. In the second paper, I conduct an experiment to investigate the effects of promotion of political violence on social media users’ affective polarization. I argue that, while threats from an out-group member leads to affective polarization by inducing fear and anger toward the out-group, threats from an in-group member contributes to affective depolarization by evoking collective partisan shame. In the third paper, I explore the relationship between offline political events and promotion of political violence on social media. I claim that offline political events concerning moralized political issues often evoke lethal partisanship, resulting in prevalence of violent rhetoric for extreme partisan expression. In the last paper, I set out build a model to forecast the occurrence of promotion of political violence on Twitter from a network perspective. Together, these four papers shed new light on aggressive political communication, mass political polarization, and detection and prevention of online political violence.