Abstract
The provision of online news provides unique opportunities for users to interact with content and with other users. One of the more common forms of interactivity involves commenting on news stories. These interactive features are often heralded for enabling virtual public discussion of current events. Yet there exists a widespread belief that these spaces fail to meet that lofty goal, instead exhibiting hostile and vitriolic discourse, which undermines the deliberative potential of online interaction. At the same time, there is a lack of consensus among researchers regarding the proper conceptualization of hostility, and its more desirable counterpart, civility. This article aims to integrate normative and contextual conceptual definitions of hostility and civility in online interactions with the news. Building on these, we develop operational definitions of the two concepts and analyze the presence of hostile and civil discussion in user comments posted to YouTube news videos. Further, we explore the possibility that the content and source of those news videos, as well as popularity and engagement metrics, may explain the degree of hostility and civility in comments.
Acknowledgements
We thank four coders—Josephine Lee, Jon Sonnheim, Jessica Bobula, and Daniel Ellman—for their careful work and dedication, Susan Anderson and Beth Bennett for their guidance and experience, and Mike Smith and the Media Management Center at Northwestern University for financial and logistic support.
Notes
1. Since the time of data collection, YouTube has replaced the “News” category with a “News and Politics” channel.
2. The original sample had 700 videos (7 days × 100 videos/day). However, after removing videos with zero comments, as well as those with all comments in a foreign language, the final sample was 515 videos.
3. The original subsample had 200 videos (5 days × 40 videos/day). However, after removing videos with zero comments, as well as those with all comments in a foreign language, the final sample was 163 videos.
4. Codebook is available upon request. Kn is a version of the Kappa inter-rater reliability coefficient.
5. It is likely that some of the disparity between civil and hostile language is explained by the greater number of words in the civility dictionary (619 words versus 351 in the hostility dictionary). Still, that ratio (1.76) is less than the ratio of civil to hostile language in the comments (3.05), suggesting that civility outweighs hostility even if we control for the number of words in the dictionaries.