Full text loading...
, Katarzyna Budzynska1
, He Zhang1
, Marie-Amélie Paquin2 and Barbara Konat3
Abstract
Offensive language affects contemporary societies by hindering communication and increasing polarisation. In this study, we apply computational linguistics to investigate offensive reactions to public figures in the climate debate on Twitter across their roles and popularity. We also use sentiment analysis to inspect the accuracy of lexical criteria in detecting negative attitudes and examine the types of social media users based on the frequency of offensive content in their posts. With an in-depth, large-scale corpus analysis comprising one million words, we demonstrate that frequent offensiveness in responses to politicians relatively rarely expresses personal attacks, and the popularity of public figures does not always come together with the highest density of offensive reactions. We also show that most users publish predominantly non-abusive posts. The study sets foundations for strategies to be employed to reduce polarisation that constitutes a threat to deliberative democracy.
Article metrics loading...
Full text loading...
References
Data & Media loading...