The next paper in this Social Media and Society 2018 session is by Michael Bossetta, Chris Zimmermann, and Anamaria Dutceac Segesten, whose interest is in patterns in post-Brexit Facebook discussions. In particular, what is the role of emotions in these discussions, and what are their implications? The project gathered data using the Vox Populi data collection, enhanced with other data.
The analysis employed SentiStrength and other sentiment detection algorithms to assess the sentiment, emotionality, arousal, core emotions, and fine-grained feelings in Facebook posts from three major Facebook pages related to the Brexit referendum; most of the discussion took place here in the week of referendum. Cross-commenting took place especially before the referendum outcomes were announced; in total, some 28% of posts were cross-posts, and 35% of these were emotional. This is somewhat less than during major commercial controversies. (The core tool here is called EmotionVis.)
Crossposting escalated in the week of the referendum, and happened mainly from the two Leave pages to Remain, not the other way around (this means there was no Leave echo chamber, then). Emotional comments made up about one third of all emotional comments, and joy and empowerment were more frequent than anger. Anger was more predominant amongst Leave supporters.
Future steps here should be to enlist human coders to validate sentiment tools, run topic models by emotion, correlate emotions and topics with engagement, compare emotions in cross-posts with those in supporter posts, explore the priming effect of the campaigns’ own posts, and examine the language of affective polarisation.