Research question/goal:

The project aimed to elucidate the macro-social and media-related conditions of mediated contestation. To this end, media debates of issues related to the public role of religion were compared in six countries (USA, Australia, Germany, Switzerland, Turkey, and Lebanon). The project covered both professional journalistic discourse (in daily newspapers, online news sites, and political bogs) and user-generated online debates (in the comment sections of news providers and in Facebook groups of partisan actors and alternative media, and on Twitter).

Journalistic items were analysed with standardised manual content analysis (N = ca. 1,700 articles, out of a proprietary database of about 2 million news items). Citizen-generated debates were captured with (semi-)automated computational content analysis (N = approx. 1.3 million user posts). The qualities of mediated contestation studied comprised

• inclusiveness of actors and ideas voiced in a debate
• civility
• justification of opinions and argumentative complexity
• discursive integration through mutual referencing between actors

The analyses have shown that journalistic discourse in majoritarian democracies (USA, AUS) is more inclusive and heavier on justifications than in consensus systems (Germany, Switzerland), but also less oriented to the common good. Contrary to popular prejudice, journalistic discourse in majoritarian countries is also more civil on average. However, citizen-generated debates online and on social media are less civil and less argumentatively complex in majoritarian systems. Such seemingly contradictory results suggest that different types of democracy foster different discursive profiles. Basic democratic functions such as broad inclusion, justification, and civility in journalism are more pronounced in majoritarian systems, while more far-reaching deliberative qualities such as justifications oriented to the common good and civility and complexity in citizen-generated debate flourish better in consensus democracies.

In addition, the project was able to show that online user posts are more argumentatively complex but at the same time less civil when they appear in online arenas geared towards plural, issue-driven debate (such as the comment section of news providers) than in arenas that are more prone to preference-driven debate among the like-minded (such as partisan Facebook groups and Twitter). Meeting opposing positions online seems to provoke stronger justification efforts but also more verbal confrontation among the dissenters.

Apart from these (and other) substantive insights, the project also developed two novel methodological procedures: (a) a semi-automated approach to classifying texts from previously unknown country contexts, the so-called Expert-Informed Topic Modeling (EITM); and (b) a computational approach to classifying large social media datasets by combining theory-informed dictionaries and ‘glass-box’ machine learning.