This article investigates different polarizing mechanisms—relational homophily and ideological partisanship—characterizing political communications using Twitter data collected during the 2017 Norwegian election. By combining two computational approches—partition-specific network analysis and quantitative analysis of language polarization—we can examine the linkages between the structure of interactions and political polarization. The results show that the Norwegian political Twittersphere is not made of isolated echo chambers but is structured around crosscutting communities of interaction. There are no signs that communities with higher degrees of polarization are the ones that display higher degrees of homophily. Yet, the degree of ideological polarization differs across communities and topics. Some topics, such as political hate and far right and economy and taxes, are more polarized than others.
Supplementary notes can be added here, including code, math, and images.