https://orcid.org/0000-0002-8044-2145
![Sample our Humanities journals, sign in here to start your FREE access for 14 days]({"type":"image" , "src" : "/sda/5939/TOC-Subject-Sample-2021-Humanities.jpg"})
ABSTRACT
Political polarization on the digital sphere poses a real challenge to many democracies around the world. Although the issue has received some scholarly attention, there is a need to improve the conceptual precision in the increasingly blurry debate. The use of computational communication science approaches allows us to track political conversations in a fine-grained manner within their natural settings – the realm of interactive social media. The present study combines different algorithmic approaches to studying social media data in order to capture both the interactional structure and content of dynamic political talk online. We conducted an analysis of political polarization across social media platforms (analyzing Facebook, Twitter, and WhatsApp) over 16 months, with close to a quarter million online contributions regarding a political controversy in Israel. Our comprehensive measurement of interactive political talk enables us to address three key aspects of political polarization: (1) interactional polarization – homophilic versus heterophilic user interactions; (2) positional polarization – the positions expressed, and (3) affective polarization – the emotions and attitudes expressed. Our findings indicate that political polarization on social media cannot be conceptualized as a unified phenomenon, as there are significant cross-platform differences. While interactions on Twitter largely conform to established expectations (homophilic interaction patterns, aggravating positional polarization, pronounced inter-group hostility), on WhatsApp, de-polarization occurred over time. Surprisingly, Facebook was found to be the least homophilic platform in terms of interactions, positions, and emotions expressed. Our analysis points to key conceptual distinctions and raises important questions about the drivers and dynamics of political polarization online.
Supplemental data
Supplemental data for this article can be accessed on the publisher’s website at https://doi.org/10.1080/10584609.2020.1785067.
Disclosure Statement
No potential conflict of interest was reported by the authors.
Notes
1. Poll conducted by the Israeli Democracy Institute. Retrieved from: http://www.peaceindex.org/indexMonth.aspx?num=308.
2. Retrieved from: https://www.mako.co.il/news-israel/local-q1_2017/Article-1a4ebc032ca6951004.htm.
3. Due to Facebook’s privacy regulations, we were able to retrieve only the contents of public pages and posts set to public.
5. Due to self-selection, participants in these groups are highly interested in politics, and not representative of WhatsApp users as a whole. Still, our findings provide rare insights into the political communication happening on this understudied platform.
6. The Facebook and Twitter data is presented on the journal’s website (Due to privacy issues, the WhatsApp data cannot be published).
7. stm was chosen for its capacity to include covariates. Estimating one model across all platforms, this allowed us to permit some topics to be more or less prevalent on different platforms.
8. In a manual content analysis, we classified those concepts and lexical choices grouped by each topic based on their fit within a) those narratives promoted by Azaria’s supporters, which presented his actions as justified and heroic in the defense against violent terrorism, (e.g., hero, terrorist, security, justified); b) those narratives advanced by those opposing Azaria, which presented his actions as immoral and denounced him as a hateful murderer (e.g., extrajudicial, murder, human rights, hateful), c) ambivalent narratives that combined considerations of both sides or were compatible with different evaluations (e.g., kill, justice, intentions), or d) neutral descriptors that gave no indication of an evaluative stance. See Appendix for detailed documentation.
9. For this classification, we decided whether a majority of top-associated words directly expressed emotions or carried strong emotional sentiment, and if so, whether negative (e.g., hate, corrupted, traitors, hypocrisy), positive (e.g., hero, solidarity, love, everybody’s child), or mixed emotions were expressed.
10. We also checked the overtime changes in a within-subject repeated measures ANOVA for those participants who contributed to all four phases, confirming the significant increase in homophily on Twitter (F(3,398) = 3.43** inbound, F(3,193) = 6.76*** outbound) and the significant decrease in homophily on Facebook (F(3,535) = 56.89*** inbound; n.s. for outbound interactions). For WhatsApp, the repeated measures ANOVAs are nonsignificant, owing to the very small number of users present in all phases.
Additional information
Notes on contributors
Moran Yarchi
Moran Yarchi (Ph.D. Hebrew University of Jerusalem) is a Senior Lecturer at the Sammy Ofer School of Communications, the Head of the Public Diplomacy program, and a Senior Researcher at the Institute for Counter-Terrorism (ICT) at the Interdisciplinary Center (IDC) Herzliya, Israel. Her main area of research is political communication, especially the media’s coverage of conflicts and public diplomacy.
Christian Baden
Christian Baden (Ph.D., University of Amsterdam) is a Senior Lecturer in the Department of Communication and Journalism at the Hebrew University of Jerusalem. His research focuses on the collaborative construction of meaning in controversial public debates in political communication and journalism, and has advanced theory and methodology in the study of dynamic political discourse.
Neta Kligler-Vilenchik
Neta Kligler-Vilenchik (Ph.D., University of Southern California) is a Senior Lecturer at the Hebrew University of Jerusalem. Her research interests include political expression in the new media environment.