Variations in Online Self-Censorship: The Analysis of the Turkish Case

• In Event: Media Influence in Authoritarian States

Abstract

As the world is going through the third wave of autocratization (Boese et al. 2022), online self-censorship has become a pressing issue. So far, our knowledge of self-censorship comes from its practice in conventional media (e.g. Bodrunova et al. 2020, Schimpfössl et al. 2020), such as TV channels and newspapers, and the existing censorship literature is heavily informed by closed authoritarian regimes, such as China, Russia, and Iran (e.g. Shen & Truex 2021, Chang & Manion, 2021, Roberts 2018) with technologically advanced methods of digital surveillance. However, not many studies have so far examined the practice of online self-censorship in countries with competitive authoritarian regimes where the access to global social media platforms has not been blocked entirely or where the online surveillance of users by the regime is not routinely implemented.
The first question of this study is whether and how political orientation of individuals (radical vs. moderate) affects the level of online self-censorship on social media. Based our existing knowledge about variations in the tone of language by different ideological groups (Frimer et al. 2019), this paper hypothesizes that users following radical parties on social media censor themselves more than those following moderate parties. The second question is whether ideological proximity to the center of political power factors into this relationship. This part of the argument stems from the point that self-censorship is a function of both one’s own political views and the ideology of those who can persecute them. The study hypothesizes that the level of self-censorship increases as the ideological proximity to those in power increases.
The study takes advantage of the new censorship law introduced on October 18, 2022 in Turkey as a threshold to measure the variation in the levels of self-censorship among users. The choice of Turkey is related to its being a typical case of a competitive authoritarian regime in which there is still the de jure freedom of media. The new Turkish law presents an empirical opportunity for measurement since it specifically criminalizes Twitter’s retweet function with its “distribution of misinformative content” clause (Akkas 2022). Using Twitter data of 60,000 users who follow a total of 12 different political parties across the spectrum, the study focuses on the number of retweets sent by users before and after the new law.
The findings show support for the hypothesis that radicals censor themselves more than moderates. There is also evidence that ideological proximity to those in power decreases the level of online self-censorship: under Turkey’s far-right authoritarian government, followers of radical-left parties tend to display self-censorship significantly more than followers of their right-wing counterparts who are ideologically closer to the parties in power. In providing the first empirical assessment of self-censorship along ideological lines in a competitive autocracy, this study sheds light on the high level of variation observable among users on Twitter. The study, therefore, challenges previous portrayals of social media as a venue of radical opposition activity and uncovers important political bias by listening to the silence of users under the threat of persecution.

Author

• Sinan Kircova, University of Southern California