Full article: Picturing Protest: Visual Framing in Authoritarian Media on Twitter
This study investigates how media in authoritarian regimes visually frame protests on Twitter/X, extending the protest paradigm theory – traditionally applied to democratic contexts – to non-democratic settings where media operate under varying degrees of state control. Leveraging over 331 million tweets from 9,331 news outlet accounts across 144 countries, we identify 124,301 protest images from 38 authoritarian countries. Through computer vision analysis, we classify these images into four protest paradigms: riot, confrontation, spectacle, and debate. We find that spectacle and debate are the most common frames in authoritarian media. We observe a recurrent cross-national pattern: domestic protests are more often depicted through the debate frame, while imagery of riot, confrontation, and spectacle appears disproportionately in coverage of foreign protests. Variations in press freedom further shape these portrayals – greater freedom correlates with more depictions of riots and confro