Visual-Verbal Framing of Misinformation Messages: International, Regional and Cross-Persuasive Communication Perspectives

We examine the visual-verbal framing of Covid-19 misinformation using a unique data set of 1500+ misinformation posts from more than 40 countries, as well as a dataset of 1500+ Covid-themed Facebook ads, collected from the US, Brazil and India. Using human coders and computer vision, we analyze the global and regional patterns of nonverbal emotional displays, blame attribution and emotional language in misinformation messages, and compare them with those used in ads. We find mostly similarities in the use of framing devices in misinformation across the world, but also some regional specificities. At the international level, several visual misinformation framing patterns emerge, including the association of images of politicians with visual displays of facial emotion. Moreover, misinformation posts and ads are revealed to use different visual-verbal framing strategies, with misinformation more likely to use visual-verbal incongruence and negative congruence, and Facebook ads more likely to use positive congruence.