The aesthetics of climate misinformation: computational multimodal framing analysis with BERTopic and CLIP

Abstract

Climate misinformation today is increasingly conveyed through multimodal content, where images and text interact to shape meaning, emotional resonance, and enhance credibility. Yet, research on climate misinformation has rarely explored how these modes work in tandem. This paper introduces a multimodal analytical framework to examine climate-related visual misinformation. Analyzing 17,848 image-text posts, we combine BERTopic, CLIP, and qualitative framing analysis to investigate how climate denialist narratives are constructed. Our findings reveal a paradoxical communication strategy: the movement appropriates scientific aesthetics–graphs, statistics, and technical imagery–to contest the scientific consensus, projecting rationality, authority, and masculine self-control. In contrast, climate advocates are depicted through emotionally charged, feminized, and irrational imagery. These contrasting multimodal framings supports a strategy of strategic depoliticization, presenting ideological claims as neutral, objective truths. We argue that understanding climate misinformation requires moving beyond factual accuracy to examine the multimodal forms through which it gains legitimacy and emotional power.