Multilingual social media analysis reveals global patterns and language imbalances in volcanic eruption coverage

Abstract Understanding how people access and engage with volcanic hazard information in the attention economy is critical for crisis response and building long-term preparedness. Volcanic eruptions can have widespread effects, potentially creating language barriers for risk communication and aid provision. Though social media offers a proxy for public attention to geohazards, previous studies have been limited to single events, short timescales, and/or monolingual analyses. Accordingly, we lack detail on how social media reflects real-world volcanic activity, how notable eruptions shape online discourse, and how language and geography influence information dissemination. This study addresses these gaps through a longitudinal, multilingual analysis of specific keywords on "Image missing" (Twitter), covering 18 languages over ~4 years. Through timeseries analysis, discrete volcanic events emerge from social media data. The 2022 Hunga Tonga–Hunga Haʻapai eruption stands out as the dominant event within the dataset across almost all languages, highlighting how notable eruptions can initiate a step-change in online discourse. However, results reveal substantial linguistic imbalance: English dominates—even for eruptions in non-English-speaking regions—disconnecting the online visibility of eruptions from their real-world significance. To foster inclusive knowledge-sharing practice, volcanologists are strongly encouraged to actively engage with diverse linguistic groups beyond the silos of their own languages.