This study explores the interplay of imagery and text in online political communication by European party leaders. We examine how visual and textual components combine to convey emotions in Instagram posts published by one hundred twenty-four mainstream and populist leaders of one hundred fourteen parties in twenty-four EU countries, covering 3 months before and after the 2019 European elections. Images are analyzed through automated face and emotion recognition algorithms to determine whether they contain positive, negative, or neutral emotions. The text accompanying each post is analyzed through a large language model algorithm. Results indicate that populist leaders, particularly right-wing ones, are less likely to combine positive textual and visual elements. Compared to mainstream leaders, right-wing populists more frequently post contents where text and image are redundantly nonpositive. Finally, emotional mismatch between textual and visual components is common among both mainstream and populist leaders.

中文翻译：

---

调整情感：欧洲政党领导人在 Instagram 上对文本和图像的比较分析

本研究探讨了欧洲政党领导人在线政治交流中图像和文本的相互作用。我们研究了视觉和文本组件如何结合以传达 24 个欧盟国家 114 个政党的 124 位主流和民粹主义领导人发布的 Instagram 帖子中的情感，涵盖 2019 年欧洲大选前后的 3 个月。通过自动面部和情绪识别算法分析图像，以确定它们是否包含积极、消极或中性情绪。每篇文章随附的文本都通过大型语言模型算法进行分析。结果表明，民粹主义领导人，尤其是右翼领导人，不太可能将积极的文本和视觉元素结合起来。与主流领导人相比，右翼民粹主义者更频繁地发布文本和图像冗余的非正面内容。最后，文本和视觉组件之间的情感不匹配在主流和民粹主义领导人中都很常见。