The Effects of Multimodal Hate Speech on Prejudice and Prosocial Behavioral Intentions

While online hate speech has become a serious problem in multimedia environments, most studies in this area have examined text-based hateful content, with less attention paid to its other visual aspects. From a multimodal perspective, we conducted an online experiment (N = 799) to investigate how multimodal hate speech (i.e., text and images presented together to convey hateful meanings) on social media affected users’ prejudicial attitudes and prosocial behavioral intentions. The results showed that participants in the text-plus-image (vs. text-only) condition felt more sympathy, which led to less implicit prejudice toward the target group and more prosocial behavioral intentions. Additionally, exposure to text-plus-image hate speech had an indirect effect on prosocial behavioral intentions through sympathy and implicit prejudice. The findings contribute to scholarship on multimodality and hate speech and provide insights for designing multimodal systems for content moderation.