Sentence voice preferences and sentiment patterns in gendered Weibo headlines on violent events: A data-driven critical discourse analysis

Drawing on Fairclough’s three-dimensional Critical Discourse Analysis framework, this study investigates gendered language differences in Weibo hot-search topics related to violent incidents between 2021 and 2024. We combine quantitative textual and sentiment analysis with discursive and socio-cultural interpretation to reveal these differences, and to uncover the possible motivations underlying them. The quantitative results reveal a consistent preference for passive-voice constructions when men are the perpetrators and women the victims, possibly muting male agency and foregrounding female suffering. At the socio‑cultural level, the results illustrate how this syntactic bias can be sustained through the interplay of media conventions, Weibo’s algorithmic logic, and the reinforcement of traditional gender hierarchies. The multidimensional sentiment analysis finds no systematic emotion differences by perpetrator gender, which may partly reflect a tendency for some male-perpetrator headlines to adopt a more neutral and fact-focused tone despite the attention-driven nature of the hot-search system. This both aligns with the goals of upholding the media’s discursive stance, and reflects the prevailing social gender conventions. Our findings highlight the subtle but impactful role of syntax in demonstrating gendered power imbalances in social media discourse.