Political communication on gender and sexuality on Twitter/X in Spain (2015–2023): An AI-assisted analysis of sentiment, toxicity and hate speech as drivers of engagement

Recent surges in anti-feminist and anti-LGBTIQ+ narratives in Spain, especially amplified on social media such as Twitter (now X), raise substantive questions about how political parties communicate about gender and sexuality and how platform dynamics reward such messaging about parties’ online communication strategies. This article presents a longitudinal, comparative analysis of 4,504 tweets posted by the official accounts of Spain’s main parties between 2015 and 2023. We apply computational techniques to analyse sentiment, toxicity, and hate speech (model-based probability scores ranging from 0 to 1) within the messages, and examine how these features relate to engagement (likes, reposts, replies, and quotes). The findings indicate that Vox is simultaneously the most toxic party and the one generating the highest level of interactions. Moreover, engagement is positively associated with toxicity and declines for more positive messages. Over time, both toxicity and hate speech show a sustained increase, while overall sentiment becomes progressively more negative, signalling growing discursive polarisation in party communication on gender and sexuality. These results suggest that negativity and toxic cues are strategically rewarded by the platform’s interaction metrics, potentially incentivising their continued use. The observed trends carry implications for public discourse, minority rights, and the quality of democratic debate, as communicative payoffs may systematically favour provocative and exclusionary framings. We discuss how these dynamics intersect with parties’ digital strategies and outline considerations for platform governance, media literacy, and future research on algorithmic visibility and affective polarisation.