A large-scale computational analysis of more than half a million public posts on X, formerly Twitter, has identified distinct language-specific conversational topics and temporal dynamics surrounding pornography discourse across an eleven-year period. Published in the journal BMC Public Health, the peer-reviewed study evaluated 634,746 public posts, described as Xeets, published between 2014 and 2024 in English, Spanish, Catalan, and Basque. Shared early to provide accelerated access to accepted research carrying a permanent digital object identifier, the study showed that reported surges in pornography consumption during the COVID-19 pandemic did not correspond to clear increases in pornography-related posts on the platform.
The authors framed their research around an important limitation in contemporary public health literature. While pornography use is widespread, empirical evidence has remained scarce regarding how pornography is framed and discussed in everyday online environments.
To compile an extensive linguistic dataset, the investigators retrieved public Xeets in English, Spanish, Catalan, and Basque using the Tweet Binder search engine. The data collection process utilized general query terms specifically related to pornography and sexually explicit content to capture relevant online material. To be deemed eligible for the study, every retrieved post was required to satisfy four specific criteria established by the research protocol: it had to be publicly accessible on the platform, contain at least one of the predefined query terms, have been posted between 2014 and 2024, and be written in one of the four designated target languages.
To analyze and organize the resulting collection of 634,746 posts, the researchers implemented machine-learning-based topic-modeling techniques. Following the automated generation of these computational models, a multidisciplinary team qualitatively evaluated and labeled the most prevalent topics within each linguistic group.
The resulting BERTopic analysis uncovered distinct thematic focal points across the four languages studied. English-language Xeets were dominated by discussions centering on pornography-related scandals as well as pornography involving young individuals. In the Spanish-language corpus, posts referred mainly to pornography consumption patterns and various forms of crimes. Meanwhile, Catalan-language Xeets frequently addressed violent or illegal pornography. In contrast, Basque-language Xeets were largely composed of explicit content, with only a smaller share of posts dedicated to broader social, legal, and educational debates surrounding pornography.
In addition to identifying distinct thematic categories, the study revealed that temporal patterns in posting activity varied across the four languages between 2014 and 2024. A notable finding concerned the period surrounding the global COVID-19 pandemic.
In their discussion and conclusions, the authors carefully delineated the boundaries of their findings, highlighting that this large-scale multilingual analysis identifies topics and temporal patterns without directly measuring underlying attitudes, personal perceptions, or user sentiments. Because the study provides descriptive observations rather than causal mechanisms or sentiment evaluations, the researchers noted that these findings should primarily serve to generate hypotheses for future public health communication research. The authors concluded that additional audience research, detailed qualitative analyses, and sentiment-focused studies will be essential before these descriptive findings can reliably inform future awareness campaigns or culturally tailored public health messaging.
Subject of Research: Topic-modeling analysis of multilingual pornography-related discourse on X
Article Title: Pornography-related discourse on X (Formerly Twitter): a topic-modeling analysis of more than half a million posts across four languages (2014–2024)
Article References: López-Crespo, J. B., Alvarez-Mon, M. A., Abelenda, F. L., Chushig-Muzo, D., Pérez-Araluce, R., Pérez-GarcÃa, J. M., Chiclana-Actis, C., Villena-Moya, A., Testa, G., Domingo-Espiñeira, J., BÅ‘the, B., Potenza, M. N., & Mestre-Bach, G. (2026). Pornography-related discourse on X (Formerly Twitter): a topic-modeling analysis of more than half a million posts across four languages (2014–2024). BMC Public Health. https://doi.org/10.1186/s12889-026-30155-9
Image Credits: AI Generated
DOI: 10.1186/s12889-026-30155-9
Keywords: topic modeling, BERTopic, X platform, social-media listening, pornography discourse, natural language processing, multilingual analysis, public health communication, COVID-19 pandemic, Tweet Binder, English Xeets, Spanish Xeets, Catalan Xeets, Basque Xeets, self-report surveys, social-desirability bias, BMC Public Health
News Source: Ophelia Keating. (November 15, 2026). Multilingual Posts on X Reveal Language-Specific Themes in Pornography Discourse. Scienmag.



