Virtual intimacy and digital desire: linguistic and thematic patterns in adult-oriented VRChat communities

This study investigates the linguistic and thematic structure of erotic discourse in NSFW (Not Safe For Work) VRChat-related Reddit communities, based on an analysis of 29,934 user-generated posts collected from public subreddits. A mixed-methods approach was employed, integrating BERTopic for topic modeling and Leximancer for semantic mapping to uncover how users express digital desire, emotional intimacy, and identity in avatar-mediated environments. The findings reveal a consistent co-occurrence of emotionally expressive terms such as “cute,” “feel,” and “better” alongside sexually explicit language including “ride,” “fuck,” and “cum.” This interplay indicates that Erotic Role Play (ERP) in virtual reality is shaped not only by sexual gratification but also by affective engagement and social interaction. The study introduces the concept of Digital Desire Grammar, offering a new framework to interpret how linguistic patterns, identity cues, and fantasy enactments are organized in immersive digital communication. In addition to its conceptual contribution, the research demonstrates an ethically informed, replicable NLP methodology for analyzing sensitive and high-context online discourse. Practical implications include recommendations for platform moderation, consent-aware design, and future applications of NLP in studies of digital intimacy.