This paper uses corpus linguistic methods and general purpose computing tools to explore short-scale lexical change in the identity terminology used in an online community for transgender men and other transmasculine people. It focuses on the rapidly changing landscape of labels for trans people, cis people, and non-binary people in a trans community on LiveJournal.com, which was a popular social media venue among trans people in the 2000s. We consider a number of questions about lexical change, including when currently popular forms (e.g. cisgender, non-binary, transmasculine, etc.) were introduced; the decline of labels that have been problematized (e.g. transgendered, transsexual); and the sociocultural discourses that contextualize and account for these changes. We also describe novel methods for social media data collection, which rely on simple custom software, which we call livecorpus. Livecorpus was built for use with widely-available cloud computing tools, meaning that it is serverless (i.e. does not require the provisioning of the analyst’s own servers) and offers flexible configuration that can be modified as data collection progresses. These methods can be applied to other social media sources that are not pre-formatted in ways that facilitate automated analysis, which in practice means we can reach further back into the history of social media-based language use. While scholars of language variation and change have tended to focus on phonological and morphosyntactic variables in unselfconscious vernacular speech rather than the lexicon, we argue that speakers’ awareness of – and metalinguistic discourses about – lexical choices makes this level of language an ideal site for considering linguistic manifestations of sociopolitical change. Far from an unfortunate exception to the normal, non-conscious process of structural linguistic transformation, these types of intentional interventions into lexical usage must be recognized as a critical component of language change.