# Characterizing English Variation across Social Media Communities with BERT Source: [Characterizing English Variation across Social Media Communities with BERT]() Collection index: [llms.txt]() Data updated: 2025-10-28T13:56:46.471Z Authors: Li Lucy, David Bamman Published: 2021-02-12T23:50:57.000Z ## Abstract Much previous work characterizing language variation across Internet social groups has focused on the types of words used by these groups. We extend this type of study by employing BERT to characterize variation in the senses of words as well, analyzing two months of English comments in 474 Reddit communities. The specificity of different sense clusters to a community, combined with the specificity of a community's unique word types, is used to identify cases where a social group's language deviates from the norm. We validate our metrics using user-created glossaries and draw on sociolinguistic theories to connect language variation with trends in community behavior. We find that communities with highly distinctive language are medium-sized, and their loyal and highly engaged users interact in dense networks. ## Recorded benchmark results A research result applies to its stated dataset, split, metric and evaluation configuration. Compare like-for-like settings, preserve the metric's unit and direction, and distinguish paper-reported results from independent reproduction. The best result recorded here is not a claim of exhaustive, current global coverage. - [Lexicon Induction on Manually validated lexicon induction 1803 words (test)](): Lexicon Induction; Manually validated lexicon induction 1803 words (test); AP: 67.9. ## Original sources and implementation [Original paper]() [PDF]()