[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"doc-seo-513633-105":3,"detail-sidebar-cat-0-en-105":81,"doc-detail-513633-en":130},{"code":4,"msg":5,"data":6},0,"ok",{"site_id":7,"language":8,"slug":9,"title":10,"keywords":11,"description":12,"schema_data":13,"social_meta":74,"head_meta":76,"extra_data":78,"updated_unix":80},105,"en","conditional-language-models-for-community-level-linguistic-variation","Conditional Language Models for Community-Level Linguistic Variation","","Conditional neural language models are used to learn vector representations for 510 online communities, capturing community-specific linguistic usage. These embeddings quantify linguistic variation across communities and are evaluated against a social-network-based community representation. Results show that the sociolinguistic embeddings align strongly with the social connectivity signal while remaining distinct from a purely connection-driven baseline. The approach targets embedding quality and interpretability for subreddit structure.",{"@graph":14,"@context":73},[15,34,56],{"@type":16,"itemListElement":17},"BreadcrumbList",[18,23,27,31],{"item":19,"name":20,"@type":21,"position":22},"https://docshare.wps.com","Home","ListItem",1,{"item":24,"name":25,"@type":21,"position":26},"https://docshare.wps.com/document/","Document",2,{"item":28,"name":29,"@type":21,"position":30},"https://docshare.wps.com/document/research-report/","Research & Report",3,{"item":32,"name":10,"@type":21,"position":33},"https://docshare.wps.com/document/conditional-language-models-for-community-level-linguistic-variation/513633/",4,{"url":32,"name":10,"@type":35,"image":36,"author":41,"headline":10,"publisher":44,"fileFormat":47,"inLanguage":8,"description":12,"dateModified":48,"datePublished":49,"encodingFormat":47,"isAccessibleForFree":50,"interactionStatistic":51},"DigitalDocument",{"url":37,"@type":38,"width":39,"height":40},"https://docshare.wps.com/thumbnails/conditional-language-models-for-community-level-linguistic-variation/513633.png","ImageObject",300,407,{"name":42,"@type":43},"CatatanPagi","Person",{"url":19,"name":45,"@type":46},"DocShare","Organization","application/pdf","2026-11-22","2026-10-06",true,{"@type":52,"interactionType":53,"userInteractionCount":55},"InteractionCounter",{"@type":54},"ViewAction",10,{"@type":57,"mainEntity":58},"FAQPage",[59,65,69],{"name":60,"@type":61,"acceptedAnswer":62},"What are the conditional language models used for in this paper?","Question",{"text":63,"@type":64},"They are used to learn community-specific vector representations (embeddings) for 510 online communities.","Answer",{"name":66,"@type":61,"acceptedAnswer":67},"How is linguistic variation between communities measured?",{"text":68,"@type":64},"The paper measures variation by comparing the learned community embeddings across communities.",{"name":70,"@type":61,"acceptedAnswer":71},"What relationship do the learned embeddings show with social connections?",{"text":72,"@type":64},"The sociolinguistic embeddings are highly correlated with a social network-based representation, though they are still different from it.","https://schema.org",{"og:url":32,"og:type":75,"og:title":10,"og:site_name":45,"og:description":12},"article",{"robots":77,"canonical":32},"index,follow",{"doc_id":79,"site_id":7},513633,1791424698,{"code":4,"msg":82,"data":83},"success",[84,88,92,96,101,106,111,115,120,123,126],{"id":22,"doc_module":4,"doc_module_name":25,"category_name":85,"show_sort_weight":86,"slug":87},"Story & Novel",90,"story-novel",{"id":26,"doc_module":4,"doc_module_name":25,"category_name":89,"show_sort_weight":90,"slug":91},"Literature",80,"literature",{"id":33,"doc_module":4,"doc_module_name":25,"category_name":93,"show_sort_weight":94,"slug":95},"Exam",70,"exam",{"id":97,"doc_module":4,"doc_module_name":25,"category_name":98,"show_sort_weight":99,"slug":100},5,"Comic",60,"comic",{"id":102,"doc_module":4,"doc_module_name":25,"category_name":103,"show_sort_weight":104,"slug":105},6,"Technology",50,"technology",{"id":107,"doc_module":4,"doc_module_name":25,"category_name":108,"show_sort_weight":109,"slug":110},7,"Healthcare",40,"healthcare",{"id":112,"doc_module":4,"doc_module_name":25,"category_name":29,"show_sort_weight":113,"slug":114},8,30,"research-report",{"id":116,"doc_module":4,"doc_module_name":25,"category_name":117,"show_sort_weight":118,"slug":119},9,"Religion & Spirituality",20,"religion-spirituality",{"id":118,"doc_module":4,"doc_module_name":25,"category_name":121,"show_sort_weight":118,"slug":122},"World Cup","world-cup",{"id":55,"doc_module":4,"doc_module_name":25,"category_name":124,"show_sort_weight":55,"slug":125},"Lifestyle","lifestyle",{"id":127,"doc_module":4,"doc_module_name":25,"category_name":128,"show_sort_weight":97,"slug":129},19,"General","general",{"code":4,"msg":82,"data":131},{"doc_id":79,"user_id":132,"nickname":42,"user_avatar":133,"doc_module":4,"category_id":112,"category_name":29,"doc_title":10,"doc_description":12,"doc_content":134,"file_id":135,"file_url":136,"file_type":137,"file_size":138,"view_count":55,"is_deleted":4,"is_public":22,"is_downloadable":22,"audit_status":22,"page_count":118,"language":139,"language_code":8,"site_id":7,"html_lang":8,"table_of_contents":140,"faqs":141,"seo_title":142,"seo_description":12,"update_tm":143,"read_time":104,"preview_page_count":33,"preview_urls":144},962090894170,"https://ap-avatar.wpscdn.com/davatar_6f874abed73319feea01a86fa6f0fab8","Conditional Language Models for Community-Level Linguistic Variation  \nBill Noble and Jean-Philippe Bernardy Centre for Linguistic Theory and Studies in Probability (CLASP) Dept. of Philosophy, Linguistics and Theory of Science University of Gothenburg  \n{bill.noble@, [jean.philippe.bernardy@}.gu.se](jean.philippe.bernardy@}.gu.se)  \nAbstract  \nCommunity-level linguistic variation is a core concept in sociolinguistics. In this paper, we use conditioned neural language models to learn vector representations for 510 online communities. We use these representations to measure linguistic variation between communities and investigate the degree to which linguistic variation corresponds with social connections between communities. We ﬁnd that our sociolinguistic embeddings are highly correlated with a social network-based representation that does not use any linguistic input.  \n1 Introduction  \nLinguistic communication requires that speakers share certain linguistic conventions, such as syntactic structure, word meanings, and patterns of interaction. Speakers assume that these conventions are common ground among their interlocutors, based on joint membership in a community (Stalnaker, 2002 ; Clark, 1996) . Such speech communities (Gumperz, 1972) range in size from the very small, like members of a friend group, to the very large, like speakers of English. However, as Eckert and McConnell-Ginet (1992) point out, it is communities of practice—deﬁned by mutual social engagement in a common activity—that are the primary locus of linguistic variation.  \nVariation is an important object of study in sociolinguistics, and is naturally amenable to computational analysis (Nguyen et al., 2016) . Most previous computational work on linguistic variation has considered variation at the level of macrosocial categories, such as gender (Burger et al., 2011 ; Ciot et al., 2013 ; Bamman et al., 2014b), age (Nguyen et al., 2013), and geographic location (Eisenstein et al., 2010 ; Bamman et al., 2014a) .In the present work, however, we investigate linguistic variation across online communities in the social media website Reddit.  \nFor this purpose, we introduce (section 2) various Community-Conditioned Language Models (CCLMs for short) . These models are conditioned on a vector representation (or embedding), which varies by community. Hence, they learn community embeddings. We report which architectures make best use of the community information (section 3), however our our primary purpose is not to improve language models in terms of perplexity, but rather to extract community embeddings that capture linguistic similarities between communities and test how the resulting embeddings correspond to the social structure of subreddits. To that end, we test the how well the community embeddings correlate with a social network-based representation of communities (section 4) .  \nThe contributions of this work are twofold. First, we develop a language model-based community embedding that we show is correlated with (but still different from) an embedding based on community membership alone. Second, the method we describe for testing the correlation between two embeddings from different models is, to our knowledge, novel to computational linguistics.  \n2 Community-conditioned language models (CCLMs)  \nWe experiment with two kinds of model architecture: simple unidirectional LSTM (Hochreiter and Schmidhuber, 1997) and a masked Transformer (Vaswani et al., 2017) . Although Transformerbased language models are considered state-of-theart, they achieve dominance partly thanks to the availability of very large data sets (e.g., Devlinet al., 2019 ; Brown et al., 2020), which are not available to us.1 Thus the LSTM is a worthy  \n1Fine-tuning existing models is not compatible with our methodology, because we fundamentally change the structure of the network by concatenating community embeddings with hidden states at various levels.  \n59  \nProceedings of the Fifth Workshop ","cbCaiiq5p0HWXKG3","https://ap.wps.com/l/cbCaiiq5p0HWXKG3","pdf",441946,"English","# Abstract\n# Introduction\n# Community-conditioned language models (CCLMs)\n## Data sets","[{\"question\":\"What are the conditional language models used for in this paper?\",\"answer\":\"They are used to learn community-specific vector representations (embeddings) for 510 online communities.\"},{\"question\":\"How is linguistic variation between communities measured?\",\"answer\":\"The paper measures variation by comparing the learned community embeddings across communities.\"},{\"question\":\"What relationship do the learned embeddings show with social connections?\",\"answer\":\"The sociolinguistic embeddings are highly correlated with a social network-based representation, though they are still different from it.\"}]","Conditional Language Models for Community-Level Linguistic Variation | PDF",1791288277,[145,149,153,157],{"page_no":22,"url":146,"width":147,"height":148},"https://res-academy.cache.wpscdn.com/upload/kdocshare/2026-11-17/8eb431dfd8d8a9918c2d6b45a72ff894.webp",983,1390,{"page_no":26,"url":150,"width":151,"height":152},"https://res-academy.cache.wpscdn.com/upload/kdocshare/2026-11-17/4bb3d0b37709a19f9f11b6ee5a5df03a.webp",982,1388,{"page_no":30,"url":154,"width":155,"height":156},"https://res-academy.cache.wpscdn.com/upload/kdocshare/2026-11-17/3bf807bad2508b4a161cfbbfeb54fc85.webp",981,1387,{"page_no":33,"url":158,"width":159,"height":160},"https://res-academy.cache.wpscdn.com/upload/kdocshare/2026-11-17/655a3d039e3f3512080e85f6e7daee18.webp",980,1385]