The structure of online social networks modulates the rate of lexical change

Proceedings of the 2021 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, pages 2201–2218 June 6–11, 2021. ©2021 Association for Computational Linguistics 2201 The structure of online social networks modulates the rate of lexical change Jian Zhu Department of Linguistics University of Michigan lingjzhu@umich.edu David Jurgens School of Information University of Michigan jurgens@umich.edu Abstract New words are regularly introduced to com- munities, yet not all of these words persist in a community’s lexicon. Among the many factors contributing to lexical change, we fo- cus on the understudied effect of social net- works. We conduct a large-scale analysis of over 80k neologisms in 4420 online commu- nities across a decade. Using Poisson regres- sion and survival analysis, our study demon- strates that the community’s network struc- ture plays a significant role in lexical change. Apart from overall size, properties including dense connections, the lack of local clusters and more external contacts promote lexical in- novation and retention. Unlike offline commu- nities, these topic-based communities do not experience strong lexical levelling despite in- creased contact but accommodate more niche words. Our work provides support for the so- ciolinguistic hypothesis that lexical change is partially shaped by the structure of the under- lying network but also uncovers findings spe- cific to online communities. 1 Introduction Lexical change is a prevalent process, as new words are added, thrive, and decline in day-to-day us- age. While there is a certain randomness at play in word creation and adoption (Newberry et al., 2017), there are also psychological, social, linguistic and evolutionary factors that systematically affect lexi- cal change (Labov, 2007; Christiansen and Kirby, 2003; Lupyan and Dale, 2010). In sociolinguistics, one structural factor that has long been recognized as influencing lexical changes is the language community’s social net- work. For example, drawing on pioneering works on social networks (Granovetter, 1977, 1983), the weak tie model of change holds that the structural properties of social networks can account for the general tendency of some language communities to be more resistant to linguistic change than oth- ers (Milroy and Milroy, 1985, 1992; Milroy and Llamas, 2013). A classic finding is that loose-knit networks with mostly weak ties are more conducive to information diffusion, thereby facilitating inno- vation and change, while close-knit networks with strong bonds impose norm-enforcing pressure on language usage, strengthening the localized linguis- tic norms (Milroy and Milroy, 1985). One compelling observation in favor of this ar- gument concerns the comparison between two Ger- manic languages, Icelandic and English. Icelandic has changed little since the late thirteenth century, which could be due to the norm-enforcing pressure inherent in the strong kinship and friendship ties. In contrast, in Early Modern London English, the loosening of network ties, accompanied by the rise of the mobile merchant class, was argued to be re- sponsible for some radical change in the language (Milroy and Milroy, 1985). This study extends network-based sociolinguis- tic research to online communities, which re- main understudied despite their expansion …

Content truncated to 500 words