The network nature of language endangerment hotspots

Language endangerment is one of the most urgent issues of the twenty-first century. Languages are disappearing at unprecedented rates, with dire consequences that affect speaker communities, scientific community and humanity. There is impetus for understanding the nature of language endangerment, and we investigate where language endangerment occurs by performing network analysis on 3423 languages at various levels of risk. Macro-level analysis shows evidence of positive assortative mixing of endangerment statuses—critically endangered languages are surrounded by similarly endangered languages, indicating the prevalence of linguistic hotspots throughout the world. Meso-level analysis using community detection returned 13 communities experiencing different levels of threat. Micro-level analysis of closeness centrality shows that more geographically isolated languages tend to be more critically endangered. Even after accounting for the statistical contributions of linguistic diversity, the structural properties of the spatial network were still significantly associated with endangerment outcomes. Findings support that the notion of hotspots is useful when accounting for language endangerment but go beyond that to establish that quantifying spatial structure is crucial. Language preservation in these hotspots and understanding why endangered languages pattern the way they do in their environments becomes more vital than ever.