Between “Threatened” and “Endangered”—The Validity of Vitality Labels

Language vitality scales are commonly used in language documentation, yet assumptions behind labels such as “threatened” and “endangered” are often taken for granted. This study adopts a computational-based approach to disciplinary discourse analysis, examining how linguists deploy “threatened” and “endangered,” terms that appear on vitality scales, across scholarly discourse. We compared vocabulary and thematic content in “threatened” and “endangered” texts using TF-IDF, topic modeling, and similarity metrics, with other linguistic terms providing contextual benchmarks. Keyword analysis shows that “threatened” and “endangered” index related but distinct orientations within linguistic research. Writing on “endangered” languages more frequently centers on documentation, archiving, and institutional research practices, while writing on “threatened” languages places greater emphasis on language vitality and ongoing sociolinguistic change. The magnitude of their difference is contextualized by contrasting “threatened” and “endangered” with other conceptual pairs within linguistics and beyond. The findings suggest that “threatened” and “endangered”, which are consecutive labels on the Language Endangerment Index, are more closely related than distinct theoretical constructs but less related than near-overlapping linguistic terms. Overall, the results demonstrate that these terms, as deployed in disciplinary practice, capture systematic distinctions in how linguistic situations are understood. Methodologically, this paper shows how computational analysis can be used as a reflexive research tool that enables us to better understand disciplinary discourse.