Are languages and dialects inventions of nationalism? On the problem of totalising metalanguage

Scholars of nationalism generally acknowledge that what counts as ‘a language’ and what as ‘a dialect’ is determined by historical and political circumstance, that both notions are idealisations of linguistic practice rather than objective entities and that the reality of language is fuzzy and complex. However, they nevertheless continue to talk about (and analyse) ‘linguistic entities’ in the same way nationalists do: as homogenous, closed systems. Paralleling Brubaker's groupist language, this paper proposes the notion of totalising metalanguage to signify all ways of talking about linguistic phenomena that reify them into unproblematically existing objects. I analyse the historical development of dialectology in Croatia and Serbia from the 19th century until today to show how dubious linguistic taxonomies have been presented in the discourse as objectively existing linguistic entities. The paper invites scholars of nationalism to seek alternative approaches to the language‐dialect dichotomy than that offered by the outdated model of Joshua Fishman.