Constructions of the type la cause/la raison/la preuve + utterance: Grammaticalization, pragmaticalization, or something else?

French makes use of a series of nominal constructions that provide a challenging object of linguistic research. They are built on a predicative NP that introduces a clause without resorting to any complementation marker such as [que] to connect the two (e.g. la cause/la raison/la preuve + utterance). At first glance, these structures could either be treated as the result of a grammaticalization process originating in a specific use whereby the noun governs the clause via a copula (la preuve est que, la raison est que, la cause est que), or as the outcomes of a pragmaticalization process occurring at the discourse level. However, a close examination of diachronic data strongly suggests that they should rather be seen as instances of a routinization of complex discourse structures. These constructions may be analysed as coupled segments (of the question–answer, deictic marker–referent type) whose traits have become more and more fixed through routinization.