Social and constructional diffusion: Relative clauses in seventeenth- and eighteenth-century Dutch
In this paper, we discuss a significant change in the morphosyntax of Dutch, arguing that bringing together evidence from (historical) sociolinguistic and constructional analyses advances our understanding of language variation and change. In so doing, we link up with recent developments in cognitive linguistics, where a “social turn” has been argued for (Croft 2000, 2009; Kristiansen and Dirven 2008; Geeraerts, Kristiansen, and Peirsman 2010; Harder 2010; Hoffmann and Trousdale 2011). The change under discussion concerns the rise of w-relativizers such as waar ‘where’ at the expense of d-relativizers such as daar ‘lit. there, where’. The change will be introduced in more detail in section 2. In section 3, we discuss previous research, and formulate two hypotheses following suggestions in the literature. First, we hypothesize that the change is a so-called change from above in the social sense, that is a change that spreads from the upper ranks of society to the lower ranks. Second, we hypothesize that the change spreads from construction to construction. We assume, therefore, that the change exhibits diffusion in at least two ways: social and constructional. We will use a socially stratified corpus of seventeenth- and eighteenth-century private letters to investigate this. The corpus will be introduced in section 4. In sections 5 and 6, we argue that both social and constructional diffusion can indeed be found in the change from d- to w-relativization. Section 5 shows that w-relativizers are more in use in letters from the upper ranks of society than in letters from the lower ranks. The private letters in our corpora are characterized by a great number of epistolary formulae. In section 6, we will argue that these formulae can be considered as constructions in the sense of construction grammar, after which we will show that w-relativizers appear to spread from formula to formula, that is from construction to construction. In section 7, we discuss the results, and moreover argue that a constructional approach may gain from sociolinguistic analyses.