Background and hypotheses
The change from d- to w-relativizers has been the topic of a vast body of recent studies in Dutch historical linguistics (van der Horst 1988, 1993; van der Horst and Storm 1991; Schoonenboom 1997, 2000; de Schutter and Kloots 2000; van der Wal 2002, 2003; Rutten 2010; Rutten and van der Wal acc.). It is often assumed or claimed that the change proceeded from indefinite to definite contexts (cf. van der Horst and Storm 1991: 115; van der Horst 1993: 300; van der Wal 2002; de Schutter and Kloots 2000: 327; Schoonenboom 2000: 137-138, 157-158), adopting the following cline of (in)definiteness (van der Horst 1988: 96). The first category would constitute the most indefinite context, whereas definite NP’s represent the most definite contexts.
Indefinite 1 the antecedent is absent or implicit: je moet doen wat ze zegt ‘lit. you must do what she says, you must do as she says’
- 2 the antecedent is an entire clause: ze deed erg haar best, wat wij heel flink vonden ‘she really did her best, which we considered very plucky’
- 3 the antecedent is a word such as iets ‘something’, niets ‘nothing’, alles ‘everything’, veel ‘much’, weinig ‘little’, genoeg ‘enough’: alles wat hij wil ‘lit. everything what he wants’, everything he wants, is er iets wat/dat ik voorje kan doen? ‘lit. is there something what/ that I can do for you?’
- 4 the antecedent is a nominalized adjective, mostly a superlative: het vriendelijkste wat/dat Jan heeft gedaan ‘lit. the most friendly what/that Jan has done, the most friendly thing Jan has done’
- 5 the antecedent is an indefinite NP: een boek dat (wat) ik mooi vind ‘lit. a book that (what) I like’
Definite 6 the antecedent is a definite NP: het boek dat (wat) ik gisteren las
‘lit. the book that (what) I read yesterday’
Counterexamples, however, are easily found. There are, for instance, eighteenth-century diaries with w-relativizers in category 3 that have d-relativizers in other categories (Rutten 2010). Therefore, it has been argued that this cline of (in)definiteness may not be able to explain all empirical facts and that a different approach drawing on construction grammar and frequency may better explain the spread of w-relativizers (Rutten 2010).
Furthermore, Rutten and van der Wal (acc.) show that some constructions subsumed under the second category are indeed among the first to adopt w- forms, but this only applies to so-called continuative relative clauses, not to appositive relative clauses with the preceding clause as antecedent in general. Moreover, continuative relative clauses are ahead of other categories, including the first category. Continuative relative clauses have the form of subordinate clauses while the semantics favors a main clause interpretation (cf. Loock 2007). Considering continuative relative clauses as schematic constructions in the sense of construction grammar, Rutten and van der Wal (acc.) show that an explanation of the spread of w-relativizers may gain from a constructionist view of the language system.
As is well-known from variationist studies, changes may also display social and stylistic diffusion, in that some (groups of) speakers may adopt new forms at a faster pace than others. With regard to changes in relativization strategies, it is often assumed that the rise of wh-forms in the history of English was a so- called change from above, ‘from the formal and literary levels of the language’ (Rissanen 1999: 295). Rissanen (1999: 293) also points out that wh-forms were well-established in appositive relative clauses in the sixteenth century, though the old form that is still found in texts representing the oral mode of discourse (cf. Bergs 2005: 181). Concerning eighteenth-century Dutch, it has been suggested that d-relativizers such as daar ‘there, where’ are more frequently found in texts closer to the spoken language such as diaries written by lesser skilled writers (Rutten 2010). Recall in this respect that present-day standard Dutch as well as most dialects only exhibit w-relativizers in locative constructions, d-forms being restricted to optional usage in a limited number of dialects (section 2).
The present study continues the line of research explained above by focusing on two main issues addressed in the literature. Can a constructionist view indeed further our understanding of the change from d- to w-relativizers? And can we identify social variation in the spread of w-forms? The corpus of seventeenth- and eighteenth-century private letters used for the present study allows us to answer these questions. The corpus is primarily designed for sociolinguistic research (section 4), and we will therefore first investigate possible social effects on the diffusion of w-forms. Since it has been suggested that w-forms were promoted “from above”, we hypothesize that we will establish social variation, with upper ranks using w-forms more frequently than lower ranks (see section 5) - despite the fact that most changes qualify as changes from below, particularly when there is no overt metalinguistic discourse on the change nor on the variation it involves, as is the case with d- and w-relativizers (van der Wal 2003).
Building on previous research, we also hypothesize that the change spreads from construction to construction through a process which might be termed constructional diffusion. We will take the opportunity to investigate this by studying a set of constructions characteristic of the language of private letters and different from the constructions studied before (cf. Rutten 2010; Rutten and van der Wal acc.).
-  The examples are taken from van der Horst (1988: 96), which is a study of the relative pronouns dat and wat. As to the choice of relative pronoun, Van der Horst distinguishes threeoptions, representing late-twentieth-century standard Dutch: wat as in catergories 1 and 2,wat/dat as in 3 and 4, indicating that both are possible, and dat (wat) as in 5 and 6, whichmeans that dat is the norm, while wat is often used, too.
-  It has been argued by van der Horst and Storm (1991), van der Horst (1993), de Schutter andKloots (2000) and Rutten (2010) that in the case of pronominal adverbs the lexical form mightbe important, either fused into one lexeme (e.g. het vuur waarin ze branden, zal niet doven ‘thefire where-in/ in which they burn, will not smother’) or separated into two lexemes (het vuurwaarze in branden, zal niet doven ‘lit. the fire where they in burn, will not smother’). This is animportant observation. In the present study, we will take into account both fused and separateforms, focusing only on the initial consonant (d or w). We hope to be able to address the lexicalform in subsequent research.