Why Should a Demonstrative Turn Into a Preposition? The Evolution of Welsh Predicative yn

This article is devoted to the anatomy of an unnatural syntactic change. It presents the life history of the Welsh predicative particle yn—its diachronic genesis in Indo-European, its synchronic status, and (much more centrally) what happened along the way, and why what happened happened specifically in Welsh. Synchronically, I give syntactic, semantic, typological, and textual arguments—some rather new—that both predicative yn and verb-periphrastic yn are adverbializers and count as grammatically polysemous subsenses of the preposition ‘in’. Diachronically, I argue that the pan-Celtic adverbializing particles yn/ent/int/ind (thence ultimately Welsh predicative yn) all derive from an article-like demonstrative *sindo-/sinto- (and not from a preposition *endo/ento). Radical categorial changes must therefore have occurred. I trace these changes and motivate a multistage metanalysis (not involving grammaticalization) whereby the original quasi-article first became an adverbializer and then was attracted into the orbit of the preposition ‘in’. Though each microstage in the process makes good structural sense vis-à-vis the evolving système of the language, the achieved macrochange is highly unnatural.^*

1. Overview

Syntactic change can sometimes be exceedingly messy. Such cases tend by their very nature to be inscrutable, hence intractable, and so are less often dealt with than the ‘cleaner’ cases more typically found in the literature, for example, in grammaticalization studies. Nevertheless, in those instances where we can actually follow (parts of) a tangled process comparatively and historically, such a case study can provide enormous insight into the sheer complexity of syntactic change—a complexity that may well be the rule and not the exception. In this article I undertake the comprehensive reconstruction of one such change: the grammatical and functional evolution of Welsh predicative yn, from Indo-European up to the present day. The multiple categorial shifts involved are in part motivated by general principles (e.g. of grammaticalization), but in much greater part are a response to the overall system of the particular language and its changing dynamics. As I seek to demonstrate, contingent, language-specific phonetic similarities across distinct morphemes, and partial functional affinities among them, motivate a series of microchanges which, though individually reasonable, come together to create an achieved macrochange that is bizarre: the metamorphosis of a demonstrative into a preposition ‘in’.

The Welsh predicative particle yn defies easy categorial analysis. Depending on what aspect it is viewed under, the particle—exemplified in 1—seems to take on a totally different character.

(1)

From a straightforward language-user’s perspective, and that of traditional Welsh grammatical description, it appears simply to be the preposition ‘in’ used in a special way; the preposition occurs in a cluster of four distinct but related uses, conditioning different mutations (in bold), as shown in Figure 1.¹ Such a polysemy approach, whereby these various yns are not accidental grammatical homophones but are in some sense synchronically ‘the same’ (despite the differences in mutation), has been expounded in detail within a cognitive framework and given a principled justification by James Fife (1990). Generative syntacticians, conversely, have almost unanimously opposed this approach, adducing differences between the more grammatical uses (B, C) and the basic spatiotemporal use (A) in support of the view that the former and the latter involve completely different elements. The particle’s synchronic status, then, is anything but clear.

Figure 1.

Uses of Welsh yn.

Seen in diachronic and comparative Celtic perspective, predicative yn becomes still more vexed. Whatever its status synchronically in Modern and Middle Welsh, it clearly does not have the same origin as the spatiotemporal preposition yn but (via adverbial yn) goes back to a form int that looks very like the definite article, as Arwyn Watkins (1962) has shown. If indeed this particle originally had to do with the article—a point requiring some argumentation; see §§5–8—then we are faced with a paradox. For, whatever the proper synchronic analysis of predicative yn, the one thing it cannot be synchronically is an exponent of definiteness: over the entire history of Welsh the particle...