The place of locatives and the limits of the arbitrariness of signs

Abstract

Locative forms are particularly wide-spread and salient in distributions that do not trivially come with a locative interpretation, such as the expletive subject in `There is a fiddler on the roof'. This phenomenon is central to issues such as the formalization of linguistic variation, the organization of the Lexicon with respect to poly-functional morphological forms and the syntactic analysis of specific constructions that come under its scope. Consequently, I begin by discussing the conceptual underpinning of such phenomena in order to make sense of this specific domain and inform research into the issues above (variation, the organization of the Lexicon, specific syntactic topics).

More concretely, I propose a way of dealing with the morphological `locativity' of expletive subjects in some languages, in keeping with contemporary guidelines. Arguing against approaches that insist on introducing `locativity' as a formal feature into the computation (and parametrizing its presence), I opt for identifying a salient `subjecthood' property and hypothesize on how locatives may come to host it.

Other distributions of locatives, such as the Danish relative `der' or the Dutch function cluster, are also discussed in terms of how they could be integrated into a robust model of variation. The resulting pattern is also argued to suggest possible modifications to our model of the Lexicon. Thus, the emerging framework of Nanosyntax is put to the test as a technology that can capture the scalarity of functional polysemy. Dutch r-words (which are notoriously poly-functional locatives) are (re)analyzed as deriving from a principled morphological split in pronominal paradigms, partially reflected in dialectal variation. Their locative interpretation remains orthogonal to the pattern of distributions.

I reach the conclusion that we have a strong case for individually explaining the various distributions of locatives through mechanisms that render their primary interpretation as epiphenomenal to their functional interpretations. As an extension of this, I advance the hypothesis that the range of distributions of locatives (and other poly-functionals) can be captured in a scalar fashion that reflects a central design feature of natural language.