Subject pronoun variation in Central Romance

David Heap

Sat. 9:00-10:40 B

Existing research into the status of subject pronouns in Northern Italian and Gallo-Romance varieties has largely focused on the distinction between grammars with (obligatory) clitic subjects, such as French and Trentino, and those with only (optional) strong subject pronouns, such as Standard Italian (Renzi & Vanelli 1982, Brandi & Cordin 1989, Safir & Jaeggli 1989). For the most part, researchers have treated these distinctions in a categorically binary fashion, in keeping with the Principles and Parameters approach to grammatical variation (Chomsky 1981, Roberge & Vinet 1989). By moving beyond a priori dichotomies which divide grammars on the basis of abstract parameters such as [+/- null subject] or [+/- subject clitic], this paper aims to achieve a more comprehensive characterization of how morphological systems can vary across a geolect continuum.

While it has been recognized that subject clitics may be unevenly distributed throughout a pronominal paradigm, and that an implicational hierarchy can apply between the different members of such a paradigm (Renzi & Vanelli 1989), the true degree of variability across subject pronoun systems has yet to be adequately explored. The relatively small sample of varieties considered (typically one or two), as well as the heterogeneous data which do not lend themselves well to comparison, have prevented any overall topological view of these phenomena from emerging from previous studies. A preliminary survey of geolectal data suggests that there in fact exists a considerable amount of variation in clitic pronoun usage even in varieties considered to be "categorically" subject clitic varieties in the existing literature.

This paper provides a more accurate and nuanced picture of subject pronoun variation in the Central Romance morphosyntactic continuum. With respect to the empirical base, a much finer grid of geolectal material is used than in previous studies: atlas questionnaire data from hundreds of points are correlated in a relational database, leading to truly comparable results for a large number of related yet different varieties. Potential factors which have been shown to interact with subject pronouns (cf. Nadasdi 1995), such as verb and clause types, tense, and the presence of another [object] clitic, are all taken into account in a GoldVarb 2 analysis of factor effects. The results show complex and variable linguistic continuum which cannot satisfactorily be reduced to the formalisms of abstract grammar.