Pronominal and null subject variation in Spanish: Constraints, dialects, and functional compensation

The occurrence of phonetically null subjects in languages such as Spanish has resulted in wide-ranging research in generative syntax and discourse analysis. However, neither discipline accounts for the statistical variability inherent in the alternation of null and lexical subjects. Working within the variationist approach to language description and explanation, the purpose of this study is to discover the significant constraints on the alternation of pronominal and null subjects in the Spanish spoken in San Juan, Puerto Rico and Madrid, Spain. In addition, I seek explanations for why these two dialects differ in the rate of subject pronoun expression. This includes investigation of the functional compensation hypothesis for Puerto Rican Spanish (Hochberg 1986) in a socially stratified study of the San Juan speech community. According to this hypothesis, a connection exists between pronominal expression and the variable deletion of second person inflectional /S/. Using strategies from both syntax and pragmatics, I first provide an explicit statement of those contexts in which pronominal and null subject variation is prohibited in order to define where such variation may occur. In turn, because Switch Reference is the central constraint on variation, a working definition of the scope and limitations of Switch Reference is provided. The major constraints which emerge include Switch Reference, Reference Chains, Set-to-Elements Saliency, Morphological Ambiguity when intersected with a switch in reference, and a Priming effect in the context of sameness of reference. Using multivariate analysis, the speakers from San Juan and Madrid show great similarity in the varbrul weights assigned to these constraints. However, they show a reverse effect associated with the referential specificity of second person TU. Whereas in San Juan Nonspecific TU favors pronominal expression, in Madrid it disfavors pronominal expression in relation to Specific TU. Finally, despite the appearance of functional compensation for Specific TU in the San Juan groups of female speakers, lower class speakers, and speakers interviewed in groups, this does not result in different rates of pronominal expression between the groups which evince functional compensation and those which do not. Therefore, the functional compensation hypothesis finds no clear support