Sociolinguistic Priming and the Perception of Agreement Variation: Testing Predictions of Exemplar-Theoretic Grammar.

Abstract

This dissertation investigates the sociolinguistic perception of morphosyntactic variation and is motivated by exemplar-based approaches to grammar. The study uses syntactic priming experiments to test the effects of participants' exposure to subject-verb agreement variants. Experiments also manipulate the gender, social status, and individual identity of the talkers to whom participants are exposed, testing the influence of social information on the perception of agreement variation. Access to social information about a speaker has been found to influence the perception of the linguistic forms they produce. Exemplar-theoretic models of speech perception accommodate these findings by positing that linguistic knowledge consists of episodic memory traces of experiences with language, and that linguistic exemplars represent rich social details. Exemplar-theoretic models of syntax likewise posit that syntactic knowledge is a function of direct experiences with language. However, syntactic exemplar theorists have not explored patterns of sociolinguistic variation, and sociolinguistically-informed exemplar-theoretic work has focused on patterns of phonological variation. This study hypothesizes that for grammatical variation that is sociolinguistically patterned, grammatical processing will show sensitivity to both social and linguistic influences in the processing context. The dissertation experiments use structural priming, a paradigm common in psycholinguistic research for exploring cognitive representations of syntactic structure. The experiments manipulate participants' exposure to variants of two subject-verb constructions that alternate commonly across English dialects: NPsg/pl+don't (The dog/dogs don't bark) and there's+NPsg/pl (There's a dog/dogs in the yard). The experiments find effects of recency, social similarity, and constructional frequency on participants' interpretation of agreement forms, supporting central features of a socially rich exemplar-based grammar. The study shows that grammatical perception is sensitive to priming, such that exposure to a nonstandard variant in the prime sentence increases the likelihood that participants will perceive a nonstandard variant in the target sentence. Priming is also differentially affected by the social dimensions of gender, social status, and talker specificity. The dissertation argues that the notions of indirect, direct, and potential indexicality capture these differences, and that they can be accommodated by a model of grammatical knowledge that includes multiple levels of abstracted linguistic and sociolinguistic categories.