Structural priming as a window into children's syntactic acquisition

What kinds of grammatical representations support children's use of their native language? Do young children rely primarily on verb-specific templates or an abstract grammar that includes grammatical categories? This issue has been avidly explored, but not settled, over the past few decades. Progress has been impeded by methodological limitations. This thesis develops a novel method for studying the mental representations that underlie children's language and if and how they change over time.;Paper 1 validates the new method. In three experiments, I show that a combination of the structural priming and eye-tracking techniques can be used to characterize the abstract grammatical representations that adults most certainly possess. Category-based representations are shown to guide adults' comprehension of sentences contrary to prior claims in the literature that comprehension may rely more on verb-specific representations than production.;Armed with a method that is sensitive to abstract grammatical representations (when they are present) I asked what representations young children (3- and 4-year-olds) use for understanding sentences. The four experiments in Paper 2 show that children's interpretation of sentences is influenced by previous sentences containing either the same or different verbs. The across-verb priming results clearly show that children as young as three use abstract representations for comprehension, even when verb-specific representations are (presumably) sufficient for the task at hand. These results adjudicate between the contradictory findings and interpretations from previous studies that used different methods.;In Paper 3, I explore the nature of children's abstract representations. Are they purely syntactic or at the interface between syntax and either conceptual or semantic representations? By manipulating the nature of the overlap between prime and target sentences, I show that 4-year-old children are sensitive to the similarities in semantic roles between events as disparate as giving, showing and loading. They (implicitly) expect these broad semantic roles to be positioned similarly in sentences containing different, but related, verbs.;Together, the results show that by age four, children possess and use abstract syntax-to-semantics mappings in ways that are similar to adults. The method described here is well-suited for exploring whether 2- and 3-year-olds' language is similarly sophisticated.