My research focuses on one of the most pervasive facts about the human language faculty, gradience, and investigates how gradient effects in language inform us about the nature of the grammar, the parser, and their relationships. I pursue this goal by focusing on phenomena that are at the core of syntactic theories, and that, at the same time, challenge existing theories of both grammar and parsing with their gradient behavior.
At the empirical level, my research focuses on core syntactic phenomena that exhibit a gradient behaviour, such as island effects and the strong/weak island distinctions. I probe gradience through a variety of experimental techniques, such as acceptability judgments, self-paced reading, maze task, and speed-accuracy trade off (SAT), and I employ statistical methods that can allow us to precisely estimate the magnitude of gradient effects, such as Bayesian methods.
At a theoretical level, my research attempts to construct computationally explicit models of grammar, processing, and their link that span the full space of possible explanations for gradience in order to determine which systems can best account for gradient facts. There are (at least) two dimensions that define the space of possible theories of gradience: whether the grammar is categorical or gradient, and whether the parser is tightly or loosely aligned with the grammar. So far, I have been deeply involved in constructing a novel theory – self-organized sentence processing (SOSP) – that combines a gradient grammatical theory and a gradient, tightly-aligned processing mechanism into a single framework .
Sandra Villata 