Congratulations to Grusha Prasad for receiving an American Psychological Association (APA) Dissertation Research Award! The title of the dissertation is “Using priming in humans and neural networks to evaluate predictions from syntactic analyses”. Very well deserved!
How do people represent sentences? Structural priming has been used widely in psycholinguistics to study this question. The logic underlying this paradigm is as follows: for two sentences, a prime and a target, which share some properties, if the processing or production of the target is facilitated by the prime, we can infer that people’s sentence representations are sensitive to at least one of the properties shared by the two sentences. Prior work has drawn on representational assumptions made by theories in syntax, a branch of linguistics studying the structure of sentences, to generate verbal hypotheses about structural properties mental representations of sentences are expected to be sensitive to and then tested these hypotheses by carefully controlling the properties shared between the prime and target. While this body of work has has revealed whether mental representations are sensitive to specific structural properties, it has not been able to directly evaluate the holistic predictions made by representational systems hypothesized by different theories in syntax. Building on this work, in my dissertation I propose a method to directly compare the psychological plausibility of the representational assumptions of different syntactic theories by generating quantitative predictions about the extent to which sentences are expected to prime each other under each of these theories. I propose to then test these predictions using a novel online behavioral paradigm.