Structure Building in Language Production

University Of Massachusetts Amherst, Amherst MA

Investigators

Abstract

Humans have the remarkable ability to express novel and complex thoughts in well-formed sentences efficiently. This uniquely human ability, the ability to produce sentences, is critical in many, if not all, aspects of human activities. However, how speakers' minds create sentences remains poorly understood. This project aims to study the cognitive mechanisms for sentence production, focusing on two fundamental properties that make human languages distinct from non-linguistic communicative systems: recursion and displacement. Human languages allow recursion; speakers can produce sentences that contain another sentence inside it, as in "The girl believes that the boy likes the dog." Human languages also allow displacement: speakers can produce sentences in which two things that should be normally close together appear far apart, as in "Which dog did the girl believe that the boy likes?", where the first phrase ("which dog") is more closely related to the farthest verb ("likes") than the closer one ("believe"). Recursion and displacement are two hallmark properties of language that contribute to its complexity and richness, but researchers know little about how speakers' minds create sentences involving them. The current project aims to fill this knowledge gap by studying how speakers of two very different languages create sentences involving recursion and displacement. This project investigates the cognitive mechanisms supporting the production of sentences involving recursion and displacement specifically through comparative psycholinguistic experimentation in two typologically distinct languages. The research team primarily uses structural priming to study what kind of building blocks speakers use and how they combine these building blocks to create sentences involving recursion and displacement. Structural priming refers to the tendency of speakers to reuse the sentence structures they have recently encountered, and this effect can be used to make various inferences about what kind of building blocks speakers use and how they combine these blocks to create well-formed sentences. The cross-linguistic data generated in this research are directly compared to better understand how grammatical differences affect cognitive mechanisms for sentence production. State-of-the-art data analysis methods involving hierarchical Bayesian modeling are used. This project involves advanced training in linguistics, cognitive psychology, and statistics for undergraduate students, graduate students, and post-doctoral scholars and also establishes international collaborations between students and researchers. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.