Doctoral Dissertation Research: Computing agreement in a mixed system - A psycholinguistic comparison of subject and object agreement

University Of Massachusetts Amherst, Amherst MA

Investigators

Abstract

This project investigates speech errors associated with the production of verbal agreement in Hindi, a widely-spoken language with a complex verb agreement system. Speaking one's native language does not usually feel like a particularly difficult or demanding task: in most situations adults do this without experiencing any noticeable difficulty. Despite this, much research into language production has shown that it is a highly complex cognitive task, and that there are many mental processes involved in translating an abstract idea to a linguistic message that we can communicate with our mouths or hands. Understanding the nature of these mental processes is important both for understanding how to develop better treatments for language disorders, as well as for developing more sophisticated natural language technology. The present project contributes to a better understanding of how language production proceeds by investigating speech errors in Hindi. Speech errors provide one of the most important sources of insight into natural language production. However, the majority of research into speech errors during language production is focused on English, potentially limiting the scope of this research. The present project fills a gap in our knowledge by investigating speech errors in Hindi, a language that has very different grammatical properties than English. In addition, the present project increases the visibility of Hindi for psycholinguistic research. Hindi remains highly understudied in psycholinguistic research on language production and comprehension, despite its many unique grammatical properties and the fact that it is the third most spoken language in the world, with over half a billion speakers worldwide and over half a million in the United States alone. Previous research on European languages has shown that speakers will sometimes produce verb agreement errors in certain grammatical contexts. For example, in the sentence 'the key to the cabinets is rusty', English speakers will sometimes produce the incorrect plural form 'are' instead of the correct is due to interference from the plural noun 'cabinets'. Research has shown that this interference is highly sensitive to the structural properties of the 'distractor' noun--'cabinets'. However, the precise structural properties that make a linguistic expression prominent and likely to create interference during language production are not well understood. Hindi can help shed light on this. Unlike other studied languages, Hindi has a 'mixed' agreement system: in some contexts, the verb agrees with the subject (as in English), and in others it agrees with the object. Here, the researchers will use structured experiments utilizing two methodologies--the spoken production paradigm and the binary choice agreement paradigm--to identify the circumstances under which Hindi speakers produce errors in verb agreement in both subject agreement configurations and object agreement configurations. In comparing error patterns for subject agreement and object agreement, this research will shed light on what factors make linguistic expressions cognitively prominent during language production. The results will be of interest to psycholinguists developing models of agreement production and to linguists interested in the grammatical generalizations that speakers acquire about their native language. 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.