Brian Dillon, University of Massachusetts Amherst
Illusory agreement phenomena: Insights from English and Hindi
“Grammatical illusions” occur when speakers and listeners seem unable to faithfully apply their grammatical knowledge during the course of analyzing or producing language (Phillips, Wagers, and Lau, 2011). The distribution of grammatical illusions across constructions and across languages has led to insights into the nature of the cognitive systems that speakers use during the course of routine language comprehension and production. In this talk, I will focus on one such a grammatical illusion: agreement attraction, or the tendency for speakers to express verb agreement with nouns other than the intended agreement target (e.g. ‘The key to the cabinets are rusty’; Bock & Miller, 1991). This phenomenon is a useful model to better understand how morphosyntactic features are bound to syntactic structure during realtime language production and comprehension. In this talk I will give a broad overview of a range of findings from our group on this phenomenon. Using acceptability judgments and eye-tracking-while-reading, we find that agreement attraction phenomena likely results both from noise in the encoding of morphosyntactic representations in working memory, as well as errors in retrieving or accessing information from working memory (work led by Christopher Hammerly, UMass Amherst). Interestingly, the nature of this interference varies in clear ways across languages. In a series of studies on the processing of agreement in Hindi (work led by Sakshi Bhatia, IIT Delhi), we show that what makes a strong agreement ‘attractor’ in English is very different than what makes a strong agreement ‘attractor’ in Hindi, in ways that seem to reflect the different grammatical properties of the agreement systems in these two languages.