Guest speaker: Christopher Hammerly (University of Massachusetts, Amherst)

Our department is pleased to welcome Christopher Hammerly, a Ph.D. candidate at the University of Massachusetts, Amherst. His research focuses on psycholinguistics and morphosyntax, and he has a particular interest in Ojibwe/Anishinaabemowin in accordance with his own heritage. He will be giving a talk, "Number representation and memory in agreement processing," in SS 560A at 3:00 PM on Friday, January 17.

Agreement attraction occurs when there is a disruption in the regular pattern of number agreement between the subject and the verb due to the presence of a number mismatching distractor noun. For example, many studies have shown that more plural verb forms are produced in the mismatch sentences in (1a) compared to the match baseline in (1b), despite the subject being singular in both cases. 

(1a) The key to the cabinets... (mismatch)
(1b) The key to the cabinet... (match)

Over the nearly 30 years that this phenomenon has been investigated, different theories have emerged to capture these effects in sentence production versus comprehension. Production errors have been explained by an encoding account (e.g. Marking andMorphing), where the number of the singular subject is misrepresented as plural in memory. Comprehension errors have been explained by a retrieval error (e.g. as implemented in ACT-R), where the attractor is mistakenly computed as the controller of agreement. In this talk, I argue for a unified processing source for agreement attraction by extending the encoding account to comprehension. This extension is based on two key findings. First, that agreement attraction in comprehension can be observed in both grammatical and ungrammatical sentences. This is predicted under current encoding accounts, but not memory retrieval accounts, and stands in contrast to previous experimental results that found attraction only in ungrammatical sentences. Second, I show that the size of agreement attraction effects is a function of syntactic distance between the attractor noun and the subject, a result directly predicted by current encoding theories.