Speaker: Juliet Stanton (MIT) Title: Trigger Deletion in Gurindji Time: Thursday, November 12th, 12:30-1:45 pm Place: 32-D461
It is generally accepted in the literature that harmony processes are myopic. For example, a regressive harmony process operating on the string […x y z…], will spread from z to y, without looking ahead to check whether it can spread all the way to x, the end of the domain. Accounting for the apparent absence of non-myopic patterns has led analysts to propose substantial revisions to the architecture of classical OT (e.g. Wilson 2006, McCarthy 2009). In this talk, however, I suggest that a non-myopic, long-distance nasal harmony process is attested in Gurindji (Pama-Nyungan; McConvell 1988). When full application of harmony would lead to an undesirable result, the trigger deletes, preventing harmony from applying altogether. Trigger deletion is predicted by frameworks that permit non-myopic interactions, like parallel OT; the existence of the Gurindji pattern suggests that this is a feature, not a bug.