Speaker: Peter Grishin (MIT)
Title: How to agree with the lowest DP
Time: Tuesday, March 15th, 1pm – 2pm
Abstract: In Passamaquoddy (Eastern Algonquian), there is an agreement marker (the “peripheral suffix”) that sits in C, c-commanding all the arguments of the verb, and demonstrably indexes the lowest clausemate DP after A-movement—for instance, in direct configurations, it’ll index the object, but in inverse configurations (involving A-movement of the object over the subject; Bruening 2001, 2005, 2009), it’ll index the subject. This poses a striking puzzle for standard probe-goal models of agreement (Chomsky 2000, 2001, a.m.o.) which all share the property of locality: a probe will agree with the closest matching accessible goal. Not here!
I propose a morphological solution to the puzzle: the peripheral suffix has agreed with all the accessible matching goals in its domain, and when it comes time to spell out the probe, we chose the newest feature bundle that the probe has acquired—we can call this “Expone Outermost”. I assume that Expone Outermost is a probe-specific morphological rule that specifies what you are to do when spelling out a terminal that has multiple distinct feature bundles on it, situating it in a broader typology of “repairs” for feature gluttony (Coon and Keine 2021): e.g. spelling out all the features individually, spelling out a portmanteau form, or choosing the most-specified feature bundle to spell out. I implement this analysis for Passamaquoddy in an Interaction-Satisfaction model of Agree (Deal 2015, to appear), analyzing the peripheral agreement as an insatiable probe with the specification [INT:φ, SAT:–] which solves the problem of feature gluttony by Expone Outermost.
Expone Outermost expands the typology of possible agreement systems—is this move justified? I argue that it is: by varying interaction and satisfaction conditions while keeping Expone Outermost constant, we predict various kinds of interesting agreement systems—and these agreement systems are attested. Narrowing the interaction condition to [INT:PART] results in a probe that indexes the lowest first or second person DP—this is attested in Aqusha Dargwa (Nakh-Dagestanian). Narrowing the interaction condition to [SAT:AUTH] results in a probe that always agrees with the lowest DP, unless there’s a first person intervener, in which case it agrees with the first person DP—this is attested in Lak (Nakh-Dagestanian). Finally, we predict that if there are contexts where we Impoverish the outermost features on the probe, the inner features should get a chance to “expose themselves”—we find this pattern in Karitiâna (Tupí-Guaraní) antiagreement, where Ā-extracting the object prevents you from realizing the features of the object, resulting in otherwise unattested subject agreement.