On agree feeding interpretation: honorification, empathy, and switch reference

Agree is one of the few core syntactic operations posited in minimalist syntax (Chomsky, 2000, 2001) and as such its nature has been subject to debate from various perspectives. While the standard assumption is that the result of Agree does not affect interpretation, this dissertation shows that Agree can not only affect PF but can also affect LF by providing arguments to some semantic predicates. In particular, I claim that some functional heads are semantic predicates that select their arguments via Agree over index features.I first establish the claim by examining Object Honorifics in Japanese. Previous syntactic literature has considered object agreement over an honorific feature borne by the object (Toribio, 1990; Hasegawa, 2017, a.o.). This approach has been supported by the observation that the distribution of the object that controls object honorific marking is constrained by the configurational cosntraints on Agree (Niinuma, 2003; Boeckx and Niinuma, 2004; Boeckx, 2006). I show, however, that the Object Honorific construction expresses an honorific relationship between the subject referent and the object referent, which cannot be reduced to the semantics of the honorific feature on the object. I claim that the head responsible for the Object Honorific marking is a semantic predicate honor, and finds its arguments via Agree. I go on to extend the idea to three other phenomena, Japanese Benefactive auxiliaries, the Direct / Inverse system in Japhug and the typology of Switch Reference. In each of these phenomena, the pivotal head establishes a relationship with some nominal expressions already in the sentence. While this relationship can be established only in a configuration where an Agree relationship can be established, each phenomenon involves components that cannot be reduced to a feature on the nominal expressions. I will show that the idea that semantic predicates can select their arguments via Agree straightforwardly captures these apparently conflicting observations. At an empirical level, this dissertation sheds light on the aspects of the four phenomena the that have been ignored or considered problematic and set aside in many of the previous syntactic studies. At a theoretical level, the conclusion of this dissertation has implications not only on the issue of the interpretability of Agree but also on the mechanism of selection and on the direction of Agree.Ph.D.Includes bibliographical reference