The Anaphor Agreement Effect (Rizzi 1990, Woolford 1999, a.o.) refers to the crosslinguistic avoidance of anaphors in agreeing positions. Looking only at languages like English, one might get the impression that the AAE is about anaphors resisting the subject position, and therefore, that it is reducible to Binding Theory. However, looking at languages where subjecthood and agreement dissociate (ergative languages, quirky-subject languages), as well as languages with no overt agreement at all (e.g. Japanese), it becomes clear that subjecthood is a red herring, and the phenomenon is indeed about anaphors not tolerating agreement.

At first glance, this seems to suggest a rather tight interaction between agreement and anaphora. This has led some (e.g. Reuland 2011) to take the AAE as evidence in support of a reduction of binding to agreement.

I will show that, upon closer inspection, and despite the aforementioned claims, the AAE provides fairly strong evidence against a reduction of binding to agreement. That’s because the reductionist view requires, somewhat paradoxically, assumptions about agreement, and about the structure of anaphoric expressions, that break their compatibility with the very mechanism of anaphora being proposed.

I will instead propose that the AAE arises due to what I call “encapsulation”: the binding index always resides on a separate, higher projection than where valued phi-features reside. In addition to accounting for the AAE, we will see that the encapsulation hypothesis enjoys broad crosslinguistic morphological support (Middleton 2018; survey of 86 languages, from 13 language families).

Finally, I will offer a proposal on the nature of phi-feature matching between anaphors and their binders, in light of the fact that agreement in phi-features cannot be what relates the two. I will present evidence that, despite persistent intuitions to the contrary, it is semantics that regulates such matching, even in cases of pluralia tantum or grammatical gender on inanimates.