What are the limits of grammar? Investigation of phi-feature resolution under coordination.

Please jlg363 [at] stanford.edu (email )for zoom link.

Abstract:

It is commonly assumed that the person/number/gender (phi) features of a coordinate structure are computed grammar-internally from the phi-features of its conjuncts:

(1) Tom_3sg and Jerry_3sg are_pl/*is_3sg/*am_1sg bouncing.

In this talk, I argue that the mechanism behind phi-resolution in coordination is not grammar-internal. Even though the elements involved in resolved agreement with coordinate structure, i.e., phi-features, coordination and agreement, are indeed grammar-internal, the resolution mechanism itself is not.

The empirical evidence I focus on comes from inter- and intra-speaker variability in the surface output of resolution:

(2) There is/are a cat and a mouse in the yard.

First, I zoom in on agreement with coordination in Polish and one existing analysis that is the most-successful in terms of empirical coverage. I show the consequences that such an analysis has on the theory of grammar in general. If our goal is to make progress on these issues, introducing certain tools into a grammatical toolkit is not a practical solution. Then, I present a broader landscape where a similar variability in the empirical data and resistance to a simple grammatical account is attested in many world-languages. While each language requires its own in-depth consideration, a recurring theme emerges where resolution of phi-features under coordination does not behave like other well-known grammar-internal phenomena.

In the final part of the talk, I sketch possible reasons for the grammar-external nature of resolution. There are phenomena not related to coordination where a similar crunching of an overabundance of phi-featural information takes place: omnivorous agreement, Person Case Constraint effects, Specificational Copular Clauses. Nevertheless, the output of this crunch is different that the output of resolution under coordination. Since coordination is the only element distinguishing the two types of crunch, I conclude that we should be looking into coordination for the reasons of grammar-external nature of resolution. I revisit the classic “quirks” of coordination and consider which of these might be the culprit.