This thesis is a study of the featural representation of nominal number. By looking at patterns of agreement when 1st and 2nd person arguments interact, I argue that the expression of morpholog- ical number does not involve dedicated plural features (e.g., [+plural]). Instead, morphological number is dependent on multiple, privative, atomic features ([INDIVIDUAL]), essentially extend- ing into morphology a standard semantic theory of nominal plurality (Link 1983, Schwarzschild 1992). A feature bundle containing more than one [INDIVIDUAL] feature is mapped to a plural exponent. The empirical discussion centers on agreement in LOCAL contexts, where a 1st or 2nd person acts on another 1st or 2nd person. In many languages, two singular arguments may result in plural morphology on the verb. I show that the plural morphology results from a single syntactic head copying phi-features from two different argument positions, "building" a plural feature bun- dle. I extend the discussion beyond local agreement contexts, and show how Individual Number can account for a large range of data. The proposed theory attempts to unify morphological and semantic representations within a Minimalist framework (Chomsky 1995, 2000, 2001)

This dissertation examines how intensional content, i.e., belief ascription, constrains antecedent- gap chains. I defend the proposal that antecedent-gap chains are intensionally uniform: the an- tecedent and the gap must refer to the same thing. The core focus is defective intervention in antecedent-gap chains(Chomsky, 2000, 2001). Previous accounts have attributed defective inter- vention to syntactic mechanisms (Chomsky, 2001; Nevins, 2004; Preminger, 2014). These ac- counts are shown to be at best entirely stipulative and at worst empirically inadequate.

I make two new generalizations concerning defective intervention. The first is that defective interveners are all attitude holders. I support this generalization by closely examining the class of tough-predicates, which permit various kinds of arguments to be projected in the syntax between the antecedent and the gap. The second generalization is that defective intervention only arises when the antecedent-gap chain connects two thematic positions. Again, I justify this generalization by looking closely at the tough-construction, which has itself inspired decades of research. I illustrate that the antecedent-gap chain in the tough-construction is “mitigated” by beliefs, deriving where the tough-construction appears, and why it appears.

The two core theoretical results of this dissertation are, I) a principled and an explanatory ac- count of defective intervention, II) a principled and explanatory account of the tough-construction.