Projecting (Your) Voice: A Theory of Inversion and Defective Circumvention

Abstract

This doctoral dissertation revolves around colloquial agreement alternations in many constructions in several languages involving postverbal thematic subjects. These alternations are productive and well-attested, and are exemplified by the following examples from English.

(1) a. There {were/was} no seats left

b. Off the plank {walk/walks} several bad pirates

c. “Moo!” {go/goes} the cows

d. What I love most about you {are/is} your outfits

This agreement alternation is notably only possible in the inverted form. When the postverbal elements in (1) appear preverbally, the optionality disappears.

(2) a. No seats {were/*was} left

b. Several bad pirates {walk/*walks} off the plank

c. The cows {go/*goes} “Moo!”

d. Your outfits {are/*is} what I love most about you

This colloquial agreement alternation has widespread consequences for theories of Agree, inversion, argument structure, and the connections among the three. Using a maximally simple formulation of Agree that effectively reduces to Minimal Search/closest Match, I show that the aforementioned agreement alternations are derivable from a single set of syntactic operations through an optional process which I dub defective circumvention, in which a probe can conditionally Agree past a featurally deficient goal and undergo sequential multiple Agree with more than one goal, given that the features of the two goals are compatible with one another. I motivate this theory primarily with data from English and Spanish, among other languages. Defective circumvention explains the agreement optionality shown in (1) as well as PCC effects which I demonstrate the existence of in various multiple subject constructions.

Given the minimal theory of Agree presented in this thesis, along with a minimal formulation of argument structure in which thematic arguments are uniformly and hierarchically externally merged, a conundrum emerges. The sentences in (1) involve A-movement (as evidenced by the effects on agreement) of an internal argument over an argument that was externally merged in a higher position. How is it that A-movement possible without the higher argument serving as an intervener? Given the understood uniformity of the projection of arguments, as well as the locality and strict minimality of this formulation of Agree, a mechanism is needed to affect the accessibility of arguments for A-operations.

As such, I demonstrate that these constructions necessarily involve inversion via smuggling, in which the internal argument is smuggled to a position above the external argument via phrasal movement of a verbal projection containing the internal argument, which makes it a viable goal for A-movement that would otherwise be blocked by the higher argument. I show that this VP smuggling to make internal arguments accessible for A-operations is characteristic of voice constructions. I present a definition of voice alternations in which voice is characterized by mediating between the mapping of thematic arguments from the positions in which they are externally merged and the potential realization of arguments at LF. Voice, then, is what allows for inversion to take place and ultimately what allows for the agreement alternation facilitated by defective circumvention. I demonstrate this analysis of inversion via detailed derivations of various inversion constructions in English, including locative inversion, quotative inversion, predicate inversion, and existential ‘there’ sentences, which I show are derived via inversion as well.