Relationship preservation

Abstract

This thesis deals with a number of puzzles related to word order, in which the co-occurence of two elements in the same clause imposes a restriction on the distribution of these elements. I suggest that elements involved in Agree relationships [Chomsky (2000, 2001)] are subject to a requirement that they be aligned with the left or right edge of a prosodic phrase, following work done in Richards (2016). I argue that there is a restriction on the opaque satisfaction of this requirement, and show that this provides a unified solution to these word order puzzles. Chapters 1 and 2 deal with movement phenomena, primarily in left-headed languages. In chapter 1, I show that some languages allow A-movement of subjects across other DPs, whereas others do not. I note that this appears to be determined by which edge of a prosodic phrase they require phrases in Agree relationships to be aligned with, and show that this is a consequence of the proposed restriction on opaque satisfaction of the alignment requirement. Chapter 2 builds on the results of chapter 1. I show that A-movement of a subject may cross another nominal in all languages, but only if there is a phase boundary between the launch site and landing site, and show how this falls out from the proposed restriction on opacity. I show also that languages that do not allow A-movement of subjects across other DPs display a similar restriction in wh-questions, and argue that this too is a result of the proposed restriction on opacity. Chapters 3 and 4 deal with the distribution of wh-phrases in languages that allow them to remain in-situ. Chapter 3 deals with co-occurance restrictions between foci and wh-phrases. I argue that these restrictions emerge as a result of a conflict between a prosodic strategy that languages might use to satisfy the alignment requirement, called Grouping, and the proposed restriction on opaque satisfaction of this requirement. Chapter 4 deals with Grouping more generally. I show that languages with Grouping have a particular prosodic signature which marks phonological phrases that contain wh-phrases, whereas languages that lack Grouping do not, and explore the consequences of this for the architecture developed in this thesis.