Words within Words: The Internal Syntax of Verbs

This dissertation is an exploration of the syntax-PF interface, in particular how functional syntactic material is mapped into different PF units, specifically the word. The empirical domain is verb and auxiliary patterns, as this particular domain has rich variation in whether tense, mood, aspect and valency changing morphology is expressed in single or multi-word constructions. On the surface the choice for an inflectional category being expressed as auxiliary or affix seems arbitrary. In this dissertation, however, I argue that it is not as arbitrary as it may appear. I explore how the expression of features in one or more words can differ systematically cross-linguistically and propose that syntactic word-formation is sensitive to syntactic domains (phases). In order to do so, I approach this problem from various angles. Chapter 2 focuses on an in-depth investigation of two head-final languages (Japanese and Turkish) and mismatches between syntactic and phonological word-hood tests. I show that these mismatches follow from particular word building processes and their interaction with syntactic domains. Hence these mismatches are not arbitrary. Chapter 3 explores the status of word building operations, and the idea that timing differences in the derivation lead to variation in auxiliary patterns. I provide evidence that auxiliary patterns in a language pattern with other phenomena that align with the timing of word building, such as matching constraints in ellipsis. Chapter 4 focuses on correlations between domains in word building and domains in the syntax, i.e., correlations between auxiliaries and phases. Finally, chapter 5 explores which inflectional features are expressed in single or multi-word expression in a cross-linguistic sample of 33 languages from 18 families. I show that inflectional categories inside the first phase do not participate in periphrasis. I show that there is more uniformity than initial appearances suggest in the expression of features as single words or periphrastic constructions, and that there may be universal aspects to the analysis. Furthermore, the apparent mismatches between phonological and syntactic tests for word-hood are in fact regular consequences of the relevant processes and their timing pointing to the need for a more integrated approach to cross-modular research.