This dissertation proposes a theory of categorial grammar called
Multi-Modal Categorial Grammar with Structured Phonology. The central
feature that distinguishes this theory from the majority of
contemporary syntactic theories is that it decouples (without
completely segregating) two aspects of syntax---hierarchical
organization (reflecting semantic combinatorics) and surface
morpho-syntactic realization---which are conflated in the single
notion of phrase structure in theories of syntax that take this notion
as a primitive.
In Chapter 2, I review three phenomena in Japanese---nonconstituent
coordination, nonconstituent clefting and four kinds of
complementation constructions---that present serious empirical
challenges to previous models of generative syntax, whether
transformational or nontransformational, and argue that all existing
analyses of these constructions in these theories are inadequate. The
problems of these previous approaches all stem from the fact that the
empirical phenomena considered here constitute cases that deviate from
what one might call the 'canonical' mode of phrasal composition,
implicitly built into the notion of phrase structure, wherein the
surface syntactic constituency transparently reflects the
semantically-oriented combinatoric structure. In theories that take
the notion of phrase structure as a primitive, such deviations can
only be accommodated by means of some kind of extensions to the basic
phrase structural component. However, such extensions are often
ad hoc and fail to capture the systematic patterns that the empirical
phenomena in question exhibit, especially when they interact with one
another and with other aspects of the grammar of the language.
This motivates us to abandon the phrase structure-based perspective
and instead adopt an architecture in which the combinatoric component
and the surface morpho-syntactic component are separated, yet interact
closely with one another. Chapter 3 presents the theory of
Multi-Modal Categorial Grammar with Structured Phonology, a formal
theory of syntax that embodies this architecture. The proposed theory
emerges as a result of unifying and integrating two most notable
features of categorial grammar as a linguistic theory, namely,
flexibility of constituency and the separation of semantic
combinatorics from surface morpho-syntax. The resulting theory
resembles and borrows many ideas from related recent variants of
categorial grammar, all of which attempt to achieve a similar
synthesis in one way or another, but it achieves this goal in the
conceptually simplest and technically most explicit way. Most
importantly, recognizing an interface component between syntax and
phonology called 'structured phonology' and working out its formal
details precisely is the major contribution of the present work.
Chapter 4 demonstrates that the present theory enables straightforward
analyses of the phenomena from Chapter 2. Specifically, different
degrees of flexibility in constituency and linear order found in
different phenomena receive a natural account, with the component of
structured phonology governing the properties of different modes of
morpho-syntactic composition. Moreover, not only does the present
approach account for the individual phenomena adequately, the analyses
of these phenomena interact with one another (and with other aspects
of Japanese syntax) to automatically make correct predictions. This
result provides strong confirmation for the present approach.