On the nature of "say" complementation
This dissertation investigates the syntax and semantics of the verb ``say'' and clausal complementation involving the verb ``say''. Clausal complementation involving the verb ``say'' is among the most common strategies implemented across the world's languages and they exhibit morpho-syntactic, semantic, and pragmatic properties that differ from other types of clausal complementation. The goal of this dissertation is to offer a syntactic analysis that offers an explanation for these differences. Each language that has ``say'' complementation also has a grammatical mechanism whose responsibility is to link verbs to form complex predicates. The null hypothesis in this dissertation is that ``say'' complementation does not involve a ``complementizer'', but instead involves a clause containing the verb ``say'' that adjoins to the matrix clause. The three questions taken up are as follows: (i) What are the morpho-syntactic and semantic properties of the verb ``say''?, (ii) What are the morpho-syntactic and semantic properties of the clause-linking mechanism, and (iii) Do ``say'' complementation structures exhibit the properties of ``say'' in a serialization structure? These questions are answered based primarily on data from Uyghur, English, and Avatime.
Chapter One introduces discussion of the main puzzles, introduces background information about Uyghur and Avatime, and introduces a brief literature review that this dissertation builds on. Methodological information is also provided within the discussion of each language.
Chapter Two introduces in-depth discussion of the verb ``say'' in English. Building upon intuitions presented in Grimshaw (2015), a morpho-syntactic analysis of the verb ``say'' as the overt realization of an abstract ``Light Verb'' SAY is provided. It is shown that ``say'' alternates between being stative and dynamic, which has effects on argument structure. More specifically, only dynamic ``say'' is capable of licensing a Goal argument and an Agent, while stative ``say'' introduces only Linguistic Material (what was said) and its source. It is further shown that ``say'' is unique with respect to the range of internal arguments that it can take relative to other predicates. Based on a ``Flavours of little v'' analysis (Folli & Harley, 2005), it is argued that stative ``say'' involves a truncated structure embedded under vBE, which lacks all syntactic structure responsible for eventive/agentive semantics, while dynamic/eventive ``say'' involves a non-truncated structure. It is finally argued that certain predicates, such as ``scream'' manner adjoins to vDO, which prevents the predicate SAY from getting pronounced. The Chapter ends with discussion of the stative/eventive alternation in Avatime, which is reflected by the presence/absence of agreement morphology in the language.
Chapter Three demonstrates that Uyghur shows the same stative versus eventive alternation observed for English, but further demonstrates that Uyghur ``say'' is unique in many ways that are distinct from English. Building upon Sudo (2012) and Shklovsky & Sudo (2014), it is argued that ``say'' is uniquely able to introduce a nominalized complement clause or a tensed complement clause, the latter of which resembles a finite (root) clause. It is proposed that the seemingly finite CPs vary in size. The larger CPs host monstrous or quotative operators that trigger Indexical Shift, which enable full feature transmission from C-to-T, yielding what looks like a root clause as it relates to case and agreement. ``Say'' additionally introduces a defective (reduced) CP, which does not allow full transmission of features, forcing the embedded subject to raise for case and resulting in default agreement on the embedded verb.
Chapter Four offers an analysis of converbial constructions in Uyghur, which is the suffix found on the ``say'' element in Uyghur ``say'' complementation structures. It is shown that the converbial suffix has two adjunction sites: VP and TP, which has interpretive consequences. Novel data demonstrates that the distribution of converbial clauses in general account for the distribution of ``say'' complementation structures, followed by demonstrating that the properties of ``say'' illustrated in Chapter Three are similarly observed in ``say'' complementation structures, offering a syntactic account for observations made in Messick (2017) and explaining various unexplained issues described in Sudo (2012) and Shklovsky & Sudo (2014). The chapter concludes by offering brief discussion of the equivent structures in Avatime, demonstrating that its ``say'' complementation structures are built upon Nuclear Serial Verb Constructions in the language, which is functionally similar to converbial construcitons in Uyghur. For both languages, I conclude that ``say'' complementation structures are truly adjunction structures where ``say'' introduces a clausal complement, not classical CPs.
Chapter Five offers discussion of Case Theory on the basis of the analysis in Chapter 4. Baker & Vinokurova (2010) and Baker (2015) introduce discussion of Sakha (Turkic) which has ``say'' complementation structures that are nearly identical to Uyghur. They argue in favor of Dependent Case Theory to account for the distribution of accusative case, on the basis of accusative case showing up in environments that seem to lack a verb capable of licensing accusative case. I demonstrate that in most environments, the verb ``say'' is present and capable of licensing accusative case. I conclude that the analysis of complementation in Chapter Four resurrects the debate between Dependent Case Theory and classical theories of case assignment, but suggest that even if we adopt Dependent Case Theory, the analysis in Chapter Four improves its explanatory power.