ABSTRACT
An investigation of the discourse function of non-canonical syntactic forms and their role in natural language generation necessarily overlaps and draws from a large amount of previous research. Of particular relevance here is corpus-based research on the discourse function of syntactic forms. This is the topic of Section 2.1, which provides a detailed picture of the necessary conditions for the use of non-canonical syntax posited in the literature. Section 2.2 summarizes and attempts to synthesize several theories of discourse structure in order to provide a framework for the claims made in Chapter 3 about the use of non-canonical syntax to communicate information about discourse structure relations. Section 2.3 is included to clearly delimit the scope of the research undertaken here given its overlap with issues of prosodic encoding of information structure ground and focus. In addition, it provides a framework for understanding the claims made in Chapter 3 about the role of focus disambiguation in the use of noncanonical syntax. The body of work in natural language generation devoted to selecting word order of major constituents in the clause in English, is summarized and critiqued in the final Section 2.4 in this chapter. Altogether, these four sections should provide the reader with the necessary background to understand the motivation for the problem set out in the previous chapter and the solutions to this problem that are presented in the following ones.
