Semantic representations of syntactically marked discourse status in crosslinguistic perspective

Abstract

This paper presents suggested semantic representations for different types of referring expressions in the format of Minimal Recursion Semantics and sketches syntactic analyses which can create them compositionally. We explore cross-linguistic harmonization of these representations, to promote interoperability and reusability of linguistic analyses. We follow Borthen and Haugereid (2005) in positing COG-ST ('cognitive status') as a feature on the syntax-semantics interface to handle phenomena associated with definiteness. Our proposal helps to unify the treatments of definiteness markers, demonstratives, overt pronouns and null anaphora across languages. In languages with articles, they contribute an existential quantifier and the appropriate value for COG-ST. In other languages, the COG-ST value is determined by an affix. The contribution of demonstrative determiners is decomposed into a COG-ST value, a quantifier, and proximity information, each of which can be contributed by a different kind of grammatical construction in a given language. Along with COG-ST, we posit a feature that distinguishes between pronouns (and null anaphora) that are sensitive to the identity of the referent of their antecedent and those that are sensitive to its type.

Focus case outside of Austronesian: An analysis of Kolyma Yukaghir

“…The grammar fragment we created with the customization system does not handle the definiteness distinction discussed immediately above because the Grammar Matrix customization system does not yet provide a library for definiteness marking. However, the Grammar Matrix's core grammar does provide support for adding such an analysis to the grammar by hand, based on Borthen & Haugereid's (2005) analysis of cognitive status of references (Bender & Goss-Grubbs, 2008).…”

Grammar Customization

“…The customization system allows us to define affixes in order to get the form of these words right, but does not yet have any provision to encode the semantic contributions of the affixes. In fact, we believe that definiteness (and more generally, information about the discourse status of referents) is best encoded as semantic feature, whether it is introduced by a separate word (determiner) or an affix (Borthen and Haugereid 2005;Bender and Goss-Grubbs 2008), and intend to add support for such marking to the system in the future. The tense/aspect library performed fine, with the exception of Hausa, where tense and aspect are closely connected to mood.…”