Abstract
This paper presents a case study in grammar engineering for linguistic hypothesis testing, focusing on the treatment of second position auxiliaries in an HPSG grammar for Wambaya. A detailed comparison of two versions of this grammar highlights the interconnectedness of linguistic phenomena, the model-dependence of linguistic analyses, and the value of computational support in calculating the consequences of differing analytical choices.
Highlights
Grammars of natural languages are highly complex objects, in which analyses of many distinct phenomena must interact in order to correctly map even the simplest sentence to its semantic representation (Bender, 2008b)
This paper presents a case study in grammar engineering for linguistic hypothesis testing, focusing on the treatment of second position auxiliaries in an HPSG grammar for Wambaya
Our models of grammars of languages are themselves complex objects, in which analytical choices are made in light of previous analytical choices, but usually could be revised provided we are willing to do the work of following the cascade of required changes to other analyses
Summary
Grammars of natural languages are highly complex objects, in which analyses of many distinct phenomena must interact in order to correctly map even the simplest sentence to its semantic representation (Bender, 2008b). Wambaya [wmb] is a non-Pama-Nyungan language of the Mirndi family (Green and Nordlinger, 2004) from the West Barkly Tablelands region of the Northern Territory of Australia.1 It was originally described and documented by Rachel Nordlinger (1998). The formal analyses described here were developed on the basis of the descriptive analyses in Nordlinger 1998 and the annotated (IGT format) examples from that work. These examples (numbering 804) became the development test suite for the implemented grammar. The grammar has relatively low ambiguity, assigning on average 12.56 parses to each item in the test set
Sign-in/Register to view highlights & summary 
Published Version (
Free)
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call