Syntactic and lexical aspects of nonconfigurationality in Nunggubuyu (Australia)

Conclusion

One way to formulate the issues is in terms of an up-or-down decision on whether such units as NP, VP, and S exist in Nunggubuyu. My view is that there is no evidence for VP, that NP is generally best considered in terms of appositional concatenation (with the possible exception of some demonstrative-noun combinations), and that S is needed but that its relationship to PNs (and INFL) is somewhat loose. However, much depends on whether the burden of proof is on those seeking to prove that units like VP exist, or on those sympathetic to the opposite position.

Whether or not these syntactic units ‘exist’ in some sense, the syntax and lexicon are obviously sharply distinct from those of strongly configurational languages like English and the other European languages which have been the basis for all prominent formal theories. If it is decided that NP, VP, and S do exist in Nunggubuyu (using various fragments of evidence, or distinct modes of argumentation), then the challenge is to proceed to account for the various facts about syntax and lexicon reported above. If the underlying structures are like English, what is it that produces such unusual features as NEG-indexing (including subjects), case-spreading (from head noun to relative clause and within relative clauses from predicates to nouns), noun-class harmony in whole/part expressions, and so forth? Why are there no adjectives or adverbs (in the English sense)? Why do quantifiers take the unusual form they do?

Hale's suggestion (1983) is that Warlpiri has a lexical Structure (LS) resembling that of configurational languages, but differs in the way this LS is projected onto phrase structure (PS). However, this model with its optional ‘nonconfigurationality’ feature does not seem capable of capturing the close relationship between lexicon and syntax in radically NC languages (where, moreover, the various available tests for a subject-VP split do not yield positive results).

Regardless of how we decide to model underlying structures in this language, NC syntax is a basic fact of observable surface structures. The various jobs that a grammar must do if speakers are to succeed in communicating must be carried out in this context. This includes the expression of complex concepts which are most easily handled by tightly knit phrasal structures, including possessor-possessum combinations, nominal conjunctions, and quantified nouns; there must also be some mechanisms for marking the scope of NEG and boundaries between main and subordinated propositions. Thus an alternative theoretical approach to NC languages is to accept the absence of tight multi-word phrasal units as a point of departure, then work (one by one) through the ways in which the language resolves the conflict between NC structure and the functional requirements just mentioned. While we have seen some ways in which the lexicon and the morphology may adjust to NC syntax, we know far too little about the cross-linguistic range of such possibilities (i.e., about the detailed typological ramifications of NC structure).