Comparing possessive classifier systems across Oceanic languages

We have conducted a free-listing experiment, across six Oceanic languages spoken in Vanuatu and New Caledonia, to investigate possessive classifiers systems. Each language in our sample has a different inventory size of possessive classifiers – Merei (2), Lewo (3), Vatlongos (4), North Ambrym (5), Nêlêmwa (20) and Iaai (23). This psycholinguistic technique is simple to conduct, and yields useful data for describing and comparing the different systems: (i) the flexibility of noun-classifier assignment, (ii) the current semantic domains of each classifier, (iii) the relative membership size of each classifier, and (iv) the level of grammaticalisation of the classifiers.

Oceanic possessive classifiers are used when the possessed noun refers to an alienable possession (Lichtenberk 1983). An important property of these systems is that a possessed noun can occur with different classifiers depending upon the relation between the referents of the possessed noun and possessor. For example, in Iaai (New Caledonia) which displays a typical system, köiö ‘water’ occurs with either the drink (1a) or the general classifier (1b). In marked contrast, in North Ambrym’s (Vanuatu) innovative system, we ‘water’ occurs only with the drink classifier (2a), not the general classifier (2b).

1a.	bele-n	köiö
	CL.DRINK-3sg	water
	‘her water’(for any purpose)

1b.	anyi-n	köiö
	CL.GENERAL-3sg	water
	‘her water (e.g. for washing with)’

Our results reveal interesting variation in these classifier systems. A cline can be established with North Ambrym having the fewest nouns that occur in multiple classifier lists, to Iaai with the largest number of overlapping nouns. This shows that North Ambrym has a relatively fixed noun-classifier assignment system as opposed to Iaai with a relatively free noun-classifier assignment.

Grammatical descriptions typically give only a few nouns to exemplify a classifier’s semantic domain. The free-listing technique offers a comprehensive overview of the semantic domains; it is also able to highlight exemplar nouns and semantic sub-domains using analyses such the Cognitive Salience Index (Sutrop 2001).

By comparing numbers of unique nouns across participant responses, relative membership sizes of classifiers can be computed. This leads to new typologies: we distinguish major classifiers, with relatively large, open membership, and minor classifiers with relatively small, closed membership.

The free-listing technique further reveals the emergence of new classifiers in some languages in our sample. We highlight how the experiment reveals inter-speaker variation in the use of inalienable nouns with some speakers treating them just as possessed nouns (3a) and others as classifiers (3b). This reveals a grammaticalisation cline, with nouns taken on properties of classifiers.

References

Lichtenberk, Frantisek. 1983. Relational Classifiers. Lingua, 60(2-3):147–176.
Sutrop, Urmas. 2001. List Task and a Cognitive Salience Index. Field Methods, 13(3): 263-376.