Typology of partitives

Abstract

This paper explores the coding patterns of partitives and their functional extensions, based on a convenience sample of 138 languages from 46 families from all macroareas. Partitives are defined as constructions that may express the proportional relation of a subset to a superset (the true-partitive relation). First, it is demonstrated that, crosslinguistically, partitive constructions vary as to their syntactic properties and morphological marking. Syntactically, there is a cline from loose - possibly less grammaticalized - structures to partitives with rigid head-dependent relations and, finally, to morphologically integrated one-word partitives. Furthermore, partitives may be encoded NP-internally (mostly via an adposition) or pronominally. Morphologically, partitives primarily involve markers syncretic with separative, locative or possessive meanings. Finally, a number of languages employ no partitive marker at all. Secondly, these different strategies are not evenly distributed in the globe, with, for example, Eurasia being biased for the separative strategy. Thirdly, on the functional side, partitives may have functions in the following domains in addition to the true-partitive relation: plain quantification (pseudo-partitives), hypothetical events, predicate negation and aspectuality. I claim that the ability to encode plain quantification is the prerequisite for the other domains. Finally, it is argued that there is a universal preference towards syncretism of two semantically distinct concepts: the proportional, true-partitive relation (some of the books) and plain quantification (some books).