Abstract

This paper proposes a pair of morphosyntactic number features, [Discrete] and [Non-Atomic], and shows how they can contribute to an understanding of how grammatical number is expressed cross-linguistically. Starting with English, where mass nominals pattern syncretically sometimes with plural count nominals and sometimes with singular ones, we use these features to improve upon a previous account (Cowper & Hall 2002), and then extend the analysis to mass-count syncretisms in Lingala and Manam and to classifiers in Western Armenian and Mandarin. We account for the cross-linguistic variation using a consistent set of features and a highly constrained theory of morphological exponence, and argue that the variation arises from differences in the syntactic structures in which the features appear and the paradigmatic systems of contrast in which they participate.