Introduction

It is a fundamental human cognitive activity to categorize objects and events in order to have a cohesive understanding of the world. While one can claim that mapping between conceptual categories and grammatical forms is arbitrary, it is undeniable that conceptual categories are revealed in different noun or nominal classifying systems in the languages of the world. Among these systems are noun classes or genders, numeral classifiers, verbal classifiers, locative and deictic classifiers, and genitive or possessive classifiers (for more extensive coverage of classifiers, see Aikhenvald, 2000).

Numeral classifiers, which are widespread across the languages of East and Southeast Asia and Oceania, are a grammatical system that reflects how speakers categorize objects that they count or quantify. They classify nouns on the basis of their referents' attributes, such as animate versus inanimate, human versus animal, and shape versus function of referents. Therefore, patterns of acquisition of classifiers are an important source of information about underlying patterns of semantic and conceptual development (Adams & Conklin, 1973; Clark, 1977; Muraishi, 1983; Craig, 1986; Matsumoto, 1987; Carpenter, 1987, 1991). Earlier, researchers like Adams and Conklin (1973) proposed that animacy and shape were universal features of numeral classifiers, stressing the possibility that classifier systems reflect basic cognitive categories despite their cultural or language-family specificity. More recently, within the context of the ongoing debate on linguistic relativity, Lucy (1992a, 1992b) and Lucy and Gaskins (2001) have asserted that the acquisition of numeral classifiers might influence conceptual development itself, because classifier languages have no syntactic distinction between mass and count nouns.