Comparing Classifier use in Chinese and Japanese

Copyright 2012 by Yue Hui Ting and Francis Bond 26th Pacific Asia Conference on Language,Information and Computation pages 264–271 Comparing Classifier use in Chinese and Japanese Yue Hui Ting and Francis Bond Division of Linguistics and Multilingual Studies Nanyang Technological University {htyue1@e.ntu.edu.sg, bond@ieee.org} Abstract Numeral classifiers present a challenge to suc- cessful machine translation. We investigate two numeral classifier languages: Mandarin Chinese and Japanese. This paper presents a quantitative analysis of classifier translations between these two languages to better understand differences in classifier usage. Keywords – numeral classifier, sortal, translation, Mandarin Chinese, Japanese, con- trastive linguistics 1 Introduction Mandarin Chinese (CMN) and Japanese (JPN) are numeral classifier languages. Numeral classifier languages express the quantity of referents by modifying a noun phrase (NP) with an obligatory numeral-classifier construction where the classifier denotes inherent referent attributes (Bond and Paik, 2000; Downing, 1996). Hence, for a numeral- classifier construction that is assigned to a noun, the numeral denotes the numerical quantity of the noun referent while the numeral classifier denotes the quality of the noun referent. Bond and Paik (2000) identified five main types of classifiers. Event classifiers classify events (Japanese: -kai 回 ‘time’; Mandarin Chi- nese: -cì 次‘time’). Mensural classifiers are em- ployed for the measurement of physical properties (Japanese: -sun 寸 ‘inches’; Mandarin Chinese: - cùn 寸 ‘inches’). Group classifiers classify group- ings of referents (Japanese: -kumi 組 ‘pair, set’; Mandarin Chinese: -shuāng 双 ‘pair’). Taxonomic classifiers effect a generic interpretation of the noun phrase (Japanese: -shu 種 ‘kind, type’; Man- darin Chinese: -zhǒng 种 ‘kind, type’). Finally, when quantifying the noun, Sortal classifiers clas- sify the type of referent that is being counted, as in (1) and (2)*. (1) JPN: pen 2-hon pen 2-CL (long, cylindrical) “2 pens” (2) CMN: 6- zhāng piào 6-CL(flat, broad) tickets “6 tickets” The numeral classifier system is organized dif- ferently for different languages. Mok et al.’s (2012) parallel studies focusing on generating sort- al classifiers found that there are differences in classifier usage for the same semantic hierarchy of noun classes, suggesting differing conceptual or- ganization between Mandarin Chinese and Japa- nese. Figure 1. Semantic Hierarchies in Chinese and Japanese For example, the semantic hierarchies in Fig.1 show that there is not always one-to-one corre- spondence between the classifier characters of * Abbr. used are CL for classifier, PTCL for particle, DET for determiner, num for numeral or numerative Animal (匹) Mammal Bird (羽) Reptile Rabbit (羽) Dog Horse (頭) Snake Fish (匹/尾) Lizard Japanese Animal (只) Mammal Bird Reptile Rabbit Dog Horse (匹) Snake (条) Fish (条) Lizard (条) Chinese 264 these two languages. For example, although the same character 匹 exists in both classifier systems, it is used differently, as a general animal classifier in Japanese and as a specific classifier for horses in Mandarin Chinese. Japanese and Mandarin Chi- nese are an interesting pair of languages to com- pare for classifiers because they share to a limited extent the same Chinese character system and oc- casionally there are one-to-one correspondences for …

Content truncated to 500 words