Title 英 語 類 別 詞 と 認 知 Author(s) 山 口, 清 美 Citation Issue Date Text Version ETD URL http://hdl.handle.net/11094/51197 DOI Rights Osaka University

English Classifiers and Human Cognition Categorization or classification is the most primary and essential mental activity and it is based on our experiences. Classification and survival are closely related, and so we have created words and expressions to represent the consequences of our categorization. Classifiers can be central expressions to reflect our points of view on referents, and thus they denote what parts of a thing a speaker focuses on. The present study seeks the potentialities of classifiers and how closely classifiers are related to human cognition, and attempts to examine classifiers in English, which is regarded as a non-classifier language. This paper consists of 8 chapters: the first chapter is the introduction; the second observes the general characteristics of classifiers and outlines the previous studies; the third represents the role of English classifiers and three types of English classifiers; the fourth focuses on one of the three types, Cognitive Classifiers; the fifth analyzes the reinterpretation of the head of classifier constructions; the sixth examines English classifiers from a typological point of view; the seventh shows corpus analyses; the final chapter is the conclusion. English classifiers denote different aspects from NUMBER and thus they tend to provide the concept of quantity together with a speaker's intention. I claim that English classifiers have characteristics that reflect a speaker's mind-set and tend to convey informative messages to an addressee. For example, among English classifiers, extended usage beyond systematic criteria is often observed. English classifiers are, moreover, open-class forms. English nouns can be employed as classifiers, and thus they may be able to convey different aspects of referents. These aspects show the diversity of a speaker's perspective such as general points of view and temporary states of referents. English classifiers can be divided into three types: "Group Classifiers", "Cognitive Classifiers" and "Predicative Classifiers". Group Classifiers are the most conventional classifiers, which denote that referents are a group or unit. Predicative Classifiers indicate the properties of referents. Cognitive Classifiers have both characteristics of group and predicative classifiers and highlight a group or unit as well as the properties of referents. English classifiers, especially, Cognitive Classifiers tend to be productive and informative and indicate high subjectivity when a speaker categorizes things and phenomena.

Cognitive Classifiers may be able to extend their categories due to schemas abstracted from prototype and extended members. The abstracted schemas can work as necessary conditions to accept referents as category members. Acceptability can be conventionalized, but a speaker's perspective and judgment may influence the acceptability in a given situation. In addition, Cognitive Classifiers may be able to extend their categories by metaphorical mapping from the core members (source domain) to the new members (target domain). For example, "A swarm of reporters followed the princess wherever she went". Mapping between the source domain and target domain is based on similarity between the properties of bees and reporters: [AGGREGATORY] [IN MOTION] [NOISY] and [AGGRESSIVE]. The selection of core members (source domain) may entail some shades of meaning. Mapping between both domains is based on similarity, that is, few similarities between core and new members show low acceptability, while many show high acceptability. The present study calls for an explanation of re-interpretation in terms of how much information classifier constructions entail, since highly productive use of English classifiers can be due to their lexical status. Instead of the grammaticalization proposed by Brems (2003), I argue that information contents conveyed by classifier expressions motivate a speaker and hearer to interpret N2 as the head of classifier constructions and that there can be a layering of interpretation depending on the quantity of contents. In brief, the more detailed information the classifier constructions represent, the more likely N2 tends to be interpreted as the head. For a comparative study, this paper examines Japanese and Chinese classifiers. Japanese classifiers represent physical or conventional properties of referents and may be categorized in terms of animacy, humanness or shape. Japanese classifiers may also be able to extend their categories by metaphorical and metonymical mapping. However, Japanese classifiers are generally less productive and their usage is more constrained in comparison with English classifiers. Chinese classifiers are basically categorized by the shape of referents and are not sensitive to distinctions between animate and inanimate. I claim that Chinese classifiers have characteristics which are similar to English classifiers rather than Japanese ones. As a reason for this, I suggest that Chinese classifiers tend to be productive and indicate a speaker's mental attitudes as is evidenced by the fact

that in some cases common nouns can be used as classifiers. In addition, Chinese classifiers function to individuate a head noun as English indefinite articles do. This paper also focuses on the relation between articles and classifiers. Both articles and classifiers are individualizers, and some languages are strongly influenced by articles and others by classifiers. For instance, French has partitive articles, which conceptualize and individuate uncountable nouns, and French classifiers can be used for enumeration of uncountable nouns. English is probably not a typical article language, because the article system is less developed than that of French and most of the language with articles, while the classifier system is more developed, especially in that English classifiers may be able to convey informative messages to an addressee. Chinese is also a classifier language, but the classifiers are, in terms of their characteristics, located in between English and Japanese classifiers due to their metaphorical mapping and high productivity. Japanese has no articles and Japanese classifiers play both the roles of individuating and enumerating referents. The British National Corpus helps the present study by providing a lot of linguistic data from people's language used in their daily lives. In addition, this paper also focuses on the transition in the use of English classifiers and examines four corpuses; LOB corpus, British English in the 60s; BROWN corpus, American English in the 60s; FLaB corpus, British English in the 90s; FROWN corpus, American English in the 90s. The usage of English classifiers in the 60s is slightly different from those in the 90s, reflecting lifestyles in those days.