國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
‧
國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
7
1.3 Organization of the Thesis
After the motivation and purpose of the present study are introduced, Chapter II presents previous studies on classifiers. Chapter III provides the cognitive
mechanisms of this research, including the concept of metonymy, the notion of metaphor, the interaction between metaphor and metonymy, idiomaticity from the cognitive perspective, and cultural constraints. Chapter IV presents data analysis based on the operation of cognitive mechanisms and cultural constraints. Chapter V concludes the thesis and points out future research.
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
‧
國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
8
CHAPTER II
PREVIOUS STUDIES ON CLASSIFIERS
Classifiers, an essential linguistic element in Chinese languages, represent overt human categorization. Some previous studies concerning classifiers will be reviewed in this chapter. First, in section 2.1 the studies on typological classifiers will be introduced (Allan 1977, Aikhenvald 2003), which define various typologies of classifiers and indicate that numeral classifiers are within the scope of the current study. Next, since classifiers are regarded as the conceptual classification of the world, in section 2.2 previous studies on classifiers in the Chinese languages will be reviewed. Studies concerning classification include Taiwanese Mandarin classifiers (Tien et al. 2002, Tai and Wang 1990, Hsu 2009, Huang and Ahrens 2003, Her and Hsieh 2010), Taiwanese Southern Min classifiers (Tai et al. 1997, Li 1998), and Taiwanese Hakka classifiers (Luo 1988, Wu 2001, Tai et al. 2001, Chen 2003, Qiu 2007, Wu 2010). Then, in section 2.3 some studies in which classifiers are
investigated from metaphorical and metonymic perspectives will be addressed, including the investigation of the metaphorical usages in Chinese classifiers in the
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
‧
國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
9
poetry of the Tang and Sung Dynasties (Ma and Zhang 2001), the investigation of the metaphorical and metonymic usages of Taiwanese Mandarin classifiers in modern poems (Chen 2009) and the investigation of the metaphorical and
metonymic usages in Taiwanese Hakka (Lee 2005, Wu 2010). Finally, in section 2.4 some remarks will be presented.
2.1 Defining Classifiers
There have been many studies on typological classifiers. Investigating more than fifty classifier languages, Allan (1977: 285) categorizes them into four types:
numeral classifier languages, concordial classifier languages, predicate classifier languages, and intra-locative classifier languages. The first type is numeral classifier languages, in which a classifier is obligatory in many expressions of quantity (e.g., Thai and Chinese). The second type is concordial classifier languages, where classifying formatives are affixed to nouns, plus their modifiers, predicates, and proforms (e.g., Bantu and Swahili). The third type is predicate classifier languages, in which verbs of motion/location consist of a theme and a stem that vary according to certain discernible characteristics of the “objects or objects conceived as
participating in an event whether as actor or goal” (Hoijer 1945: 13) (e.g., Navajo verbs). Examples of the three types taken from Allan (1977: 286) are shown in what
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
‧
國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
follows:
(1) a. Thai: khru. lɑ̂.j khon [teacher three person] ‘three teachers’
b. Bantu: ba-sika ba-ntu bo-bile [ba+have+arrived ba+man ba+two] ‘Two men have arrived.’ Here ba- is the plural human classifier.
c. Navajo: béésò sì-ʔɑ̨́ [money perfect-lie of round entity] ‘A coin is lying there.’
béésò sì-nìl [money perfect-lie of collection] ‘Some money is lying
there.’béésò sì-l̴tsòòz [money perfect-lie of flat flexible entity] ‘A note is
lying there.’The fourth type is intra-locative classifier languages, where noun classifiers are embedded in some of the locative expressions which obligatorily accompany nouns in most environments. For example, Toba has a set of locative noun-prefixes for objects, i.e., coming into view, going out of view, out of view, and in view. For objects in view, there are three prefixes which classify the accompanying nouns based on the arrangement and/or shape of their referents, that is, vertical extended object in view, horizontal extended object in view, and saliently three-dimensional object in view. According to Allan, there are few languages of this type. The other two languages are Eskimo and Dyirbal.
In addition to Allan’s (1977) study, Aikhenvald’s (2003) study also presents a full typology of classifiers. As the paradigm type, numeral classifiers, including for example those in Chinese, are within the scope of the current study. The properties
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
‧
國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
11
of numeral classifiers taken from Aikhenvald (2003: 98) are given in (2) below:
(2) a. The choice of a numeral classifier is predominantly semantic.
b. Numeral classifier systems differ in the extent to which they are grammaticalized. Numeral classifiers can be an open lexical class.
c. In some numeral classifier languages not every noun can be associated with a numeral classifier. Some nouns can take no classifier at all; other nouns may have alternative choices of classifier, depending on which property of the noun is in focus.
Regarding the property of (2a), Allan (1977: 285) also indicates that classifiers denote some salient perceived or imputed characteristics of the entity to which an associated noun refers. With respect to the property of (2b), Adams (1989) explains that the way numeral classifiers are used often varies from speaker to speaker, depending on their social status and competence. To illustrate the property of (2c), Aikhenvald (2003) exemplifies classifiers in Mandarin Chinese. A distinction is made between specific classifiers and the general classifier ge. Specific classifiers typically mark the first mention of a new item; once reference is established, subsequent mentions take the general classifier or constructions where no classifier is required (Erbaugh 1986: 408).
2.2 Classifiers as the Conceptual Classification of the World
Since the classifier system reflects human conceptual structure and
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
‧
國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
categorization, studies on classifiers in the Chinese languages are largely concerned with classifying them into different types. According to He (2008), the numeral classifiers in Chinese are classified into four main categories: mingliangci (名量詞)
‘noun measures’,4 dongliangci (動量詞) ‘verb measures’, jianzhi liangci (兼職量詞)
‘twofold measures’, and fuhe liangci (複合量詞) ‘compound measures’. Noun measures include geti liangci (個體量詞) ‘individual measures’,5 jihe liangci (集合 量詞) ‘group measures’, bufen liangci (部分量詞) ‘partitive measures’, zhuazhi
liangci (專職量詞) ‘specific measures’, jieyong liangci (借用量詞) ‘borrowed
measures’, linshi liangci (臨時量詞) ‘temporary measures’, and duliangheng
liangci (度量衡量詞) ‘standard measures’. Verb measures, co-occurring with
actions, are used for counting the times of the action. Twofold measures refer to measures that can be regarded either as noun measures or as verb measures.
Compound measures are composed of a noun measure and a verb measure, in which the former is used for the entity while the latter measure is used for the action.
For noun measures specifically, many studies focus on classifying nouns into various categories of classifiers according to the prototypical semantic features shared by classifiers and their corresponding nouns. For instance, nouns which are long and flexible are classified into the category of the classifier tiao (條) ‘stripe’,
4 Noun measures include both classifiers and measure words in this thesis.
5 Individual measures refer to classifiers in the following in this thesis.
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
‧
國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
13
and nouns which are round and spherical are classified into the category of the classifier ke (顆). Tien et al. (2002), conducting psycholinguistic experiments focusing on Chinese classifiers, indicate that each classifier demands specific semantic features on their corresponding nouns. When the prototypical features of nouns are manipulated, the selection of a given classifier will be shifted. This study hence evidences that the selection of classifiers is not arbitrary but determined by shared prototypical semantic features with their corresponding nouns.
Different from the study of Tien et al. (2002), Tai and Wang (1990) examine nouns within the category of the classifier tiao (條) ‘stripe’ according to family resemblance (Wittgenstein 1953) and divide them into three groups by semantic extensions: central members, natural extension, and metaphorical extension. The central members (e.g., fish and pants) denote three-dimensional concrete objects with a long shape, and the natural extension (e.g., rivers and roads) refers to entities which possess a visible long shape but which have only two-dimensions. While central members and natural extension are used to classify concrete visible entities with a long shape, the metaphorical extension (e.g., news and laws) is grounded on the imagined long shape of an entity via the creative mind of human beings. Tai and Wang (1990) thus conclude that classifiers are not an arbitrary linguistic device of categorization but reflect human categorization based on both the salient perceptual
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
‧
國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
properties of entities and human’s imagination.
Concerning the relationship of classifiers and their corresponding nouns, Hsu (2009) proposes an interactive model of classifiers and nouns grounded on the prototype effects (Rosch and Mervis 1975), experiential view of categorization (Johnson 1987, Lakoff 1987), and the intercategorical continuity (Kleiber 1900).
With this model, Hsu (2009: 43) argues that the interactions of classifiers and their nominal referents can affect the members of nominal referents. For example, the classifier duo (朵) ‘prosperity’ prototypically combines with flowers. While when the feature charming originally belonging to the noun flowers feedbacks to duo (朵), it can be used to combine with beautiful and charming things, like smiles as in yi
duo weixiao (微笑) ‘a smile’. The relationship of the classifier and the noun comes
from the temporary metaphor, but the motivation of this metaphor is the interaction of features. If this metaphor is conventionalized, the feature charming would become one member of the features clustering in duo (朵). Therefore, anything that is beautiful and charming can be linked to duo (朵) metaphorically, such as wanxia (晚霞) ‘sunset’ and zitai (姿態) ‘posture’.
The relationship of classifiers and their corresponding nouns is also investigated in Huang and Ahrens’ (2003) study, in which classifier coercion of nouns is
‧
proposed. Employing Pustejovsky’s concept of qualia structure,6 they propose a tripartite classifier system in Taiwanese Mandarin, i.e., individual, kind, and event.
Following Pustejovsky’s theory, Huang and Ahrens (2003) assume that individual classifiers can coerce nominal semantic types, and that semantic coercion can be predicted through a well encoded qualia structure. For example, with the noun
dianhua (電話) ‘telephone’, the individual classifier ju (組) ‘set’ selects the Formal
role of telephone, i.e., the telephone machine itself while the individual classifier
xian (線) ‘line’ selects the Telic role of telephone, i.e., the line for the phone.
Furthermore, Huang and Ahrens (2003) propose that classifiers can type-shift nouns to a kind reading and an event reading. Unlike the highly idiosyncratic selection of the individual classifiers, the kind classifiers select a broad class of nouns. For instance, the kind classifier yang (樣) selects the kinds defined by shape and appearance, as in san yang shuiguo (三樣水果) ‘three kinds of fruit’. In the same vein, the event classifiers coerce an event type reading from nominals. The semantic nature of events is that they are temporally anchored. Thus, temporal reference is an integral part of the semantics of events. For example, the event classifier chang (場) refers to scheduled and regularly occurring events, as in zhe chang dianying (這場電
6 Qualia structure is a set of semantic constraints by which we understand a word when embedded within the language. These constraints contain four basic roles: Constitutive, Formal, Telic, and Agentive. Constitutive constraints involve the relation between an object and its constituents, or proper parts. Formal constraints distinguish the object within a larger domain. Telic constraints refer to purpose and function of the object. Agentive constraints refer to factors involved in the origin or bringing about of an object (Pustejovsky 1995: 85ff).
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
‧
國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
影) ‘this movie’. In sum, a noun can result in various interpretations due to the
complex semantic content. Also, it can occur with different specific classifiers.
However, as Huang and Ahrens (2003) put it, it is the classifier that selects the relevant properties of the noun and coerces the appropriate meaning. That is, the selected classifier coerces a particular meaning of its concurrent noun.
In addition to relationship of classifiers and nouns, the discrimination between classifiers and measure words is also a controversial issue. Tai and Wang (1990: 38) define classifiers and measure words in what follows:
A classifier categorizes a class of nouns by picking out some salient perceptual properties, either physically or functionally based, which are permanently associated with entities named by the class of nouns; a measure word does not categorize but denotes the quantity of the entity named by noun.
In a different study, Lyons (1977: 463) indicates that a classifier is “the one which individuates whatever it refers to in terms of the kind of entity that it is while a measure word is the one which individuates in terms of quantity”. Aikhenvald (2003: 115) also states that “while sortal classifiers categorize nouns in terms of their inherent properties such as animacy, shape, and consistency, measure words are used for measuring units of countable and mass nouns”. Employing the Aristotelian distinction between essential and accidental properties and the Kantian distinction
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
‧
國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
17
between analytic and synthetic propositions, Her and Hsieh (2010: 15), furthermore, characterize the distinction between classifiers and measure words as follows:
A classifier indicates an essential property of the noun, and can be paraphrased as the predicate concept in an analytic proposition with the noun as the subject concept; a measure word indicates an accidental property in terms of quantity, and can be restated as the predicate concept in a synthetic proposition with the noun as the subject concept.
In addition to semantic discrimination given above, Her and Hsieh (2010) propose two sets of refined and reliable tests to substantiate the distinction between classifiers and measure words in Taiwanese Mandarin, i.e., numeral/adjectival modification and de(的)-insertion. In terms of the test of numeral/adjectival
modification, measure words block numeral and adjectival modification to the noun, while classifiers do not. Therefore, numeral stacking is possible only for measure words as xiang (箱) ‘box’ in yi xiang shi ke pingguo (一箱十顆蘋果) ‘one box of ten apples’ but it is not acceptable for classifiers as ge (個) in *yi ge shi ke pingguo (*一個十顆蘋果). Furthermore, stacking antonymous adjectives is perfectly fine for measure words, such as xiang (箱) in yi da xiang xiao pingguo (一大箱小蘋果) ‘one big box of small apples’ since the measure word xiang (箱) ‘box’ blocks the
adjective da (大) ‘big’ from modifying the noun pingguo (蘋果) ‘apple’. In contrast, stacking antonymous adjectives is impossible for classifiers, such as ke (顆) in *yi
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
‧
國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
da ke xiao pingguo (*一大顆小蘋果) in that the classifier ke (顆) does not block the
adjective da (大) ‘big’ from modifying the noun pingguo (蘋果) ‘apple’. Therefore, the noun pingguo (蘋果) ‘apple’ is simultaneously modified by the two antonymous adjectives big and small, contributing to contradictory meaning. With respect to the test of de(的)-insertion, increased computational complexity in the classifier phrase increases the acceptability of de (的) intervention. Given one is the least complex number, the de(的)-insertion is acceptable for measure words such as xiang (箱) in
yi xiang de shu (一箱的書) ‘one box of books’ whereas it is not acceptable for
classifiers, such as ke (顆) in *yi ke de pingguo (*一顆的蘋果). Since the tests are more reliable and accurate, the distinction between classifiers and measure words of the data in the current study will be based on them.
Similarly, various studies on classifiers in other Chinese languages, including Taiwanese Southern Min and Taiwanese Hakka, are largely concerned with classifying them into different types. Tai et al. (1997) adopt the experimental
paradigms of ranking and listing developed by Rosch (1975) to collect production of classifiers in Taiwanese Southern Min from illiterate, monolingual native speakers of Taiwanese Southern Min. Based on the data collected, they construct categorical structures for classifiers in Taiwanese Southern Min and propose that the distinction between classifiers and measure words in Taiwanese Southern Min is a continuum
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
‧
國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
19
from prototypical classifiers to prototypical measure words, including individual measures, partitive measures, group measures, container measures, and standard measures. Analyzing the classifier system in Taiwanese Southern Min, Li (1998) claims that the more prototypical classifiers are, the more they function as a
categorization tool, and that the more prototypical measure words are, the more they function as a measurement.
With respect to the classifiers in Taiwanese Hakka, Luo (1988) divides them into two main categories: noun measures and verb measures, in which the former are used to measure nouns and the latter are used to modify verbs. Noun measures contain duliangheng minliangci (度量衡名量詞) ‘standard measures’(e.g., gin1 (斤)
‘kilogram’), danweici (單位詞) ‘individual measures’ (e.g., tiau5 (條) ‘stripe’), jiti
minliangci (集體名量詞) ‘group measures’ (e.g., doi2 (堆) ‘heap’), jieyongmingci
minliangci (借用名詞名量詞) ‘measures borrowed from nouns’ (e.g., ang1 (盎)
‘jar’), and jieyongdongci minliangci (借用動詞名量詞) ‘measures borrowed from verbs’ (e.g., quan7 (擐)); verb measures include zhuanyong dongliangci (專用動量 詞) ‘specific measures’ (e.g., bai1 (擺) ) and jieyong dongliangci (借用動量詞)
‘borrowed measures’ (e.g., ba1 zhong2 (巴掌) ‘palm—slap’).
Focusing on noun measures specifically, Wu (2001), Tai et al. (2001), and Chen (2003), following the study of Tai et al. (1997), elicit data on the production of
‧
Taiwanese Hakka classifiers from illiterate, monolinguial native speakers of Taiwanese Hakka. They limit their collection of Taiwanese Hakka classifiers to individual measures, partitive measures, and group measures, which serve the function of understanding human categorization. As the classifier systems in Taiwanese Southern Min, Tai et al. (2001: 3) propose that the distinction between classifiers and measure words in Taiwanese Hakka is a continuum from prototypical classifiers to prototypical measure words, as illustrated in Table 1.
Table 1. A continuum from prototypical classifiers to prototypical measure words (Tai et al. 2001: 3)
Individual Measures
tiau5 (條) gin1 (根), gi1 (枝)
Partitive Measureskuai5 (塊) pien2 (片), si1 (絲)
Group Measuresdoi1 (堆), kiun5 (群), tiab8 (疊)
Container Measuresvon2 (碗), bui1(杯), ang1 (盎)
7 Standard Measureschag4(尺), gin1 (斤), bong3 (磅)
Based on prototype theory (Allan 1977, Pinker 1989, Rosch 1975) and experiential view (Johnson 1987, Lakoff 1986), they identify the cognitive basis of classifiers in Taiwanese Hakka and construct their categorical structures. In their studies, the
7 Since the original container measure ping (瓶) ‘bottle’ in the study of Tai et al. (2001: 3) does not exist in Hakka, we replace it with ang1 (盎) as the corresponding word in this thesis.
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
‧
國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
21
prototype theory explains the classification of nouns to a given classifier. That is, the salient perceptual properties of nouns determine the selection of a classifier. For example, nouns whose salient characteristics are very thin and slight, such as leaves and chipped ginsengs, are classified into the category of the classifier pi5 (皮), which denotes the feature of thinness and slightness. Moreover, the experiential view accounts for the fossilized or conventionalized collocation of classifiers and nouns, which seem apparently arbitrary, unable to be analyzed in terms of shared semantic features. For instance, the noun tien5 (田) is specifically classified with the classifier kiu1 (坵) in the sense that fields are paramount to the agricultural society in Taiwanese Hakka cultures.
Since classifiers in different languages reflect culture-specific constraints, there has been much research concerning the comparison of classifier systems in
Taiwanese Mandarin, Taiwanese Southern Min, and Taiwanese Hakka (Wu 2001, Tai et al. 2001, Chen 2003, Qiu 2007). Based on prototype theory (Allan 1977, Pinker 1989, Rosch 1975) and experiential view (Johnson 1987, Lakoff 1986), Wu (2001), Tai et al. (2001), and Chen (2003) comparing the similarities and differences between the classifier systems in Taiwanese Mandarin, Taiwanese Southern Min, and Taiwanese Hakka, demonstrate the variation of classifiers specific to each language to unravel human categorization from different cultural backgrounds. As
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
‧
國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
an individual measure, for instance, the classifier tou (頭) ‘head’ is selected to classify cattle in Taiwanese Mandarin as in yi tou niu (一頭牛) ‘a cow’ and to classify trees in Taiwanese Hakka as in rhid4 teu5 shu7 (一頭樹) ‘a tree’; however, this classifier does not exist in Taiwanese Southern Min. Chen (2003) further analyzes the classification of classifiers and nouns from the functional perspective and attributes the shared classifiers in Taiwanese Southern Min and Taiwanese Hakka to the identical culture and agricultural life they shared in early days. For instance, the noun fish is classified with the classifier tiao (條) in Taiwanese Mandarin based on the salient perceptual property of its long shape; however, it is classified with the classifier mui1 (尾) both in Taiwanese Southern Min and
Taiwanese Hakka in the sense that the feature of its tail is accentuated when people caught fish by the tail in early times. Such an example exhibits that the selection of classifiers is determined by people’s interaction with the environment.
Aside from the comparison of the classifier systems between Taiwanese Mandarin, Taiwanese Southern Min and Taiwanese Hakka from the cognitive perspective and the experimental view, Qiu (2007) compares the syntactic structures of the clasisifer systems of them and indicates some uqnique structures of classifiers shared by Taiwanese Southern Min and Taiwanese Hakka, such as the structure [Quantifier-Classifier-Modifier-Classifier], which is not found in Taiwanese
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
‧
國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
23
Mandarin. As in the sentence na1rhid4 tiau5 tai7 tiau5 shiu1gon1 loi5 (拿一條大條 水管來) ‘fetch me a big water conduit’, the two classifiers tiau5 (條) ‘stripe’
function differently: the first classifier in rhid4 tiau5 (一條) ‘a stripe’ functions as a measurement while the second classifier in tai7 tiau5 (大條) ‘big stripe’ functions as a categorization tool. Qiu’s (2007) investigation on classifiers in Taiwanese
Mandarin, Taiwanese Southern Min, and Taiwanese Hakka demonstrates that the variety of classifiers is richer and the syntactic structures are more complex in Taiwanese Southern Min and Taiwanese Hakka than those in Taiwanese Mandarin.
Furthermore, Wu (2001) examines classifiers in Southern Hakka and Northern Hakka in Taiwan and finds that they are approximately identical. The only difference between them is individual measures. For example, the noun trousers is classified with the classifier tiau5 (條) in Northern Hakka whereas it is classified with the classifier liang1 (領) ‘collar’ in Southern Hakka; the noun white gourds is classified with the classifier tiau5 (條) in Northern Hakka whereas it is classified with the
Furthermore, Wu (2001) examines classifiers in Southern Hakka and Northern Hakka in Taiwan and finds that they are approximately identical. The only difference between them is individual measures. For example, the noun trousers is classified with the classifier tiau5 (條) in Northern Hakka whereas it is classified with the classifier liang1 (領) ‘collar’ in Southern Hakka; the noun white gourds is classified with the classifier tiau5 (條) in Northern Hakka whereas it is classified with the