Chapter 1 Introduction
1.3 Organization of the Thesis
國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
14
could lead to grammatical sentences, raw data of Japanese is needed. So one online Japanese corpus, JpWac in Sketch Engine, and Google Search are used to do the tests.
JaWac is a Japanese Web Corpus which contains 409,384,405 tokens and
333,246,192 words. The data is collected from 49,544 documents. All searched data is segmented according to the part-of-speech. JpWac is used to identify the nouns
adjacent to possible Japanese numeral Cs. In Google Search, the language is set to Japanese so that grammatical sentences that are commonly used are collected to examine the syntactic structure and to find the true Cs in Japanese.
1.3 Organization of the Thesis
This thesis is organized as follows. In the first chapter, the motivation and purpose of identifying Japanese numeral Cs are discussed, and the method of data collection is also presented in the introduction of this thesis.
In Chapter 2 on literature review, traditional and recent definitions of classifiers are first discussed to determine a proper and evident way for identifying true Cs in Japanese. Then, the various lists of Japanese Cs in previous studies and grammar books are reviewed. The definitions of each work are mentioned as well. The third part of the chapter touches on the issues of eliminating classical Cs in contemporary Japanese.
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
15
In Chapter 3, the main focus is to introduce a method of examining all the possible Japanese Cs, and then some syntactic and semantic tests are provided to ensure that the data collected from previous works is re-analyzed carefully. Chapter 4 provides the analysis of the Japanese classifier inventory for the present result and includes a bottom-up categorization figure of Cs. The last part of Chapter 4 contains the statistics on the usage frequency for Cs.
In the last chapter, some suggestions in teaching Japanese as a second language will be provided.
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
16
Chapter 2 Literature Review
Japanese numeral Cs has long been discussed in previous studies and traditional grammar books. However, the definitions of numeral Cs in previous linguistic studies are different, and it often leads to different standards for identifying Cs.
Most of the studies agree that Cs indicate some “inherent nature” of objects.
That is to say, Cs may occur with a set of specific nouns, and the nouns share some common features.
Nonetheless, different studies provide different ideas of the nouns’ “inherent nature”, and the definition of Cs is still not clear cut enough to identify Cs out of Ms (measure words). In addition, some elements such as pure nouns are counted as Cs in previous studies. Because of the different definitions and standards identifying the word class “C”, Cs listed in previous studies could be very different from one study to another.
In this chapter, previous studies are reviewed for three general aspects. First is to review the difference in the definitions of C in previous studies. Then, the
theoretical framework on which this thesis is based is addressed. Since Cs could be viewed as a kind of noun categorization device, previous studies discussing how
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
17
Japanese Cs categorize nouns is presented in this chapter as well.
The second aspect is to review the lists generated in previous studies. Since dissimilar definitions would lead to different standards for identifying Japanese Cs, and the main goal of the thesis is to identify the inventory of Japanese numeral Cs, the C lists discussed in previous studies are reviewed.
For the last part of the literature review, issues in contemporary Japanese and teaching Japanese as a second language will be reviewed. After the list of true Japanese Cs inventory is established, it can also be applied to practical issues of Japanese language.
2.1 Definitions of Japanese Numeral Classifiers
2.1.1 Common consensus of Japanese numeral classifiers
Traditionally, the discussions of Cs may contain not only common views but also some discrepancies. There is a common consensus suggesting that Cs will indicate some nature of nouns. For example, Aikhenvald (2000) reported that Cs are elements that categorize nouns by the inherent properties of the entities. Senft (2000) also claimed that the Cs might categorize nouns based on semantic criteria.
Therefore, Cs could select nouns with specific features – a true C could only collocate with a set of specific nouns. Lucy (1996) stated that more than one noun
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
18
could co-occur with one C, and those nouns share some features in common. Huang and Ahren (2003) studied Chinese Cs like 條, 朵, 回, etc., and suggested that Cs could “coerce” the adjacent nouns. Philip (2004) studied Japanese syntactic structure, and he also stated that there is semantic agreement between Japanese Cs and the adjacent nouns. Example (7) is quoted from Philip (2004; 163). The factor that makes (7a) grammatical and (7b) ungrammatical is the selection of the C and the semantic feature of the NP 学生 (“students”). In (7b), the morpheme “匹” is a C commonly used with animals such as cows and dogs, so “匹” could not co-occur with the human NP “学生”. However, in (7a) the C “人” would co-occur with human nouns. Thus, only example (7a) is grammatical. A specific Japanese C that will coerce a set of specific nouns, the Japanese C “人” in (7a) matches the human NP like “学生” or
“女”(woman), while the Japanese C “匹” in (7b) matches animal NP like “猫” (cat) or
“犬”(dog).Thus Philip suggested that the syntactic structure of Cs is conditioned by the semantic requirement, and he has proved that Cs may label some nature of the nouns.
(7) a. 三 人 の 学生 が 来-た。
san nin no gakusei ga ki-ta
3 C[+human] GEN student NOM come-past tense “Three students came.”
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
19
b. *三 匹 の 学生 が 来-た。
san biki no gakusei ga ki-ta
3 C[-human] GEN student NOM come-past tense “Three students came.”
In addition, there are features of Cs that do not cause a debate: the
“individuation” and the “counting” characteristics. Quine (1968) defined Cs from a philosophical point of view and suggested that the idea of original nouns like “牛” in Chinese is only a grouping of “cattle” in our mind, but by adding a C (for example, a cow “一隻牛”), the grouping term becomes an individuative term, making the thoughts distinct from the mass idea in conception.
For counting characteristic, Ueda (2009; 125) provided the following example (8) to explain that Cs denote countability feature of nouns. The C in example (8) is
“つ”, it is generic C and it can co-occur with many nouns but not with uncountable nouns. Since the noun “水” (water) in (8b) is uncountable, this makes (8b)
ungrammatical. This example proves that Cs denote the “countability” feature of NP.
(8) a. 一 つ の コップ。
Hito tsu no coppu 1 C GEN cup “one cup”
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
20
b. *一 つ の 水。
Hito tsu no mizu 1 C GEN water “*one water”
Nevertheless, the identification of Cs is still debatable because part of the definition is vague, making it hard to distinguish a C from M or other morphemes based on the common views of Cs.
2.1.2 Traditional views of classifiers and measure words
With regard to the discrepancies in the definitions of C’s, the idea of “indicate come inherent nature of nouns” is vague. Moreover, it is unclear what the standard is for distinguishing classifiers (Cs) from measure words (Ms). Some of the studies might draw a clear distinction between Cs and Ms, while some might not.
In studies that do not differentiate Cs from Ms, they might also view
morphemes indicating quantity or volume of the nouns as Cs. For example, Lyons (1977) suggested that Cs are obligatory in syntactic structure; they may indicate the nature of the entity. He divided Cs into sortal and mensural Cs. Sortal Cs could indicate the categorization of nouns, such as “條” (coerce long, thin objects) in Chinese. In contrast to sortal Cs, mensural Cs are elementally individuate nouns in terms of the quantity, such as “箱” (it means “box of something”) in Chinese. Croft
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
21
(2001) further divided the mensural Cs into four types – partitive, measure, group, and arrangement Cs. Chao (1968;585) also included both Cs and Ms in the morpheme
“measures”, and he divided “measures” into 9 types – group measures, partitive measures, and container measures, etc. Chao treated true Cs as a type of “measures.”
Although Gil D. (1994) ruled that mensural numeral Cs (like “箱” in Chinese) may occur with nouns of low countability, mensural numeral Cs are still treated as one type of Cs. Nonetheless, it is still unclear how the features of “quantify” and
“kind” can be considered “inherent features” of nouns. Some of the previous studies do not treat mensural Cs as true numeral Cs, but rather view them as Ms.
Although the studies discussed do not give a clear cut distinction between Cs and Ms, in actuality, the nature of “條” and “箱” in Chinese is not the same. “條” in Chinese indicates the long-shaped feature of NP (for example, “一條蟲” is
grammatical in Chinese, 一條書 is not), while “箱” does not coerce specific nouns (for example, 一箱蟲, 一箱書 are both grammatical in Chinese). It is necessary to make a further distinction between Cs and Ms.
Although some researchers have defined rules to distinguish Cs and Ms, the rules are unclear. For example, Tai and Wang (1990) suggested that Cs are words that categorize nouns by indicating some “salient properties” of the nouns, while Ms indicate the “quantity” of the nouns and, thus, could not be noun categorization
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
22
devices. Tai (1992) also stated that in Chao (1968)’s analysis, the “9 types of measures” could be divided into Cs and Ms. He thought that Chao (1968)’s
“individual measures” are actually true Cs, group measures, partitive measures, and container measures, etc., belong to Ms.
In addition, Cheng and Sybesma (1998) claimed that Cs are used to count countable entities, while Ms cannot denote individuality and they are used to modify mass nouns. Thus, they considered Ms as “mass-classifiers”.5 These studies tend to distinguish between Cs and Ms.
If the standard rules distinguishing C/M are not clear enough, it may lead to difficulties in differentiating Cs from Ms. In studies discussed previously, they suggest that Cs may denote some “salient properties” of nouns, while Ms may denote the “quantity” of nouns, the standard of judging whether a property of noun is “salient”
is unclear. If an M could indicate some features of nouns, the M could unintentionally be treated as a C. In example (9), all three sentences are grammatical in Chinese, the C ‘條’ in (9a) indicate long, thin objects, so the NP ‘繩子’ could be used with the C.
In (9b) and (9c,) the morpheme “包” (it means “bag of something”) does not refer to the inherent properties and individuality of specific nouns, and it denotes the quantity
of nouns, so normally it will be treated as an M in (9b). However, in the case of (9c),
5 The terminologies in Cheng and Sybesma (1998) were not Cs and measure Ms, they called them
“classifiers” and “massifiers”(or mass-classifiers).
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
23
it seems that “包” might accidentally indicate the property of “垃圾袋” (garbage bag).
Thus, the morpheme ‘包’ could probably be wrongly identified as a C. The problem of this definition results in the confusion over what degree we could view a
morpheme as denoting the “salient properties of the nouns.” The standard in identifying classifiers should be refined to be clearer and more precise.
(9) a. 一 條 繩子 (in Chinese) yi tiao shengzi
1 C-long rope
“One rope.”
b. 一 包 繩子 (in Chinese) yi bao shengzi
1 M-bag rope
“One bag of ropes.”
c. 一 包 垃圾袋 (in Chinese) yi bao lecedai
1 M-bag garbage bag
“One bag of garbage bags.”
Studies analyzing Japanese Cs may face the same difficulty in identifying C/M.
Philip (2007) reported that Cs are used to denote individuality, and he categorized
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
24
Japanese Cs into individual, measure, monetary, temporal, and degree Cs. Similar to the idea presented in this paper, Paik and Bond (2002) suggested that there are five types of Japanese numeral Cs: sortal (could be further divided into kind, shape and property classifiers), event, mensural, group, and taxonomic classifiers, as shown in (10). This typology is a mixture of both Cs and Ms.
(10) a. sortal classifiers：つ (tsu, generic classifier for individual items) b. event classifier：回 (kai, denoting times)
c. mensural classifiers：センチ (senchi, it means “centimeters”) d: group classifiers：群 (mure, it means “a group of”)
e: taxonomic classifiers：種 (syu, it means “a kind of”)
Although Tai(1994) suggested that Cs coerce nouns by their “inherent” or
“permanent” properties, while Ms refer to relatively “contingent” or “temporary”
properties of nouns. Nevertheless, to what degree the reference to nouns’ inherent property is sufficient to determine whether the elements are Cs? Additionally,
“relative” is not an absolute value, which resulting in a gray area between Cs and Ms.
Therefore, the distinctions between Cs and Ms in traditional analysis are not clear enough to differentiate Cs and Ms from each other.
2.1.3 The theoretical framework of the thesis
The theoretical framework of this thesis is based on the principles of identifying
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
25
Cs in Her (2010a, 2010b, 2011, 2012). Her provided two principles to determine whether the morphemes are true Cs or not. The principles are described in (11).
(11) a. C/M Distinction in Set Theory
Given a well-formed phrase [Num K N], X is the set of properties denoted by K, and Y is the set of properties denoted by N, Ki is C iff X⊂Y;
otherwise, K is M.
b. Mathematical Distinction of C/M
Given [Num X N], X=C iff X =1; otherwise, X=M.
The first principle (11a) indicates that Cs should refer to some properties of nouns, the morpheme K is a C if and only if the morpheme K can denote the
properties of the adjacent nouns. This principle in (11a) was consistent with the other studies.
The second principle (11b) provides a further constraint: Cs are equal to the multiplicand 1. In previous works, it is not debatable to state that Cs may indicate the individuality of countable nouns. However, the distinction between Cs and Ms is still vague if we only apply the first principle (11a). With the help of the second principle (11b), it will be much easier to distinguish Cs from Ms, since Her sets a precise standard to differentiate between Cs and Ms.
These principles will solve the problem with a clear distinction between Cs and
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
26
Ms. For example, in (12), the NP in (12a) and (12b) is “cigarette”, but being a C or M can lead to different ideas. In (12a), the C “本” in Japanese (the spelling is ほん hon /hɔn/) indicates one dimensional long feature of the cigarette, and the number of cigarettes in (12a) is 1x1=1, implying that there is only one cigarette in (12a). In (12b), the M “箱” does not specify any inherent property of the cigarette, and the number of
“箱” is not necessarily 1 cigarette either. “One box of cigarette” could elicit the idea of “one or more cigarettes.” The number of cigarettes in (12b) 1xM=not necessarily 1, therefore “箱” is a M.
(12)a. タバコ 一 本 tabako ip pon
cigarette 1 C 1x1=1 cigarette “one cigarette”
b. タバコ 一 箱 tabako hito hako
cigarette 1 M 1xM=not necessarily 1 cigarette “one box of cigarette”
Moreover, Her (2012) also proposed the concept of “profiler” to explain the characteristic of Cs. The following example (13) is quoted from Her (2012; 1673).
Her suggested that different Cs could refer to separate semantic attributes of nouns. In
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
27
example (13), “尾”, “條” and “隻” are all Cs in Chinese – the differences lie on the properties they denote. In (13a), “尾” may indicate the “tail” of a fish, and in (13b) and (13c), “條” and “隻” may profile specific “long-shaped” and “animacy” features of the fish. That is to say, Cs would not always indicate all features of NP.
(13)a. 一 尾 魚 (in Chinese) yi wei yu
1 C-tail fish
“One fish” (profiles the tail)
b. 一 條 魚 (in Chinese) yi tiao yu
1 C-long shape fish
“One fish” (profiles the long-shape) c. 一 隻 魚 (in Chinese) yi zhi yu
1 C-animacy fish
“One fish” (profiles animacy)
Her addressed that Cs are “profilers.” They could profile some or all features of an object. The theoretical frame with “profiler” will also solve the problem in
example (3). Recall the example (3) in chapter one (P.12), it appears that the C “発”
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
28
could not be the modifier of the NP (pistol), but the modifier of the VP (shooting event), leading to the traditional assumption suggesting Cs are modifiers of VPs. Once Cs are viewed as “profiler”, then the C “発” can refer to the shooting event only.
In the following example (14), (14a) presents the principle while (14b) and (14c) are examples in Japanese. (14a) states that if the element between the numeral and noun is a C, then the C could be deleted without changing the meaning, but a M could not. In example (14b), the semantic property of C “匹” (C for animals) is x1
multiplicand, so the deletion of C will not change the number of dogs it counts.
However, in (14c), the element between the numeral and noun is an M “箱”, which means “a box of cigarette”, the number of cigarette is not necessarily one, so the deletion will lead to misunderstanding. This syntactic test can be used to test Japanese Cs.
(14)a. C/M Distinction in Numeral Quantification Scope
Given a well-formed [Num K N], if Num scopes over N, then K = C;
otherwise, K = M.
b. 三 匹 犬 = 三 犬 san biki inu san inu 3 C dog three dog ‘three dog’
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
29
c. 一 箱 タバコ ≠ 一 タバコ hito hako tabako hito tabako 1 M-box cigarette one cigarette
‘One box of cigarette’ ‘*One cigarette’
Another principle (15) is from Her (2012; 1672). Her claimed that in Chinese the adjective modifying C could modify the following noun as well, but an adjective could not modify both the M and the noun. Japanese C does not support the
modification of an adjective, so this may not be applicable to Japanese.
(15) a. C/M Distinction in Adjectival Modification Scope
If either [Num A-K N] = [Num K A-N] or [A-K-de N] = [A-N]
semantically and A refers to size, then K = C, and K ≠ M.
b. 一 大 顆 蘋 果 = 一 顆 大 蘋 果 (in Chinese) yi da ke pingguo yi ke da pingguo
1 big C apple 1 C big apple
‘one big apple’ ‘one big apple’
c. 一 大 箱 蘋 果 ≠ 一 箱 大 蘋 果 (in Chinese) yi da xiang pingguo yi xiang da pingguo
1 big M apple 1 M big apple
‘one big box of apples ’ ‘one box of big apples’
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
30
Her (2012) also presented some syntactic tests for distinguishing Chinese Cs and Ms. These syntactic tests are useful to identifying Chinese Cs and Ms, and some of them could be applied to test Japanese Cs in this thesis. Her’s model of
distinguishing Cs from Ms, and Her’s idea of using Cs as “profilers” with some syntactic test will be the theoretical basis for this thesis.
2.1.4 Other elements that are not true classifiers
To define true Cs in the lists from previous studies, the main issue is the
definition of Cs and how to distinguish them from Ms. Her’s model is used to identify the Japanese Cs. In addition, there are other elements (such as echo Cs, pure nouns, units of measurement, etc.) that could possibly be identified as Cs in previous works.
Figure 1 summed up how 蔡 (2010; 20) described the traditional category of Cs as counters in Japanese grammar books. She, on the other hand, suggested that counters “助数詞” are in fact one type of nouns, and a majority of the traditional Japanese grammar books do not provide further categorization within counters.
Figure 1. Noun classifications of Japanese grammar book in 蔡 (2010; 20)
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
31
Since Japanese counters “助数詞” are words along with numerals, the counter list provided by Iida (1999) and NHK dictionary are relatively huge. Also, previous studies, such as T’sou (2001), might not strictly distinguish the terminology of
“counters” and “classifiers”. In T’sou (2001), T’sou used the terminology “counters or classifiers” for Japanese Cs. This indicates the fact that there is no exact distinction between Japanese counters and Cs in previous studies.
In traditional Japanese grammar, Cs are in a mixed category of counters.
However, the syntactic behaviors of units of measurement, pure nouns, echo classifiers, etc. are not the same. So they cannot be treated as the same syntactic element.
First of all, with regard to the echo Cs, Sornlertlamvanich, Pantachat, and Meknavin (1994) suggested that there is a stage in classifier generation where Cs may be created as many as nouns. Tai and Wang (1990) also reported that Cs are originally nouns. As a result, it is reasonable to say that some of the Cs may be created by repeating the nouns. Downing (1984, 1996) proposed that echo Cs are “repeaters”,
First of all, with regard to the echo Cs, Sornlertlamvanich, Pantachat, and Meknavin (1994) suggested that there is a stage in classifier generation where Cs may be created as many as nouns. Tai and Wang (1990) also reported that Cs are originally nouns. As a result, it is reasonable to say that some of the Cs may be created by repeating the nouns. Downing (1984, 1996) proposed that echo Cs are “repeaters”,