4. Analysis
4.1 Data Analysis in Languages
4.1.4 Optional classifier languages
國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
Teribe follows the strict CPCD principle; classifiers and plurals do not co-occur in a noun phrase.
(78) a. domer kl-öbö /doglo mya Man CL.ANIMATE -some/ CL. ANIMATE three
‘some/ three men’ (Quesada. 2000: 52)
b. *domer-ga kl-öbö /doglo mya Man PL CL.ANIMATE -some/ CL. ANIMATE three
‘some/ three men’ (Quesada. 2000: 52)
4.1.4 Optional classifier languages
In this section, we investigate the languages whose classifiers are optional according to Gil(2008), including Hatam, Tuvaluan, Hungarian, Turkish, Ainu, Khmer, Indonesian, Tetun and Chantyal. However, The real classification of classifiers in such languages are not necessarily optional. The role of classifiers in each language is according to collected data.
4.1.4.1 Hatam
Both classifiers and plurals are optional in Hatam. Four kinds of classifiers are introduced by Reesink(1999): -ngud for animal; -njon for flying species except for insects; -mon for tree; -ngan for seeds. Since the data are deficient, we can only assume the range of application of classifiers is ranked as inanimate classifiers based
‧
on the description of Reesink (1999).
(79) Optional classifiers
There are two ways of number marking in Hatam. One is plural marker; the other is reduplication. A plural marker -nya can optionally (based on WALS online database) apply to animate common nouns.
(80) Range of application of plural marker a. Human
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
However, the plural marker -nya does not attach to nouns, but to the final position of noun phrases (Reesink 1999).
(82) Krau misien ni-de-nya Grab dog 3GS-POS-PL
‘He grabbed his dogs.’ (Reesink 1999: 50 )
In addition to plural marker -nya, there is a collective -bat in Hatam.
(83) Collective
dani kin di-sut-bat-nya
I with 1SG-friend-COLL-PL
‘I and my friends’ (Reesink 1999: 50 )
The co-occurrence of classifiers and plurals is grammatical in Hatam.
(84) Di-kindig-bat-nya i-bou can kin di-cig 1SG-older.sib-COLL-PL 3PL-head two with 1SG-father
‘My two older brothers and my father….’ (Reesink 1999: 83)
4.1.4.2 Tuvaluan
The data in Tuvaluan is so limited that we can only know little about its classifiers. We found the same reference as WALS online database, Tuvaluan: A Polynesian Language of the Central Pacific (Besnier 2000). There are only five classifiers introduced by Besnier (2000): toko, tua, tao, tau, and tuu. toko is for human;
tua and tao is for flat things that are piled up; tau is for coconuts in bunches; tuu is for groups of birds or kinfish.
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
(85) Range of application of classifiers a. Human classifier: toko
te toeaina i te toko lua i te feituu ki tai the old-man at the Num two at the side to lagoon ‘the second old man towards the lagoon’ (Besnier 2000: 570)
b. Inanimate classifier: tua E tua faa ana gatu
Nps N-layer four her clothes
‘[She] is wearing four layers of clothing’ (Besnier 2000: 570)
Although Besnier introduced classifier system in Tuvaluan, he stated that the classifiers are classifier-like elements. Further, from personal communication with Besnier, he suggested that Tuvaluan is not an ideal target when exploring classifiers.
Therefore, Tuvaluan is ignored in this study.
4.1.4.3 Hungarian
Classifiers are optional in Hungarian and include measure words, general classifiers, human classifiers, inanimate classifiers, and shape classifiers.
(86) Optional classifiers a. With a classifier
2 szem gyöngy
‘2 pearls’ (Csirmaz and Dékány 2010: 7)
b. Without a classifier 2 gyöngy
‘2 pearls’ (Csirmaz and Dékány 2010: 7)
‧
‘seven people belonging to a crew’ (Csirmaz and Dékány 2010: 17)
d. Inanimate classifier: fej Hungarian, and applies to all common nouns.
(88) a. Human
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
In addition to the plural marker, there is also a collective -ék in Hungarian
(89) Collective: ék Péter- ék Peter-COL
‘Peter and his friend(s)/ family’ (Kenesei, Vago, and Fenyvesi 1998: 254)
The co-occurrence of classifiers and plurals is grammatical as in (94). However, plurals can not co-occur with numeral as in (95). Thus we assume that in a noun phrase, [Num+CL+PL] is ungrammatical. Classifiers and plurals can co-occur only if there are no numerals.
(90) ex-ek a szem-ek rohadt-ak this-PL the CL-PL rotten-PL
‘These (ones) are rotten.’ (Csirmaz and Dékány 2010: 13)
(91) három takaró-(*k) three blanket-PL
‘three blankets’ (Csirmaz and Dékány 2010: 13)
4.1.4.4 Turkish
Classifiers are optional in Turkish and include human classifiers, animate classifiers, and inaimate classifiers.
(92) Optional classifiers Without a classifier:
beş çocuk
‘five children’ (Kornfilt1997: 265)
‧
(93) Range of application of classifiers a. Measure word
Nouns are transumeral in Turkish. A plural marker lar/ler can optionally apply to all nouns. When the number is denoted by other elements as numeral, plural markers can not occur.
(94) Transnumeral nouns Çarşıda CD aldım
‘I’ve bought a CD/ CDs in town.’ (Aarssen 2001 : 85)
(95) Range of application of plurals a. Human
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
As for the co-occurrence of classifiers and plurals, it is ungrammatical in Turkish (Kornfilt p.c.). Kornfilt, an expert of Turkish, states that classifiers do not co-occur with plurals in general. If classifiers and plurals do co-occur, the classifier is no longer a classifier, but a noun.
4.4.1.5 Ainu
Classifiers are optional in Ainu and include human classifiers and inanimate classifiers.
(96) Optional classifiers
tu okkaypo rupne sike ki pa two young-man big luggage do PL
‘Two young men carried much luggage.’ (Bugaeva, p.c.)
(97) Range of application of classifiers
a. Human classifier:-n(after vowel)/-iw(after consonant) tu-nv iwan-iw
‘two (people)’ ‘six people’
b. Inanimate classifier: -p tu-p
‘two (thing)’ (Bugaeva 2012: 471)
Nouns are transnumerals in Ainu. A plural marker -utar can optionally apply to all common nouns.
(98) Range of application of plurals a. Human
ainu-utar
‘men’ (Patrie 1982 p.131)
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
b. Animate chikap-utar bird PL
‘birds’ (Kindaichi& Chiri 1936: 30 )
c. Inanimate čise-utar
‘houses’ (Patrie 1982: 131)
From personal communication with the expert of Ainu, Bugaeva, the co-occurrence of classifiers and plurals is grammatical in Ainu as in (99).
(99) a. okkaypo utar tu-n sike ari.
young.man PL two-human.CLF luggage put.down.PL
‘These two young men put their belongings down’ (Bugaeva, p.c.) b. ne pa ta acapo utar tun ne kimun a wa
this year LOC uncle PL two-human.CLF COP go.to.the.mountains.to.hunt PERF ‘This year, two uncles went to the mountains to hunt’ (Bugaeva, p.c.)
4.1.4.6 Khmer
Classifiers are optional (Gilbert 2008) in Khmer, including measure words, human classifiers, animal classifiers, inanimate classifiers and shape classifiers.
(100) Range of application of classifiers a. Measure word:
/sac pii kiilou/
Meat two M
‘two kilos of meat’ (Ehrman & Sos 1972: 17)
‧
d. Inanimate classifier: bͻn-dtub ‘rooms’(Gilbert 2008: 201) e. Shape classifier: groab
kñom ñam bey groab haʉy
‘I took three pills’ (Gilbert 2008: 208)
Nouns are transnumerals in Khmer. Its plural marker can optionally apply to all common nouns.
(101) Transnumerals
a. khla ‘tiger or tigers’ (Gorgoniyev.1966: 68) b. baha ‘question or questions’ (Gorgoniyev.1966: 68)
There are two ways to denote plurality: one is plural marker; the other is reduplication. But the reduplication form can only apply to human nouns.
(102) a. Plural marker: tǝla:y
Based on a native speaker of Khmer, Soksan Ngoun, there is no words such as tǝla:y. He supposed that the word only used in ancient language such as poem.
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
They no longer use it in daily life. Thus we focus on the reduplication form in following discussion.
The co-occurrence of classifiers and plurals is grammatical in Khmer. Mr.
Soksan indicates that there are two ways to express ‘two men’.
(103) a. Proas bei nak man three CL
‘three men’ (Soksan, p.c.)
b. Proas proas bei nak man man three CL
‘three men’ (Soksan, p.c.)
4.1.4.7 Indonesian
Classifiers in Indonesian are optional and include measure words, general classifiers, human classifiers, animate classifiers, inanimate classifiers, and shape classifiers.
(104) Optional classifiers Without a classifier
dua orang
‘two people’ (Sneddon.1996: 132) (105) Range of application of classifiers
a. Measure word se-bungkus rokok One M cigarette
‘a packet of cigarettes’ (Sneddon 1996: 138)
b. General classifier: buah (Sneddon 1996: 136)
‧
Classifiers are optional in Indonesian, but they are obligatory when co-occuring with numeral ‘one’ (Sneddon 1996).
Nouns are transnumerals in Indonesian. The ways to express plurality is
The co-occurrence of classifiers and plurals is ungrammatical in Indonesian. A native speaker of Indonesian, Johnny, states that it is ungrammatical to say dua biji
‧
telurtelur ‘two eggs’. Besides, we can find dua biji telur from Google search engine, but no examples were found for dua biji telurtelur.
4.1.4.8 Tetun
Classifiers in Tetun are optional and include measure words, human classifiers, animate classifiers, and shape classifiers.
(108) Optional classifiers (109) Range of application of classifiers
a. Human classifier: na’in c. Shape classifier: lolon
tais lolon ida Cloth CLS:trunk one
‘one (handwoven) cloth’ (van Klinken 1999: 105)
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
Although classifiers are optional in Tetun, it is a polite expression if there are classifiers (van Klinken. 1999). Thus, (108a) is more polite than (108b).
In addition, the frequency of the appearance of classifiers varies according to numerals. For example, classifiers are rarely used with ida ‘1’ (van Klinken. 1999).
This phenonmenon is quite similar to plural –s in English which is used except for ‘1’.
Nouns in Tetun are transnumerals. A plural marker -sira can optionally apply to all common nouns. Although it is optional, there are always a plural marker when expressing plurality (van Klinken. 1999).
(110) Transnumerals:
Ami lori kuda 1PE bring horse
‘We bought a horse/the horse/horses’
(van Klinken, Hajek, and Nordlinger 2002: 19) (111) Optional plurals
a. With a plural marker livru sira
‘the books’ (van Klinken, Hajek, and Nordlinger 2002: 20)
b.Without a plural marker
ata nia-k rua ne’e slave 3S-POS two this
‘these her two slaves’ (van Klinken 1999:.133)
(112) Range of application of plurals a. Human:
klosan sia
‘maleservant’ (van Klinken 1999: 113)
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
b. Inanimate:
livru sira
‘the books’ (van Klinken, Hajek, and Nordlinger 2002: 20)
In addition to the plural –sira, there is a Portugese plural -s in Tetun due to language contact. Even more, -s sometimes can co-occur with –sira as in (113) .
(113) livru-s sira book PL PL
‘the books’ (van Klinken, Hajek, and Nordlinger 2002: 20)
The plural marker –sira can also be used as collective as in (114).
(114) Simao sira
Simao ASSOC.PL
‘Simao and his family/friends/colleague’
(van Klinken, Hajek, and Nordlinger 2002: 20)
There is no clear example of the co-occurrence of classifiers and plurals. But
Hajek, the expert of Tetun, states that he only collected the data as feto nain rua ‘two women’ from native speakers of Tetun. So it is more likely that the co-occurrence of
classifiers and plurals is ungrammatical.
4.1.4.2 Chantyal
The classifier system of Chantyal is borrowed from Nepali and divided into two types: human -jana and non-human -ta. There is only measure words in its own language as in (115). And the classifiers are optional based on WALS online database.
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
(115) Range of application of classifiers a. Measure word: gilas
yek gilas ca one glass tea
‘one glass of tea’ (Noonan 2003: 320)
b. Animate classifier:-ta
tin-ta jmmy naku-ma three-CL all dog-PL
‘all three dogs’ (Noonan 2003: 318)
With regard to number marking, a plural marker -ma can optionally apply to all common nouns. But no representative examples were found. In addition, the plural marker -ma can also be used as collective as in (116).
(116) Collective:
Ram-ma
Ram-ASSOC.PL
‘Ram and his family/companions’ (Noonan 2003: 318)
The co-occurrence of classifiers and plurals is grammatical as in (117).
(117) tin-ta jmmy naku-ma three-CL all dog-PL
‘all three dogs’ (Noonan 2003: 318)
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
4.2 A Coordinate Representation of Language Distribution
Based on the data analysis, we rank the range of application of classifiers and plurals onto a scale to find the completion of both systems as we have mentioned in Chapter 3. The results are shown as in Figure1.
Figure 1. The Range of Application of Classifiers and Plurals
(M: measure word、P: Plural personal pronoun、OP:optional、OB: obligatory、H:
human、A in X axis:animal、A in Y axis: animate、NA: Inanimate、S: shape)
Along the X axis, classifiers are scaled in ranking as (23) Measure words <
Optional classifiers (human) < Optional classifiers (animal) < Optional classifiers (inanimate) < Optional classifiers (shape) < Obligatory classifiers (human) <
Obligatory classifiers (animal) < Obligatory classifiers (inanimate) < Obligatory classifiers (shape); along the Y axis, plurals are scaled in ranking as (25) Plural
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
personal pronoun < optional human plurals < optional animate plurals < optional inanimate plurals < obligatory human plurals < obligatory animate plurals <
obligatory inanimate plurals. The higher ranked one will imply the existence of the lower one, so it represents that the system in such language is more complete.
However, part of our data are insufficient, thus we adopt the categorization in WALS online database to compensate for such absence so as to complete the figure.
For example, the necessity of classifiers is unclear in Nivkh, and we assume that it is obligatory based on WALS online database; the range of application of plurals is unclear in Vietnamese, and we assume that the plurals can apply to all nouns based on WALS online database.
From figure1, we can find that the range of application of classifiers and plurals are in certain complementary distribution except for Mokilese and Kathmandu Newar.
There are no language in the lowest ranked of classifiers and plurals, and only 2 languages containing the highest ranked of classifiers and plurals. In other 19 languages, if the classifier system is stronger, the plural system will be weaker, and vice versa. So the distribution is along a negative slope which means that the range of the usage of classifiers and plurals are in complementary distribution. Therefore, the degree of the violation of the CPCD principle in the 19 languages (except for Mokilese and Kathmandu Newar) are not strong.
‧
In addition, we also summarize the co-occurrence of classifiers and plurals in a noun phrase in 20 languages (except for Tuvaluan and Kham) based on the data as in Table 3. We do not take Tuvaluan and Kham into consideration because Tuvaluan is not an ideal target when examining the issues of classifiers, and Kham only contains measure words.
Table 3 The Categorization of Co-occurrence Co-Occurrence co-occurrence of classifiers and plurals number more than those languages which do not allow the co-occurrence of classifiers and plurals. Thus, the co-occurrence of classifiers and plurals is not accidental. To account for the co-occurrence of these languages, we will provide a syntactic analysis in Section 4.3.
‧
mismatches with the categorization provided by WALS online database, and we make a comparison of the categorization in WALS online database and in the data analysis in this paper as in Table 4.Table 4. Language Categorization
Languages Categorization in WALS online database
Categorization From Data Analysis
a. Mandarin Classifiers: Obligatory Plurals: Only human nouns,
Plurals: animate nouns, obligatory e. Belhare Classifiers: Obligatory
Plurals: All nouns, optional in inanimates
e. The same
f. Mokilese Classifiers: Obligatory
Plurals: All nouns, always h. Nivkh Classifiers: Obligatory
Plurals: All nouns, always
‧
o. Tuvaluan Classifiers: Optional
Plurals: All nouns, always distribution in different aspects as following.
‧
Table5. The Phenomenon Involved in the Complementary Distribution of Classifiers and Plurals
Thus, there are two ways to analyze the category of classifiers and plurals. One is that classifiers and plurals are different heads of projection (Li 1999, Huang 2009, and Fassi Fehr 2005). The other is that classifiers and plurals are the same category (Borer 2005, and Her 2012a).
When classifiers and plurals are different heads of projection as in (118), we can explain the languages with both classifiers and plurals and the co-occurrence in a noun phrase.
(118)
(Li 1999: 87)
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
However, from the viewpoint of typology, such co-occurrences of classifiers and plurals are rare. There is a tendency for languages to have only one of the system.
Therefore, in this paper, we adopt Borer (2005) and Her’s (2012a) claim that classifiers and plurals are the same category, as in (119).
(119)
(Vinet and Liu 2008: 361) In this structure, classifiers and plurals occupy the same syntactic node, so it is reasonable for them to be in complementary distribution in majority of languages.
However, it is impossible for classifiers and plurals to co-occur in the nominal structure as in (119). If such were the case the prediction would be that the co-occurrence of classifiers and plurals such as san ge xuesheng men ’three students’
and san-nin gakusei-tati ’three students’ would be ungrammatical. However, among the 22 languages in Sections 4.1.3 and 4.1.4 , 14 of them, Chinese, Japanese, Taba, Kathmandu Newar, Belhare, Nivkh, Vietnamese, Ulithian, Jacaltec, Hatam, Hungarian, Ainu, Khmer and Chantyal allow classifiers and plurals to co-occur.
Therefore, the co-occurrence of classifiers and plurals is indeed grammatical, rather
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
than accidental.
Thus, the nominal structure should be revised to explain the 14 languages as following:
(120)
In this new nominal structure, we can account for the majority of languages with either classifiers or plurals because they are the same category; and furthermore, we can include the co-occurrence cases due to the consideration that they are co-head.
For example, san ge xuesheng men ’three students’ in the syntactic structure is as following.
(121)
Therefore, the structure as (120) is acceptable thus far in explaining the data in the 22 languages.
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
4.4 Possible Explanations
The co-occurrences of classifiers and plurals are grammatical in languages, but they should be regarded as marked cases because they are co-head structure. Here are two possible reasons for the marked cases to occur: one is language contact; the other is language change.
De Leon (1987) proposed that a classifier language, Tzotzil, has more obligatory number marking due to the influence of Spanish. In this paper, we also find similar cases in our target languages. Classifiers in Chantyal are borrowed from Nepali;
certain of the classifiers in Belhare are also borrowed from Nepali; the plural system in Tetun is affected by the plural from Portuguese. In addition, the change of plurals in Chinese is also a good example for the influence of language contact. The plural -men was marked on human nouns in the past, such as xueshengmen ‘students’ or laoshimen ‘teachers’. But the range of application of -men by the younger generation
has been expanded to include animal or parts of inanimate nouns, as xiaogoumen
‘dogs’ or zhuozimen ‘tables’. This phenomenon may be owing to the learning of
English. Children learn the plural -s in English, and apply the range of usage of -s to -men. Therefore, with language contact, the system of classifiers or plurals may become complex.
As for language change, Chinese is also a good example. In ancient Chinese,
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
classifiers were optional. Nouns could be directly adjacent to numerals as (122a).
However, in modern Chinese, there must be a classifier between er zi and er shan as in (122b) and (122c).
(122) a. 命 夸 娥 氏 二子 負 二 山 ming kua e shi er zi fu er shan.
‘Ask kua e shi’s two sons to carry the two mountains’ (Leizi 列子)
b. 兩 個 兒 子 liang ge er zi
‘two sons’
c . 兩 座 山 liang zuo shan.
‘two mountains’
During the era prior to Qin Dynasty, the number of classifiers increased in line with the setting up of the measurement system. And more classifiers were produced in Han Dynasty. Since then, classifiers become prosperous until nowadays(Wang, 1989).
As for the plural system, there were many plural markers in ancient Chinese, such as bei, deng, cao, and men.
(123) a. 此 客棧 正 為 我 輩 而 設 ci ke zhan zheng wei wo bei er she
‘the hotel is for us’ (shui hu zhuan 水滸傳)
b. 料 爾 等 米 粒 之 珠 liao er deng mi li zhi zhu
‘I guess you are tiny as rice.’ (feng shen yan yi 封神演義)
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
c. 爾 曹 身 與 名 具 滅 不 廢 江 河 萬 古 流 er cao shen yu ming ju mie, bu fei jian he wan gu liu
‘Your body and fame will vanish, but the river will flow on forever’
(Tu-Fu 杜甫, xi wei liu jue ju 戲為六絕句) d. 看 他 門 得 人 憐 秦 吉 了
kan ta men de ren lian qin ji le.
‘Look at them, those lovely parrots.’ (XinQiJi 辛棄疾, qian nian tiao 千年調).
In the Song Dynasty and the Yuan Dynasty, mén, mei and mèn are the major plural marks. However, mei and mèn gradually disappeared, and only mén is applied in modern Chinese (Zhang, 1987). Thus, the plural system was stronger in ancient Chinese than in modern Chinese. From the variation in the system of classifiers and plurals, it is reasonable for both system to co-occur in a certain period of time.
In the Song Dynasty and the Yuan Dynasty, mén, mei and mèn are the major plural marks. However, mei and mèn gradually disappeared, and only mén is applied in modern Chinese (Zhang, 1987). Thus, the plural system was stronger in ancient Chinese than in modern Chinese. From the variation in the system of classifiers and plurals, it is reasonable for both system to co-occur in a certain period of time.