The Ordering of Mandarin Chinese Light Verbs

The Ordering of Mandarin Chinese Light Verbs
Chu-Ren Huang1 and Jingxia Lin1
1
Department of Chinese and Bilingual Studies, The Hong Kong Polytechnic University, Hong
Kong
[email protected], [email protected]
Abstract. Two or more are sometimes found to co-occur in both Taiwan and
Mainland Mandarin Chinese, e.g., jiāyǐ and jìnxíng in duì xuéshēng jiāyǐ jìnxíng
yǐndǎo ‘to guide students’, and a particular ordering is often preferred, e.g., jiāyǐ
jìnxíng over jìnxíng jiāyǐ. This study argues that the order of the light verbs is
closely associated to two kinds of information that the verbs specify, i.e. aspectual eventive information and argument information. On one hand, a light verb
that denotes aspectual eventive information tends to occur before light verbs
without such information. On the other hand, a light verb with less argument information is more likely to occur after other light verbs and closer to the event
complement. The two semantic constraints form a general principle for the ordering of light verbs. The findings of this study can also contribute to a finergrained classification of Chinese light verbs.
Keywords: ordering of serial light verbs; aspectual eventive information; argument information
1
Introduction
Previous literature has not yielded a clear-cut and unified definition for the notion of
“light verbs” (cf. [1-3]; among others). In Chinese as well, studies differ from each
other in terms of the number of light verbs that Chinese has, cf. [4-7] and others. Taking a “loose” but comprehensive perspective, this paper defines light verbs as semantically impoverished verbs that may contribute information about event shape (e.g.,
beginning or ending of an event), but specify little about the kind of event under description (cf. [5]; [7]; among others). The event, i.e. the predicative content of a light
verb construction, mainly comes from the event-denoting element that is taken as
complement by the light verb. For instance, in the construction jìnxíng tǎolùn proceed-discuss ‘discuss’, the event of discussion is denoted by the complement tǎolùn
‘discuss’, whereas the light verb jìnxíng ‘proceed’ indicates a process aspect of the
event. With such a definition, the light verbs discussed in this paper not only include
the typical ones such as jiāyǐ ‘inflict’, jìnxíng ‘proceed’, and zuò ‘do’, but also the less
discussed verbs such as kāishǐ ‘start’, jiéshù ‘finish’, and jìxù ‘continue’.
In both Mainland and Taiwan Mandarin, two or even more light verbs may occur
together in a complex predicate; such co-occurrences are not abundant, but they are
sometimes found in governmental or institutional documents where the language is
expected to be formal and standard, as in (1).
(1) a. Rénmínbì huìlǜ
tiáozhěng bìxū jǐnshèn, yǒu bùzhòu,
RMB exchange.rate
adjustment must cautiously have step
jiànjìnde
jiāyǐ
jìnxíng
gradually
inflict proceed
‘Adjustment of RMB exchange must be carried out in a cautious, step-by-step,
and progressive way.’ (http://www.gov.cn/zlft2011/zbzy.htm)
b. Yòu'éryuán jiāng lùxù
kāishǐ jìnxíng bàomíng gōngzuò
kindergarten will successively start proceed registration work
‘The kindergartens will begin registration successively.’
(http://jtjy.e21.edu.cn/content.php?id=9362)
Table 1 Google tokens of light verb sequences (2012-4-10)1
Light verb
Mainland Taiwan
Example
sequence
10,300
8,900
jiāyǐ jìnxíng kòngzhì
jiāyǐ jìnxíng
inflict proceed
inflict proceed control
jìnxíng jiāyǐ
8,060
1,500
jìnxíng jiāyǐ chóngxiū
proceed inflict
proceed inflict rebuild
523,000
35,400
jìnxíng zuò wèijìng jiǎnchá
jìnxíng zuò
proceed do
proceed do gastroscopy
zuò jìnxíng
285,000
16,200
zuò jìnxíng jiǎyá xiūfù zhìliáo
do proceed
do proceed denture repair treatment
19,200
8,700
jiāyǐ zuò gèzhǒng wéitiáo
jiāyǐ zuò
inflict do
inflict do all.kinds.of fine-tuning
zuò jiāyǐ
4,670
1,070
zuò jiāyǐ qūbié
inflict do
do inflict distinguish
55,700
7,330
kāishǐ jiāyǐ zhěngdùn
kāishǐ jiāyǐ
start inflict
start inflict rectify
jiāyǐ kāishǐ
1,310
157
jiāyǐ kāishǐ jiǎnchá
inflict start
inflict start inspect
kāishǐ jìnxíng 10,400,000 2,380,000 kāishǐ jìnxíng bǐsài
start proceed
start proceed game
jìnxíng kāishǐ 83,000
38,800
gǎizào xiūshàn gōngchéng jìnxíng kāishǐ
proceed start
reform repair project proceed start
18,200,000
2,320,000
kāishǐ zuò shíyàn
kāishǐ zuò
start do
start do experiment
zuò kāishǐ
18,300
18,800
NA
do start
1
Although the token numbers are based on raw data where invalid examples are also counted,
the Mainland and Taiwan data consistently shows that a preferred ordering exists for each
pair of the light verbs.
More importantly, Internet data shows that the light verbs tend to follow some preferred orders when they co-occur. Table 1 presents raw data from the Google searches
for the co-occurrences of jiāyǐ ‘inflict’, jìnxíng ‘proceed’, and zuò ‘do’, and kāishǐ
‘start’ in Mainland and Taiwan websites; it shows that in both varieties of Chinese,
although examples of both orders are found when two light verbs co-occur, a particular order is often preferred over the other: jiāyǐ jìnxíng inflict proceed, jìnxíng zuò
proceed do, jiāyǐ zuò inflict do, kāishǐ jiāyǐ start inflict, kāishǐ jìnxíng start proceed,
and kāishǐ zuò start do are all more frequently used than the reversed orders.
Furthermore, the frequencies in Table 1 give rise to an ordering hierarchy for the
four light verbs, as given in (2).
(2) kāishǐ ‘start’ > jiāyǐ ‘inflict’ > jìnxíng ‘proceed’ > zuò ‘do’
The hierarchy illustrates that when two of the four light verbs co-occur, their leftto-right order tends to follow the hierarchy from left to right, i.e. “kāishǐ ‘start’ + jiāyǐ
‘inflict’/ jìnxíng ‘proceed’/ zuò ‘do’”, “jiāyǐ ‘inflict’ + jìnxíng ‘proceed’/ zuò ‘do’”,
and “jìnxíng ‘proceed’ + zuò ‘do’”.
The orderings presented in Table 1 as well as the hierarchy in (2) indicate that
there are internal differences among Chinese light verbs, and such differences may
not only provide an explanation to the order of the light verbs, but also shed light on a
better characterization and classification of the verbs. Therefore, a study is necessary
to discover the factors that bring about the differences and the ordering.
2
Semantic Properties Determining the Order of Chinese Light
Verbs
This study proposes that the relative order of Chinese light verbs is closely associated
to two basic semantic properties of the verbs: aspectual eventive information and
argument information. Aspectual eventive information refers to the information about
the overall (viewpoint) aspectual shape of event structure, including inception, process, completion, and so on. Argument information refers to the information about the
arguments or participant roles involved in the event, e.g., agent, goal, location, theme,
and so on. Eventive information and argument information are provided with different
representation in some theories of verbal semantics. For instance, according to the
MARVS (Module-Attribute Representation of Verbal Semantics, [8]), eventive information and argument information are represented as Event Module and Role Module, respectively. Furthermore, more refined specifications on a particular event or
argument are represented as Event-internal Attributes and Role-internal Attributes,
which thus enables the differentiation of synonyms even when they share the same
Event/Role Modules. The MARVS is able to provide a succinct but critical description of verbs, especially near synonyms. For instance, tóu, zhì, rēng, and diū all denote an event of throwing, but [9] argue, within the framework of MARVS, that these
verbs can be differentiated based on their aspectual and argument information: in
terms of aspectual information, diū ‘throw’ specifies a bounded process, whereas all
other three verbs only denote activities with a starting point (Event Module); in terms
of argument information, diū ‘throw’ and rēng ‘throw’ do not specify goal infor-
mation, whereas tóu ‘throw’ and zhì ‘throw’ specify direction and goal, and thus can
take spatial NPs as direct object (Role Module); in addition, tóu ‘throw’ and zhì
‘throw’ can be further differentiated that the former specifies the kind of goal associated to the event, as it often collocates with container-like locative NPs (Role-Internal
Attributes). The MARVS demonstrates the significance of the eventive and argument
information that a verb specifies in determining the verb’s syntactic distribution and
behavior.
The rest of this section discusses the eventive and argument information specified
by Chinese light verbs, and proposes the principle that determines the order of cooccurring light verbs.
2.1
Eventive Information of Chinese Light Verbs
It has been found in many languages that light verbs may contribute viewpoint aspectual information to the event described by the light verb construction ([10-11]; among
others). Different viewpoint aspectual systems have been proposed by previous studies (cf. [8]; [11-13]), and this paper follows [8] for Chinese light verbs. [8] identify
five atomic event structures ---- boundary, punctuality, process, state, and stage ---and argue that these atomic event structures can combine and generate twelve event
types with different degree of complexity. In Chinese, kāishǐ ‘start’, jìxù ‘continue’,
jiéshù ‘finish’, and jìnxíng ‘proceed’ are among the light verbs that denote aspectual
meanings. For instance, kāishǐ ‘start’ specifies inception. Evidence can be found with
the types of complements that kāishǐ ‘start’ takes. For example, both shuìzháo ‘fall
into sleep’ and dǒngde ‘get to understand’ are understood as instantaneous events, but
kāishǐ ‘start’ is only compatible with dǒngde ‘get to understand’. The incompatibility
of kāishǐ ‘start’ and shuìzháo ‘fall into sleep’ is probably because kāishǐ ‘start’ denotes an inceptive meaning, but an event of shuìzháo ‘fall into sleep’, in contrast to an
event of dǒngde ‘get to understand’, usually does not have a specific starting point (cf.
[10]). Unlike kāishǐ ‘start’, the light verbs jìnxíng ‘proceed’ and jìxù ‘continue’ specify a process. For example, both jìnxíng ‘proceed’ and jìxù ‘continue’ are compatible
with the continuative marker xiàqù, as in bǐsài jìxù xiàqù ‘the game continues’ and
diàochá jìnxíng xiàqù ‘the investigation continues’, cf. kāishǐ ‘start’ that is rarely
found with xiàqù.
We propose that when two light verbs occur together, the one with an aspectual
meaning occurs before the verb without such a meaning. For instance, jiāyǐ ‘inflict’
does not clearly specify any aspectual information, as evidenced by the fact that it
usually cannot be suffixed with aspectual markers (??jiāyǐ-le ‘perfective’/zhe ‘durative’/guò ‘experiential’), so jiāyǐ ‘inflict’ is preceded by kāishǐ ‘start’ when they cooccur. Furthermore, if both light verbs specify some aspectual meanings, their order is
determined by the logical relation of the aspectual meaning. For instance, although
both kāishǐ ‘start’ and jìnxíng ‘proceed’ denote aspectual information, kāishǐ jìnxíng
start proceed is preferred over jìnxíng kāishǐ proceed start because the former is consistent with the logical order that the inception comes before the process.
Nonetheless, the constraint of aspectual information is unable to determine all orderings that are found in Chinese light verbs. For instance, the constraint predicts that
jìnxíng ‘proceed’ will occur before jiāyǐ ‘inflict’ as the former contains aspectual
information, but as shown in Table 1, jiāyǐ jìnxíng inflict proceed is more frequently
used than jìnxíng jiāyǐ. Therefore, some additional constraint is needed. Section 2.2
introduces the new constraint.
2.2
Argument Information of Chinese Light Verbs
As introduced in Section 1, light verbs are usually devoid of concrete meanings, and
thus the information about the event under description usually comes from the complements taken by the light verbs. In Mandarin Chinese, the event-denoting complements can be deverbal nominals or event NPs, as in jìnxíng dì yī cì huìyì ‘to have the
first meeting’ and jìnxíng diàochá ‘to investigate’, respectively. However, as observed
in many previous studies, light verbs are semantically bleached to different degrees,
and thus differ in terms of the types of complements they can take. Table 2 is a comparison of the semantic and syntactic features of the complements to jìnxíng ‘proceed’
and jiāyǐ ‘inflict’ based on previous studies ([7]; [14-15]; among many others) and
corpus data.2
Table 2 Semantic and syntactic features of the event complements to jìnxíng ‘proceed’
and jiāyǐ ‘inflict’ (2012-4-14) (“?” marks that the use is unnatural)
Event complements
jìnxíng
jiāyǐ
‘proceed’ ‘inflict’
√
√
Semantic Formal and spontaneous
event (fēnxī ‘analyze’)
features
Durative event
√
√
(fēnxī ‘analyze’)
Event involving interaction
√
??
of the agent and patient
(gōutōng ‘communicate’,
hézuò ‘cooprate’)
√
√
Syntactic Transitive verb (kòngzhì ‘control’)
features
Intransitive verb (xiāoshòu ‘sell’)3
√
???
NP (dìyī xiàng yìchéng
√
????
‘The first item on the agenda’)
VP (wánshàn fúwù ‘improve service’)
????
????
Event complement at subject position
√
????
(huìyì zhèngzài jìnxíng ‘the meeting
is in progress’)
Table 2 shows that the event complements in jiāyǐ ‘inflict’ constructions are usually transitive and the events they describe do not involve any interaction between the
agent and the patient, e.g., gōutōng ‘communicate’ and hézuò ‘cooprate’. This indicates that jiāyǐ ‘inflict’ still maintains some of its literal meanings, that is, the event
associated with jiāyǐ ‘inflict’ is usually “imposed” by an agent onto the patient. In
2
Regional variations can be found in Taiwan and Mainland jìnxíng ‘proceed’ and jiāyǐ ‘inflict’,
but they are not discussed in this paper as the differences are relatively minor.
3
Xiāoshòu ‘sell’ is usually used as an intransitive verb in Mandarin Chinese, cf. mài ‘sell’
which is a transitive verb.
contrast, jìnxíng ‘proceed’ is more semantically bleached as it selects a larger variety
of complements.
In a double-light-verb construction, however, only the selectional restriction of
the second light verb, i.e. the verb closer to the complement, must be satisfied. For
instance, Table 3 is a summary of the semantic and syntactic features of the complements to jiāyǐ jìnxíng inflict proceed from Mainland and Taiwan Internet data (200
Baidu examples for each variety); the table shows that the types of complements to
jiāyǐ jìnxíng inflict proceed are very similarly to those to jinxing ‘proceed’, cf. Table 2.
For instance, intransitive verbal complements and complements denoting events involving interaction of the agent and patient are still found in the constructions despite
the presence of jiāyǐ ‘inflict’, cf. Table 2.
Table 3 Semantic and syntactic features of the event complements to jiāyǐ jìnxíng
inflict proceed (2012-4-14)
Event complement
jiāyǐ jìnxíng
inflict proceed
Mainland Taiwan
Semantic Formal and spontaneous event
√
√
features
(fēnxī ‘analyze’)
Durative event (fēnxī ‘analyze’)
√
√
Event involving interaction of the agent
√
√
and patient (miànduìmiàn de jiāoliú ‘faceto-face communication’, tǎolùn ‘discuss’)
Syntactic Transitive verb (kòngzhì ‘control’)
√
√
features
Intransitive verb (tuīlǐ ‘deduce’, pèiduì
√
√
‘make a pair’)
NP (yǒuxiào de tiáojié ‘effective
√
√
regulation’)
VP (Pínglián xuéqí chéngjī ‘assess
NA
√
semester academic grades’)
Event complement at subject position (Zīyuán
√
√
de pèizhì jīběn shàng tōngguò shìchǎng
jiāyǐ jinxing ‘Allocation of resources is
usually carried out through the market’)
Therefore, we propose that in order to optimally satisfy the selectional restriction
between the light verb and the complements, the light verb that allows a larger variety
of complements, i.e. the verb that is more bleached, tends to occur closer to the complement in a double-light-verb construction. In this sense, jìnxíng ‘proceed’ is expected to occur after jiāyǐ ‘inflict’ although it is more specific than jiāyǐ ‘inflict’ in
terms of aspectual information, cf. Section 2.1.
2.3
The Ordering Principle
Sections 2.1 and 2.2 introduce the two important semantic properties of Chinese light
verbs and their effect on the ordering of light verbs: (a) a light verb with eventive
information tends to occur at the beginning of a light verb construction, and light
verbs with different aspectual information follows the logical order of the aspect
meaning; (b) a light verb that is more semantically bleached tends to occur closer to
the complement. The two semantic constraints complement each other and form a
general principle for the ordering of light verbs.4
The ordering principle is able to account for the hierarchy in (2): on one hand,
kāishǐ ‘start’ precedes the other three light verbs because it specifies an inceptive
meaning; on the other hand, although jìnxíng ‘proceed’ is also associated with an
aspectual meaning, it is relatively more bleached than jiāyǐ ‘inflict’ and thus tends to
follow jiāyǐ ‘inflict’, and it is relatively less bleached than zuò ‘do’ and thus tends to
precede zuò ‘do’. Furthermore, although the co-occurrence of three light verbs is rarely found in the corpus data, the principle is able to provide a unified account for their
relative order. As illustrated in (3), the order of kāishǐ ‘start’, jìnxíng ‘proceed’, and
zuò ‘do’ is consistent with the hierarchy in (2).
(3) a. Yǐ
yǒu zhìshǎo liǎngjiā
kāishǐ jìnxíng zuò
already have at.least two.Classifier start proceed do
shì shāng jiāoyì de
yánjiū hé chóubèi
city business trade Modifier research and preparation
‘There are at least two brokerage firms who have begun the research and
preparation of city business transaction.’
(http://archive.news.sina.com.tw/破冰-推進-做市商-441910)
b. Gè zǔ
tóngxué tǎolùn yīxià, tǒngyī-le
yìjiàn,
each group fellow.student discuss a.little.bit agree-Perfective opinions
zài kāishǐ jìnxíng zuò shíyàn
then start proceed do
experiment
‘Students of each group should have a discussion to reach an agreement, and
then begin the experiments.’
(http://www.edudown.net/teacher/jiaoan/xqita/200607/7923.html)
3
Conclusion
This preliminary study investigated the relative order of co-occurring light verbs in
both Mainland and Taiwan Mandarin Chinese. It proposed that the ordering is codetermined by the eventive information and argument information that the light verbs
specify. The proposal of this study will also shed light on a better classification of
Chinese light verbs. In future studies, we will provide a more detailed representation
of the verbal semantics for each Chinese light verb in order to achieve a finer-grained
classification. In addition, a more comprehensive comparison of Mainland and Taiwan light verbs will be carried out in order to describe the possible regional differences for better cross-strait communication.
4
Note that in Table 1, the frequency difference between jiāyǐ jìnxíng inflict proceed and jìnxíng
jiāyǐ proceed inflict is much larger in Taiwan Chinese than in Mainland Chinese. This indicates
that the constraint of argument information may play a more important role in Taiwan Chinese,
but due to space reason, regional variations are not discussed in this paper.
Acknowledgments. We are very grateful to Ge XU for his help with the extraction of
the Baidu data. This work was supported by PolyU project 1-ZV8E.
4
References
1. Jespersen, O.: A Modern English Grammar on Historical Principles. Allen & Unwin, London (1942)
2. Grimshaw, J., Armin, M.: Light verbs and θ-marking. In: Linguistics Inquiry (1988) 205232
3. Butt, M., Geuder, W.: On the (semi)lexical status of light verbs. In: Corver, N., van
Riemsdijk, H. (eds.): Semi-lexical Categories. Mouton De Gruyter, Berlin New York
(2001)
4. Yin, S.C..: Shilun Nianzhuo Dongci [A tentative study on bound verbs]. In: Zhongguo
Yuwen (1991)
5. Zhu, D.X.: Xiandai Shumian Hanyu de Xuhua Dongci he Mingdongci [Light verbs and
verbal nouns in Modern literary Chinese]. In: Selected Papers of Zhu Dexi. Commercial
Press, Beijing (1999)
6. Li, L.D.: Xiandai Hanyu Dongci [Modern Chinese verbs]. China Social Science Press,
Beijing (1990)
7. Diao, Y.B.: Xiandai Hanyu Xuyi Dongci Yanjiu [A study on Modern Chinese abstract
verbs]. Liaoning Normal University Press, Shenyang (2004)
8. Huang, C-R., Ahrens, K., Chang, L-L., Chen, K-J., Liu, M-C., Tsai, M-C.: The moduleattribute representation of verbal semantics: From semantics to argument structure. In:
Computational Linguistics and Chinese Language Processing (2000) 19-46
9. Liu, M-C., Huang, C-R., Lee, C., Lee, C-Y.: When endpoint meets endpoint: A corpusbased lexical semantic study of Mandarin verbs of throwing. In: Computational Linguistics
and Chinese Language Processing (2000) 81-96
10. Butt, M.: The Structure of Complex Predicates in Urdu. CLSI, Stanford (1995)
11. Allerton, D. J.: Stretched Verb Constructions in English. Routledge, London New York
(2002)
12. Comrie, B.: Aspect. Cambridge University Press, Cambridge (1976)
13. Smith, C.S.: The Parameter of Aspect. Kluwer Academic Press, Dordrecht (1991)
14. Lü, S.X.: Xiandai Hanyu Babai Ci [800 words in Modern Chinese]. Commercial Press,
Beijing (1980)
15. Lu, F.B.: Duiwai Hanyu Changyong Ciyu Duibi Lishi [Comparative illustrations of common Chinese words and expressions]. Beijing Language and Culture University, Beijing
(2000)