國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
validity of explicating the interaction between modal construction and modal
interpretation, we extract more modal co-occurrence samples in GigaWord Corpus,
which contains about 1.1 billion Chinese characters, including more than 700 million
characters from Taiwan’s Central News Agency, and nearly 400 million characters
from China’s Xinhua News Agency. For the sake of comparison and emphasis, some
self-created sentences and examples in other studies are also included in this thesis. In
addition, in the present study, the data source of all examples from corpus or other
literatures will be noted followed by the example. Otherwise, the instances are
introspected if no data source specification.
1.3 Outline of Thesis
Chapter I gives the brief but insightful introduction of this study. Chapter II reviews
previous literatures on Modality issue including syntactic approach and
semantics-pragmatics interface approach. Chapter III explicates different modal
interpretations of Chinese polysemous modal verbs. Chapter IV accounts for the
multiple-modal construction in Mandarin Chinese. Chapter V concludes this study.
‧ 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
Literature Review
In this chapter, we will mainly review two types of studies on Modality in
Mandarin Chinese: a syntactic approach and a semantics-pragmatics interface
approach. The syntactic studies investigate the syntax of modals within the framework
of Chomskian approach while the semantic-pragmatic studies explore various senses
of modal interpretation via the examination of interaction between intrinsic lexical
semantics and structural implication of utterance. In the first part, we will begin with
the syntactic exposition and then point out their deficiency in capturing the insight of
Chinese modal system. Before going into the discussion of syntactic framework, we
will first explicate the criteria of Chinese modal verbs for determining the final list of
Chinese modal verbs in the present research. In the second part, in terms of semantic
and pragmatic research, we will compare the descriptive-oriented works done by
Lyons (1977), Coates (1983) and Palmer (1990, 2001) with the relevance analysis
postulated by Sweetser (1990), Klinge (1993) and Papafragou (2000).
2.1 Syntactic Approach
2.1.1 Syntactic Status of Chinese Modal Verbs
Modality can be expressed by different syntactic categories. In English, for
example, modality can be interpreted as modal auxiliaries such as may, might, and
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
should, in modal verbs such as assume or allow, and in modal adverbs such as
possibly or certainly. In Chinese, on the other hand, Tang and Tang (1997) specify that
Chinese modal system can be expressed in modal particles such as ba in ‘ta zou le ba 他走了吧’ and ne in ‘ni shuo hao bu hao ne?你說好不好呢?’ and in modal adverbs
such as huoxu ‘maybe’ and dagai ‘perhaps’ and so on. However, their status and the
criteria for identifying them are still controversial issues in Chinese linguistics. Some
argue that Chinese modal verbs belong to verbs, but are different from full verbs (Lin
and Tang, 1995; Tang, 2000); others argue that Chinese modals can precede full verbs
like adverbs or behave like auxiliaries between subject and predicate (Tsai and Porter,
2008). None of them nonetheless can provide a holistic account for the syntactic
status of Chinese modal verbs. So we are wondering if there are any valid criteria for
differentiating these categories in Chinese modal system.
Li & Thompson(1981: 172-4) suggest that modal verbs in Chinese are modal
auxiliaries and they possess following features:
A. They do not take aspect markers.
B. They cannot occur before the subject as English modals do in question.
C. They cannot be nominalized.
D. They cannot be modified by intensifiers, such as hen很, geng更.
As for the above criteria, some problems arise. For example, we don’t agree with the
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
criterion D as there are many counterexamples to refute this claim. For instance, in
this sentence 這個人很能吃苦 Zheigeren hen neng chiku ‘This person is capable of
taking hardship’, the intensifier hen很 modifies the modal verb neng能. Xiaoming bi
Xiaoli genghui change 小明比小李更會唱歌 ‘Xiaoming can sing better than Xiaoli’,
in this sentence, intensifier geng更 modifies the modal verb hui會. Ni bi ta gengneng
chiku 你比他更能吃苦 ‘You can bear more ordeals than he’, in this sentence, the
intensifier geng更 modifies the modal verbs neng能. On the other hand, Tang(2000)
pointed out that in traditional syntactic analyses, modal verbs and modal adjectives
are “auxiliaries”. They listed seven syntactic characteristics of modal verbs and
adjectives in order to differentiate them with modal adverbs in Chinese. These
include:
A. They can appear alone in an answer or a predicate.
B. They can have A-not-A question forms and can be negated.
C. They can be the focus in a cleft sentence and appear after shi是.
D. When they are predicates, they can appear in ‘shi……de’ construction and be the
focus when nominalized.
E. Modal adjectives can be modified by intensifiers and they can appear in
comparative structures.
F. They can appear either before or after a negative adverb.
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
G. They can coexist with other similar or different kinds of modal verbs or modal
adjectives.
These criteria seem to be quite reasonable compared to Li and Thompson. However,
Tang didn’t make a clear separation between modal verbs and modal adjectives in his
paper. On the contrary, both of them have the same syntactic status based on his
analysis. In addition, the existence of adjective category in Mandarin Chinese is
disputed (Huang, 2008). Thus, we accept his criteria for Chinese modal system but
only admit that there only exists the category of modal verbs and modal adverbs in
Mandarin Chinese. In what follows, we adopt the above criteria and make a detailed
comparison among common verbs and modal verbs and modal adverbs:
modal verbs and common verbs
I. Modal verbs must be used in the construction “ModV + V + N”, not “ModV +
N”. That is to say, modal verbs cannot take nominal object, while common verbs can.
Sentences (6a) and (7a) are acceptable, whereas (6b) and (7b) are not.
(6) a. 他能開車
Ta neng kai che.
He able to drive car “He is able to drive a car.”
b. *他能車
‧
“You are able to write words.”
b. *你能字
Ni neng zi.
You able to words
“*You are able to words.”
II. Modal verbs cannot combine with the aspect markers of –le了,- zhe著,- guo過.
In other words, they do not allow the construction “ModV + le/zhe/guo + V” while
common verbs can allow the construction “V + le/zhe/guo + V”. Hence, sentence (9)
is ungrammatical in Chinese, but sentence (8) is acceptable.
(8) 他學過唱歌
“He has become able to sing”
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
III. Modal verbs cannot be reduplicated, while common verbs can:
(10) a. 你想想再做決定
Ni xiangxiang zai zuo jueding You think think then make decision
“You think carefully and then make a decision.”
b. *你想想去台北
Ni xiangxiang qu Taipei You want want go Taipei
“You want to go to Taipei.”
Xiang in sentence (10a) is a common verb, so it can be reduplicated; xiang in sentence
(10b) is a modal verb, so it cannot be reduplicated.
IV. Modal verbs can be modified by the intensifiers, while common verbs cannot:
(11) a. 他很/非常會唱歌
He hen/feichang hui chang ge He very can sing song
“He can sing very well.”
b. *他很/非常唱歌
He hen/feichang chang ge He very sing song
“He sings very well”
(12) a. 他很/非常能講故事
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
Ta hen/feichang neng jiang gushi He very be able to tell story
“He is capable of telling the story very much.”
b.*他很/非常講故事
Ta henfeichang jiang gushi.
He very tell story “He tells the story very well.”
modal verbs and modal adverbs
Some linguists claim that modal verbs can precede full verbs like adverbs because
those function words that can only act as adverbial modifier are adverbs. Thus,
“ModV + V” construction is analyzed as the modifier-head structure. Actually, the
following characteristics show that both of them have different syntactic properties:
I. Modal verbs can be directly negated by the negative adverbs bu ‘not’, fei
‘improperly’, while adverbs cannot. Bu can only occur before neng as in (13b),
not before dangran as in (13a), because neng is a modal verb, dangran ‘of course’
is an adverb.
(13) a. *我們不當然能告訴他這個事實
Women bu dangran neng gaosu ta zheige shishi We not certainly can tell he the truth
“We certainly cannot tell him the truth.”
b. 我們當然不能告訴他這個事實
‧
“We certainly cannot tell him the truth.”
II. Modal verbs can appear alone as a reply to questions, while adverbs cannot.
(14) a. — 他會說英文嗎?
To sum up the above analysis, we can formulate four criteria for generating the
category of Chinese modal verbs1:
1 Here we claim that the candidates of Chinese modal verbs do not have to exhibit all of these four
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
I. Modal verbs must take predicate object.
II. Modal verbs cannot be reduplicated.
III. Modal verbs cannot be suffixed by -le, -zhe, -guo aspects.
IV. Modal verbs can be modified by the degree of adverbs, such as hen ‘quite’, geng
‘even more’, feichang ‘extremely’, and they can also be negated by negative
adverbs, such as bu, fei.
We contend that Chinese modal verbs are subclass of common verbs and they
have their own syntactic characteristics, which are different from common verbs and
modal adverbs. It is helpful for us to have a deeper understanding of criteria for
differentiating these categories in Chinese modal system. Thus, according to the
above discussion we can list typical Chinese modal verbs as follows: hui 會, neng 能,
yao 要, gan 敢, ken 肯, xiang 想, dei 得, yinggai 應該, keyi 可以, nenggou 能夠, yuanyi
願意, keneng 可能, bixu 必須. In the present study, we will follow Wu’s (2009) study
to focus on polysemous modal verbs, hui 會, neng 能, yao 要, yinggai 應該, keyi 可以,
for investigating their modal interpretation and co-occurrence.
2.1.2 Generative Approach on Chinese Modals
As for the discussion on syntactic analyses, two types of categories are identified:
lexical category and functional category. Lexical categories provide the descriptive
criteria. The more characteristics the Chinese modals confirm to these criteria, the more they belong to typical modal verbs; otherwise, they belong to non-typical modal verbs.
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
content and the basic argumental structure, whereas functional categories determine
the grammatical specifications such as Tense and modality. They contribute to the
determination of the interpretation and the form of linguistic expressions. In what
follows we will specially focus on the previous Chinese modal literatures relating to
these two analyses and then contend why both of them cannot fully explicate the
nature of Chinese modal system.
In terms of the lexical analysis of Chinese modals, Lin and Tang (1995) argue that
modals in Chinese can be analyzed as raising and control verbs, because modals can
occur in the sentence-initial and sentence-final position. See the following examples
they indicate:
(16) [IP 張三i 不 [VP 敢 [CP[IPPROi 睡覺 ]]]]
Zhangsan bu gan shuijiao Zhangsan not dare sleep “Zhangsan dare not sleep.”
(17) ‘You shouldn’t do it this way.’
a. [IPNeg[VPMODAL[CP[IP … ]]]]
不 應該 你 這樣 做 Bu yinggai ni zheyang zuo not should you this way do
b. [IP[CP[IP你 這樣 做] i ] 不 [VP應該 ti ]]]
ni zheyang zuo bu yanggai you this way do not should
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
c. [IP你i不 [VP應該 [CP[IP ti 這樣做 ]]]]
ni bu yanggai zheyangzuo you not should this way do
Lin and Tang (1995)
Under the derivational framework, they propose the structure in (16) for the control
type modals and (17a) for the raising type modal. Both types take a CP as the
complement. When the complement raises, sentence (17b) with modals in the final
position is derived while sentence (17c) in which the subject raises to the IP-Spec
position is yielded if the CP is infinitive. According to such derivation, Lin and Tang
(1995) capture the possibilities with modals in the sentence initial position, sentence
final position and the position between the subject and the predicate. So it is predicted
that all modals can be in the sentence initial and final positions since modals are
lexical verbs that take a CP as the complement.
However, this expectation cannot be a generalization of Chinese modals as Hsu
(2004) argued in his study. First, although some modals may occur either in sentence
initial or sentence final position in daily conversation, not all modals can behave in
this way (*hui mingtian xiayu ma? ‘Will it rain tommorow?’). Such word order is also
regarded as marginal by many native speakers, e.g. % keyi Zhangsan bu qu shangke
‘Zhangsan is not required to go to school’. Moreover, Lin and Tang (1995) also argue
that their lexical category analysis is mainly based on the proposition that it is quite
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
uncommon for an INFL element to occur in sentence final position. Nevertheless, in
English, the modal auxiliary will as an INFL element is possible to occur sentence
finally, e.g. ‘They said he may tend to leave and tend to leave he will’. Also given by
their assumption, all modals can occur sentence finally; however, only YINGGAI
‘should’ and KEYI ‘may’ are allowed to occur in sentence final position even they are
marginal and restricted in A-not-A form while other modals do not. Such analysis
overgeneralizes the marginal possibilities to every Chinese modal and the discussion
above all doubt on Lin and Tang’s (1995) argument.
On the other hand, instead of treating all epistemic modals as raising verbs and all
deontic modals as control verbs, Lin and Tang (1995) indicated that not all deontic
modals are control verbs so that they suggest a more detailed division for deontic
modals. They argue that speaker-oriented deontic modals such as yinggai ‘should’ and
keyi ‘may’ should accompany with epistemic modals and be separated from other
subject-oriented modals such as neng ‘can’ and hui ‘be able to’. It is because these
speaker-oriented deontic modals do not impose selectional restriction on subjects as
other subject-oriented modals do. Thus, they claim that (18b) involves a raising-type
deontic whereas (18c) involves a control-type deontic.
(18) a. [他i [可能/應該 [ti吃過飯了]]] (Raising type epistemic modals) [tai[ keneng/yinggai [ti
chi guo fan le]]]
he may/should eat-ASP rice ASP
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
‘He may/should have resigned.
b. [書i [可能/應該 [ti 漲價了]]] (Raising type deontic modals) [shui[keneng/yinggai [ti zhangjia le]]]
book may/should raise-price ASP
‘The price of books/the book may/should have been raised.’
c. [他i [敢/願意/能 [PROi 吃飯]]] (Control type deontic modals) [tai [gan/yuanyi/neng [ PROi chi fan]]]
he dare/will/can eat rice
‘He dare/will/can eat rice.
(Lin & Tang 1995: 72)
However, Hsu (2004) argued that lexical verb analysis is not the only way to account
for the selectional properties of dynamic modals in (18c). Analyzing modals as
functional categories can also be compatible with such selectional restrictions. For
example, the English modal auxiliary can has the same property as shown in (19):
(19) *The door can break.
Sentence (19) is only tenable when the door uses his “leg” to break something but it is
not a common case. In other words, modals in English also pose selectional
restrictions on their subjects even if they are treated as functional categories.
Therefore, this kind of thematic property does not have to be a lexical property as
argued by Lin and Tang (1995). To sum up, the analysis of modals as raising and
‧
control verbs leads to the wrong prediction of Chinese modal verb distribution. Hence,
the discussion above shows that the analysis of modals as raising and control verbs
does not hold.
In terms of functional analysis of Chinese modals, on the other hand, Hsu (2004)
proposed that Chinese modals should be divided into three-level modality layers
reflected in the clausal hierarchy with the help of facts from preverbal adverb, the
cleft shi, the particle suo, the Aspect system in Chinese, and the non-finite clauses and
passivization2. Thus, there are projections of modals split in the INFL domain as
shown in (20):
(20) The tree structure of Chinese modal construction TP
Huang (2009) also has similar assumption that Chinese modal system only has
2 See Hsu (2004: 40-58) for more detailed discussion of these evidences in favor of split INFL hypothesis.
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
two categories: modal auxiliaries and modal adverbs. They are all directly merged in
distinct functional projections in syntactic structure and there exists a rigid hierarchy
among modals, thus accounting for the sequencing constraints on multiple-modal
occurrence. This approach though seems to provide a resolution to the wrong
prediction raised by Lin and Tang (1995) and better account for the order restriction of
these three modal expressions, some problems still remain. Both studies only
hypothesize the existence of such functional projection and define the syntactic
position of modal auxiliaries depending on the interaction of other syntactic elements
such as different layers of modal adverbs and aspect marker le2 (Huang, 2009: 45-56).
Nevertheless, they do not explain how this hierarchy is motivated and what enables
the rigid Epistemic>Deontic>Dynamic sequence. Such hierarchy also cannot predict
some semantically unacceptable modal co-occurrences. Consider the following
examples:
(21) ‘You will be able to earn much money.’
a. 你 會 能 賺大錢
ni hui neng zhuan da-qian you will can earn much-money
b. ?你 能 會 賺大錢
ni neng hui zhuan da-qian you can able to earn much-money
‧
hui and neng are polysemous modal verbs and they both possess deontic and
dynamic expression. In (21a) hui and neng are interpreted as deontic and dynamic
expression respectively due to prediction of such hierarchy and occupied on their
defined position as in (20). However, the same situation in (21b) results in
semantically unacceptable modal co-occurrence for most people. Huang (2009: 79-81)
also admits his analysis wrongly predicts the multiple-construction of obligation
modals bixu/yinggai-yao and bixu/yinggai-dei3.
2.2 Semantics-Pragmatics Interface Approach
2.2.1 Descriptive-oriented Approach
Many semantic criteria have been proposed for the definition of modality. Due to
its complexity, it is hard to give a simple and clearly definable description.
Nevertheless, according to Hsieh (2005), it is generally accepted by Lyons (1977:
452) that Modality basically refers to the speaker’s point of view or mental attitude
toward propositions. It conveys the propositional judgments through interpretations
such as evaluative, probability, possibility and necessity. The following studies on
Modality are also based on Lyons’ definition. Palmer (2001) presents a more general
survey of modality as a typological category. He draws attention to the subjective
nature of modality, and thereby defines it as ‘the grammaticalization of speakers’
3 At the very beginning Huang adopted VP-fronting and VP-ellipsis tests in which these two can be only licensed by modal auxiliaries but not by modal adverbs. However, the result of which bixu/yinggai-yao passes the tests whereas bixu/yinggai-dei fails forced him to treat dei is a modal adverb, which is contradictory to general intuition.
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
attitudes and opinions’. Bybee and Fleischman (1995) also attempt to characterize
modal systems cross-linguistically. Their argument treated the concept of modality
modal systems cross-linguistically. Their argument treated the concept of modality