The Categorization of Dou (都) in Chinese: A Study from a Cross-linguistic Perspective

In: Bulletin of Chinese Linguistics
View More View Less
  • 1 Harbin Institute of Technology
  • 2 哈爾濱工業大學(深圳)人文與社會科學學院 yingzhang@hit.edu.cn

Compared with prototypical universal quantifiers in other languages of the world, dou in Mandarin Chinese presents more complicated semantic behaviors. One of the most disputed issues is what are the relations between dou expressing “universal quantification” (uq) and dou expressing “scalar trigger” (sca). First-hand data that comes from 40 languages demonstrates that Mandarin Chinese is the only language that employs the same form for “universal quantification” and “scalar trigger”. The empirical evidence strongly suggests that uq dou and sca dou are different, and the two functions uq and sca lack universal conceptual correlations. The special polysemous behavior of Mandarin dou, is proved to come from two language-specific reanalysis processes in dou’s diachronic development which also supports the two-dou claim. The study thus instantiates how a cross-linguistic perspective provides insights to explain long-standing language-particular issues. Besides, it is also argued that the cross-linguistic approach is promising in predicting if a future research is on a right track as it can steer us through overgeneralization and undergeneralization.

Abstract

Compared with prototypical universal quantifiers in other languages of the world, dou in Mandarin Chinese presents more complicated semantic behaviors. One of the most disputed issues is what are the relations between dou expressing “universal quantification” (uq) and dou expressing “scalar trigger” (sca). First-hand data that comes from 40 languages demonstrates that Mandarin Chinese is the only language that employs the same form for “universal quantification” and “scalar trigger”. The empirical evidence strongly suggests that uq dou and sca dou are different, and the two functions uq and sca lack universal conceptual correlations. The special polysemous behavior of Mandarin dou, is proved to come from two language-specific reanalysis processes in dou’s diachronic development which also supports the two-dou claim. The study thus instantiates how a cross-linguistic perspective provides insights to explain long-standing language-particular issues. Besides, it is also argued that the cross-linguistic approach is promising in predicting if a future research is on a right track as it can steer us through overgeneralization and undergeneralization.

1 Introduction

The adverb dou in Mandarin has long been an area of fascination due to its complex and distinctive syntactic and semantic properties. Until very recently, discussions on Mandarin dou still took center stage. One of the most disputed topics is how many dous there are in Chinese, and why and how the different meanings are subsumed under one grammatical morpheme in Chinese. Xiandai Hanyu Babaici (《現代漢語八百詞》), as in many other descriptive works, lists the following three uses of dou.
Universal Quantification

(1)他們是老師。

tāmen dōu shì lǎoshī

3pl dou cop teacher

‘They are all teachers.’

Scalar Trigger

(2)這麼簡單的問題,三歲小孩知道。

zhème jiǎndān dewèntísān suìxiǎohái dōuzhīdào

sosimple aux question three-year small child dou understand

‘So simple a question that even a three-year-old child can understand (it).’

Already

(3)老師走了,他才來。

lǎoshīdōu zǒu letā cáilái

teacher dou go pfv 3sg not until come

‘He did not come until the teacher had already gone.’

A considerable amount of research proposes that there is only one semantic characterization of dou (Lee 1986; Cheng 1995; Jiang 1998; Pan 2006). In these studies, it is argued that both the readings of “scalar trigger” and “already” could be regarded as extensions of the first meaning “universal quantification” in particular contexts.1

However, other studies reveal that the first use is more objective whereas the latter two uses contain subjective evaluations (Wang 1999, 2001; Jiang 2003; Zhang 2005). Therefore, it is argued in these studies that there are two subtypes of dou, in which dou 1 carries the function of “universal quantification” as illustrated by (1) and dou 2 denotes a more subjective meaning of “being abnormal or inappropriate” as shown by (2) and (3). As for the relation of these two uses of dou, some studies describe the “scalar trigger” use as a pragmatically reinforced subtype of the “universal quantification” use (Xiao 2005; Jiang 2011) while others hold an opposite view about this and argue that the “universal quantification” use is not embedded in its “scalar trigger” use (Shank 2004; Chen 2008).

Other studies compromise between the above two opinions, putting forward the claim that there is only one dou in Chinese, but it can be attributed to two types according to the syntactic/semantic differences (Jiang and Pan 2013; Li 2013; Shen 2015). The two subtypes of dou in these studies basically correspond to dou 1 and dou 2 as categorized by the second approach.2

In fact, if we exclusively rely on internal observations within a single language, it would be rather difficult to evaluate which view among the above-mentioned three fares the best. The first approach emphasizes the consistency among the different uses whereas the second emphasizes their differences. Although the third approach captures more behaviors of dou, it is largely a rearrangement of the ideas previously known, with no new evidence provided.

To solve this long-standing dilemma, we need to take recourse to external evidence—regularities gained through cross-linguistic comparison. If the different functions3 delivered by dou are demonstrated to reflect some universal conceptual connections (i.e. the functions are repeatedly expressed by one grammatical morpheme across different languages), it will suggest that at the level of human conceptualization, they are intrinsically interconnected and thus subsumable, thereby justifying the oneness of dou in Chinese. However, if the functions are proved to lack a universal correlation (the functions tend to be expressed by different grammatical morphemes in different languages), chances are high that different dous should be identified.

In this research, we will examine the conceptual correlations of “universal quantification” and “scalar trigger” by comparing the cross-linguistic data collected from 40 languages, and apply the insight obtained from cross-linguistic comparison to solving the long-standing categorization issue of dou in Chinese.

2 Three Related Functions

If we compare Chinese dou with its counterparts in other languages, one of the most striking differences is that dou, as a universal quantifier in Chinese, can also take over the function of “scalar trigger.” By contrast, the universal quantifiers in other languages cannot extend their functions to “scalar trigger.” The “scalar trigger” function in these languages is often delivered by another group of grammatical morphemes—“additive particles.” For instance, in the Zhuang language, the function of “scalar trigger” is expressed by the word hix, but hix is not the word used to express “universal quantification,” but a word that would be simply rendered in English as also—an expression for the function of “additive focus.” Therefore, we feel obliged to take the function of “additive focus” into consideration as well to prepare the ground for further cross-linguistic comparisons. In the following, the three related functions—“additive focus,” “scalar trigger” and “universal quantification”—will be explained in detail.

2.1 Additive Focus (add)

“Additive focus” is commonly expressed in English by means of too, also and as well’ (Gil 2004). It means to add a new circumstance to the antecedent, in order to show consistency. In nature, it is an existential operator.
(Jino: Tibeto-Burman; Sino-Tibetan)

(4) nə42 le55 ɑ, ŋɔ42 33 le33ɑ.

2sg go prt 1sg add go prt

‘If you go, I will go too.’

This could be simply expressed as ‘[[you]target1 go] ∧ [[I]target2 go]’ and there is a consistency shown in between the two target members. Basically, the additive focus does not affect the truth value of the original sentence, but it does bring a presupposition that at least one of the alternative values under consideration in a context satisfies the same predicate. This alternative value could be made explicit like “you” in sentence (4), but could also be implicit, that is to say, existing in the context or in the mutual knowledge held by the speaker and the addressee. Discussions on additive focus particles, like König (1991: 62) and Shank (2004: 31), mainly focus on this presupposition, but we would like to point out further that apart from the existential presupposition it invokes, additive focus also suggests similarity between the alternative value and the focus that the additive particle introduces. In fact, for the most typical additive particle in Chinese, ye, its function is most widely accepted as “showing similarity (類同)” (Ma 1982).

2.2 Scalar Trigger (sca)

The concept of “scalar trigger” is sometimes called “scalar additive focus” (König 1991: 66), which is seen as a special case of “additive focus.” In fact, “scalar trigger” also refers to an addition of focus, but the difference is that the individual that has been added is located at an extreme of a pragmatic scale. As argued by Fillmore, Kay, and O’Connor (1988) in the “scalar model,” one uses an operator to trigger the highest/lowest polarity so as to realize the pragmatic entailment. The higher on the scale the expression is located, the more informative it is, and thus it is easier to entail relevant expressions. For example:
(Lachi: Kra, Tai-Kadai)

(5) to55ɕo44 tei55 ljuŋ55   vuɑ35, li44 tjou55 sei31 31 ɕu55 ljo31.

past 1pl handicap very one clf character even know neg

‘It was quite a handicap in the past that we didn’t know even one word.’

By addressing the extreme situation, la³¹ triggers the extreme situation of the event of “not knowing a word,” which is much more informative than those located lower in the scalar model, like “not knowing two/three … words”. It is easy to see that this sentence is equal to “we knew none of the words” in terms of truth value. Also, the example sentence actually contains the negative polarity item (npi), which is one of the two types of what we called “scalar trigger” here. As suggested by Gast and van der Auwera (2011), different forms may be used to denote “scalar trigger” in affirmative contexts and in negative contexts, especially for languages in Europe. Therefore, another category that contains the positive polarity item (ppi) is also identified in this study. For instance:
(Dong: Kam-Tai; Tai-Kadai)

(6) lieen yɑoc nyɑc buh hɑemk.

even 1sg 2sg even ask

‘You ask even me.’

To lay some groundwork for further discussion, we would like to differentiate two subtypes of ppi. As is well known, the implied meaning of a sentence can be divided into the conventional implicatures and the conversational implicatures. As the former have been conventionalized in the grammaticalization process, it is not defeasible by adding an opposite context. The latter have not been conventionalized, and therefore the implicatures could be cancelled. The ppi in our study only refers to the Conventional ppi but not Conversational ppi.

2.3 Universal Quantification (uq)

The term “universal quantification” here is used in the same sense as generally accepted in formal semantics. It shows that a property/action is satisfied by every member of the quantified domain.
(Jing: Mon-Khmer, Austro-Asiatic)

(7) kuŋ2ɳɑu1 kuŋ 3 lɑːi6  joi2

everyone  all   come pfv

‘All have come.’

In the above example, kuŋ 3 marks the action lɑːi 6 as one undergone by all participants expressed by kuŋ 2 ɳɑu 1, reflected in the English translation of “all”; therefore, all the participants in the domain possess the same property of the predicate. As is also the case in the categorizing practice in formal semantics, there are two instantiations of universal quantification: those with the property of totality/maximality and those with the property of non-totality/non-maximality.4 The former includes expressions like English every/each, which allow no exception to the quantification of all the members. In contrast, the latter has more tolerance of exceptions, like Chinese dou, which does not always apply the exhaustivity quantification to the np (Wu and Tao 2014; Xu 2014).

3 Cross-linguistic Representation Patterns

In this section, we will first introduce the procedure and example sentences that we adopted to investigate the 40 languages and then present the different connecting patterns by analyzing the cross-linguistic data received.

3.1 Investigating Procedure and Language Data

The investigating procedure is basically onomasiological. We will start from the three concepts of “additive focus,” “scalar trigger” and “universal quantification,” and try to figure out their formal expressions in different languages. After this, the different correlation patterns of the concepts and the forms will be compared in order to see in what fashion these concepts are correlated.

For the three concepts that we have analyzed above, example sentences have been designed to help the interviewees to find the counterparts in their mother tongues. Most of the examples are extracted from instantiations of the concepts in reference grammars, and, as we will see, more than one example sentence is offered under each of the concepts so as to capture all the syntactic environments in which the concepts may occur.

Additive Focus

  1. i.If you go, I will go too.
  2. ii.My younger brother is also sick.

Although in English different words are employed to show the different syntactic levels of the focus, there is no such difference in most other languages in my data sample.

Scalar Trigger

  1. i.He is so poor that he doesn’t even have a place to sleep.
  2. ii. Even a small child knows this.
  3. iii.She gave him lots of things, even this/this too.

In light of the detailed canvass of scalar triggers provided in the previous section, we find that sentence i is an example of an npi polarity trigger while sentence ii is an instance of a ppi polarity trigger. As for sentence iii, the alternatives of “even this” and “this too” are meant to confirm whether the ppi polarity trigger is a conventional one or a conversational one. When the investigation proceeds to this item, a more detailed context is offered by the interviewer to further differentiate the two types of implicatures.

Universal Quantification

  1. i. Everyone has come.
  2. ii.They all need to work hard.
  3. iii.The playground is full of young people.

From sentence i to sentence iii, the restriction on the totality property of the quantified scope gets progressively looser. In this way, we are trying to provide all the possible environments for the maximality universal quantifiers and non-maximality universal quantifiers in different languages.

Based on the above-mentioned procedures and examples, we interviewed more than 40 interviewees about their mother tongues at ucla and blcu,5 and collected data from 40 languages that come from 10 language families. The 40 languages are comprised of 14 Indo-European languages, 8 Sino-Tibetan languages, 6 Niger-Congo languages, 3 Austronesian languages, 3 Turkic languages, 2 Tai-Kadai languages, 1 Austroasiatic language, and 1 Afro-Asiatic language as well as Korean and Japanese. Table 1 offers a breakdown of the sources of the language data.

T000001

3.2 Three Types of Cross-Linguistic Representation Patterns

Now let us look into how the 40 different languages use different forms to express the three concepts. There are mainly four types of representations found in the 40 languages we investigated.

Representation Pattern 1

The three concepts “additive focus,” “scalar trigger” and “universal quantification” are coded separately by three distinct grams in 17 languages in our data pool.

To facilitate further comparisons with other types, we can diagram this type of representation by using the blocks to show the allocation pattern as shown in Fig. 1:

Figure 1
Figure 1

Distributive Pattern of Type 1

Citation: Bulletin of Chinese Linguistics 10, 2 ( 2019) ; 10.1163/2405478X-01002005

The 17 Type 1 languages and their respective forms for the three concepts are shown in Table 2:

T000002

Apart from the 17 languages listed in Table 2, German also uses different lexical forms to code the different notions individually; however, it goes deeper for the concept of “scalar trigger” and tends to use different words to express the ppi and npi scalar triggers. For the former, selbst is used whereas for the latter, nicht einmal is employed. For the current purpose, we will categorize German into Type 1 as all these grams in German are mono-functional, as in the other Type 1 languages. Therefore, for Type 1, we have 18 languages altogether, which constitutes 45 percent of the data sample. If we zoom in to see the difference in distribution among the language families, we find that, for the Indo-European family and the Niger-Congo family, the Type 1 languages occupy exactly half of the sample from the family, namely 7 out of 14, and 3 out of 6, respectively. But for the Sino-Tibetan language family, from which 8 languages/dialects are sampled, only one language belongs to Type 1.

Representation Pattern 2

There are 18 languages in our sample that use additive focus particles to express the concept of “scalar trigger” (including both the ppi scalar trigger and npi scalar trigger).

The representation pattern is presented in Fig. 2:

Figure 2
Figure 2

Distributive Pattern of Type 2

Citation: Bulletin of Chinese Linguistics 10, 2 ( 2019) ; 10.1163/2405478X-01002005

The 18 languages and relevant data are shown in Table 3:

T000003

One language offers a slightly different picture from Pattern 2 – the Italian language. Fig. 3 shows how it differs from the typical Type 2 language.

Figure 3
Figure 3

Distributive Pattern of Italian

Citation: Bulletin of Chinese Linguistics 10, 2 ( 2019) ; 10.1163/2405478X-01002005

Italian is the same as the other languages listed in Table 3 in using a separate lexical item to code the concept of universal quantification. However, two different forms are employed in affirmative sentences and negative sentences respectively in expressing the concept of scalar trigger. As it also shows a strong connection between the notion of “additive focus” and that of “scalar trigger,” it is categorized into Pattern 2 as well.

If Italian is included, we have a total of 19 languages belonging to Type 2. These 19 languages constitute 47.5 percent of our language sample. In terms of different language families, for Indo-European languages and Niger-Congo languages, languages of Pattern 2 count for almost half of the sample from each language family, namely 6 out of 14, and 3 out of 6 respectively. If we compare the ratios of Type 1 and Type 2 in the two language families, the two categories are evenly divided in the sample languages. That is to say, for a language coming from the Indo-European family or the Niger-Congo family, there is approximately a 50 percent chance that it uses three distinct forms to denote the notions of “additive focus,” “scalar trigger” and “universal quantification”; and a 50 percent chance that it employs the additive focus particle to denote the concept of “scalar trigger,” or at least the “scalar trigger” in affirmative sentences, but a different form for “universal quantification.”

However, among the eight members of the Sino-Tibetan family, four members show Pattern 2. All of them are non-Mandarin Chinese varieties; while Mandarin Chinese demonstrates a unique representation pattern—Type 3.

Representation Pattern 3

There is only one example in our data pool that shows Pattern 3, namely the Southwestern Mandarin dialect spoken in Chongqing. In this dialect, the universal quantifier, which is identical in form to that in Standard Mandarin, can also be employed to express the function of “scalar trigger.”

Pattern 3 can be represented in one schema as in Fig. 4:

Figure 4
Figure 4

Distributive Pattern of Type 37

Citation: Bulletin of Chinese Linguistics 10, 2 ( 2019) ; 10.1163/2405478X-01002005

The most prominent feature of Type 3 is that an identical form is employed for the concepts of “scalar trigger” and “universal quantification,” whereas in the previous two types, the two concepts are expressed by distinct forms.

Apart from the two sub-dialects of Mandarin Chinese mentioned above, the Min dialect spoken in Taipei is another example of Type 3 but with a slight deviation. It is different in that only the npi scalar trigger can be expressed by the universal quantifier whereas the ppi scalar trigger shares the same form with the additive focus particle. The pattern can be illustrated as in Fig. 5.

Figure 5
Figure 5

Distributive Pattern of Min Dialect in Taipei

Citation: Bulletin of Chinese Linguistics 10, 2 ( 2019) ; 10.1163/2405478X-01002005

As the universal quantifier in this dialect does not exclusively express the function of “universal quantification” like those in the previous two categories, it can deemed as a special case of Type 3.

Representation Pattern 4

We now turn to the last possibility, which is to express all the concepts with one and the same multifunctional gram. The only member in our data sample that falls into this type is the Yue variety of Chinese (Cantonese) in the Sino-Tibetan family.8 Cantonese dou is different from Mandarin Chinese dou in that it can also express the meaning of “additive focus.” For example:

(8) Haih a, ngoh dou yauh heui.

is prt 1sg add have   go

‘That’s right, I went too!’ (Matthews and Yip 1994: 281)

Furthermore, the concept of “ppi scalar trigger,” “npi scalar trigger” and “universal quantification” can also be expressed by Cantonese dou, just like its counterpart in Mandarin Chinese.

ppi Scalar Trigger

(9)Gam yuhksyun dou yauh ge!

so ugly p.sca exist prt

‘What could be so ugly?’ (Matthews and Yip 1994: 283)

npi Scalar Trigger

(10)Yat lap seng dou mouh.

one clf sound n.sca not-have

‘There’s not even a sound.’ (Matthews and Yip 1994: 272)

Universal Quantification

(11)Leuhng ga dou leng.

Two clf uq nice

‘They’re both nice.’ (Matthews and Yip 1994: 263)

Therefore, the representation pattern for Type 4 is much more concise than the previous ones as is shown in Fig. 6.

Figure 6
Figure 6

Distributive Pattern of Type 4 (Cantonese)

Citation: Bulletin of Chinese Linguistics 10, 2 ( 2019) ; 10.1163/2405478X-01002005

We can put aside the differences inside the concept of scalar trigger for now and have a general picture of the four types of cross-linguistic representations of the concepts “additive focus,” “scalar trigger” and “universal quantification.” Table 4 summarizes the contrast. (The three special cases which involve the differentiation of ppi and npi are excluded for now.)

T000004

The above contrast clearly demonstrates that the universal quantifiers in Mandarin Chinese can take over one more function of “scalar trigger” and no other language behaves the same way. A fact that is worthwhile noting is that the situation in Mandarin Chinese is by no means representative of the Chinese language in this respect. Our investigation shows that many other regional varieties of Chinese do not take on the same polysemous behaviors as Mandarin dialects. For example, Wu dialects and Jin dialects use the same form for add and sca, but a separate item for uq. In Xiang dialects, the same finding has been reported. For example, Xiao (2005: 101) states that in Changsha (長沙), Changning (常寧), Loudi (婁底), Xinhua (新化) and Hengshan (衡山), the scalar use is also expressed by an additive focus particle rather than the universal quantifier. All these cross-linguistic data strongly suggest that the concepts of universal quantification and scalar trigger are not universally connected, and chances are high that the presentation pattern of Mandarin Chinese is due to some language-specific development rather than language-universal reasons.

4 Diachronic Evidence

In this section, we will take a cursory look at the diachronic development of the markers for uq and sca in Chinese, aiming to examine the existing diachronic research evidence to see whether the development of Mandarin Chinese dou is due to language-specific reasons or due to some language-universal reasons, i.e. whether the process is mere happenstance in a particular language, or follows certain regularities that can be replicated across various languages.

Although Mandarin Chinese dou is widely studied in the synchronic perspective, discussions on historical relations between uq dou and sca dou have barely been touched upon. Only Zhang (2005), to the best of our knowledge, has examined the historical data and provided detailed analysis on the relations between uq dou and sca dou from a diachronic perspective.9

In what follows, we briefly introduce how the sca dou derived from uq dou as discussed in Zhang (2005).

When dou performs the function of “universal quantification”, it demonstrates that a certain property/action is satisfied by every member of the quantified plural arguments. In certain historical periods of Chinese, the plural arguments that appear in the subject position could be linked by a conjunction derived from a verb, like the word lian (連) ‘plus, to connect.’ In this case, the whole structure takes the form of “NP1 lianNP2 dou VP”. Historical records show that it is from this structure that the scalar trigger dou was derived (Zhang 2005).

As suggested by Zhang (2005), after dou had developed the function of universal quantifier, the quantified components might take diversified forms. The “NP1 lianNP2 dou VP” is one of them, where “NP1 lianNP2” literally means “the first thing, plus the second.” (12), (13) and (14) are examples from classical novels written in Early Mandarin:

(12)面頰連項都有疤痕。(《水滸傳》 第69回)

miànjiá lián xiàng dōu yǒu   bāhén

face connect with neck  dou have  scar

‘There are scars all across the face and the neck.’

The verbal status of lian may still be retained in the above sentence, and dou acts to quantify the two arguments: “the face” and “the neck.” However, the two arguments do not carry the same weight semantically; that is to say, “the scars” exist primarily in “the face,” and reach “the neck” as a supplement location. Hu and Shi (2007: 76) also point out that “compared with NP1, the connected NP2 is mostly secondary to the speaker…. NP2 is often the unexpected or unconventional object to the speaker.” Therefore, NP2 is more informative than NP1. When NP1 and NP2 differ in the degree of informativeness, they are located at different points in the “scalar model,” which may easily trigger the pragmatic inference that we have discussed in the introductory part of the “scalar trigger” section. Therefore, due to the higher degree of informativeness provided by NP2, “lianNP2 dou VP” could entail “NP1 lianNP2 dou VP” from the perspective of pragmatic inference. Driven by the motivation of economy, the more concise form “lianNP2 dou VP” wins out over the complete form “NP1 lianNP2 dou VP”, giving rise to a new structure, i.e., “lianNP2 dou VP.” For instance:

(13) 連我腳上穿的鞋都不見了。(《金瓶梅》第28回)

lián    wǒ jiǎo shàng chuān de   xié  dōu  bú jiàn le

including  1sg foot on  wear  aux shoe dou neg see sfp

‘Even the shoes on my feet are not seen any more.’

With the frequent use of the structure “lianNP2 dou VP,” the pragmatic mechanism of labeling the most unexpected object to trigger the scalar inference is absorbed by dou. Therefore, the structure “NP2 dou VP” could independently express this inference with or without lian. As a result, the surface structure is further condensed by omitting lian, and at this point the scalar trigger dou comes into being.10

(14) 纔知道鞋子都跌掉了。(《醒世姻緣》第3回)

cái   zhīdào   xiézi dōu diē diào le

only then knowshoe dou fall off sfp

‘Only then did (I) know that even (my) shoes had fallen off.’

From the above diachronic process, we find that there were two reanalysis operations in the development from uq dou to sca dou. During the reanalysis processes, the surface structure has been changed twice. In the first change, NP1 was omitted and lian has undergone a grammaticalization from a conjunction to a preposition; in the second change, lian is omitted and the scalar meaning was shifted from the whole structure liandou to the individual word dou.

The above-mentioned fine-grained diachronic analysis has widely been accepted in Chinese linguistics. Besides, as it is an independent study from our research, it serves as neutral evidence for us to conclude that the development from uq dou to sca dou can only be a language-specific phenomenon, because we cannot expect other languages to perform the same two omitting operations exactly as that of Chinese.

In fact, it is a prevalent fact in the field of typology that in most languages of the world, the scalar additive particles are mostly derived from the additive particles (König 1991, Gil 2004). So far, we have not found any reports in other languages which claim a universal quantifier has developed into an additive scalar function.

To further demonstrate why this language-specific change from uq dou to sca dou proves the two-dou claim, a reverse example explaining how language-universal change proves conceptual correlation is employed in the following discussion. Consider the example of English have to as shown in example (15).

(15)A: Ingrid has to arrive before we do (because she promised to help with the preparations).

B: Ingrid has to arrive before we do (because she set out an hour earlier). (Bybee and Pagliuca 1985: 74)

As is shown, have to can sometimes be ambiguous. In sentence (15A), it produces an obligation reading which is known as “deontic” modality; while in sentence (15B), it is compatible with an epistemic reading and known as “epistemic” modality. Bybee and Pagliuca (1985: 73) examined the historical data and argued that “the epistemic sense is a metaphorical extension of obligation to apply to the truth of a proposition: X (a proposition) is obliged to be true.”

“Metaphorical extension,” as stated above in the semantic change from deontic have to to epistemic have to, is usually deemed as a conceptual phenomenon, and thus often termed as “conceptual metaphorization” (Traugott and Dasher 2001: 29). If we compare the conceptual metaphorization process of English have to with the reanalysis process of Chinese dou, two major differences can be detected. (1) The metaphorization process of have to is at the conceptual level whereas the reanalysis process of dou entails operations on the surface structure. (2) The reanalysis of dou involves processes of loss, while the grammaticalization of have to is mainly an extension from one conceptual domain to another conceptual domain without formal loss. If we put these two semantic change processes into a cross-linguistic canvass, it is easy to infer that the metaphorization process, like that of have to, is more likely to be a cross-linguistic phenomenon; while the reanalysis process, like that of dou, is less likely to be observed across different languages. The reason is that different languages are comparable at the conceptual level but not at the structural level. Besides, as “loss is unpredictable and irregular” (Traugott and Dasher 2001: 39), the reanalysis process, such as that of dou which involves two operations of structural loss, is hard to be replicated across languages.

In fact, that metaphorization works as a language-universal grammaticalization mechanism has long been verified. By reviewing the theoretical issues, Traugott and Dasher (2001: 75) stated that research on metaphor, or the metaphorization process, actually starts from cross-linguistic studies, like Greenberg, Ferguson and Moravcsik (1978). So far, we have analyzed that metaphorization can be cross-linguistic. In the following, we will employ historical syntax and grammaticalization studies across different languages on the field of modality to verify that a metaphorical process, such as that of have to, is indeed a cross-linguistic phenomenon. Throughout the recorded history of Chinese, we find lots of language forms similar to English have to which can express both the meaning of obligation and epistemic possibility. Sun (1996) and Yin (2014) examined the history of de (得) and dang (當), and concluded that both of these two forms developed their epistemic meaning from the deontic meaning. Yin (2014: 204) also analyzed how metaphor plays a part in the grammaticalization process. Apart from English and Chinese, many language forms in other languages can also perform both obligation and epistemic possibility. By examining 25 languages in the world, Bybee and Pagliuca (1985) generalized a cross-linguistic tendency that the obligation sense precedes the epistemic sense, and it is also clearly demonstrated that metaphorical extension is one of the important mechanisms in generalization.

The semantic change from “deontic” to “epistemic” recurs so often and across totally unrelated languages that such regularity is by no means an accident; it proves that deontic and epistemic are intrinsically bound, i.e. deontic and epistemic have universal conceptual correlations. Typological studies at the synchronic level have also proved this (Van der Auwera and Plungian 1998). This kind of typological finding can provide insights on many issues, including those that cannot be easily solved by a language-specific perspective. In this sense, if we are required to evaluate the relation between deontic dang and epistemic dang in Chinese, the answer is evident that they should be subsumed under one category.

By the above contrast between Chinese dou and English have to, we have illustrated how the diachronic semantic change of dou confirms the language-specific property of the connections between uq dou and sca dou, which again verifies the synchronic conclusion of the two-dou claim.

So far, the synchronic analysis, the cross-linguistic comparison and the diachronic development of dou all point to the same conclusion: the notions uq and sca are not universally connected. The behavior of Mandarin Chinese dou in employing the same form to denote these two notions is proved to be language-specific, which arises from an unusual reanalysis in the development. Therefore, uq dou and sca dou should be deemed as two different representations.

5 The Categorization of Dou

The reason why the categorization of dou has caused much controversy in Chinese linguistics is that the uq function and the sca function are a pair of conceptually different but pragmatically related functions. In the above we have proved from various perspectives that these two functions lack a universal connection; therefore, uq dou should be recognized as distinct from sca dou. In the following, we will present why the opposite one-dou-claim is problematic.

In the literature on dou, pragmatic theories are often evoked to provide a solution for inferring a uq reading from the sca use. Jiang (1998, 2009, 2011) and Xiang (2008) have illustrated from the perspectives of “scalar model” and “pragmatic scale” that scalar dou induces an ordered sequence of a set. By pointing to the member that is located at the end point of this scale, the dou sentence entails that the members that are located higher (or lower depending on the specific context) on the scale will all have the property as the scalar member that dou marks. That is how it connects to the uq reading. To illustrate this point, Xiang (2008: 227) provides the following example:

(16)連傻瓜都知道這個。

Lián shǎguā dōu zhīdào zhègè.

even idiot dou know  this

‘Even idiots know this.’

Xiang indicates that “speakers can accent on the word idiot to emphasize on the intended message that everybody knows how to do it.” (ibid) However, in our opinion the scalar dou sentence does not intend to emphasize the universal quantification reading that “everybody” knows something. Instead, it is used to emphasize a specific member that is higher on this scale (most likely the reference is the listener in the conversation), in comparison to the scalar member—the “idiot,” is more likely to know this. A more explicit example that supports our account is from Xiao (2005: 99), which is also cited in Jiang (2011: 452).

(17) A:你知道活字印刷的發明者是誰嗎?

nǐ zhīdào huó   zì yìnshuā de fāmíngzhě  shì shéi ma

2sg know movable type printing  aux inventor    cop who sfp

‘Do you know who is the inventor of the movable type printing?’

B:小學生都知道。

xiǎoxuéshēng dōu zhīdào

pupil even know

‘Even a pupil knows this.’

(18)小學生都知道(活字印刷的發明者是誰),

xiǎoxuéshēng dōu zhīdào (huó  zì   yìnshuā de fāmíngzhě shì shéi)

pupil    even know  movable type printing aux inventor cop who

‘Even a pupil knows who is the inventor of the movable type printing.’

(我知道的比小學生知道的多),

(wǒ zhīdào de bǐ xiǎoxuéshēng zhīdào de duō)

1sg know aux cm pupil   know aux more

‘What I know is more than what a pupil knows,’

(所以,我當然知道活字印刷的發明者是誰)。

(suǒyǐ, wǒ  dāngrán zhīdào huó  zì yìnshuā de fāmíngzhě shì shéi)

therefore 1sg  sure  know movable type printing aux inventor cop who

‘Therefore, of course I know who is the inventor of the movable type printing.’

The above instance clearly demonstrates that there is a syllogistic chain which lets us attain the pragmatic inference of dou. In this syllogistic chain, “pupil” is set as an extreme member of the scale, from which the property of another particular member on the scale (the reference is “I” in the above utterance) is deduced. Therefore, it is not the whole complementary set of {pupil} that is focused; instead, it is a particular intended member in the complementary set that is in focus. In this sense, we can find that the sca dou is still an existential operator in nature, and it is different from the uq dou. Therefore, we do not think it suitable to regard sca as a pragmatically reinforced subtype of uq.

Based on the evidence from synchronic data and a diachronic survey, we have demonstrated that uq and sca lack a universal conceptual correlation and the semantic derivation of dou from uq to sca is due to a particular reanalysis process. Therefore, no direct conceptual correlation exists between dou that expresses the uq function and dou that expresses the sca function. Besides, we have also demonstrated from the opposite angle that it is inappropriate to see sca as a special case of uq as the focus of the sca expression is a particular member on the scale rather than the whole set. In these two approaches, we prove that the uq dou and the sca dou should be regarded as two distinct dous in Chinese.

6 Discussion

In this research, we have provided an account for the long-standing puzzle about how many dous there are in Mandarin Chinese and how the different functions of dou are conceptualized in the same language. Based on the first-hand data from 40 languages, we show that employing the same form for uq and sca is an exclusive phenomenon in Mandarin Chinese. No other language or dialect exhibits the same behavior. The empirical data strongly suggests that uq dou and sca dou are different. This claim is further supported by the diachronic development of dou, from which we argue that dou absorbs the scalar semantics through two reanalysis processes and this change is highly language-specific rather than language-universal.

The process of identifying the uniqueness of Mandarin dou has also demonstrated the advantage of a cross-linguistic comparison. In fact, only by fully understanding universality shared by different languages can one best recognize the particularity in specific language. The cross-linguistic comparison is just one of the best candidates to cater for this goal. Besides, this perspective can also help us evaluate existing proposals and even make predictions for future work. From the cross-linguistic perspective, we can tell whether or not a generalization is at the right level of generality. As pointed out by Haspelmath (1997: 7), “cross-linguistic studies can help steer us through the Scylla of overgeneralization and the Charybdis of undergeneralization.” Sometimes one is prone to explain a language-particular phenomenon by a high-level generalization. Alternatively, one may erroneously invoke a low-level generalization to account for a language-universal phenomenon. With a cross-linguistic perspective, however, we can avoid this kind of wrong prediction. If the cross-linguistic data converge to show a common regularity, we may seek to account for it based on universal factors; on the other hand, if a phenomenon is only found in an individual language, then a particular explanation is called for (ibid). In this sense, the cross-linguistic exploration on dou as shown above can also predict if on-going research is on the right track. For instance, an account at the language-specific level is believed to be the more fruitful direction than its counterpart for research inquiries that involve the relation of uq dou and sca dou.

Apart from the uniqueness of Mandarin Chinese dou, Cantonese dou is also identified to present very rare semantic behaviors. Cantonese dou can not only deliver the functions of uq and sca by one form, but also it can express the function of add as well. Naturally, we may raise a question about the origin of this add function. Does it reflect a natural semantic extension, or is it merely a stand-in for the vanished ye in Cantonese? Szabolcsi, Whang and Zu (2014: 148) pose a similar query: “We do not know whether the gap in Mandarin is due to the blocking effect of ye ‘also, even’ or, conversely, dou spills over to fill a vacant spot in the absence of ye in Cantonese.” However, no further discussion is offered in their research. Peng and Chen (2010: 109) argue that Cantonese dou expressing additive focus is a rather recent phenomenon, and it is derived from the universal quantification use of dou. However, this derivation is admitted by the authors to be uncommon, as it deviates from the widely accepted mechanisms of grammaticalization. Apart from Cantonese dou, we have also found a parallel circumstance in a reference grammar of the Lachi language (Li 2000), where it is shown that la 31 can also express the three concepts all by itself. However, evidence is insufficient for us to tell whether this circumstance is due to conceptual correlation or simply a historical incident induced by non-conceptual factors. We believe this line of inquiry deserves further pursuit.

Abbreviations

1

first person

2

second person

3

third person

add

additive focus

aux

auxiliary

clf

classifier

cm

comparative marker

cop

copula

n.sca

npi scalar trigger

neg

negative

npi

negative polarity item

p.sca

ppi scalar trigger

pfv

perfective

pl

plural

ppi

positive polarity item

prt

particle

sca

scalar trigger

sfp

sentence final particle

sg

singular

uq

universal quantification

I would like to express my sincere gratitude to Professor Min Zhang, who patiently supervised me in this project. I also thank two anonymous reviewers for their thoughtful comments and suggestions on earlier versions of it. Of course, all the remaining errors are mine. The work has been supported by Hong Kong Research Grants Council (Project No. 641913).

References

  • Bybee Joan L. , and Pagliuca William . 1985. Cross-linguistic comparison and the development of grammatical meaning. Historical Semantics and Historical Word Formation, ed. by Fisiak Jacek , 5983. Berlin: Mouton.

    • Search Google Scholar
    • Export Citation
  • Chen Liping . 2008. Dou: Distributivity and Beyond. Ph.D. dissertation, the State University of New Jersey.

  • Cheng Lisa Lai-shen . 1995. On dou-quantification. Journal of East Asian Linguistics 4: 197234.

  • C.J. Fillmore, , Kay P. , and O’Connor M.C. . 1988. Regularity and idiomaticity in grammatical constructions: The case of let alone . Language 64: 501538.

    • Search Google Scholar
    • Export Citation
  • Gast Volker and van der Auwera Johan . 2011. Scalar additive operators in the languages of Europe. Language 87.1: 254.

  • Ge Jiacai . 葛佳才. 2005.《東漢副詞系統研究》。長沙:岳麓書社.

  • Gil David . 2004. Riau Indonesian sama: Exploration in macro functionality. Coordinating Constructions, ed. by Haspelmath Martin , 371–426. Amsterdam/Philadelphia: Benjamins.

    • Search Google Scholar
    • Export Citation
  • Greenberg Joseph H. , Ferguson Charles A. and Moravcsik Edith . (eds.) 1978. Universals of Human Language. Stanford: Stanford University Press.

    • Search Google Scholar
    • Export Citation
  • Gu Feng . 谷峰. 2015. 〈“都”在東漢有沒有語氣副詞的用法?〉,《中國語文3: 230239.

  • Haspelmath Martin . 1997. Indefinite Pronouns. Oxford: Oxford University Press.

  • Hu Xiaoping and Shi Jinsheng . 胡曉萍 史金生. 2007. 〈連類介詞的語法化〉,《語法化與語法研究》(三):6585.

    • Search Google Scholar
    • Export Citation
  • Jiang Jing . 蔣靜. 2003. 〈“都”總括全量手段的演變及其分類〉,《漢語學習4: 7276.

  • Jiang Yan . 蔣嚴. 1998. 〈語用推理與“都”的句法語義特徵〉,《現代外語1: 1024.

  • Jiang Yan . 蔣嚴. 2009. 〈梯級模型與“都”的語義刻畫〉,程工、劉丹青主編《漢語的形式與功能研究》123132北京:商務印書館.

  • Jiang Yan . 蔣嚴. 2011.〈“都”的形式語用學〉,蔣嚴主編《走進形式語用學》:432456。上海:上海教育出版社.

  • Jiang Jingzhong and Haihua Pan . 蔣靜忠 潘海華. 2013. 〈都的語義分合及解釋規則〉,《中國語文1: 3850.

  • König Ekkehard . 1991. The Meaning of Focus Particles. London/New York: Routledge.

  • Lee Thomas Hun-tak . 1986. Studies on Quantification in Chinese. Ph.D. dissertation, University of California, Los Angeles.

  • Li Wenshan . 李文山. 2013.〈也論“都”的語義複雜性及其統一刻畫〉,《世界漢語教學3: 319330.

  • Li Yunbing . 李雲兵. 2000.《拉基語研究》。北京:中央民族大學出版社.

  • Ma Zhen . 馬真. 1982.〈說“也” 〉,《中國語文4:283287.

  • Matthews S. and Yip V. . 1994. Cantonese: A Comprehensive Grammar. London and New York: Routledge.

  • Pan Haihua . 潘海華. 2006. 〈焦點、三分結構與漢語“都”的語義解釋〉,《語法研究與探索》(十三)。北京:商務印書館.

  • Peng Xiaochuan and Chen Qiping . 彭小川 陳啟萍. 2010. 〈廣州話“類同義”副詞歷時演變的特點與機制〉,《暨南學報(哲學社會科學版)3: 105109.

    • Search Google Scholar
    • Export Citation
  • Peyraube Alain . 2006. On the history of some adverbs of scope and quantity in Chinese, Studies in Chinese Language and Culture: Festschrift in Honor of Christoph Harbsmeier on the Occasion of His 60th Birthday, eds. by Anderl Christoph and Eifring Halvor , 269283. Oslo: Hermes Academic Publishing.

    • Search Google Scholar
    • Export Citation
  • Shank Scott . 2004. Domain Widening. Ph.D. dissertation, Vancouver: The University of British Columbia.

  • Shen Jiaxuan . 沈家煊. 2015.〈走出都的量化迷途:向右不向左〉,《中國語文1: 317.

  • Sun Chaofen . 1996. Word Order Changes and Grammaticalization in the History of Chinese. Stanford: Stanford University Press.

  • Szabolcsi A. , Whang J.D. , and Zu V. . 2014. Quantifier words and their multi-functional parts. Language and Linguistics 15.1: 115155.

    • Search Google Scholar
    • Export Citation
  • Traugott Elizabeth C. and Dasher Richard B. . 2001. Regularity in Semantic Change. Cambridge: Cambridge University Press.

  • van der Auwera Johan and Plungian Vladimir A. . 1998. Modality’s semantic map. Linguistic Typology 2.1: 79124.

  • Wang Hong . 王紅. 1999.〈副詞“都”的語法意義試析〉,《漢語學習6: 5560.

  • Wang Hong . 王紅. 2001. 〈語氣副詞“都”的語義、語用分析〉,《暨南大學華文學院學報2: 4145.

  • Wu Haiping and Tao Hongyin . 2014. Doing generalizations in Mandarin conversation: The case of plural NP+dou expression. Paper presented at the 2nd Annual Conference of the American Pragmatics Association, October 17–19, la.

    • Search Google Scholar
    • Export Citation
  • Xiang Ming . 2008. Plurality, maximality and scalar inference: a case study of Mandarin dou . Journal of East Asian Linguistics 17.3: 227245.

    • Search Google Scholar
    • Export Citation
  • Xiao Shuangrong . 肖雙榮. 2005.〈“都”的全稱量項義與隱含差比義〉,《湖南人文科技學院學報6: 98104.

    • Search Google Scholar
    • Export Citation
  • Xu Liejiong . 徐烈炯. 2014.〈“都”是全稱量詞嗎?〉,《中國語文6: 498575.

  • Yang Rongxiang . 楊榮祥. 1998. 〈總括副詞“都”的歷史演變〉,《北大中文研究1:311332.

  • Yin Chunyi . 尹淳一. 2014. 〈“祖堂集”情態動詞及其語法化研究〉,復旦大學博士論文.

  • Zhang Yisheng . 張誼生. 2005. 〈副詞“都”的語法化與主觀化──兼論“都”的表達功用和內部分類〉,《徐州師範大學學報(哲學社會科學版)1: 5662.

    • Search Google Scholar
    • Export Citation
1

Jiang (1998) applies Relevance Theory to prove that the “scalar” meaning and the “already” meaning are produced by the whole structure that contains dou rather than dou itself. What dou performs in these two contexts could both be subsumed into the first meaning “universal quantification.”

2

Jiang and Pan (2013) further differentiate two subcategories of dou 1 and dou 2 respectively according to whether there is an “exclusive” reading attached.

3

As demonstrated by reviewing the previous studies on the categorizing issue of dou, the core of the debate is whether the non-scalar use of dou (“universal quantification”) and the scalar uses of dou (“scalar trigger” and “already”) should be ­differentiated or not. In this paper we will mainly address the relations between “universal quantification” and “scalar trigger.” The function of “already” will be set aside in this paper with no harm.

4

There are of course other ways to convey universal quantification, such as reduplicated classifiers in Mandarin like 人人renren, ‘everyone.’ Such expressions are irrelevant to our theme and will be left unaddressed in this study.

5

ucla and blcu stand for “University of California, Los Angeles” and “Beijing Language and Culture University” respectively.

6

The original characters of the three grams in Min (Yun’ao) are not clear.

7

Ye in Mandarin Chinese can also express the npi scalar trigger. The overlapping pattern of dou/ye in this function is worth pursuing in a separate study.

8

Second-hand data from the reference grammar of the Lachi Language has also provided data which shows the pattern of Type 3 (Li 2000)

9

There are of course a lot more diachronic studies on Chinese dou, as discussed in Yang (1998), Ge (2005), Peyraube (2006) and Gu (2015). However, these pieces of research show how the uq dou came into being, and what is the relation between dou as a universal quantifier and dou as modal adverb expressing the meaning of “completely,” which we can harmlessly ignore here.

10

For a more detailed discussion, the reader is referred to Zhang (2005).

If the inline PDF is not rendering correctly, you can download the PDF file here.

  • Bybee Joan L. , and Pagliuca William . 1985. Cross-linguistic comparison and the development of grammatical meaning. Historical Semantics and Historical Word Formation, ed. by Fisiak Jacek , 5983. Berlin: Mouton.

    • Search Google Scholar
    • Export Citation
  • Chen Liping . 2008. Dou: Distributivity and Beyond. Ph.D. dissertation, the State University of New Jersey.

  • Cheng Lisa Lai-shen . 1995. On dou-quantification. Journal of East Asian Linguistics 4: 197234.

  • C.J. Fillmore, , Kay P. , and O’Connor M.C. . 1988. Regularity and idiomaticity in grammatical constructions: The case of let alone . Language 64: 501538.

    • Search Google Scholar
    • Export Citation
  • Gast Volker and van der Auwera Johan . 2011. Scalar additive operators in the languages of Europe. Language 87.1: 254.

  • Ge Jiacai . 葛佳才. 2005.《東漢副詞系統研究》。長沙:岳麓書社.

  • Gil David . 2004. Riau Indonesian sama: Exploration in macro functionality. Coordinating Constructions, ed. by Haspelmath Martin , 371–426. Amsterdam/Philadelphia: Benjamins.

    • Search Google Scholar
    • Export Citation
  • Greenberg Joseph H. , Ferguson Charles A. and Moravcsik Edith . (eds.) 1978. Universals of Human Language. Stanford: Stanford University Press.

    • Search Google Scholar
    • Export Citation
  • Gu Feng . 谷峰. 2015. 〈“都”在東漢有沒有語氣副詞的用法?〉,《中國語文3: 230239.

  • Haspelmath Martin . 1997. Indefinite Pronouns. Oxford: Oxford University Press.

  • Hu Xiaoping and Shi Jinsheng . 胡曉萍 史金生. 2007. 〈連類介詞的語法化〉,《語法化與語法研究》(三):6585.

    • Search Google Scholar
    • Export Citation
  • Jiang Jing . 蔣靜. 2003. 〈“都”總括全量手段的演變及其分類〉,《漢語學習4: 7276.

  • Jiang Yan . 蔣嚴. 1998. 〈語用推理與“都”的句法語義特徵〉,《現代外語1: 1024.

  • Jiang Yan . 蔣嚴. 2009. 〈梯級模型與“都”的語義刻畫〉,程工、劉丹青主編《漢語的形式與功能研究》123132北京:商務印書館.

  • Jiang Yan . 蔣嚴. 2011.〈“都”的形式語用學〉,蔣嚴主編《走進形式語用學》:432456。上海:上海教育出版社.

  • Jiang Jingzhong and Haihua Pan . 蔣靜忠 潘海華. 2013. 〈都的語義分合及解釋規則〉,《中國語文1: 3850.

  • König Ekkehard . 1991. The Meaning of Focus Particles. London/New York: Routledge.

  • Lee Thomas Hun-tak . 1986. Studies on Quantification in Chinese. Ph.D. dissertation, University of California, Los Angeles.

  • Li Wenshan . 李文山. 2013.〈也論“都”的語義複雜性及其統一刻畫〉,《世界漢語教學3: 319330.

  • Li Yunbing . 李雲兵. 2000.《拉基語研究》。北京:中央民族大學出版社.

  • Ma Zhen . 馬真. 1982.〈說“也” 〉,《中國語文4:283287.

  • Matthews S. and Yip V. . 1994. Cantonese: A Comprehensive Grammar. London and New York: Routledge.

  • Pan Haihua . 潘海華. 2006. 〈焦點、三分結構與漢語“都”的語義解釋〉,《語法研究與探索》(十三)。北京:商務印書館.

  • Peng Xiaochuan and Chen Qiping . 彭小川 陳啟萍. 2010. 〈廣州話“類同義”副詞歷時演變的特點與機制〉,《暨南學報(哲學社會科學版)3: 105109.

    • Search Google Scholar
    • Export Citation
  • Peyraube Alain . 2006. On the history of some adverbs of scope and quantity in Chinese, Studies in Chinese Language and Culture: Festschrift in Honor of Christoph Harbsmeier on the Occasion of His 60th Birthday, eds. by Anderl Christoph and Eifring Halvor , 269283. Oslo: Hermes Academic Publishing.

    • Search Google Scholar
    • Export Citation
  • Shank Scott . 2004. Domain Widening. Ph.D. dissertation, Vancouver: The University of British Columbia.

  • Shen Jiaxuan . 沈家煊. 2015.〈走出都的量化迷途:向右不向左〉,《中國語文1: 317.

  • Sun Chaofen . 1996. Word Order Changes and Grammaticalization in the History of Chinese. Stanford: Stanford University Press.

  • Szabolcsi A. , Whang J.D. , and Zu V. . 2014. Quantifier words and their multi-functional parts. Language and Linguistics 15.1: 115155.

    • Search Google Scholar
    • Export Citation
  • Traugott Elizabeth C. and Dasher Richard B. . 2001. Regularity in Semantic Change. Cambridge: Cambridge University Press.

  • van der Auwera Johan and Plungian Vladimir A. . 1998. Modality’s semantic map. Linguistic Typology 2.1: 79124.

  • Wang Hong . 王紅. 1999.〈副詞“都”的語法意義試析〉,《漢語學習6: 5560.

  • Wang Hong . 王紅. 2001. 〈語氣副詞“都”的語義、語用分析〉,《暨南大學華文學院學報2: 4145.

  • Wu Haiping and Tao Hongyin . 2014. Doing generalizations in Mandarin conversation: The case of plural NP+dou expression. Paper presented at the 2nd Annual Conference of the American Pragmatics Association, October 17–19, la.

    • Search Google Scholar
    • Export Citation
  • Xiang Ming . 2008. Plurality, maximality and scalar inference: a case study of Mandarin dou . Journal of East Asian Linguistics 17.3: 227245.

    • Search Google Scholar
    • Export Citation
  • Xiao Shuangrong . 肖雙榮. 2005.〈“都”的全稱量項義與隱含差比義〉,《湖南人文科技學院學報6: 98104.

    • Search Google Scholar
    • Export Citation
  • Xu Liejiong . 徐烈炯. 2014.〈“都”是全稱量詞嗎?〉,《中國語文6: 498575.

  • Yang Rongxiang . 楊榮祥. 1998. 〈總括副詞“都”的歷史演變〉,《北大中文研究1:311332.

  • Yin Chunyi . 尹淳一. 2014. 〈“祖堂集”情態動詞及其語法化研究〉,復旦大學博士論文.

  • Zhang Yisheng . 張誼生. 2005. 〈副詞“都”的語法化與主觀化──兼論“都”的表達功用和內部分類〉,《徐州師範大學學報(哲學社會科學版)1: 5662.

    • Search Google Scholar
    • Export Citation

Content Metrics

All Time Past Year Past 30 Days
Abstract Views 0 0 0
Full Text Views 594 246 20
PDF Downloads 145 93 11