Chapter 2 Literature Review
This chapter consists of three sections. In section 2.1, we will examine three theories on the categorization of modalities since ke is canonically used as a modal. Section 2.2 reviews previous studies on grammaticalization. We will discuss certain key notions about the motivations, mechanisms and characteristics of grammaticalization. Section 2.3 reviews previous studies of ke, including dictionary sources, diachronic studies and synchronic studies.
2.1 Theories on modality
There have been various categorizations of modality in the literature (Lyons 1979, Palmer 1990, Bybee et al. 1994, Coates 1983, Hofmann 1993, Li 1982, Hsieh 2003, etc.). In this section, we will review Lyons (1979), Palmer (1990) and Bybee et al. (1994). Lyons’s categorization of modalities is two-fold, Palmer’s three-fold and Bybee et al.’s four-fold.
2.1.1 Lyons (1979)
Lyons (1979) divided modality into epistemic modality and deontic modality. He incorporated the traditional notions of modal logic, i.e.
possibility and necessity, and further subcategorized each modality into two types. Therefore, in the category of epistemic modality, we have epistemic possibility and epistemic necessity; in the category of deontic modality, we have deontic possibility and deontic necessity.
Epistemic modality concerns the speaker’s evaluation and attitude toward
the truth of proposition. It indicates the speaker’s opinions rather than report
facts. Speaker’s choice of epistemic modals reflects the degree of his/her commitment to the truth of the proposition. The choice of a modal that expresses epistemic possibility, such as may, indicates a lower degree of commitment while a modal that expresses epistemic necessity, such as must, signals a greater commitment. Consider examples (1) and (2).
(1) Epistemic possibility: Alfred may be unmarried. (Lyons 1979: 797) (2) Epistemic necessity: Alfred must be married. (ibid)
Deontic modality, on the other hand, concerns “the necessity or
possibility of acts performed by morally responsible agents” (Lyons 1979). Put simply, it concerns permission (deontic possibility) and obligation (deontic necessity). The deontic sources of permission and obligation can be from some person or some institution that has the authority or from certain social norms or regulations. Examples of deontic modality are as follows:
(3) Permission: You may open the door. (Lyons 1979: 839) (4) Obligation: You must open the door. (ibid)
Lyons’ categorization of epistemic and deontic modalities are widely accepted and adopted in later studies of modality. He, however, did not have much discussion on modals that express ability and volition. Traditionally, in the two-fold categorization of modality, modals of ability and volition are recognized as deontic modality (Sweetser 1990).
2.1.2 Palmer (1990)
Unlike Lyon’s bipartite categorization, Palmer’s categorization of
modality is tripartite. He singled out modals of ability and volition from the
category of deontic modality and grouped them as dynamic modality. The
result is a three-fold model: epistemic modality, deontic modality and
dynamic modality. Like Lyons, Palmer also subcategorized each modality in two subtypes, i.e. possibility and necessity.
Palmer’s definition of epistemic modality is similar to Lyon’s. It concerns the speaker’s “judgments about the possibility, etc., that something is or is not the case” (Palmer: 1979: 50).
Palmer’s deontic modality also includes permission (deontic possibility) and obligation (deontic necessity) but his definition of deontic modality is narrower than Lyons’s. He only regarded as real deontic modals those which are employed by speakers who have the authority to give permission or lay obligation. Therefore, he excluded most modals that have I or we as the subject, since we don’t usually give permission to, or lay obligation upon ourselves. He also excluded modals that indicate circumstantial necessity, where the necessity or the obligatoriness of the action comes from the circumstances, not from the speaker. Palmer referred to these modals as modals of dynamic necessity.
Palmer’s dynamic possibility covers a wide range of modality. It includes neutral/circumstantial use, ability use and extended uses such as implication use and private verb use. Neutral/circumstantial use simply indicates the degree or extent that an event is possible, usually with an implication of the circumstances; subject-oriented ability use indicates the subject’s internal ability or necessary qualities; implication use not only indicates what one can do but also suggest that what should be done, and private verb use refers to verbs of sensation and perception.
(5) Neutral/circumstantial use: You can only get the job if you don’t want it. (Palmer 1990: 84)
(6) Ability use: They can’t speak a word of English, of course, not
a word, but, you know, they can say what they like. (ibid: 85)
(7) Implication use: I can tell you the reference, if that’s any help, of the matter. (ibid: 86)
(8) Private verb use: I can see the moon. (ibid)
While the neutral and the ability uses of dynamic modality are clear, the implication and the verb use seem confusing. Modals of implication use are in fact modals of neutral use; the implication is simply a contextual
interpretation. When one says I can tell you the reference, it is obvious that s/he does not merely state the possibility, which would be really odd and violate the rules of conversation; instead s/he says this to indicate that s/he is willing to give the information. Such interpretation is quite pragmatic and can’t be regarded as an independent use as the neutral use of dynamic modality. As for the private verb, we might as well treat them as one type of modals of ability rather than separate them as a different kind since when one can see something, it means that his/her eyes are functioning well and it is the necessary characteristics or power of eyes to see things.
Palmer has noted that the indeterminacy between circumstantial and subject-oriented (ability) dynamic possibility and between deontic modality and dynamic necessity. He recognized a gradation between dynamic modality and deontic modality and argued that sometimes it is merely a matter of terminology. What we call deontic is simply “subjective deontic” while what is dynamic is simply “objective deontic” (Palmer 1990: 131).
Palmer’s categorization is clearer than Lyons’s in the way that he
recognized the essential differences between dynamic modality and deontic
and epistemic modalities. He also noted the indeterminacy of modalities,
which had to do with their diachronic relations. However, his criteria for the
categorizations of modals seem to involve more than the modal itself. For example, dynamic necessity concerns the subject of the modal, the private verb use involves the types of verbs, and the implication use is simply the pragmatic interpretation from the context. In this way, we are not sure if the criteria of the categorization of modals are based on the senses of the modal itself or even involve the semantics of the subject, the verb and the pragmatic factors. Though there is a danger of over-categorization, a detailed
classification like Palmer’s may help us recognize better the nuances of uses of modals.
2.1.3 Bybee, Perkins and Pagliuca (1994)
Bybee et al. divided modality into four types: agent oriented modality, speaker oriented modality, epistemic modality and subordinating modality.
Agent oriented modality “reports the existence of the internal and external conditions on an agent with respect to the completion of the action expressed in the main predicate” (Bybee et al. 1994: 177). It concerns
obligation, necessity, ability, and desire. Modals of obligation will extend to report intention; modals of desire will extend to express willingness. Modals of ability will generalize to express root possibility, which indicates the general enabling conditions that are not exclusively restricted to internal conditions but concerned with external, social, or physical circumstances. This is similar to Palmer’s circumstantial use of dynamic possibility. In fact, Bybee et al.’s agent oriented modality includes Palmer’s dynamic possibility, dynamic necessity and part of deontic modality.
Speaker oriented modality imposes conditions on the agent to complete
the predicate action. It concerns “directives, such as ‘commands, demands,
requests, entreaties (all of which are ‘mands’) and warnings, exhortations and recommendations” (Bybee et al. 1994: 179). Notions included in speaker oriented modality are imperative, prohibitive, optative (concerning wish or hope), hortative (concerning encouragement and incitement), admonitive (concerning warnings), and permissive (concerning permission). Though grouped as a separate type, modals of speaker-oriented modality are usually modals of agent-oriented modality used in directives. Clearly, Bybee et al.’s speaker oriented modality equals Palmer’s deontic modality.
Likewise, Bybee et al.’s epistemic modality is the same as Lyons’s and Palmer’s. However, instead of the traditional subcategorization of possibility and necessity for epistemic modality, Bybee et al. have proposed three degrees of epistemic modality, i.e. possibility, probability and inferred certainty.
(9) Possibility: I may have put them down on that table; they’re not in the door. (ibid: 180)
(10) Probability: The storm should clear by tomorrow. (ibid)
(11) Inferred certainty: There must be some way to get from New York to San Francisco for less than $ 600. (ibid)
As for modals of subordinating modality, they are modals of epistemic and speaker-oriented modalities used in subordinating clauses. The types of subordinate clauses where modals may occur can be complement clauses, as in (12), in concessive clauses, as in (13) and in purpose clauses, as in (14).
(12) I suggested that he should call you immediately. (ibid)
(13) Although he may be a wise man, he has made some mistakes in the past. (ibid)
(14) We are working hard now so that we can take the summer off. (ibid)
The functions/meanings of the subordinating modals are basically the same as
when they are used as agent-/speaker- oriented modalities. Therefore, we will not devote a particular discussion to the modals used subordinately.
Bybee et al.’s framework has categorized modality in four types, i.e.
agent-oriented modality, speaker-oriented modality, epistemic modality and subordinating modality. In addition, Bybee et al. have noticed the diachronic relations among these four modalities. Agent oriented modality may extend to speaker oriented modality when used in directive, and it also gives rise to epistemic modality. With their diachronically oriented discussion of modality, the idea of how the different modalities are applied and related can be clearly captured, which is not discussed in Palmer’s and Lyons’s synchronic studies.
2.1.4 Interim summary
In this section, we have reviewed three theories of modality: Lyons (1979), Palmer (1990) and Bybee et al. (1994). Lyons’s deontic modality includes a wide range of modals. It contains both the more objective dynamic modality and the more subjective deontic modality. Likewise, Bybee et al.’s
agent-oriented modality encompasses a wide range of modals too. The
obligation and necessity uses of agent-oriented modality overlap with
Palmer’s deontic modality and dynamic necessity. Bybee et al.’s speaker
oriented modality can be confusing when it comes to questions of asking
permission, in which case, the deontic source comes from the addressee, not
the speaker. Thus, it should be seen as addressee-oriented modality (Palmer
1990). Furthermore, Bybee et al.’s subordinating modality is in fact the
employment of agent or speaker oriented modals in subordinate clause, and
therefore they do not in fact make a separate category of modality. Palmer’s
implication use is a pragmatic aspect of modals and therefore modals of
implication use can be either modals of root possibility or intention in Bybee et al.’s terminology.
Since our study is data driven, we will not adopt any specific framework of modality. For the analysis of ke’s modal uses, when the use of modal ke expresses a possibility of action which is determined by external objective circumstantial factors, we will refer to such use as a (dynamic) circumstantial modal. When ke expresses a possibility of action which concerns more with the appropriateness of action judged by the speaker than with the external circumstances, we will refer to such use as a (dynamic) rational modal (Palmer 1990) 1 . When ke expresses a possibility of action which depends on the speaker’s will or his identification with social norms and moral standards, we refer to such use as a deontic modal. When ke expresses a possibility which indicates the degree of the speaker’s commitment to the truth of the
proposition, we refer to such use as an epistemic modal.
2.2. Previous Studies on Grammaticalization
Grammaticalization is viewed as the dynamic, unidirectional, historical process whereby a lexical word assumes a grammatical function (Traugott 1989, Heine et al. 1991, etc.). It consists in “the increase of the range of a morpheme advancing from a lexical to a grammatical or from a less
grammatical to a more grammatical status” (Kuryłowicz 1975:52). In other words, a lexical term in its process of grammaticalization develops in the direction toward abstraction: a concrete word will acquire an abstract meaning when a less abstract word will acquire a more abstract meaning.
Such direction is unidirectional and irreversible.
1
The distinction between rational and deontic uses will be discussed with more detail in section 3.0.
In what follows, some key notions of grammaticalization discussed in previous researches will be reviewed for the hope that they will facilitate our discussion of ke’s grammaticalization in Chapter 4.
2.2.1 Metaphor and Metonymy
Metaphor and metonymy have been recognized as the two major
mechanisms of grammaticalization (Traugott & Heine 1991, Heine et al. 1991, Hopper & Traugott 1993, Bybee et al. 1994, etc.). Generally speaking,
metaphor is considered to be responsible for the early stage of
grammaticalization in the lexical end while metonymy is regarded to be responsible for the later stage of grammaticalization in the grammatical end.
The mechanism of metaphor is highly valued in the traditional
cognitive-semantic approach to the study of grammaticalization with the emphasis on how the cognitive concepts are mapped onto the language
structure. The mechanism of metonymy, on the other hand, is highly valued in the more recent discourse-pragmatic approach, which has its focus on how discourse functions are realized in the grammatical patterns (Givón 1979). In the following paragraphs, we will first talk about the mechanism of metaphor and then move on to the discussion of metonymy.
Metaphor is seen as a cognitive strategy to express abstract concepts by means of concrete ones. That is, it is a transfer form a concrete domain to a less concrete domain or from a less abstract to a more abstract domain. Such transfer is said to be iconic and analogical in principle. One famous type of metaphorical transfer is referred to as categorical metaphor (Heine et al.
1991):
PERSON > OBJECT > SPACE > TIME > QUALITTY
In this cline, items on the left are used to express items on the right. This cline not only shows how metaphor works in general, but also indicates the direction of grammaticalization, i.e. from the left end to the right end. Several important concepts of grammaticalization can be captured in this cline. First, items on the left are more concrete than those on the right and are less
grammaticalized. Second, concepts that are related to human ( PERSON ) are less grammaticalized than those that are nonhuman ( OBJECT ). Third,
concepts that are spatially oriented are less grammaticalized than those that are temporally oriented. The Chinese verb 過 guo can best exemplify the shift from space to time and then to quality. Originally guo denotes a spatial
movement, as in 過了彰化 guo le Zhanghua ‘go past Changhua (name of county)’. Later, it can denote temporal movement, as in 過了中午 guo le zhongwu ‘past noon, after noon’. Finally it can denote the crossing/violation of rules or norms, as in 過分 guo fen ‘cross the line’. As we can see,
metaphorical transfer is a cross-domain extension from a concrete to an
abstract one, during which the image schemata is preserved. In the case of guo, the concept of passing is preserved (cf. Wu 2003).
While metaphor is an analogical and cross-domain mapping, metonymy, on the other hand, is congruous and within-domain. Metonymic process is recognized as the association between congruous concepts in the continuum of speech. It is inferential in essence and serves to strengthen the degree of informativeness of speech. Metonymy is often used to explain the rise of induced interpretation from contexts and conversational implicatures. One often discussed example of metonymy is the grammaticalization of the phrase
‘be going to’ into a marker for intention. The motion verb go in its progressive
form indicates a future time frame, and when it co-occurs with the purposive
to, the intention sense is inferred since a planned event has to occur, if it will, in a later time frame than that of the utterance time.
The process of grammaticalization was considered goal-oriented. The exploitation of old means for novel uses is due to the situation that such novel uses find no expressions in the existing language system (Heine et al. 1991, Traugott 1982). Therefore, the mechanism of metaphor is favored since it concerns the employment of concrete objects, which are usually old, existing means, to express abstract concepts, which are usually new to the language users. Such explanation, however, is recently argued against with the evidence from cross-linguistic and discourse studies, which prove grammaticalization to be more mechanic than functional (Bybee et al. 1994, Hopper & Traugott 1993, 2003). Evidence shows that the process of grammaticalization is a cumulative effect of the constant employment of language in discourse.
Moreover, the view that grammaticalization is goal-oriented goes against the uniformitarian principle (Hopper & Traugott 2003).
Since metaphor failed to explain certain phenomena in
grammaticalization, researchers started looking for explanation in the factors of context and pragmatics in the process of grammaticalization. As a result, metonymy is gaining its significance in studies of grammaticalization. In recent literature, metonymy comes to be considered a more fundamental mechanism than metaphor, which was seen in many ways grounded in
metonymy (Heine et al. 1991). For example, the spatial-temporal metaphor of the English phrase going to is now considered metonymic in nature since a movement from one space to another will certainly involve the passing of time.
Therefore it is natural to infer a temporal reading from a spatial reading.
While metaphor and metonymy are different mechanisms, they are not
exclusive; instead, they are complementary in the sense that metaphor gives rise to conceptual transfer when metonymy to context induced interpretation.
Given its analogical nature, metaphor is responsible for paradigmatic changes when metonymy, given its associative power on the basis of congruity of continuous speech, is more explanatory for syntagmatic changes.
2.2.2 Semantic Bleaching and Pragmatic Strengthening
As we have said, the process of grammaticalization involves the emptying
of semantic specifics of lexical words. Emptying of specifics generalize the
meaning of words, which will facilitate the process of grammaticalization
(Hopper and Traugott 2000, Xing 2003). Once the meaning is generalized,
employment of words becomes less restricted and they can be used in more
contexts. Since the meaning is reduced, context will have to come into play in
the interpretation of meanings. The pragmatic aspect of words is thus getting
more and more important in their application. Once the context and the
pragmatics are getting more and more deciding power in the interpretation of
words, and the context induced meaning occurs in high frequency and in
various contexts, the newly gained meaning/function will be added to the
words and established as the basics, i.e. semanticized. The example of be
going to serves as a good illustration of pragmatic strengthening. While the
direction sense of be going to is getting bleached in I’m going to get married
(Hopper & Traugott 1993, 2003), the phrase infers an intention reading with
the purposive to, and the future meaning. Once the future meaning is added in
the sense of be going to, it can apply in sentences whose meaning does not
involve any motion at all, such as It is going to rain, or in sentences which
would have resulted in semantic anomaly, as in He’s going to come.
In the literature of grammaticalization, there are different models concerning the relationships of semantic bleaching and pragmatic
strengthening. They are bleaching model, containment model, and loss and gain model. Bleaching model (Heine et al. 1991) contends that the process of grammaticalization is the process of emptying the specifics of meanings, by which only the core general sense is kept. However, in the later stage, chances are that the source is impossible to detect in highly grammaticalized terms. On the contrary, containment model (Givón 1973, Bybee & Pagliuca 1985), which is revised as prototype extending model, holds that a grammatical term, much as it is modified and emptied of its specifics, is related to the core meaning of its lexical source. Sweetser (1988) proposed the loss and gain model, which argues that the loss of source meaning in the lexical domain is compensated by the gain in the grammatical domain by means of metaphorical transfer, during which the image schemata of the source meaning is reserved.
2.2.3 De-categorization, Re-categorization and Reanalysis
In the process of grammaticalization, one common phenomenon is reanalysis as a result of metonymic association, that is, the structure of
sentence is rearranged, which, however, does not involve the change of surface
structure (Langacker 1977, quoted from Traugott 2003). Hopper & Traugott
(1993) illustrated reanalysis with the example back of the barn, which was
originally parsed as [[back] of the barn], where back is part of the barn and of
denote possession, or part/whole relation. But it is now reanalyzed as [back of
[the barn]], where back of behaves more like a preposition and of does not
indicate possession anymore. In the process of reanalysis, we see that back is
de-categorized from a noun and is re-categorized into a preposition.
The phenomena of de-categorization, re-categorization and reanalysis are significant in the grammaticalization of Chinese verb. Xing (2003) argued that in the process of grammaticalization of Chinese verbs, grammatical
morphemes developed from verbs undergo three stages of reanalysis, i.e. verb serialization, verb decentralization and functionalization. According to Xing, the source structure of a main verb before its process of grammaticalization is diagrammed as:
(NP 1 ) + V (+NP 2 )
And the first reanalysis, that is, verb serialization, is diagrammed as:
(NP 1 ) + V 1 (+NP 2 ) + V 2 (+NP 3 )
The structural change is simply the combination of two source structures and the more semantically significant verb will be decided as the main verb while the other will be the secondary verb. This process, as the second reanalysis, is diagrammed below:
Case A: (NP 1 ) + V main (+NP 2 ) + V secondary (+NP 3 ) Case B: (NP 1 ) + V secondary (+NP 2 ) + V main (+NP 3 )
The secondary verb will go on to change to a function word:
Case A: (NP 1 ) + V main (+NP 2 ) + G [rammatical Word] (+NP 3 ) Case B: (NP 1 ) + G [rammatical Word] (+NP 2 ) + V main (+NP 3 )
Xing has illustrated the three-stage reanalysis with the Chinese verb 將 jiang.
Stage 1: 將秦軍為前行
Jiang Qin jun wei qian xing.
‘(He) commanded the army of Qin and led them to go forward’
Stage 2: 將妻子放到店中
Jiang qizi fan zhi dian zhong.
‘Taking his wife, he settled her in the store.’
Stage 3: 寶玉將黛玉喚醒
Baoyu jiang Daiyu huan xing.
‘Baoyu woke up Daiyu.’
In stage 1, jiang functions as a verb, meaning ‘command’, co-occurring with another verb xing ‘go, walk’. They are separate actions in succession.
While in stage 2, the action that jiang, meaning ‘take’, denotes is to fulfill the action denoted by the verb fang. The agent’s main purpose is to settle his wife but not just to take her with him. Hence, jiang denotes merely a secondary action. However, Xing did not provide enough contexts to support her argument that jiang is the secondary action. In stage 3, jiang does not function as a verb but as an object marker. The only action in the example of stage 3 is huanxing ‘to wake up’. It would be odd if we see jiang as a verb and interpret the sentence as ‘Baoyu took Daiyu and woke her up,’ since Daiyu was not taken to any place. She simply stayed where she was asleep and Baoyu just woke her up. Jiang in this stage functions purely as an object marker.
2.2.4 Subjectification and Intersubjectification
Subjectification involves the shift of perspectives from the syntactic
subject to the speaking subject. When a lexical term begins its course of
grammaticalization, its pragmatic aspect of meaning is strengthened and
enriched. Pragmatics concerns how language is used in context, and it has to
do with the speaker’s choice of words to express his subjective attitudes and
opinions. Thus, pragmatic strengthening of a lexical term is the strengthening
of subjectivity (Traugott 1995). It is by the process of pragmatic strengthening
that a conversational inference becomes semanticized and meanings come to
encode the speaker’s attitude and opinions rather than objective facts.
Traugott (1982) has argued that the meaning changes that a word undergoes in the process of grammaticalization is from the propositional meaning, to the textual meaning and then to the expressive meaning. The proposition meaning of a word is internal and does not encode much of the speaker’s attitude while the textual meaning of a word indicates the way the speaker decides to organize the discourse to present ideas and the relationship between ideas. When a word has acquired its expressive meaning, instead of contributing its original semantics to the proposition of the sentence, its main function is to show the speaker’s personal attitudes toward the proposition. It is a process toward greater and greater subjectivity. Traugott (1989)
elaborated the direction toward subjectivity in grammaticalization by proposing three tendencies:
Tendency I: Meanings based in the external situation > meanings based in the internal (evaluative/perceptual/cognitive) situation.
Tendency II: Meanings based in the external or internal situation >
meanings based in the textual and metalinguistic situation.
Tendency III: Meanings tends to be increasingly based in the speaker’s subjective belief state/attitude toward the proposition.
Tendency I applies to most of the cases of semantic degradation or
amelioration, where a neutral term assumes a negative or a positive meaning.
The word boor, for example, with its original neutral meaning as “farmer’,
develops a pejorative meaning of ‘crude person’. Tendency II is illustrated
with the word while. With its original meaning as ‘the time’, while develops a
new function as a connective, meaning ‘during’. Tendency III applies to the
shift of the word very, originating from the French word verai ‘true’, to a
scalar particle in English, as in the very height of her career (Traugott 1989).
Examining these three tendencies in the perspective of metaphor and metonymy, since metaphor concerns the representation of concepts while metonymy concerns the speaker’s way of presenting his ideas, we can see that tendency I and tendency II involve the mechanism of metaphor when
tendency III involves the mechanism of metonymy (Traugott & König 1991).
With the acquisition of subjective meaning, some words will further undergo intersubjectification. Different from subjectification which centers meanings on the speaker, intersubjectification centers the focus on the
addressee (Traugott 2003). Common cases of intersubjectification can be seen in uses of deixis, honorifics and hedges. One example given by Traugott is the stance adverb actually in Actually, I will drive you to the dentist. The word actually not only indicates the speaker’s attention to the addressee but also implies the speaker’s awareness to the addressee’s counterexpectation; it is therefore used to mitigate the potential disagreement by the addressee.
Traugott proposed that the path of development is from nonsubjective to subjective and then to intersubjective. She contended that
intersubjectification can be seen as an extension of subjectification but is not a necessary step in grammaticalization.
2.2.5 Grammaticalization of modalities
In the chapter of introduction, we mentioned that cross-linguistic
evidence indicates that modals are derived from lexical verbs. The diachronic development of modal auxiliaries from lexical verbs involves not only
semantic attrition but also loss of syntactic, morphological and phonological
characteristics (Hopper and Traugott 1993, Roberts and Roussou 1999, Beths
1999, Wu 2003, Xing 2003, Heine et al. 1991).
Take *sculan for example (Beths 1999, Traugott 1994). Syntactically,
*sculan functioned as a main verb and took a noun phrase as its complement.
When changing to a modal auxiliary, it lost its ability to take nominal complements and started to take infinitival clauses. Semantically, *sculan’s verbal meaning is ‘to owe someone something’ and now as a modal it means
‘shall’, indicating future tense. In the case of Chinese, the Mandarin desire verb 要 yao originally means ‘want’ and takes a noun as its complement. It later comes to co-occur with another verb to form a serial verb construction.
Gradually its semantic primacy/importance has been reduced and served as a modal which indicates intention, modifying the verb to its right. Later on, it has acquired another modal meaning indicating prediction. In an even later stage of its grammaticalization, yao is used as a conditional marker meaning
‘if’ (cf. Chang 1996).
When a lexical verb becomes a modal auxiliary, the process of grammaticalization does not stop; instead, the modal auxiliary will still continue its course of grammaticalization. A dynamic modal can acquire a deontic function (Palmer 1979, Bybee et al. 1994); a deontic modal can
develop into an epistemic modal (Traugott 1989, Heine et al. 1991, Bybee et al.
1994, Li 2004, Peyraube 1991) and an epistemic modal can become more and
more subjective or start to serve a concessive function (Traugott 1989, Bybee
et al. 1994). Based on their cross-linguistic evidence, Bybee et al. (1994)
proposed paths of grammaticalization that modalities may undergo.
Path I:
protasis
ability root possibility epistemic possibility comp 2 to think permission concessive
Path II:
future purpose
intention comp to want, order obligation
imperative
probability comp to think concessive Path III:
Desire Purpose Intention Comp to want Movement toward future Imperative
Probability Comp to think
Concessive
The above three paths can be summarized as:
Speaker-oriented
Agent-oriented Subordinating Epistemic
One thing deserving to note is that Bybee et al. argued that for an ability modal to acquire an epistemic meaning, the transition to root possibility is prerequisite, but that no evidence can be found that a deontic modal would
2
‘Comp’ refers to ‘complementizer’
change directly into an epistemic modal.
2.3 Previous Studies of Ke
In this section, we will first examine the definition and functions of ke listed in four dictionaries, i.e. 語言大典 Yuyan Dadian (henceforth, YDD) (1990), 漢語大字典 Hanyu Da Zidian (henceforth, HDZD) (1990), 辭海 Cihai (henceforth, CH) (1979) and 大辭典 Da Cidian (henceforth, DCD) (1985). Additionally, some grammar books on Chinese grammatical words will also be reviewed. Then we will review previous studies on ke as follows:
section 2.3.2 examines the diachronic studies of ke in modern Chinese, and section 2.3. 3 deals with the synchronic studies which focus on the
development of ke, including studies with the specific focus of ke in older texts.
2.3.1 Dictionaries
In the four main dictionaries we are to examine, YDD provides the most comprehensive meanings of ke; twenty-one meanings are given in this dictionary while six are given in CH, twelve in DCD and eighteen in HDZD.
Except for YDD, which categorizes the meanings of ke on the basis of syntactic categories, the other three dictionaries simply provide different
interpretations of ke in various contexts. All of them are simply a listing of meanings and little connection is made among them; both diachronic and synchronic usages of ke are listed together without differentiation.
Concerning the overlapping of meanings and functions, we will first
examine the interpretations of ke given by YDD, and interpretations not
documented in YDD will be mentioned in following sections.
2.3.1.1. Yuyan Dadian (1990)
YDD categorizes the uses of ke into three categories, i.e. ke as a verb, ke as an adjective, and ke as an adverb. First we will discuss its verb use.
2.3.1.1.1 Ke as a verb
As a verb, ke denotes eight meanings. They are ‘permit’, ‘can, be able to,’
‘may, can,’ ‘be worth,’ ‘be correct, right’ ‘accord with,’ ‘may not, can not’ and
‘recover’. Each meaning will be illustrated with one example.
(i) Permit:
(15) 始皇可其議, 收去詩書百家之語以愚百姓. (史記 李斯列傳) Shihaung ke qiyi, shoqu shishu baijiazhiyu yi yu baixing.
‘The emperor approves the suggestion of confiscating all the literary books to keep the subject ignorant.’
(ii) Can, be able to:
(16) 彼蒼者天, 殲我良人, 如可贖兮, 人百其身! (詩經 秦風 黃鳥) Bi cang zhe tian, jian wo liangren, ru ke shu xi, ren bai qi shen.
'Thou azure Heaven there! Thou art destroying our good men. Could he have been redeemed? We should have given a hundred lives for him.' (Legge 1971)
(iii) May, can:
(17) 可遇不可求.
Ke yu bu ke qiu.
‘(We) can only rely on chance but (we) can not have it with force.
(iv) Be worth:
(18) 亦可畏也, 伊可懷也! (詩經 豳風 東山)
Yi ke wei ye, yi ke huai ye.
'These thoughts made us apprehensive, and they occupied our breasts.' (Legge 1971)
(v) Be correct, right:
(19) 然則古之無變, 常之毋易, 在常古之可與不可. (韓非子 南面)
Ran ze gu zhi wu bian, chang zhi wu yi, zai chang gu zhi ke yu buke.
‘Whether the old traditions are to last long or not and whether the social norms are to be abided long or not, it all depends on whether they are right or wrong for the time.’
(vi) Accord with:
(20) 其味相反, 而皆可於口. (莊子 天運) Qi wei xiangfan, er jie ke yu kou.
‘Their flavors are different, but they all suit our taste.’
(vii) May not, can not:
(21) 異哉, 試乃可已. (尚書 堯典) Yi zai, shi nai ke yi.
‘Well but.... Try him and then you can have done with him.' (Legge, 1971)
(viii) Recover:
(22) 待軍師病可，行之未遲. (三國演義) Dai junshi bing ke, xing zhi wei chi
‘Let’s wait until the military counselor recovers from his illness, then it’s still not too late to set off.’
Examining the eight verbal senses that YDD has provided, we found two
problems with the categorization. First, most of the verbal meanings are
similar; they are only different in their contexts, which contribute to the
different pragmatic interpretations. For example, the first, the fifth and the
sixth senses all derive from a core meaning ‘agree with, accord with’. The first interpretation of ‘permit’ comes from the subject of the sentence where the subject is the king. Under the context where the subject has the power to ratify proposals, the interpretation of ke as ‘permit’ would be more proper than
‘agree with’. Likewise, the fifth interpretation of ‘correct’ comes from the context where whether rules and social norms are right or wrong depends whether or not their content agrees with/ is in accordance with the time. As for the eight interpretation of ‘recover’, it is an extended meaning from the
“agreeing” sense, meaning ‘acceptable’. The ‘recover’ interpretation of ke simply derives from the context where the topic is the conditions of someone’s illness. Second, the second, the third and the seventh senses are rather
modal-like than verbal. The negative interpretation of the seventh sense is merely a conditional interpretation implied in the context. While the second and the third senses are listed separately, they both indicate a circumstantial possibility.
2.3. 1.1.2 Ke as an adjective
Ke has four meanings in this syntactic category. They are ‘ordinary, light,’
‘facing, against,’ ‘full, all,’ and ‘able’. Each meaning will be illustrated with one example.
(i) Ordinary, light:
(23) 白日裏可些, 到晚來越淒慘過活不得. (嬌紅記) Bairili ke xie, dao wan lai yue qican guohuobude.
‘In the daytime it (the pain) is still light and ordinary, but when the night comes, it gets more heartbreaking and unbearable.’
(ii) Facing, against:
(24)日斜疏竹可窗影, 正是幽人睡足時. (題繼祖蟠室)
Ri xie shuz hu ke chuang ying, zheng shi yoren shui zu shi
‘When the sunlight slants and the shadow of bamboos is projected onto the window shade, it is time for a hermit like me to take a nice nap.’
(iii) Full, all:
(25) 他還是把一肚子話可桶兒都倒出來. (兒女英雄傳) Ta haishi ba yiduzi hua ketonger dou dao chulai.
‘He nevertheless poured (to us) all of the thoughts he had in mind.’
(iv) Able:
(26) [這是]可變換的.
[Zheshi] ke bianhuan de.
‘(This is) changeable.’
Similar problems exist in the category of adjective. Both the ‘ordinary’
sense and the ‘full’ sense in fact share an ‘agreeing, acceptable’ meaning. The
‘ordinary’ interpretation comes from the comparison of the pain in the daytime and in the nighttime. Compared with the unbearable pain in the nighttime, the pain in the daytime is bearable/acceptable, and thus, ordinary and lighter. The ‘full’ interpretation of ke only exists in the fixed phrase ketonger, which attempts to convey the image that as many thoughts as a bucket can contain (figuratively) is poured out. This is also a certain degree of agreeing, i.e. an accordance with the capacity of a bucket.
As for the ‘facing’ and the ‘able’ interpretations, the former functions
more as a verb than as an adjective while the latter functions more as a modal
than as an adjective. In (24), the image of accordance/agreeing still remains
while the context implies a ‘facing’ interpretation. In (26), ke indicates a
possibility of change. It is similar to the ‘may, can’ and ‘may not, can not’
senses listed in the verbal category.
2.3.1.1.3 Ke as an adverb
As an adverb, ke has nine different interpretations. They are ‘but,’ ‘about, approximately,’ ‘just,’ ‘again,’ ‘certainly, really,’ ‘interrogative marker,’ ‘high degree,’ ‘a lot,’ ‘finally, eventually’ and ‘only’. Each meaning will be illustrated with one example.
(i) But:
(27) 勞動辛苦, 可大家幹勁十足.
Laodong xinku, ke dajia ganjinshizu.
‘The job is laborious, but everyone is very energetic.’
(ii) About, approximately:
(28) 重可千斤.
Zhong ke qianjin.
‘It weights about a thousand jin.’
(iii) Just:
(29) 戰城南, 死郭北, 野死不葬鳥可食
Zhan chengnan, si guobei, yesi buzang niao ke shi.
‘Due to the war, there are many dead bodies around the castle.
‘The dead bodies are laid unburied, just serving as the food of birds.’ (樂府詩集)
(iv) Again:
(30) 我去莊院人家燙熟了酒, 吃了呵, 可來取我這把鐵鍬. (老生兒)
Wo qu zhuanyuan renjia tang shu le jiu, chi le he, ke lai qu wo
zheba tieqiao.
‘I first went to one of the manors, warmed the liquor, drank it and again I went there to fetch my spade.’
(v) Certainly, really:
(31) 那本書我可沒拿.
Nabenshu wo ke meina
‘I really did not take that book.’
(vi) Interrogative marker:
(32) 你可聽過他殺過人?
Ni ke ting guo ta sha guo ren?
‘Have you heard that he killed someone?’
(vii) High degree:
(33) 人可多了.
Ren ke duo le.
‘What a big crowd we’ve got here!’
(viii) Finally eventually:
(34) 我們等了你好久, 你可回來了!
Women deng le ni haojiu, ni ke huilai le.
‘We have been waiting for you for ages. Thank God you are finally back!’
(ix) Only:
(35) 這可是千真萬確的!
Zhe ke shi qianzhenwanque de!
‘This is only as sure as fate!’
As can be seen from the examples, though defined differently, the
‘certainly,’ ‘interrogative marker,’ ‘high degree,’ ‘finally, eventually,’ and ‘only’
senses in fact serve to add emphasis and carry a subjective meaning, while the
‘just’ and ‘again’ senses in fact indicate a dynamic possibility. The meanings of
‘again’ and ‘just’ are merely contextual inferences, which do not seem to have to do with ke’s use. As for the ‘approximately’ meaning, we prefer to see it as ke’s verbal use of ‘agree with,’ with the interpretation ‘the total weight agrees with 1000 jin,’ i.e. the weight reaches 1000 jin. We do not think of
‘approximately’ as a proper interpretation. It at best is a contextual inference.
With regard to the ‘but’ sense, as will be discussed, it is a derived meaning from the emphasis use.
2.3.1.2 Hanyu Da Zidian (1987)
Unlike YDD, HDZD does not classify the functions of ke by syntactic category, but merely presents the different interpretations of ke. Among the eighteen interpretations presented in HDZD, two are not documented in YDD.
The first one is equivalent to 所 suo, serving as a functional word before a verb to signify the state or outcome of the action denoted by the verb.
(36) 無可有, 以大豆一斛相助. (笑林) Wu ke you, yi dadou yihu xiangzhu.
‘I’ve got nothing but fifty deciliters of soybeans to offer you.’
In the above example, ke is used to indicate what the verb can achieve, which is shown with the word preceding ke, i.e. wu, ‘nothing, none’.
Another interpretation in HDZD that is not mentioned by YDD is ‘how come, how is it possible’, usually used in a negative sense.
(37) 可知年四十, 猶自未封侯. (詩詞曲語辭匯釋) Ke zhi nian sishi, youzi wei fenghou.
‘How come I would know that I should not be offered any office at
the age of forty?’
In the above example, though ke serves as an interrogative marker for rhetorical question, yet the answer is always the negative, i.e. the
speaker/writer himself did not know that he should get himself into a
miserable state. Ke in this function implies the speaker’s counter-expectations.
This function is similar to the ‘interrogative marker’ function of ke. However, when example (32) can be a real question, ke in the above example is never a real question nor is addressed to a second party but merely an interjection when the speaker is having reflections.
Both of the senses are in fact not new to us. The suo interpretation of ke actually denotes a possibility, a negative one, of the predicate action you. Thus, it is a modal usage. The second interpretation functions similarly to the
emphatic use mentioned in section 2.3.1.1.
2.3.1.3 Da Cidian (1985)
DCD lists twelve interpretations of KE, four of which are not documented in YDD and HDZD. They are ‘good’, ‘not bad, passable’, ‘strength’ and ‘should’.
These senses are illustrated respectively in examples (41), (42), (43) and (44).
(i) Good:
(38) 阻而鼓之, 不亦可乎? (左傳) Zu er gu zhi, bu yi ke hu?
‘We can just attack them when they are held back, isn’t it perfect?’
(ii) Passable:
(39) 可也; 未若貧而樂, 富而好禮者也. (論語) Ke ye; wei ruo pin er le, fu er hao li zhe ye.
‘They will do, but they are not equal to him who though poor is yet
cheerful and to him who though rich loves the rules of propriety.’
(iii) Strength:
(40) 僕亦無一可耶! (後漢書 朱暉傳) Pu yi wu yi ke ye!
‘I don’t have any strength either.’
(iv) Should:
(41) 及平長, 可娶妻, 富人莫肯與者, 貧者平亦斥之. (史記 陳丞相傳) Ji Ping zhang, ke qu qi, furen mo ken yu zhe, pinzhe Ping yi chi zhi.
‘When (Chen) Ping grew up and should get married, the rich would not want to marry their daughters off to him while Ping himself would feel shame to marry a girl from an impoverished family.’
While the fourth interpretation of ‘should’ seems a valid one, the other three do not. The ‘good’ and the ‘not bad’ interpretations are similar to the
‘acceptable’ interpretation, while the ‘strength’ interpretation is similar to the possibility interpretation in that one’s strength implies what one can do, i.e. a possibility.
2.3.1.4 Cihai (1979)
Cihai provides only six senses of ke, all of which are noted in the above three dictionaries. They are ‘permit’, ‘about’, ‘right, proper’, ‘certainly’, ‘how is it possible,’ and the use as a question marker.
2.3.1.5 Interim summary
In this section, we have examined four dictionaries, i.e. Yuyan Dadian,
Hanyu Dazidian, Da Cidian, and Cihai. All of them try to offer an exhaustive
list of the uses of ke in both ancient and modern Chinese. None of them,
however, account for the relations among the various meanings and functions of ke. Most of the senses are merely contextual interpretations and can not be seen as established semantics. A unified and systemic explanation for
phenomenon of the multi-functional ke is hence needed.
In the next section, we will review some diachronic studies of ke.
2.3.2 Diachronic studies of Ke
In this section, we will review three studies: Peyraube (1991), Duan (1992) and Li (2004). Peyraube focuses on how ke is used in Standard Classic
Chinese, which corresponds to our classification of Old Chinese. Duan examined a broader range of data, approximately from Early Old Chinese to Late Middle Chinese. Li examined both diachronic and synchronic data but put more focus on the diachronic data.
2.3.2.1. Peyraube (1991)
Peyraube (1991) had his focus on the modal use of ke in Standard
Classical Chinese, i.e. from the period Shangshu and Shijing to that of Lunyu.
According to Peyraube, ke, and its equivalent keyi, are the most common modal auxiliaries used in Classical Chinese and also the most difficult ones to describe. He argued that ke mainly serves as non-epistemic modality, i.e.
deontic possibility (permission) and deontic necessity (obligation or
requirement). The use of ke to denote epistemic modality might be derived from their deontic values.
Peyraube has identified four uses of ke in Standard Classic Chinese. The
first modal use expresses permission, the second obligation or requirement,
the third epistemic possibility, and the fourth ‘worthy of’, which is neither
deontic nor epistemic.
(i) Permission:
(42) 燕可伐與? (孟子 公孫醜) Yan ke fa yu.
‘Can Yan be attacked?’
[Is it permissible to do it?]
(ii) Obligation or requirement:
(43) 賢者之為人臣也, 其君不賢則故可放與? (孟子 盡心)
Xian zhe shi wei renchen ye, qi jun bu xian ze gu ke fan yu ‘(When) a good man serves as subordinate, (if) his prince is not good, ought he, naturally, to be exiled?’
(44) 晉不可不善也. (國語 周語) Jin bu ke bu shan ye
‘(The State of) cannot but be good.’
(iii) Epistemic possibility:
(45) 其或繼周者雖百世可知也 (論語) Qi huo ji Zhou zhe, sui baishi ke zhi ye
‘That which (would) also inherit (from the regime of) Zhou can be known even though after one hundred generations.’
(iv) ‘Worthy of’:
(46) 雖小道必有可觀焉. (論語 子張) Sui xiaodao bi you keguan yan
‘Though (it is a) small craft, is has necessarily (something) that is worthy of being admired.’
While Peyraube recognized (42) as a deontic one, we have a different opinion.
In (42), the question is addressed by a minister to Mengzi, who in fact does
not have any power to decide whether the country Qi is permitted to set on the country Yan or not. The criteria of “permission” here are based rather on the circumstances than on the will of Mengzi. Therefore, it is more likely Palmer’s circumstantial possibility. As regards the deontic use, Peyraube remarked that the sense of deontic necessity is emphasized when ke is inserted between negative words, as (44) has shown. As for the epistemic possibility, the interpretation of ke in (45) as epistemic possibility seems plausible, but we prefer the dynamic circumstantial possibility reading. In (45), the sentence is said by Confucius to answer his disciple’s question whether one can
see/predict things in ten generations/ten generations ahead 3 . Confucius responded that we can know the things of ten generations ahead by examining what is inherited now. The dynamic circumstantial possibility reading seems more reasonable than the epistemic one.
Peyraube has noted that ke’s deontic values are more basic and representative in Pre-Qin language. The epistemic value of ke is a later development. He then proposed two hypotheses of the path of changes that Chinese modals undergo. The first one is from non-epistemic to epistemic and the second one is from weak epistemicity to strong epistemicity. These
changes are triggered by two factors. One is frequency of occurrence in
contexts and the other is weakening of deontic semantics and strengthening of the focus of speakers’ belief and attitude to the proposition. Peyraube claimed that the strengthening of the focus of the speakers’ belief and attitude is a case of strengthening of informativeness.
Appealing as his observation may sound, Peyraube did not provide a very satisfying argument for his assumptions. He simply mentioned them in
3
The exact words of the question are ‘十世可知乎?’.
passing. We did not see how the different uses of ke are related, and how the contexts and semantic weakening and pragmatic strengthening contribute to the change of meaning. Furthermore, since Peyraube confined his discussion to the modal use of ke, other uses of ke were not covered in his study, and thus the whole picture of the uses of ke is not accomplished yet.
2.3.2.2 Duan (1992)
In his book on grammatical words in ancient Chinese 4 , Duan (1992) classified the uses of ke in two syntactic categories, i.e. modal auxiliary and adverb. The modal ke can express four meanings and the adverb ke has seven meanings.
As a modal auxiliary, ke serves to express circumstantial possibility, subject-oriented possibility, i.e. ability, obligation, and worthiness of behavior or characteristics.
(i) Circumstances:
(47) 官爵可買, 則商工不卑也矣. (韓非子 五蠹) Guanjue ke mai, ze shanggong bu bei ye yi
‘If one can get a government position by paying money, then businessmen and laborers are not at all low in social status.’
(ii) Ability:
(48): 可勝也, 而弗能居也. (左傳 哀公) Ke sheng ye, er fu neng ju ye
‘We are able to defeat them but we can not settle in their place.’
(iii) Obligation:
4
Duan did not specify that exact period, but from his examples we can see that the period Duan examined is
approximately from Early Old Chinese to Late Middle Chinese.
(49) 天下方未定,故可因遂就宮室. (史記 高祖本記) Tianxia fang wei ding, gu ke yin sui jiu gongshi
The world is not settled yet, so we should repair and build the palace in the nick of time.
(iv) ‘Worthy of’:
(50) 當水石峻激相搏處, 最為可玩. (朱熹 百丈山記) Dang shuishi junji xiangbuo chu, zui wei ke wuan.
‘Where the river violently breaches the rocks is most worth visiting sights.’
As an adverb, ke is used to express approximation of quantity,
certainty/emphasis of statements, turn/disjunction of meanings, and ‘just in time’, the latter two of which are later developments.
(i) Approximation:
(51) 項羽之卒可十萬. (史記 高祖本記) Xiangyu zhi zu ke shiwan
‘The amount of Xiangyu’s army is about one hundred thousand people.’
(ii) Certainty:
(52) 吾懼燕人掘吾城外塚墓, 僇先人, 可為寒心. (史記 田單列傳)
Wu ju Yanren jue wu chengwai zhongmu, lu xianren, ke wei hanxin.
‘I’m afraid that the Yan people will dig at the tombs outside of the castle and disgrace our ancestor’s bodies. It really mortifies us and hurt our hearts.’
(iii) Turns of meanings:
(53) 吾與足下相知久矣, 可不復甚解. (答李應書)
Wu yu zuxia xiangzhi jiu yi, ke bu fu shen jie.
‘We have been friends for a long time yet I don’t know you very well.’
(iv) Just in time:
(54) 由來碧落銀河畔, 可要金風玉露時. (辛未七夕) You lai bi luo yinhe pan, ke yao jinfeng yulu shi
‘(It is) when the wind is just about to blow and the frost is just about to form (that they meet) at the bank of the Milky Way in the sky.’
As an adverb, ke can serve to indicate three different moods. They are imperative mood, as in (55), interrogative mood, as in (56) and the mood of rhetorical questions, as in (57).
(55) 買得棺木未? 可速買. (逸史 樂生) Mai de guanmu wei, ke su mai.
‘Have you bought the casket yet? you should go pick one quickly!’
(56) 桃花一簇開無主, 可愛深紅愛淺紅? (江畔獨步尋花) Taohua yicu kai wu zhu, ke ai shenhong ai qianhong?
‘The peach blossom blooms beautifully and everyone can enjoy the view of it. Which color of flower do I prefer after all, the dark red one or the light red one?’
(57) 有君如彼其信也, 可無歸乎? (韓非子 外儲說左上) You jun ru bi qi xin zhe, ke wu gui hu?
‘With a king so true to his promises as him, how can one not serve him?’
Duan has covered ke’s functions as a modal and as an adverb, and he also mentioned that ke’s adverb use of is a later development. While Duan’s
categorization of ke’s modal uses seems plausible, his categorization of ke’s
adverb uses is somewhat confusing. First, some of the adverb uses seems
different simply owing to the contexts. For example, the function of ke in (54) is exactly the one in (52). They both add emphasis to the following
action/situation. Second, Duan mentioned three moods that the adverb ke can mark but he did not explain either how these three moods are marked by ke.
To us, ke in (55) and (57) seem more likely to be a modal. In (55), ke has a deontic value, meaning ‘should’ and in (57), ke has a dynamic value, meaning
‘can.’ In (56), on the other hand, ke functions as an adverb with a similar meaning to (49). It is used to highlight the speaker’s indecisiveness to choose.
The moods of these three sentences are not decided by ke, but by the contexts.
Though Duan used extensive data of ke in different periods of Chinese, he did not discuss the diachronic relations between these various functions of ke.
We did not know how ke’s modal uses are related and which use occurs earlier and which occurs later. Moreover, the relations between ke’s modal uses and adverb uses are still left unexplained.
2.3.2.3 Li (2004)
In his typological study of modality in English and Chinese, Li (2004) discussed the semantic development of modal ke, and he observed three modal uses of ke. He also noticed some post-modal uses of ke.
First, Li noted that as a modal, ke can express three types of modality, i.e.
epistemic modality, participant-internal modality, i.e. ability, and
participant-external deontic modality 5 , as illustrated in (58), (59) and (60), respectively.
(i) Epistemic modality:
5
