CC BY
11
I. A. Mel'cuk
Université de Montréal, Montréal
THE EAST/SOUTHEAST ASIAN ANSWER TO THE EUROPEAN PASSIVE
Как все мы знаем, он таковский, Наш славный Виктор Эс. Храковский! Свой нежный шлю ему привет, Желаю жить сто двадцать лет!1
1. The Problem Stated
This paper tries to answer a seemingly simple question: Is there a passive voice in Mandarin Chinese? A number of descriptive grammars, reference books, manuals and special papers speak about the passive voice in Chinese, indicating, however, its particularities with respect to what is called passive in many European languages. I will start with a blunt answer:
I No, there is no such a thing as passive voice in Mandarin Chinese.
1 'As we all know, he's like that, / Our glorious Viktor Es. Khrakov-skij! / I am sending him my tender greetings, / And I wish that he live hundred twenty years!' — However, while formulating this desire, I feel slightly embarrassed. The fact is that way back, in 1948, the Soviet Union was celebrating the 70-th anniversary of the Greatest Leader of all times and peoples, the Most Famous General and the Most Beloved Father of scientists and athletes, as well as the First Linguist, Comrade Joseph V. Stalin. In the middle of the festivities, students of the Moscow Foreign Languages Institute Maurice Thorez prepared, as was the custom, a hand-written wall newspaper dedicated to the event. It featured a poem created by a local Homer, in which the author wished that "the Great Stalin might live hundred thousand years." Everything seemed perfect, when, suddenly, the school's Communist Party Bureau ordered the paper removed from the wall and destroyed. The editor got a Party reprimand. "Why should we limit our greatest Leader longevity?" he was told. Many years later, I have my doubts: Maybe the Party Bureau was right after all? Maybe we really should not limit?
The goal "our" languages achieve by using the passive is reached in Chinese - and several Southeast Asian languages - in quite a different way. The rest of my paper demonstrates why this is so.
What is needed to establish whether there is or is not the passive voice in Chinese? I will try to answer by following the example of a Soviet-era military medical cadet of the last century, who was asked at the final examination at his Academy: "What do you need to give an enema?" He became famous for his prompt answer: "First, you need an enema; second, you need an anus; third, you need to apply the first to the second." In the same vein, I need, first, a definition of the passive; second, a precise description of the relevant Chinese facts; third, I have to apply the first to the second - and bingo!
Consequently, the paper is organized in an obvious way: Section 2 presents a definition of the passive voice; Section 3 describes the construction called passive in Chinese; Section 4 applies the definition of the passive to the Chinese facts to achieve and buttress the conclusion that what we see is not the passive voice, but an essentially different phenomenon; Section 5 sketches the situation in a couple of other Southeast Asian languages facing a similar problem; and Section 6 represents the conclusion.
It is not by chance that I chose to write on the passive voice for Xrakovskij's Festschrift: Viktor Xrakovskij is one of those scholars who pioneered intensive and extensive investigations into the problems of voice in many languages. He also contributed to the domain several important and influential studies: [Xrakovskij 1974, 1975, 1981], to name but a few (they were republished, with corrections and additions, in [Xrakovskij 1999]); see also his "summing-up" paper [Xrakovskij 2004]. Therefore, my article is a tribute to his long-standing, assiduous and fruitful work in the domain.
2. The Passive Voice
To propose a rigorous definition of the passive voice, four steps are needed: formulating the principles on which such a definition must be based (2.1) and then give the definitions of diathesis (2.2), of voice (2.3), and of the passive (2.4).
2.1. Conditions on a Good Definition
Since this paper is essentially based on definitions of the concepts used, it is worthwhile to dwell on the concept of scientific definition
itself. I will consider definitions of linguistic concepts, although what is said might well apply in other fields (see [Mel'cuk 2006a]).
2.1.1. Substantive Requirements. First of all, the definition of a concept must satisfy the three substantive requirements related to the question "What exactly is to be defined?":
• A definition of X must be oriented towards prototypical cases of X; "deviant" cases are to be covered by additional special conditions.
• The phenomenon X one wants to define must be defined as a particular case, or a subclass, of a more general phenomenon Y. In other words, a definition must be strictly deductive - that is, to be an Aristo-telian/Boetian analytical definition of the form "X is a Y which is Z," where Y and Z have been defined previously.
• Specific differences Z - properties that define Z as a subclass of Y - must be reduced to the simplest2 defining features possible, so that they ensure a systematic hierarchical class inclusion. For instance, consider a commonly used definition of ergative construction:
Definition 1: *Ergative Construction
Ergative construction [= X] is a transitive verb predicative construction [= Y] such that [= Z]:
(i) its Direct Object is marked in the same way as the Subject of an intransitive verb;
(ii) its Direct Object is marked by the nominative case;
(iii) its Subject is marked by a special case different from the nominative.
This definition satisfies our first and second substantive requirements above, but fails the third one: its Z consists of three independent properties each of which can be absent. Taking out one of them produces a new definition that defines... what? Something for which there is no name and which does not belong to a previously defined subclass. What is indeed a transitive verb predicative construction for which Condition (i) is not satisfied - its Direct Object [= DirO] is not marked (under specific circumstances) in the same way as the Subject of an intransitive verb, but Conditions (ii) and (iii) are satisfied? Such a situation is found, for instance, in a Papuan language, Motu [Lister-Turner, Clark 1931: 28-30], in common sentences of type (1b):
2 'Simplest' is to be construed liberally enough: simplest, but such that allows for a sensible classification.
(1a) Sisia+na vada e + la dog PATH PERF 3SG go 'The dog has gone'.
(1b) Sisia+ese boroma+0 vada e +kori+a
dog ERG pig NOM PERF 3SGSUB bite 3SGOBJ 'The dog has bitten the pig'.
(1c) Sisia+ese mero+na vada e +kori+a
dog ERG boy PATH PERF 3SGSUB bite 3SGOBJ 'The dog has bitten the boy'.
In Motu, the DirO is marked by the nominative, except for human nouns: with them, it is marked by the pathetive case (a special case found also in some Malayo-Polynesian and Australian languages: see [Mel'cuk 1988: 180-181]). The intransitive Subject in Motu is always marked by the pathetive (^ nominative!), and the transitive Subject, by the ergative case.
According to the letter of Definition 1, (1b) is not an ergative construction (because its DirO is not marked the same way as the intransitive Subject). But what is it? This type of construction clearly belongs to a subclass of transitive verbal constructions that also includes the ergative construction in the sense of Definition 1; however, this subclass has no definition and no name. The construction in (1c), which is very close to (1b), belongs nevertheless to a different subclass (of transitive verbal constructions), to which we cannot refer: it does not have a name, either. Worse, these two subclasses do not form a common subclass within the class of transitive constructions. In order to avoid such violations of step-wise consistent hierarchical classification, I propose to define first the most general subclass of verbal predicative constructions that includes the Definition 1 ergative construction as a particular case (see [Mel'cuk 1988: 182, 251, 258ff; Mel'cuk 2006b: 269ff]).
Definition 2: Ergative Construction
Ergative construction [= X] is a verbal predicative construction [= Y] whose Subject is marked by a case different from the nominative [= Z]3.
3 The nominative being, of course, the case of naming.
Proceeding from Definition 2, I would say that in (1a) we have an intransitive ergative construction, in (1b) a transitive ergative construction with a nominative DirO (the most current type), and in (1c) a transitive ergative construction with a non-nominative DirO. The er-gative construction in the sense of Definition 1 is then a transitive ergative construction with a nominative DirO that coincides with the intransitive nominative Subject - a very particular case.
This way of defining guarantees a systematic inclusion of concepts in the corresponding subclasses, without missing important intermediate classes.
2.1.2. Formal Requirements. Second, a definition must be formally correct in the three respects, related to the question "How do we define what we define?":
• A definition must be formal - that is, it should be applicable verbatim (= mechanically).
• A definition must be rigorous - that is, it should contain only concepts which either have been defined prior to it or else are indefinable.
• A definition must be adequate - that is, sufficient and necessary, covering all the phenomena that are perceived as subsum-able under the corresponding concept, and nothing but such phenomena.
Given my topic, I will not dwell on this question any more, but in the present paper I accept these six conditions as postulates.
2.1.3. Prototypical Passives. As prototypical cases of passives, I take the passive in Latin, Armenian and Swahili:
LATIN
(2a) i. Serv +i reg +em
slave PL.NOM king SG.ACC porta+0 + nt + 0 carry PRES.IND 3PL ACT 'The slaves are carrying the king'.
ii. Rex a(b) serv + is
king-SG.NOM by slave PL.ABL porta+0 +t + ur carry PRES.IND 3SG PASS 'The king is being carried by the slaves'.
ARMENIAN
(2b) i. Asot +0 + 0 + a namak+er + 0 + a
Ashot SG NOM DEF letter PL NOM DEF gr +0 + ec + 0 write ACT AOR IND.3SG 'Ashot wrote the letters'4.
ii. Namak+ er + 0 +a gr +v + ec + in letter PL NOM DEF write PASS AOR IND.3PL Asot +0 + i + 0 koymic
Ashot SG DAT NON-DEF from.side 'The letters were written by Ashot'.
SWAHILI
(2c) i. Wa+tanzania wa+na + sem +0 + a
II Tanzania II PRES speak ACT DECLAR Ki+swahili VII Swahili
'Tanzanians speak Swahili'.
ii. Ki+swahili ki + na +sem + w + a
VII Swahili VII PRES speak PASS DECLAR na Wa+tanzania
with/by II Tanzania 'Swahili is spoken by Tanzanians'.
Here is what can be stated about these prototypical passives:
1. There is no propositional semantic difference between Sentences (i) and (ii). They show, of course, a communicative semantic difference: in Sentence (i), the Actor is the Sem-Theme (= Topic) of the sentence, while in Sentence (ii), the Patient is. The passive is used to express communicative information.
2. The crucial syntactic difference between Sentences (i) and (ii) is as follows:
4 Note the following particularity of the nominal case system in Modern Armenian: it does not have an accusative, so that the Subject and the DirO are both marked by the nominative. A similar situation is found in other languages, for instance, in Romanian and Nivkh. Here, the DirO is case-marked the same way as the Subject of an intransitive verb. However, the transitive predicative construction of these languages is never called ergative!
• In Sentence (i), the Actor is expressed as the Deep-Syntactic Actant I/the Surface-Syntactic Subject, and the Patient, as the DSyntA II/the Direct Object.
• In Sentence (ii), the Actor is expressed as the Deep-Syntactic actant II/the Surface-Syntactic Agent Complement, and the Patient, as the DSyntA I/the Subject.
It is impossible to explain here the concepts of Semantic Actant [= SemA], Deep-Syntactic Actant [= DSyntA], and Surface-Syntactic Actant [= SSyntA], even if they are essential for the discussion; see [Mel'cuk 2004a, 2014: Ch. 7].
3. The crucial morphological difference between Sentences (i) and (ii) consists in the difference between the forms of the Main Verb: in Sentence (ii), but not in Sentence (i), it has a special suffix, which marks the communicative and syntactic modification, stated above; this suffix is the marker of the passive. As a result, we obtain the opposition of active vs. passive forms. All other morphological differences observed in the verb and the actantial nouns are automatic consequences of that difference.
Based on active ~ passive opposition, we must call passive such verbal forms that are semantically identical to active forms, but syntactically entail the transformation characterized above. To describe this transformation in formal terms, the concept of diathesis is needed.
2.2. Diathesis
Each lexeme (= a word taken in one well-defined sense) that expresses a predicative meaning has actants: SemAs, DSyntAs and SSyntAs. What interests us here is the correspondence between the SemAs and DSyntAs of a lexeme. For instance, the noun JOY (X's joy over Y) has two SemAs: 'X', who experiences the feeling, and 'Y', which is the cause and the object of that feeling; JOY also has two DSyntAs: DSyntA I (implemented on the surface by a possessive form or by a phrase with OF), which expresses 'X', and DSyntA II (implemented by a prepositional phrase with OVER), which denotes 'Y'5.
Definition 3: Diathesis
The diathesis of a lexeme L is the correspondence between the SemAs and the DSyntAs of L.
5 There is a huge literature on the concepts of diathesis and voice, which I cannot survey even cursorily. I base this exposition on my own work - in particular, [Mel'cuk 2004b] and [Mel'cuk 2006b: 181-262].
The noun JOY has the following diathesis: X ^ I, Y ^ II; it can also be written as
X Y
I II
In many languages, some verbal lexemes (e.g., transitive verbs) can have more than one diathesis: such is exactly the case in Latin, Armenian, and Swahili. One of the diatheses corresponds to the basic, lexicographic form of the verb, while the other one corresponds to the passive - that is, a form "derived" from the basic form by the corresponding affix. This other diathesis can be written as X ^ II, Y ^ I, or as
X Y
II I
Now we can say that passivization is the following modification of the basic diathesis:
X Y
I II
X Y
II I
Three operations on diatheses - permutation of DSyntAs with respect to SemAs, suppression of DSyntAs, and referential identification of SemAs (with violation of the correspondence between SemAs and DSyntAs in the basic form) - produce, for a binary basic diathesis, 12 possible distinct modifications, including the zero one (see, e.g., [Mel'cuk 2004b: 293-300, 2006a: 184-191, 2006b: 194-209]). The zero modification of the basic diathesis corresponds to the active, and the simplest permutation produces the diathesis that corresponds to the passive.
2.3. Grammatical Voice
At this juncture, the definition of grammatical voice seems straightforward.
Definition 4: Grammatical voice
Grammatical voice is a verbal inflectional category whose gram-memes (= particular voices) mark the modification of the basic diathesis and are themselves formally marked on the verb.
Note that formally marked on the verb does not necessarily mean 'marked on the verb by an affix': a modification of the basic diathesis
can be marked by a structural word such as an auxiliary verb (as in (3a)) or an invariable particle (as in (3b)); for instance:
ENGLISH / FRENCH / GERMAN (3a) The letter was written by John himself. = La lettre a été écrite par Jean lui-même. = Das Brief wurde von Hans selbst geschrieben lit. 'The letter became by Hans himself written'.
(3b) ALBANIAN (ë = /9/)
Active vs. Passive
'I open' hapa ~ 'I am being opened' u hapa
'YouSG open' hape ~ 'YouSG are being opened' u hape
'He opens' hapi ~ 'He is being opened' u hap
'We open' hapëm ~ 'We are being opened' u hapëm
'YouPL open' hapët ~ 'YouPL are being opened' u hapët
'They open' hapën ~ 'They are being opened' u hapën
Thus, voices can have analytical forms, just like any other inflectional category.
But can there be a change of the basic diathesis that is not marked on the verb, but on one of its actants? Yes, such a situation is logically possible, and it exists, for instance, in Ancient Chinese [Jaxon-tov 1965: 47, 1974: 201]:
(4a) Sha ren '[He] killed [a] man'. ~
Shayu ren '[He] was.killed by [a] man'.
(4b) Cheng bao min, de bao cheng.
city.walls protect people virtue protect city.walls 'City walls protect people, the virtue protects city walls'.' ~
Min bao yu cheng,
people be.protected by city.walls cheng bao yu de.
city.walls be.protected by virtue
'People are protected by city walls, city walls are protected by virtue'.
The diathesis of the verbs SHA 'kill' and BAO 'protect' in the first members of the pairs of sentences in (4) changes in their occurrences in the second members of these pairs. However, since this change of
the basic diathesis is not marked on the verb, second sentences of the pairs in (4) do not represent grammatical voice. These sentences manifest the pseudo-passive construction; a genuine passive construction needs a passive verbal form.
2.4. Passive
Definition 5: Passive voice
Passive is a grammatical voice that marks a change of the basic diathesis such that it involves the permutation of DSyntA I.
In other words, a passive voice necessarily entails demotion of DSyntA I. Summing up:
The passive voice of verb V is a synthetic or analytical form of the lexeme V that expresses a change of the basic diathesis of V such that consists in permuting at least V's DSyntA I.
We have the first piece of our puzzle. Let's move to the second one.
3. Chinese "Passive Construction"
Unfortunately (for me), I do not know Chinese, and in what follows I proceed strictly from the data available in printed sources. In the first place, these are [Li, Thompson 1981; Hashimoto 1988; Ren 1993; Paris 1998; Huang 1999; Huang et al. 2008], from which I take my examples. (I modified these examples a bit, to make them easier to understand; among other things, I replaced the Chinese human names by English ones. Also, as in many publications, I do not indicate the tones.) Here is a typical example of what is currently called passive sentence / construction in Chinese:
(5) Mary bei tufei dasi- le baba bandits kill PERF father
lit. 'Mary BEI bandits killed father'. = 'Mary lost her father to
bandits'.
(The lexical unit BEI cannot be properly glossed before its meaning and syntactic function are clarified.)
Huang [1999], following in some respects [Hashimoto 1988], demonstrates that the lexeme BEI, commonly called "the passive marker," is an auxiliary verb with a very vague meaning ~ '[to] undergo
[that]' or '[to] be affected by', which has a rather syntactic function; generally speaking, what follows BEI is a full normal clause with its own syntactic subject, etc. As a result, sentence (5) is best literally translated as 'Mary «underwent.that» bandits killed father'. Here are Huang's four arguments for this description (again, I slightly reformulated and rearranged them).
1) BEI is not a preposition. In spite of many traditional approaches that classify BEI as a preposition (e.g., [Alleton 1973:121-122; Li, Thompson 1981: 365; Ren 1993: 127ff; Paris 1998: 358ff])6, the presumed passive marker BEI is by no means a preposition introducing an agent noun complement. Consider, for instance, (6a), a very common type of sentence containing a clause introduced by BEI (= a BEI-clause):
(6a) Mary zuotian bei John da - le
yesterday underwent hit PERF
'Mary was hit by John yesterday'.
One cannot say that here BEI forms a prepositional phrase with JOHN, for at least three reasons:
- The presumed prepositional phrase *bei John cannot be positioned in any other slot in the sentence, while normal prepositional phrases can appear in all these slots - except for the position between BEI and the subject of the subordinate clause (b vs. c):
(6b) *Bei John Mary zuotian da-le. / *Mary bei John zuotian da-le. / *Mary zuotian da-le bei John.
(6c) Zai jiali Mary bei John da-le. / at home
Mary zai jiali bei John da-le. / Mary bei John zai jiali da-le. / Mary bei John da-le zai jiali.
But not *Mary bei zai jiali John da-le, see below, before (14), on the particularities of the clause introduced by BEI.
- Very often (actually, more often than not) BEI is not followed by a noun, but by the subjectless verb:
6 As far back as half a century ago, [Solnceva 1962: 66-67] clearly stated that bei cannot be considered a preposition, but rather represents a defective verbal form.
(6d) Mary bei da-le 'Mary was beaten'.
In this case, it is customary to speak of an ellipsis of the noun; however, the ellipsis of a noun between a preposition and a verb is a rather exceptional situation in Chinese. Moreover, this is not a contextual ellipsis: the missing indication of the agent is absent from the syntactic (and the semantic) structure of the sentence: it cannot be "restored" from previous context.
- BEI cannot be repeated in coordination, although this repetition is near-obligatory for prepositions in Chinese [Hashimoto 1988: 331-332]:
(7a) Mary bei quinren huaiyi wairen zhize
relative suspect stranger criticize 'Mary is suspected by relatives [and] criticized by strangers'.
vs.
*Mary bei quinren huaiyi bei wairen zhize.
(7b) Mary zai xuexiao xuexi, zai jiali xiuxi in school study in home rest
vs.
*Mary zai xuexiao xuexi jiali xiuxi.
2) Syncategorematic (= subject-oriented) lexemes in the BEI-clause.
- The BEI-clause can contain the adverb GUYI 'intentionally', as in (8):
(8) Mary bei John guyi da -le
intentionally hit PERF lit. 'Mary underwent.that John intentionally hit [her]'.
Since GUYI can semantically bear only on the syntactic subject and in (8) it bears on JOHN, the use of GUYI shows that JOHN is the syntactic subject in the BEI-clause.
- The BEI-clause can also contain the reflexive pronoun ZIJI ~ 'self, which can be coreferential only with the syntactic subject:
(9) Neifeng xin bei John daihui ziji - de jia qu -le this letter bring.back self 's home go PERF 'This letter was brought back by Johni to hisi house'.
This gives another indication to the effect that JOHN is the subject of the BEI-clause (rather than the complement of the "preposition" BEI).
3) Syntactic phenomena characteristic of normal clauses. The BEI-clause can feature various constructions that show it to be a normal clause, with a subject, a Main Verb, and even with a DirO:
- Coordination with gapping, illustrated by (10):
(10) Mary bei John ma -le liang-sheng
scolded PERF two times Peter da -le san -xia
hit PERF three times 'Mary was scolded twice by John and hit three times by Peter'.
- So-called "long-distance dependencies", where the element coref-erential with the subject of BEI does not depend on the Main Verb of the BEI-clause directly, but through a string of subsequent dependencies:
(11) Mary bei zhengfu pai jingcha zhuazou- le
government send police arrest PERF lit. 'Mary underwent.that government sent police arrested'. = 'Mary was arrested by the police on government's orders'.
As one can see, the understood (but not expressed) DirO 'her' in the BEI-clause depends on 'arrest', which in its turn depends on 'send'.
- The verb in a BEI-clause can have its own DirO ([Chappell 1986: 274, 277]; this paper specifies the conditions imposed on this DirO):
(12a) Ta bei diren dashang - le tui he enemy hit.wound PERF leg
lit. 'He underwent.that enemy hit.wounded leg'. = 'He had his leg wounded by the enemy's fire'.
(12b) Ta bei pengyou kai - le yi ge wanxiao he friend play PERF a CLAS joke
lit. 'He underwent.that friends played a joke'. = 'He had a joke played on him by his friends'.
(12c) Yifu bei shao-le yi ge dong cloths burn PERF a CLAS hole
lit. 'Cloths underwent.that [fire] burnt a hole'. = 'The fire burnt a hole in his cloths'.
4) Resumptive pronoun in the BEI-clause. A BEI-clause can contain a resumptive pronoun coreferential with the subject of BEI in the role of DirO - in at least two types of context7:
• If the resumptive pronoun is not sentence-final, but is followed by a lexical expression, as in (13a):
(13a) Mary bei John da -le ta san -xia
hit PERF she three times lit. 'Mary underwent.that John hit her three times'. 'Mary was hit by John three times'.
• If the resumptive pronoun is turned into an Affected Object, introduced by the preposition BA and preceding the Main Verb, see (13b):
(13b) Mary bei John ba ta da - le
she hit PERF
lit. 'Mary underwent.that John her hit'. = 'Mary was hit by John'.
This would not be possible if the Main Verb of the BEI-clause were (in any sense) passive8.
Taking all this into consideration, we have to accept Hashimoto's and Huang's proposal, summarizing it in the following two points:
• Syntactically, BEI is the Main Verb of the sentence; it is a bivalent auxiliary verb and roughly means something like 'undergo that' = 'be affected by'.
Prof. Yin finds that sentences (13a-b) "do not sound natural".
8 A resumptive pronoun cannot appear in a bei-clause as a DirO (co-referential with the subject of bei) if this pronoun turns out to be clause-final element:
(i) Mary bei John da-le *ta lit. 'Mary bei John hit her'. = 'Mary was hit by John'.
There is still another argument supplied by [Hashimoto 1988: 335] against the sentences with the bei-clause being a "passive" construction: the possible absence of the "active" counterpart. Thus, for (ii) there is no correspondent sentence without bei:
(ii) Kanshou bei fanren pao-le. ' [The] jailer suffered because of [the] criminal's escape'.
This argument is, however, invalid: "passives without actives" are not a rarity at all. Recall the Japanese passive, let alone verba deponentia of Classical languages.
• The BEI-clause is a normal "active" clause with a transitive Main Verb, although this clause has a few special properties: it can lack an overt subject not in a contextual controlled ellipsis; its own DirO most often - although by no means always! - has the same referent as the subject of BEI; if this DirO occupies the last linear position in the sentence, it cannot be expressed by a resumptive pronoun, but otherwise it can; its subject cannot be preceded by a prepositional phrase, which otherwise is quite common, cf. (6c) above; etc.9
Therefore, a Chinese "passive" sentence is, if literally glossed, something like this:
(14a) Mary bei da-le
lit. 'Mary underwent/that [some.people] hit [her]'.
(14b) Mary bei John da-le
lit. 'Mary underwent.that John hit [her]'.
(14c) Mary bei John da-le ta san-xia
lit. 'Mary underwent.that John hit her three.times'.
BEI is the Main Verb of the whole sentence10; the surface-syntactic structure of (14c) is as follows:
9 [Huang 1999: 11]:
(i) John bei Mary zaixuexiao da-le 'John was hit by Mary at school'.
vs.
(ii) *John bei zai xuexiao Mary da-le 'John was hit at school by Mary'.
10 It is true that bei does not have several properties of normal Chinese verbs: thus, it does not accept aspect suffixes -le, -guo, -zhe, etc. But then, several Chinese lexemes that are traditionally accepted as auxiliary verbs - for instance, shi 'let', neng 'can' or yao 'want' - do not have these properties, either [Li, Thompson 1981: 172ff; Hashimoto 1988: 339-340]. Also bei alternates with genuine verbs gei 'give', jiao 'be called' and rang 'let, allow' [Li, Thompson 1981: 506]:
(i) Wo gei/jiao /rang ta tou -le liang kuai jian
i got he steal perf two dollar money
lit. 'I underwent/allowed that he stole two dollars [from me]'.
On the other hand, [Hashimoto 1988: 340] indicates that the continuous aspect suffix -zhe is found on bei in texts, even if rarely.
Figure 1. Surface-Syntactic Structure of Sentence (14c)
BEI
subjectival auxiliary
MARY c
-h-
MARY
O
JOHN
1. The bidirectional dashed arrow represents the coreferentiality of the two lexemes.
2. The syntactic relation "auxiliary" in Chinese is different from analogous Synt-relations in European languages, where "auxiliary" SyntRels serves to construct analytical forms of lexemes.
I challenge anybody to produce a different syntactic structure for this sentence; I think this is simply impossible.
The discussion in Section 3 gives us the second piece of the puzzle: the BEI-construction is not passive; it is a normal "active" clause introduced by the auxiliary verb BEI, which has very little semantic content in Modern Chinese and is used mostly for syntactic and communicative purposes. The schematic form of a BEI-clause is as follows:
'X undergoes.that Y Z-s X' o X BEI Y Z(V, transitive) [X]
Now we have to apply the first piece of the puzzle to the second one. If something does not walk like a duck and does not quack like a duck, why call it duck? Perhaps simply because it reminds us of a duck (for instance, a small goose). Yet this is not a compelling reason. The BEI-construction is not passive at all: it is built around a transitive verb that does not change its only diathesis and, quite naturally, has no marking.
It is interesting to mention a linguistic phenomena happening in Chinese now: bei begins to be increasingly used to introduce such verbs as zisha 'commit suicide' or ziyuan 'volunteer', for instance, Ta bei zisha lit. 'He underwent committing.suicide', to mean that he was murdered, the murder disguised as a suicide; Ta bei zuyuan lit. 'He underwent volunteering' - that is, he was forced to volunteer (see http://news.xinhuanet.com/english2010/indepth/ 201002/16/c_13176690.htm).
No matter how sporadic such usages, they show that bei is not perceived by speakers as a preposition.
But it does resemble a passive construction in some essential respects; let us see in which.
4. Affected-Subject Construction in Chinese
The BEI-construction is used when the Speaker wants to speak about X and say about X that Z done by Y happened to X or at least somehow affected X. Suppose that Y Z-ed (e.g., offended) X, and the Speaker chooses to communicate this talking about X; since in Chinese, as a general rule, the syntactic Subject must be the Theme of the sentence, he has to say X bei Y Z. In this way, the BEI-construction does two things with one blow: communicatively, it turns 'X' into the Theme; syntactically, it turns 'X' into an Affected Subject. This is what makes this construction similar to the passive of many languages: both the BEI-construction and the passive construction fulfill (almost) the same communicative and syntactic roles. However, the similarity, even identity, of roles fulfilled by two linguistic phenomena does not entail the similarity, let alone identity, of the phenomena themselves. Should we consider English prepositions to be case markers simply because they often play the same role as cases (marking syntactic dependencies)?
The Chinese construction in question should by no means be called passive; I would suggest for it the straightforward name of Affected-Subject Construction. In addition to its being explicit and clear, this term has another advantage: it forms a pair with the name of another Chinese construction, described, e.g., in [Li, Thompson 1981: 463ff]: the BA-construction. Let us start with an example:
(15) Wo ba chabei nongpo-le
I as.for tea.cup break PERF
lit. 'I as.for tea.cup broke'. = 'I broke the tea cup'.
If a nominal N (here, CHABEI 'tea cup') is intended to be the Direct Object, it must follow the Main Verb (here, NONGPO 'break'), as all DirOs do. But, if supplied with the preposition BA ~ 'as for', it ceases to be a DirO and must precede the verb. (The BA-construction is subject to several constraints: the verb in this construction must express an action that really affects the referent of N; N itself must be definite or generic, but not indefinite specific; etc. This, however, is not relevant for my purposes here.) The sentence element implemented by BA + N cannot be considered a DirO: first, it is difficult to call a nominal
introduced by a preposition a direct object11, and second, more importantly, a sentence with the BA-construction can contain a genuine DirO (boxed in (16)):
(16a) Wo ba juzi bo - le ^ I orange remove PERF skin
lit. 'I orange removed skin'. = 'I removed the skin from the orange'.
(16b) Wo ba John bang - le yiangzhi jiao I tie.up PERF two foot
lit. 'I John tied up two feet'. = 'I tied up John's two feet'.
Therefore, the BA + N phrase must be given a special name; I propose to call it the Affected Object. The syntactic relation that links it to the Main Verb also cannot be called direct-objectival; I propose affected-objectival.
The Affected-Object Construction also thematizes the nominal involved, just as the Affected-Subject Construction does; this enhances its similarity with the latter. [Huang et al. 2008: 155-162] emphasizes the parallelism of both constructions in several respects. But of course this parallelism is not complete: to begin with, BA in the BA-construction is a preposition, while BEI in the BEI-construction is a verb; BA does not carry any propositional meaning, and BEI does (albeit not much); there are other differences as well. But this is beyond the limits of our discussion.
5. Affected-Subject Constructions in Southeast Asian Languages
To drive the nail completely home, I will consider what is called "passive" in Vietnamese, based on [Truong 1970] and [Tam 1976]. Here are examples borrowed from [Tam 1976]:
11 A few cases where a preposition marks a DirO are known. Such is, for instance, the preposition et in Hebrew, which marks exclusively definite DirOs and does nothing else; the preposition z- 'as' marking the DirO in Classical Armenian; or the prepositions a in Spanish and pe in Romanian, which are necessary for animate DirOs under particular conditions of referen-tiality and specificity. However, in none of these cases is the preposition-marked DirO compatible within the clause with another DirO.
(17a) Nga danh Nam 'Nga beat Nam'. ~
Nam bi12 Nga danh 'Nam was beaten by Nga'.
(17b) Nga khen Nam 'Nga congratulated Nam'. ~
Nam duac Nga khen 'Nam was congratulated by Nga'.
BI and BU"0C are auxiliary verbs with meanings, respectively, 'undergo, suffer' and 'receive, benefit from'; accordingly, they produce sentences with adversative or beneficial/neutral meaning.
The same situation obtains in Lao, Khmer and Thai [Tam 1976: 442]:
LAO
(18a) Khacaw khaa muu khoi they killed friend I
'They killed my friend'.' ~
Muu khoi thyyk khacaw khaa friend I undergo they killed 'My friend was killed by them'.
KHMER
(18b) Kee bombaek kbaal knom they break head I 'They broke my head'.' ~
Knom trzw kee bombaek kbaal (knom) I undergo they break head I
'I got my head broken by them'.
THAI
(18c) Dek tii maa child hit dog 'The child hit the dog'. ~
Maa thuuk dek tii dog undergo child hit 'The dog was hit by the child'.
A similar situation is observed in Burmese:
12 The auxiliary verb bi represents the Chinese bei, borrowed into
Vietnamese.
(19) Cuydo yai'+di du+go
I hit DECLAR he DirO
'I hit him'. ~
Du cuydoi dyai'+go khang +ya + di he my blow DirO experience receive DECLAR lit. 'He experience.received my blow'. = 'He was hit by me'.
The difference with the preceding three languages is that what corresponds to the BEI-clause is nominalized in Burmese: 'I hit' ^ 'my blow'; however, in our framework this is irrelevant.
6. The Problem Solved
Summing up, the Chinese construction with the marker BEI is not passive; the category of voice does not exist in Chinese. The BEI-construction can be called the Affected-Subject construction. The same recommendation applies to similar constructions in Vietnamese, Thai, Lao, Khmer, and Burmese.
Acknowledgments
The first version of the paper was read and criticized by L. Ior-danskaja; the subsequent versions were submitted to M. Alonso Ramos, D. Beck, L. Iordanskaja, S. Kahane, P. Magistry, J. Milicevic, A. Polguere, A. Tremblay, Y. L. Wang, and H. Yin. I am deeply grateful to all of them for the proposed corrections and constructive criticisms.
Abbreviations
ii/vii — Noun Class II/VII; abl — ablative case; acc — accusative case; act — active voice; aor — aorist (a verbal tense); clas — classifier; dat — dative case; declar — declarative mood; def — definite; diro — direct object; erg — ergative case; ind — indicative mood; nom — nominative case; non-def — non-definite; pass — passive voice; path — pathetive case; perf — perfect (a verbal tense); pl — plural; pres — present (a verbal tense); sg — singular; sgobj — singular of the object; sgsub — singular of the subject.
References
Alleton 1973 — V. Alleton. Grammaire du chinois. Paris: Presses Universitaires de France, 1973.
Chappell 1986 — H. Chappell. The passive of bodily effect in Chinese // Studies in Language 10, 1, 1986. P. 271-296.
Hashimoto 1988 — M. Hashimoto. The structure and typology of the Chinese passive construction // M. Shibatani (ed.). Passive and Voice, Amsterdam — Philadelphia: John Benjamins, 1988. P. 329-354.
Huang 1999 — C.-T. J. Huang. Chinese passive in comparative perspective // Tsing Hua Journal of Chinese Studies 29, 1999. P. 423-509.
Huang et al. 2008 — C.-T. J. Huang, Y.-H. A. Li, Y. Li. The Syntax of Chinese. Cambridge: Cambridge University Press, 2008.
Iordanskaja, Mel'cuk 2007 — L. Iordanskaja, I. Mel'cuk. Smysl i socetaemost' v slovare (Meaning and Cooccurrence in the Lexicon). Moskva: Jazyki russkoj kul'tury, 2007.
Jaxontov 1965 — S. Jaxontov. Drevnekitajskij jazyk (Ancient Chinese). Moskva: Nauka, 1965.
Jaxontov 1974 — S. Jaxontov. Nekotorye passivnye konstrukcii v kitajskom jazyke (Some passive constructions in Chinese) // A. Xolodovic (ed.). Tipologija passivnyx konstrukcij. Diatezy i zalogi, Leningrad: Nauka, 1974. P. 195-202.
Li, Thompson 1981 — Ch. Li, S. Thompson. Mandarin Chinese. A Functional Reference Grammar. Berkeley: University of California Press, 1981.
Lister-Turner, Clark 1931 — R. Lister-Turner, J. B. Clark. A Grammar of the Motu Language of Papua. Sydney: Pettifer, 1931.
Mel'cuk 1988 — I. Mel'cuk. Dependency Syntax. Theory and Practice. Albany, N.Y.: The State University of New York Press, 1988.
Mel'cuk 2004a — I. Mel'cuk. Actants in semantics and syntax. I-II // Linguistics 42, 1, 2004. P. 1-66; 42, 2, 2004. P. 247-291. (See a Russian translation in [Iordanskaja, Mel'cuk 2007: 39-184].)
Mel'cuk 2004b — I. Mel'cuk. Opredelenie kategorii zaloga i iscislenie voz-moznyx zalogov: 30 let spustja (Definition of the category of grammatical voice and a calculus of possible voices: 30 years later) // V. Xrakovskij, A. Mal'cukov, S. Dmitrenko (eds.). 40 let Sankt-Peter-burgskoj tipologiceskoj skole. Moskva: Znak, 2004. P. 286-314. (See also in [Iordanskaja, Mel'cuk 2007: 184-210].)
Mel'cuk 2006a — I. Mel'cuk. Calculus of possibilities as a technique in linguistic typology // F. Ameka, A. Dench, N. Evans (eds.). Catching Language. The Standing Challenge of Grammar Writing, Berlin — New York: Mouton de Gruyter, 2006. P. 171-205. (See also: http://olst.ling. umontreal.ca/pdf/Melcuk2006_DixonPaper.pdf)
Mel'cuk 2006b — I. Mel'cuk. Aspects of the Theory of Morphology. Berlin — New York: Mouton de Gruyter, 2006.
Mel'cuk 2014 — I. Mel'cuk. Semantics: From Meaning to Text. Vol. 3. Amsterdam — Philadelphia: John Benjamins, 2014.
Paris 1998 — M.-C. Paris. Syntaxe et sémantique de quatre marqueurs de transitivité en chinois standard: BA, BEI, JIAO et RANG // A. Rousseau (ed.). La transitivité, Villeneuve d'Ascq: Presses Universitaires de Septentrion, 1998. P. 357-370.
Ren 1993 — X. Ren. Syntaxe des constructions passives en chinois. Paris: Langages Croisés, 1993.
Solnceva 1962 — N. Solnceva. Stradatel'nyj zalog v kitajskom jazyke (Proble-my morfologii) (The Passive Voice in Chinese (Morphological Problems)). Moskva: Izdatel'stvo vostocnoj literatury, 1962.
Tam 1976 — Duy Le Tam. Vietnamese passive. Papers from the 12th Regional Meeting of the Chicago Linguistics Society, 1976. P. 438-449.
Truang 1970 — Van Chinh Truang. Structure de la langue vietnamienne. Paris: Paul Geuthner, 1970.
Xrakovskij 1974 — V. Xrakovskij. Passivnye konstrukcii (Passive constructions) // A. Xolodovic (ed.). Tipologija passivnyx konstrukcij. Diatezy i zalogi. Leningrad: Nauka, 1974. P. 5-45.
Xrakovskij 1975 — V. Xrakovskij. Iscislenie diatez (Diathesis calculus) // Diatezy i zalogi. Leningrad: Leningradskoe otdelenie Instituta jazyko-znanija AN SSSR, 1975. P. 34-51.
Xrakovskij 1981 — V. Xrakovskij. Diateza i referentnost' (Diathesis and re-ferentiality) // V. Xrakovskij (ed.). Zalogovye konstrukcii v raznostruk-turnyx jazykax. Leningrad: Nauka, 1981. P. 5-38.
Xrakovskij 1999 — V. Xrakovskij. Teorija jazykoznanija. Russistika. Ara-bistika (Linguistic Theory. Russianistics. Arabistics). Sankt-Peterburg: Nauka, 1999.
Xrakovskij 2004 — V. Xrakovskij. Koncepcija diatez i zalogov (isxodnye gi-potezy — ispytanie vremenem) (Concepts of diathesis and grammatical voice (starting hypotheses and test of time)) // V. Xrakovskij, A. Mal'-cukov, S. Dmitrenko (eds.). 40 let Sankt-Peterburgskoj tipologiceskoj skole. Moskva: Znak, 2004. P. 505-519.
