Документ взят из кэша поисковой машины. Адрес оригинального документа : http://www.philol.msu.ru/~otipl/new/main/courses/syntax_rus/fowler.pdf Дата изменения: Tue Feb 16 11:22:01 2010 Дата индексирования: Thu Apr 8 09:37:06 2010 Кодировка:

American Association of Teachers of Slavic and East European Languages

Oblique Passivization in Russian Author(s): George Fowler Source: The Slavic and East European Journal, Vol. 40, No. 3 (Autumn, 1996), pp. 519-545 Published by: American Association of Teachers of Slavic and East European Languages Stable URL: http://www.jstor.org/stable/310146 Accessed: 05/02/2010 15:14
Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at http://www.jstor.org/page/info/about/policies/terms.jsp. JSTOR's Terms and Conditions of Use provides, in part, that unless you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you may use content in the JSTOR archive only for your personal, non-commercial use. Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at http://www.jstor.org/action/showPublisher?publisherCode=aatseel. Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed page of such transmission. JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about JSTOR, please contact support@jstor.org.

American Association of Teachers of Slavic and East European Languages is collaborating with JSTOR to digitize, preserve and extend access to The Slavic and East European Journal.

http://www.jstor.org


OBLIQUE PASSIVIZATION IN RUSSIAN
George Fowler, Indiana University

1. Introduction. Russian verbs can take NP complements in four cases: Accusative, Genitive, Dative, and Instrumental. The Accusative case is the most productive, and can be considered the default case marking for direct objects under any syntactic analysis. The precise status of complements in the other three cases remains an open question. This article addresses one empirical question that bears directly on this issue: the extent to which oblique-complement verbs permit the formation of passive constructions, creating pairs of related sentences such as those given in (1):
(1) a. BopicoB ynpaBJiaeT ia6pHKOHiNSTR.

'Borisov manages the factory.' b. Oa6pHKaynpaBJIaeTca BopIOpcoBbIM. 'The factory is managed by Borisov.' Despite the considerable interest that oblique passivization holds for the syntax of Russian case, there are only scattered references to it in the literature, and various contradictory views have been expressed. Freidin (1992: 206-07) states that it never occurs. Ruziicka (1967: 1730) and Comrie (1980: 217) mention it as a sporadic possibility. Neidle (1988: 169, 172) states that it occurs only rarely and exceptionally, while Siewierska (1988: 254-55) remarks that it is a regular phenomenon throughout Slavic for the Genitive and Instrumental cases.' The present study provides Russian evidence of the general validity of oblique passivization. I ultimately argue that Genitive and Instrumental complements in Russian should be regarded as direct objects, despite their oblique case marking, while Dative complements are fundamentally distinct from other oblique verbal complements, in terms of either syntactic configuration or semantic roles. This conclusion confirms Siewierska's general statement for Russian, while the evidence in support of this conclusion permits us to formulate a more precise conception of "direct object" in Russian.
SEEJ, Vol. 40, No. 3 (1996): p. 519-p. 545 519


520

Slavic and East European Journal

2. The Range of Oblique Verbal Complements. The data considered in this article include only verbs and complements that manifest what Russian grammars call "strong government" (Svedova 1970: 490). I exclude the kind of weak government exemplified in (2), even though standard dictionaries give [KeM] a label for Instrumental government midway down the as entry for pa6omamb 'work'. (2) OH pa6oTaeT MHrKeHepoMINSTR. 'He works as an engineer.' The Instrumental NPs that accompany such verbs are not closely associated complements, but rather circumstantial adverbial NPs-they are not essential to the verb's core meaning or usage, but rather supplement it; moreover they are fully productive-not lexically specified in any way. I also exclude oblique NPs associated with verbs in -ct, such as 3aaumambcH 'study; occupy oneself' or 6oambcH 'be afraid of': they are not candidates for passivization under any circumstances due to the morphological fact that the verb already contains the potential voice marker -ce, which preempts its addition in the creation of a derived passive. Also ruled out are verbs which are commonly identified as copular or copulalike, e.g., 6blmb 'be' or cmamb 'become'. Finally, I will not consider the Instrumental second predicate with three-place verbs, as in (3).
(3) a. MbI C'qwTaeM eroAcc nypaKOMINTR.

'We consider him a fool.' b. Mbi BbI6paJI eroACCnpe3HleHTOMINsTR. 'We elected him president.' These verbs already have an Accusative object, and Russian verbs, naturally, do not simultaneously take two direct objects. Moreover, the Instrumental complement can never become the subject via passivization, so the question of the status of the Instrumental complement with respect to this process does not arise. These exclusions leave us with a core set of two-place verbs that take complements in three cases: Genitive, Instrumental, and Dative. Instrumental-complement verbs form a rather semantically consistent set united by the notion of 'control' in one form or another.2 An inventory of 17 verbs from standard dictionaries is given in Table 1.3 In the standard language all of these verbs obligatorily require that their complements be marked with the Instrumental case-no variation in casemarking is acceptable. Genitive complements, however, usually alternate with the Accusative case, primarily as a result of the semantics of the


Table 1. Instrumental-Complement Verbs BenaTb/- 'manage, be in charge of' BepxoBoAHTb/- 'lord it over' BjianeTb/- 'own, possess' BopotaTb/- 'have control of' RopoKHITb/- 'value' 3aBenoBaTb/- 'superintend' 'misuse' 3JnoynoTpe6JIHTb/3J.oynoTpe6HTb KoMaHaoBaTb/-'be in command of, command' o6naaTb/- 'possess, be possessed of'

'take oBsJaneBaTb/oBJIaneTb npaBHTb/- 'rule over, gover npeABonHTeJIbCTBOBaTb/- ' npeHe6perTt,/npeHe6petb ' pacnoJIaraTb/- 'have at one pyKOBoAHTb/- 'lead, guide, ToproBaTb/- 'deal in, trade ynpaBJIHTb/-'manage, adm of, run'


522

Slavic and East European Journal

Verbs Table 2. Genitive-Complement
'achieve' AoCTHraTb/JOCTHrHyTb, JOCTHqb

)Ka>KaTb/- 'thirst for' wKJaTb/- 'wait for, expect'
)KejnaTb/no)KejiaTb'desire'

HCKaTb/- 'seek' HaKynaTb/HaKynHTb 'buy a lot of'

owKaaTb/- 'expect'
npocHTb/nonpocHTb

'request'

'deserve' 3acny>KHBaTb/3acJIy>KHTb H36eraTb/H36eKaTb 'avoid'

'demand' Tpe6oBaTb/noTpe6oBaTb XOTeTb/- 'want'

object. The Genitive-complement verbs that we will consider are listed in Table 2. In Table 2 the verb HaKynumb 'buy a lot of' is taken as representative of the large number of quantitative verbs in the prefix Ha- (na6pamb 'gather a lot of', Ha2osopumb 'say a lot of', etc.), as well as other genitive-inducing prefixal formations.4 They take the Genitive, as illustrated in (4a), except when an overt quantifier is present, as in (4b). In that case, the complement NP is Accusative, and the Genitive case marking on the quantified noun is due to the quantifier, rather than the verb.
(4) a. OH HaKynIJI KHirGEN.

'He bought a whole lot of books.'
b. OH HaKynIJI KyqYACCKHIrGEN.

'He bought a bunch of books.' Russian monolingual dictionaries (the Slovar' sovremennogo russkogo literaturnogo jazyka [SSRLJ] and others) characterize Genitive-complement verbs inconsistently. A verb like wc&amb 'wait', which takes both Genitive and Accusative complements, is characterized as both transitive and intransitive, depending solely upon the case of the object, while verbs in na-, which take the Genitive more consistently, are labeled as exclusively transitive.5 The dictionaries take this line because they regard the Genitive case with Ha-verbs as an instance of the partitive Genitive. This policy is questionable, since the Genitive is obligatory with Ha-verbs and optional, even highly marked, with ordinary partitive objects.6 Finally, a considerably longer list of representative verbs that take lone Dative complements is given in Table 3. Many verbs of communication and gesture could be added to Kcuamb 'nod' in Table 3: maxamb 'wave', uop2amb 'wink', annao&upo6amb 'applaud', etc. However, these are fundamentally intransitive verbs that need not take any VP-complement at all. They differ radically from most of the verbs in Table 3, which are two-place predicates and therefore "feel" transitive: 3a6u6oeamb 'envy', teuwamb 'bother', Mcmumb 'take revenge against',


in ObliquePassivization Russian Verbs Table 3. Dative-Complement
'favor' 6njaronpaTcTBOBaTb/BepHTb/noBepHTb/- 'believe' BHHMaTb/BHaTb/'heed, hear' [arch., poet.] Bo3pa)KaTb/Bo3pa3HTb 'object' 'injure' BpeanTb/noBpe.aHTb BTOpHTb/- 'echo, repeat' rpo3HTb/npHrpo3HTb 'threaten' nIocaKxaTb/Joca1aHTb'annoy, vex' 'envy' H3MeHMTb/H3MeHHTb 'betray, 'nod' KHBaTb/KHBHyTb 'flatter' JIbCTHTb/nOJIbCTHTb noapa>KaTb/- 'imitate' noaXOJHTb/HnoaoOTH 'suit, fit' nOMOraTb/nOMOIb 'help' nOTaKaTb/- 'indulge'
3aBHIOBaTb/-

523

npeamuecTBOBaTb/- 'precede' npenaTCTBoBaTb/BOcnpeHnTCTBOBaTb

'hinder,impede, hamper'
npHHaajaneKaTb/- 'belong to' npoTHBoaeicTBOBaTb/- 'oppose, counteract' npoTHBOpeHnTb/- 'contradict'

paaeTb/nopaaeTb'oblige'
cjneaoBaTb/nocjneosBaTb follow' 'serve' CnJIyKHTb/nOCJIy)KHTb cogeAcTBoBaTb [biaspectual] 'assist' cooTBeTCTBOBaTb/- 'correspond' conyTcTBOBaTb/- 'accompany' cotyBcTBOBaTb/- 'sympathize' yro>)IaTb/yroaITb 'please, oblige' yrpo)KaTb/- 'threaten' 'satisfy' yaosJBjeTBopaTb/yaoBsJeTBOpHTb

to' be unfaithful

MemaTb/noMemaTb 'bother, disturb, hinder'

and the others all exemplify strong government. Another class of verbs that has been systematically omitted from Table 3 encompasses verbs of command and permission, such as npuca3amb 'order', 3anpemumb 'forbid',
no36solumb

'permit', etc. These verbs take infinitive complements

along

with the Dative NP, which gives them the argument structure of three-place predicates. 3. Passive Constructions in Russian. Now that we have defined the range of verbs to be considered, let us establish the inventory of passive processes that can be applied to them. Following Babby (1993: 5-11), I take passivization to refer to a remapping of a verb's argument structure onto syntactic configuration such that the verb's external semantic role (e.g., Agent) is internalized, i.e., incorporated into the verb itself; the maximum set of constructions in which this occurs in Russian is enumerated in (5). (5) Passive sentences Perfective (based on short-form past passive participle) Imperfective (based on verb in -cs) Participial clauses Past passive participle Present passive participle


524

Slavic and East European Journal

The most canonical and least disputable passive construction is the perfective sentential passive, containing a past participle and a form of the verb 'be', as illustrated in (6). (6) a. Pa6oTHHKHNoM 3aKOHUIJIM KOHCepBaI,ILoAcc. 'The workers finished the conservation works.'
b. KoHcepBai4HI,NOM 6bIJa 3aKOH'ieHa pa6oTH1KaMHINSTR.

'The conservation works were finished by the workers.' [Siewierska 1988: 245] The present analysis is compatible with two conceivable views of passivization: 1) passive sentences like (6b) are derived directly from their active equivalents, such as (6a), through an explicitly syntactic process, as in older transformational grammar (e.g., Babby and Brecht 1975); or 2) passive sentences result from a morpholexical process that adds the suffix -en prior to lexical insertion (e.g., Babby 1993).7 Under either approach, passive and active sentences correspond precisely because the arguments of the verb have been remapped. This is illustrated descriptively in (7), where the direct mapping of the active sentence gives way to the crossing pattern of the corresponding passive.8 (7) a. Active Agent Action Patient b. Passive Agent Action Patient Subject Verb

Subject Verb

Object

Adjunct

The one constant to all passive sentences in Russian is that the notional Patient occurs as the grammatical subject.9 The treatment of the Agent argument is not consistent. In many sentences, it occurs as an Instrumental NP, e.g., in (6b). However, agentless passives are also possible; one example is given in (8).
WTO6bI 6bIJa 3aKa3aHa BOBTOPHHK, (8) Bama KHHiraNOM noJIyIH,Tbee
BoBpeM5 .

'Your book was ordered on Tuesday in order to receive it in time.' This sentence is clearly passive on formal and semantic grounds: even though the Agent is not expressed overtly, it can be inferred that the action was performed by someone; we can readily add an Instrumental phrase such as Mouu compybnuKco 'by my coworker'; and it takes a purpose clause whose implicit subject is identical to the implicit agent of the upper clause.


in ObliquePassivization Russian

525

Passives formed from imperfective verbs by addition of the particle -cH are a bit more disputable.10 I follow Siewierska (1988) in assuming that some verbs in -ca are indeed passives, at least when the action of the verb is to be interpreted as agentive, as in (9).
(9) flocyaNoM MOeTC5Icjiy>aHKoHINSTR.

'The dishes are washed by the servant-girl.' Of course, the vast majority of verbs in -Caare not passives; there are also intransitives (nonagentive), true reflexives, reciprocals, and lexicalized verbs like 6oWmbcH 'be afraid'. A thorough overview of the different types of mappings between semantics and syntax attested among Russian verbs in -Ca is given by Brecht and Levine (1984). In a survey of the typological properties of passive constructions, Keenan (1985: 254) states that there is no formal distinction in Russian between "reflexives"'1 and -cH passives without overt agents, so that the distinction must be made on purely semantic grounds: a verb is passive when an agent is inferred. However, the semantic distinction between agentive and nonagentive -cS constructions-fuzzy and unsatisfactory when no overt agent is present-is in fact reinforced by a formal distinction: -cH passives do not take perfective pairs12(instead, the alternative sentential passive incorporating the past passive participle is used), while other verbs in -cI do have perfective counterparts: intransitives (e.g.,
HaqUHambcFHlHaambca

'begin'), reflexives (y.Mbilambcal/yMbimbCH'wash up'), reciprocals (ecmpeaambcl/ecmpemumbcJ 'meet'), etc. Accordingly,

I assume that -cH passives do "count" in considering the range of oblique passivization. Passive participles do not strictly fit the schematic definition of passivization in terms of the mapping between arguments and grammatical relations, as given in (7), because they have no overt grammatical subject. Nevertheless, they do host an opposition between active and passive formations which is parallel to the distinction between active and passive sentences. We might combine participles and sentential predicates into one all-encompassing voice opposition in two acceptable ways. First, we could simply define the "subject" of a verbal participle to be that NP which imposes gender and number agreement on it. This is intuitively reasonable, since the subjects of clauses also impose gender and number agreement on finite verbs. Sentences lacking subject-verb agreement are impersonal, or subjectless, and thus exhibit default agreement (neuter singular), but participial clauses are never "subjectless", and therefore always show gender and number agreement. Alternatively, we could avoid referring to subject by identifying a passive construction as one in which the Patient argument takes over the unmarked grammatical function ordinarily associated with the Agent argu-


526

Slavic and East European Journal

ment. The choice between these alternatives depends upon one's theoretical inclinations; both formulations pick out the same mapping between argument structure and syntactic relations. Just as with sentential passives, passive participles can be classified according to aspect: past passive participles are formed productively from perfective verbs,13while present passive participles are formed from imperfective verbs.'4 The distribution of passive forms with respect to aspect is summarized in (10). (10) Imperfective Sentential -cs reflexive passive Participial Present passive participle Perfective -en passive Past passive participle

Having defined the inventory of constructions that can be considered passive, let us now consider the extent to which these passive constructions occur with oblique-complement verbs. 4. Passivization of Oblique-Complement Verbs. A number of the Instrumental-complement verbs listed in Table 1 form reliable -cHpassives; examples of these forms are given in (11).15
(11) a. PyccKaa apMHaynpaseJiacb rocyaapeM 13 IleTep6ypra. KyTy30BbIM, c ero lITa6oM, H

'The Russian army was run by Kutuzov, with his staff, and by the sovereign from Petersburg.' [Tolstoj] b. TeM CaMbIM OJIHH H3 HaH6ojieeBbICOKHX ypOBHei AI3bIKa npeHe6pezaemc HIccJIejoBaTeJIaM4. 'In this way, one of the highest levels of language is neglected by researchers.'
c. Pycb aoJiro npaeuJacb BaparaMH.

'Rus' was ruled for a long time by the Varangians.' d. BpeMa OTnyCKa qacTO3aoynompe6JUiemcs CTyneHTaMH. 'Vacation time is often misused by students.'
e. B AMep1Ke HHocTpaHHbIeI3bIKHBoo6iOHe HHKeM He

osiabesawomcn. 'In America foreign languages aren't mastered by anyone at all.' A somewhat greater percentage of these verbs form present passive participles; several examples are given in (12).


in ObliquePassivization Russian (12) a.
...

527

oTpjII, npeoh3ooumeIlbcm6yeAbfiU OTBa)IKHbIM

led by a courageous commander...' [SSRLJ] b. KYTY3oBqpe3 cBoero Jia3yTuHKKa nonyqi4J nepBoro Ho5I6p5I
H3BeTHe, cTaBHinmee
KoAtaHyeAtyFo

KoMaHJIHPOM... a detachment 4...

H4M apMiLo

B O·TI14

6e3BbIxoAjHoe

nojooeKeHH4e.

'Kutuzov through his scout received on November 1 information that placed the army commanded by him in an almost inextricable
position.'

[Tolstoj]
c. HpeHe6pe2aeMbtu' To6oH' MojioJo0H YeneoBeKoKa3arcic OTJIH'HbIM
HOMOL1AHHKOM.

'The young man disdained by you turned out to be an excellent assistant.' The incidence of passive constructions with Instrumental-complement verbs is summarized in Table 4. In Table 4 (as well as in Table 5 below) plus signs indicate that the passive construction occurs for the given verb; minus signs signify that it does not of Verbs Table 4. Passivization Instrumental-Complement Imperfective
-cii

Perfective
-eH Past Pass.

Pres. Pass.

Passive
BeJxaTb/BepxoBoJHTb/BnageTb/BopoqaTb/jOPO>)K4Tb/-

Participle Passive Participle
-

3aBegoBaTb/3JIoynOTpe6JIi.Tb/3JIoynoTpe6H4Tb
KOMaHJIOBaTb/-

+
-

+
+

+

+

o6nagaTb/oBjiaJ1eBaTb/o0Bjaa,eTb

-

npaBHTb/npeJBoxHTen bCTBoBaTb/npeHe6peraTb/npeHe6pes b
pacnoJaraTb/-PyKOBORHTb/ToproBaTb/--

+ +
-

+ +
+

+

+

ynpaBR%Tb/-

+

+


528

Slavic and East European Journal

occur; and question marks indicate hesitation or serious conflict among my informants. Blank spots indicate that no form is expected, e.g., based on the distribution of passive forms spelled out in (10). We will return to the minus signs in section 5 (and note that they represent over 50% of the possible passive forms). There is no definitive printed source for determining whether or not a given present passive participle actually occurs; the data in Table 4 represent informant consensus. Genitive-complement verbs are even more consistent in forming passive constructions than Instrumental-complement verbs. Examples of sentential passives involving Genitive-complement verbs are given in (13), while several participial examples are given in (14). (13) a. luenjb 6bina bocmuzHyma. 'The goal was achieved.' [Riuicka 1967: 1730]

b. OInITb npHI3aAyMajicITeJIOBeK Hai- co6oio, H nopemIJi nICKaTb cxiaCTH4I CeMeHHOH B >KH3HH,3aBCrqLeH,nOBHRMOMy, MeHbIme Bcero OTBHeIIIHHX cJIyiaHiHOCTeH, HOHe aajiaToro, ITO ucKaJZocb, H ceMeHa I )KH3Hb.

'Again the man fell into thought about himself, and he decided to seek happiness in family life, which depended the least, it would seem, on external chance, but even family life didn't provide what was being sought.' [N. V. Selgunov]
c. Ha Hocy 6bIJIa Ha6pocaHa 3eMaI.

'A lot of earth was tossed onto the bow [of the ship].' [Goncarov]
d. Ero BbICTynnJeH4e owcubaemcqBCeMI c orpOMHbIM HeTepneHmeM.

'His appearance is awaited by everybody with great impatience.'
e. Boo6iue, B HaImeM u36e?aaocb Bce, xTO MorFJoJaBaTb JaOMe niHmy Boo6paxKeHHIo.

'In general, in our house everything was avoided that could give food for the imagination.' [Saltykov-Scedrin]
f. Tpe6yeMbiHi 3eIeKT
BBeJeHMH

. . . ocmuaemcS

HaMHnocpefCTBOM

B onpegeJieHHe

nocJIeHIHHXyCJIOBHl. ..

'The required effect . . . is achieved by us through introduction into the definition of the latter the condition. . .'
(14) a. ... yBaxeHHe, aaciaycennoe cojinaTaMHHa nojie 6HTBbI...

'. . respect earned by soldiers on the field of battle . .'
b. MbI noJIylHJIH e.jaeMbie HaxaJIbCTBOM pe3yJIbTaTbI.

'We obtained the results desired by our superiors.' The examples in (13-14) have been selected because informants agree that in the corresponding active sentences the Genitive case would be most
natural, e.g., for (13a) oocmuzHymb
4eauGEN

'achieve the goal' is more


in ObliquePassivization Russian Verbs of Table 5. Passivization Genitive-Complement Imperfective
-ca Pres. Pass.

529

Perfective
-eu Past Pass.

Passive
JocTHraTb/JIOCTHrHyTb, OCTHqb +

Participle Passive Participle
? + +

)KaxKaaTb/wKmaTb/-

--

>KeJlaTb/HO>KeJlaTb 3acjiy)KHBaTb/3aciy)KHTb H36eraTb/H36e>KaTb
HCKaTb/HaKynaTb/HaKyHTb o)KHUaTb/-

+ +
+ + +

+ ? +
? +

+
?7 +

+
? +

npocHTb/nonpocHTb Tpe6oBaTb/noTpe6oBaTb
XOTeTb/-

+
--

+

+ ?

+ +

acceptable than ?0ocmuzHymb 14eIbAcC.16 Example (13b) is especially instructive. First, ucKamb 'seek' is paired with the abstract noun ctacmue 'happiness', with which we would expect the Genitive. Second, and most convincing, the active and passive counterparts both occur in the same sentence, and the active example is in the Genitive. Moreover, some speakers, including two of my informants, strongly prefer the Genitive case with most of the variable verbs in Table 5, and their passivization judgments do not diverge significantly from those of the other informants, for whom the Genitive is less predominant (but still possible in all cases). A summary of the data is given in Table 5. With both Instrumental- and Genitive-complement verbs, we have found there to be considerable potential for the creation of passive constructions. But when we turn to the set of Dative-complement verbs, we find hardly any trace of any passive forms. Consider the typical examples in (15) and (16), where the ungrammatical (b) sentences have been created by brute-force passivization of the normal (a) sentences:
AeiecTBITeJIbHOCT". (15) a. Ero cBseeHstH COOTBeTCTBy1OT

'His information corresponds to reality.'
b. *JeHiTBHTeJ1bHocTb COOTBeTCTByeTc5Iero cBeAeHHIMH.

*'Reality is corresponded to by his information.'
(16) a. HBaHOBnoMor HamieMycoTpyJAHKy.

'Ivanov helped our associate.'


530

Slavic and East European Journal b. *Ham COTpyaXHHK noMOceH HIBaHOBbIM. 6biji

*'Our associate was helped by Ivanov.' Passivization of Dative complements is clearly counter to the intrinsic grammatical system of Russian; informants presented with such examples have trouble even deciphering the intended meaning. Clearly the status of passivization with Dative-complement verbs is an order of magnitude different from the situation with Instrumental- and Genitive-complement verbs.18 I will return below to the reason why Datives should be so different in this respect from other oblique-complement verbs. The ungrammaticality of (15b) and (16b) notwithstanding, a few scattered traces can be found of passive constructions based on Dativecomplement verbs from Table 3. Three sentential examples are given in (17), and three participles are given in (18).
BnoJIHe, AaBHO (17) a. Tbi omoMu4eie

A KeM H KaK-He Bce Jib paBHo?

'You are avenged fully, long ago, But by whom and how-does it really matter?' b. Kopa6JIb6biJino6pexcOen... 'The ship was [irreparably] damaged. . .'

[Lermontov] [Fedin]

c. He:KcaHOB, BepoaTHo, caM He noAo3pesaJI, Ao KaKoi cTeneHH ero caMoJIio6He 6bIJIo noJIbineHo ee o6xocAjeHHeM c HHM. 'Nezdanov, probably, didn't even notice himself to what extent his pride was flattered by her behavior with him.' [Turgenev] K (18) a. CBAmIIeHHHK, npeomuecmeyeMblu JabAKOHOM, npH6JIHaKaJIca IepKBH.

'The priest, preceded by the deacon, approached the
church.' [Cexov] b. '. .. yzpoxaeAaq npi o6cTpeJie CTOpOHa... the side threatened during shelling. . .' '.. [RG 1:668] c. ... CTHxoTBopeHHII, 3HaeMbIXBCeMH Hal3yCTb H CTOJIb HeylaqHo noMHHyTHO noOpaMaeAbtx. . . 'poems, known by heart to all and continually imitated so unsuccessfully. . .' [Puskin]

Vinogradov (1982: 186) states that present passive participles such as
those in (18) were deliberately and artificially created in the second half of the 18th century to translate certain French participles, e.g., npeouecmbyeMbli corresponding to French precede. Several of them have remained in the language as isolated forms associated with the paradigms


Oblique Passivization in Russian

531

of a few verbs, but they have never achieved significant productivity; a few other isolated examples from Vinogradov are given in (19). (19) a. ... eKcenI xopomo ycJaycen 6bITbxoqeT... 'if he wants to be well served' [Vinogradov 1982: 186; "Trubnja", 1769; cf. Fr. servir] b.. . ecJI TbIycTyniHIb MHe3CTOHHo, yzpowcaeMyio
CIIrI3MyHJIOBbIM

BJIaCTOJIKO6HeM...

'.. . if you yield to me Estonia, threatened by Sigmund's lust for power. . ." [Vinogradov 1982: 186; Karamzin; cf. Fr. menacer] Vinogradov classes such examples together with phrasal calques from French, such as oH xeacmajacHuMembnepo 3ojomoe, cf. French qu'il se vantait d'avoir une plume d'or. Some such translations caught hold in Russian, such as npeotuecmeyeMublu and yzpocaeMbtu; others did not, such as ycjiymce in (19a). The examples in (17) have separate explanations. The verb Mcmumb/ omoMcmumb 'avenge; take revenge against', although currently a Dativecomplement verb, can take an Accusative object as an archaic variant. Since the only examples I have uncovered are as old as the one in (17a), it is reasonable to suggest that they represent the older variant of the verb's government. On the other hand, noepeoumb 'injure' participates in two
aspectual pairs: epeoumblnoepeoumb and no6pexoamb/no6peoumb. The

first pair takes the Dative, but the second takes the Accusative. As the distinct meaning of the latter is 'irreparablydamage an inanimate object', we can conclude that the participle no6pexMeHHbtiuis derived from nospeoumb from the second pair, and the apparent anomaly in (17b) disappears. Finally, the verb .jbcmumblno.abcmumb has an archaic variant which takes the Accusative, as attested in the expression ibcmumb ce6R Ha)eacouio 'console oneself with the hope', or such examples as (20):
(20) a. Ce6a H
Bsac HaAeaKloi Jibimy, xTO CbIH Bamn XKIB.

'I console myself and you with the hope that your son is alive.' [L. Tolstoj]
b. OTIaJIeHHoe 3aBoeBaHi4e MrJIO JIbCTIHTb MecToJlO6I4BOrO FeHpiHxa, HOOHHe I4MeJ cpeAcTB OTnpaB4Tb BOiiCKO.

'A distant conquest could flatter the vainglorious Genrix, but he did not have the means to send an army.' [N. Polev] has a separate entry in several Russian dictionarMoreover, noalbuteHHMbl ies, where it is defined as 'satisfied by something flattering to oneself', which is not a directly passive participle within the paradigm of Jibcmumb/


532

Slavic and East European Journal

noAbcmumb, but rather represents a departicipial adjective (Babby 1993:

19-20). Thus, this example too can be dismissed as archaic or misleading. It has been shown in this section that a significant portion of Instrumentaland Genitive-complement verbs permit passivization, while traces of passivization with Dative-complement verbs are minimal and generally explicable as archaisms or loan translations from French. In this respect, Instrumental and Genitive complements pattern with Accusative direct objects, while Dative complements do not. 5. Unattested Passive Forms. If passivization is in principle possible with Instrumental- and Genitive-complement verbs, but excluded for those verbs that take Dative complements, why are passive forms not attested for all of the verbs in Tables 1 and 2? In section 6 an analysis is presented under which the former are classed as syntactic direct objects, differing from Accusative objects only in surface morphological case marking. The oblique passivization data presented thus far demonstrate that Instrumental and Genitive complements could be direct objects. In this section it is shown that oblique passivization respects general limitations on passivization of Accusative direct objects, and that the extension of the class of direct objects to encompass these oblique NPs is motivated not only positively (the occurrence of passive constructions), but also negatively (passivization is limited in the same way with both Accusative and Instrumental/Genitive objects). First, it is worth noting an important methodological point: in cases of linguistic variation where one variant represents the occurrence of a certain process (in this case, passivization) and the other represents the failure of that process to occur (the absence of certain potential passive forms in Tables 1 and 2), it is almost invariably preferable to assume that the grammar allows for the process (i.e., passivization is possible with Instrumental-and Genitive-complement verbs), while other extra-grammatical factors (in this case, the semantics of the verb and Patient) may interfere with the process. The reverse hypothesis is difficult to countenance: that passivization should be grammatically excluded with Instrumental- and Genitive-complement verbs, but something-semantics, pragmatics, speakers' whim-triggers the formation of a variety of ungrammatical passive forms, overcomes the grammatical prohibition against them, and "bootstraps" them to acceptability. However, when a process cannot occur (here, passivization with Dativecomplement verbs), and the apparent exceptions can be explained away on historical or other grounds, we may be confident that the grammar rules out the process. Passivization is not a universal litmus test for determining whether or not a verb takes a direct object; transitivity is a necessary, but not sufficient condition for passivization. Many uncontroversially transitive verbs in Rus-


in ObliquePassivization Russian

533

sian fail to exhibit any trace of passivization. So, for example, the passive sentence in (21) is ungrammatical:19
(21) *3TH (PaKTbI 3HaIOTCS BCeMI HaUIHMI CTyjeHTaMM.

'These facts are known by all our students.' Two semantic aspects of the VP in (21) are relevant: the predicate contains an atelic stative verb, and the object is an unaffected Patient. Briefly, stative predicates are opposed to dynamic predicates, in that the character of the situation denoted by the verb does not vary internally (Comrie 1976: 48-51). Thus, 3Hamb 'know' refers to an essentially static situation: the subject's knowledge is fixed in any sentence containing this verb. Statives are canonically imperfective in Russian; when they become perfective, they must generally become dynamic as well. The verb y3Hamb 'find out, learn' is a perfective verb which differs from 3Hamb along the stative/dynamic axis, but minimally in other semantic respects: it describes a situation in which an initial state of non-knowledge, or unconfirmed knowledge, gives way to a state of knowledge. The perfective is also telic, i.e., it builds toward a terminal point (Comrie 1976: 44-48), in the case of y3amb, the attainment of knowledge. Unaffected Patients are those in which the action of the verb is not transmitted directly onto the Patient; for example, the object of see is not affected by the verb in the situation described by the situation Oleg saw the vase, whereas it is directly affected in Oleg broke the vase. Not coincidentally, these two semantic categories influence passivization in a host of different languages. A useful summary of various typological considerations involved in passivization is provided by Keenan (1985), who establishes a number of implicational universals which indicate that passivization of VPs incorporating stative verbs and/or non-affected objects is typologically secondary to other types of passivization. Keenan generalizes that all languages which have a recognizable passive construction apply it to activity verbs, but not all apply it to stative verbs. Moreover, although not stated in such terms by Keenan, it is clear that the stative/dynamic opposition is not an all-or-nothing proposition. Some languages permit any syntactic direct object whatsoever to be passivized, even with highly stative verbs, as in the Bantu language Kinyarwanda, where the highly stative active clause in (22a) gives rise to the passive equivalent in (22b): (22) a. Ishaati i-fit-e ibifuungo bibiri. shirt it-have-ASP buttons two 'The shirt has two buttons.' b. Ibifuungo bibiri bi-fit-w-e n'ishaati. buttons two they-have-PASS-ASP by shirt 'Two buttons are had by the shirt.' [Keenan 1985: 250]


534

Slavic and East European Journal

English permits passivization of some, but not all stative verbs. Verbs characterized as "highly stative" by Keenan, such as cost, weigh, be, and possessive have, generally do not permit passivization in English, as in (23a-b). Yet other semantically stative verbs do permit passivization, as in (23c-d). (23) a. b. c. d. *Two *One These These buttons are had by the shirt. dollar was costed by the newspaper. facts are known by all our students. apartments are owned by landlords who live in London.

Keenan further notes that "distinct passives in a language may vary according to degree of affectedness of the [passive] subject. . ." (1985: 269), although he characterizes this conditioning factor as less important than others. This variation is demonstrated by the Russian translations of the English sentences cited previously to illustrate the distinction between affected and nonaffected Patients: (24) a. Ba3a 6bIja pa36iTa OJIeroM. 'The vase was broken by Oleg.' b. ?*Ba3a 6biJIayBHIeHaOJIeroM. 'The vase was seen by Oleg.' The first example, containing an affected Patient, is markedly better than the second, in which the Patient is nonaffected. Note that the difference in acceptability is not correlated with the stative/dynamic opposition, as the perfective yeuOemb is not stative (in contrast to its imperfective partner 6uMemb), but is rather a telic achievement verb, and thus eligible for passivization. Russian, then, is a language in which this factor is relevant to the formation of passives, although it is not important in English. These distinctions enable us to account for most of the missing passives in Table 4. Some of these Instrumental-complement verbs are clearly
statives: 6isaoemb 'own, possess', oopoMcumb 'value', pacnonozamb 'have

at one's disposal', etc. But what is the difference among the various verbs denoting 'control, manage, run"? Statives and dynamic activities, as strictly semantic notions, can conveniently be viewed as prototypes (Langacker 1987). Statives are generally atelic, which accounts for the fact that they do not naturally form aspectual pairs. Certain verbs from this set, such as eeoamb 'manage, be in charge of', are associated with the stative prototype, and thus reflect a conceptualization along the lines of 'be in control/charge of' or 'have the status of being in charge'; other verbs are associated with the opposite pole, and thus conceptualize the situation as 'actively engage in the managing/control of'. The dynamic prototype in-


in ObliquePassivization Russian

535

volves an affected Patient, while the Patients of stative verbs are unaffected (the verb establishes a static relationship between the subject and object, rather than applying an action directly to the Patient). A few
verbs, such as pyKco6Oumb 'lead, guide, direct' or 3aeeoosamb 'superin-

tend, be in charge of', can be regarded as associated cognitively with both poles, and therefore they have spotty passive properties.20 One possible problem is mopzosamb 'deal, trade', which is clearly an activity. Most likely it really does not belong in this list. It often occurs with no NP complement, as in (25): ToproBaJiI. (25) Eme HaEmH npaneJnbc KHTaeM 'Our forefathers were already trading with China.' We can regard mopzosamb as a simple intransitive verb; any Instrumental NP associated with it is a circumstantial adjunct, just as in (2). The sole difference is that by virtue of its lexical meaning, mopzosamb tends to take such an adjunct NP rather more frequently. The same approach also helps us understand the missing Genitives. The verbs xomemb 'want', a.mMcamb 'thirst for', and cejiamb 'desire' are statives with unaffected objects; hence, they do not naturally form passives. The verb xcamb may be either a stative or an activity, but in any case the object is unaffected. A further consideration is morphology. Present passive participles are heavily influenced by morphological factors. For example, mcaamb cannot readily form one because the lack of mutation of the stem-final consonant, combined with the non-syllabic stem, forces an unpalatable choice between two awkward-sounding forms: *"WceXMbiu *XMOMblU.Zaliznjak (1977: and lists additional morphological limitations on the formation of this parti86) ciple; a case in point is npa6umb, which belongs to a stem-type (the suffix -u) characterized as "difficult" (Russ. 3ampyoHumejibHo).The occurrence of a clausal passive in (llc) demonstrates that this verb is semantically suited for passivization; therefore the absence of a present passive participle must be ascribed to this well-established morphological difficulty. Simi'avoid' does not form a past passive larly, the perfective verb u36ecMamb it would be semantically suitable (and the imperfective participle, although pair exhibits some evidence of passivization). One contributing factor is that the formation of participles from verbs of this stem type can be morphologically awkward; RG (1: 671) notes that the forced stress shift to the syllable preceding a stressed -a in the infinitive, as in u36ecamb *u36etanHbti, is difficult. Moreover, the entire stem class is overwhelmingly intransitive, and there is considerable vacillation in the formation of participles from verbs in this class which are converted to transitive through addition of a prefix.


536

Slavic and East European Journal

Thus, we may conclude from the data in section 4 that obliquecomplement verbs comprise not one, but two separate groups. Verbs which take Instrumental and Genitive complements pattern together with ordinary transitive verbs which take Accusative direct objects, while Dativecomplement verbs are separate. In the final section we can take up the question of why this should be so. The failure of some Instrumental- and Genitive-complement verbs to form passives reflects both natural processes governing the formation of all passives in Russian and certain morphological facts about individual verbs which take oblique complements.21 6. Discussion. Passivization is a morpholexical operation that relates argument structure to syntactic organization; see Babby (1993) for a thorough treatment of the issues that arise. Accordingly, the set of verbal complements eligible for rearrangement must be characterizable in morpholexical terms. In order to account for the oblique-passivization data spelled out in this article, two provisions must be made: 1) Dative complements of twoplace verbs must be distinguished formally from other oblique-case complements (Genitive and Instrumental); and 2) Genitive and Instrumental complements must be associated with Accusative complements. We could simply state the necessary generalization in terms of surface case marking: "Dative complements do not undergo passivization". However, this statement is strictly stipulative; a more satisfying account must suggest why the Dative case is special. Such an explanation can be sought by identifying Dative NPs as syntactically distinct from Accusatives, or as semantically unique. Under a syntactic account, two general possibilities have been suggested in the literature, as represented in (26). (26) a. Franks 1995 VP :V ^<
V NPr ^"NPDAT
NPACCIGENIINSTR

b. Bailyn 1995 VP
VI V NP NPACCIGENIINSTR
V NPDAT

The structure in (26a) postulates that Dative NPs are higher within the syntactic tree than Accusatives (Franks 1995), while the reverse proposal in (26b) has also been argued for (Bailyn 1995). In either case, it is plausible to assume that the semantic value of Dative indirect objects is mapped to the distinctive syntactic configuration. The specific syntactic analysis of Dative case assignment depends upon theoretical and empirical factors which would take us far beyond the scope of the present article. For our purposes, it is sufficient to note that Datives are formally distinct, that Genitive and Instrumental complements occur in the position of Accusative direct objects, and that Dative NPs have the same semantic range in


in ObliquePassivization Russian

537

two-place verbs of the class listed in Table 3 as when they are canonical indirect objects. Let us now consider the last point. Datives cooccur with Accusative direct objects naturally in three-place verbs, as in sentences like those in (27).
(27) a. IHBaHNOM KHIIrYAcC 6paTYDAT. aJI

'Ivan gave the book to his brother.'
b. 4BaHNoM KyniHI 6paTyDAT KHHrYACC.

'Ivan bought his brother a book.' Two canonical semantic functions of the Dative are illustrated in these examples: the Dative NP in (27a) is a Recipient, or indirect object proper, while in (27b) it is a Benefactive. It is possible to unite these two functions into one vague category as well, but the question of whether there is one universal semantic invariant for the Dative case is not crucial for the present argument. The point illustrated by (27) is that the Dative case is semantically distinct from the direct object, and that both direct and indirect objects can cooccur in three-place predicates. The same semantic value of Recipient/Benefactive occurs consistently in verbs which take a lone Dative complement, as enumerated in Table 3. For example, the complements of noMozamb/nomotb 'help', Abcmumb/
noAscmumb

'flatter', no6xo6umblnoOoumu

'suit', CJaycumb/nocanyJcumb

'serve', yzoxc6amblyzo6umb 'please, oblige' are all clear Benefactives, while the complements of Ku6amb/Ku6Hymb 'nod' and other verbs of communication can be thought of as a variety of Recipient. On the other hand, the complement of verbs such as 6ocac6amblocaaumb 'annoy, vex', Metuamb/
nometuamb 'bother', HaooeaamblHaooecmb 'tire, "feed up" ', or y2pocamb

'threaten' could be termed a Malefactive, i.e., a Benefactive with a minus sign instead of a plus sign. If the arguments in a three-place predicate like aamb 'give' are mapped to the syntactic constituents as in (28a), which follows the model of (7), then lone Dative complements must reflect the mapping in (28b). (28) a. Three-place predicate Agent Action Patient Recipient/Benefactive Subject Verb Object Indirect Object

b. Dative-complement predicate Agent Action Recipient/Benefactive Subject Verb Indirect Object


538

Slavic and East European Journal

Syntactic indirect objects are automatically marked with the surface Dative case, which can be viewed as the default selected in the syntax for arguments in this class. In this respect, indirect objects differ sharply from direct objects, which display not only the default Accusative case marking, but also lexical case marking for Genitive and Instrumental (Babby 1984). While Dative complements are syntactically and semantically distinct from Accusative direct objects, Genitive and Instrumental arguments are not distinct in either respect. Semantically they are indistinguishable from ordinary Patients (this is precisely the idea behind Babby's lexical case marking), while syntactically they do not cooccur with Accusative objects. (Although Instrumental NPs are often found alongside Accusative objects, as in H6sa pe3aJ xie6 HOJCOM 'Ivan cut the bread with a knife', these are adjuncts and not candidates for direct object status.)22 The data on oblique passivization provide substantial support for the general theory of case elaborated by Leonard Babby during the 1980s.23 Under this theory, grammatical cases such as the Russian Accusative are automatically assigned to certain syntactic configurations. However, individual lexical items may supersede the structural default by superimposing their own specific case requirements. Thus, the verbs in Table 1 obligatorily overrule the default Accusative and require the Instrumental, while the verbs in Table 2 superimpose the Genitive. The relation between lexical and configurational case assignment can be summarized as in (29). (29) a. Agent Action Patient Recipient/Benefactive Object Indirect Object Acc i Gen Instr Dat Structural Default Lexical Superimposition

Subject Verb Nom

Under this account, Genitive and Instrumental complements of verbs which assign oblique lexical case are eligible for passivization because in syntactic terms, they are actually direct objects, with a superficial veneer of oblique morphological case.24Following Franks (1995: 33-34; 349; forthc.) and a host of general syntactic literature on passivization, I assume that passive morphology "absorbs" the case assigned to the object; it moves to subject position in order to receive case. In terms of (29), this means that the structural default case is absorbed. In cases of oblique passivization, elimination of the structural default case assignment by the verb means that


in ObliquePassivization Russian

539

lexical superimposition cannot take place: the input to that process has been eliminated by case absorption.25 If Instrumental and Genitive complements are actually direct objects, they should exhibit other syntactic behavior which associates them with direct objects and differentiates them from indirect objects. Just such behavior is illustrated by facts involving control of second predicates, or predicate nominals. Nichols (1981: 68) observes that the only possible controllers are subjects, direct objects, and "inverse subjects";26 simple examples are given in (30). (30) a. OH meJi BeceJIbIi. 'He walked along happy.'
b. CIIhTaIOTezo reHmeM.

'They consider him a genius.'
c. MHe 3Jecb BeceJIo aKJIOCbpe6eHKOM.

'Life was happy for me here as a child.'

[Nichols 1981: 68]27

Second predicates controlled by direct objects occur in two forms: a default Instrumental and an agreeing Accusative, as illustrated in (31).
(31) a. MbI BIUejJI ezo cnlI4M,INSTR.

'We saw him sleeping.'
b. CHaxaJia uatuuny B3BseIemIHBaT nycTyioACc.

'First they weigh the truck empty.' If Instrumental and Genitive complements are actually syntactic direct objects, as argued here, they should potentially be able to serve as controllers for second predicates. The possibilities are somewhat limited, as the majority of verbs do not support the additional predication required for this structure.28However, whenever it is possible to construct sentences which are not semantically anomalous, they are indeed well-formed; two examples of second predicates controlled by oblique objects are given in (32).
(32) a. OHH 136eraJIi
eZOGEN nIIbHOrOGEN/?nIbAHbMINSTR.

'They avoided him when he was drunk.'
b. OHH npeHe6peraJn
InbHOHINSTR. eOINSTR

'They disdained her when she was drunk.'29 On the other hand, Dative complements cannot control second predicates under either agreement pattern, as illustrated in (33).
(33) a. OHH 3aBHJoBajI e.JyDAT *IIbYIHOMyDA 7/*nIIbHbIlMNSTR.

*'They envied him when he was drunk.'


540

Slavic and East European Journal b. OHH CJny)KHJI euyYDAT *IIbIHOMyDA 7/*IInbHbIMINSTR.

*'They served him when he was drunk.' Thus, Dative complements are distinguished from Genitive and Instrumental complements: only the latter are empowered to control second predicates. This is a central property of Accusative direct objects, and serves as powerful confirmation that the oblique complements of Genitive- and Instrumental-assigning verbs are in fact direct objects. 7. Conclusion. This article has investigated the extent to which the phenomenon of oblique passivization occurs in Russian. It has been shown that, as a rule, Genitive- and Instrumental-complement verbs do indeed form passives, while Dative-complement verbs do not. Missing passive forms can be accounted for by typological factors which affect the formation of passives in many languages: stative verbs and unaffected Patients tend to inhibit the formation of passives. A theory of case assignment such as that of Babby (1984) or Fowler (forthc. b) permits a principled account of how grammatical case can be superseded by lexical case due to specific properties of individual lexical verbs. Viewed in this light, the phenomenon of oblique passivization is limited to structural objects, regardless of their surface case marking. Strong confirmation of their object status is provided by their ability to control second predicates. Dative complements are structurally distinct from direct objects, Accusative or oblique. The exact nature of the structural distinction between Datives and other NPs was left unspecified. The data considered in this paper enable us to give a new and unified definition of "transitivity": a transitive verb is one which takes a structural direct object, regardless of its surface case marking. This definition resolves a long-standing conflict in traditional Russian grammars, where transitivity is keyed to the surface case of the verbal complement: Accusative-complement verbs are transitive, as are those whose complements bear the partitive Genitive, the Genitive of negation, or the quantitative Genitive due to the prefix Ha-; but verbs like ucKamb 'seek' or u36ewcamb 'avoid' are not regarded as transitive, nor are Instrumental-complement verbs such as ynpasnAimb 'manage'. This article has demonstrated that the latter two types can profitably be regarded as transitive as well, governing direct objects that happen to be marked in an oblique surface case. The occurrence of passivization is the best evidence of direct object status, but strong confirmation is provided by the ability of Genitive and Instrumental complements to control second predicates.


Oblique Passivization in Russian

541

NOTES
1 Siewierska gives four examples from Polish, but adduces no additional data or literature to confirm her assertion. 2 This vague lexical notion of 'control' cannot simply be superimposed on the syntactic category of direct object to trigger the Instrumental case automatically. Rather, the common semantic thread shared by these verbs is part of the organization that speakers impose on relatively arbitrary lexical information. Full justification of this claim is beyond the scope of the present article, but it is argued at length in Fowler (forthc. b). The assertion that the semantics of 'control' makes the Instrumental "predictable" can be confronted with one piquant counter-example: the verb KoHmpoJaupo6amb, which takes the Accusative case. However, this example is a trifle unfair, because it really means 'subject [an individual] to control' rather than merely 'control', and could thus plausibly be excused from the list in Table 1. At any rate, the important point is that semantic generalizations of this sort are best reserved for the mental organization of the lexicon, rather than employed as cornerstones of formal syntax. 3 Table 1, as well as the subsequent inventories of Genitive-and Dative-complement verbs, is intended to be fairly complete and fully representative, but not necessarily exhaustive. Verbs are given as the traditional imperfective/perfective aspectual pairs, despite the fact that this lexicographic tradition obscures the essence of the relation between perfective and imperfective verbs; cf. Fowler (forthc. a); dashes indicate that the expected pair does not occur. Aspect is relevant chiefly in considering the various passive forms that arise for each verb. It is interesting to note that so many oblique-complement verbs are unpaired. 4 Not only are the established verbs (i.e., recorded in standard dictionaries) in the category too numerous to enumerate, but new examples can be formed so productively from existing transitive verbs that they form a completely open class. 5 For example, SSRLJ characterizes HaKynumbfrom (4) as "nepex. (tmo u tezo)". The variation in government registered in the dictionary refers precisely to the pattern illustrated in (4): SSRLJ, which is generally rich in textual examples of usage, gives not a single example for verbs in Ha- with this quantitative meaning in which an NP without an overt quantifier occurs in the Accusative case; the Accusative is reserved for overt quantifiers, as in (4b). 6 RG (1980, 1: 614) accepts the idea that verbs are still transitive even when negated objects are Genitive; and it even refers to verbs which host an active Accusative/Genitive 'wait', npocumb 'request', or xomemb 'want', as "transitive". variation, such as jdcaamb 7 I incline toward a mixed morphological/syntactic analysis along the lines of Franks [forthc.]; however, nothing here crucially depends upon this assumption. 8 This diagram was suggested by discussion in Brecht and Levine (1984: 118-19); it is also a variant of the "theta-grids" adopted by Babby (1993). 9 Other languages, including Polish and Ukrainian among the Slavic family, exhibit various types of impersonal passives where this remapping is not found. 10 Contrasting views of the issues surrounding reflexive passives are given in Siewierska (1988) and Gerritsen (1988). Siewierska concludes that a reflexive passive must be accepted for Russian (though not for all other Slavic languages), in view of the possibility of expressing the Agent overtly via an Instrumental phrase. Gerritsen argues that the term "passive" is a misnomer when applied to verbs in -cs in Russian, in view of the variety of semantic possibilities. 11 Keenan uses "reflexive" as a catch-all term for "everything except passive". 12 RG (1: 616) includes a note to the effect that it is barely possible for some -cs passives to form a perfective. None of the examples given is particularly convincing as a passive, e.g., CTaKaHbI. IOJIbIOTCi KHIITKOM (i) CKopo H33TOrO caMOBapa KPYTbIM


542

Slavic and East European Journal 'Soon the glasses will be filled up with fiercely boiling water from this samovar' [Kataev] Past passive participles can also be formed, in principle, from simplex (unprefixed) imperfective verbs. I will ignore this possibility here, because: 1) the productivity of this formation is somewhat controversial; and 2) only one of the unprefixed imperfective Genitive- or Instrumental-complement verbs considered in this paper forms such a participle; cf. fn. 17. Apparent present passive participles formed from perfective verbs, such as 6onycmubltu 'allowable', are lexicalized adjectives of participial origin; cf. Townsend (1975: 235). Sources are identified for textual attestations; examples with no indicated source were provided by informants. The Genitive-complement data can be a bit more difficult to evaluate, because many of the verbs occur in active sentences with both Accusative and Genitive complements, depending on the semantics of the NP (greater definiteness, abstractness, specificity, etc., evokes the Accusative, while less of these semantic properties tends to encourage the Genitive). Miloslavskij (1981: 76-77) goes so far as to elevate this Accusative/Genitive variation to the status of a separate case, dubbing it the Attendant case (fcOameabHblinabec). However, if the claim is not made that passive sentences are derived directly by syntactic transformation from underlying active sentences, then it is not necessary to identify the case of the unpassivized NP, since it occurs only as the subject. The fact that it is so difficult to formulate rigorous rules for the case of the complement to certain verbs (Icaamb 'wait for, expect', ucKamb 'seek', etc.) is in itself an argument that the complements have the same syntactic status, regardless of the surface case marking; the variation in case comprises a semantically meaningful opposition only if that is the only variable involved. Some of my informants were willing to create and use a past passive participle from
ucKamb 'seek'.

13

14 15 16

17

18 I refrain from repeating the list of verbs from Table 3 in a separate table here, as there would be nearly uniform minus signs; the rare attested form are discussed below. 19 Note, however, the passive participle derived from this same verb by Puskin in (18c); this usage is yet another French-influenced archaism. 20 Syntax of the generative type is generally taken to reflect discrete categories and choices, while semantics may involve more complicated or fuzzier oppositions. For example, the Russian Genitive of negation is simple in plain morphosyntactic terms: the object of a negated verb is either Accusative or Genitive. Yet the semantic trigger for this simple formal dichotomy is immensely complex, as demonstrated most vividly in Timberlake (1975) and documented in exhaustive detail in Mustajoki (1985), Mustajoki and Heino (1991). Similarly, the dichotomy at issue here is discrete: a passive form either does or does not occur. The variation between speakers reflected in Table 4, as well as the borderline productivity of some passivization processes discussed here, shows that this simple morpholexical process is the tip of a much more complicated semantic iceberg. 21 The relatively small number of oblique-complement verbs makes these isolated morphological factors more prominent here than in the language as a whole. 22 There is exactly one exception: the verb JuuMumb, which takes both Accusative and Genitive arguments, as in (i):
(i) CysbHiNOM JIHInHJIH eroACC rpaKlaaHCTBaGEN.

23

'The judges stripped him of his citizenship.' This example is problematic for any theory of case-assignment in Russian. I assume that since only the Accusative object can passivize, zpawcaaHcmeais not the same kind of lexically oblique direct object as the other examples considered in this paper. Instead, it must be a true oblique argument. An excellent illustration of the kind of data Babby's case theory is designed to account for


Oblique Passivization in Russian

543

24

25

26

27 28 29

is given in Babby (1984), where it is presented in a relatively non-theory-specific way; a survey of its application to a range of problems is provided in Freidin and Babby (1984). While Babby has continued to develop his approach to case in more recent publications, and others have extended it in various respects (e.g., Bailyn 1995; Franks 1995; Fowler forthc. b), the crucial conceptual core is already in place in these two articles. Indeed, it is not obvious that there is any principled reason why the Dative case could not also be assigned lexically to direct objects. Examples of passive phenomena associated with Dative-complement verbs were given in (17) and (18), and were explained away as archaisms or artificial elements in modern Russian. If that explanation is unappealing, the alternative of exceptional lexical case superimposition is also conceivable. An anonymous SEEJ reviewer asks why there is no violation of the Projection Principle, which states generally that lexical information must be respected at all levels of the syntactic representation of a sentence. The verbs in Tables 1 and 2 lexically specify an oblique case, but it is suppressed under passivization. Several answers are possible, but one is the following: lexical case specification is a morphological replacement operation, which states that syntactic Accusative case is replaced by morphological Instrumental or Genitive case. If syntactic Accusative case is absorbed by passive morphology, which is a well-accepted facet of the analysis of passive constructions, then this morphological replacement operation is never triggered. In other words, lexical case marking does not state that the verb must assign Instrumental or Genitive case to its object; rather, it states that the object, if it is case-marked, must not appear in the structural default Accusative case. Inverse subjects are Dative NPs in the function of logical subject (Russian cy6beKm, as opposed to no6eAewau4ee 'grammatical subject'). It is a controversial question whether or not they should be viewed as subjects in the formal grammatical sense, and well beyond the scope of this article. Controllers are italicized. Note that the range of possible controllers is the same for both nouns and adjectives as predicate nominals. Nichols states (1981: 70) that her corpus included only one oblique controller, but it was the prepositional phrase c HUM 'with him', and this is irrelevant to the issue at hand. Informants prefer the agreeing form in (32a), but we can be certain that the morphological forms are Genitive and not Accusative because ue6ezamblue6ecamb obligatorily governs the Genitive. The Genitive nbRHozo cannot be a postponed NP-internal modifier rather than a second predicate, for two reasons: first, pronouns generally do not take NPinternal adjectives; and second, the English gloss accurately captures informant judgments as to the meaning of this sentence, which corresponds only to a second predicate. In (32b) we cannot tell whether the Instrumental case of the second predicate is an agreeing or default form, but it does not matter: the crucial point is that it can be controlled at all.

WORKS CITED
Babby, Leonard H. 1984. "Case Conflicts and Their Resolution". Cornell Working Papers in Linguistics 6: 1-21. Babby, Leonard H. 1993. "A Theta-Theoretic Analysis of -En- Suffixation in Russian". Journal of Slavic Linguistics 1.1: 3-43. Babby, Leonard H. and Richard D. Brecht. 1975. "The Syntax of Voice in Russian". Language 51: 342-66.


544

Slavic and East European Journal

Bailyn, John. 1995. A Configurational Approach to Russian "Free" Word Order. Cornell University Ph.D. dissertation. Brecht, Richard D. and James S. Levine. 1984. "Conditions on Voice Marking in Russian", in Michael S. Flier and Richard D. Brecht, eds. Issues in Russian Morphosyntax, 118-37. Columbus, OH: Slavica. Comrie, Bernard. 1976. Aspect. Cambridge: Cambridge University Press. Comrie, Bernard. 1980. "Nominalizations in Russian: Lexical Noun Phrases or Transformed Sentences", in Catherine V. Chvany and Richard D. Brecht, eds. Morphosyntax in Slavic, 212-20. Columbus, OH: Slavica. Fowler, George. [forthc. a]. "An Articulated Theory of Aspect and Prefixation in Russian". To appear in Jindrich Toman, ed. Formal Approaches to Slavic Linguistics 3. Ann Arbor, MI: University of Michigan. Fowler, George. [forthc. b]. The Major Case Constructions of Russian. Munich: Lincom Europa. Franks, Steven. 1995. Parameters in Slavic Morphoxyntax. Oxford: Oxford University Press. Franks, Steven. [forthc.]. "Empty Subjects and Voice-Altering Morphemes in Slavic", to appear in International Journal of Slavic Linguistics and Poetics, 39-40. Freidin, Robert. 1992. Foundations of Generative Syntax. Cambridge, MA: MIT Press. Freidin, Robert and Leonard H. Babby. 1984. "On the Interaction of Lexical and Structural Properties: Case Structure in Russian". Cornell Working Papers in Linguistics 6: 71-105. Gerritsen, Nelleke. 1988. "How Passive Is 'Passive' -sja?" Dutch Contributions to the Tenth International Congress of Slavists, Sofia. Linguistics, 97-179. Amsterdam: Rodopi. (Studies in Slavic and General Linguistics, 11.) Keenan, Edward L. 1985. "Passive in the World's Languages", in Timothy Shopen, ed. Language Typology and Syntactic Description, v. 1, 243-81. Cambridge: Cambridge University Press. Langacker, Ronald. 1987. Foundations of Cognitive Grammar, v. 1. Stanford: Stanford University Press. Larson, Richard K. 1988. "On the Double Object Construction". Linguistic Inquiry 19.3: 335-91. Miloslavskij, I. G. 1981. Morfologiceskie kategorii sovremennogo russkogo jazyka. Moscow: Prosvescenie. Mustajoki, Arto. 1985. Padez dopolnenija v russkix otricatel'nyx predlozenijax 1: Izyskanija novyx metodov v izucenii staroj problemy. Helsinki: Helsinki University. (Slavica Helsingiensia, 2.) Mustajoki, Arto and Hannes Heino. 1991. Case Selection for the Direct Object in Russian Negative Clauses. Helsinki: Helsinki University. (Slavica Helsingiensia, 9.) Neidle, Carol. 1988. The Role of Case in Russian Syntax. Dordrecht: Kluwer. Nichols, Johanna. 1981. Predicate Nominals: A Partial Surface Syntax of Russian. Berkeley: University of California Press. (University of California Publications. Linguistics, 97.) [RG] 1980. Russkaja grammatika, 2 v. Moscow: Nauka. Ruzicka, Rudolf. 1967. "Korrelation und Transformation", in To Honor Roman Jakobson: Essays on the Occasion of His Seventieth Birthday, 11 October 1966, III: 1709-33. The Hague: Mouton, 3 v. (Janua linguarum. Series maior, 31-33.) Siewierska, Anna. 1988. "The Passive in Slavic", in Masayoshi Shibatani, ed. Passive and Voice, 243-89. Amsterdam: Benjamins. [SSRLJ] 1950-63. Slovar' sovremennogo russkogo literaturnogojazyka. Moscow: Akademija nauk SSSR, 17 v. Svedova, N. Ju., ed. 1970. Grammatika sovremennogo russkogo literaturnogo jazyka. Moscow: Nauka.


Oblique Passivization in Russian

545

Timberlake, Alan. 1975. "Hierarchies in the Genitive of Negation". Slavic and East European Journal 19.1: 128-38. Townsend, Charles E. Russian Word-Formation.Cambridge, MA: Slavica, 1975. Vinogradov, V. V. 1982. Ocerkipo istorii russkogo literaturnogojazyka XVII-XIX vekov, 3rd ed. Moscow: VysSaja skola. Zaliznjak, A. A. 1977. Grammaticeskij slovar' russkogo jazyka. Moscow: Russkij jazyk.