21.1 Informal Characterization of Voice
Grammatical voice is a prominent feature of the Slavic verb, and of the verb in general: it exists in one-third of the world’s languages (Reference HaspelmathHaspelmath 1990: 28, Reference Siewierska, Dryer and HaspelmathSiewierska 2013), including all Indo-European languages. It is a multifaceted phenomenon, involving three major components of language: semantics, syntax, and morphology. This makes the study of voice both interesting and challenging.
Voice is a verbal inflectional category; the values of this category, grammemes, are in principle obligatory for at least some classes of verbs in a language that possesses it.Footnote 1 The following example illustrates the basic voice opposition in Rus: active ~ passive; the verb is in the active voice in (1a), and in the passive (in 1b-c): reflexive passive in (1b), and participial passive in (1c).Footnote 2
(1)
a. Nemeckie rabočie (po+) stroi +l +i German worker-pl.nom perf build act.past pl školu. (Rus.) school-sg.acc ‘German workers built the school.’ b. Škola stroi +l +a +s′ school-sg.nom build act.past sg.fem refl (nemeckimi rabočimi). German worker-pl.ins ‘The school was being built (by German workers).’ c. Škola by+l +a po +stro +en school-sg.nom be-past sg.fem perf build pass-part +a (nemeckimi rabočimi). sg.fem German worker-pl.ins ‘The school was built (by German workers).’
Informally, voice can be characterized as follows.
All voice forms of a verbal lexeme (in corresponding tenses, aspects, etc.) are propositionally synonymous. This does not mean, however, that an active sentence and its passive counterpart are fully synonymous; they differ with respect to their communicative structure (a.k.a. information structure) and are not always interchangeable in texts. Different voice forms are used (in a sentence) to communicatively promote/demote or suppress the expression of different actants [core dependents] of the verb.
Voice involves simultaneously two sets of roles of verb actants: the semantic (Agent, Patient, etc.) and the syntactic ones (Subject, Direct Object, etc.). That is, each specific voice is characterized by a particular correspondence of these two types of roles. This correspondence is known as (verbal) diathesis; alternative terms are alignment and linking.Footnote 3 Active diathesis is characterized by the correspondence Agent/Subject ~ Patient/DirO (1a), and the passive one by the correspondence Agent/Agentive Complement (often left out of but implied by the sentence) ~ Patient/Subject (1b)–(1 c). Thus, we can say that voice grammemes affect the argument structure of the verb by modifying its basic diathesis; this is, by default, the active diathesis.
Not all specific voices are compatible with all verbs in a language. The compatibility depends on the verb’s lexical meaning, its syntactic features such as transitivity, active/stative character, and the inflectional categories of tense and aspect.Footnote 4 Also, a verb’s combinability with specific voices is contingent upon the features of its core dependents in the sentence (Subject, DirO), such as animacy, definiteness, and referential status; we will return to these features in due course.
Voice values are regularly indicated on the verb in the sentence, albeit not necessarily by markers exclusively used by voices (see immediately below). The verb is supplied with a special voice marker (an affix, a clitic, a particle, etc.) or appears in a special construction. Core verbal dependents are marked as well: in Slavic languages with inflected case forms (all except Mac. and Bul.), the Subject of a transitive verb in the active voice is in the nominative, and the DirO in the accusative.
There is syncretism of expressive means used with voices; some means of expression are shared among different specific voices and with some phenomena similar to, but nevertheless distinct from, voice. (There are historical reasons for this, to which we will turn later.) This poses a problem of teasing apart different specific voices, on the one hand, and distinguishing verbal voices from voice-like inflection and derivation, on the other. (In some broad conceptions of voice – Reference CreisselsCreissels 1995: 264–300, Reference Givón and GivónGivón 2001: 90–173, Reference PlungjanPlungjan 2003: 191–224 – some of these phenomena are considered as voice.)
21.2 Definition of Voice and a General Inventory of Specific Voices
There is a huge literature on grammatical voice in Slavic (and elsewhere). Yet, little agreement has been reached as to the definition of the phenomenon, and, consequently, the number and type of specific voices in individual Slavic languages. In various approaches voice is characterized more or less broadly, depending on the aspect of the phenomenon that is considered vital by a particular researcher: structural (how do voices impact the realization of the verb’s argument structure?) or functional (how are voices used in discourse?), the latter type of characterization tending to be more inclusive, subsuming under the label voice several phenomena that structural approaches consider distinct from it. However, few approaches offer a formal definition of voice, relying instead on a consensual understanding and informal characterizations.
A structural characterization of voice is proposed in (Reference Kulikov and SongKulikov 2012: 371): “The category of voice is determined on the basis of the concept of diathesis as follows: voice is a regular encoding of diathesis through verbal morphology.”
Reference PlungjanPlungjan (2003: 196) characterizes grammatical voice in functional terms, as
the verbal category whose grammemes indicate a change in the communicative rank of the participants of the situation [denoted by the verb-JM]. “Active” or “zero” voice corresponds to some initial rank structure […], while “oblique” or “derived” voices (the passive voice far from being a unique possibility) indicate the transfer of the status of the highest rank participant from one verbal argument to another.
Specific voices are sometimes characterized via a prototype and divergencies from it; thus, Reference SiewierskaSiewierska (1984: 28) defines the canonical, personal passive as having “(a) an overt subject with semantic content; (b) a corresponding active construction; (c) the subject of the passive corresponding to the P[atient] of the active”; see also Reference ShibataniShibatani (1985, Reference Shibatani1988), Reference HaspelmathHaspelmath (1990: 25ff.), and Reference Dixon and AikhenvaldDixon & Aikhenvald (2011: 46ff.).
Invoking the absence of consensus on how voice should be defined, Reference BabbyBabby (1998: 56) even proposes to replace the notion of voice, definable as “a subset of a base verb’s derived diatheses,” by that of “alternation based on the full diathetic paradigm”, since “the traditional definitions of voice often boil down to finding names for diatheses with certain arbitrarily determined formal properties” (Reference BabbyBabby 1998: 6).
The problem of divergent characterization of voice is compounded by terminological differences, as noted by many researchers, Slavists and others. Thus, Reference Fried, Lyngfelt and SolstadFried (2006: 86) mentions “a wealth of often confusing and conflicting terminology” in the domain of voice, and similar remarks can be found in Reference Dixon and AikhenvaldDixon & Aikhenvald (2011: 84–85).
Voice typologies generally focus on the opposition active ~ passive (sometimes adding to the inventory reflexive/middle and, perhaps, antipassive), further subdividing the passives along the axes plain ~ reflexive, personal ~ impersonal, and analytic (periphrastic) ~ synthetic (morphological). This holds for the bulk of the literature on Slavic voices (Reference SiewierskaSiewierska 1984, Reference Siewierska and Shibatani1988, Reference Franks and FranksFranks 1995, and many others), and for that on voice in general (Reference Perlmutter and PostalPerlmutter & Postal 1977, Reference Keenan, Dryer and ShopenKeenan & Dryer 2007, Reference Kazenin, Haspelmath, König, Oesterreicher and RaibleKazenin 2008, to mention just a few). Reference Kulikov and SongKulikov (2012) proposes a more developed voice typology, similar to the one to be presented here.
Our definition of voice and the inventory of logically possible specific voices are taken from Reference Mel‘čuk, Comrie and PolinskyMel‘čuk (1993, Reference Mel‘čuk2006a), studies that build on the characterization of voice put forward in Reference Mel‘čuk and XolodovičMel‘čuk & Xolodovič (1970) and writings of the Saint Petersburg typological school (Reference XolodovičXolodovič (ed.) 1974, Reference XrakovskijXrakovskij 1978, Reference Xrakovskij1981, Reference GeniušienėGeniušienė 1987). This is a deductively developed, cross-linguistically valid system of definitions, based on a dependency approach to syntax (Reference TesnièreTesnière 1959, Reference Mel‘čukMel‘čuk 1974 and 1988).
Central to this definition of voice is the familiar concept of diathesis; however, it is specified in terms of semantic actants and ‘generalized’ syntactic actants (known as deep-syntactic actants), rather than in terms of semantic and surface-syntactic roles.
Semantic actants of a lexeme L are the arguments of the corresponding predicate ‘L’ (L’s meaning); they are represented by variables: X stands for the first semantic actant, Y for the second one, and so on. Semantic roles of the actants (Actor/Agent, Patient, Theme, Goal, etc.) have no theoretical status and are used only informally, for ease of exposition. (This is because the role an actant plays within L’s meaning is determined via a decomposition of L’s meaning.)
‘Generalized’ syntactic actants of L normally correspond to L’s semantic actants; they are represented by Roman numerals I–VI. Each generalized syntactic actant stands for a cluster of specific surface-syntactic roles: syntactic actant I subsumes the subject of a verb and subject-like syntactic roles (e.g. that of a nominal complement), syntactic actant II covers the first object (direct, indirect, or oblique, as the case may be), syntactic actant III the second object, and so on.
This way of defining the diathesis allows for a more precise description of voice. In point of fact, a calculus of specific voices is not possible in terms of surface-syntactic roles, as their implementation is too variegated cross-linguistically. Moreover, sole reliance on surface-syntactic roles does not allow for an easy distinction between voices and certain phenomena superficially similar to voices (see Section 21.7).
Definition 1: Diathesis
A diathesis of a lexeme L is a correspondence between its semantic actants and its generalized, or deep, syntactic actants.
The basic [lexicographic] diathesis of a lexeme is part of its Government Pattern (Rus modelʹ upravlenija), specifying types of syntactic constructions in which this lexeme can participate as the syntactic governor. For instance, the basic diathesis of a bivalent transitive verb appears as follows (the diathesis is shaded):
| X ⇔ I | Y ⇔ II | DIATHESIS |
|
| –subjectival→NNOM | –dir.objectival→NACC |
|
An ‘oblique’ diathesis is obtained by a modification of the basic one; one possible oblique diathesis for a bivalent transitive verb is shown below (the implementation of the Agentive Complement is generalized to cover the two options existing in different Slavic languages: a noun in the instrumental case and a prepositional phrase).
| X ⇔ II | Y ⇔ I |
| –passive.agentive→NINSTR <PREP NCASE> | –subjectival→NNOM |
Realizations of these two diatheses are illustrated in (2a) vs. (2b) for the verb BCS podržati ‘[to] support’.
(2)
a. MedijiX⇔I su podržali media-pl.nom
be-pres.3pl support-act-part.pl.masc
idejuY⇔II. (BCS) idea-sg.acc
‘The media supported the idea.’ b. IdejaY⇔I je podržana od idea-sg.nom
be-pres.3sg
support-pass-part.sg.fem
from (strane) medija X⇔II. part-sg.gen
media-pl.gen
‘The idea was supported by the media.’
Definition 2: Voice
Voice is an inflectional category of the verb such that its grammemes mark a modification of the basic diathesis of a verbal lexeme L (without changing the propositional meaning of L).
The basic diathesis of L corresponds to a zero modification thereof; it expresses the grammeme active. The oblique diathesis of a bi-actantial transitive verb given above is obtained by a bilateral permutation of the verb’s syntactic actants relative to its semantic actants; it expresses the grammeme (full promotional) passive.
There are three logically possible types of diathetic modification: (1) permutation of deep-syntactic actants of L with respect to its semantic actants; (2) suppression of deep-syntactic actants of L; (3) referential identification of semantic actants of L.
Any of these three operations (or a combination thereof) applied to the basic diathesis of the verb (which represent a zero diathetic modification) gives us four possible major voice types: active, passive, suppressive, and reflexive; see Table 21.1.
Table 21.1 Twelve voices logically possible for a bivalent verb, eight of which are found in Slavic
| VOICE CLUSTER | GRAMMEME (= SPECIFIC VOICE) | DIATHESIS | |
|---|---|---|---|
| I. ACTIVE | [1] ACTIVE | X ⇔ I | Y ⇔ II |
| II. PASSIVES | [2] FULL PROMOTIONAL | X ⇔ II | Y ⇔ I |
| [3] AGENTLESS PROMOTIONAL | X ⇔ – | Y ⇔ I | |
| [4] PARTIAL DEMOTIONAL | X ⇔ III | Y ⇔ II | |
| [5] FULL DEMOTIONAL | X ⇔ II | Y ⇔ III | |
| [6] PATIENTLESS DEMOTIONAL | X ⇔ II | Y ⇔ – | |
| III. SUPPRESSIVES | [7] SUBJECTLESS | X ⇔ – | Y ⇔ II |
| [8] OBJECTLESS | X ⇔ I | Y ⇔ – | |
| [9] ABSOLUTE | X ⇔ – | Y ⇔ – | |
| IV. REFLEXIVES | [10] SUBJECTLESS | [X = Y] ⇔ II | |
| [11] OBJECTLESS | [X = Y] ⇔ I | ||
| [12] ABSOLUTE | [X = Y] ⇔ — | ||
Voice types II–IV each have several grammemes, based on subtypes of diathetic modifications involved (see the second remark below). Thus, voice can be regarded as a super-category (Reference Haspelmath, Müller-Bardey, Booij, Lehmann and MugdanHaspelmath & Müller-Bardey 2004) and major voice types as voice clusters (Reference Kulikov and SongKulikov 2012).
For a bivalent verb, there is a total of 12 logically possible specific voices. Out of these, Slavic bivalent verbs have eight, generally speaking (voices attested in Slavic are bolded in Table 21.1).
Remarks
1. The modifying adjectives in the names of specific voices are to be interpreted as follows:
| promotional | = Y is promoted from II to I and, at the same time, X is demoted from I to II; |
| demotional | = X is demoted from I to III and Y is not promoted (i.e. it remains II); |
| full | = all actants are involved; partial = not all actants are involved; |
| agentless | = the Agentive Complement cannot be expressed; |
| patientless | = the Patient cannot be expressed; |
| subjectless | = only Y⇔II can be expressed; |
| objectless | = only X⇔I or [X=Y]⇔I can be expressed; |
| absolute | = no actant can be expressed. |
2. A full passive is necessarily promotional, and vice versa. A partial passive is necessarily demotional, but the converse is not true: a demotional passive can be full or partial.
3. Full demotional passive (No. 5) with bivalent verbs has not been found in the world’s languages (Reference Mel‘čukMel‘čuk 2006a: 203); however, it is possible with monovalent (intransitive) verbs; see example (11b) below.
4. In the cases where a diathetic modification results in the absence of DSyntA I, and, therefore, the absence of a full-fledged syntactic subject at the surface-syntactic level (Nos. 4–6, 7, 9, 10, and 12), a dummy subject – a semantically empty lexeme (Reference Mel‘čukMel‘čuk 2006b: 469–516) – is called for in languages/constructions where the verb shows agreement with the subject. The presence of the dummy subject, which can have a phonological realization (Ang. IT, Fr. IL, Ger. ES, etc.) or, as is the case in Slavic, be phonologically null (Mel‘čuk 2006b: 469–516), explains the modifier ‘impersonal’ in some of the names for specific voices usually found in the linguistic literature.
5. Here are the names by which ‘oblique’ voices (i.e., other than the active) identified in Table 21.1 usually go in the literature: [2] basic/ canonical/personal passive; [3] short/truncated passive, potential passive; [4] impersonal passive, impersonal construction; [7] desubjective, impersonal reflexive passive, active impersonal; [9], [12] impersonal/deagentivized construction; [11] reflexive.
Voice, as defined by the type of calculus presented above, can also be applied to monovalent and plurivalent verbs. In what follows, we will cursorily consider these types of verbs as well.
21.3 Voice Types in Slavic
Not all Slavic languages have all the voices indicated in Table 21.1. Within any given Slavic language, there is no one verb that has all the voices in principle possible in that language; there are no voice grammemes that are compatible with all the verbs in the language. The determining factor is, above all, transitivity (Reference Hopper and ThompsonHopper & Thompson 1980): the basic passive and the (objectless) reflexive are formed only from transitive verbs; other voices can be formed also from intransitive verbs. The semantic type of the verb plays a role, as well.
Examples of specific voices possible for a Slavic bivalent transitive verb follow (the numbering 1-2, 1-3, etc. refers to voice names in Table 21.1):
(3)
1-2 Rus. prinesti ‘to bring’; issledovat‘ ‘to study/to investigate’ a. Mal‘čikX⇔I prinës kniguY⇔II. boy-sg.nom bring(perf)-past.sg.masc book-sg.acc ‘The boy brought the book.’ b. KnigaY⇔I byla prinesena book-sg.nom be-past.sg.fem bring(perf)-pass-part.sg.fem mal‘čikomX⇔II. boy-sg.ins ‘The book was brought by the boy.’ c. UčënyeX⇔I issledujut èto javlenieY⇔II. scientist-pl.nom study(imperf)-pres.3pl this phenomenon-sg.acc ‘Scientists study this phenomenon.’ d. Èto javlenieY⇔I issledujetsja učënymiX⇔II. this phenomenon-sg.nom study(imperf)-pres.3sg.refl scientist-pl.ins ‘This phenomenon is studied by scientists.’
(4)
(5)
(6)
1-7 Pol. czytać ‘to read’ a. JanX⇔I przeczytał książkęY⇔II. Jan-sg.nom read-past.sg.masc book-sg.acc ‘Jan read the/a book.’ b. Przeczytano książkęY⇔II (*przez JanaY⇔II). read(invar) book-sg.acc (*by Jan) ‘The book has been read.’
(7)
1-9 BCS igrati ‘to dance’ a. U klubu su sviX⇔I igrali in club be-pres.3pl all-nom dance-act-part.pl.masc salsuY⇔II. salsa-sg.acc ‘Everyone in the club danced salsa.’ b. U klubu se igralo (*salsuY⇔II).Footnote 5 in club refl dance-act-part.sg.neu (*salsa) ‘There was dancing in the club.’
(8)
1-11 Bul. kăpja ‘to bathe’ a. MajkaX⇔I šte kăpe AnaY⇔II. mother will bathe Ana ‘Mother will bathe Ana.’ b. AnaX+Y⇔I šte se kăpe (sama). Ana will refl bathe (herself) ‘Ana will bathe (herself).’
(9)
1-12 Pol. czesać ‘to comb’ a. MatkaX⇔I uczesała mnieY⇔II. mother-sg.nom comb-past.sg.fem I-acc ‘Mother combed my hair.’ b. Uczesano się. combed(invar) refl ‘People/someone has combed their hair.’
Monovalent intransitive verbs can have, in addition to the active, one regular and productive oblique voice: subjectless suppressive (equivalent in this case to absolute suppressive), illustrated in (10). Marginally, some such verbs admit full demotional passive, possibly in dialectal and substandard usage, as is the case with the Rus. example in (11).
(10)
1-7 BCS živeti ‘to live’ a. Ovde ljudiX⇔I lepo žive. here people-pl.nom well live-pres.3pl ‘People live well here.’ b. Ovde se lepo živi. here refl well live-pres.3sg ‘One lives well here.’/ ‘Life is good here.’
(11)
1-5 Rus. xodit‘ ‘to walk’ a. Skol’ko jaX⇔I tut xodil! how.much I-nom here walk-past.sg.masc ‘How much I walked here!’ b. Skol’ko mnojX⇔II tut bylo xoženo! how.much I-ins here be-past.sg.neu walk-pass-part.sg.neu ‘How much walking was done by me here!’
In some Slavic languages, plurivalent intransitive verbs from specific semantic classes can be marginally used in partial demotional passive and subjectless suppressive; see Section 21.4 for examples.
Unlike some other typologies of Slavic voices, the one presented here does not include objectless suppressive (No. 8 in Table 21.1) and reciprocal. Sentences in (12) illustrate what is usually taken to be objectless suppressive (a.k.a. deobjective, absolutive, antipassive); those in (13) illustrate the reciprocal.
(12)
(13)
Oni se vole <poštuju, mrze>. (BCS) they refl love-pres.3pl <respect-pres.3pl hate-pres.3pl> ‘They love <respect, hate> each other/one another.’
The forms in question are not voice forms. Our definition of voice requires that voice grammemes do not affect the propositional meaning of lexeme L to which they are associated, that is, that no semanteme (a chunk of meaning) be added to L’s meaning in the process. Alternatively, we can say that the situation described by L (roughly, ‘who does what to whom’) does not essentially change under diathesis modification. The constraints on the actants of L (their distribution/correlation between them), on the other hand, or the conditions of L’s use, may be affected.
According to this definition, neither (12) nor (13) are instances of voice, since in both cases there is addition of meaning with respect to that of the base lexeme: the meaning of aggression in the former case (Reference IsraeliIsraeli 1997: 109–125),Footnote 6 and that of symmetry/interaction in the latter. Therefore, in line with Reference Mel‘čukMel‘čuk (2006a: 206ff.), we treat such examples as instances of verbal derivation, more specifically, as particular cases of detransitivization.Footnote 7 While one can define voice in a way as to include both of these verbal forms, such a definition would lump together phenomena that in our view are distinct.
On objectless suppressive treated as a voice form in Slavic, see Reference Rivero and Milojević SheppardRivero & Milojević Sheppard (2003) and Reference JanicJanic (2014, Reference Janic2016), for instance; an overview of its use in Slavic and Baltic languages can be found in Bondarenko (2022). The publications on the reciprocal and its relation to the objectless reflexive (which in Slavic languages make use of the same formal marking on the verb – in yet another instance of syncretism of expressive means, mentioned earlier) include Reference NedjalkovNedjalkov (2007), dealing with reciprocals in Polish, Bulgarian, and Russian (and many other, non-Slavic, languages), Reference König, Gast, König and GastKönig & Gast (2008) and Reference Letuchiy, Kosta and SchûrcksLetuchiy (2011).
Before we turn to a description of Slavic oblique voices, we need to comment on the fact that some voice types in our typology (subjectless and absolute suppressive, absolute reflexive, Nos. 7, 9, and 12) are not always explicitly qualified elsewhere as instances of voice and are called simply ‘impersonal’ or ‘deagentivized’ constructions (cf. fourth and fifth remarks after Table 21.1). While it is true that all these voices use the impersonal construction (i.e. a construction featuring a dummy subject or lacking one altogether) as one of their means of expression, impersonalization is independent of diathetic modifications (in our framework, they pertain to two different levels of representation – surface- vs. deep-syntactic, respectively) and therefore not relevant for the calculus of possible voice types. In addition to implementing different voice types, the impersonal construction can implement different verbal derivations; here are some examples.
(14)
a. Mirisa +l +o je na vlagu. (BCS) smell act-part sg.neu be-pres.3sg on dampness ‘There was a smell of dampness.’ b. Juče se iš +l +o na izlet. yesterday refl go act-part sg.neu on outing ‘Yesterday people/we went on an outing.’ c. Jovanu se iš +l +o na izlet. Jovan-sg.dat refl go act-part sg.neu on outing ‘Jovan felt like going on an outing.’
The verb in sentence (14a) is in the active (voice), that in (14b) in the absolute suppressive (voice), and the one in (14c) in the dispositional (derivation). At the same time, each of these sentences is ‘impersonal’, lacking an overt, referential syntactic subject.
21.4 Slavic Oblique Voices
As has been seen, oblique voices come in clusters: passives, suppressives, reflexives; we will take them in turn.
21.4.1 Passives
21.4.1.1 Full Promotional Passive
This passive type, usually going by the labels basic, canonical, or personal passive, is formed from transitive verbs; it comes in two variants: participial (periphrastic) passive and reflexive (pronominal) passive. Both variants are common to all Slavic languages except modern Polish, where the reflexive passive is obsolete (Reference Fehrmann, Junghanns and LenertováFehrmann et al. 2010: 211, Reference Holvoet and Linde-UsiekniewiczHolvoet & Linde-Usiekniewicz 2015: 110).
Participial passive construction features the auxiliary ‘be’ (Pol. additionally has zostać ‘become’; colloquial Sorb uses wordować ‘become’) and the lexical verb in the passive participle form; see examples (1c), (2b), and (3b) above. Reflexive passive uses a reflexive marker – a suffix (-sja/-s‘) or a clitic (se, się, etc.), which is added to the active form of the lexical verb; see examples (1b) and (3d) above. The auxiliary agrees with the syntactic subject in person, number, and (in the past tense) gender; the participle shows agreement with the subject’s number and gender. In Russian, the participle must be short (only predicative).
The Patient/Subject is in the nominative case. The optional Agentive Complement is marked by the instrumental case in E. Slavic and Slk., by a prepositional phrase in Pol. (przez ‘by/through’ NACC), and S. Slavic (BCS od (strane) ‘from (part.of)’ NGEN; Bul./Mac. ot/od strana na N; Sln. s strani NGEN), and in either way in Cze (with a preference for the instrumental phrase, at least according to the speakers I asked).Footnote 8 See examples in (16)–(18).
Constraints on the Verb
Participial and reflexive passives are distributed as a function of verbal aspect and tense.
The clearest distribution is found in Rus., where in an overwhelming majority of cases the participial passive is formed from perfective verbs and the reflexive passive from imperfective verbs. Imperfective participial passive, found sporadically, is subject to strong lexical, syntactic and contextual constraints (Reference Maslov and NedjalkovMaslov 1988, Reference Knjazev and NedjalkovKnjazev 1988, Reference Knjazev2007, Reference Borik, Gehrke, Lenertová, Meyer, Šimík and SzucsichBorik & Gehrke 2016); the same is true for the perfective reflexive passive (Reference PercovPercov 2003, Reference LevineLevine 2010).
Participial passive with imperfective verbs is allowed in varying degrees in Cze., Pol., and Bul.; however, a general preference for this variety of the passive is to be used with perfective verbs. This is linked to the stative/resultative flavor of the periphrastic passive (Reference Guentcheva, Xrakovskij, Mal’čukov and DmitrenkoGuentcheva 2004: 109). At the same time, reflexive passive is preferably made from imperfective verbs, owing to its predominantly actional/dynamic character (Reference Guentcheva, Xrakovskij, Mal’čukov and DmitrenkoGuentcheva 2004: 109). Aspectual oppositions are the least strong in BCS, where both types of promotional passive are in principle allowed from both perfective and imperfective verbs.
In Pol., the choice of the auxiliary is contingent on the aspect of the main verb: the zostać passive is used with perfective verbs, the być passive with both imperfective and perfective verbs. According to Reference Siewierska and ShibataniSiewierska (1988: 250), Pol. passives with zostać are always actional; those with być are actional with imperfective verbs and can be either stative or resultative if formed from perfective verbs.
In Slavic languages, only imperfective verbs can convey the actual present meaning (i.e. refer to the present moment), and this independently of voice. Non-actual present (iterative, injunctive, hypothetical) can be conveyed by both imperfective and perfective verbs.
In BCS, additionally, there is no participial passive in the present with the actual reading (Reference Đurković and HendriksĐurković 2004: 78, Reference TanasićTanasić 2014: 101ff.); BITI ‘to be’PRES + VPASS.PART expresses the past (cf. example (2b)).Footnote 9 (This means that actual present facts can be reported in the passive only with imperfective verbs, i.e. in the reflexive passive.) An analogous situation obtains in Mac. (Reference TopolińskaTopolińska 2003: 22).
Also, the use of BCS perfective verbs is restricted in the reflexive passive of the past and, to a lesser extent, the future. The acceptability of the reflexive passive in these tenses depends on the verbs involved, as well as on the number and definiteness of the noun expressing the Patient. It may also be subject to individual/regional variation. These constraints are illustrated in Table 21.2 for the verb cˇitati ‘to read’ in the passive: čitati (imperfective) and pročitati (perfective).
Table 21.2 Interaction of voice, aspect, and tense in BCS (Milićević 2022: 300)
| V(Imerf) | V(Perf) | |
|---|---|---|
|
|
|
| Past |
|
|
| Future |
|
|
Reference SpalatinSpalatin (1973: 121) speaks about “more or less prominent reluctance to use perfective verbs in the se-passive.” We may add that this reluctance seems less prominent in Cro.; thus, in sentences such as (15), taken from a contemporary Croatian text, Serbian speakers would sooner use the periphrastic passive forms (in this case, određeni su and nisu uključeni, respectively).
(15)
a. Pojedini tipovi konstrukcija odredili certain type-pl.nom of.constructions determine(perf)-act-part.pl.masc su se na temelju … (Cro.) are refl on basis.of … ‘Certain construction types were determined based on … ’ b. U raspravu o konstrukcijama nisu se in debate of constructions not.are refl uključili glagoli tog tipa. include(perf)-act-part.pl.masc verb-pl.nom of that type ‘Verbs of that type were not included in discussion about the constructions.’
In Bul., the participial promotional passive is restricted tense-wise: it invariably appears with the past tense morphology (Reference Dimitrova-VulchanovaDimitrova-Vulchanova 1999: 24).
For verbal aspect and tense/mood in Slavic, see Chapters 9 and 10 in this volume.
Constraints on the Actants
In all Slavic languages, the expression of the Agent in passive clauses is optional with the periphrastic passive; with the reflexive passive, it is optional in some languages and blocked in others.Footnote 10 The Agent often remains unexpressed (since one of the discourse functions of the passive is, exactly, Agent backgrounding; see Section 21.5).
With the periphrastic passive, agentive phrases are unrestrictedly used only in E. Slavic; elsewhere, they tend to be limited to journalistic and scientific/technical registers.
(16)
Agentive phrases with the reflexive passive are possible in E. Slavic, USo., Bul., and, to some extent, Mac., but are disallowed in Pol., Cze., Slk., BCS, and Sln. (Reference Fehrmann, Junghanns and LenertováFehrmann et al. 2010: 214). Note, however, that in BCS agentive phrases are increasingly used with the reflexive passive, even though this practice is still ‘officially’ proscribed (Reference Piper, Antonić, Ružić, Tanasić, Popović and TošovićPiper et al. 2005: 371; cf. (17c)).Footnote 11
(17)
a. Šaty so runje (wot wowki) šija. (USo.) clothes-pl.nom refl right.now (by grandmother) sew-pres.3pl ‘Clothes are being sewn right now (by grandmother).’ b. Šaty se práve šijí (*babičkou). (Cze.) dress-sg.nom refl right.now sew-pres.3sg (*by grandmother) ‘The dress is being sewn right now.’ c. Penzioni sistem u Srbiji finansira se pension system-sg.nom in Serbia finance-pres.3sg refl samo od (strane) Države. (Ser.) only from part state-sg.gen ‘The pension system in Serbia is financed only by the State.’ d. Izborot se vrši od selection-sg.def refl preform-pres.3sg from strana na Komisija za verifikacija […] (Mac.) part at commission for verification ‘The selection is made by the Verification Commission.’
Semantic and referential features of the Agent seem to be the least constrained in Rus. Thus, the Agent of the Rus. reflexive passive can be not only ‘human’, as it usually is elsewhere in Slavic, but also ‘non-human’ and even ‘inanimate’ (Reference SiewierskaSiewierska 1984: 181).
(18)
a. Jagody klevalis’ pticami. (Rus.) berry-pl.nom peck-past.pl.refl bird-pl.ins ‘Berries were being pecked by birds.’ b. List’ja vzmëtyvalis’ vetrom. leaf-pl.nom sweep-past.pl.refl wind-pl.ins ‘Leaves were being swept by the wind.’
For properties of Agents in passive clauses cross-linguistically, see Reference Siewierska, Baker, Brown, Chumakina and CorbettSiewierska & Baker (2013).
As for the Patient’s features, they do not seem to be restricted: its referent can be inanimate, animate, or human with both variants of the periphrastic passive.
21.4.1.2 Agentless Promotional Passive
Agentless promotional passive, a.k.a. short/truncated/potential passive, has an anonymous or generalized human Agent whose surface expression is blocked, and a Patient expressed as the syntactic subject with which the verb agrees. This passive type exists across the Slavic domain, but is only marginally present in Polish (which, as we have seen, has lost the basic reflexive passive).Footnote 12
(19)
a. Bel’ëY⇔I stiraetsja v gorjačej vode. (Rus.) laundry-sg.nom wash-pres.3sg.refl in hot water. ‘The laundry is washed in hot water.’ b. SzarlotkaY⇔I się łatwo kroi. (Pol.) apple.pie-sg.nom refl easily cut-pres.3sg ‘The apple pie cuts easily = is easy to cut.’ c. KrompirY⇔I se oljušti, potato-sg.nom refl peel(perf)-pres.3sg iseče (se) na kockice i (Ser.) cut(perf)-pres.3sg refl on dices and ‘Potato is peeled, diced and … ’ d. Go poučija lugjeto: prašanjetoY⇔I him taught people question-pl.def taka ne se postavuva. (Mac.) thus not refl put-pres.3sg ‘People explained to him: you don’t/one shouldn’t ask the question that way.’
The implied presence of the Agent is corroborated by the possibility to add to the sentence an Agent-oriented adverb, for instance, pažljivo ‘carefully’ in (19c).
The short passive is usually used with imperfective verbs, but this usage is not exclusive, at least not in BCS, where perfective verbs are normal in prescriptions (operating instructions, recipes, and the like), as in (19c). It tends to carry a generic (habitual, iterative) or modal (potential, injunctive) meaning and often requires the presence of a manner/locative adverb or the negation.Footnote 13
21.4.1.3 Partial Demotional Passive
Partial demotional passive from transitive verbs, known also as impersonal (transitive) passive and -to/-no impersonal, is characteristic of Ukr. and N. Rus. dialects. (For the Pol. -no/-to construction, which is a subtype of suppressive, see immediately below.)
(20)
a. XlopcjaY⇔II bulo znajdeno u košyku boy(N, masc)-sg.acc be-past.sg.neu found(invar) in basket likarjamyX⇔III.doctor-pl.ins (Ukr.) ‘A baby was found in a basket by doctors.’ b. Cju operacijuY⇔II bude vykonano this operation(N, fem)-sg.acc be-fut.3sg carried.out(invar) vidomym xyrurgomX⇔III. well.known surgeon-sg.ins ‘This operation will be carried out by a well-known surgeon.’ c. TabirX⇔II bulo zajnjato camp(N, masc)-sg.acc be-past.sg.neu occupied(invar) amerykans’kym vijs’komY⇔III. American troop-sg.ins ‘The camp was occupied by American troops.’
The Ukr. demotional passive is almost exclusively formed with perfective verbs and has actional, rather than resultative, semantics, the latter being associated in this language with the canonical passive (Reference Nedashkivska AdamsNedashkivska Adams 1998). The main verb is invariant, in the so-called predicative form, ending in -o (the old neuter agreement marker, different from the current neuter -e). The auxiliary buloPAST.SG.NEU ‘was’/budeFUT.3SG ‘will be’ indicates the tense and the mood. The Patient is retained in the accusative case and the Agent, which carries the feature ‘human’, may but need not be expressed. Since no semantic actant is implemented as the deep-syntactic actant I (which should correspond to the subject), an empty zero 3sg neuter subject must be used at the surface-syntactic level for agreement purposes.
In N. Rus. dialects, the Patient in the partial demotional passive may also be marked by the nominative case (Reference TimberlakeTimberlake 1974, Reference RonkoRonko 2017), as in (21b); intransitive verbs can participate in the construction, as well (21c).
(21)
a. U njejX⇔III tëlkuY⇔II bylo at her heifer-sg.acc be-past.3sg.neu zarezano. (N. Rus.) slaughter-pass-part.sg.neu ‘A heifer was slaughtered by her.’ b. Pereexano bylo dorogaY⇔II tut. cross-pass-part.sg.neu be-past.sg.neu road-sg.nom here ‘There has been crossing over the road there.’ c. U menjaX⇔II uže vstato bylo. at me already get.up-pass-part.sg.neu be-past.sg.neu ‘I had already gotten up.’
In Rus., Ukr., BCS, and Bul., partial demotional passive is possible with specific classes of intransitives (verbs of saying/perception) and may be limited to colloquial style. In (22a) we see a demotional passive form of a tri-valent verb.
(22)
a. MneZ⇔III bylo ukazano I-dat be-past.sg.neu point.out-pass-part.sg.neu na ošibkuY⇔II (direktoromX⇔IV). (Rus.) at error-sg.acc director-sg.inst ‘The error was pointed out to me by the director.’ b. Pro ceY⇔II (namiX⇔III) vže hovorylosja. (Ukr.) about that we-ins already speak-past.sg.neu.refl ‘That has been already talked about by us.’ c. Za tovaY⇔II veče se govori (od nasX⇔III) about this already refl talk-past.3sg by us na minalato săbranie. (Bul.) at last meeting ‘This has been already talked about by us at the last meeting.’
21.4.2 Suppressives
The suppressive cluster comprises two subtypes: subjectless suppressive (a.k.a. desubjective), which retains the Patient and suppresses the expression of the Agent, and the absolute suppressive, with which neither the Patient nor the Agent can be expressed. The implied Agent carries the features ‘indefinite’ and ‘human’.
Suppressives are the best developed in Pol, where they can be implemented both by the -no/-to construction and the reflexive one; in other Slavic languages, only the reflexive variant is regularly present.
In Slavic linguistics, both suppressive subtypes are usually called active impersonal/indefinite constructions, on the account of their semantic similarity with the constructions featuring indefinite human pronouns (Eng. one, Ger. man, Fr. on, etc.) and 3pl indefinite active constructions, as well as the fact that they lack an overt referential syntactic subject.
21.4.2.1 Subjectless Suppressive
Examples of Pol. subjectless suppressive follow: the -no/-to variant in (23a–23c), and the reflexive one in (23d).
(23)
a. (*Zostało) znaleziono chłopca w koszu become-past.3sg.neu found(invar) boy(N.masc)-sg.acc in basket (*przez lekarzy). (Pol.) by doctors ‘A boy was found in a basket (*by doctors).’ b. Rozkaz wykonano. order-sg.acc executed(invar) ‘The order was executed.’ c. Żeby uratować dziecko, wybito szybę. in.order.to save child broken(invar) pane-sg.acc ‘In order to save the child, the window-pane was broken.’ d. Tę książkę czytało się z przyjemnością. that book-sg.acc read-past.sg.neu refl with pleasure ‘That book was read with pleasure.’
Reference KibortKibort (2004: 241–340) provides an extensive discussion of the two types of Polish suppressive (called -no/-to impersonal and reflexive impersonal, respectively.)
The Pol. -no/-to construction differs from the Ukr. one in the following respects: (1) it does not allow for an auxiliary (cf. (23a) vs. (20a)); (2) the verb, in the (resultative) past, doesn’t agree with anything, so there is no need to postulate a syntactic subject; (3) the implied Agent cannot be individualized, that is, it is interpreted as ‘people’ or indefinite ‘they’ and typically excludes the Speaker (Reference Siewierska and ShibataniSiewierska 1988: 271). For a detailed comparison of the two cognate constructions, see Reference BlevinsBlevins (2003) and Reference LavineLavine (2005).
Pol. reflexive subjectless suppressive can be used in all tenses/moods; at the surface-syntactic level, there is a 3sg empty zero subject, with which the verb agrees.
Subjectless suppressive from transitive verbs is attested also in Sln. and Cro., but not in Ser. or the other S. Slavic languages (Reference Franks and FranksFranks 1995: 347ff., Reference Lenardič, Marušič, Mišma and ŽaucerLenardič 2020, Reference KučandaKučanda 1992, Reference Oraić RabušićOraić Rabušić 2017, Reference Uhlik and ŽeleUhlik & Žele 2020). It may be confined to the informal style/stylistically marked.
(24)
a. Dokumente se redno preverja. (Sln.) document-pl.acc refl regularly check-pres.3sg ‘Documents are regularly checked.’ b. U vrijeme kad sam bio ministar, mene se in time when I was Cabinet member, I-acc refl žestoko kritiziralo. (Cro.) strongly criticize-act-part.sg.neu ‘At the time I was a Cabinet member, I was strongly criticized.’
BCS and Sln. allow subjectless suppressive from some intransitive verbs (mostly verbs of saying/perception and mental activity verbs), as do W. Slavic languages and, more limitedly, Russian.Footnote 14
(25)
a. Meni se nikad ne v(j)eruje. (BCS) I-dat refl never not believe-pres.3sg ‘I am never believed.’ b. O njem se je veliko govorilo. (Sln.) about he-loc refl be-pres.3sg a.lot speak-act-part.sg.neu ‘He was spoken of a lot.’ c. Ob ètom slučae mnogo govorilos’ about that case much speak-past.3sg.neu.refl v poslednee vremja. (Rus.) in last time ‘This case has been much talked about lately.’
21.4.2.2 Absolute Suppressive
Absolute suppressive from intransitively used transitive verbs (unergatives) is available everywhere but in Rus. and Bel. In some contexts, a temporal or locative adverbial is required.
(26)
a. Tutaj tańczono <się tanczyło>. (Pol) here danced(invar) <refl dance-past.sg.neu> ‘There was dancing <People/We danced> here.’ b. Minalata godina zdravo se pieše. (Bul.) last year-sg.def much refl drink-past.3sg ‘People/We drank a lot last year.’ c. *Pracavalasja do ranicy. (Bel.) work-past.3sg.neu.refl till morning ‘People/We worked till morning’.
Absolute suppressive from monovalent intransitive verbs (unaccusatives) is unavailable in East Slavic and freely used in the remaining Slavic languages except in Bulgarian, where its acceptability is dubious for some speakers (Reference Fehrmann, Junghanns and LenertováFehrmann et al. 2010: 213).
(27)
a. Płakano bez końca. (Pol.) cried(invar) without end ‘There was endless crying.’ b. Najviac sa dnes umiera na srdcové choroby. (Slk.) mostly refl today die-pres.3sg on heart diseases ‘Today people die mostly from heart diseases.’ c. ?Prez srednevekovieto se e umiralo through middle.ages refl be-pres.3sg die(imperf)-act-part.sg.neu ot čuma. (Bul.) of plague ‘During the Middle Ages people used to die from plague.’
Reflexives
Reflexives also come in two variants: objectless reflexive and absolute reflexive. Both are characterized by referential identification of the Agent and the Patient and subsequent suppression of syntactic actants.
With the objectless reflexive, the only syntactic actant that can be expressed corresponds to the surface subject (which controls the agreement of the verb):
(28)
Absolute reflexive blocks the expression of both syntactic actants (i.e. neither the surface subject nor the direct object is possible), thus placing the focus solely on the event described by the verb. The only examples of this voice in Slavic languages known to me come from Polish:
(29)
Ubrano <wykąpano, umyto> się. (Pol.) dressed(invar) <bathed(invar), washed(invar)> refl ‘Someone/They dressed <bathed, washed>.’
While objectless reflexive is considered in linguistic literature as a type of voice, absolute reflexive is considered the same way as the suppressives, that is, as an (active) impersonal construction. On null/unspecified subjects and impersonal constructions, see Chapter 20 in this volume; on null subjects and passive, see Reference Franks and FranksFranks (1995: 339–355).
Table 21.3 recapitulates the distribution of grammemes of the seven oblique voices possible with transitive verbs in Slavic. (As noted previously, some of these voices are limitedly possible with some intransitive verbs as well.)
Table 21.3 Slavic oblique voices for bivalent transitive verbs per branch/language
| Full prom. pass. | Agentless prom. pass. | Part. dem. pass. | Subjectless suppr. | Absolute suppr. | Objectless refl. | Absolute refl. | |
|---|---|---|---|---|---|---|---|
| E. Slavic | All | All | Ukr., N. Rus. | — | Ukr. | All | — |
| W. Slavic |
|
| — | Pol. | All | All | Pol. |
| S. Slavic | All | All | — | Sln., Cro. | All | All | — |
21.5 Discourse Functions of Voices
Voice alternations do not change the verb’s semantics (or do so only slightly) – they do not express propositional meanings; what they do express is the communicative perspective on the situation described by the verb: it is seen from the angle of a different participant or as purely eventful. Therefore, specific voices can be used to express the communicative, a.k.a. information, structure (Reference Mel‘čukMel‘čuk 2001, Féry & Ishihara, eds 2016) and establish thematic progression (topic ~ comment patterns) in discourse/text. This is indeed the primary discourse function of voices.Footnote 15
Voice alternations, in particular the active ~ passive one, have been much discussed in the literature in terms of the relative discourse prominence of the participants of the situation described by the verb (i.e. actants of the verbal lexeme). Passivization is variously viewed as spontaneous demotion of the Agent (Reference Comrie, Cole and SaddockComrie 1977); Agent backgrounding (i.e. making it referentially obscure; Reference Fried, Lyngfelt and SolstadFried 2006: 86); Agent defocusing (Reference ShibataniShibatani 1985); detopicalization of the Agent and topicalization of the Patient (Reference Givón and GivónGivón 2001). For Reference HaspelmathHaspelmath (1990), the main effect of passivization is the inactivation of the situation, from which foregrounding/backgrounding effects on the participants follow automatically. Reference IsraeliIsraeli (1997: 170–198) describes discourse uses of the Russian reflexive passive with and without the expression of the Agent. For instance, the situations in which the Agent of the reflexive passive is not mentioned include the following ones: (a) it is not necessary to mention the Agent because it is self-evident, unimportant or generalized; (b) the Agent is deliberately omitted in order to obscure it because it is antagonistic (to the Speaker), to avoid naming oneself (as Agent) out of modesty, to represent the action as taking place by itself, etc. Reference Dixon and AikhenvaldDixon & Aikhenvald (2011: 48) maintain that “[T]he situations in which passive is typically used include: to avoid mentioning the A(gent) argument; to direct attention onto the O(bject), rather than on the A; to place a topic (which is underlying O) into surface S(ubject) function; to focus on the result of the activity.”
The calculus of possible voices presented in Section 21.2 above clearly establishes discourse prominence of the verb’s actants, which is indicated in the names of specific voices (second and third remarks after Table 21.1). With respect to the non-marked active, the situation is as follows.
Promotional passive, based on a bilateral permutation of the verb’s actants, allows the Patient to advance to the most important surface-syntactic role, the subject, from which the demotion of the Agent follows automatically.
With demotional passive, the Agent is relegated to a less prominent syntactic role, that of an adjunct, without the concomitant promotion of the Patient, which keeps its rank.
Objectless reflexive focuses on the Agent, seen as acting upon himself or in his own interest, and eliminates the referentially identical Patient from the picture.
Subjectless suppressive anonymizes the Agent and precludes its expression in syntax altogether without promoting the Patient, which nevertheless acquires more prominence as the only remaining actant in the clause.
Absolute suppressive and absolute reflexive anonymize the Agent and do not allow either of the actants to be expressed, thus emphasizing the event itself.
As for the distribution and frequency of specific voice forms in texts, it depends in the first place on the language of the text and text genre/type (individual preferences also play a role). This issue has been researched especially in technical/scientific and academic writing, L2 teaching and learning, as well as in contrastive and translation studies; here again, the focus has been on the use of active vs. passive forms. It is well known, for instance, that the occurrence of passive forms in English general texts is higher than in corresponding Russian texts (Reference Comrie and ComrieComrie 1987), that BCS overwhelmingly prefers the active in all but technical/scientific texts, and so on. Also, passive tends to have undeservedly bad press in the pedagogically oriented literature, where it is subject to a virtual blanket ban, regardless of the type/style of the text; see, for instance, Reference PullumPullum (2014) for a critique of such views of the English passive.
21.6 Historical Development of Voice in Slavic
Both the participial and the reflexive construction used today to implement voices in Slavic have existed throughout the history of these languages, dating back to Proto-Slavic and even Proto-Indo-European (PIE). Most accounts of diachronic development of Slavic voices focus on the canonical passive; this is what we will do as well.
Passive syntactic patterns have been reconstructed for PIE, although the evidence for the existence of dedicated passive morphology is present only in late PIE and only in some dialects (Reference Kulikov, Lavidas, Kulikov and LavidasKulikov & Lavidas 2015: 119). Slavic languages (together with Greek, Germanic, and Romance) belong to the so-called syncretic type, in which the means of expression of the passive have continued to be shared with other phenomena – namely, with stative/resultative and middle constructions. (Languages of anti-syncretic type, which have developed specialized passive morphology, include Indo-Aryan, Iranian, and Armenian.) On voice syncretism, see Reference BahrtBahrt (2020) and references therein.
Participial passive arose through a reanalysis of an earlier Indo-European resultative construction, “a construction that describes a resulting state with undergoer-orientation” (Reference Toyota, Mustafović, Abraham and LeisiöToyota & Mustafović 2006: 201ff.). This explains a predominantly ‘resultative flavor’ of the periphrastic passive, mentioned earlier, and its potential ambiguity with purely resultative ‘copula + adjective’ constructions (see Section 21.7).
Reflexive passive used to be a part of the Indo-European middle cluster, a set of constructions with distinct semantics but having in common the “affectedness of (the referent of) the subject” (Reference KemmerKemmer 1993), opposed to the active. The middle cluster included, in addition to passive forms, decausative/inchoative verbs, inherent reflexives/reciprocals, and subject-oriented auto-benefactives (Reference Kulikov, Lavidas, Kulikov and LavidasKulikov & Lavidas 2015: 105), all marked with the same reflexive marker that can be traced back to the PIE reflexive pronoun *s(w)e- (Lat se, Sanskrit sʋáyam, Old Slavic sebe).Footnote 16 This situation has largely persisted in modern-day Slavic languages, even though the original voice opposition ‘active ~ middle’ had long disappeared and was replaced by the ‘active ~ passive’ one.
In Proto-Slavic, the main exponent of the passive was the periphrastic construction, but reflexive passives have been reconstructed, as well (Reference Siewierska and ShibataniSiewierska 1988: 257ff.). For the passive in Old Church Slavonic, well documented, see Reference LuntLunt (2001), Reference Krause and SlocumKrause & Slocum (2013), and Reference Malicka-KleparskaMalicka-Kleparska (2016), among others.
21.7 Some Voice-Like Phenomena in Slavic
As we just mentioned, the constructions implementing voices are not uniquely used for this purpose: the same construction can express a number of other phenomena. Therefore, in some cases it may not be immediately clear whether we are dealing with voice or something that looks like it.
Formally identical constructions: passive [‘be’(Aux) + pass-part] vs. stative/resultative [‘be’(Copula) + Adj]
(30)
a. Prozor je zatvoren (pažjivo). (BCS) window be(Aux)-pres.3sg close-pass-part.sg.masc ‘The window was closed (carefully).’ b. Prozor je zatvoren (ceo dan). window be(Cop)-pres.3sg closed(Adj)-sg.masc ‘The window is closed (all day).’
The ambiguity of the type illustrated in (30) is due to a dual nature of the passive participle – verbal and adjectival – manifesting itself in different degrees, depending on the verb’s lexical meaning (Reference TanasićTanasić 2014: 178–179). It can be resolved in favor of the passive reading by the addition of an agent-oriented adverbial, such as ‘carefully’ (30a); the stative/resultative reading, on the other hand, is compatible with a durative adverbial like ‘all day’ (30b).
(31)
a. Dveri zakryvajutsja (storožem v 9 časov večera). (Rus.) ‘The doors are closed (by the watchman at 9pm).’ b. Dveri zakryvajutsja (sami soboj). ‘The doors close/get closed by themselves.’
A sja-verb form in the imperfective aspect with an inanimate subject can be either a passive form of ZAKRYT’ (V, trans) or a form of ZAKRYT’SJA (V, refl); see Padučeva (2003: 137–139). In (31a), the ambiguity is resolved thanks to the agentive complement ‘by the watchman’, which allows only for a passive reading, while the phrase ‘by themselves’ in (31b) supports only a decausative interpretation of the reported event, portrayed as occurring spontaneously or being somehow the “responsibility” of the subject.Footnote 17
Various diagnostics have been developed to help tell apart such superficially identical constructions; see, for instance, Reference KallulliKallulli (2007) and Reference Alexiadou and DorianAlexiadou & Dorian (2012).
Different (albeit semantically closely related) lexemes presenting diathetic alternations NOT formally marked on the verb (cf. Levin 1993).
(32)
a. namazati puter na hleb ~ namazati hleb puterom (Ser.) ‘to spread butter on bread’ ~ ‘to spread bread with butter’ b. kritikovati Vladu za njenu lošu politiku ~ kritikovati Vladinu lošu politiku ‘to criticize the Government for its bad policies’ ~ ‘to criticize the Government’s bad policies’
Involuntary state construction, a.k.a. dispositional/desiderative (Reference Franks and FranksFranks 1995: 364–369, Reference Rivero and Milojević SheppardRivero & Milojević Sheppard 2003, Reference Frąckowiak and RiveroFrąckowiak & Rivero 2008, Reference Marušič and ŽaucerMarušič & Žaucer 2014), which adds meaning.
(33)
The added meaning is not the same across Slavic – ‘feel like’ (33a, b) vs. ‘be able to’ (33c, d, e) – nor is the syntactic construction implementing it: (33a, d, e) vs. (33b, c).
Indefinite and zero subject constructions (Reference Mel‘čukMel’čuk 2006b), which have semantically full (non-empty) and phonetically zero subjects: ‘something’/‘natural force’ (34) or ‘some people’ (35):
(34)
a. Ego ubilo molniej. (Rus.) he-acc kill-past.sg.neu lightning-sg.ins ‘He was killed by lightning.’ b. Mist zneslo povodkom. (Ukr.) bridge-sg.acc sweep.away-past.sg.neu flood-sg.ins ‘The bridge was swept away by the flood.’
(35)
a. [32-godišnjak, pijan; vozio kamion i skrivio nesreću.] (colloquial Cro.) A 32 year old, drunk; he drove a truck and caused an accident. Uhitilo ga i odmah arrest-act.part.sg.neu he-acc and immediately strpalo u zatvor na 14 dana. pack-act.part.sg.neu in jail for 14 days ‘He was arrested and immediately put in prison for 14 days.’ b. Murata oslobodilo od vojske. (colloquial Bos.) Murat-sg.acc free-act.part.sg.neu from military ‘Murat was freed from military service.’
21.8 Summary of the Main Issues in the Study of Slavic Voices
As this chapter has shown, there is a wealth of scholarship on grammatical voice; however, it is characterized by widely divergent approaches, resulting in descriptions that are difficult to compare and evaluate.
Some of the issues discussed in the literature on grammatical voice (in Slavic and elsewhere) follow.
Morphological Issues
Is voice an ‘ill-behaved’ inflectional category? Do all verbs have a voice in a language that possesses this category? For instance, can deponent verbs (inherent reflexives, etc.) be said to have voice at all? This is the problem of the application of grammemes of voice to specific lexemes (or, put differently, the regularity/productivity of voices). As we have seen, not all voices are equally applicable to all lexemes of a language and one can speak of ‘major’ and ‘minor’ voices (e.g. full promotional passive vs. partial demotional passive in BCS, respectively). This is a serious problem that in a sense throws a shadow on the whole idea of voice as an inflectional category.
Is formal marking of specific voices a defining feature of the category of voice?
Semantic/Syntactic Issues
To what extent are specific voices allowed to change the propositional meaning of the base verb? What exactly does it mean that they should not do so ‘substantially’?
Is the agentless promotional passive a full-blown voice in Slavic? Is there an antipassive voice in Slavic? Is reciprocal a voice or a special case of the reflexive/a particular derivation?
How do specific voices relate to impersonal constructions? Is impersonalization relevant for establishing the inventory of voice types or is it, rather, an ancillary operation which participates in the surface-syntactic implementation of some specific voices?
Discourse-related Issues
How/why are specific voices used in discourse? What factors – discourse prominence, information structure, text genre, style, etc. – are in play?
What is the frequency of use of oblique voices (with respect to the unmarked active)?
Notional and Terminological Issues
While these issues are not often discussed, they are indeed crucial for a better understanding and fruitful exchanges among linguists.
An integrative/unifying approach to voice in Slavic (and beyond) has yet to be found.
