Old Church Slavonic as a language with the middle voice morphology.
Old Church Slavonic (OCS) is described in some major sources dealing with the grammatical description of the language as a language characterized by two voices: active and passive. Such an account is offered by e.g. Brajerski (1966), Lunt (2001), and Krause & Slocum (2013). In these sources the category of voice is based on mixed semantic and morphosyntactic criteria. The subject functioning as the performer of an activity is implicitly essential for building the active voice, while the patient-like function of the subject together with the verbal complex based on the auxiliary verb and the passive participle of the lexical verb are building blocks of the passive voice. Lunt (2001) explicitly includes reflexively marked verbs in OCS into the passive voice structures. (1) It has to be stressed here that the voice system has been given very little attention in these sources, and the active-passive juxtaposition is taken for granted. We will argue in this text that OCS is indeed a system with two voices (at the point when a transitory three-voice system is being formed). However, the voice which functions side by side the active voice is the middle voice, not the passive voice. We will present an analysis showing that the so-called reflexive forms in OCS are in fact middle forms, fulfilling the semantic criteria for middle predicates, as presented in Kemmer (1993). (2) At the same time, we will link the voice system with changing valency values of the verbs, (3) (see Babby 1998, Alexiadou & Doron 2012) rather than with specific inflectional patterns (as was the case in Classical Greek). If the latter were considered, OCS would have just the active voice as OCS verbs reveal a type of inflection not connected with their meaning. However, following Schenker (1988), we believe that systematic pairing of a specific form with equally systematically occurring meaning is a minimum and sufficient condition for proposing the existence of a linguistic category. Such is the case of the reflexively marked verbal forms in OCS: they appear in clauses with a single affected participant in the event (even if this participant affects itself). (4) The middle voice is distinguished from the passive voice both formally and semantically. The passive voice implies the presence of another, distinct argument involved in the event and it is not marked with the reflexive formative. This voice is still rarely used in OCS, but in time it will replace the middle voice of OCS in modern Slavic languages. Such an understanding of the concept of voice allows us to capture an interesting property of OCS - the correlation of the reflexive morphology marker with the specific middle semantics and valency, (5) as well as with the scarcity of passive structures. The reflexive formant se marks middle forms and middle forms are almost exclusively marked by the reflexive se (barring few deadjectival and denominal inchoatives).
The paper will begin with a short presentation of Kemmer's (1993) conception of middle verbs, as we will argue that OCS "reflexive" predicates answer her characterization of middle predicates. She devotes her work to the analysis of the category of middle verbs, which are characteristically marked in various world languages. The semantic class of predicates, the types of event structures they appear in, as well as the ways of morphological marking of the category and its status in particular language systems are analyzed by Kemmer (1993) in great detail.
The gist of this paper lies in the matching of OCS verbs marked with the reflexive clitic se with Kemmer's brands of middles to show that the OCS verbs answer the semantic description of middles. Then we show that event structures in which OCS middles appear are characteristic for middle formations and that the marking they bear is morphological, and not syntactic in nature. This allows us to claim that OCS is equipped with the active--middle voice system. Additionally, the position of the passive voice in OCS is briefly discussed.
2. Kemmer's (1993) verbs with middle meaning
Kemmer's (1993) point of departure into the area of middle predicates is the semantics of such verbs (see also Condovardi 1989), following the tradition of Lyons (1969). Kemmer (1993) stresses that a critical issue which should be taken up in characterizing forms as middle is their occurrence with an affected participant in the event. The participant undergoes some change of state, or is just in a state, where some shades of the middle meaning may answer this description more or less literally. This participant is the subject of the middle clause (Kemmer 1993: 8).
Another essential feature of middle predications is "low degree of the elaboration of events" (Kemmer 1993: 8), in other words, the events spelled-out by middle verbs are largely limited to the participation of the subject, which is affected by the event at the same time, or is in the state introduced in the event, and the roles of other participant are eliminated or downplayed. With some middles (e.g. facilitative ones) circumstances characterizing events are also eliminated, such as e.g. time and place considerations.
Kemmer (1993: 16-21) characterizes particular groups of middle verbs which are frequently equipped with specific morphology in languages of the world and considers middle verb clauses as a transitory link between fully elaborated transitive event structures with articulated agentive participants (typically accompanied by patient participants) and typical mono-argumental intransitives with a non-affected subject argument. This view is challenged for instance by de Schepps (2010), according to whom the relationship between transitivity and intransitivity constitutes a whole network of relationships, non-linear in nature. De Schepps views the conception of the transitory middle voice as an oversimplification of a much more complex theoretical problem. In this text we do not issue any claims as to universal tendencies in the voice systems of the world, however the particular claim made by de Schepps that reflexive and reciprocal meanings constitute the step between transitive and intransitive structures is not upheld by OCS material, in which anticausatives (and not reflexives or reciprocals) outnumber other groups of reflexively marked verbs. It seems, however, that OCS morpho-syntax supplies the material which is in agreement with Kemmer's (1993) views on middle semantics and event structure. Judging by the uniform reflexive marking and relatively unified semantics of OCS verbal categories, it seems that the broad concept of middle verbs as proposed by Kemmer plays an important role in the morpho-syntactic system of this language.
Polish examples will be given below as an illustration of Kemmer's classes of middles because English does not offer any formal markers identifying middle verbs, while such marking is available in Polish in the form of the reflexive clitic sie. Kemmer's middles include: grooming verbs (czesac sie 'comb'), verbs of body movement without changing the position of the whole body (non-translational middles, e.g. klaniac sie 'bow'), self-benefactive middles (prosic sie, colloquial, 'beg'), naturally reciprocal events (witac sie 'greet'), translational motion verbs (przechadzac sie 'stroll around'), emotion middles (gniewac sie 'be angry'), verbs of emotional speech (modlic sie 'pray'), cognitive middles (zastanawiac sie 'wonder'), spontaneous events (psuc sie 'get spoiled'), logophoric middles (widzieli sie bogatymi '[They] saw themselves rich'), anticausatives (tamac sie 'break') and facilitative middles (dispositional middles) (prac sie tatwo 'wash easily').
Kemmer (1993) stresses it in her work that middle verbs can be deponents, i.e. they need not have transitive correlates in the system of a language. The intransitive verbs with suitable semantics and morphology which appear in appropriate event structures qualify as middles equally well as the verbs with transitive correspondents based on the same roots (see also Rokoszowa 1978, 1979 for very similar views).
In the next section we will show that the reflexively marked verbs in OCS answer Kemmer's semantic characteristics and have other properties of middle verbs.
3. Old Church Slavonic middle verbs
3.1. Morphological marking
Before we embark on the introduction of particular verbal classes in OCS, let us enlarge on the morphological marking of these verbs. The verbs we are preoccupied with are marked with the presence of the morphological clitic element se, which according to traditional grammars (see e.g. Lunt 2001) constitutes a pronoun in the accusative case--a short form of the full pronoun sebe, which in OCS may mark either the accusative case or the genitive case. We will show in Section 4 that there is little ground, if any at all, to treat se as a pronominal element. Consequently, we consider it to be a derivational formative characterizing the middle voice, i.e. the head of the middle voice projection in the formation of middle voice constructions. The presence of the formative will be shown to correlate with the middle semantics of verbs, as characterized by Kemmer (1993), and with their intransitive characteristics, as well as with simple event structure of the clauses with such verbs. Se will be shown to be independent of pronominal arguments that may appear in OCS clauses and to possess different properties than other pronouns.
As far as the phono-morphological status of se is concerned, it appears to be a clitic, which typically follows the verb in OCS clauses, but, occasionally, it may appear in other places in the clause. This distributional freedom testifies to its non-suffixal nature (although, as will be stated in Section 4, there is some evidence suggesting that it may be at the beginning of its way to becoming a suffix). Below we illustrate some positions occupied by se in sentences: (6)
(1) a. Post-verbal position: ne protiviti se zulu not oppose.INF REFL evil.DAT.SG 'Do not oppose the evil.' b. Pri-verbal position: cto se pecete which.ACC.SG REFL care.IND.PRES.ACT.2NDPL Why do you worry?' c. Other: dastu bo se vamu give.IND. because REFL you.DAT.PL PRES. ACT. 3rd. SG 'because it is given onto you'
Because the positioning of se is relatively free (although the post-verbal location is definitely preferred), we classify it as a clitic element. Let us stress here that the positions characteristic of the reflexive clitic do not coincide with the positions of pronominal clitics in Early Slavic (see e.g. Jung & Migdalski 2014, Migdalski & Jung 2015).
Among the verbs marked with se in OCS, the most numerous group encompasses anticausative verbs, i.e. the intransitive verbs which appear in the clauses whose subjects undergo changes. In some morpho-syntactic systems, anticausatives have corresponding transitive verbs based on the same root. The subjects of the transitives are interpreted as causers of the changes of their internal arguments, the very same arguments that could function as subjects of anticausative intransitives. In OCS the prevailing majority of middle verbs belong to this group. Below we include just some examples of such middles in clausal context, so the simplicity of anticausative event structures can be appreciated. In particular, such middles are never accompanied by agentive arguments realized in the same way as in passive structures (i.e. as arguments in the instrumental case, see Section 6 for details).
(2) Anticausatives: a. ize viznesetu se who.NOM.SG rise.IND. PRES. ACT. 3rd. SG REFL sumeritu se i fall.IND. PRES. ACT. 3rd. SG REFL and REFL sumerejei fall.PART. PRES. ACT. NOM. SG viznesetu se rise.IND. PRES. ACT. 3rd. SG REFL 'Who rises will fall, falling--will rise' b. nicutoze taino eze ne nothing. secret. which not NOM.SG NOM.SG avitu se appear.IND.PRES.ACT.3RD.SG REFL 'All secrets will be revealed.' c. cto hostc aste juze i which. want.IND. in order to also ACC.SG PRES.1ST. SG Vuzgore se enflame.IND.AOR.ACT.3RUSG REFL 'which I want to have come into flames' d. nacese kunizunici i begin.IND. AOR. learned.NOM.PL. M and ACT. 3rd. PL farisei ljute eko gnevati se pharisee.NOM.PL.M. strongly.ADV.POS as be angry.INF. PRES REFL 'The learned and the pharisees began to get very angry.' e. greje se pri svesti warm.PART. PRES. ACT. NOM. SG REFL at flame.LOC. SG 'heating up at the flame' f. sadu [...] iskorenitu se plant.NOM. SG uproot.IND. PRES. ACT. 3rd. SG REFL 'The plant will get uprooted.' g. ispluni se vreme fulfill.IND. AOR. REFL time.NOM.SG ACT. 3rd. SG 'The time got reached.'
As the examples above show, anticausatives are created extensively in OCS: for instance, English translations have to render examples (2 b, f, g) above with passive structures. Anticausatives appear with subjects in the nominative case, as is evidenced by the examples with lexically specified, non-pronominal subjects (2 b, d, f, g), and the subjects are affected in the evens. Neither agents, nor other causers of the events are regularly mentioned in the event structures.
Apart from the verbs which render events whose subjects go into some state, a class of stative verbs is characterized by the same clitic morphology in OCS. These verbs situate their subjects in a given state. Lyons (1969: 373) describes such verbs as middles, and in OCS they show all the characteristics of middles: a single subject argument situated in a state, simple event structure, no causer of the state, the se clitic. Some examples of the use of such stative verbs are given in (3) below:
(3) a. da ne avisi se clvkmu in order not seem.IND. REFL person.DAT.PL PERS.2ND.SG poste se fast.PART.PRES.NOM.SG REFL 'not to seem fasting to people' b. odenie ego bliscaje clothes. his.GEN. glisten.PART.PRES.ACT.NOM.SG NOM.SG 3RD.SG se REFL 'His clothes glisten.' c. na podunebeskojo svititu se on sky.POS.ACC.SG shine.IND.PERS.ACT.3RD.SG REFL '[It] shines in the sky.'
The clauses in (3) above specify the states in which the subject arguments appear, rather than changes of states, however they assert something about the arguments, and, consequently, can be classified among middle verbal uses.
3.4. Subject experiencer verbs
Another extensive group of middle verbs consists of subject experiencer verbs--i.e. emotion middles in Kemmer's (1993) terminology. (7) Like with the previous groups of cliticized verbs, the clauses with subject experiencer verbs show characteristics of middle predications: the appropriate semantics, morphological marking and simple event structure:
(4) a. blazneho se o nemi doubt.IND.IMP.ACT.3RD.PL REFL about he.LOC.3RD.SG '[They] doubted about him.' b. ne boite se not be afraid.IMP.2ND.PL REFL 'Do not be afraid!' c. dse [...] veseli se soul.VOC.SG be merry.IMP.PRES.ACT.2ND.SG REFL 'Oh soul, be merry!' d. divi se za wonder.IND.AOR.ACT.3RD. SG REFL about neverustvo ihu doubt.ACC.SG.N they.GEN.3RD.PL '[He] was amazed about their doubts.'
The subjects of the above verbs are affected as a result of emotional changes that they undergo. The fact that these changes are not physical does not seem to have any bearing on their middle status in OCS.
3.5. Other verbs with middle semantics
Apart from the (most numerous) groups introduced above, se clitics accompany some other categories of verbs which, according to Kemmer (1993), are frequently morphologically marked middles. Examples in (5) below include non-translational motion verbs, as well as a reciprocal:
(5) a. Non-translational motion verbs: ne vuzvratitu se not turn back.IND.PRES.ACT.3RD. SG REFL '[He] does not go back.' sily nebeskyje dviginotu se power.NOM.PL heaven.POS.NOM.PL raise.IND.PRES.ACT.3RD.PL REFL 'Heavenly powers rise.'
b. A reciprocal verb:
suburase se vu kupe gather.IND.AOR.ACT.3RD.PL REFL in group.ACC.SG '[They] gathered together.'
As Kemmer (1993) stresses, not all middle verbs possess corresponding transitive counterparts. She presses the point that it is erroneous to look for middle characteristics only among the verbs which have corresponding transitives (see also Rokoszowa 1978, 1979). Middle characteristics are independent of transitivity, as the external argument of the transitive verb--in those cases where transitives exist--in any case is not realized in the middle predication. Thus among the middles in OCS we find such deponent verbs as e.g. in (6) below:
(6) Deponents: klanese se emu bow.IND.IMPERF.ACT.3 REFL him.DAT.3RD.SG '[He] bowes to him.' pomoli se oticu 8 pray.IMP.PRES. ACT.2ND. SG REFL father.DAT.SG 'Pray to your father! togda nacqtu rotiti se i then begin.IND.AOR.3RD.SG seear.INF.PRES REFL and kleti se vow.INF.PRES REFL 'Then [he] began to swear and vow.'
These verbs can be classified among middles because of their characteristic semantics: Kemmer (1993) notices that in various languages middle marking is observed on verbs of speech actions of the emotive type, and the verbs in (6) definitely belong to this class.
Quite unexpectedly, the middle marking se also appears in impersonal structures of a certain type. These impersonals are formed in a specific way: no subject argument is present in the clause, the verb is typically in the third person singular, although exceptionally the plural form appears. (9) The only argument which is spelled out in such clauses is the applied argument, i.e. the argument in favor of which the event takes place. Notice that in this way the argument is also affected by the event so described, so, semantically speaking, we are also dealing here with a middle situation. The verbs which appear in these impersonal structures are not typical middle verbs though, but the ones that fall outside the 'notional middle' group. Below we give a few examples of such verbs:
a. njoze mero merite which.INST.SG measure.INSTR.SG measure.IND.PRES.ACT.2ND.PL vizmeritu se vamu measure.IND.PRES.3RD.SG REFL you.DAT.2ND.PL 'With which measure you measure, so it will be measured for you.' b. prosite i dastu se ask.IMP.PRES.ACT.2ND.PL and give.IMP.PRES.ACT.2ND.SG REFL vamu you.DAT.2ND.PL 'Ask and [it] will be given to you.' c. tlucete i otvruzetu se knock.IMP.PRES.ACT.2ND.PL and open.IND.PRES.ACT.3RD.SG REFL vamu you.DAT.2NDPL 'Knock, and [it] will open for you.'
The verbs in the examples above do not come easily under the category of middles. Neither giving (7b), nor measuring (7a) result in significant changes of states of their direct arguments. Opening may be seen in this light, but it may also be conceived of rather in terms of giving access to some place, and it is used in this sense in (7c), rather than for the change of state of an argument from closed to open. However, as we have stated above, one argument in the above propositions is seen as affected, and it is the applied argument. Consequently, we would like to claim that the impersonal constructions illustrated above spell out in OCS these event structures which do not lend themselves easily to conceptualization in terms of the internal argument being affected. Some arguments, however, more precisely the beneficiaries of such events, could be conceptualized as affected, and hence se middle structures come into play.
Kemmer (1993: 148) mentions impersonal structures as possible cases of middle situations, but she refrains from discussing the point in any detail. Similarly, Rokoszowa (1978, 1979) mentions impersonals as possible cases of 'reflexive' voice (middle voice) in Polish.
The fact that superficially non-middle verbs can appear in such structures that fill the gap in the formation of the middle voice constitutes an additional argument in favor of stating that OCS is a language with a middle voice grammar component.
In Section 3 we have shown that the verbs marked in OCS with se represent middle semantics and event structure. Additionally, the impersonal se structures fill up the gaps in the middle voice paradigm. What remains to be argued for is the status of the clitic element, which we want to see as a morphological marker of the middle voice constructions, and not as a pronominal argument of a transitive verb.
4. The status of se
In descriptive grammars of OCS (e.g. Brajerski 1966, Lunt 2001, Krause & Slocum 2013), se is presented as a reflexive pronoun in the accusative case, a variant of sebe, which, in turn, is syncretic with the reflexive pronoun in the genitive case. It has to be made clear, however, that the full form does not ever accompany the verbs with middle characteristics, so, in fact, we have no evidence that sebe is a full variant of se.
Most of the verbs which appear in the same clauses as sebe take the genitive case in OCS, so they could not co-occur with se, which according to descriptive sources realizes the accusative case. However, for instance the verbs: supasti 'save' and sutvoriti 'consider' take complements in the accusative case. They may also be conceived of as capable of appearing in the middle voice since they have the appropriate semantics, with an affected direct object (supasti), or being a verb of cognition (sutvoriti). (10)
(8) sebe ne mozetu supasti self.GEN.3RD.SG not can.IND.PRES.3RD.SG save.INF.PRES. ACT '[He] cannot save himself.' a sebe pogubitu but self.GEN.3RD.SG lose.IND.PRES.3RD.SG 'but [he] will lose himself ni sebe dostoina not self.GEN.3RD.SG deserving.POS.GEN.SG sutvorihu priti ku consider.IND.AOR.ACT.1ST.SG come.INF.PRES.ACT to tebe you.DAT.2ND.SG '[I] did not consider myself deserving enough to come to you.'
Nevertheless, they do not appear with se, which could be expected were se and sebe weak and strong forms of the same pronominal argument.
Some other evidence against treating se and sebe as variants of the same pronoun comes in the form of such structures which admit se and sebe side-by-side. The sentence in (9) below suggests strongly that se is a word-formational element: (11)
(9) da otuvruzetu se sebe in order to renounce.IND.PRES.ACT.3RDSG REFL self.GEN.3RDSG 'in order to renounce oneself'
Still another piece of evidence is supplied by the verb naceti 'begin', which is often spelled together with the clitic element, testifying to the functional, rather than referential status of the morpheme: (12)
(10) nacese moliti i begin.IND.AOR.ACT.3RD.PL beg.INF.PRES.ACT he.ACC.3RD.SG otiti otu predelu ihu go.INF.PRES.ACT from border.GEN.PL they.GEN.3RDPL '[They] began beg him to go away from their borders.'
Similarly, if se was a pronoun, it should become the subject of passive clauses, which occasionally appear in OCS. However, when the middle structure is made passive (which happens quite exceptionally), se stays with the lexical verb, as if no argument rearrangement took place:
(11) sotona sotono izgonitu na se (13) satan.NOM.SG satan.ACC.SG chase.IND.PRES.ACT.3RD.SG on REFL razdelilu se estu divide.PART.RESULT.ACT.NOM.SG REFL be.IND.PRES.ACT.3RD.SG 'Satan chases Satan away, quarreling with himself
All the grammatical phenomena we have enumerated in this section point univocally to se as a morphological formative, and not a pronominal argument. It has to be stressed additionally that to have just one form of a pronoun for all persons and numbers in its weak form goes against the grain of OCS as a whole. The language is very strongly inflectional, with pronouns inflected for person, case and number. Se would be unprecedented in this system, having just a single morphological shape.
5. Event structures for middle voice
The data in the last two sections suggest that an extensive group of verbs in OCS realizes middle semantics and is marked with a morphological formative in a uniform way. As these examples show, also the event structures into which the verbs are inserted remain simple. The external causer argument does not appear with the relevant verbs. The research into event structures reflected by particular classes of verbs dates back to Vendler (1967), Dowty (1979), Rappaport Hovav & Levin (1998, 2001). Recent extensive analyses of the event structure associated with verbs with middle semantics carried out on non-Slavic languages attribute to them the causative element which is associated with such verbs (see e.g. Kallulli 2007, Koontz-Garboden 2009, Alexiadou et al. 2015). The arguments used as justification for such a position come mostly from the area of prepositional phrases that can accompany such verbs. These phrases, although introduced by different prepositions than the agentive arguments in passive structures, spell out the arguments that perform the function of (or analogical to) external arguments (Agents, causes) in middle clauses. Consequently, there arises the problem whether in OCS the event structure expressed by clauses with middle verbs also includes the Agentive participant. Such a concept would allow us to view the voice phenomenon as abstaining from changing the basic meaning of a proposition expressing the middle meaning with respect to the active clause meaning (as for instance Mel'cuk's 1993 conception of the notion of voice would require).
The closest that middle verbal structures come to the realization of the causer argument are sentences with otu phrases. Contrary to Madariaga (2010), we do not see these phrases as agentive in nature. They contain the arguments which are typically not human, and even not animate. Thus they do not reflect the canonical external arguments of the corresponding transitive verbs. They usually evoke natural phenomena (see also Malicka-Kleparska 2015):
(12) ististi se out prokazy clean.IND.AOR.3RD.SG REFL from pestilence.GEN.SG '[He] cleaned himself from pestilence.'
The possibilities of introducing the agentive participant into the event structure with an affected subject are very limited in OCS, and reserved for infrequent analytic passive constructions, to be discussed in the next section.
6. OCS passives
In OCS passives are few and far between. They are analytic structures in which the auxiliary verb is in the tensed personal form, while the lexical verb takes the shape of the past passive participle. The original objects of active verbs appear as subjects in passive constructions, while the original agentive subjects may take the form of instrumental phrases. Consequently, the event structures of transitive active and passive clauses are not that different, unlike middle clauses, which lack one of the key participants--the Agent/causer. Below we give some examples of passive structures, but we stress once again that they are quite rare in the body of data which we consider here:
(13) vise mine predana sotu all me.DAT.1ST.SG give.part.past.pass.nom.pl be.IND.PRES. ACT.3RD.PL oticmi moimi father.INS.SG my.INS.SG 'All are told to me by my father.' vise telo [...] vuvruzeno all.NOM.SG body.NOM.SG throw.PART.PAST.PASS.NOM.SG bodet vu geono be.IND.FUT.ACT.3RDSG in hell.ACC.SG 'The whole body will be thrown into hell.' vinemlete milostyne vaseje ne care.IMP.PRES.ACT.2NDPL good deed.GEN.PL your.GEN.PL not tvoriti predu clvky da do.INF.PRES.ACT before person.INSTR.PL in order vidimi bodete imi see.PART.PRES.PASS.NOM.PL be.IND.FUT.ACT.2NE.PL they.INS.3RDPL 'Take care so that you do not do good deeds for them to be seen by people.'
Although both passive and middle constructions may target the same affected arguments as their subjects, OCS expresses such situations predominantly by using the middle voice.
The picture of the voice morpho-syntax of OCS, as it emerges from the data, shows the prevailing two-voice system, consisting of the active and middle voices. The third voice--passive --is in its initial stages of development. This image tallies with the proposal by Brondal (1943), who claims that grammars with three-voice systems constitute a phase in the development of two-voice systems. We believe that OCS is precisely such a system in flux, in which the middle voice will give way and in effect the active-passive oppositions of Present-day Slavic languages will arise.
Suggestions concerning the possibility of a third voice in Slavic have already been voiced for instance by Rokoszowa (1978, 1979), although the picture of the "reflexive" voice which Rokoszowa presents differs from Kemmer's (1993) middle structures. The "reflexive" voice is defined by personal subject participants only, so it is significantly less general than the view we have adopted here. Rokoszowa attributes the more traditional distinction into active and passive voice to the approach to voice morpho-syntax as a transformational phenomenon (see e.g. Chomsky 1957, Kurylowicz 1964). Namely, if we consider voices as resulting from simple rearrangements of sentence structure, then the passive voice is a natural mirror reflection of the active voice, with the original Agent being expressed as an oblique argument, while the original object - as the subject. The middle voice, with its mono-argumental structure, is neither here nor there in this picture. Rokoszowa (1978, 1979) stresses the fact that the problem of transitivity should be dissociated from the concept of voice: once we give up the transformational view, we may have a totally different picture. She also claims that the transformational cline results in dispositional middles being recognized and described extensively in the literature on morpho-syntax (although their actual uses are not frequent), as their place in the opposition is clear--they imply the agentive participant of the event.
Following Rokoszowa's (1978, 1979) and Kemmer's (1993) reasoning we claim that voice systems are not based on transitive--intransitive oppositions/correspondences, but on types of participants and complexity of event structures. Such an approach to voice is reasonable especially in the approaches to morpho-syntax which base verbal structures on roots and are non-lexicalist (see e.g. Pylkkanen 2008, Alexiadou 2010, Alexiadou & Doron 2012). Each verb is built from scratch, upwards from the root element, independently of any other existing lexical items. Active and middle verbs differ in the presence of two voice projections: active or middle, while the properties of these projections impose characteristic event structures. Consequently, the morpho-syntax of voice makes space for deponent verbs, which include "notional" middle predicates, i.e. such that take affected subjects, have simple event structure and characteristic morphological marking.
Alexiadou, Artemis. 2010. On the morpho-syntax of (anti)causative verbs. In Rappaport Hovav, M., Doron, E. & Sichel, I. (eds.), Lexical semantics, syntax, and event structure, 177-203. Oxford: Oxford University Press.
Alexiadou Artemis & Doron Edith. 2012. The syntactic construction of two non-active voices: passive and middle. Journal of Linguistics 48. 1-34.
Alexiadou, Artemis, Elena Agnostopoulou & Florian Schafer. 2015. External arguments in transitivity alternations. A layering approach. Oxford: Oxford University Press.
Babby, Leonard. 1998. Voice and diathesis in Slavic. (http/seelrc.org/glossos) (Accessed 2015-12-12).
Brajerski, Tadeusz. 1966. Jezyk staro-cerkiewno-siowianski. Lublin: Towarzystwo Naukowe Katolickiego Uniwersytetu Lubelskiego.
Brondal, Viggo. 1943. Essais de linguistique generale. Copenhagen: Munksgaard.
Chomsky, Noam. 1957. Syntactic structures. The Hague & Paris: Mouton.
Codex Marianus. (http://www.helsinki.fi/slaavilaiset/ccmh/) (Accessed 2015-12-12).
Condoravdi, Cleo. 1989. The middle: where semantics and morphology meet. MIT Working Papers in Linguistics 11. 18-30.
Corpus PROIEL. (http://www.tekstlab.uio.no:3000/users/sign_in) (Accessed 2015-12-12).
De Schepper, Kess. 2010. The space between one and two. Transitives, intransitives, and the middle voice. In Brandt, P. & Garcia Garcia, M. (eds.), Transitivity: Form, meaning, acquisition, and processing, 191-208. Amsterdam & Philadelphia, PI: John Benjamins.
Dowty, David. 1979. Word meaning and Montague grammar: The semantics of verbs and times in generative semantics and in Montague's PTQ (Studies in Linguistics and Philosophy 7). Dordrecht: Reidel.
Fried, Mirjam. 2006. Agent-backgrounding as a functional domain: Reflexivization and passivization in Czech and Russian. In Lyngfelt, B. & Solstad, T. (eds.), Demoting the agent: Passive, middle, and other voice phenomena, 83-110. Amsterdam & Philadelphia, PI: John Benjamins.
Jung Hakyung & Migdalski Krzysztof. 2014. On the degrammaticalization of pronominal clitics in Slavic. FASL 23. Michigan Slavic Publications. 418-437.
Kallulli, Dalina. 2007. Rethinking the passive/anticausative distinction. Linguistic Inquiry 38. 770-780.
Kemmer, Suzanne. 1993. The middle voice. Amsterdam & Philadelphia, PI: John Benjamins.
Koontz-Garboden, Andrew. 2009. Anticausativization. Natural Language and Linguistic Theory 27. 77-138.
Krause, Todd & Slocum, Jonathan. 2013. Old Church Slavonic online. (http://www.utexas.edu/cola/centers) (Accessed 2015-12-12).
Kurylowicz, Jerzy. 1964. The inflectional categories of Indo-European. Heidelberg: Carl Winter.
Lunt, Horace G. 2001. Old Church Slavonic grammar. Berlin & New York, NY: Mouton de Gruyter.
Lyons, John. 1969. Introduction to theoretical linguistics. Cambridge: Cambridge University Press.
Madariaga, Nerea. 2010. The development of Indo-European middle-passive verbs, a case study in Ancient Greek and Old Church Slavonic. Indogermanische Forschungen. Zeitschrift fur Indogermanistik und allgemeine Sprachwissenschaft 115. 149-178.
Malicka-Kleparska, Anna. 2015. A case for two voices in Old Church Slavonic--reflexively marked OCS verbs. Linguistics Beyond and Within 1. 137-154. (http://lingbaw.com/index.php) (Accessed 2015-12-26).
Malicka-Kleparska, Anna To appear. Dispositional middles in English and Polish with detours into other Slavic languages. English versus Slavic. Bern: Peter Lang.
Manney, Linda. 1998. Middle voice in Modern Greek. Amsterdam & Philadelphia, PI: John Benjamins.
Mel'cuk, Igor. A. 1993. The inflectional category of voice: towards a more rigorous definition. In Comrie, B. & Polinsky, M. (eds.), Causatives and transitivity, 1-146. Amsterdam & Philadelphia, PI: John Benjamins.
Migdalski, Krzysztof & Hakyung Jung. 2015. Diachrony of cliticization patterns in Slavic. Paper presented at the conference of the Societas Linguistica Europea, Leiden, 2-4 September 2015.
Pylkkanen, Liina. 2008. Introducing arguments. Cambridge, MA: The MIT Press.
Rappaport Hovav, Malka & Beth Levin. 1998. Building verb meanings. In Butt, M. & Geuder, W. (eds.), The projection of arguments and compositional facors, 97-134. Stanford, CA: CSLI Publications.
Rappaport Hovav, Malka & Beth Levin. 2001. An event structure account of English resultatives. Language 77. 766-797.
Rokoszowa, Jolanta. 1978. Medium a strona bierna. Biuletyn PTJ 35. 85-102.
Rokoszowa, Jolanta.1979. Uwagi o kategorii strony. Studia Gramatyczne 3. 97-127.
Schenker, Alexander. 1988. Slavic reflexive and Indo-European middle: A typological study. In Schenker, A. (ed.), American contributions to the Tenth International Congress of Slavists, 363-383. Sofia, Columbus.
Tabakowska Elzbieta. 2003. Those notorious Polish reflexive pronouns: A plea for middle voice. Glossos. 1-18.
Vendler, Zeno. 1967. Linguistics in philosophy. New York: Cornell University Press.
Anna Malicka-Kleparska, John Paul II Catholic University of Lublin
Department of English
Catholic University of Lublin
Al. Raclawickie 14
(1) Lunt (2001: 158) writes that the passive voice in OCS has no uniform form. It is either expressed by the verb byti plus the passive participle or by a reflexive verb.
(2) See also Manney (1998), who puts forward a similar conception of the middle voice in Modern Greek and points out different shades of meaning of various reflexive middles in this language, some showing greater "agentivity" of the argument realized in the clause (which could be compared with the reciprocals discussed in this text), some showing the argument as an undergoer of spontaneous change (like in the anticausatives in this text) and realized in Greek by middle inflectional reflexives. The approach to middle verbs as a class varying semantically, though realized by the same (reflexive) morphology is also presented by Tabakowska (2003) for Modern Polish.
(3) This conception of voice does not follow the strain of thought that views voice as a meaning preserving inflectional paradigm, as it is presented by e.g. Mel'cuk (1993).
(4) Schenker (1988) additionally claims that reflexives are the manifestation of the middle semantics in Slavic, although he does not identify reflexives with voice phenomena.
(5) Middle voice understood in this way has to be distinguished from constructions which are more frequently named "dispositional middles", which appear in various Indo-European languages. In these constructions the presence of another argument is implicit and these structures most generally specify the disposition of the subject argument to be involved in a situation named by the verb--see e.g. Fried (2006) for an analysis of such constructions in Czech and Russian. We have found no dispositional middles in OCS, but as these constructions are relatively rare in modern Slavic languages (see also Malicka-Kleparska to appear), their lack in the data constitutes no definite proof that they did not exist there.
(6) All the examples are taken from Codex Marianus in the Corpus Cyrillo-Methodianum Helsingiense, available online and entered in the references. The glosses have been verified against the glossing in the corpus PROIEL, also available online, while the transliteration system has been adopted from Lunt (2001). We have decided on giving examples in transliteration because the precise phonological shape of the sentences we have quoted as illustrations is largely immaterial, while the extensive use of various symbols for presumably identical/similar phonological units in the original is unnecessarily confusing.
(7) Cf., however, Madariaga (2010), who gainsays the middle voice structure of such predicates in OCS. She claims that they are reflexive passive structures, so decidedly different than middle ones, first of all because they allow agentive arguments to appear in clauses with such verbs. In Section 5 we will give our reasons for claiming otherwise. See also the discussion in Malicka-Kleparska (2015).
(8) Moliti 'pray' (the non-prefixed form) is sometimes used without the clitic element, although the cliticized forms prevail.
(9) See e.g a plural verb in the reflexive impersonal construction:
predadetu i jezykmu i porogajotu sell. IND. PRES. ACT. 3. PL him.ACC.3RD.SGpagan.DAT.PL and curse. IND. PRES. ACT. 3. PL emu
REFL he.DAT.3RD.SG ' [They] sold him to the pagans and he was cursed.'
(10) Here we give just one meaning of this verb, which in OCS was used in many senses. However, in none of them it was accompanied by se.
(11) For additional argumentation against the pronominal status of se see also Malicka-Kleparska (2015).
(12) This fact also shows that the clitic may be on the verge of becoming a suffix (as is the case in Present-Day Russian with the morpheme -sja).
(13) As pointed out by an anonymous reviewer the status of the element se introduced here by a preposition in unclear. We make here no claims about the status of the element se appearing after the preposition. Any attempt at specifying its status in OCS would require additional studies.
|Printer friendly Cite/link Email Feedback|
|Publication:||SKASE Journal of Theoretical Linguistics|
|Date:||Dec 1, 2016|
|Previous Article:||Searching for competing patterns in morphological derivation: the case of adjective borrowing.|
|Next Article:||Stress rules in loan words in Bedouin Jordanian Arabic in the north of Jordan: a metrical account.|