CC BY
106
J.E. Lavine
Anti-Burzio Predicates: From Russian and Ukrainian to Icelandic’
This paper extends previous work on Transitive Impersonals in Russian and Ukrainian to two constructions with similar properties in Icelandic: the Fate Accusative and New Passive. It is argued that the appearance of accusative in the absence of a nominative-marked NP or Agent occurs on a particular configuration of v-type heads, namely when the transitivity feature (Cause) appears structurally distinct from the head responsible for projecting the external argument (Voice). The Cause head, an accusative "probe", is thereby potentially active in the absence of an Agent, so long as the predicate is causative.
Key words: Event Structure, Fate Accusative, New Passive, Transitive Impersonal, v-type heads, voice, cause, Natural Force, split-vP, Accusative survival.
1. Introduction
This paper explores the appearance of structural accusative case in predicates lacking a higher nominative-marked argument. In previous work, I have referred to such structures as instantiating “Accusative First” syntax [19; 22], in order to draw an explicit contrast with the widely assumed rule of “Nominative First” syntax, which states that accusative is assigned only after nominative is already discharged; i.e., the appearance of accusative is dependent on the prior assignment of nominative [6, p. 80-81; 27; 31, p. 102105; 33; 40]. Nominative-First syntax formalizes the correlation known as “Burzio’s Generalization,” given in (1):
(1) BURZIO’S GENERALIZATION [9, p. 178; 10, p. 139]
A verb (with an object) Case-marks its object if and only if it theta-marks its subject.
Burzio’s Generalization states a correlation between a predicate’s external thematic property (Agent) and its internal case-assigning property (Accusative) - two properties, as has been widely observed, that have no principled connection. So while Burzio’s Generalization accounts for a wide range of facts, such as NP-Movement in the case of passives and monadic
1 I thank the organizers of the Typology of Morphosyntactic Parameters 2012 for inviting me to present my work at their Moscow conference. This paper is much improved as a result of the many productive discussions I had with the conference organizers and participants. I gratefully acknowledge Johannes Gisli Jonsson, Halldor Sigurdsson, and Jim Wood for assistance with the Icelandic data. Some of the ideas presented here were developed in collaborative work with Leonard Babby and Robert Freidin.
Филологические
науки
Лингвистика
unaccusatives, we might also imagine a structure in which the projection (or non-projection) of an Agent argument would have no bearing on the accusative case-assigning potential of a predicate. Consider, for example, the “Fate Accusative” in Icelandic, given in (2), which violates Burzio’s Generalization:
(2) Icelandic fate accusative
a. Batinn rak a land.
boat.the-Acc drove to land
‘The boat drifted ashore.’ [34]
b. Strompinn bles af husinu.
chimney.the-Acc blew off house.the
‘The chimney blew off the house.’ [41, p. 145]
I will argue that the Fate Accusative is a type of Transitive Impersonal: a predicate in which accusative is assigned to the object in the absence of an Agent subject. I will show that accusative in the Icelandic Fate construction is not quirky - it is not a special property of the verbs in (2a-b) (contra [35; 41]; nor is accusative in (2a-b) the product of a special “Fate-v” occurring only in Icelandic (contra [28; 34; 35]). Rather, we will see that accusative appears in this construction as a predictable consequence of the arrangement of more general v-type heads in the language.
In Section 2 I provide an introduction to the “constructionist” approach to verbal meaning, which lays out the framework I adopt for event structure and argument structure. Subsequent sections focus on my analysis of the following “anti-Burzio” predicates: (i) the Icelandic Fate Accusative (3); (ii) the Russian Transitive Impersonal (4); (iii) the Ukrainian Transitive Impersonal (5); and (iv) the Icelandic “New Passive” (6):
(3) Icelandic: fate accusative (repeated from (2a))
Batinn rak a land.
boat.the-Acc drove to land
‘The boat drifted ashore.’ [34]
(4) RUSSIAN: transitive impersonal
Плот унесло волной. [5]
(5) Ukrainian: transitive impersonal1
Хату було спалено блискавкою.
house-Acc was burned.down-pAss lightning-iNs
‘The house was burned down by a strike of lightning.’
1 The Ukrainian Transitive Impersonal is passive in form only. There is no external argument (Agent) at any stage of the derivation, so passivization, as an operation that acts on an Agent argument, “dethematizing it,” cannot be said to have taken place. The Ukrainian construction, on my account, is a dyadic unaccusative (a two-place predicate, both arguments of which are internal to the VP). The verb-form is glossed as passive (pass) for etymological reasons only. Ukrainian also has a genuine impersonal passive, discussed below in Section 5.
(6) Icelandic: new passive
Pad var skodad b^linn af bifvelavirkjanum.
expl was inspected-pAss car.the-Acc by car.mechanic.the
‘The car was inspected by the mechanic.’ [16, p. 294]
The examples in (3-6) share the property of Accusative First syntax. Accusative occurs in each in the absence of a higher nominative argument. The goal in what follows is to provide a unified account for precisely how “independent accusative” occurs under a particular theory concerning the arrangement of v-type heads. An elaborated (“split-”) vP will be proposed in contrast to the earlier minimalist model of Chomsky 1995 (ch. 4), given in (7):
(7) vP
NPagotT'''''" , The representation of v in (7) was designed to
. provide a phrase-structural account of Burzio’s
v VP Generalization: v is endowed with the dual [akt™] properties of projecting an external argument and
—I NP-Acc assigning accusative case, precisely the two
-------* properties that Burzio’s Generalization sought to
unify as a single phenomenon. The elaborated vP to be advanced shortly is designed to unbundle these properties, such that the appearance of accusative case and the projection of an external argument are formally dissociated in the syntax. I show that a split-vP admits the anti-Burzio predicates in (3-6), while preserving the effects of Burzio’s Generalization elsewhere, where Nominative First syntax occurs.
2. V-Roots and v-Heads
The “constructionist” approach to verbal meaning states that different event structures are built via the interaction of lexical roots and functional heads. Let us take the argument-projecting property (the projection of an Agent) to be a function of the “v-voice” variety of v [18, c. 11] and the transitivity property (accusative) to correlate with causation, expressed as “v-cause”, another variety of v [14; 19; 20; 21; 30; 36]. Consider, for example, the different event structures that can be constructed above the single lexical entry for the root verb Vbreak in (8a-b).
(8) English: ±voice/cause
a. The vase broke.
-voice, -cause (non-agentive; anticausative)
b. The child broke the vase.
+voice, +cause (agentive; causative)
On a constructionist analysis, the syntactic mapping of a verb’s arguments is determined by the functional structure in which verbal roots are inserted,
Филологические
науки
Лингвистика
rather than by a verb’s lexical semantics alone. For example, the lexical entry
event, which is true of some affected object. It follows that verbal predicates may differ with respect to their accusative case-assigning potential depending on differences in v, rather than as idiosyncratic properties of root V.
The crucial point in terms of variation between languages is that the properties of voice and cause can appear either bundled on a single v-type head (as in (7)) or be distributed across two heads. Note that in this latter case the transitivity property can be operational regardless of the setting for voice [2; 7; 8; 15; 30]. That is, in the event that voice is not argument-projecting (-voice). accusative appears, so long as the event is caused (+cause), as schematized in the split-vP in (9):
(9) SPLIT-vP
V-voiceP In contrast to English (8a-b), Russian
(11) [± voice/cause] :
a. +voice, +cause (agentive, transitive predicate)
b. -voice, -cause (non-agentive anticausative)
c. -voice, +cause (Transitive Impersonal; non-volitional causation)
d. *+voice, -cause
Our focus is the case of (11c), in which accusative appears in the absence of an Agent. This is the configuration, which, by hypothesis, gives the Russian and Ukrainian Transitive Impersonals (and which is crucially absent in English and other languages without a split-vP). The Icelandic Fate construction and New Passive will be treated as structurally parallel, with respect to v-heads, with Russian and Ukrainian. Note finally that (11d) does not occur: if v-voice projects a sentient, volitional Agent, then, by implication, the event set in motion by the Agent is caused.
for root Vbreak is neither inchoative nor causative; it asserts only a breaking
allows the combination of v-heads illustrated above: [-voice, + cause]. This gives non-volitional causation, as in (10), in which a rolling ball, rather than an Agent, sets the event in motion:1
(10) Russian: transitive impersonal
[+v-gause] VP
VP [-VOICE, +CAusE]
Вазу разбило мячом.
>.NP-acc The possible combinations of v-heads are
given in (11).
1 Compare the ungrammatical English sentence in (i) (on the expletive interpretation for it) with grammatical (10) in Russian:
(i) *It smashed the vase by the wind.
To review, the case of separate settings for voice and cause, as in Russian (10), indicates the need to tease apart the voice and cause properties of v in (7) (similar to the distinct Predication and Transitivity heads in Bowers [7]). In (10), v-cause is identified as active by the presence of instrumental мячом. Indeed, the impersonal construction, in the absence of active voice, forces the reading of ball as setting the event in motion out of human control.
3. The Icelandic Fate Accusative
An immediate empirical problem posed by the Icelandic Fate Accusative is that accusative appears to occur in apparent one-place predicates [34-36], superficially resembling basic unaccusatives (and anticausatives), with no causer argument. It is not clear, then, what sets the event in motion (i.e., how v-cause, the source of accusative, enters the structure). That one-place predicates do not form Transitive Impersonals is illustrated by the Russian data in (12): accusative does not occur with the monadic unaccusative сгореть ‘burn down’ in (12a), in contrast to its two-place counterpart сжечь in (12b), in which the causer argument lightning sets the event in motion, thereby activating v-cause and its accusative reflex (дачу).
(12) Russian: transitive impersonal
a. *Дачу сгорело. b. Дачу сожгло молнией.
Returning to the Icelandic construction, the licit Fate Accusative in (13) superficially resembles ungrammatical (12a) in occurring without an overtly expressed causer:
(13) Icelandic: fate accusative
Batinn fyllti.
boat.the-Acc filled
‘The boat swamped.’
However, note crucially that the reading in (13) is not anticausative, as shown in (14), in which accusative on the sole argument is ungrammatical with dedicated anticausative morphology (-st):
(14) Icelandic: fate accusative Baturinn (*batinn) fylli-st. boat.the-NoM *acc filled-ANTicAus
‘The boat filled up.’ [35, p. 203]
The Fate Accusative in (13) also fails ‘by itself’ modification, otherwise acceptable in anticausatives:
(15) Icelandic: fate accusative
Batinn rak ad landi (*af sjalfu ser).
boat.the-Acc drove at land by itself
[Intended: ‘The boat drifted to land (on its own)’] [39, p. 121]
Филологические
науки
Лингвистика
As in Icelandic (15), the Transitive Impersonal in Russian is likewise incompatible with ‘by itself/on its own’ modification, as indicated in (16), thereby demonstrating that the predicate is two-place and necessarily caused, even if the non-Theme causer argument is left unexpressed, which is most typically the case for the Icelandic expression:
(16) Russian: transitive impersonal
Дачу сожгло (*саму по себе).
Indeed the Russian Transitive Impersonal occurs as well without the overt expression of its causer, which, regardless, is interpreted as present in the argument structure, but sometimes left unexpressed for the simple reason that it is unknown to the speaker, as in the examples in (17):1
(17) Russian: transitive impersonal
a. Летчик ... сбросил тысячу галлонов топлива, и [самолет сразу
потянуло вверх]. [3, p. 29]
b. Давило грудь, сжимало виски. [4, p. 24]
c. Почему Европу замело, а Россию заморозило?
Ottosson notes [28, p. 148] for the Fate construction that “these verbs are marked for accusative subject only when the agent side argument is natural force”. We can then conclude, preliminarily, that the Fate construction is a kind of Transitive Impersonal, in which the non-Theme argument is necessarily interpreted as a Natural Force causer, thereby activating the accusative probe v-cause.2 It follows that anti-Burzio predicates are necessarily dyadic and causative.
1 Indeed, in some Transitive Impersonals, such as in Polish, exemplified in (i), no expression of the causer whatsoever is licit, though, on the present account, there is no valency reduction - (i) is crucially interpreted as caused, rather than as anticausative:
(i) polish: transitive impersonal
Wyrzuciio lodkq na brzeg (* przez fale / *od fall).
threw-N.sG boat-Acc onto shore by/through waves-Acc from waves-GEN ‘The boat got thrown onto the shore.’ [17]
[I thank B. Cetnarowska, B. Citko, K. Dziwirek, p.c., for additional judgments]
On the theory that (i) involves no valency reduction (the non-Theme causer is precisely what activates v-cause), it remains unclear what prevents its syntactic realization, a problem I leave for future research. The problem is compounded by the fact that “verbs of pain and physical discomfort” in Polish also participate in the Transitive Impersonal, but with the normal expression of the oblique causer, as in (ii):
(ii) polish: transitive impersonal
Zemdliio mnie z nerwow.
nauseated-N. sg me-acc from nerves-GEN
‘I felt sick from nervousness’ [B. Cetnarowska, p.c.]
2 Each functional head in the syntax is a predicate of sorts, identified with a particular argument (cf. [8]. Think of v-cause as an unsaturated predicated until an argument enters the structure that is capable of independently setting the event in motion. Secondary Instrumentals, discussed in
[2], that require human manipulation, like ‘fork’, would thus fail to “saturate” [= license] the
I will argue in the next section that Transitive Impersonals are not formed via a valency reduction operation, but rather via a systematic alternation in the predicate’s argument structure. Transitive Impersonals instantiate the very underspecification in argument structure that is countenanced and predicted by constructionist syntax. That is, while the “affected” argument is fixed as the Theme, the predicate’s second argument, on a split-vP, is freely merged high, as an Agent (in Spec, v-voice and realized as nominative), or low, as a non-volitional causer (and realized by an oblique case, if the causer is known and relevant to the discourse, while activating v-cause).
4. Toward an Analysis
As suggested above, the Russian Transitive Impersonal and Icelandic Fate Accusative are both formed from verbs that otherwise appear as transitive two-place predicates:
(18) Icelandic
a. agentive transitive
Bondinn rak hestana ^ burtu.
farmer.the-NOM drove horses.the-Acc away ‘The farmer drove the horses away.’
b. FATE AccusATivE
Batinn rak a land.
boat.the-Acc drove to land ‘The boat drifted ashore.’
(19) Russian
agentive transitive
a. Безработный австриец отрезал себе ногу.
[Аргументы и факты; 2012, March 27]
transitive impersonal
b. Ему отрезало ногу. (not by a human agent)
The examples in (18-19) show variable behavior with respect to the source of causation. The idea is that the structure in (9) accommodates both volitional and non-volitional causation in the form of two distinct v-heads. The causer argument can either merge “high,” as the Agent argument of v-voice, or low, as a VP-internal non-volitional causer. v-cause enters the structure and licenses accusative, so long as the event is caused.
Babby & Lavine [5] observe that this systematic, variable realization of the causer argument must target verbs that are potentially, but not obligatorily,
cause head. Only causative obliques, expressed overtly or merely understood, saturate the v-cause predicate, licensing its presence in the structure and giving rise to its morphosyntactic reflex of licensing (probing) accusative (cf. also the licensing of Borer’s 2005 Aspect head).
Филологические
науки
Лингвистика
agentive. In ungrammatical (20) the verb ампутировать ‘amputate’ is necessarily agentive - it has no non-volitional usage, in contrast to otrezat' ‘sever’ in (19):
(20) *£Му ампутировало ногу.
[Intended: ‘He got is leg amputated off.’]
It appears plausible, then, to treat the Fate Accusative in Icelandic as a two-place predicate, of the appropriate, potentially-agentive type, whose nonTheme argument is variably merged low as a causer, and thereby interpreted as a Natural Force, rather than Agent. The unspoken Natural Force argument sets the event in motion (and activates the accusative probe in v-cause), now explaining the otherwise mysterious appearance of accusative without further stipulation.
In point of fact, it appears that the Natural Force argument can be stated explicitly, so long as it is not introduced by agentive af ‘by’ (H. Sigurdsson. pc.):
(21) Icelandic: fate accusative
Batinn rak a land vegna mikils vinds / i
boat.the-Acc drove at land because.of heavy winds in vindinum. wind
‘The boat drifted to land due to the heavy winds.’
That the accusative is not quirky (lexically stipulated) in the Fate construction, as suggested initially by Zaenen & Maling [41], is demonstrated by the fact that it does not survive passivization, as quirky case does elsewhere in Icelandic:
(22) Icelandic
a. fate accusative
Strompinn bles af husinu.
chimney.the-Acc blew off house.the ‘The chimney blew off the house.’
b. passive
Strompurinn var blasinn af husinu.
chimney.the-NOM was blown off house.the
‘The chimney was blown off the house.’ [41, p. 145]
Additional evidence that the Fate Accusative is necessarily sensitive to non-volitional causation comes from Experiencer predicates. Experiencer predicates are causative only when they occur with a subject Causer argument [29, p. 55-60]. Otherwise, the non-Experiencer argument is interpreted as a Stimulus or Target of Emotion [29], rather than a Causer. As a result, we predict that accusative should not be possible in the Icelandic Fate construction or the
Russian and Ukrainian Transitive Impersonal, when formed from Experience verbs, since, as impersonals, they lack a Causer subject (Spec, v-voiceP is not filled; and v-cause is not identified). Compare the licit appearance of accusative in (23a), which is causative, with the ungrammatical occurrence of accusative in the Experiencer predicate in (23b):
(23) ICELANDIC
a. FATE ACCUSATIVE
Fiskimennina rak a land vegna mikils vinds.
fishermen.the-Acc drove to land because.of heavy winds
‘The fisherman drifted ashore due to heavy winds.’
b. EXPERIENCER PREDICATE
*Fiskimennina hraddi vegna mikils vinds. fishermen.the-Acc frightened because.of heavy winds
[Intended: ‘The fishermen were frightened because of the heavy
winds.’] [J. Jonsson, p.c.]
The same facts are given for the Russian Transitive Impersonal in (24) and the Ukrainian Transitive Impersonal in (25):
(24) RUSSIAN
a. TRANSITIVE IMPERSONAL
Мальчика ударило током.
b. EXPERIENCER PREDICATE1
*Мальчика напугало игрушкой. [37, p. 426]
(25) UKRAINIAN
a. TRANSITIVE IMPERSONAL
Через загрозу вибуху було затримано поїзд. due.to threat of.bomb was delayed-PASS train-Acc
‘A train was delayed due to a bomb threat.’
b. EXPERIENCER PREDICATE
*Пасажирів було роздратовано новиною. passengers-Acc was annoyed-PASS news-iNS [Intended: ‘Passengers were annoyed at the news.’]
Recall that both the ungrammatical impersonal psych predicates and the grammatical Transitive Impersonals in (23-25) are dyadic unaccusatives, but only in the case of the (a) examples is a causative sub-event asserted.
1 Zimmerling [42] correctly points out that the ungrammaticality of (24b) is due to the fact that игрушкой cannot receive the necessary uncontrolled causer interpretation. As mentioned above, it is interpreted as the “target” or “stimulus” of the emotion denoted by the verb [29]. Zimmerling suggests the improved Experiencer predicate in (i):
(i) ?Мальчика напугало вспышками молнии.
Though still degraded, вспышками молнии 'flashes of lightning' cannot be manipulated by humans and, it follows, is a better candidate for the necessary uncontrolled causer reading.
Филологические
науки
Лингвистика
To review, the analysis thus far treats the Icelandic Fate Accusative as a two-place predicate, the non-Theme argument of which is necessarily interpreted as a Natural Force, which sets the event in motion and introduces into the structure v-cause, which in turn licenses structural accusative case. The Fate construction, it follows, is not an Icelandic-specific quirky phenomenon, but rather a Transitive Impersonal, in the sense described for Russian and Ukrainian. It now follows why accusative is impossible in anticausatives (perhaps universally): these are one-place predicates that assert only the root verbal meaning in relation to its complement; as in the case of Experiencer predicates, no causative sub-event is asserted.
The connection between causation and accusative is further revealed in the Ukrainian examples in (26). Note the difference in object case and interpretation in the Impersonal Passive in (26a) versus its agreeing passive participial counterpart in (26b):
(26) UKRAINIAN
a. IMPERSONAL PASSIVE
Дерево зрубано.
tree-Acc cut.down-PAss .[—agr]
‘The tree was felled.’
b. AGREEING PARTICIPIAL PREDICATE
Дерево зрубане.
tree-NOM.N.SG cut.down-PASS.N.SG
‘The tree is cut down.’
The agreeing passive predicate in (26b) asserts only the state of the tree, not the event that brought about this state. The Impersonal Passive in (26a) asserts the presence of an unidentified argument with cause-to-V semantics, thus accounting for the presence of v-cause and the accusative marking on the Theme. Only (26a) is appropriate when followed by це сталося в понеділок ‘this happened on Monday’, where the pronoun це ‘this’ refers to the event. In (26b) there is no such event for це to identify.1 Likewise, accusative in (27) is degraded because there is no person (voice) or Natural Force (cause) that can concentrate the meaning of a word in its root:
(27) Ukrainian: transitive impersonal
???Основне значення слова зосереджено
basic meaning-Acc of.word concentrated-PAss.[-AGR] в корені. in root
1 Alla Nedashkivska Adams [1] notes that the agreeing passive in contemporary usage is stative in meaning (an adjectival passive), while the Transitive Impersonal form is now standard for the dynamic, verbal passive.
[Intended: ‘The basic meaning of the word is concentrated in its root.’]
[32, p. 142]
Compare (27) with its nominative (agreeing) counterpart in (28). Since subjectless statives lack a sufficiently complex event structure (i.e., they lack the crucial causative sub-event that gives rise to accusative), nominative appears instead. (28) is a monadic unaccusative, like (26b), which pattern like anticausatives and Experiencer predicates with respect to the non-appearance of accusative.
(28) Основне значення слова зосереджене в корені.
basic meaning-NOM.N.sG of-word concentrated-N.sg in root
The independent operation of v-cause makes a further prediction: that accusative may potentially co-occur with passive voice. We now turn to the passive cum accusative construction in Icelandic and Ukrainian.
S. Accusative Survival under Passivization
Accusative survival in agentless predicates has been described in this paper in terms of the arrangement of v-heads in the language; namely, as a function of v-cause operating independently of v-voice (see, also, [21]). Passives are causative (since they contain an implicit Agent), so v-cause functions as an accusative probe (in split-vP languages), giving the impersonal passive in Ukrainian (29):
(29) Ukrainian: impersonal passive
a. Батька буде вбито дитьми.
father-Acc will.be killed-PAss children-iNs
‘Father will be killed by his children.’
b. Табір було зайнято американським військом.
camp-Acc was occupied-PAss American troops-iNs
‘The camp was occupied by American troops.’ [38, p. 117]
Icelandic has a new impersonal construction, passive in form, which has been variously described as passive [13; 16; 35] or as an innovative active [25; 26]. Regardless of the status of the “New Impersonal,” we might conjecture that it arose as a reflex of the syntax described for the Fate construction and Transitive Impersonal (viz, the split-vP). Examples are given in (30) ((30b) is repeated from (6)):
(30) ICELANDIC NEW PASSIVE1
1 The New Passive in Icelandic is an innovative construction, not used by all speakers, and considered “sub-standard” by some. Observe that the New Passive in Icelandic, in addition to appearing with accusative on the Theme argument, does not undergo NP-Movement to subject position, forcing the appearance of an expletive subject to maintain the language’s V-2 syntax. Note further the non-application of the definiteness effect, familiar from the Icelandic canonical
Филологические
науки
Лингвистика
a. Pad var barid litla strakinn. expl was beaten-PAss little boy.the-Acc
‘The little boy was beaten.’ [13, p. 174]
b. Pad var skodad bUinn af bifvelavirkjanum.
expl was inspected-PAss car.the-Acc by car.mechanic.the
‘The car was inspected by the mechanic.’ [16, p. 294]
Let us further consider the role that causation plays in the unexpected appearance of accusative. Recall that in the case of Experiencer predicates (and other statives), the Transitive Impersonal construction was shown to fail in Ukrainian (25b; 27). This was explained in terms of the event structure of these predicates, namely, the lack of a causative sub-event. The same facts hold for Icelandic, as shown in degraded (31-32):
(31) Icelandic: new passive (experiencer predicate)
* Pad var hratt mennina af eldfjallinu.
expl was frightened-pAss men.the-Acc by volcano.the [Intended: ‘The men were frightened by the volcano.’]
[H. Sigurdsson, p.c.]
(32) Icelandic: new passive (stative)
?? Pad var samt alltaf att marga hesta
expl was still always-pAss owned many horses-Acc [Intended: ‘Many horses were still owned.’] [16, p. 302]
While I assume that there is no subject of any sort in these passives, i.e., that they are truly impersonal as spelled out, the question of their subject position remains a point of considerable discussion. Maling [25] assumes a null thematic pro subject, such that expletive Pad must then occur on the left periphery. Zimmerling [42] and Legate [24] both posit zero, non-referential nominal subjects. For Legate (and Maling), this has the effect of maintaining Burzio’s Generalization for these constructions. In the terms developed here this would amount to saying that v-voice performs its argument-projecting function (i.e., by assigning an external argument), so that, in accordance with Burzio’s Generalization, the appearance of accusative follows naturally on the object argument.1 The immediate cost to such an analysis is that we lose the correlation between causation (most importantly, non-volitional causation) and accusative. Recall the distinction in (12), repeated in (33):
(33) Russian: transitive impersonal
a. *Дачу сгорело. b. Дачу сожгло молнией.
passive. In the New Passive, definite DPs can appear licitly post-verbally [16, p. 302]. We continue to focus only on the survival of accusative under passivization.
1 Zimmerling’s [42] zero subject account is posited to explain agreement facts; he explicitly rejects Burzio’s Generalization.
The verbs in (33) differ only with respect to whether or not they contain a causative sub-event, as is the case with сжечь in (33b), accounting for its licit formation of the Transitive Impersonal.1
Note further that if the zero subject bears phi-features (agreement features), as proposed by Zimmerling [42], then the zero element can be maintained for Ukrainian only on the assumption that we introduce a fourth gender in the language, specifically for this purpose. As is well known within Slavic, the n.sg. ending for Ukrainian adjectives and participles in the modern language is /-e/, and not /-o/. It is unclear, then, how a zero subject in the Ukrainian construction agrees with a participial ending, whose dedicated function is nonagreement. One might also wonder whether the zero subject is structurally compatible with the neutral OV word order in the Ukrainian Impersonal Passive; i.e., the preverbal “EPP”-position is already occupied as a result of NP-Movement [23]; likewise is the zero subject compatible with the overt expletive in the New Passive construction in Icelandic, which also occurs in preverbal position?
On a final note, Legate [24] explicitly treats the New Passive in Icelandic as structurally and functionally parallel to the Ukrainian construction. What they share, she posits, in addition to their unexpected transitivity, is a nominal element in Spec, v-voice, “restricting the Agent position.” This nominal element, according to Legate, is smaller than a full DP, preventing its phonetic realization, while at the same time capable of licensing a passive by-phrase. However, the New Passive in Icelandic fails to function like the Ukrainian construction is one important way: the Ukrainian Transitive Impersonal is not necessarily passive; that is, it can occur in the absence of an Agent argument (see fn. 1, p. 50), as in (5), repeated below in (34):
(34) Ukrainian: transitive impersonal
Хату було спалено блискавкою.
house-Acc was burned.down-PAss lightning-iNs ‘The house was burned down by a strike of lightning.’
1 I dispute Zimmerling’s [42] assertion that monadic unaccusatives like замерзнуть and сгореть are stative predicates, and, as such, fail to participate in the Transitive Impersonal construction for this reason. While we have seen that statives do not form Transitive Impersonals, замерзнуть, to take Zimmerling’s example, is not stative in the relevant sense. Consider the case of in x time modification in (i), which holds of accomplishments, but not states [12]:
(i) RUSSIAN: ANTICAUSATIVE
Прошлой зимой река замерзла за три недели.
It is only when we build a causative sub-event above замерзнуть, giving dyadic заморозить, that we can get a Transitive Impersonal, as predicted:
(ii) RUSSIAN: TRANSITIVE IMPERSONAL
Едва лодки спустились до поселка Одиночного, как реку заморозило...
[http://mustagclub.ru; 2011, June 15]
Филологические
науки
Лингвистика
An additional, non-passive, example is given in (35):
(35) Ukrainian: transitive impersonal
Машина злетіла в повітря, [тіла злочинців було
car-NOM flew-up into air bodies-Acc of-criminals was
розірвано]. torn-PAss
(after a terrorist attack.) ‘the car flew upwards into the air and the bodies of the terrorists were torn to pieces.’
[http://ua.euronews.com; 2013, July 10]
The New Passive in Icelandic, on the other hand, is necessarily agentive (i.e., it only appears as a passive - never as an agentless Transitive Impersonal). Note ungrammatical (36), which is the hypothetical New Passive formed from the Fate Accusative in (2):
(36) ICELANDIC: NEW PASSIVE
*Pad var rekid batinn a land.
expl was driven-PAss boat.the-Acc to land [Intended: ‘The boat was driven to shore’ (by a natural force)]
[H. Sigurdsson, p.c.]
It follows that an analysis directed at suppressing the Agent argument, such
as that advanced in Legate [24], while initially plausible for Icelandic, does
not extend to agentless, passive-like impersonals in Ukrainian. Assuming that v-cause potentially occurs independently in both languages, giving accusative in Agent-suppressed (or agentless) constructions, the difference between the Icelandic and Ukrainian constructions appears to reduce rather to the features of v-voice. While in Icelandic, v-voice obligatorily projects an Agent (whose realization is suppressed in some way by passive voice), v-voice in Ukrainian may or may not project an Agent argument, even in the presence of overt passive morphology. In the Ukrainian Transitive Impersonals in (34-35), the passive morphology is merely pleonastic. v-cause is activated by a Natural Force argument - lightning - in (34), and by an unstated force having to do with an explosion in (35). The implied Natural Force in Icelandic (36) is not sufficient to set the event in motion; the New Passive is necessarily agentive.
б. ConcIusion
“Anti-Burzio” predicates are unaccusatives that exhibit the transitivity property (accusative). The appearance of accusative in the absence of an Agent argument occurs in languages with an elaborated “split-vP” dominating lexical VP, such that a cause head can operate in the absence of voice (or regardless of the setting of voice as active or passive). The causative theory of accusative for Russian and Ukrainian extends naturally to the Icelandic Fate Accusative,
now accounting for the obligatory Natural Force reading in terms of features and arrangements of independently motivated v-type heads. Varieties of v distinguish between an Agent - an intentional actor, merged in Spec, v-voice -and an inanimate (natural) force capable of initiating a change of state, thereby identifying v-cause. Accusative survival under passivization follows without stipulation in languages in which v-cause operates independently of the setting of v-voice as active or passive. To be sure, the theory of anti-Burzio predicates proposed here, described in terms of an elaborated vP, is by no means a sufficient condition for passive cum accusative constructions, but it is advanced as a necessary condition, regardless of whether and how the construction is realized on the continuum of passive and passive-like constructions.
References
1. Adams A.N. Transitivity, pragmatics, and discourse in the morphosyntax of contemporary Ukrainian: Ph.D. dis. University of Pittsburgh, 1998.
2. Alexiadou A. et al. The properties of anticausatives crosslinguistically // Phases of interpretation / Ed. M. Frascarelli. P. 187-211. Berlin, 2006.
3. Babby L. A theta-theoretic analysis of adversity impersonal sentences in Russian // Formal Approaches to Slavic Linguistics. 1994. № 2. P. 25-67.
4. Babby L. Prolegomenon to any future typology of impersonal sentences // Hypothesis A / Hypothesis B: Linguistic explorations in honor of David M. Perlmutter / Eds. D. Gerdts, J. Moore, M. Polinsky. Cambridge, MA, 2010. P. 19-40.
5. Babby L., Lavine J. Impersonalization as an alternation. Paper presented at the 35th Annual Conference of the German Linguistic Society. 2013.
6. Borer H. Structuring Sense: The Normal Course of Events. NY, 2005.
7. Bowers J. Transitivity // Linguistic Inquiry. 2002. № 33. P. 183-224.
8. Bowers J.. Arguments as relations. Cambridge, MA, 2010.
9. Burzio L. Italian syntax. Dordrecht, 1986.
10. Chomsky N. Knowledge of language: Its nature, origin, and use. NY, 1986.
11. Chomsky N. The Minimalist Program. Cambridge, MA, 1995.
12. Dowty D. Thematic Proto-Roles and argument selection // Language. 1979. № 67. P. 547-619.
13. Eythorsson T. The New Passive in Icelandic really is a passive // Grammatical change and linguistic theory: The Rosendal papers / Ed. Th. Eythorsson. Amsterdam, 2008. P. 173-219.
14. Folli R., Harley H. Flavors of v // Aspectual inquiries / Eds. P. Kempchinsky, R. Slabakova. Dordrecht, 2005. P. 95-120.
15. Harley H. External arguments and the Mirror Principle: On the distinctness of Voice and v // Lingua. 2013. № 125. P. 34-57.
16. Jonsson J.G. The new impersonal as a true passive // Advances in Comparative Germanic Syntax / Eds. A. Alexiadou et al. Amsterdam, 2009. P. 281-306.
17. Kibort A. On three different types of subjectlessness and how to model them in LFG // Proceedings of the lFg06 / Eds. M. Butt, T.H. King. Stanford, 2006.
18. Kratzer A. Severing the external argument from its verb // Phrase structure and the lexicon / Eds. J. Rooryck and L. Zaring. Dordrecht, 1996. P. 109-137.
Филологические
науки
лингвистика
19. Lavine J. Case and events in Transitive Impersonals // Journal of Slavic Linguistics. 2010. № 18. P. 101-130.
20. Lavine J. What does structural accusative mean? An argument against the Aspectual Theory // Formal Approaches to Slavic Linguistics. 2011. № 19. P. 66-84.
21. Lavine J. Passives and Near-Passives in Balto-Slavic: On the Survival of Accusative // Non-canonical passives / Eds. A. Alexiadou and F. Schafer. Amsterdam, 2013. P. 185-211.
22. Lavine J., Franks S. On Accusative First // Formal Approaches to Slavic Linguistics. 2008. № 16. P. 231-247.
23. Lavine J., Freidin R. The subject of defective Tense in Slavic // Journal of Slavic Linguistics. 2002. № 10. P. 253-289.
24. Legate J. A cline of Passives (ch. 4). Ms., University of Pennsylvania. 2013.
25. Maling J. From passive to active: Syntactic change in progress in Icelandic // Demoting the Agent: Passive, Middle and Other Voice Phenomena / Eds. B. Lyngfelt, T. Solstad. Amsterdam, 2006. P. 197-223.
26. Maling J., Sigurjonsdottir S. The ‘New Impersonal’ construction in Icelandic // Journal of Comparative Germanic Linguistics. 2002. № 5. P. 97-142.
27. Marantz A. Case and licensing // ESCOL’91 / Eds. G. Westphal, B. Ao. H-R. Chae. NY, 1991. P. 234-253.
28. Ottosson K. A feature-based approach to thematic roles // Papers from the Tenth Scandinavian Conference of Linguistics / Ed. V. Rosen. Bergen, 1988. P. 136-150.
29. Pesetsky D. Zero syntax. Cambridge, MA, 1995.
30. Pylkkanen L. Introducing arguments. Cambridge, MA, 2008.
31. Richardson K. Case and aspect in Slavic. Oxford, 2007.
32. Shevelov G. The syntax of modern literary Ukrainian. The Hague, 1963.
33. SigurQsson H. Case: abstract vs. morphological // New perspectives on case theory / Eds. E. Brandner, H. Zinsmeister. Stanford, 2003. P. 223-267.
34. SigurQsson H. The Nom/Acc alternation in Germanic // Comparative Studies in Germanic Syntax / Eds. J.M. Hartmann, L. Molnarfi. Amsterdam, 2006. P. 13-50.
35. SigurQsson H. On the New Passive // Syntax. 2011. № 14. P. 148-178.
36. Svenonius P. The nanosyntax of the Icelandic passive. Paper presented at the Lund Grammar Colloquium. 2005.
37. Tsedryk E. Case and agreement in Russian adversity impersonal constructions // Formal Approaches to Slavic Linguistics. 2004. № 12. P. 419-439.
38. Wieczorek D. Konstrukcii tipa mnoju oderzano lysta v sovremennom ukrainskom jazyke // Z problemow ewolucji wspolczesnych j^zykow slowianskich w aspekcie socjolingwistycznym / Ed. A. Bartoszewicz. Warszawa, 1989. S. 113-118.
39. Wood J. Icelandic Morphosyntax and Argument Structure: Ph.D. dis. New York University, 2012.
40. Woolford E. Burzio’s Generalization and markedness // New perspectives on case theory / Eds. E. Brandner, H. Zinsmeister. Stanford, 2003. P. 301-329.
41. Zaenen A., Maling J. Unaccusative, passive and quirky case // Proceedings of West Coast Conference on Formal Linguistics. 1984. № 3. P. 317-329.
42. Zimmerling A.V. Transitive impersonals in Slavic and Germanic: Zero subjects and Thematic Relations // Computational Linguistics and Intellectual Technologies. 2013. № 12. P. 723-737.
