Passive with and without Passive Morphology: A Minimalist Approach to Russian and Hungarian

Edit Jakab, Princeton University

The purpose of this paper is to examine how Russian and Hungarian express the notion of 'passivity' in light of the fact that Russian possesses passive morphology while Hungarian does not, and to give a Minimalist account of the different strategies of the passive.

In the function of the passive (PASS) Hungarian has the option to use verbal adverbs or adverbial participles (GER) as in (1), while Russian utilizes its passive morphology (2):

(1) az ablak be van csuk-va
the window PREF is close-GER [be = prefix to the verbal adverb]
'the window is closed'

(2) okno zakry-t-o
window close-PASS-N
'the window is closed'

Contra de Groot I will argue that (1) indeed stands for passive in Hungarian since it exhibits the semantics and the single universal property of the passive, i.e., the suppresion of the Agent argument (or the implicitization of the external theta role in Babby's terms). Among de Groot's reasons for (1) not being passive are that (i) it can be derived from one-place (monadic or intransitive) predicates, and (ii) it does not allow the overt expression of the Agent, i.e., a "by-phrase". (i) Even though "canonical" passive can be formed from transitive verbs only, it is not a universal constraint on passivity: passive from intransitive verbs can be found cross-linguistically, e.g., in German (3):

(3) es wurde getanzt
it was danced
'they danced'

(ii) The implicitized Agent argument does not always have to have an overt manifestation cross-linguistically; there are several languages that do not allow it at all -- this is actually the raison d'etre of the passive, i.e., that the logical subject can remain unexpressed or rather covert. For example in Finnish the implicitized subject never licenses a by-phrase:

(4) kirja luetaan (*lapselta)
book-nom read-PASS (*child-abl)
'the book is read (*by the child)'

In this paper I will propose, following Boeckx 1997, that it is the aspectual richness of the passive morphology (-en or in the same context the Hungarian -va/-ve suffix of the verbal adverb) that permits the logical subject to remain covert (or suppressed/unexpressed). This type of passive construction is derived solely from perfective verbs in both Russian and Hungarian. Hence de Groot's observation that constructions such as (1) are "aspectual" (stative or resultative) exactly reinforces and does not undermine the argumentation that (1) is passive. De Groot's other argument against (1) being passive, that it requires copula support and is therefore a non-verbal predicate, can be also dismissed since auxiliaries are very common in passive constructions; this is the case in many Slavic languages where the participle adjoins to the auxiliary be.

I will argue that the logical subject which is licensed by the rich aspectual features of the passive morpheme is pro (a specific null item with arbitrary [3rd person only] interpretation). Since this licensing is very local, pro must be located in AspP (Aspect Phrase) to be as close to the passive morpheme as possible; for this reason AspP must be contained in vP. Given that accusative Case features are normally checked in Spec-vP, which equals AspP, and that pro is associated with Asp, pro will prevent the accusative Case features from being checked in Spec-vP, and will force the object NP to move to T to be licensed. Also since pro contains all the necessary information being licensed by rich morphology, it is less specified than the object (it lacks phi-features), hence the object will be attracted to T on the basis of the principle of Attraction according to which elements with more specified features have priority. This explains the canonical nominative Case of the underlying object.

In my talk I will support this argument with additional evidence from Russian and Hungarian, and I will explore in detail why morphology other than passive is able to express passivity. The possibility or impossibility of a by-phrase will also be further investigated.