According to Knjazev, in (8a) the construction okna byli zakryty «the windows were shut» is an objective resultative, whereas in (8b) the construction okno bylo zakryto «the window was shut» is an actional passive. In other words, the construction in (8a) has the meaning of a state and implies: a) on the semantic level, one participant only about which a contingent property is predicated through the past passive participle; b) on the syntactic level, the structure is of the predicative type where the predicate, even though it has the form of a past passive participle, has the status of an adjectival determiner. On the other hand, the construction in (8b) has the meaning of a resultative state having, on the semantic level, three characteristics [DesclSs & GuentcMva 1993: 91]: 1. it implies a preceding event and the existence of an agent (specified or not); 2. it determines a property of the patient; 3. the property is not necessarily contingent upon the implied preceding event. On the syntactic level, the auxiliary operates on the past passive participle, associated with an abstract passive predicate which includes the notion of an unspecified agent, in view of its transformation into a verbal unit [ibid].

The resultative state, which I have just defined, must not be confused with the resultant state. Indeed, in languages such as Bulgarian, where there is an overt expression of the perfect (9b) and of a periphrastic passive (9a), these two notions are clearly distinguished:

Although these two forms may appear in the same context, each has its own meaning: (9a) denotes a resultative state as defined above; it permits therefore to draw attention to the patient and to the characteristic which is attributed to it by the passive predicate; if the verbal form is a reminder of the implied event, it is in order to signify that at its origin is an agent. On the contrary, (9b) is an overt expression of the perfect; it denotes a resultant state: that is, a state which is brought about by an event and which is contiguous to this event.

It has often been pointed out that the periphrastic passive tends to be constructed with a perfective past participle, whereas the reflexive passive tends to use the imperfective form. On this subject Siewierska [1988: 247] notes that in Slavic languages, with the exception of Polish, the periphrastic passive including an imperfective participle rarely appears, and quotes Czech and Serbo-Croatian where the constructions are said to be used mainly in scientific texts. This affirmation is not wholly justified. In Russian, the contrast is not any clearer: the constructions with an imperfective past passive participle are sporadically attested [Maslov 1988; Poupynin 1996: 131] and are subject to strong lexical, syntactic and contextual constraints; they are allowed in varying degrees in Czech, Serbo-Croatian, Polish, and above all Bulgarian.

The «be»-perfective passive is claimed to convey both the meaning of a state resulting from a previous action and that of an action. The main argument for justifying such an analysis comes from a distributional property of the form, founded upon the compatibility of the perfective form with localisation markers. According to Khra-kovskij [1991; 151–154] and Knjazev [1988:350–351], the actional passive is compatible with time adverbials and adverbial phrases which are precise indications of the temporal interval relative to the preceding event and place markers ((10), (11)) whereas the resultative state is compatible with those which mark duration (12) or iterativity.

It is well known that Polish is the only Slavic language which has two auxiliaries to form the periphrastic passive voice: the zosta'c passive which imposes a perfective participle and whose meaning is characterised as «actional», and the «be»-passive, which allows both the perfective participle and the imperfective and which, depending on the context, may take on either a so-called stative meaning or a so-called actional meaning:

Siewierska [1984: 129, 1988: 251] notes that recourse to (13c) implies that the room needs repainting, whereas (13a) does not. If one admits that the passive resultative state determines a property of the patient while at the same time implying a preceding event, and if one takes into account the meaning of zosta'c «to become», one easily understands that such inferences may be made. Zosta'c directly links the resultative state to the event serving to highlight the transitional character of the event and to state that the patient's property has been acquired prior to the act of speaking; the adverbial expression is therefore taken as included in the transitional event, which would explain why the periphrastic construction with zosta'c is incompatible with dva razy «two times», mnogo raz «many times», etc. On the other hand, the «be»-passive with a perfective verb gives priority to the resultative state by only referring to the implied event, leading to ambiguity as to whether the resultative state belongs to the patient (meaning a resultative state) or whether the resultative state only serves to hark back to the event at the origin of the patient's affectation. The adverbial phrase in (13a) is presented as incident to the event, which allows one to understand why iteration is permitted.