'Agent defocusing' revisited
This article presents some descriptive generalisations about the distribution of passive and impersonal constructions in five European languages (Italian, Spanish, Polish, Danish, and Modern Greek). The constructions analysed include (but are not limited to) passive/impersonal constructions in which a reflexive marker is used (Italian, Spanish, and Polish), or in which there is a bound morpheme historically connected to the reflexive pronoun (Danish); so-called impersonal passives, i.e. constructions in which there is an unequivocal formal relationship of the predicate with passive morphology, but either there is no patient, or the patient is marked in the same way in which it is marked in the active sentence (e.g., the -no/-to construction in Polish, the impersonal passive in Danish); man-clauses, i.e. impersonal active clauses having some general noun (“man”, “people”, etc.) as subject (Danish). The corpus consists of Umberto Eco’s novel Il nome della rosa, and its Danish, Modern Greek, Polish, and Spanish translations.Much in the spirit of Myhill (1997), it is argued that the distribution of these constructions in texts is sensitive to the distinction between different types of agent defocusing: there are some discourse conditions under which a given construction is preferred to others, and this can only be captured through a careful inspection of texts. Three basic situation types are identified which are systematically associated with certain modes of expression in each of the five languages of the sample: these are configurations of prototypical semantic features and discourse conditions concerning the agent, the patient (if present), and the nature of the event. The statistical significance of the correlation between situation types and construction types is carefully evaluated, and the results are encouraging: in each of the languages of the sample there is a consistent division of labour among passive and impersonal constructions. This allows us to conclude that the facts surveyed are hardly accidental, and to advance some descriptive generalisations, of both a formal and a functional nature. It is suggested that there is a functional cline of agent defocusing ranging from those cases in which the agent is easily recoverable from the context to those in which it cannot be identified but generally. This cline is correlated with a syntactic gradient of patient and agent coding, although the goal of finding perfect correlations between function and form in the realm of discourse is to be abandoned.