The Unaccusative Hypothesis and participial absolutes in Italian. Perlmutter’s generalization revised

The syntax of participial clauses has been an important issue in syntactic theory, especially since Perlmutter (1978) put forth the Unaccusative Hypothesis. This is the object of a lively debate, one of the most controversial aspects being whether unaccusativity effects are better explained in syntactic or semantic terms (cf. e.g. Sorace 2000). Focusing on Italian, this paper argues that participial clauses are subject to both syntactic and semantic constraints and that, contrary to the claims of semanticist approaches to unaccusativity (à la Van Valin 1990), the former take precedence over the latter. Demonstrably, an Aktionsart constraint requiring that the predicate involved be telic operates in a different fashion for the two classes of unaccusatives vs. unergatives. This fact, however, becomes clear only once a commonplace is refuted: Italian unergatives are not altogether ungrammatical in participial clauses, as currently assumed in the wake of Perlmutter (1989). They are so only within a syntactically defined subset thereof, that of noncontrolled participial constructions, while in the complementary subset of controlled participial clauses, occurrence of unergatives is syntactically allowed and semantically constrained. This indicates that the two constructions obey distinct syntactic conditions. Formalization of these, carried out in Relational Grammar, automatically solves a number of problems, within Italian and across Romance. Most notably, it affords a natural solution to the problem of the voice of Romance participial constructions. It also allows a proper understanding of the Romance facts under discussion against the background of a cross-linguistic typology of alignment systems.