Construction of Meaning Workshop

UNACCUSATIVE MISMATCHES IN ITALO-ROMANCE

Delia Bentley
University of Manchester

Friday, January 20, 3:30 PM MJH Rm 126

Sponsored by the Stanford Humanities Center/Mellon Foundation
Graduate Research Program

Starting from the assumption that, in Italo-Romance, split intransitivity manifests the tension between a type of argument marking which is principled in syntactic terms (accusative alignment), and one which is principled in semantic terms (active alignment), I consider three kinds of unaccusativity mismatches. These mismatches concern:

* The domains of different diagnostics within a given language (e.g., in Italian, the perfective auxiliary `essere' (be), the marking with `si' and past-participle agreement mark all reflexives as unaccusatives, but only some classes of reflexives license `ne'-cliticization of their argument).

* The domains of a given diagnostic across different languages (e.g., the past participle agrees with the subject of unaccusatives and passives in Italian but only with the subject of passives in Sicilian).

* The domains of a given diagnostic within a given language (e.g. whilst, in Italian, some classes of verbs categorically require perfective `essere' (be), others do not).

Cross-linguistic analysis of split intransitivity suggests that unaccusativity is determined by semantic factors, and that the semantic parameters of split intransitivity vary across languages in interesting ways (Van Valin 1990). The variation of the semantic parameters of split intransitivity is certainly significant vis-a-vis the unaccusativity mismatches which are exhibited by Italo-Romance, but it is not the only factor. Rather, an analysis of this group of closely related languages suggests that the unaccusativity mismatches arise primarily from the conflict of two driving forces; one is semantic, and is directly related to the semantic foundation of unaccusativity, whilst the other is syntactic, and strives to obliterate the manifestations of the semantic principle. Individual Italo-Romance languages manifest different synchronic results of the historical tension between these two forces.

In this paper I argue that (i) whilst the unaccusativity mismatches do not constitute evidence against the hypothesis that unaccusativity is syntactically encoded (Levin and Rappaport Hovav 1995), they challenge the analyses which deterministically associate each manifestation of unergativity (e.g., the perfective auxiliary `avere' (have)) with unergative syntax across languages; (ii) the semantic parameters which are relevant to split intransitivity combine with syntactic, and, in some cases, pragmatic constraints to characterize each diagnostic, and this explains why the domain of a given diagnostic may differ from that of other diagnostics within and across languages.