A syntactic approach toward the interpretation of some distributional frequencies: comparing relative clauses in Italian corpora and in elicited production

RGG (2014) 36: 1-28 A SYNTACTIC APPROACH TOWARD THE INTERPRETATION OF SOME DISTRIBUTIONAL FREQUENCIES: COMPARI NG RELATIVE CLAUSES IN ITALIAN CORPORA AND IN ELICITED PRODUCTION Adriana Belletti Università degli Studi di Siena Cristiano Chesi Istituto Universitario di Studi Superiori di Pavia A robust set of experimental results from previous studies on both production and comprehension of subject relatives (SRs) and object relatives (ORs) in Italian have confirmed the well known different status of SRs and ORs holding cross-linguistically in both children and adults, with ORs harder than SRs, in various dimensions. One crucial finding of these results concerns Italian-speaking adults who, in elicited production tasks, tend not to produce ORs in a systematic way and resort to the production of alternative structures: the privileged alternative is represented by use of passive in the relative, leading to the production of up to 90% of passive object relatives (PORs) in the studies considered.The contribution of this article is primarily comparative in nature, bringing into the picture a new dimension: a corpus study of (headed) SRs and ORs in standard Italian to be compared with the results from elicited production. The animacy feature is also manipulated in a new elicitation experiment adapting previous designs. Results indicate that, on the one hand simple frequency based considerations cannot be at the source of the ample resort to PORs in the elicited productions, as PORs are rather infrequent in the corpora of spontaneous production investigated; on the other hand, ORs with an inanimate relative head are relatively frequent in the same corpora, yet manipulation of the animacy feature does not play a role in favoring the elicited production of ORs headed by an inanimate noun phrase. We propose that the grammatical dimension in terms of intervention locality may offer a crucial key in interpreting the complex shape of the results and highlight that simple distributional frequency factors remain unreliable as for the expectations they can generate.