[SynSem] Mauro Viganò (SFL)

retour Séminaire Syntaxe et Sémantique

Mauro Viganò (SFL)

Italian reduced structures: contrasting structural and elliptical accounts with an experimental investigation

Italian reduced structures, as described by Cecchetto and Donati (2022), are colloquial sentences that have full illocutionary force despite being clearly “reduced” (they display no external argument and no finite verb). Two examples of these structures are the Bare Noun Reduced sentences (e.g., “Nonna guarita” = grandma healed) and the Dislocated Reduced sentences (e.g., “Guarita, la nonna” = the grandma, healed). Analysing these structures, two possible accounts can be proposed. According to C&D account, they are generated as small as VPs and they are interpreted as sentences by the labelling by feature sharing mechanism postulated by Chomsky (2019) for full clauses (thanks to the agreement between the nominal part and the verbal participle). On the contrary these structures could be derived through phonological deletions from fully-fledged sentences (e.g., elision of auxiliaries, articles, etc.). C&D’s analysis makes predictions that have not been experimentally tested: for example, if only a VP is projected, radically reduced structures with unergative and transitive verbs should not be possible, as they would require the activation of the vP layer, which by hypothesis is absent. The same reasoning applies to all the functional projections in the middle field. For example, sentential negation, if it requires a NegP layer, should induce ungrammaticality too.
The talk will present an experimental study that was performed to test these predictions considering the effect of three structural factors on the acceptability judgement of these sentences: structure type (Bare Noun Reduced vs. Dislocated Reduced), predicate valency (intransitives vs. transitives), and structural layer (VP vs. NegP vs. vP).
For all the factors, the analysis identified main effects and significant interactions. The post-hoc analyses showed significantly higher acceptability ratings for all the conditions characterized by the VP layer (i.e., affirmative passive and unaccusative predicates) in comparison with NegP and vP layers conditions, regardless of the structure and the valency factors. This main finding supports the validity of the C&D account; however, some results were not predicted. The NegP layer conditions (i.e., negative passive and unaccusative predicates) reported significantly higher acceptability ratings in comparison with most vP layer conditions (i.e., affirmative transitive active and unergative predicates). Moreover, Bare Noun Reduced Structure with transitive predicates (e.g., “mangiato la pizza” = eaten-MASC the pizza-FEM) showed significant better ratings in comparison to all the NegP layer conditions. The modulation in the degradation of acceptability in presence of negation and in this particular case of auxiliary omission do not apparently match the C&D analysis.

During the talk, possible explanations for these unexpected findings will be explored.