The study presented here investigates the comprehension of reflexive sentences on the part of Italian-speaking preschoolers. The aim is to verify what we refer to here as the “reflexive passive hypothesis” (Belletti 2020), according to which, at the earliest stages of development, children entertain the hypothesis that Italian can express passive through the reflexive morphology. This is a cross-linguistically attested option which is present in adult Italian in a partly different and more limited way, and which becomes plausible by assuming that children do not treat reflexive si as a reflexive, but as a middle/passive voice morpheme.

The experiment consists of a sentence-picture matching task as 2AFC task that targets comprehension of reflexive structures by means of drawings displaying reflexive-transitive pairs of actions. The interpretation of the reflexive prompt used to describe the pictures is thus assessed in a situation that is ambiguous between a true reflexive and a middle/passive reading of si. In this way, it is possible to verify whether this specific condition hinders application of Principle A of the binding theory (Chomsky 1981).

Our results confirm the nontarget-like performance with the reflexive (a fact also observed in independent works by Smith 2021 for comprehension; Manetti 2013; Belletti & Manetti 2019 for production). Interestingly, all children oscillated between the reflexive and the middle/passive reading of si – a behaviour that can be explained as a result of the competition between Principle A and the “reflexive passive hypothesis”. Hence, as predicted in light of Belletti (2020), Italian 3-4-year-olds consider si to be interpretable both as a true reflexive and as a middle/passive voice morpheme.