The study reports the result of an experiment targeting use and preference for postverbal subjects in Italian by 8 multilingual heritage speakers whose dominant language is Turkish. All participants belong to the old Italian Levantine community of Istanbul, whose members have so far maintained a multilingual linguistic repertoire that include Italian, French and Greek, as well as Turkish.
The theoretical background of the study is the Interface Hypothesis (IH), according to which linguistic phenomena at the interface between syntax and discourse are particularly vulnerable in the acquisition of heritage languages due to a variety of factors that are often difficult to disentangle, including underspecification of interpretable features, processing costs of bilingualism itself, and input effects.
The aim of the study was to assess the role of these factors in the HS’ grammars compared to those of a control group of 10 Italian monolingual. A two-parts timed, computer-assisted test was carried out: the first part consisted in combining given words to formulate an appropriate answer to a given question, in wide or narrow focus conditions (e.g. Perché Giovanni è felice? Ieri/suo fratello/tornare). The second part consisted in choosing the most appropriate answer to a given question between two options, in wide focus conditions (e.g. Chi era prima al telefono? Non lo so: Mario ha risposto/Non lo so: ha risposto Mario).
Collected data illustrate an internal distinction in the experimental group depending on the role of Turkish as language of primary and secondary education. More specifically, subjects of the experimental group who had not attended eight consecutive years of education in Turkish differ from their monolingual counterpart only in the use of postverbal subjects [W(18) = 55.5, Z = -3.535, p = .001], whereas those who had continuously attended Turkish schools from the age of 6 to the age of 14 show vulnerability both in use [Z2 = 57.9, p =.001] and preference [Z2 = 11.27, p = .010] of the same interface.
The proposal is put forward that for the former the source of non-target behaviour is instability in linguistic performance that emerges only in production due to its higher cognitive burden; in the latter, instead, the exposition to a Turkish input in critical years may have affected the linguistic representation, thus emerging both in production and interpretation of postverbal subjects. Effects of cross-linguistic influence, then, appears to emerge only under particular input conditions related to the speaker’s educational development and life choices.