Several studies on heritage language (HL) acquisition investigate a single linguistic structure, showing how language exposure or cross-linguistic effects affect its acquisition. Here, we consider HL speaking children's mastery of several linguistic structures using a cloze-test. We examine how their language competence is affected by language exposure variables and age. We tested 180 children between the ages of 8 and 16, living in Switzerland and speaking European Portuguese as HL and French, German or Italian as their societal language. The items of the cloze-test cluster around two levels of difficulty, with the items at the second level corresponding to structures that are acquired late in Portuguese monolingual acquisition. Older age and a greater amount of formal instruction in the HL lead to better performance. The role of the amount of formal instruction varies based on the level of difficulty of the target structures. Cross-linguistic influence does not affect the results.

The present study aims to understand which factors contribute to different patterns of use of referring expressions by bilingual children, by considering the triangulation between language experience and proficiency, executive functions and cross-linguistic effects. We analyze reference use in Greek in the context of a narrative elicitation task as performed by 125 children of different language combinations, including Greek–Albanian, Greek–English and Greek–German. We calculate, for each child, an index of language experience that combines a proficiency measure with background questionnaire information. After identifying the occurrences of underinformative (underspecified) and overinformative (overspecified) referring expressions in the production of each child, we investigate to what extent each pattern of reference use is affected by language experience, cross-linguistic effects and executive functions. The study aims to shed some new light on the nature of overspecification and underspecification in bilingual reference production and, more in general, to model variation in reference use among bilingual children.