Despite the fact that bilingualism is extremely common around the world, few CALL (Computer Assisted Language Learning) systems have been designed that specifically address the needs of children who are acquiring several languages from birth or at an early age (simultaneous bilingualism).

Recent research has shown that bilingual children may have a cognitive advantage over monolinguals. However, bilingualism acquisition and development is a challenging adventure, and few individuals become truly balanced bilinguals. Often, the imbalance is linked to unequal amounts and/or quality of inputs that a child receives in each language. When each language is only used for some specific purposes, a specialization operates: each language becomes specialized in the domain in which it is mostly used (for example at home). Finally, when formal education is exclusively provided in one language, literacy in the other language is not achieved.

The purpose of our research is to uncover new interaction paradigms to be deployed in CALL systems to support the acquisition of balanced bilingualism by young children and facilitate biliteracy and biculturalism.

In our presentation, we will discuss three different approaches. First, we propose to apply multimodal interaction techniques, for example speech analysis and prosody, to emphasise language specific cues in order to help a child differentiate between several language systems. Secondly, we discuss the possible advantages to re-enacting multilingual learning contexts and environments in educational software. In particular, we will illustrate the approach showing examples from a prototype that implements the "OCOL" (One Character One Language) model, a simulation of the popular OPOL (One Parent One Language) bilingual parenting strategy. Finally, we propose to borrow techniques from Artificial Intelligence to assess language proficiency and support biliteracy.