"Susana keeps mixing her two languages; she's semilingual!"
"Pierre doesn't know either language well; he combines them all the time!"
I cringe, as do other language specialists, when I hear comments such as these concerning children who know and use two or more languages. In fact, there are a number of factors that can account for the active presence of the other language when children are speaking a particular language.
First, children may simply be in the process of becoming bilingual. When that is the case, their first language may well intrude upon their new language in the form of interferences, that is deviations from the language being spoken due to the influence of the other language (see here). In addition, even when they are speaking to monolinguals, they may well call on their other language through code-switching, that is they will bring in a word, a phrase or a sentence from the language not being used (see here). After all, their interlocutors might just understand parts of what they say in the other language and since communication is crucial, why not try?
A second factor is that, even when the acquisition of the other language has stabilized, children are often dominant in one of their languages. This is true of very young bilinguals as well as of older ones. For instance, it is relatively rare that children acquiring two languages simultaneously receive as much input in each of their two languages. One language is usually the stronger one and it has a tendency to influence the weaker language. With time, stop-gap elements such as code-switches will diminish in number as the non-dominant language is heard more, is used increasingly and is better known.
A third factor concerns the language mode - monolingual or bilingual - children are in when they are communicating (see here). It is still unclear when very young children who are acquiring their two languages simultaneously start controlling language choice, i.e. which language to use with whom and for what, as well as code-switching. It happens quite early on, and they become adept at it, but there seems to be a short period of adjustment for the appropriate mechanisms to be in place. Slippage can take place during this period, hence the intermingling of languages.
For slightly older children, the crucial question will be whether they are with interlocutors (parents, siblings, friends) who understand their languages, even if incompletely. Children are quite pragmatic: if they know the person with whom they should be using a particular language has some knowledge of the other language, then they may well bring it in if they need it. Getting the message across is their foremost concern. And more generally, if children are raised in bilingual families where code-switching is frequent, then they too will intermingle their languages.
As children come into increasing contact with speakers of just one language, the amount of intermingling will be reduced. In addition, caretakers can influence the course of things by putting their children in monolingual environments, whenever possible, where people know and use just one language. This will allow the children to receive language input that does not contain elements of the other language, and it will increase their knowledge of the language.
With time, bilingual children will learn how to adapt their language mode to the situation they are in and the person they are speaking to. They will become proficient at navigating between a monolingual mode, where just one language is used, and a bilingual mode, where the intermingling of languages is possible .... and is accepted.
Grosjean, F. Linguistic aspects of childhood bilingualism. Chapter 16 of Grosjean, F. (2010). Bilingual: Life and Reality. Cambridge, Mass: Harvard University Press.
"Life as a bilingual" posts by content area: http://www.francoisgrosjean.ch/blog_en.html
François Grosjean's website: www.francoisgrosjean.ch