The Economist recently published an article on infant bilingualism, which challenged one of the great assumptions of the multilingual parenting community: that of "one parent, one language" as the best means of language transfer.
The basic idea is supposed to be that the two parents speak different languages to the child, and the child picks them both up. Many extend this into a three-language pattern by having a third language that the parents speak to each other, so that there's "Mum's language", "Dad's language" and "family language".
The article cites one François Grosjean, a professor emeritus from the field of psycholinguistics and speech processing. The way the article puts it, one parent one language is not "a must", and that "the child must experience regular monolingual situations in each language". In Grosjean's terms, there must be a "need factor" -- the child must feel that a particular language is necessary in certain given context.
The argument is not very well developed in the article, or even on Grosjean's own website. He doesn't really talk directly about the weakness in the one-parent-one-language model, but from the subtext, I would interpret it as follows:
A child in the one-parent-one-language household may well twig that both parents understand one or both of the languages, and therefore could end up picking only one language to speak in.
Grosjean proposes instead a model of home language and outside language (ie the local language used when out shopping etc), but I don't see how this addresses the problem of the child twigging that one language is enough, which in this case would be the outside language (because everyone they know speaks it, whereas in their minds, only two or three people speak the home language).
The other problem that Grosjean fails to address is a matter not directly pertaining to language, but a pragmatic one of education. He talks about the erroneous blaming of language problems on bilingualism, but he has forgotten to the lesson of the Germans in the US.
Home language and outside language was the strategy by default, because schools used English almost exclusively. The German-speakers lived in a state of diglossia where they could talk about everyday matters in what became called "kitchen German", but discussing more serious matters required English.
This was already a dangerous situation for the German of the immigrants, as not being able to speak about important matters trivialises a language in the eyes of its speakers, but the language survived... for a while.
The final nail in the coffin was schooling. With initial literacy being taught exclusively in English, all spelling lessons assumed that the child would know certain a certain base of common household vocabulary, and these words were used in teaching. But if you only speak German at home, your child won't know the word "faucet" (en_gb: "tap"). The establishment of the time misdiagnosed the problem as one of bilingualism interfering with language development, instead of simply identifying a different set of common words to use in initial teaching. Parents were talked into dropping the German at home for the sake of their children's development and the transmission of German within the US stopped.
This problem still hangs over the home-language-and-outside-language family. Is your child going to have problems at school when the teacher holds up a picture of an apple, and the only word your child knows for it is pomme or manzana? How do you prevent your child's confidence from being damaged by sudden exposure to a raft of unknown words?