I’ve been saying this for years. My colleague Mary Anne Nericcio says she’s been saying this for 30 years– I guess I’ve been saying it for about that long too! As part of our Diagnostic Markers of Language Impairment in Bilingual Children project, funded by the NIH (NIDCD) we screened some 1200 children who spanned the range from monolingual Spanish speakers to monolingual English speakers and looked to see whether children in the middle (bilinguals) were more likely to fall in the risk range more often than monolinguals. They don’t.
The findings of our paper appear in the American Journal of Speech-Language Pathology if you want to read the nitty-gritty, but I’ll give you the highlights here. What we did is screen a bunch of kids in Utah and Texas over a 3-year period. Children were ALL screened in Spanish and English. I’ve mentioned this procedure before– we wanted to know how much exposure was associated with their actual performance (we’re working on a paper on this currently, and I’ll tell you about it once it’s done). And a problem sometimes in bilingual research (and testing bilingual children) is that we can make mistakes about what a child’s better language is, and kids often have distributed knowledge so it’s best to test in both. As part of the screening procedure we interviewed parents asking them what language their child uses and hears each hour of the day to estimate how much they are exposed to each language per week. Right now I think we’re going to call this questionnaire the Language Interview Parent Survey (LIPS)– too cute??
Our results showed that on average, bilinguals scored lower than monolinguals in both languages. That is, bilinguals scored lower in Spanish than Spanish functional monolinguals and lower in English than English functional monolinguals. We call these kids functional monolinguals, because even if they have as much as 20% exposure to the language they really can’t communicate in it, tell a story, or respond in that language.
Anyway, you might be asking how the bilinguals can score lower in two languages and STILL not have an increased risk for language impairment. Good question. Although it seems a little counter-intuitive, this is indeed the case, they scored lower, but they didn’t fall below the cutpoint any more often than the monolinguals. How was this possible? We looked at the patterns of failure. Our cuts were set at the 25th percentile. And we considered children to “fail” the screener (which, clinically would indicate need for additional testing, because screeners usually are too short to be diagnostically accurate) if they fell below the 25th percentile on at least 3 of the 4 subtests. Predictably, the monolinguals failed in the language they didn’t know– but they scored the highest in the language they did know. To fail, they needed to score low in one of the subtests in the language they did know. Bilinguals were all over the place in terms of their patterns. When they failed, it could be in Spanish OR in English. When they passed, they might have a low score in one or two subtests (in Spanish or English or both) and score high on two or more subtests (again, Spanish, English or both). I think that because of this pattern, the averages for the bilinguals were lower because their lowest scores could be in either language, not just in one. So, they might do well in Spanish morphosyntax and English semantics, or English morphosyntax and Spanish semantics, or Spanish morphosyntax and semantics, or English morphosyntax and semantics. Whew!
Clinically, I think this means that unless children are functionally monolingual, you really need to test in both languages. You also can’t expect that their strongest language is consistent across domains. In fact, “mixed dominance” seems to be more the norm than not.