Does bilingualism hurt children with DLD?

You know I’m gonna say no. But, it’s important to establish what does happen and to do so with data. After several studies we have enough data to look at this question more carefully with a set of children with developmental language disorder (aka: language impairment; specific language impairment; or primary language impairment) who had varying levels of exposure to Spanish and English.It’s important to look at the question controlling for level of exposure because this could skew results. Often, studies of bilingual DLD are relatively small because it’s difficult to find kids with DLD who are bilingual. DLD occurs in about 7% of the population; bilingualism in the U.S. is between 20% and 40% of the child population (depending on where you are); Spanish speakers are about 80% of the 20-40%, so bilingual DLD is about 1.4-2.8% of the total child population, so it takes time.

Why do we have to control of level of exposure? 

Well, the classification isn’t really bilingual vs. monolingual, although many studies of bilingual DLD do divide children up that way from necessity. We’ve divided into dominant Spanish, balanced, and dominant English knowing of course that among people exposure is continuous– where do you draw the dominant Spanish vs. balanced line? To a certain extent it’s arbitrary. We’ve also divided into 5 groups (functional monolingual English, dominant English, balanced, dominant Spanish, and functional monolingual Spanish). With typical kids, it’s easier to see the continuous nature of exposure and we’ve done this with our screening studies using regression and using GAM modeling.

It’s harder to use a continuous approach with children who have DLD because there aren’t as many of them. But, it’s important to do so. You want to be able to compare children with similar levels of exposure to really understand the effects of DLD independent of the effects of bilingualism. Otherwise we could be comparing balanced bilingual children who have 50/50 levels of exposure to each language with perhaps children with are bilingual but have 70/30 levels of exposure to L1 and L2. This is why a continuous approach is better. We can look at whether children’s performance is parallel across levels of bilingualism, or if they are non-parallel.

In this first example, we can see that the two lines representing ability groups are parallel. There are differences between the two groups based on ability, and when tested in the L2 those who have more exposure to the L2 score higher regardless of language ability status. This is a situation where exposure does not additionally affect performance.

In this next example, you can see that there are differences between children with and without language impairment, but the two lines are not parallel. The line for language impairment is steeper. The difference between the two groups at points A and B are different. With B being smaller.  This kind of pattern would show that language impairment would be more affected by bilingualism.

A third possible pattern is one where bilingualism advantages children with language impairment. The line of the children with DLD (or language impairment) is flatter compared to typical children. Here the differences between more balanced bilinguals (A) with and without language impairment is smaller than that for those who are more monolingual (B), This pattern shows that language impairment is not affected by bilingualism, and shows a relative advantage.

In our study, we looked at the performance of 600 bilingual children with and without language impairment on two kinds of measures (semantics and morphosyntax) in two languages (Spanish & English). For the younger kids, we used the BESA, and for the older kids with used the BESA-ME. We compared them on their standard subtest scores. There were 400 children with typical development and 100 children with DLD. What we found were patterns that corresponded to the first and third options described above. Either the lines were parallel indicating that there was no additional effect of bilingualism on impairment, or there were smaller differences for the kids in the middle (balanced bilinguals) and typical kids at the same level of exposure compared to those on the monolingual end in the language they were tested in. 

This finding adds one more piece to the puzzle regarding bilingual DLD. Here (again) we see that bilingual children with DLD are not disadvantaged relative to typical children once we control for level of bilingual exposure. And in some instances, they may be advantaged over their more monolingual peers with DLD. So, moving them toward monolingualism may not help (and could hurt).

 

Advertisement