We’ve had a paper out for a couple of years now and I’ve been meaning to blog about it, but for some reason other things have taken priority. The question that we addressed in this study is the extent to which English assessment of children who are Spanish-English bilinguals would be useful for identification of DLD.
Most of U.S. SLPs are monolingual English speakers– and what I mean by most is practically EVERYONE! Something like 96% of speech-language pathologists in the U.S. speak only English. The problem is that 20% of the school population speaks other languages in addition to English. In some places, like California, it’s 40% of kids who are bilinguals. Spanish is one of the most common languages, followed by Mandarin and Vietnamese. Some districts have up to 300 languages represented.
It’s helpful if an SLP is bilingual because they would be able to conduct speech and language assessment in English and another languages. But, most can’t. So, what can we do? We know that tests normed for English monolinguals are probably not appropriate for children who are dominant in another language. They may score low on such a test because they don’t have enough experience with English– or because they have a developmental language disorder. But, we wouldn’t be able to tell. This is why we want to test kids in both their languages. But, what if it was possible to test kids in English and get reliable results? This is what we set out to do in a study published in LSHSS.
We had Spanish-English speakers with and without DLD between the ages of 7 and 9;11. We identified DLD on the basis of testing in both languages, specifically the semantics subtest of the BESA-ME, narrative language samples, parent and teacher report. We looked at items from the BESA-ME (which is our experimental upward extension of the BESA) English morphosyntax subtest. We further divided children by language exposure. The three groups were English dominant bilinguals (using and hearing English 60% to 100% of the time), balanced bilinguals (using and hearing Spanish and English 40%-60% of the time), and Spanish dominant bilinguals (using the hearing Spanish 60% to 100% of the time). We looked at the performance of children with and without DLD in each of the three language groups.
Kids with the highest levels of English exposure performed the best (no kidding), followed by balanced bilinguals and finally Spanish dominant kids. More importantly, we found that there was separation of kids with and without DLD that was related to item type. If you look at the figure, you can see that for example, the difference for DLD and typical Englislh dominant kids on the copula was small, but within balanced and Spanish dominant kids the difference was greater. Irregular past forms separated English dominant kids, but the difference for balanced and Spanish dominant kids was much less.
When we did discriminant analysis based on the items that best separated kids with and without DLD across the different exposure levels. We found that this composite of items has 79% sensitivity and 88% specificity. The same cut off however classified Spanish dominant kids with 90% sensitivity and 71% specificity. Our target is 80% or higher sensitivity and specificity– so this result is not very good. But, when we used a different cut for the Spanish speakers, the sensitivity was 84% and specificity was 80%. These results are promising and shows that it might be possible to get good information when testing kids in English. Hwever, we have to make sure to pick the right items and to identify the correct cut offs. Those derived for monolinguals will likely not work well. And don’t forget, testing in both languages will get you the most accurate information.
We are continuing to work on this issue and we think that we can do better than this– for now I think the results displayed in the figure can help us decide what kinds of items might work for kids who are still in the process of learning English. If you only speak English, you can get reliable information in English, but you need to pay attention to the specific forms that might work. To get the most accurate diagnostic you should combine this information with assessment in the first language.