Across both these posts, presentations, chapters and journal articles, I often say that we need to test children in both of their languages. I think that many of us know that. The question however is what do you do with that information once you’ve obtained it.When I first started learning about bilingualism and speech-language assessment, I was taught to first determine the child’s first and stronger language and to test in that language. It was assumed that everything that children knew in the stronger language was also known in the other language. But, we now know that’s not true. Pearson is one of the first people who described bilingual children’s divided vocabulary knowledge. Colleagues and I have replicated and extended these findings using a category generation task. It makes sense that children would have divided knowledge. Children learn words to meet linguistic demands. Bilingual children do the same. But, the demands that they need to meet are divided by language. For example they need to learn words in the home language to interact at home and words in English to interact at school. While some of the words may cover the same concept others may entail different content because of the different activities.
We have also learned that dominance is not static. Luk and Bialystok have a paper whose title I like, “bilingualism is not a categorical variable.” In our own work, we find that children will look “dominant” in one language for one kind of language task, but “dominant” in another language for another kind of task. This means that the way I was trained anyway, to identify the dominant language and test in that one may not best capture everything a child knows. Discrepancies across domains may be due to children’s accelerated learning in that language for one or another domain. This kind of variability means that test validity is questionable when children are tested in only one language (and compared to monolingual norms).
My friend and colleague Sam Ortiz was in town this week. Yesterday, he presented here in Austin at Region 13, on the topic of “Assessing Second Language Learners Using Cross Battery Assessment.” Over dinner we discussed issues and challenges in assessment of ELLs in both speech-language pathology and school psychology. And there are a lot of similarities. School psychologists too, have been taught to determine the dominant language and test in that. The assumption is that whatever you get is valid. But, we don’t know that that is really the case–bilinguals know TWO languages, and they might have learned some things in the less dominant language (which is ignored if children are tested in only one language). In addition, even if one gives a “non-verbal” test, it is not completely free of culture, nor is it completely free of linguistic content.
In development of the BESA, (which is now available for purchase) we tried to take these issues of dominance into account. Our first idea had been to find items that worked similarly across Spanish and English. But, that didn’t work– we would have ended up with too few items. We ended up selecting items that demonstrated differences between children with and without language impairment across different levels of language use and exposure.
Another thing we did is to compare performance by level of use/exposure to determine if children should be tested in one or both of their languages. We found that over 30% exposure to one language (and below 70% to the other) children are more likely to present “mixed” profiles of dominance, so we recommend testing in both. The BESA test kit will include the language use/exposure questionnaire to help make this determination in practice.
When kids are tested in both, we do a couple more things. For semantics, we accept responses in either language– we (as clinicians) stay in one language (so, no translating or codeswitching during the test) but a correct response is counted as correct, regardless of language. The other thing we do is to take the higher score. So, in phonology is Spanish is higher, we take that score. For language assessment we combine the higher semantics score with the higher morphosyntax score for a language composite. The composite may be Spanish + English; Spanish + Spanish; English + English; or English + Spanish. This way we’re giving kids every chance to demonstrate what they know. And you know what? These composites and “best language scores” provide the absolute best classification compared to single language scores.
There may be other ways to combine the two languages that capture what a child can do. We’re exploring those, but so far this works pretty well!