Testing in Two Languages

When we test bilingual children we need to be able to do so in both of their languages. We can to look at speech and language in each of their two languages and we use this information to determine if their language production is like that of their typical (bilingual peers).

In the area of lexical-semantics we know that children who have exposure to two languages often show patterns of lexical knowledge consistent with their divided exposure. They may know home words in the home language and school words in the second language. It makes it difficult to test in only one language, but how do we take account of both their languages?

One of the observations we’ve made in many years of testing bilingual kids is that it is difficult at times for them to switch between languages– especially when they’ve been using English in diagnostics. This doesn’t mean of course that kids don’t codeswitch, they do and they do so during testing, but switching between languages on demand is hard.

In a recent study, we gave children the EOWPVT in both English and Spanish. We gave the tests allowing them to respond in either language but tried to elicit the target in the language of testing– that is, English- or Spanish. If they responded in the nontarget language, we recorded the response, and asked if they knew it in the target language. If they could not recall the word at all we asked if they knew it in the other language. 

I explained before that we scored the two tests 3 different ways each: single language scoring, conceptual scoring within test (this was the score where we counted correct in either language during the test), and conceptual scoring across tests (this is where we compared each item across the two tests and credited correct responses in either one).  One really interesting finding was that the highest scores were obtained when we counted items across the two administrations of the test.

This is important to know and I think has implications for how we test bilinguals. On vocabulary or semantics measures it may be faster to ask kids to switch in the moment– but if they don’t, or can’t it doesn’t mean they don’t know the word in the other language. It may mean that they are so “locked in” to one of their two languages that it’s more difficult to call up the word in the other language. In other words, there is a cost to switching.

In terms of making recommendations for how to test a bilingual child, I think I would block sessions (or parts of sessions) by language. Also, I would probably start with the child’s home language before starting in the school language. We conducted a study several years ago where we found that kids were more likely to switch from Spanish to English during testing but not from English to Spanish (even if they knew the answer in Spanish). Also, I think it’s fine to record spontaneous responses in the other language, but perhaps wait till testing in that language is over before filling in what they know in the other language. Asking kids switch back and forth during testing may inadvertently disadvantage bilinguals leading to underestimation of what they know.





, , , , ,

  1. #1 by Na'ama Yehuda on May 2, 2019 - 7:52 am

    Good post on an important topic.
    As someone who often is called to test children in more than one language (Hebrew/English), I agree with the need to: a. test in an immersive way (I use one language across ALL interaction in one testing session, and another language across ALL interaction in another testing session), b. start with the perceived-stronger language (this may often be the home-language in young children, and sometimes is the school-language in older children, but I rely on intake for that decision), c. not require/elicit responses in both languages at the same time (this is problematic not only on vocabulary/retrieval basis, but also on morphological and sentence structure basis, and only serves to facilitate iatrogenic hybrids IMO, and by that can miss true morpho-syntactic issues in children).
    There are few tests that are standardized and normalized for particular language combinations, and even those don’t necessarily take into account successive bilingualism over concurrent bilingualism and the possible lack of ‘equal exposure’ between languages. So, assessing language in BOTH, and then assessing the ‘cumulative’ language — not only in vocabulary measures but also in sentence-structures, semantics, morphology, syntax, narrative and discourse rules, often gives the best idea of whether language is WNL at least in one of the languages, and what issues are present in either or both.

  2. #2 by Sandra on May 3, 2019 - 6:23 pm

    Thank you so much for providing this great insight!

  3. #3 by Mary Constant on June 1, 2019 - 7:34 am

    That’s great!

  4. #4 by Shocked at Cherry Hill Private Investigator on June 1, 2019 - 7:38 am

    I like your experiments in bilingual. It will benefit people in many ways.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

%d bloggers like this: