We have a fairly new article accepted for publication in the International Journal of Bilingualism and Bilingual Education. Even though it’s not yet published it’s available through the Journal’s forthcoming articles list.
As part of an NIH funded project, we screened about 750 children (actually we now have screened 1200 kiddos, but when we wrote the article were still in the process of screening so the analysis is based on the numbers to that point–still it’s a lot of kids). We developed a screener based on the Bilingual English Spanish Assessment that we’d previously worked on. The screener is called the Bilingual English Spanish Oral Screening (get it? get it??). It takes about 15-20 minutes to give in both languages (compared to the full version of the test this is about 1/4 of the time). The BESOS includes morphosyntax and semantics sections. If you want to know more about the development of the BESA (from which the BESOS is derived see here and here for morphosyntax; and here for semantics. (And yes, the BESA (but not the screener) includes phonology and pragmatics).
Anyway, in this study we gave the screener to all the kids regardless of whether they thought they didn’t know English or Spanish. Children were preschool and kindergarten age (between 4;6 and 5;6). We did stop testing a subsection if they gave us no response to 5 items in a row (we’re not totally cruel, it’s just that sometimes kids know more than they think– more than their parents and teachers think too!). We were interested in seeing what factors were associated with knowing something, anything in a language. We also wanted to know what factors were associated with higher scores.
The first analysis we did focused on knowing more than 0 (we called this the zero analysis). Here, we found that input and output made a difference in knowing something. For knowing some Spanish both hearing and using the language mattered. For knowing some English only using the language mattered, but not as much hearing the language. Age mattered as well– that is, older children tended to score non-zero in comparison to younger children in the group.
Similar factors predicted higher scores as well, but in slightly different patterns. These patterns depended on both language and domain (morphosyntax and semantics). So, for Spanish morphosyntax children that used Spanish more had higher scores. In comparison for semantics both hearing and using Spanish was related to higher scores. SES as indexed by participation in the free or reduced lunch program was associated with semantics, but not morphosyntax.
English patterns are a bit different. For morphosyntax both hearing and using English was related to higher scores. But, semantics scores were related to using English. This patterns is different from that for higher Spanish scores. Also, mother high school and college education was related to English morphosyntax scores. For semantics, only mother college was related to higher scores. Note that parent’s level of education wasn’t related to Spanish performance.
What does all this mean? Well, first we can’t assume that the same factors that predict higher scores in one language are the same ones that are associated with higher scores in the other language. Also, hearing a second (and first language) and using it both contribute to language learning but again in different ways. Children need to both hear and practice a language to learn it well– but how much they need probably depends on the immediate and longer term goal. Children may need a certain “threshold” amount of hearing and practicing a language in order to start learning and using it (although they won’t use it perfectly). This isn’t that different from learning a first language. Of course children learning a second language already have a pretty good handle on their first language. But, to learn the second language more deeply they need more experience with it. This is why I think we see so much variation in children who are in the process of L2 learning– they are highly variable in when they first had exposure to L2, how much exposure they get, and how much they use it (and in what circumstances). I think our study brings us closer to understanding these complex factors in L2 learning and emphasize the idea that what they learn depends on a number of experiential factors.