by Brian A. Goldstein
“I live on a one-way street that’s also a dead end. I’m not sure how I got there.”
At the recent 2012 convention of the American Speech-Language-Hearing Association (ASHA) in Atlanta, Georgia, USA, I attended many talks focusing on multilingual speakers. Overall, the presentations were of high quality, and I learned a great deal. I noticed, however, that many of the presenters mentioned, and sometimes measured, the children’s language “dominance.” Yes, I purposely put that word in quotes.
Although “dominance” is an intuitively pleasing word/notion, it is ultimately unsatisfying. I am not sure the word “dominant” is an appropriate one to use because, as a unit of measure, it neither describes nor explains. It puts us down a one-way road on which there is oncoming traffic and that is a road we don’t want to travel. To continue the metaphor, the oncoming traffic is the non-“dominant” language. That is, when we aim our headlights at the “dominant” language, then that is the only language that gets the attention. The “other” language ends up in our blind spot, and thus, we might focus only on the one language that is considered “dominant.” Such blinded attention serves to view the bilingual as two monolinguals in one, a notion against which Grosjean (e.g. 1989) has repeatedly, vigorously, and convincingly argued.
“Dominance” is suspect for a variety of reasons, some of which were addressed in this blog before. First, there is not a singular definition. It has been used to describe proficiency, age of acquisition, range of levels, functional strength, scores on standardized tests, etc. (Butler and Hakuta, 2006). Second, it is notoriously difficult to measure (Valdés and Figueroa, 1994). For example, to even attempt to measure it, Valdés and Figueroa indicate that the examiner must measure performance in each of the two languages across tasks, contexts, settings, modalities, and functions and then compare the individual’s total performance for each language. This set of tasks necessitates parallel and equivalent instruments in the two languages. Needless to say, few of these types of instruments exist. Finally, such tests of dominance have been criticized roundly for their relatively poor psychometric properties (e.g. MacSwan, 2005, see item 112 in the Contents; MacSwan and Rolstad, 2006; Mahoney and MacSwan, 2005). Example criticisms include:
- lack of a theoretical foundation
- weak validity and reliability
- lack of adequate operational definitions or norms
- use of arbitrary cut-off scores
- assuming “dominance” is stable over time
- poor classification rates
The result of such relatively poor psychometrics properties is an over-representation of bi-/multilinguals in special education programs, which renders these tests, “...at best arbitrary, at worst, dangerous” (p. 376, Shuy, R. (1978). Problems in Assessing Language Ability in Bilingual Education Programs. In H. La Fontaine et al. (eds.), Bilingual Education. (pp. 376-381). Wayne, NJ: Avery Publishing.).
These concerns provide the backdrop in terms of one of my apprehensions about language “dominance”. I am concerned that once, and if, “dominance” is determined, it will never be re-evaluated, especially for multilingual children who have communication disorders and are enrolled in intervention. That is, once the child is determined to be “dominant” in Language A, she will always be seen as “dominant” in Language A. Language B will not be merely the road less taken but will be the road never taken (my deepest apologies to Robert Frost). Children’s language skills change significantly over time, especially for multilingual children whose input and output vary based on a host of factors including age, community, parents, siblings, educational status, etc. (e.g. Rojas and Iglesias, in press).
Thus, “dominance,” even if it is psychologically responsible (Menn, 1992), is a moving target in the same way that one adjusts their driving based on the type of road, driving conditions, and make of car. We would be better off examining, and re-examining, multilingual children’s language use (i.e. how often they use each language) and language proficiency (i.e. how well they use each language). By better off, I mean examining constructs that are reliable and valid and aid us in understanding the variables that impinge on language acquisition in multilingual children. As Valdés and Figueroa (1994, pp. 66-67) said, “[m]easuring bilingual ability and bilingual proficiency is both complex and problematic...the best we can do is use a series of measures that together might provide guidance for describing an individual as more like or unlike other bilingual individuals...”
Furthermore, examining language skills in multilinguals needs to be specific to each language domain (i.e. phonology, morphology, syntax, semantics, lexicon, and pragmatics). That is, an omnibus notion of “dominance” is equivocal at best and simply does not seem to be supported by the available evidence, at worst, as I have alluded to in a previous post. For example, Ball, Müller, and Munro (2001) examined the production of the trill /r/ in 44 Welsh-dominant and 39 English-dominant children aged 2;6-5;0. They found that production of the trill /r/ differed depending on language “dominance” (quotes mine). Welsh-dominant children acquired the trill earlier than English-dominant peers.
There are other studies that indicate that “dominance” is not a critical factor in accounting for language skills (phonological, in this case). Law and So (2006) examined 100 Cantonese- and Putonghua-speaking bilingual children, aged 2;6-4;11. The children were grouped as either “Cantonese dominant” or “Putonghua dominant” (again, quotes are mine). They found that irrespective of language dominance, Cantonese phonology was acquired earlier than Putonghua phonology. Such equivocal results indicate that labeling a child as “dominant in English” or “dominant in Amharic” simply does not seem warranted as it neither describes nor explains. Finally, Paradis (2001) examined syllable omissions in French-English bilingual 2-year-olds. In 4-syllable target words, English-dominant bilinguals preserved higher frequency of 2nd syllables than did French-dominant bilinguals, but French-dominant bilinguals preserved a higher frequency of 3rd syllables than did English-dominant bilinguals.
Going forward, the picture might even be more complicated than this. Take phonology, for example, a multilingual child could have a superior ability in producing long words in Language A compared to Language B but superior ability in producing complex syllables in Language B than in Language A. If that is the case, then in what language is the child “dominant?” The point is that even defining “dominance” is more complicated than it seems. What is seemingly intuitive is actually quite difficult and nuanced. I am suggesting that we need to examine the child’s skills across each domain in all of their languages and relate those outcomes to their language use and language proficiency. That way, regardless of the car you drive, you will not be traveling down a one-way street with oncoming traffic. Just remember:
“Understanding is a two-way street.”
Brian A. Goldstein is Dean of the School of Nursing and Health Sciences and Professor of Speech-Language-Hearing Sciences at La Salle University, Philadelphia, PA, USA.
© Brian A. Goldstein 2013