ACQ Vol 13 no 3 2011

Another area of related evidence comes from a larger body of research regarding bilingual education. This line of research provides insight into the nature of language learning associated with bilingualism. Slavin and Cheung (2005) reviewed 17 studies across a range of bilingual education models (mostly involving Spanish-English bilinguals), concluding that “bilingual instructional models [produced more favourable learning] over those that eliminated [the] native language” (p. 280). The conclusion was that the evidence is mounting in favour of bilingual learning producing superior results for bilingual children, even when the second language is dominant in the education system. The research confirms that maintenance in the first language is a predictor of future proficiency in the second language as well as a powerful tool to assist in the transfer of literacy knowledge from one language to the next. Carlo et al. (2004), Combs, Evans, Fletch, Parra, and Jimenez (2005), and Rolstad, Mahoney, and Glass (2005) are other useful sources for this related evidence. Furthermore, the benefits of bilingual education models are likely to apply to children with language difficulties as well (Culatta, Reese, & Setzer, 2006). The problem of “evidence” In complex areas, such as child language and multiculturalism, building a body of evidence that adheres to the NHMRC standards is problematic. The levels of evidence tables such as that of the NHMRC make judgements about “strong” and “weak” evidence based on medical models that require reduction to controllable variables with tightly defined populations and simple interventions. Such systematic control is not well suited to child language difficulties, where clinical populations are poorly defined, and where complex and variable interventions are used (Law, Campbell, Roustone, Adams, & Boyle, 2008). Furthermore, clinical principles in the field of speech and language pathology favour adaptation of interventions to individual needs (Speech Pathology Australia Code of Ethics, 2008), which makes such variables even harder to control. Pring (2004) indicated that randomised control trials in the first instance are not appropriate as we need to develop a strong body of foundation research at the case-study and small control group level. He outlined a progression, wherein specific therapies are developed for well-defined groups, tested first in small-scale efficacy then effectiveness studies, and results disseminated to clinicians for clinical application, before any larger scale studies should be attempted. Fey (2006) made the point that “the motivation for higher level studies and the justification for sponsoring them financially generally comes from studies that have already produced encouraging results using less costly, lower level research designs [that were high in quality]” (p. 318). Certainly premature RCTs may be conducted by glossing over problems, resulting in unusable results, such as could be said of Glogowska, Roustone, Enderby, and Peters (2000). Another incompatibility that emerges is that cultural and linguistic diversity is about variation from the norm, whereas the evidence level system is about the norm and about subsuming variation within a group to produce statistically robust results over large numbers. But the concept of “normal” populations of minority groups within a dominant culture, especially recent migrant or refugee groups, is problematic. Types and degrees of bi-or multilingualism vary, the amount of identification and practice of originating culture to dominant culture varies, and both of these are

subject to sometimes rapid change over time and can be very difficult to judge. This issue presents us immediately with a dilemma. If we try to produce the evidence based on the usual criteria, we run the risk of not finding strong evidence due to the inherent difficulty of running large group studies, and therefore making our arguments look weak. We also may find ourselves concentrating on the evidence that is of least value, that is, those aspects which can be subjected to large group similarities, when the main interest is in variation. Not only do the levels of evidence downplay the significance of the lower levels, they fail entirely to deal with qualitative data. An important aspect of evidence based practice and research is considering the client and family preferences and needs, and with this aim, the client experience should form a major part of the research. Much of this research, along with much “outcome” research, should be qualitative (Kovarsky & Curran, 2007). However, this research is sparse at best, and discounted even when it is conducted and published. The positivist and therefore culture-bound value system bound up in the levels of evidence is also incompatible with cultural competence perspectives, and needs ongoing critical scrutiny (Martin, 2009; Kovarsky & Curran, 2007). Clinical bottom line After finding and reviewing the evidence, weighing up both its strengths and limitations, you draw a number of evidence based conclusions to guide your practice and thinking around this complex issue. Reaching a clinical bottom line is important to be able to challenge the current perspectives of your colleagues and the prevailing monolingual model of practice. Guided by the evidence, the key findings that you want to communicate are: • Children with language difficulties learn bilingually at least as successfully as monolingually . The intervention studies reviewed here have indicated that when language intervention is provided to bilingual children in both languages, the children were capable of learning the two languages to at least as good a level as one, and there is some evidence to suggest their achievement can be superior in bilingual intervention (Kay-Raining Bird et al., 2005; Seung, Siddiqi, & Elder, 2006; Thordardottir et al., 1997; Tsybina & Eriks-Brophy, 2010; Waltzman et al., 2003; Wauters et al., 2001). Of particular importance is that there is no evidence supporting the argument that performance is worse as a consequence of bilingual intervention compared to monolingual. • Use bilingual clinical intervention, rather than just encourage bilingualism . Kohnert (2010) concluded that systematic support for the home language(s) of young children with language impairment is critical to the long- term success of language intervention. She holds that encouragement of home language use is not going far enough, and fails to recognise the significance of the child’s social, emotional, and cognitive development taking place within the cultural context of the family. SPs not only can, but should conduct bilingual interventions. • Clinicians can conduct interventions when they do not speak all the languages competently . It is evident that a range of teamwork options are available that can support a shift away from monolingual practice. Family involvement is consistent with family-centred practice principles and the research suggests that children successfully learn language targets with parent-based

152

ACQ Volume 13, Number 3 2011

ACQ uiring Knowledge in Speech, Language and Hearing