The reading and mathematics performance of language-minority children in Luxembourg, Serbia and Europe: Is school instruction in their mother tongue important?

In order to explore the complex reality of the importance of mother tongue for the reading and mathematics performance of language-minority children, I conducted three studies. The aims of the studies were threefold:

a) to investigate the reading and mathematics performance of language-minority preschool children in Luxembourg and Serbia (Study I).
b) to identify predictors for early reading performance in the majority language (Study II).
c) to conduct a meta-analysis on the effectiveness of bilingual as opposed to submersion programs in promoting the academic achievement of language-minority children (Study III).

All three studies were theoretically anchored in the psycholinguistic constructs of Jim Cummins, as in OECD (2006, 2010) results reporting the underachievement of minority students in Europe. On the strength of this theory and drawing from empirical evidence, I propose three hypotheses:

a) Hypothesis 1: language-minority children will score significantly lower in both reading and mathematics than language-majority children (Study I).
b) Hypothesis 2: the significant predictors for early reading will be whether the test was conducted in the child’s mother tongue or not, the child’s gender, the level of education of the child’s parents, the child’s range of vocabulary, the child’s phonological awareness, competence in mathematics and the child’s behaviour (Study II).
c) Hypothesis 3: bilingual education programs that include language-minority children’s mother tongue in school instruction are superior to submersion programs that exclude the children’s mother tongue in school instruction in promoting their academic achievement (Study III).

International data show that those language-minority students who do not speak the school language of instruction at home are, on average, one year behind their native peers (Stanat & Christensen, 2007). This gap hampers student’s academic achievement, which in turn restrict the student’s opportunities in the labour market. The purpose of Study I was, therefore, to investigate the performance in reading and mathematics of language-minority preschool children in Luxembourg (N=174) and Serbia (N=159). MANOVA results showed that in Luxembourg, Portuguese children performed significantly lower than both native Luxembourgish children and other minority children. However, with regard to the testing of vocabulary and rhyming words - sections of the test which are evidently loaded with Luxembourgish-specific words - Portuguese and other minority children scored significantly lower than Luxembourgish children. I speculate that language of instruction can be one of the reasons for their possible low performance. In Serbia, Roma performed significantly lower than Hungarians, Serbs and other minority children. This finding may suggest that there are other variables, such as the socio-economic backgrounds of the children that may contribute to the low performance of both the Portuguese and Roma as language-minority groups at school. Thus, Hypothesis 1 is partly confirmed.

Reading skills provide a crucial foundation for children’s success at school (Lonigan, Burgess, & Anthony, 2000) and beyond (Miles & Stipek, 2006). Good progress in reading and mathematics in the earliest years constitute the most important factors which continue to play a role at the age of 11 (Tymms, Jones, Albone, & Henderson, 2007). Study II, involving preschool children from Serbia (N=159) and Luxembourg (N=174), examines the predictive value of the child’s gender, the child’s mother tongue, the level of education of the child’s parents, the child’s range of vocabulary, the child’s phonological awareness, competence in mathematics and the child’s behaviour for early reading skills. For the Serbian sample, multilevel models showed that whether the test was administered in the child’s mother tongue at the age of 5 or not and competence in mathematics were the most significant predictors for early reading at the age of 7 after controlling for age, gender, vocabulary, phonics and behaviour. For the Luxembourgish sample, gender, vocabulary, phonological awareness and competence in mathematics at the age of 5 were significant predictors for reading at the same age, after controlling for age and the mother tongue. The level of parental education in the Serbian sample and the children’s behaviour in both samples proved not to be significant. Thus, Hypothesis 2 is partly confirmed.

The education of language-minority children becomes increasingly important in today’s society. Five previous meta-analyses investigated the effectiveness of bilingual programs in promoting academic achievement of language-minority children in the United States. The present meta-analysis (Study III) investigates seven European studies on the topic. Results indicate a small positive effect (g=0.23) for bilingual over submersion programs on the academic achievement of language-minority children (also see Rolstad, Mahoney & Glass, 2005, 2008). This meta-analysis, therefore, appears to support bilingual education in Europe, the education that includes the mother tongue of language-minority children in the school instruction. Thus, Hypothesis 3 is confirmed. However, the results are restricted due to the small number of studies. More published studies in bilingual education in Europe are needed.