The Impact of Multilingualism on Children’s Reading/Writing Skills and Scholastic Performance

While research on multilingualism has shown both, positive (e.g. inhibition; Coderre et al., 2013), and negative (e.g. vocabulary; Bialystok et al., 2008) effects on cognition and language proficiency, its influence on scholastic achievement appears to be largely negative (Hoffmann et al., 2018; Martini et al., 2021). Children in Luxembourg are educated in a multilingual educational system. In Kindergarten, the main teaching language is Luxembourgish. This switches to German for literacy acquisition in elementary school, with French taught as a second language. Despite its small size, Luxembourg is also highly multi-cultural, boasting 170 nationalities (The Government of the Grand Dutchy of Luxembourg, 2023). Thus, many of the children in the school system do not speak the language(s) of instruction at home. Data from the Luxembourgish national school monitoring program reveals significant differences in German reading comprehension in grade 3 depending on the language spoken at home. Because Luxembourgish is linguistically close to German, Luxembourgish-speaking children generally perform better than children who do not speak Luxembourgish at home (Hoffmann et al., 2018; Martini et al., 2021).

Furthermore, the language-based differences in children’s scholastic performance complicate the diagnostic process of children with potential learning disorders, such as dyslexia and/or dyscalculia. In Luxembourg, the language in which children are screened and diagnosed for potential learning disorders is usually identical to the main language of instruction at school, which at time of diagnosis (typically grade 3) is German. It is therefore difficult to distinguish poor performance based on potential difficulties with reading/writing or mathematics from poor performance based on low language proficiency in the test language. Furthermore, the diagnostic tools currently employed in Luxembourg are developed in countries with primarily one language of instruction, challenging the validity of the diagnostic process in a multilingual population (Ugen et al., 2021).

We have thus developed a comprehensive reading/writing test battery adapted to the Luxembourgish educational curriculum and multilingual environment. Children’s potential language proficiency differences in the test language (German) are taken into account using simplified instructions with reduced language load, multiple examples, varying degrees of difficulty of the test materials, as well as the construction of distinct language-group norms, depending on the language(s) spoken at home. This helps avoid over-diagnosis of reading and writing disorders in children who do not speak the language(s) of instruction at home and underdiagnosis of children who do. The developed test battery assesses children’s performance in key domains relevant for reading and writing comprising phonological skills, (non)word and text reading (fluency and accuracy), reading comprehension, writing, and vocabulary. Furthermore, we link children’s performance in the newly developed test battery to their performance in the Luxembourgish national school monitoring program.

Methodology, Methods, Research Instruments or Sources Used
We have tested 214 children during the pre-test phase of the project (February – June 2022; age 8 – 12; M = 9.59; SD = 0.68; 95 girls) and will test approximately 735 children during the validation and norming phase (February – June 2023). All children attend grade 3 in public primary schools in Luxembourg. The distribution of classes participating in the project covers all 15 regions of the country, resulting in a representative sample of the Luxembourgish school population.

Children complete the 9 sub-tests of the novel reading/writing test battery, which includes precursor skills: Rapid Automatized Naming (RAN), non-word phoneme segmentation, non-word phoneme deletion; reading skills: word and non-word reading, text reading and comprehension; writing skills: gap dictation and text dictation; as well as a receptive vocabulary task. The vocabulary and writing skills are assessed in a group setting (all children complete the tasks together in the classroom), the precursor and reading skills are assessed individually in a quiet room in the school. The total testing time (group test + individual tests) does not exceed 90 minutes per child. All tests are conducted by trained test administrators following a standardised procedure.

The pre-test data were analysed per sub-test using Repeated Measures Analysis of Variance with language group as the between-subject factor and results of the sub-test (per category where applicable) as the within subject factor. Significant main effects of language group were explored using post-hoc pairwise-comparisons (Bonferroni corrected t-tests). Four language groups were created based on the frequencies of the reported language(s) spoken at home: Luxembourgish/German monolingual, Luxembourgish/German bilingual, Romance language (e.g., French, Portuguese, Spanish) mono- and bilingual, Other language (e.g., English, Slavic) mono- and bilingual.

The results of each sub-test of the novel reading/writing test battery were also correlated with children’s performance on German listening and reading comprehension in the Luxembourgish national school monitoring programme (Bonferroni corrected Pearson correlations).
Conclusions, Expected Outcomes or Findings
The results of the pre-test phase show that children, who speak Luxembourgish or German at home outperform children who speak a Romance or Other language at home. Particularly, significant differences between language groups were observed for: word reading accuracy (F(3,190) = 4.94, p = .003); word reading fluency (F(3,190) = 4.59, p = .004); text reading accuracy (F(3,190) = 8.73, p < .001); text reading fluency (F(3,190) = 11.50, p < .001); text comprehension (F(3,190) = 12.45, p < .001); gap dictation (F(3,180) = 10.52, p < .001); text dictation (F(3,180) = 18.22, p < .001). The significant main effects of language highlight the need for separate language group norms for screening and diagnostic purposes. The lack of main effects of language for non-word phoneme deletion, non-word phoneme segmentation, and non-word reading indicate that the sub-tests using non-words were successfully constructed to account for language proficiency effects.

Significant Pearson correlations between the school monitoring results of German listening (.28 < |𝜌| < .59) and German reading comprehension (.24 < |𝜌| < .65) and the majority of the newly developed sub-tests of the reading/writing test battery were also observed. These correlations provide a measure of construct validity, illustrating the significant link between children’s scholastic performance and performance in the novel reading/writing test battery.

We expect to replicate these initial findings with a larger sample of children during the validation and norming phase of the project and supplement our data analyses with more detailed results highlighting the distribution of scores per sub-test based on language spoken at home and its effect on scholastic performance as assessed by the Luxembourgish national school monitoring program.
References
Bialystok, E., Craik, F., & Luk, G. (2008). Cognitive Control and Lexical Access in Younger and Older Bilinguals. Journal of Experimental Psychology: Learning Memory and Cognition, 34(4), 859–873. https://doi.org/10.1037/0278-7393.34.4.859
Coderre, E. L., van Heuven, W. J. B., & Conklin, K. (2013). The timing and magnitude of Stroop interference and facilitation in monolinguals and bilinguals. Bilingualism, 16(2), 420–441. https://doi.org/10.1017/S1366728912000405
Hoffmann, D., Hornung, C., Gamo, S., Esch, P., Keller, U., & Fischbach, A. (2018). Schulische Kompetenzen von Erstklässlern und ihre Entwicklung nach zwei Jahren. In T. Lentz, I. Baumann, & A. Küpper (Eds.), Nationaler Bildungsbericht (pp. 84–96). University of Luxembourg & SCRIPT.
Martini, S., Schiltz, C., Fischbach, A., & Ugen, S. (2021). Identifying Math and Reading Difficulties of multilingual children: Effects of different cut-offs and reference group. In M. Herzog, A. Fritz-Stratmann, & E. Gürsoy (Eds.), Diversity Dimensions in Mathematics and Language Learning (pp. 200–228). De Gruyter Mouton.
The Government of the Grand Dutchy of Luxembourg. (2023, January) Society and culture – Population Demographics. https://luxembourg.public.lu/en/society-and-culture/population/demographics.html
Ugen, S., Schiltz, C., Fischbach, A., & Pit-ten Cate, I. M. (2021). Lernstörungen im multilingualen Kontext. Diagnose und Hilfestellungen. Melusina Press. https://doi.org/10.26298/bg5s-ng46