The dynamics of multilingual assessment: exploring the impact of linguistic accommodations on science achievement

This study examines the impact of linguistic accommodations on the science performance of multilingual pupils. In a randomized controlled trial conducted in Flanders (Belgium), pupils aged 9-12 were assigned to one of three conditions: a control group taking a science test in the language of schooling (n = 64), a group receiving a written bilingual test in both the pupils' L1 and language of schooling (n = 64), and a group with a written bilingual test accompanied by read-aloud accommodations in both languages (n = 69). The hypothesis posited that pupils in accommodated conditions would outperform those in non-accommodated conditions. However, univariate analysis of variance did not reveal significant differences between conditions, suggesting that accommodations did not lead to higher test scores. Subsequent multiple linear regression within the condition involving the bilingual test with read-alouds examined how within-group variance impacted accommodation effectiveness, considering both main effects and interaction effects. Results indicate that proficiency in the L1 and frequency of read-aloud use in the L1 significantly predict science performance. Notably, for pupils who frequently used read-alouds, the significant interaction effect with L1 proficiency suggests an amplified beneficial effect on the test scores when pupils are more proficient in their L1.