The role of language in word-problem solving: a meta-analysis

Abstract

The relationship between language proficiency and word-problem solving has been extensively studied in the last three decades with one main finding: language proficiency is directly proportional to word-problem solving scores. Researchers have focused on language accommodations to standardized tests to level the playing field for nonnative speakers. Interestingly, several meta-analyses on language accommodation have noted that there are no significant effects on scores. At the same time, most research has reported significance and not effect sizes, which does not allow to establish comparisons between studies. Therefore, the purpose of this meta-analyses was to study the effect size of studies conducted in the US and the rest of the world to provide a new population effect size and identify possible moderator variables that have not been considered before. The main findings were that the differences in scores between native speakers and nonnative speakers (g=.360) were less pronounced than previously reported (g=.604). The participants’ age and the language of instruction and testing moderated a small percentage of the effect size, which leads to the need to include specific information about the participants to provide a clearer picture of this relationship and eventually provide language accommodations that have significant effects on scores.