As some Programme for International Student Assessment (PISA) studies claimed that native speaking (NS) students outperform language minority (LMi) students, far-reaching inferences have been drawn by policymakers. However, previous PISA assessments were not appropriate because they only included a dichotomous home language variable. The main objective of this study is to gain a better understanding of how students’ language background and use are related to academic achievement.

The PISA data from 2012 provides a unique opportunity to fill this research lacuna as it includes a more elaborated questionnaire on language background and use.

Multivariate three-level analyses are conducted on PISA 2012 data from 18 countries, covering about 5,000 schools and 120,000 students.

The results show that there is an achievement gap between LMi and NS students for both reading and math. After controlling for students and school characteristics, the LMi–NS achievement gap narrows, but remains significant. This holds true for most countries. However, language use per se is not the cause of underachievement: LMi students who more often speak a minority language with their parents do not achieve less. In some countries, speaking a minority language more often with parents is actually positively related to math and reading achievement. Nevertheless, speaking the instruction language in the school context is positively associated with math and reading achievement.

This study revealed that the relation between language use and academic achievement is more complex than it was conceptualized in most previous PISA studies. Scholars need to go beyond the dichotomous approach to achieve a better understanding of language use. Our results show that linguistic diversity could function as an asset for academic performance, at least if a good balance between focus on minority languages at home and instruction language at school can be found.