"You don’t improve a language by speaking it less"

New research shows that Flemish nursery pupils are underperforming in both maths and language. In De Afspraak on VRT Canvas, this immediately sparked renewed calls for more use of home languages other than Dutch in education. It is an idea that keeps resurfacing in education debates. But it remains a bad idea.

Supporters of home-language teaching often argue that children who are more proficient in Arabic, for example, will also become more proficient in Dutch. By that logic, speaking a home language in the classroom would strengthen Dutch language development. That is simply not true. Even among highly gifted learners, language proficiency does not automatically transfer from one language to another. No one becomes better at a language by speaking it less. The strongest predictor of language proficiency remains exposure and practice. For children who do not speak Dutch at home, school is often their only opportunity to learn the language properly.

Speaking less Dutch also does little for the home language of many linguistically disadvantaged children, especially when the teacher does not speak that language. In practice, multilingualism in Flanders often proves to be a utopia; the reality is more often one of no language mastery at all. The simplistic idea of transfer between languages can just as easily be reversed: if we focus maximally on Dutch, then Arabic should automatically improve too. It is the same reasoning, yet supporters of home-language education never make that argument.

Another frequently cited argument points to international studies where the school language improved after introducing another home language. But those older examples are not relevant to today’s super-diverse Flemish context. In the rare success stories, the children involved were multilingual but still formed a relatively homogeneous group — for instance, children who migrated from Mexico to the southern United States. They received carefully designed bilingual education, taught by teachers fully fluent in both languages, supported by highly targeted teaching methods. None of that can realistically be applied in Flanders, where teachers cannot reasonably be expected to master the fifteen different home languages spoken in a single classroom — a key condition for those programmes to succeed.

Homogeneous group

There is also no empirical evidence showing that systematically using another home language in Flemish classrooms improves Dutch language proficiency. What is often left unsaid is that advocates of home-language education already tested this approach in Ghent more than a decade ago. Even then, language diversity was too great to involve an entire school, so the project focused on a relatively homogeneous group: bilingual children in Ghent who spoke Turkish at home. The conclusion was clear: neither Dutch nor Turkish language skills improved. Teachers and parents were unconvinced. In our experience, it would now be difficult to get such an experiment involving children past an ethics committee.

Today, strong command of Dutch among migrant pupils is too easily dismissed as unattainable. Those low expectations become a self-fulfilling prophecy. In this new study, teachers expected no successful school trajectory for 30 per cent of the nursery children — even though they were barely five years old. Home-language education then presents itself as an easy but superficial solution. That socially engaged openness carries with it a deeply tragic and deterministic view of children. As though this generation of children with a migration background were somehow the first in history incapable of learning Dutch. Of course that is not the case. Across the world, both privileged and disadvantaged children learn the language of the environment in which they grow up.

Language sponges

The real challenge is to make that environment as language-rich as possible — and as early as possible. Children’s brains are language sponges. Alongside frequency of exposure and practice, the age at which learning begins is one of the strongest predictors of later language proficiency. The earlier, the better, because the brain’s flexibility decreases after the age of six. Abandoning Dutch as the exclusive language of schooling illustrates a broader problem in education policy: well-intentioned and socially inspired ideas often lead to approaches that appear empathetic, yet ultimately undermine social mobility.

Integration, access to the labour market, higher education, social participation, social cohesion and citizenship all require knowledge of Dutch when living, learning and working in Flanders. The PIRLS study shows that only 52 per cent of Flemish children always speak Dutch at home. In many other countries, the home language is far more often the same as the school language, including in migration countries such as England, where the figure is 69 per cent. Flemish education is paying the price for a wider failure in integration policy, where non-commitment too often prevails. If we expect education to shoulder its social responsibility, then exposure to Dutch must be maximised.

High expectations

It was telling that the three intelligent and successful women discussing this issue on television had each learned Dutch exceptionally well within their own migration stories. Their parents held high expectations and considered speaking Dutch important. Flemish Education Minister Zuhal Demir, Professor Els Consuegra and journalist Fatma Taspinar have all spoken about this repeatedly in interviews.

Educational achievement is far more strongly linked to the language proficiency of children and their parents than to their income. Social inequality is, to a large extent, language poverty. Children who do not speak Dutch at home lag roughly a year behind children who do. That is true even for children without a migration background. The language of schooling matters far more than migration itself, because language carries thought, knowledge and instruction. It is the driving force behind both educational disadvantage and social mobility.

So what nursery education needs is not less Dutch, but more. An occasional word of Turkish is no disaster, but systematically prioritising home languages is practically unworkable, inefficient, and no substitute for structured, intensive teaching in the language of schooling. Our teacher training programmes should equip teachers for exactly that task, with ambitious learning goals for every child.