Language of instruction and literacy outcomes: A note on data collection

Language of instruction and literacy outcomes: A note on data collection


By Maya Alkateb-Chami, PhD candidate, Harvard University

With 9% of the world’s children of primary school age being out of school, of those who reach the end of primary school only 58% achieve minimum proficiency in reading. An important factor to consider in this striking discrepancy is linguistic discordance: the mismatch between the language children speak at home and the language of instruction (LOI) at school. More strategic data collection and reporting is needed to understand this problem and devise effective solutions.

Language matters for learning

In a study published earlier this year, where I analysed data from 56 countries, I showed a compelling link between higher linguistic discordance and lower literacy outcomes at the national level. Countries with higher rates of students being taught in a LOI that is not their home language also have higher rates of children unable to read with understanding.

The analysis was based on data made available by the UNESCO Institute for Statistics (UIS) on SDG indicators 4.1.1 (percentage of students who achieve minimum proficiency in reading) and 4.5.2 (percentage of students who have their first or home language as LOI). The latter measure is constructed using the language of school tests as a proxy for the LOI, compared with the responses of students who indicated that they speak the language of the test at home.

The findings provide a wide comparative view, validating existing research on the impact of language barriers on educational outcomes. For example, in South Africa, one study demonstrated that early education in a child’s home language improves literacy rates in English later on. Similarly, in the United States, a study found that students who receive instruction in their home language alongside the dominant language will exhibit better reading performance in subsequent grades. In Cameroon, a study revealed that switching the LOI to students’ home language in early primary education reduced school dropout rates.

The devil is in the details

However, collecting and reporting language-related data is not simple. For example, in Singapore, the linguistic discordance statistic is based on whether students in grade 4 reported that they spoke the language of the TIMSS assessment (English) at home ‘always’, ‘almost always’, ‘sometimes’, or ‘never’. Students considered as having the same mother tongue as the LOI were those who responded with “always” or “almost always” which was the case for 48% of respondents. But another 48% reported speaking English at home “sometimes”, indicating that the language of instruction was not entirely foreign to them. While this case is unique, it demonstrates the limits of a binary measure of students’ home language matching the LOI at school.

One may conclude that we need more information about the situation at hand to better analyse the results. Indeed, reporting on the four options for levels of language familiarity could be helpful. However, it has been argued that more knowledge does not necessarily lead to more understanding. For example, with regard to language use in the classroom, we could consider which language(s) are employed for what purposes and for how long each is used by teachers as well as students, differentiated by speaking in the class plenary, in small groups, and one-on-one. We could also look at language(s) used to teach various subjects, the language(s) of textbooks and assessments, as well as whether one or more of these languages change from one grade to another. If we add the closer-to-reality view that languages are dynamic and fluid rather than static codes, things get even more complex.

A need for strategic data collection and reporting

More data can sometimes obscure our understanding. The right balance needs to be struck between reporting too little and too much information for comparative research purposes.

The UIS choice of using the language of textbooks as a proxy for LOI is not only practical given data availability but also wise, because language learning goals are embedded within it to a certain extent. Supplementing this indicator with one focused on the language of assessments for different subjects at key educational stages, such as the end of primary school and the end of secondary school, would deepen our understanding of language learning aims in different countries. This is especially helpful because having cross-subject exit examinations in a language different from the LOI used in primary education has been observed to have a washback effect on linguistic practices and perceptions in the classroom, influencing teachers’ classroom usage of language toward that of the exit examinations and resulting in lower learning gains in the language with low-stake assessments.

With respect to sampling, national data can obscure details about minority students. It is helpful to measure linguistic discordance experienced by speakers of non-dominant languages in addition to data based on country-level sampling of students. Furthermore, capturing information about out-of-school children would provide a more comprehensive view than solely examining responses from children currently enrolled in school. This is particularly important because LOI itself could play a role in pushing students out of school.

While we need to improve data collection and reporting substantively, the internationally comparative data recently curated by UIS on students having their home language as the LOI is an important resource that should be expanded in geographic scope. The most significant impact of such data is their potential for spotlighting the issue of linguistic discordance, as is also done in the second Spotlight continental report on Africa, which dedicates a section on language.

This window of opportunity is particularly crucial because misconceptions about language learning, ideological biases regarding the value of different languages and dialects, and political influences often converge, leading to ineffective language-in-education policies. And while language is not everything is education, there is enough evidence to show us that attention to it is key to achieving quality education for all.

 



Source link