Revisiting the effects of language-of-instruction policy in Ethiopia

• In Event: From Early Reading to Language Policy to AI: Global Perspectives on Education

Proposal

Children in Sub-Saharan African (SSA) countries tend to face education where the language of instruction (LOI) is not their mother tongue but a local language and a European language. Those countries are common in the large numbers of the local languages. The governments tend to adopt limited numbers of local languages as LOI for limited grades from grade 1 and a European language for the later grades. In the multilingual settings, LOI is an expected factor to affect education attainment and learning outcomes through education. Psycholinguistic theories assume that education in students’ mother tongues is more effective on learning particularly in the beginning grades, as the pupils basically have better skills of the languages. Human capital theory expects that more effective education increases schooling periods. Thus, education in pupils’ mother tongues for wider grades from grade 1 is assumed to enhance education attainment and learning outcomes.

However, pro-mother tongue LOI policies are not necessarily more effective on those outcomes. For example, the effects of those policies depend on the quantity and quality of other education inputs, such as textbooks and teachers corresponding to the mother tongues. Some previous studies quantitatively examining pro-mother tongue LOI policies in SSA countries actually argued the negative effects. Furthermore, the suggestions even from the studies focusing on the same policy change in Ethiopia are mixed. Recently sophisticated knowledge of causal inference with difference-in-difference (DiD) approaches implies that the inconsistent conclusions about the policy change in Ethiopia partly come from methodological problems: The defined control groups include already-treated individuals and the estimation models are not compatible with multiple time points of treatment, which can cause bias.

A goal of this study is to re-examine the effects of the policy change in LOI in Ethiopia happening after the political upheaval in 1991. My study and the previous studies focusing on Ethiopia are common in terms of the focal policy change and roughly share the same outcomes and approach of causal inference (i.e. DiD). On the other hand, I escape from the methodological problems that the previous studies may have by separating the units of analysis based on their ethnicities linking to their mother tongues, defining the comparison groups for each ethnicity as the people who were never treated or not yet treated and adopting a different estimation model suitable for treatment at multiple time points. What I investigate are (1) the relative effects of a LOI policy to extend the width of the grades in the mother tongue of a ethnic group from grade 1-6 to 1-8 by replacing English and the relative effects of policies to change the LOI for grade 1-6 and 1-8 respectively from (2-1) a local language that is not the mother tongue for a ethnic group and (2-2) both the local language and English to its mother tongue on the highest grade completed and literacy in the mother tongue among each ethnic group. The dataset of analysis in my study is repeated cross-sectional consisting of data collected through the Demographic and Health Survey Ethiopia 2000, 2005, 2011 and 2016.

Ethiopia has at least 77 local languages. Its previous policy of LOI before the political upheaval in 1991 selected only one local language and English as the LOI for grade 1-6 and the later grades respectively. Triggered by the regime change in 1991, the country's LOI policy was gradually replaced by the fractionalized ones for both the regions and local languages. The uniqueness of the new LOI policy in Ethiopia in SSA includes the adoption of local languages as LOI for a comparatively wider grades (e.g. grade 1-8) and the expansion of the local languages to be adopted as LOI from one to more than 10.

My study makes three contributions to the existing studies. First, I reconfirm the effects of a pro-mother tongue LOI policy replacing a local language with the mother tongue of children in Ethiopia by applying a different design of a DiD approach that seems to be more reliable according to recently developed knowledge of DiD approaches. Second, I examine the effects of another pro-mother tongue LOI policy extending the width of grades in students’ mother tongue to grade 1-8 by replacing English, which was not yet investigated in the previous studies. The effects of LOI policies adopting student’s mother tongue especially for longer grades beyond grade 4 have not yet reached a consensus among the empirical studies. Those examinations are meaningful for Ethiopia where the LOI policy was more frequently changed in SSA. Examining the effects of the two types of the LOI policy may also be significant to other SSA countries in the future. Compared to Ethiopia, the LOI policies in other SSA nations tend to limitedly adopt local languages as LOI in terms of the numbers and the widths of the applied grades. Lengthening the width of grades in pupil’s mother tongues may become more seriously considered, as shortage of skillful people and expansion of enrolment in secondary and higher education become remarkable. Substituting a local language for the mother tongue of a language group can be a practically tentative option of LOI policies under restrictions of resources.

Third, I shed light on the effects of the LOI policy in Ethiopia not only on average among the all units of analysis but also among each subgroup of the units by their birth cohorts. Those investigations suggest a change in the effects of the policy by time periods, which can happen, for instance, by improvement in the quantity and quality of other education inputs necessary to the LOI (e.g. teachers who speak the language and textbooks corresponding to the language). Examining the time-average effects of a new LOI policy only for the beginning several years may mislead about the potential effects of that LOI policy. Heterogeneity in the effects of a governmental LOI policy in SSA on the schooling and learning outcomes is not yet well studied.

Author

• Takashi Yamaguchi, University College London, Institute of Education