Association for Public Policy Analysis & Management

Most sub-Saharan African countries use colonial language as the medium of instruction. For example, the language of instruction in Malawi is a local native language (Chichewa) until the 4^th grade, but changes to English after the 5^th grade. In principle, the use of colonial languages could accelerate globalization and foster economic development. However, this language policy in primary education imposes a great learning barrier to children who do not have fluency in colonial languages such as English or French. We study the causal effects of the language of instruction on 4^th and 5^th graders in primary school who experience a language transition from Chichewa to English in the context of a summer learning program in rural Malawi. We design and implement a field experiment which randomize the language of instruction during a 7-week-long summer school program in which local regular primary school teachers provide math and social studies lessons.

Specifically, 4,960 students in 4^th and 5^th grades were randomly assigned to the “English group”, “Chichewa group” or the control group. 400 students are assigned to the “English group” where a teacher conducts lessons in English with a textbook written in English. 455 students are assigned to the “Chichewa group” where a teacher conducts lessons in Chichewa with a textbook written in Chichewa. 4,105 students are assigned to the control group who do not participate in the summer learning program.

The data we use includes 1) the baseline and follow-up surveys, 2) the teacher survey, 3) the summer school program data including student attendance, homework scores, and weekly quiz scores, and 4) the regular final exam data in the semester following the summer school program. The quiz during the summer school program was administered in English for the “English group” and Chichewa for the “Chichewa group”. However, the regular final exam in the fall semester are administrated in English for all students.

During the summer school, we find that those in the “Chichewa group” performed better than those in the “English group” in both math and social studies. Impacts are particularly larger in social studies.

Results in the regular semester are interesting. The summer school program improved overall test score by 0.2 standard deviation. Improvements are larger in math and social studies test scores (0.28 and 0.29 standard deviation, respectively), which were covered by the program. The program also improved test scores in Chichewa (0.11 standard deviation) which was not covered (spill-over effects). More importantly, those in the “Chichewa group” performed better than the “English group” in math test. However, we do not find such impacts on social studies where the exam was administrated in English.

In sum, we show that teaching in the mother tongue significantly improves primary school students’ learning outcomes. Given that fluency in the colonial language is not required for most people’s daily lives in developing countries, our findings imply that delaying the transition from a local native language to colonial language to later grades could help improving learning outcomes of primary school children in developing countries.