A very interesting paper by Prof Tarun Jain of ISB.
The summary is given at Ideas4India. The basic idea is students prefer instructions in their mother tongue . If this does not happen due to policy the learning suffers:
Till the mid-19th century, education in India mostly followed indigenous systems without resorting to standardisation, and was largely restricted to the social and economic elites. Subsequently, when East India Company officials started focussing on administration of their conquered territories, they introduced formal education to train potential employees for clerical positions, as well as to create acceptance of Western traditions and colonial rule. Following Wood’s Despatch1 in 1854, most provinces used the dominant vernacular as the medium of instruction. For instance, Tamil was used as the medium of instruction in most districts of the Madras Presidency, except for Telugu in the districts that form the present state of Andhra Pradesh.
Consequently, the numerically-dominant mother tongue was the same as the official language of the province in some districts (matched districts) but not in others (mismatched districts). For example, in the state of Hyderabad, the medium of instruction until the 1950s was mostly Persian/Urdu, which was at odds with the mother tongue of many residents in the Nizam’s2 Dominion. Writing in 1886, Motaman Jung of the Educational Department of Hyderabad stated “Considering the population, schools ought to exist, as far as mass education is concerned, in the following ratios: Telugu 4, Mahratti 3, Canarese 1, Persian 1. Facts however show the reverse of this viz., Persian 4, Mahratti 2, Telugu 1, Canarese 0.”
Figure 1. Mother tongue vs. official language of provinces in South India: Matched and mismatched districts
Comparing matched districts to mismatched ones is like a natural historical experiment on the effects of using mother-tongue instruction in schools on long-run educational outcomes. District-level data from the Census of India shows that mother-tongue instruction led to persistent increases in educational achievement that lasted well past Independence till the 1980s.
The impact was greatest in primary and secondary schooling, which is mostly conducted in the vernacular. Specifically, the literacy rate in districts where students were taught in their own language was 18% higher than in mismatched districts, and the middle school completion rate 25.2% greater. The difference in university education was very small, which is not surprising since most post-secondary instruction in the colonial era was conducted in English.
In 1956, following the recommendations of the Fazal Ali Commission, the government reorganised the state boundaries in South India on strict linguistic lines. A key policy change that was implemented in all states after reorganisation was to extend the official language of the state as the medium of instruction in schools. This meant that more students were now offered schooling in their mother tongues. Did districts that were mismatched to begin with ‘catch-up’ in subsequent educational achievement?
The district-level Census data shows that reassignment had a large and significant impact on literacy, middle-school completion, and matriculation rates. Hitherto mismatched districts experienced higher growth rates than matched districts as they caught up with the latter. Specifically, the growth rate was 37% higher for literacy, 56% greater for middle-school completion, 60% for matriculation, and 51% for college graduation. Thus, while colonial-era provinces – that had both linguistic minority and majority areas – suffered from differences in long-term educational outcomes, the linguistic reorganisation of states seems to have remedied these differences.