Reading achievement in Kenya: the language factor

This thesis addresses the schooling challenges posed by children living in African communities where several languages are used and where the language of instruction at school is not their home language. Its focus is the dynamic between languages spoken at home and in the school environment, particularly, how the home language may affect early reading skills. The study is situated in urban Kenya, where Kiswahili is the official language of instruction in lower grades, with English used later on, while many households use other languages at home (e.g. Kikuyu, Kikamba). This dissertation uses survey data collected in 2012 by USAID. The study explores the extent to which the impact of being taught in Kiswahili in lower grades, depends on whether or not Kiswahili is the pupil's home language. School fixed effects are used to control for unobservable factors at the school level. Results from assessments run over a population of pupils who speak Kiswahili at home against those who do not within the same schools are compared. The Kiswahili literacy scores of pupils who speak Kiswahili at home are .206 standard deviation higher than those of pupils who do not speak it at home. The same students also achieve .247 standard deviation more in English, suggesting that speaking the same language at home and at school may also help reading acquisition in another language. The thesis then investigates reading skills' interactions between Kiswahili and English, to see if there is cross-language transfer. Seemingly Unrelated Regression is used to account for equations' cross-correlation. An interdependence is demonstrated between the two languages. The relationship between reading skills in the two languages is not constant and the transfer is stronger when tied to proficiency in English. The results further suggest that having a home language other than Kiswahili is not detrimental to language transfer once a certain proficiency is reached in English and in Kiswahili. Finally, the study examines the linguistic composition of a pupil's peer circle at school and the effect of the group's linguistic diversity on reading outcomes. The peer effect is isolated using a linguistic fractionalization index. This is done in different grades within the same school. Results show that peer effects on Kiswahili scores are mediated by linguistic diversity at school. As the peers' linguistic diversity increases, peers' Kiswahili scores decrease, which negatively affects pupils' own score. On English scores, peer effects are not found to be conditional on linguistic diversity. Findings further show that low achievers are more affected by peer effects than high achieving pupils