Language of assessment matters: Early learning outcomes when preschool children are tested in isiXhosa mother tongue vs in English: The language of learning and teaching

Background: Many young children in South Africa are enrolled in early learning programmes (ELPs) that use a different language of learning and teaching (LOLT) from their mother tongue. In which language should they be assessed? Aim: To investigate the effect of the language of testing on Early Learning Outcomes Measure 4&5 (ELOM 4&5) performance. Setting: Preschool Programmes. Methods: The nationally standardised ELOM 4&5 was administered to isiXhosa-speaking children (39 boys and 46 girls) attending English LOLT programmes (mean age = 62.12 months; standard deviation [sd] = 4.36). The language of assessment order was counterbalanced (English first and isiXhosa second or vice versa). A mixed linear model with fixed and random effects was fitted with ELOM 4&5 Total score at assessment time 2 in language 2 (English or isiXhosa) as the dependent variable. The model included the following predictors: ELOM 4&5 Total score in the language assessed at time 1, days between assessments, sex, age in months and ratings of Task Orientation. Results: Children performed better in isiXhosa regardless of the language of administration order. Children assessed in English first performed better when tested in isiXhosa second. Conclusion: Early Learning Outcomes Measure 4&5 test scores of English LOLT isiXhosa-speaking children in ELPs are likely to be more valid indicators of their ability when children are tested in their mother tongue language. Contribution: This first South African study to investigate the effects of language of test administration on ELOM 4&5 performance in children attending English LOLT programmes indicates that isiXhosa speakers should be assessed in their mother tongue as required by the national home language assessment policy.