Home-Language Dynamics and Statistical Modelling of Language Allocation in Winterveldt Schools

Abstract

Language of learning and teaching (LoLT) is a crucial equity concern in South African education. Although education policy emphasises mother-tongue instruction, the selection of vernacular languages in schools is often guided by administrative convenience or historical precedent and not by empirical evidence. This study examines home-language (L1) dynamics in Winterveldt, a multilingual area north of Pretoria, where Setswana is the primary vernacular in public schools. Using learner-level administrative records from eight schools (N = 1 947), the study estimates the distribution and dominance of L1s and evaluates the alignment between learners’ reported L1s and the languages offered for instruction. A statistical modelling framework integrating descriptive analysis, dominance indices, and optimisation-based allocation rules is developed to support fair and unbiased language allocation decisions. Findings indicate that Setswana does not uniformly represent the dominant L1 across schools and that a substantial proportion of learners experience a mismatch between their L1 and the school vernacular. Model-based allocation settings exhibit that alternative language structures could significantly increase the proportion of learners accommodated while improving equity across language groups. The study contributes a replicable, data-driven decision-support style for language policy execution in multilingual school contexts.