Do the Annual National Assessments in Mathematics Unfairly Assess English Language Competence at the Expense of Mathematical Competence?

Abstract

This chapter emerges from the author’s broader PhD study that investigated the nature of the linguistic complexity of the Grade 4 Department of Education Annual National Assessments (ANA) test items and how learners (with a poor command of the language of learning and teaching) experience them. It compares first-language-isiXhosa learners’ performance in solving selected ANA test items with a high linguistic complexity index in a written unmediated assessment with their performance in the same items with linguistic mediation in the form of task-based interviews (including translation into isiXhosa). This chapter reports on the performance of learners who registered significant improvement in performance with the aid of linguistic support. It draws on three data sources: content analysis of the 2013 mathematics ANA test items, learner performance on these items in the 2013 written ANAs and task-based interviews with the selected learners.

Findings from the task-based interviews point to enhanced mathematical performance due to linguistic mediation among learners on the selected items. Content analysis points to serious linguistic difficulty of test items which compromise English-language learners’ performance, thus calling into question the validity of such assessments for these learners. The study recommends consideration of the linguistic complexity of test items and trialling of test items on English-second-language learners prior to their use in national assessments. Such considerations are especially pertinent at the Grade 4 level where the majority of South African learners would have had only a few months of mathematics instruction in English before sitting for these assessments in English.