The argument for evaluating monolingual language tests for equivalence across language groups

Abstract

The demonstration of scalar equivalence in language proficiency tests (which can be viewed as monolingual language tests) has often been deemed as unnecessary as it is argued that the biases associated with the language of a test (used across multilingual language groups) will not occur. However it is increasingly acknowledged that scalar equivalence is as important in monolingual language tests as it in multilingual language tests. This paper will provide empirical support for the argument that the meaning of tests scores across groups (scalar equivalence) is as important in monolingual language proficiency testing as it is in any other cross-linguistic testing. The authors will present research conducted on the equivalence of an adapted English version of a standardised academic language proficiency test (Woodcock-Munoz Language Survey, WMLS, 2001), with its intended use being across English-first-language speakers and isiXhosa-first-language speakers. More specifically, the focus will be on an item bias analysis across the English- and isiXhosa-first language speakers for all the sub-tests of the adapted English version of the WMLS.