VST as a reliable academic placement tool despite cognate inflation effects

Abstract

Abstract Knowing academic words is essential to achieving success at university, so reliable tests are needed to estimate students academic vocabulary. Recently, attention has been drawn to how cognates may overestimate vocabulary test results, making such tests unsuitable for placement. Here, we investigated whether a combination of results from two tests could be used for academic placement purposes despite cognate inflation effects. We assessed the receptive vocabulary knowledge of 106 Polish English majors (B2 level or higher) via the Vocabulary Size Test (VST; Nation & Beglar, 2007) and a tailor-made checklist Academic Vocabulary Test (AVT). The comparison of cognates and non-cognates in both tests evinced a clear cognate inflation, which distorted the scores. However, a regression analysis indicated that scores in the VST significantly predicted students scores in the AVT. Furthermore, cluster analyses combining both tests’ scores identified the VST threshold at or above which successful performance in the AVT was highly likely. Thanks to the identification of this threshold, we argue that the VST can be reliably used for placement purposes with English majors in Poland. We also explain why some low-frequency VST bands become disproportionally easy due to the existence of cognates typically present across languages.