Article published in:Onderzoek ontmoet onderwijs 2
[Toegepaste Taalwetenschap in Artikelen 66] 2001
► pp. 79–90
De Vertekening van het Antwoordgedrag in de JA/NEE Woordenschattoets
This article reports on an experiment in which the Yes/No vocabulary test is used with French-speaking learners of Dutch. In previous experiments with Francophone subjects the data were characterised by high false alarm rates which may be due to a response bias. In the current experiment, we aim to get a better picture of the influence of the instruction on the subjects' response behaviour. Two groups of learners received a Yes/No vocabulary test that differed only in the formulation of the instruction. The test was followed by a translation task in order to obtain a concurrent validation. It is shown that a stringent instruction leads to a significant decrease of the false alarmrate. However, it did not result in a higher concurrent validation. These results illustrate the need for a formula that can take into account the response bias. Signal Detection Theory is described as a theoretical model that allows an explicit distinction between response bias and vocabulary proficiency. However, specificities of the Yes/No task have to be considered in order to gain sufficient insight into the merit of the model.
Article language: Dutch
Published online: 24 March 2014