Speech Recognition Technology in Language Testing: Does Taking the Test in an English-Speaking Environment Matter?

Abstract

AbstractThis study investigates a particular use of an application of speech recognition technology in the assessment of English proficiency. The use of the application, called Versant English Test, is examined in the context of a country where English is not the first language of communication, in order to determine whether or not English as the first language of the country in which the test is taken could have a bearing on the test result. As suggested by Chun (2006), this study compares the results achieved by test takers in a non-English speaking environment with those obtained by different test takers in an English speaking environment. To be able to decide whether the Versant is more prone to setting-related bias than other English proficiency tests, the Versant test scores are correlated with the TOEFL scores of the test-takers in a non-English speaking setting and the correlation coefficient is then compared with that achieved in an English-speaking environment. The results suggest that the correlation between the Versant and TOEFL in a non-English-speaking environment is not significantly different from that obtained in an English-speaking environment.