Does gender, test medium, or attitude matter? Analyzing test takers' responses to technology-mediated speaking tests

Abstract

In response to increasing concerns about fairness and justice in computerized oral proficiency testing, this study investigated the relationship between test taker attributes and test delivery media of two different technology-mediated speaking tests. Two hundred eight non-native English speakers in a major public US university participated in this study. The results of a multiple regression analysis showed that test takers’ attitudes toward the test delivery media was the most influential predictor of results on the technology-mediated speaking tests, followed by a gender-test delivery medium interaction effect. Further analyses of the qualitative information identified the perceived advantages and disadvantages of two different technology-mediated modes and a human-interlocutor-mediated mode. The results indicated that a live face-to-face interview mode was the most favored in measuring oral proficiency, followed by a computerized mode and an audio-tape-mediated mode. Implications are drawn for appropriate use of technology in measuring oral proficiency.