Exploring the relationship between self-assessments and opic ratings of oral proficiency in french

Abstract

The present study analyzed the self-assessed spoken French language abilities that students said they ‘can do’ in relation to the ACTFL proficiency scores they received on an oral proficiency interview by computer (OPIc). A secondary aim was to assess different scales that have been used to convert OPIc ratings to numeric scores. French university students (N = 216) of varying proficiency levels rated a series of can-do statements related to speaking skills. They then completed the ACTFL OPIc test, which was rated by certified ACTFL raters. A series of regression analyses showed that the strength of the relationship between self-assessment and OPIc ratings was strongly influenced by the type of numeric scale used: When data were ranked ordinally and analyzed using an ordinal regression, a majority (65%) of variance in OPIc scores was explained by self-assessment scores. Analyzed using linear regression, when scores were converted to equal-interval scales, self-assessment scores explained approximately 30% of variance. On a graduated scale that reflected the increasing distances between ACTFL (2012) proficiency levels, only 20% of variance was accounted for.