Given the important consequences of oral language proficiency tests, it is important to ensure the reliability of the scoring by inter-raters. However, this measure doesn’t identify biases shared by a group of judges, for example, collective bias triggered by stereotypes about the ethnicity of the speaker. Three foreign students (Syrian, Taiwanese, and Brazilian) were recorded reading a text with nine grammatical errors. The recorded texts were assessed by 343 native French speaking university students who were asked to note the errors and overall proficiency in French language. Judges who thought that the first speaker was ‘Arabic’ judged her differently than those who attributed a different origin to her. Based on the perceived ethnic affiliation, stereotypes and social representation could thus be triggered and affect the judgments of foreign language proficiency by untrained judges.