Working-class people are severely under-represented in elected office. Yet experimental and other studies reveal little sign that voters prefer politicians from upper and middle classes -- indeed, often the preference is for those from working-class backgrounds and occupations (Carnes & Lupu, 2016; Elsässer, 2024; Kevins, 2021; Vivyan et al., 2020). One route to reconciling these findings is to suggest that voters' biases operate not via static features of a candidate's CV, like occupation, but instead through other 'living' markers of class such as accent. The latter is well established as a source of prejudice and stereotyping in the highly class-conscious context of the UK.

In this paper, we report a 'verbal guise' or aural vignette experiment in which the same political text is spoken by actors in five common English accents. Respondents were randomly assigned to one of the accents and then asked a number of questions about the candidate that they had just heard. We address two main research questions: first, are accents a cue for class? second, do they have consequences for candidate (trait) evaluations? The answer in both cases is 'yes'. Strikingly, most of the accent effect is direct, i.e. not mediated by class. In either case, though, accent popularity is *negatively* associated with social class , with a particular rejection of the politician with the most upper-middle-class accent. On this trial, at least, we reject the hypothesis that working-class candidates are handicapped by the way they talk.