Do accents—the way that language is pronounced—shape social and economic interactions? We answer this question using an experiment embedded in an online survey of 6,000 Colombian adults. Respondents evaluated paired profiles in which audio introductions were randomly assigned to feature either a high- or low-class accent, while income, education, and other attributes were independently randomized. We find a sizable accent premium: speakers with high-class accents are 5–16 percentage points more likely to be chosen as friends, business partners, colleagues, or bosses. This premium is significantly larger among respondents with high socioeconomic status, consistent with an in-group favoritism capable of reproducing inequality. By varying the information we present to respondents, our experiment allows us to conclude that the premium cannot be attributed solely to inferences about income or education. We further show that the premium vanishes for high-class foreign accents, suggesting that class cues are culturally specific and difficult for outsiders to detect. Finally, we document that respondents systematically associate high-class accents with multiple proxies of social status and that they elicit more deferential treatment. Overall, our findings reveal that accents function as a form of capital: culturally specific linguistic signals that reproduce social hierarchies, with implications for labor markets and efforts to promote mobility and integration.
