Psych Safety, by Iterum

Evidence for the Reproduction of Social Class in Brief Speech

Kraus et al (2019) · Class read from speech

Kraus et al · Primary Research · Power & Equity, Voice & Silence, Culture & Context · Proceedings of the National Academy of Sciences · 2019 · Open access

Open this neighbourhood on the live map →

Kraus, Torrez, Park and Ghayebi examine how a person's social class is accurately perceived and reproduced by mundane patterns in brief speech. In five US studies, strangers judged class above chance from brief speech, in study 1 from seven words, and how people spoke carried more class signal than a transcript of what they said; people with hiring experience who heard short pre-interview recordings rated higher-class candidates more favourably.

Studies 1 to 4 used speakers from the International Dialects of English Archive. In study 1, 229 perceivers heard seven words spoken out of context by each of 27 speakers and judged college graduation correctly 55.49 per cent of the time against 50 per cent chance, less accurately than race, age or gender. In study 2, controlling for age, gender, race and pitch, higher-class speakers deviated less from the digital speech standards of sites such as Google and Amazon, as did speakers whom 568 other listeners rated higher in class. In study 3, 200 perceivers judged higher-class speakers closer to their own ideal of English. In study 4, 302 perceivers heard or read 90 seconds of self-description: perceived and actual class correlated at 0.21 for speech but at minus 0.22 for transcripts, a reversal they call unexpected.

In study 5, 274 people with hiring experience, recruited through Prolific and screened by two questions, heard some and read transcripts of others among 20 candidates from the New Haven community, chosen from 110 for the widest class disparity, answer 'How would you describe yourself?' with explicit mentions of class removed. With no information on qualifications, they judged higher-class candidates more competent, a better fit, more similar to themselves and more hireable (D of 0.62 to 0.69), and offered them higher starting salaries and signing bonuses. Perceived class favoured the higher-class candidates in both formats, more so in speech, and the hiring analyses use the speech condition only; the main text reports no hiring ratings for transcripts, so it does not show that the voice, rather than the words, drove them. They are ratings on set scales (salary between 55,000 and 65,000 dollars), not live hiring decisions.

They stress that accuracy here means any minimal tendency above chance, well short of error-free perception, and argue that it was enough in study 5 to bias hiring judgements. They leave open whether class cues inform judgements of competence and fit or the reverse, offer cultural matching as a possible explanation from exploratory analyses, suggest that class cues may carry racial inequality, given the overlap of class and race in the US, or the reverse, and call for work on class cues independent of and alongside racial stereotypes. All participants were US residents.

The paper does not address psychological safety; the bridge is ours. Read beside Levon and colleagues' accent study, which found a smaller bias confined to some listeners, we take it as evidence that how someone sounds can count for or against them before what they say is weighed, so the risk of speaking up may differ by class from the first words.

Kraus, M.W., Torrez, B., Park, J.W. and Ghayebi, F. (2019) 'Evidence for the reproduction of social class in brief speech', Proceedings of the National Academy of Sciences, 116(46), pp. 22998-23003.

Save this paper to a bibliography, keep notes on it, and export citations. Create a free account →

Related psychsafety.com articles (5)

Related papers (4)

View the source paper →