What did we find?
We examined public attitudes to both accent ‘labels’ - familiar names and categories of accents - and people’s responses to actual recorded voices. We then looked more specifically at the effects of accent bias in a professional setting. Our work led to three key findings.
What did our results show?
First, accent bias exists. There is a very established and enduring ‘hierarchy of accents’ in the UK. Second, this bias is sensitive to context – people’s responses are more nuanced when listening to actual speech. And lastly, that, with increased awareness, people can almost fully suppress this bias under certain conditions – for example in a professional recruiting context.
Attitudes to labels - an enduring hierarchy of accents
While our work suggests that there is some reduction in accent bias – the distances between the highest and lowest rated accents are smaller in our study than in earlier studies – our results show a persistent hierarchy of accents: one that penalises non-standard working-class and ethnic accents and upholds the belief that national standard varieties are the most prestigious.
Attitudes to voices - bias is affected by context
Overall, the results of our work with audio recordings show a similar accent hierarchy to that observed with accent labels survey, and the two surveys together provide strong evidence of systematic bias against certain accents in England and, conversely, of a general belief that Received Pronunciation (RP) is most suitable for professional employment.
However, the surveys also indicate that respondents are sensitive to context: we see less bias overall in the audio recordings survey, where patterns are somewhat weaker and are more specific to particular groups of respondents. This suggests that while accent bias exists, effects may be attenuated when evaluating an actual speaker in a
real social context.
Attitudes at work - ‘switching off’ bias to focus on expertise
Our work with lawyers showed that when asked to assess candidates in a controlled setting, with the ability to listen carefully to the content of responses and perhaps with heightened awareness of a need to be fair, legal professionals have the ability to switch off biases and attend very well to the quality of content.
However, our study does not model accent bias throughout the profession, only a simulation of a small part of trainee hiring. The study does not give us a full picture of the potential for accent bias in many other aspects of professional life. The 2015 Social Mobility Commission report points to many other aspects of ‘cultural fit’ in hiring decisions. We therefore cannot conclude that speakers of more ‘standard’ accents would not ultimately still have a better chance at being hired.
The present study has shown that accent bias is pervasive but, under certain conditions, people in positions of power have the capacity to resist this effect.
In other words, we are not slaves to our ‘natural’ or automatic responses: our ideologies are more boldly on display when asked to react to disembodied labels, less so when listening to real voices, and even less so when acting in a professional capacity and under observation.
Further work/next steps
The field of Psychology has shown that bias – systematic deviations in perception or judgement – is a fundamental property of human cognition. It is not only present in those who are prejudiced, it arises in all humans due to a need to quickly process complex information as well as for emotionally-based motivations (Tversky and Kahneman 1974; Bless, Fielder, and Strack 2004). Belief formation and human behaviour is extensively influenced by a wide range of such biases.
Bias becomes discrimination when we allow these biases, or cultural baggage, to govern our judgement of unrelated traits such as intelligence or competence.
Our work offers hope that people may be quite flexible in their ability to prevent their automatic biases from affecting their judgement. One goal of unconscious bias training can therefore be not the complete elimination of long-held, and often unconscious, preferences but the management of their impact on judgement.
In continuing work we are developing packages of interactive materials that showcase these themes in a practical way for students, professionals, and HR departments. We are also testing a range of training interventions to understand how best to increase awareness of language bias and fair access.