Hearing skin color: The connections between language and race

What is race? What does it mean to be White, Black, Asian, Latina/o, or any other identifiable race? Most of us probably think of it as something marked on our bodies. It’s the pigment of our skin, the shape of our eyes, the size of our noses or brows. Of course, we’re not wrong about these ideas. It’s a well-documented fact that people do assign others a category of race relying on features like skin color and facial features (see, for example, this study).

However, physical features are not the whole story. We use other features to racially categorize people. As a linguist, I’m particularly interested in how good we are at using language as a way of assigning race to others.

One of the best ways to show that we use more than physical characteristics to assign race to people is to show that people can sense others’ races through language in situations where physical features are not available to them.

The first context where we can observe that people hear race comes from experiments where participants are exposed to recordings of different voices and asked to identify the speaker according to race. It is not terribly surprising that research shows people can do this quite accurately when we hear a sentence or two from a speaker. What is perhaps surprising though is how little input people need in order to accurately judge the race of a speaker. In one experiment, participants were exposed to only the word “hello” and asked to decide what the speaker’s race was. Participants were surprisingly accurate at judging the race of the speaker based only on the word “hello”. The green cells in the tables below show when participants correctly guessed the speaker’s identity. Listeners were particularly good at identifying the Chicano English and ‘standard’ (White) English voices from the word “hello”, although a sizable portion were unable to classify the African-American English (AAE) speaker.


But that’s not all. Another study took this a step further. The researchers had 50 Black men and 50 White men record a single vowel. Listeners heard two one-second vowels. They were told one vowel was from a White man. The other one was from a Black man. The listeners had to decide which was which. On average, they could do it about 60% of the time (one participant was able to identify them 72% of the time). This finding is impressive when we consider that listeners only heard a one-second long vowel. These studies suggest we can make reasonably accurate judgments about people’s races based on their voices extremely rapidly after a single word or maybe only a single sound.

Another compelling context where we can observe people making judgments about race without visual cues comes from research on whether and how blind people perceive and experience race. Because we know that people are generally capable of classifying people into races based on voices, it is not surprising that blind people are aware that race exists. However, because they cannot see race we might believe that it is a different and less significant aspect of their perceptions of other people.

However, research by Osagie K. Obasogie suggests that this is a misconception (check out this interview, in which Obasogie explains his research). In fact, based on his interviews with the blind, Obasogie believes that blind people not only have “as significant an understanding of race as anyone else” but also that blind people have a conception of what race is that is very much in line with how sighted people experience race visually. For example, one White, blind respondent, from the study, Jens, expressed his perception of Black people, “Most black people look pretty much the same with a few exceptions. Of course it always depends on the person, but in general, they look pretty much the same I think” (p. 596). What is interesting is how, despite being unable to see, Jens has been socialized into the ways we tend to see physical characteristics as more diverse within our own race (this is sometimes known as own race bias and has a long history of social psychological research). What is interesting then is that even in the absence of the visual cues that mark race for sighted people, through relying on experiences derived from their other senses, blind people are able to experience race in a manner that is strikingly similar to sighted people.

The research I’ve discussed above suggests that people’s physical characteristics are simply one route we can take to arrive at a judgment of people’s race. Indeed it seems especially likely that language allows us to assign people into different races. One interesting implication of this is that we experience non-human characters in media as members of different races. Rosina Lippi-Green offers an excellent case study of this in her book English with An Accent by closely examining the characters in Disney animated films. Lippi-Green concludes that Disney casts voice actors with linguistic aspects that mark them as a particular race in order to draw on existing stereotypes or associations, whether negative or positive, about racial groups. This is particularly clear when we look at how non-White voices are sometimes used in Disney films.

The Lion King (1994) offers a clear example of how non-human characters can be placed in racial categories on the basis of language and voice in order to draw on audiences’ stereotypes. Fairly clear lines are drawn using language between the positive lion characters (for example, Simba and Nala) and the villains (Scar and the hyenas). In particular, the positive characters use ‘standard’ US English (even in cases where they are voiced by African Americans like James Earl Jones). In contrast, we find more linguistic diversity in the voices of the negative characters. Scar, for example, speaks with a clearly British accent.

Of greatest interest to the issue of hearing race are the hyenas Shenzi and Banzai. Shenzi, voiced by Whoopi Goldberg, speaks African-American English. Banzai, voiced by Cheech Marin, frequently uses Latino-accented English, including occasional code-switching into Spanish (e.g., “¿qué pasa?”). It appears then that the Lion King enlists non-White voices for these characters to accentuate their similarities to tough, inner-city racial minorities and to tap into audiences’ fears of street crime and gang activity (I looked at the common association between crime and racial minorities in a previous post).

Lippi-Green interprets the general message of the separation in language as follows “The message is a familiar one: AAVE [African-American Vernacular English] speakers occupy the dark and frightening places, where Simba does not belong and should not be; he belongs on the sunny savannah where SAE [Standard American English] speakers like his father live” (Lippi-Green, 2011, p. 122). It is particularly interesting how in this case in terms of physical features the voice actor (James Earl Jones) playing Simba’s father is African American. However, in the context of the movie with visual information about Jones missing, the character he gives voice to, Mufasa, is less likely to be perceived in this racial category than the AAE-speaking Shenzi is. The lack of clear connection then between language and physical features highlights the arbitrariness of racial categories.


Returning to the question of what race is, I believe the research I talked about above lends important insights to it. Our ability to place others in racial categories using non-visual information like spoken language suggests that what we might think of as race (skin color, eye shape, and other physical features) are not its sole or even its essential properties. It also underscores the racist nature of criticisms of African-American English (AAE) such as those expressed about Rachel Jeantel during her testimony in the George Zimmerman trial. Despite the fact that the languages we speak are not determined by genetics, our ability to racially categorize people on the basis of their language suggests that what critics of AAE are really airing are not aesthetic preferences for a particular language variety but preferences for a particular racial category.

Language’s ability to signal race can and has been manipulated to exploit racial stereotypes as can be seen from The Lion King example. This suggests we should be wary of what voices are used for what purposes in much the same way many are sensitive to the use of racial stereotypes in advertising (but see here for more subtle uses of racial stereotypes in advertising that we may not notice). We also need to be aware of the potential for linguistic cues about race to be used in areas like housing discrimination. The race we hear when we hear voices may be a particularly insidious way to trigger our implicit racist biases because it avoids aspects of race that we are more consciously aware of like skin color. Hence, an awareness of how we respond implicitly to language is important for overcoming racial discrimination.

Tagged with: , , , , , , , , , , , ,
Posted in Human migration, Ideology and social change, Language and race, Linguistic diversity, Prescriptivism and language prejudice
3 comments on “Hearing skin color: The connections between language and race
  1. Great post! The Lion King example was fascinating. I’m definitely going to pay more attention to how characters are being voiced in the future. Thanks for all the insight.


  2. […] groups who have been historically oppressed or denied access to power (Disney movies are complicit in this too). But that is a topic for a whole book… And indeed, there are some great books about it, […]

  3. Theresa says:

    Very nice blog youu have here

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

Enter your email address to follow this blog and receive notifications of new posts by email.

Join 465 other subscribers
Follow linguistic pulse on Twitter
%d bloggers like this: