Can we tell people’s social class just by listening to them?

Popular wisdom tells us not to judge a book by its cover. But when we hear someone –on the phone, out in the hallway, or sitting behind you on the bus – we immediately get an impression of the person: male or female, gay or straight, young or old, upper or lower class. Willingly or not, these impressions may be translated into inferences and judgments, from our part, upon the speaker: How loud! How clever! How rude!

Does the way we speak, such as the way we pronounce our Rs or certain vowels, contribute to these overall impressions, in addition to the content of whatever is being said? Is it possible to tell someone’s social class, sexual orientation, place of origin just by listening to them?

Linguists have been interested in learning on what kinds of linguistic cues we base our inferences and how they come to be attached to certain social meanings and stereotypes.

Not getting an apartment

A study in Montreal in the 1960s showed that speakers of both French and English in this bilingual community had consistent unconscious reactions to anglophones and francophones. A few fluent bilingual speakers were recorded reading a passage in both languages, and the recordings were later on submitted to multiple “judges” who listened to the audioclips and evaluated the speakers on a number of personal traits, but without knowing that they listened to each speaker twice. Their judgments differed according to the language in the recordings: English was associated with being confident, ambitious, intelligent, and good-looking, while French was associated with being religious and kind.

English was associated with being confident, ambitious, intelligent, and good-looking

In California, a group of researchers conducted a series of phone surveys to tell whether dialect identification is possible based solely on the voice. In one experiment, a researcher called landlords with available property for rent and asked to set an appointment to visit the location; each landlord was called three times, each one in a different dialect, in a randomized order: standard, Chicano or African-American English. Different voice guises achieved different rates of success in setting up an appointment for different city areas, which shows that prospective tenants may be discriminated against on the basis of their voices

The listeners attributed different areas of residence and social status depending on the R pronunciation in the recording: working-class neighborhoods and lower social status when they listened to the “redneck-r”.

In a recent study in São Paulo, I investigated which social meanings are attached to two different pronunciations of R at the end of a syllable – the so-called “redneck-r” and “Paulistano-r” –, apart from stereotypes related to speakers’ geographical origins. The listeners, again not knowing that they were rating the same speakers, attributed different areas of residence and social status depending on the R pronunciation in the recording: working-class neighborhoods and lower social status when they listened to the “redneck-r”. They also attributed both positive and negative personal characteristics such as “sophisticated” and “snobbish” when a speaker was listened to with “Paulistano-r”, and “simple” and “hardworking” when using “redneck-r”.

But the results also showed that listeners rate the speakers differently according to their own social characteristics; for example, dwellers of working-class areas don’t differentiate R pronunciations as drastically as those who live in upper-class areas, and migrants consider “redneck-r” just as Paulistano as “Paulistano-r”. Our own social experiences are undoubtedly at the basis of these judgments.

How do we know (or think we know)?

Social perceptions vary considerably among individuals and social groups, but they are also highly structured. Social and personal traits tend to cluster in socially-related attributes, such as “Valley girl-annoying-snobbish-spoiled”, “religious-close to family”, “hardworking-trustworthy-solidary”, and “redneck-accented-working class”.

Clearly being upper-class is not a synonym of being highly educated, nor is receiving formal education a synonym of being smart. But in people’s social experience, most upper-class individuals they’ve met hold a college degree, or they may consider someone unintelligent for not mastering “everyone-has-learned-it-at-school” type of knowledge.

It seems that when one trait come to be associated with a specific language feature, the whole cluster of attributes is open for the speaker-listeners to make inferences. It remains to be investigated, however, where this network of inferences stop: in São Paulo, although R pronunciation is perceptually related to speakers’ level of education, it does not spread to perceptions of speakers’ intelligence, even though other studies have shown a close correlation between perceptions of education and intelligence – as is the case, for instance, of perceptions on standard and nonstandard nominal agreement in São Paulo.

When one trait come to be associated with a specific language feature, the whole cluster of attributes is open for the speaker-listeners to make inferences. It remains to be investigated, however, where this network of inferences stop.

While making correct judgments about people’s social and personal traits isn’t 100% accurate, we do make inferences about individuals we meet in our everyday lives, partially based on the way they speak. These inferences are quite automatic, even when taking the utmost care not to fall into pre-judgments. Mapping associations between linguistic traits, social meanings and listeners’ characteristics may help understand the workings of social prejudice.