Understanding how people detect social class from speech requires taking a cultural psychological perspective

In “Evidence for the reproduction of social class in brief speech,” Kraus et al. (1) present a series of 5 compelling studies showing that perceivers can detect the social class of speakers at above-chance rates. They further demonstrate that perceivers infer the social class of speakers by comparing their speech to “ideal speech standards.” Although perceivers were able to detect targets’ social class in either spoken or written text, this detection was most likely to occur for spoken text. This suggests that, over and above content, some cues specific to verbal transmission (e.g., pronunciation or accent) signal social class. In addition, perceivers not only infer social class from little information, but they also use this categorization to make judgments about a potential job candidate’s fit and competence.

Going beyond prior work suggesting that perceivers can detect social class (2, 3), the current studies make 3 important theoretical contributions. First, the authors provide evidence that one informational mechanism for detecting social class from speech is comparing that speech to ideal speech standards conveyed by educational and societal norms (1). Second, the current studies document that social class detection does not require an actual social interaction, or even a conversation between 2 people, and can be detected in as few as 7 words. Third, this research links social class detection to its consequences for social class stereotypes about fit and competence, and to downstream outcomes such as hiring (4).

Kraus et al.’s (1) work is practically important because it suggests that directly communicating specific social class cues in social interactions or interviews (e.g., sailing or a first-generation student group; ref. 3) is not necessary to detect social class. Their findings further suggest that organizational attempts to blind resumes (e.g., by removing informational cues of social class) are unlikely to conceal it fully. This process of social class detection—and associated stereotypes about fit and competence—may be one reason why it is so hard to increase social class diversity in educational and workplace settings.

In this commentary, we adopt a cultural psychological perspective to make sense of the findings and their implications. To fully understand how people detect social class from speech, we suggest that it is critical to consider the impact of the cultural context of the researchers, as well as the cultural or social class backgrounds of the targets and perceivers in the social class detection paradigm.

The Researchers’ Independent Cultural Context

First, it is essential to consider how the highly independent cultural context of the United States may shape the authors’ and other scholars’ interpretation of and reaction to the findings (5). Study 1 shows that perceivers are able to detect the social class (i.e., guess the educational attainment) of speakers 55% of the time (1). The authors refer to 55% as accurate and suggest that this finding is surprising. We believe that this perspective would be widely shared by researchers in the United States and that it is a culture-specific interpretation. In the United States, powerful cultural narratives, such as meritocracy, the American Dream, and pulling oneself up by one’s bootstraps, suggest that social class background does not influence people’s experiences or outcomes in life (6, 7). The idea that social class does not matter would imply that social class does not exist, is invisible, or, at the very least, would be difficult to detect. The prevalence of these narratives and how they shape Americans’ understanding of social class can help to explain why a number that is slightly above chance would be framed as surprising and an indication of accuracy. To illustrate the point, if perceivers were able to accurately guess gender or race only 55% of the time, we would be surprised about their inaccuracy. Likewise, in a cultural context where social class is more strongly institutionalized and accepted as a meaningful social category (e.g., India), researchers might also interpret a 55% rate of detection as surprisingly inaccurate. Of course, considering how scholars’ cultural contexts shape the framing of results does not imply the findings are any less important. On the contrary, considering how researchers from other cultural contexts may frame the results suggests additional implications of the present work.

The Perceivers’ and Speakers’ Cultural Backgrounds

Second, we suggest that it is important to consider how various intersecting cultural factors shape how perceivers interpret the social class of others. We propose that the process of social class detection hinges on not only what content is communicated in the speech but also how culture shapes the process of making sense of that information (8). A range of cultural factors likely influence this process, such as the social class of the perceiver and target, as well as the mainstream and local cultural standards that people may use to evaluate speech. Below, we suggest some specific questions that future research should consider to more deeply explore how culture shapes this process of social class detection.

How Does Social Class Shape Accuracy?

One important question is how the perceiver and target social classes shape accuracy in the detection of a target’s social class. We could conceptualize accuracy in terms of the target’s social class (that is, Is it easier to detect higher vs. lower social class targets?), the perceiver’s social class (that is, Are higher versus lower social class individuals better at detecting social class?), or their interaction (that is, Are higher versus lower social class individuals better at detecting the social class of higher- versus lower-class targets?). Future research should examine the systematic patterns involved in the judgments that perceivers make, and whether and how those judgments are informed by perceiver and target social classes.

Examining these patterns of accuracy may help us not only to better understand the process of social class detection but also to reveal how people learn or improve their detection over time and the purpose that this detection serves. For example, if future research were to find that higher (vs. lower) social class individuals are more accurate in detecting others’ social class, we might then examine the reason for this, which would give us insight into the function of accurate social class detection. One possibility is that higher social class individuals learn the codes or rules to detect social class through an educational system that teaches them the “right” or ideal standard for how to speak. Higher social class individuals might become especially aware of and attuned to the signals of social class, given their frequent role as institutional gatekeepers, who often decide whom to admit or offer opportunities.

Do People Use a Local or Mainstream Standard for Speech?

A second important question is how people’s own cultural backgrounds shape the standard they use to detect social class in speech. In particular, when perceivers process speech, how does their own social class—and accompanying cultural norms for speech—shape how they interpret social class? One possibility is that everyone—irrespective of their own social class—interprets the same speech cues as indicators of high versus low social class. For example, one might compare speech to the ideal standard and ask, “Does she pronounce that word in the right way?”

In “Evidence for the reproduction of social class in brief speech,” Kraus et al. present a series of 5 compelling studies showing that perceivers can detect the social class of speakers at above-chance rates.

The authors do indeed find some evidence that comparison to this ideal standard helps to explain how people infer social class from speech. Evidence supporting the ideal standard might imply that higher-class people have the power to institutionally promote their own standards for ideal speech more broadly through a shared educational system (9, 10).

Another complementary possibility is that perceivers might also evaluate the degree to which a speaker adheres to their local social class norms. For example, people might judge the correctness of speech based on their own local experience and what sounds “right” or normative in their own context. The authors do not explore this possibility in their study (1), but it is a process that researchers could explore in the future. If people were to use their own local standards, this might imply that parents and teachers in local social class contexts play an important role in defining what is appropriate and normative in that particular context (10). That is, people may have their own cultural codes that they consider and value as normative, appropriate, and adaptive in their local social class context. Future research should consider the relative role of each of these processes and disentangle when people might evaluate speech by comparing it to their own speech (a local standard), or by comparing it to a broader, mainstream standard that is shared across social classes (a mainstream standard).

In sum, the current research represents an important first step toward understanding how people detect the social class of speakers. Taking a cultural psychological perspective, we suggest that it is critical to consider the impact of cultural context in shaping how scholars interpret the accuracy or inaccuracy of social class detection. We further suggest that future research should explore a wide range of cultural factors that likely inform how people interpret and respond to others’ speech. Beyond the 2 questions posed above, research might also explore how social class intersections with race, gender, and other social group memberships might impact the process of social class detection. Extending and deepening our understanding of social class detection will help us to better understand how American society reproduces inequality, as well as how we can begin to reduce it.