An open access publication of the American Academy of Arts & Sciences
Summer 2023

Linguistic Profiling across International Geopolitical Landscapes

John Baugh
View PDF

Voice recognition lies at the heart of linguistic profiling, a discriminatory practice whereby goods, services, or opportunities that might otherwise be available are denied to someone, typically sight unseen, based on the sound of their voice. The technology that faithfully recreates one’s voice during phone conversations provides the basis on which nefarious, if not illegal, voice-derived discrimination occurs. These denials often go undetected because callers typically believe that the declination of their request for an apartment or a job or a loan is valid; that is, they do not necessarily assume that they were turned down because of negative stereotypes about their speech. I debunk a long-standing myth that exists among well-educated native speakers of the dominant language(s) in the countries where they live: namely, that such individuals speak without an accent. After dispelling this prevalent falsehood, I explore various forms of linguistic profiling throughout the world, culminating with observations intended to promote linguistic human rights and the aspirational goal of equality among people who do not share common sociolinguistic backgrounds.

John Baugh, a Fellow of the American Academy since 2021, is Professor of Psychology, Anthropology, Education, English, Linguistics, and African and African American Studies at Washington University in St. Louis. He is the author of Linguistics in the Pursuit of Justice (2018), Beyond Ebonics: Linguistic Pride and Racial Prejudice (2000), and Out of the Mouths of Slaves: African American Language and Educational Malpractice (1999).

The miracle of human speech is a double-edged sword that can be weaponized in situations when a person’s speech reveals demographic information that evokes negative prejudicial reactions.1 Alternatively, some positive benefits of linguistic profiling accrue on those occasions when people recognize speech belonging to someone who is well-known to them, or from a favored sociolinguistic background. These familiar voices tend to be viewed more positively when heard by listeners who share similar language traits. Family members who live together recognize each other’s voices, seldom needing any further identification, with rare exceptions, such as an adult child who sounds nearly identical to one of their parents. A combination of biological and sociological circumstances gives rise to these familial linguistic similarities; parents typically serve as the linguistic models for their children, whose vocal tracts often match that of their mother or father.

Every child who learns to speak has no memory of doing so. One of the most difficult of all human accomplishments—becoming a fluent speaker of a language—takes place so early in life that our long-term memory is not yet fully engaged.2 Thus, depending on the social circumstances under which a child learns their first language, they are unlikely to comprehend the relative status (or lack of it) when viewed within the sociolinguistic totality of the speech community in which they live. Those among us who experience pathological speech disorders know well that insensitive people often mock their manner of speaking, a painful personal experience known to anyone who has ever been told, “You talk funny.” As with perceptions of beauty, the belief that someone talks funny is relative to the ear of the beholder.3 What may sound funny to one person could easily be the source of discomfort and potential discrimination to another.

Children raised in circumstances in which their parents or caregivers are well-educated, fluent speakers of the dominant language(s), wherever they may be living, frequently come to believe that they speak without an accent.4 These self-perceptions of accent-free speech are always wrong, but are persistent and prevalent because the dominant groups hold the reins of political power and therefore set the linguistic standard(s) by which others are judged. Some nations, like France and Spain, have official languages that are protected by scholarly academies responsible for maintaining the linguistic purity of their beloved language. In other cases, as with English, there are no established academies but rather a set of socially dominant groups who unofficially establish norms of “correctness” or “standards.”

Language academies reinforce perceptions that some people may speak a language properly, without an accent, while others who are perceived to speak with an accent are often viewed as speaking that same language improperly.5 From a scientific linguistic point of view, notions of proper speech and correct grammar are misnomers, perpetuated by those who seek to control the inevitable tides of linguistic change that impact all languages worldwide. Much like the imperceptible movement of the earth’s plates, language change is also constant, and dimensions of that change are frequently undetectable, while others (for example, the creation of slang or new pronoun usage) stand out as might earthquake tremors that splinter the ground under our feet.

Before one can fully appreciate the consequences of linguistic profiling, it is vital to understand that language prejudice is relative and most impactful when standing on the shoulders of ill-founded fallacies of linguistic (and racial) superiority.6 Thus, the weaponization of language is most formidable when wielded by members of the well-educated elite who may not fully comprehend their prejudicial reactions to others whose language backgrounds are substantially different from their own or, worse, they may indeed be aware of their linguistic privilege and use it to their personal advantage.7

The first discoveries related to linguistic profiling were unearthed quite by accident through calls to inquire about prospective rental properties. The National Fair Housing Alliance regularly sent housing testers to view properties in person, frequently noting that minority housing testers routinely fared less well. These tests were expanded to include telephone calls, in which European Americans and members of American minority groups would call prospective landlords asking for appointments to visit rental properties, revealing a similar pattern of racial bias. Minority callers were denied access with far greater frequency than was the case for white callers. A group of social scientists became interested in evaluating these trends, and performed a series of experiments that confirmed the existence of linguistic profiling. The results proved that callers from different racial and linguistic backgrounds received (or didn’t receive) an appointment to view a property based exclusively on the sound of their voice.8

Ensuing experiments went further, demonstrating that some prospective landlords used answering machines to help screen calls. In those instances, the property managers never answered their phones; all calls initially went to voice mail. Upon listening to their messages, the property managers only returned calls to white callers. In a striking contrast, Black callers—who never had the opportunity to speak with anyone—did not receive a return call.9 The tactic of using an answering machine to help screen calls was presumably employed to offer the perpetrators of these crimes with a defense of never having met nor even spoken to the caller. How, then, might a plaintiff prove that a defendant landlord was guilty of racial discrimination when no direct personal contact had taken place? This deniability allowed landlords to use the tactic regularly throughout the United States.10

Evidence of linguistic profiling in housing markets is not confined to the United States, but it is always based on linguistic discrimination wherever it exists. In four German cities, callers who had Turkish names were less likely to be granted an appointment to view properties than were callers with Anglo-American names. Further, the group with American names, while treated somewhat better than callers with Turkish names, had less success than callers whose names were recognizably German and who were far more likely to be given an opportunity to view rental properties.11

Somewhat related to these examples of linguistic profiling, sociologists also explored differential access to homeowner’s insurance. Using quite similar methods to those used by fair-housing testers, the sociologists questioned whether minority homeowners might have equal access to insurance policies; alas, they do not. A nontrivial difference between renting a property and purchasing home insurance as a homeowner stems from the fact that the majority of home-insurance policies are discussed between an agent and client by phone prior to a policy being written and sent to the homeowner. This fact intersects with other historical evidence of linguistic profiling in this arena:

If race is not a factor or insurers cannot detect race over the phone, then in the initial telephone conversation there should be no association between the race of the caller and when the applicant is asked about the location of the home. But if the insurer can detect the race of the applicant and race (or racial composition of neighborhood) matters, a question about where the home is located would likely be posed earlier in the conversation with Black applicants than with Whites.12

Racially biased discrimination resulting from phone calls, while extremely problematic, represents only one form of linguistic profiling. Other manifestations of discrimination based on language affect different groups in various social circumstances. Language discrimination in the workplace, for example, can result in hostile work environments for speakers of nondominant languages in different parts of the world. In one well-documented case in the United States, an employer imposed harsh restrictions on any employees’ use of Spanish, or any language other than English, while at work, even when on break with fellow employees who shared fluency in another language.13

The employer argued that his efforts to confine all employee communication to English was beneficial because it would promote inclusivity among all employees since a significant number of workers were monolingual English speakers. Ensuing conversations that explored this rationale more thoroughly exposed substantial linguistic chauvinism on the part of the employer, who admitted that some monolingual English-speaking employees feared that their colleagues who did speak Spanish could employ it as a means of exclusionary, if not derisive, conversation. However, this fear was based entirely on speculation, and largely concerned conversations, such as in the lunchroom, that were entirely unconnected to their work.

Since private employers have tremendous latitude to dictate policies associated with their workplaces, Spanish speakers had no alternative but to comply with this demand, even when their private conversations had nothing whatsoever to do with the job. The employer seemed to be impervious to the fact that bilingual employees should be free to use whatever language they prefer during conversations with others who share their linguistic competence if that conversation is unrelated to their job or taking place at a time or location within the workplace when the conversation is completely dislocated from anything having to do with their assignments.14

All the examples of linguistic profiling that I have considered thus far reveal interlocking connections among people from diverse language backgrounds who share different roles and responsibilities. And in an institutional context, these dynamics often provide opportunity for a language or dialect to become weaponized in ways that may either break the law or deny a person of their civil rights, linguistic human rights, or both.

Some of the research on perceptual dialectology is highly informative in this regard.15 Every language variety evokes different reactions among speakers who do not share the same dialectal background for that language.16 The larger the linguistic footprint associated with any given language, the more likely it will be that differences of opinion prevail that expose a patchwork of perceptions that vary along numerous sociological and demographic dimensions, such as region, class, education, race, and religion, among other traits including sexual orientation or speech impediments. These traits may not only differentiate speakers of a given language, but do so in ways that provide alternative sociolinguistic perceptions of speech. Perceptions regarding those who employ local speech patterns are inevitably relative psychological constructs determined by the myriad of factors that individuals maintain from one region to another within every speech community. Moreover, the concept of a speech community, which is a basic construct of linguistic science, has evolved throughout human history as technological advances promoted increasingly rapid and distant travel, resulting in massive linguistic contact among people who were historically dislocated from one another slightly more than a century ago.

Advances in technology did more than promote language contact among people who spoke different languages, or different dialects within a single language. The invention of writing followed by its companion invention the printing press gave rise to increasing numbers of people who could read and write. The growing need for educated citizens throughout the world produced new mechanisms that served to offer the hope of greater social equality at the very same time that differential access to unequal educational opportunities continued to perpetuate the established social class order worldwide, regardless of the political orientation of the language in question, or the nation-states that used it.

Educational opportunities in England illustrate this point wonderfully, due in no small measure to the long-standing reign of its royal family and the array of educational and religious institutions that have evolved there. At the height of the British Empire—that is, when the sun never set on lands that had been colonized by Britain through a combination of military might and increasing global trade—access to educational opportunities was determined in large measure by virtue of a person’s social status at birth. Oxford University and Cambridge University were not available to the majority of English citizens who lived in different locations, with an eclectic mixture of languages and dialects that still linger on the tongues of regional descendants of their bygone ancestors.

Clearly, the English were not alone as far as matters of colonial and linguistic expansion were concerned. Holland, Spain, France, and Portugal, among other European nations, have left their indelible linguistic imprint on distant lands that echo aspects of the linguistic expansionism that became an inevitable artifact of the subjugation of non-Europeans, either through enslavement or other forms of social dislocation that dislodged Indigenous peoples and their languages in deference to political domination imposed from afar. These historical facts are common precursors to newer educational ventures in countries once inhabited by European colonizers. Fluency in the language of one’s oppressor became an ironic means to personal betterment throughout the world, often resulting in the suppression and diminishment of Indigenous languages or, worse, their eventual demise. It is against this neocolonial backdrop that history has witnessed another form of linguistic profiling that has been perpetuated, either indirectly or by design, in schools throughout the world. In England, where uneducated masses living in poverty were unable to avail themselves of educational opportunities, elite academies eventually gave way to a burgeoning educational system that exposed new forms of differential access to educational opportunities whose quality was fundamentally determined by one’s wealth or lack of it. Again, the English are not unique when it comes to allocations of educational opportunities based on wealth, which is pervasive worldwide. However, the longevity of unequal educational opportunities in England stands out because of the expansiveness of their former empire, procured at a time when England’s naval might was the primary determinant of its global power.

The United States, along with many other former British colonies, created schools that replicated models of economically driven, differential access to education. In fact, educational and linguistic disparities in the United States have been exacerbated by long-standing decentralized policies. Each state has the authority to regulate public education within its jurisdiction, while school funding within states is differentiated largely by local property values, resulting in a disjoined national education system that varies widely in content, resources, and quality. While it may be understandable that each state devotes a portion of its curriculum to historical state-centric studies, there are also different approaches to the teaching of various subjects, including language-related subjects, be they related to English, other languages, or how best to educate children who are profoundly deaf or who experience pathological speech disorders. Therein, I find fertile ground for sowing the seeds of uninformed linguistic profiling, based variously on misguided perceptions of linguistic elitism, authority, and superiority as means through which less influential speakers are castigated or treated in other discriminatory ways.17

Language attitudes alone do not account for many of these educational disparities. Some states maintain an ethos of equal educational opportunity by assigning identical textbooks for all students throughout the state, regardless of their linguistic background. The underlying assumption is that if students are required to adopt the same textbook, then they share equal access to the same pedagogical content. However, informed educational linguistic scholarship has shown that students from different language or dialect backgrounds may benefit from pedagogy that is modified to account for their unique cultural and linguistic backgrounds.18

Keeping in mind that most educational systems throughout the world are designed to maintain the political status quo, cultural and linguistic modifications to education were once employed during apartheid in South Africa under the guise that children can learn best when doing so in their native language. While the principle of supporting mother-tongue education has clear benefits, it is also important to fully understand the sociopolitical circumstances under which such policies operate, as well as their far-reaching consequences for the students who attend schools that do not share a common language. In 1953, South Africa’s apartheid government acted upon the Eiselen Commission Report, produced in 1951, which encouraged the government to take charge of the education of Black South Africans as a way to control the socioeconomic development and, by extension, the political future of the country.19

The Bantu Education Act was designed to ensure that Black South Africans would not have direct or sustained access to the same educational opportunities that were offered to the minority-ruling white South Africans, who were either native speakers of Afrikaans or English, the two South African languages that received official governmental recognition prior to the fall of apartheid.20 As was the case for nearly every institution within South Africa, the Bantu Education Act was designed to help maintain racial segregation while simultaneously making sure only white South Africans had access to the languages of power and political influence. Those policies were dramatically transformed after apartheid ended. Under President Nelson Mandela’s leadership, South Africa adopted a new national language policy with eleven official languages, taking care to still include Afrikaans and English in the hope that doing so might increase the likelihood of racial healing, bolstered by the new, more inclusive recognition of nine additional Indigenous languages that were native to South Africa long before Afrikaans or English was spoken there.21

These educational exemplars from nations where overt racial segregation was once the law of the land serve as a stark reminder that government policies under­write many of the cultural and linguistic discrepancies that create and maintain racial segregation, perpetuating distinctive Black language usage, as well as widespread discrimination against Black people for the way they speak. Although instances of linguistic profiling against Black people differ from country to country based on the specific historical sociolinguistic circumstances of the nation in question, Black linguistic equality remains elusive at best anywhere in the world. The combination of policies and prejudices that can be traced to colonization and the denial of human rights that resulted from the African slave trade have only exacerbated these trends while also creating a climate in which perceptions of well-spoken Black people are considered to be those who have mastered the languages and/or dialects of their (former?) oppressors.22

The various forms of linguistic profiling I have described thus far can be evaluated and examined in a variety of ways, including descriptive techniques, survey research, and experimental studies. The latter method was employed quite recently regarding potential bias against Uyghur speakers of Mandarin, or Putonghua, as a second language.23 Carefully designed experiments that used name-based priming examined how listeners thought they were hearing either a Korean, Uyghur, or a nondescript person with a Chinese surname, all of whom were portrayed as second-language (L2) Mandarin speakers. By employing the classical matched-guise task design, in which participants listen to multiple speakers and assess them based on various characteristics, the researchers explored alternative reactions to the same speech, albeit associated with different ethnic surnames. The study in question contemplated the relative employability of an L2 Mandarin speaker and did so by comparing three experimental conditions: the first condition had no social priming; the second condition contrasted speakers with Chinese or Korean surnames (written in Chinese characters); and the third and final condition introduced these voices as belonging to either a Chinese or Uyghur surname (also written in Chinese characters).24

The results were significant, showing that the speech being primed as belonging to a Uyghur surname was perceived to be that of someone who was hardworking, but who was also deemed as less likely to be hired compared with the nondescript L2 Mandarin speaker depicted with a Chinese surname. Of considerable importance here, this matched-guise test revealed linguistic profiling based on surnames. The bias was not the product of differences in speech styles, which were controlled and held constant. As such, we now have learned that some forms of linguistic profiling exceed actual differences in linguistic behavior and can merely be triggered by the belief that a person is a speaker from a devalued group.25

Nearly two decades ago, when studies of linguistic profiling first began, building on the experimental foundations of matched-guise tests and perceptual dialectology, every effort was made to ponder how best to help speakers belonging to marginalized groups gain more fair and equitable access to housing, employment, education, justice, and medical care in speech communities where bias against nondominant linguistic groups had been documented.26 With the passage of time, we have come to recognize that the relief needed to advance linguistic human rights and increase access to equal opportunities throughout the world may also rely on the goodwill of those who are in positions of political influence and power within their respective societies. Faced with global evidence of linguistic prejudice that varies from one country to another, it is imperative that greater linguistic benevolence be bestowed on those who are often powerless to detect or challenge when their voice—or profound deafness and the use of sign language—triggers unwelcome, if not illegal, reactions that restrict their access to opportunities routinely afforded to anyone perceived as speaking without an accent.