Michael Erard on the latest form of racial profiling: Can you be discriminated against because of the way you speak?
In April 2001, a Californian named James Johnson began to suspect that the owners of an apartment he wanted to rent in the San Francisco Bay Area were ignoring his phone calls because he was African-American. Johnson hadn’t met the landlords in person or told them about his race. But he had spoken to them by telephone, and he assumed that they could identify his race from the way he spoke. So he sued, alleging racial discrimination.
In Johnson v. Jensen, which will be heard in upcoming months in the U.S. District Court for Northern California, the defendants will likely argue that they had no idea Johnson was black. Having only heard Johnson’s voice, they couldn’t have discriminated against him on the basis of his race. Race, after all, is something one sees, not hears.
Or is it? John Baugh, an African-American linguist at Stanford University, is on Johnson’s side. In Johnson v. Jensen, he will provide expert testimony to support the claim that people can often identify a speaker’s race from speech alone. In recent years, Baugh has emerged as one of the leading experts on the intersection of race and language, largely as a result of his research into this sort of “linguistic profiling,” a term he coined.
Johnson’s case offers a potential breakthrough in discrimination law. No judge has yet recognized linguistic profiling as a distinct form of racial discrimination, and many lawyers feel that an official ruling would have a significant effect on future litigation. Shanna L. Smith, the president of the nonprofit National Fair Housing Alliance, predicts that if the concept of linguistic profiling is vindicated in court, the number of housing complaints will rise significantly.
For skeptics, this is not promising news. John McWhorter, an African-American linguist at the University of California at Berkeley and the author of Losing the Race: Self-Sabotage in Black America, argues that linguistic profiling is a phenomenon largely dreamed up by civil rights crusaders desperate for fresh battlefields. “We’re at a point in this country where great progress is being made in race relations,” he says. “There’s still racism, but naked discrimination is a thing of the past. So if you’re going to look for racism, it’s going to get more and more subtle.” He worries that Baugh is exaggerating the problem: “No one before 1964 was running around complaining about linguistic profiling, because it was a marginal issue. It still is.”
In 2002, however, James Johnson and others think there’s something to complain about. The practical question is: Can they prove it in court?
At a linguistics conference last April at the University of Texas at Austin, Baugh spoke about the topic of linguistic profiling. In the course of his talk, without warning, he switched seamlessly from the Standard English dialect he had been speaking to Black English and then to Chicano English. The scholarly audience, accustomed to his “professional voice,” burst out laughing. Baugh learned to speak all three dialects fluently while growing up in the inner cities of Philadelphia and Los Angeles, where he was raised by well-educated parents. In his first book, Black Street Speech (1983), he recalls hearing his mother talk on the telephone: “As I grew older,” he writes, “it became fairly easy to tell if Mom was talking to someone black or someone white, based on her speech alone.”
Baugh became seriously interested in this phenomenon in 1987, when he was 38 and looking for a place to live in an affluent suburban section of Palo Alto, California. Using his professional voice, he phoned landlords and consistently received appointments. But when the landlords met him in person, he says, things quickly changed. “That’s when I was told the apartment was no longer available,” he recalls, “or there had been some mistake.” He decided to conduct an experiment, making a large number of phone calls to landlords in the Bay Area, varying the calls among his three dialects. As he predicted, when he spoke in Standard English he set up far more appointments in predominantly white areas than when he spoke in his other dialects.
Over the next ten years, this type of telephone audit became a standard way for social scientists to gauge patterns of housing discrimination. In 2001, using similar techniques, Douglas Massey and Garvey Lundy, two sociologists at the University of Pennsylvania, published a study about access to rental housing in Philadelphia and found significant disparities. Around that time, Baugh began thinking about how such cases might be handled in a courtroom. Without hard scientific evidence, he figured, plaintiffs in housing discrimination cases would be at a loss to defend their intuition that their race had been identified over the phone. “From a legal standpoint,” he says, “that’s real problematic. Any defense attorney can say, Why are these Black English speakers? And a plaintiff doesn’t have anything to say about that, except that they sound more black.”
In response, Baugh has made a systematic study of people’s ability to recognize dialects, often using recordings of his own voice in experiments. In 1997, he and two fellow linguists played recordings of a variety of his speech samples to a group of 421 university students: Eighty-four percent identified his Black English as such, 86 percent recognized his Standard English, and 91 percent identified his Chicano English. “To my chagrin,” Baugh quipped to the audience at the Austin conference, “my Black English is judged to be one of the least authentic styles.” From a scientific standpoint, though, the significant fact was that all the rates of identification were high. Even when listeners were played recordings of Baugh saying nothing but “hello” in his three dialects, an astonishing 72 percent of them accurately identified the dialect in question. Results like these suggested that linguistic profiling was possible—and could be proved in court.
But does linguistic profiling really need a scientific defense? The novelty of Baugh’s involvement has created a stir of excitement around Johnson’s case, but in truth courts have long acknowledged that race can be identified by speech—at least in a variety of criminal cases. In the 1912 case Rhea v. State, the Supreme Court of Arkansas upheld a murder conviction that hinged on testimony from a witness who said he had heard a “cultured” white voice shout in a crowd. Lawyers for the convicted murderer W.S. Rhea argued that the testimony should have been inadmissible, but the Arkansas justices ruled otherwise. “It is not a mere matter of opinion,” the court wrote, “[that] one [can]…recognize and know the difference between the voices of persons of different nationalities, and between that of a white man and a negro.”
More recently, in the 1999 case Clifford v. Kentucky, the Kentucky Supreme Court upheld the drug conviction of a black man named Charles Clifford on similar grounds. In that case, an undercover narcotics detective had purchased drugs from Clifford and secretly transmitted their conversation via microphone to a police surveillance team. The resulting recording was inaudible, but a police officer who had monitored the surveillance equipment testified that he had heard a black man’s voice make the sale. Since Clifford was the only black man in the room with the undercover detective, he was convicted.
In upholding the conviction, the Kentucky Supreme Court argued that a witness who testifies about the racial characteristics of a speaker’s voice is no less credible than a witness who testifies on other matters of human experience: the height of a suspect; the speed at which a car was traveling; the smell of gasoline. “We perceive no reason,” the court ruled, “why a witness could not likewise identify a voice being that of a particular race or nationality, so long as the witness is personally familiar with the general characteristics, accents, or speech patterns of the race or nationality in question.”
If racial identification by voice has been acknowledged in murder and drug cases, why hasn’t it played a more conspicuous role in housing discrimination cases? Baugh worries that the reason may be simple racial bias: In criminal and drug cases, where blacks are disproportionately defendants, courts have allowed the notion of racial identification by voice; in housing discrimination cases, where blacks are disproportionately plaintiffs, racial identification by voice has not been ruled on. The imbalance works to the disadvantage of blacks in both situations.
Some fair-housing experts, however, have a less cynical explanation for the imbalance. Linguistic profiling has not been addressed in fair-housing cases, they argue, because by and large the notion of linguistic profiling hasn’t been needed. Connie Chamberlin, the president of Housing Opportunities Made Equal (HOME), one of the most successful fair-housing organizations in the country, notes that her organization has been winning cases for years without the help of scholars like Baugh. When HOME receives a complaint, she explains, it typically begins by placing telephone calls to the landlord in question.
More effectively, though, HOME also sends teams of “testers” to visit the landlord in person and inquire about housing availability. The testers are pairs of people who differ in race, but who present nearly identical career paths, income, and marital status. In court, they testify on the contrasting ways in which they were treated (or mistreated). As a result, there isn’t necessarily a need for fair-housing cases to be decided solely on voice evidence. James Johnson’s lawsuit, for instance, also contains testimony from testers who visited the landlord he contacted, and is therefore winnable without engaging the issue of linguistic profiling.
Chamberlin concedes, however, that voice evidence has an impact in typical cases—albeit in subtle ways. Fair-housing lawyers often play tape recordings to juries of the phone calls made by prospective tenants. But since the tapes provide only one form of evidence among others introduced, the lawyers don’t need to instruct the jury to draw conclusions about racial identification—though jurors surely do so as a matter of course. “It’s our experience,” Chamberlin explains, “that when juries are confronted with taped evidence, they have no difficulty deciding what the housing provider would have known on the phone. There isn’t a need to be as technical as linguistic profiling makes it appear.” Stephen Dane, a Toledo-based lawyer who has litigated housing discrimination cases for 15 years, concurs: “In my mind, you don’t need an expert to tell a jury that you can tell race by voice. I think everybody knows that.”
Even if housing discriminations suits have been won for years without the assistance of scientific studies like Baugh’s, some fair-housing advocates think that a firm legal precedent acknowledging linguistic profiling can’t do any harm, and will only make litigation easier. As the linguist Dennis Preston argues, “We need a judicial decision that says we don’t have to test this more than once.” Still, practitioners like Chamberlin and Dane remain cautious about Baugh’s high-profile involvement. “More power to him if
he wants to strengthen the general understanding of the validity of linguistic profiling,” Chamberlin says. “But if the precedent becomes that you can’t prove a fair-housing case without an expert, that’s doing us no favors.”
Evidence of linguistic profiling may eventually be forced to take center stage, however, as interactions between buyers and sellers occur increasingly on the telephone. Many insurance and home mortgage companies, for instance, conduct negotiations and transactions entirely by phone. Shanna L. Smith of the National Fair Housing Alliance has urged such companies to hire linguists to train their employees so they’ll be aware of inferences they draw from a speaker’s voice. “I say to the big companies,” she explains, “that this training helps if you have a class-action lawsuit, since you can diminish or eliminate the damages. If it’s just an individual employee discriminating, you can’t punish the company, just the employee.”
For his part, Baugh anticipates that a favorable ruling on linguistic profiling would help serve the cause of a range of non-Standard English speakers, not just blacks. “One thing I didn’t expect when I started this,” he says, “is that linguistic profiling affects the deaf, and it extends to sexual orientation as well. It’s much bigger than racial profiling.” As for McWhorter’s argument that linguistic profiling is being blown out of proportion, Baugh is dismissive. McWhorter’s point that linguistic profiling wasn’t discussed in 1964 is irrelevant, Baugh argues, since the notion of racial profiling didn’t exist then either. Ironically, he adds, the increasing economic and social parity of blacks and whites that McWhorter celebrates has been a factor in the emergence of linguistic profiling cases. “The Mr. Johnsons of the world,” Baugh says, “now have the money, now have the credit, and they’re saying enough is enough.”