have some slippery ways of denying their guilt. When charged with administering
job or school application tests that are culturally biased, they claim
the tests are "merit-based." When racist cops stop and search
an innocent young Black man driving through a white neighborhood, they
cite crime statistics to back up their actions. And when racist landlords
reject potential Black tenants on the basis of a phone call, they say
they had no way of knowing the callers were Black.
in the case of the landlords, it's becoming increasingly difficult to
crawl behind the color-blind cover. Research by John Baugh, a professor
of education and linguistics at Stanford University, and others has
demonstrated that racial identification by speech takes place all the
time, and it has had several legal implications.
example, in 1999 racial identification by speech was used to convict
a Black man. A Kentucky Supreme Court judge, hearing an appeal by the
man, convicted for drug trafficking in a sting operation, ruled that
it was proper for a white police officer to identify a suspect as Black
solely on the basis of the voice he heard in an audio transmission from
a wired cop.
officer testified that in his 13 years on the force he had had numerous
conversations with Black males and knew the voice of a Black man when
he heard one. The judge, upholding the conviction, stated that no one
suggested it was improper for the officer to identify one of the voices
he heard as being that of a female. Thus, "We perceive no reason
why a witness could not likewise identify a voice as 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."
this case the defendant was done in by "linguistic profiling."
Surely you've heard of its sister, racial profiling, a practice infamously
employed by police officers, who stop and search Blacks simply because
they fit a "profile." In linguistic profiling, however, the
racial cues are aural rather than visual.
it TWB - Talking While Black. A person has a telephone conversation
with someone he has never seen before and draws a conclusion about the
race of that person based solely on the way the person sounds.
particularly insidious about that - on the face of it. We can usually
tell if a person is a man, a woman, young or old, a Southerner or a
Latino in the space of a five-minute telephone conversation. But linguistic
profiling is somebody "acting upon that racial or demographic imprint
in a criminal way by denying [the victims] access to a business transaction
that should not be in any way biased, based on a person's racial background,"
own personal experience with linguistic profiling occurred a few years
ago, when he was looking for a place to live in California. He would
call up in response to an ad in the paper, but when he would show up,
he would learn that the apartment was unavailable. He believes that
it is because over the phone, when he uses his "professional voice,"
he sounds White. When he appeared in person, he was handed all sorts
of excuses for why he could not rent - none being, of course, the obvious
fact that he was Black.
Baugh went about trying to prove what he had suspected. Having grown
up in the inner city, in Philadelphia and Los Angeles, Baugh was exposed
to a variety of ethnic dialects and considers himself "linguistically
dexterous." He began telephoning renters and would say, "Hello,
I'm calling about the apartment you have advertised in the paper."
He would make some calls using his professional voice. Other times he
would modify his voice, repeating the same sentence with the same grammar
but with an intonation that was unmistakably Black. He made more than
100 calls and found that his "Black" voice got half as many
calls back as his "White" voice. It did not matter that when
Baugh used his Black voice he was speaking perfect, standard English.
if a speaker on the telephone sounds African-American, he is subject
to the same kind of racial discrimination as he might be in a face-to-face
least Baugh, when speaking in his professional voice, got called back
and was able to make
it to the second stage of the interview process. But what about those
African-Americans who call about apartments, or jobs, or loans who never
get called back and have no idea why? After all, they may be well educated,
and gainfully employed; in other words they look great on paper. What
could possibly put them at a disadvantage? According to Baugh and others,
it is simply TWB. The work that he has done with the National Fair Housing
Alliance (NFHA), a civil rights organization that focuses on housing
discrimination, bears this out.
National Fair Housing Act makes it illegal to deny housing, loans or
insurance to anyone on the basis of race. In loans and insurance, particularly,
most of the transactions are conducted over the phone. Very often whether
you are able to obtain financing at a reasonable rate
rests on how the person at the other end perceives you by your voice.
Sometimes the first thing out of an insurance agent's mouth, once he
or she has guessed that the caller is Black, is "Have your ever
had any claims against you? Have you ever cancelled?" says the
NFHA's executive director Shanna Smith.
based on linguistic profiling has been difficult to prove. Unless there
is some smoking gun - a written telephone message, saying the caller
sounds Black,or Mexican or whatever - judges have been reluctant to
hear such cases. Baugh's research has been employed by the NFHA to prove
that renters, loan companies and the insurance agents do treat callers
differently based on racial identification by voice.
one experiment Baugh and others tape-recorded the same phrase,"Hello,
I'm calling about the apartment you have advertised in the paper,"
changing the phonology - the sound or accent - of the voice, but always
using standard, grammatical English. All the speakers were adults. The
subjects of the experiment then had to identify as many social demographics
of the speakers as they could, whether they were men or women, Northerners
or Southerners, Black, White or Latino, young or old. (Baugh spoke three
different times, using his African-American voice, his professional
voice, and his Latino dialect.)
75 percent of the time in the Latino case, over 80 percent of the time
in the African-American case, and over 88 percent of the time in the
instances in which Baugh used his professional voice, the subjects identified
the speakers correctly as either Mexican or Puerto Rican, Black or White.
Baugh, using his three voices, was able to demonstrate that just by
manipulating intonation, he could lead people to very different conclusions
about the speaker.
another experiment Baugh's colleague spliced the word, "hello"
out of the complete phrase. The result was over 90 percent recognition
with accuracy for racial identification, using that word alone.
NFHA chose its linguistic testers based upon whether a "control"
person was able to identify correctly the race or national origin of
the tester over the phone. Evidence was gathered
in states where it is legal to tape a phone conversation. When an African-American
tester would call about renting an apartment, the landlord would lie
and say it was already rented. When a White tester followed up, the
apartment was once again available and an appointment would be set up
for him to come see the place.
is so easy for landlords to get away with this kind of discrimination,
because all they have to say is, "Oh, you know what? There are
three people ahead of you. Why don't you give me your name, and if those
fall through, I'll call you." To the caller that sounds legitimate.
"African-Americans and Latinos simply don't report that,"
says Smith. "That's why we do testing. We can catch that sophisticated
lie that's used to deny housing."
a lawsuit filed last fall, in San Francisco's U.S. District Court, lawyers
for the plaintiff, James Johnson, attempt to show how linguistic profiling
was used to deny him an opportunity to seek better housing in the neighborhood
where he lived. Johnson, who lives in San Leandro, CA and was looking
for a larger apartment two years ago, saw a For Rent sign in front of
an apartment complex on his block, and called the number listed on the
sign to inquire. He got a voice-mail message, instructing callers to
leave their name and telephone number, which he did, adding that he
had worked at Kraft Foods as a supervisor for 20 years.
Johnson did not receive a response to his message, he called repeatedly
and left several messages at the same telephone number. He never got
a call back. Exasperated, he gave up. Months later, Johnson saw a For
Rent sign posted again in front of the same apartment complex and called
the number to inquire - five or six times, leaving voice-mail messages
each time. Again, no reply.
asked a friend, a Latino who "sounds white," to call about
the apartment. The friend called, left a message and got a call back
on the same day from one of the owners of the complex. She told the
friend that an apartment was available. Johnson then filed a housing
discrimination complaint with the local fair housing center. The center
conducted an investigation, with five different testers - two Black
and three White - calling the apartment complex owners about availabilities
and leaving their names and numbers in voice-mail messages. The White
testers all got called back the same day. The Black testers' calls were
NFHA has filed suits charging Prudential with racial discrimination
against African-Americans in Milwaukee, Richmond, Toledo, Washington,
D.C. and Chester, Pa. All of the testing in those cases was done over
the telephone, says Smith. In some instances African-American testers
would call repeatedly to inquire about insurance, but were never called
back. The white testers calling the same agents had glaringly different
experiences. They were given quotes and encouraged to purchase the insurance.
Furthermore, says Smith, when an insurance agent thinks that the person
on the other end is white, the agent will market to that customer a
whole array of products that the African-American caller will never
get to hear about - auto, along with home-owner's insurance, for example
- which can lower the premium.
cases like these, Baugh's research is used to counter claims by the
defendants that they have no way of telling whether a caller is Black
or not, and that it is even racist to suggest as much. But common sense
tells us this is not true. Science confirms it.
who's screening out Black callers? Very likely people who are less educated
and earn less money than those on the other end. In some instances a
White person conducting business over the phone may be asked to practice
linguistic profiling by his or her supervisor. Smith mentions a case
in Alabama where a White apartment manager contacted the local Fair
Housing group to report that the apartment owner told her if she suspected
that a caller was Black, to tell him or her nothing was available. Smith
also recalls a conversation she had with a White woman working for an
employment services company, who had been instructed to get callers
to say the word "ask." If they pronounced it "ax,"
that was one way of identifying them as African-American. (Baugh says
that particular pronunciation is most commonly associated with Blacks.)
Smith, who is White, says, "We get to hear these things all the
only way to put a stop to linguistic profiling is for those who know
about it to report it and make the offenders pay. No need for Blacks
to hire speech coaches. What's called for is not more pear-shaped tones,
but more organizational muscle, exercised by the likes of the NFHA and
its local affiliates.