Linguistic Profiling and Fair Housing

Overview

Massey’s research: America’s large urban areas remain highly segregated, with 41% of Black Americans in hyper-segregated neighborhoods (all-Black high-density areas near other all-Black neighborhoods) and an additional 18% in high-segregation conditions.
Historical framing: Post-World War II housing discrimination by property owners, mortgage institutions, and insurers; today, discriminatory practices persist in new forms.
California case as an example of tyranny of the majority: 1963 Rumford Fair Housing Act (prohibited racial discrimination in housing); 1964 Proposition 14 attempted to amend the state constitution to allow discrimination; Prop 14 passed with 65% but was later struck down as unconstitutional by the California Supreme Court and ultimately by the U.S. Supreme Court in May 1967.
Core concept: tyranny of the majority vs. protection of minority rights in the U.S. constitutional framework; link to the Bill of Rights and Civil Rights legislation.
Core laws and actors: Civil Rights Act of 1968 (Fair Housing Act), Office of Fair Housing and Equal Opportunity (FHEO), state agencies, non-profit civil-rights organizations, and housing advocates.
Practical focus: What counts as discrimination and how it is prohibited in housing markets (rental, sale, lending, services, advertising, and related activities).
Emergent theme: Despite legal protections, discrimination remains pervasive in housing markets; rationales fall into victim vulnerabilities and profit-motive screening.
Linguistic profiling as a tool for discrimination: accent and dialect as basis for discriminatory treatment in housing market interactions (especially over the phone).

Historical and legal context

Rumford Fair Housing Act (1963): prohibited racial discrimination in housing practices in California.
Proposition 14 (1964): sought to revise the state constitution to allow discrimination (text cited: non-denial/limit/abridge right to sell/lease/rent by absolute discretion to choose). This would have created a constitutional right to discriminate.
Prop 14 vote outcome: Passed with 65% majority; later struck down by California Supreme Court as unconstitutional (violation of the Fourteenth Amendment).
U.S. Supreme Court affirmation: In May 1967, the Supreme Court affirmed the California ruling.
Primary sources and references: Noel & Cheng 2009; Oppenheimer 2020:118.
Tyranny of the majority: Courts acted to protect minority rights even when public opinion opposed them.
Bill of Rights as a counterweight to majority rule; Civil Rights Act of 1968 (Fair Housing Act) broadened protections to include a wider set of discriminatory behaviors and protected classes.
Federal enforcement: Office of Fair Housing and Equal Opportunity (FHEO) administers and enforces federal fair-housing laws; aims to ensure equal access to housing.
Related actors and mechanisms: state organizations, dozens of non-profit civil-rights groups, and fair-housing advocates.

Prohibited actions under federal and state fair housing law

Refusal to rent or sell housing;
Taking steps to make housing unavailable (e.g., removing property from market to avoid protected categories);
Setting different terms, conditions, or privileges for sale or rental (e.g., higher deposits for a particular race);
Providing different housing services or facilities (e.g., restricting access to amenities);
Falsely denying that housing is available for inspection, sale, or rental;
For-profit practices such as blockbusting or denying access to or membership in housing-related services;
Refusing to make a mortgage loan; refusing to provide information regarding loans; imposing different loan terms (rates, points, fees);
Discriminating in property appraisal; refusing to purchase a loan; setting different terms for purchasing a loan;
Threatening, coercing, intimidating, or interfering with anyone exercising fair-housing rights or assisting others who exercise those rights;
Advertising or statements indicating a limitation or preference based on race, color, national origin, religion, sex, familial status, or handicap.
Note: The prohibition against discriminatory advertising applies to single-family and owner-occupied housing that is otherwise exempt from the Fair Housing Act.

Why discrimination persists

Victim vulnerabilities: victims may be unaware of rights or how to pursue redress.
Profit motive: property owners, managers, contractors, and lenders seek higher profits even when discriminatory behavior is illegal.
Katrina case: after Katrina, Latino communities faced heightened vulnerability; contractors recruiting Latinos for reconstruction projects; lack of government oversight; Natalia, a Latina teacher fluent in English, experienced discrimination; discrimination audits reveal subtler forms of bias.
Discrimination audits (tester methodology): government and non-profit agencies track discrimination by sending testers with identical information but different racial/ethnic/accented identities to compare treatment (e.g., Korean-American applicant vs. Anglo applicant for mortgages).
Natalia’s case: experiences with rental agents documented and compared to Anglo non-Latina; discrimination often masked by a friendly voice; many cases go unreported.
Resulting insight: as protections tighten, screening becomes subtler and more difficult to detect.
National Fair Housing Alliance (NFHA) view: discriminatory practices in real estate and housing markets contribute to segregation and wealth gaps (education, employment, home ownership, wealth accumulation).

Linguistic profiling as a specific mechanism

Insidious method: screening by accent over the phone before in-person visits; “Sounds like discrimination” is a key illustration.
John Baugh: sociolinguist studying English varieties and discriminatory practices against non-dominant language speakers; focus areas include African American Vernacular English (AAVE) and Chicano English (ChE).
Personal narrative (Baugh): looked for an apartment around Stanford; used three varieties of English; formal academic English via telephone led to a promise to visit; in-person contact resulted in denial.
Language subordination model: the ideology that if you sound like “us,” you’ll be accepted; people who speak SAE finish with easier access, regardless of actual race.
Variation is natural; problem arises when language features are used as a basis for discrimination.

Research program and methodology (linguistic profiling)

Ford Foundation-funded four experiments (Purnell, Idsardi, and Baugh, 1999) to determine if race/ethnicity can be inferred from speaking voice without visual cues.
Experiment 1: Baugh collected data by calling five SF Bay Area neighborhoods; used the same sentence with variations to mask identity and used different return numbers and pseudonyms for each of the three calls to each landlord; hypothesis: in Anglo-majority communities, fewer affirmative responses to inquiries in AAVE or ChE.
Experiment 2: same sentence, 20 native speakers of targeted varieties recorded; three Baugh utterances recorded for comparison; 400 Stanford students listened and guessed race, ethnicity, sex.
Experiments 3 and 4: focus on fine phonetic/acoustic details; isolate the word "hello" from Baugh’s recordings to minimize external factors; duration constraint: x ext{ (duration)} = 414 ext{ ms}; this allows natural perceptual studies with minimal utterance length.
Key finding summary (Purnell et al., 1999): null hypothesis rejected; dialect-based discrimination exists; ethnic group affiliation is recoverable from speech; very little speech suffices to discriminate dialects; some phonetic correlates of dialects are recoverable from short speech.
Four major conclusions (Purnell et al., 1999):
1) Dialect-based discrimination takes place.
2) Ethnic group affiliation is recoverable from speech.
3) Very little speech is needed to discriminate dialects.
4) Some phonetic correlates or markers of dialects are recoverable from a very small amount of speech.

Experiments: details and findings

Area-level patterns (Experiment 1 data): population breakdown in five SF Bay Area locales (East Palo Alto, Palo Alto, Woodside, Oakland, etc.).
- East Palo Alto: roughly equal numbers of Anglos, African Americans, and Chicanos/as; Palo Alto: very small Chicanos/as and African Americans; Woodside: predominantly Anglo.
- Oakland: largest proportion of African Americans; East Palo Alto: largest proportion of Chicanos/as.
- In terms of responses to inquiries: ChE inquiries had low success overall (as low as 20%) in some areas; East Palo Alto yielded the best response to ChE but still only about 60% success for ChE inquiries, illustrating discrimination even in areas with higher minority populations.
- Corollary: discrimination is not limited to Anglo-dominated areas;Anglo residents still receive more favorable outcomes in many cases.
Table 17.1: stereotype judgments about intelligence among four race/ethnicity groups (Los Angeles area data; scale 1 = intelligent, 3.4 = average, 7 = not intelligent).
- Groups being stereotyped: Blacks, Latinos, Asians, Whites.
- Table layout (values are average judgments):
- Whites judging Blacks: 3.79; Latinos: 3.96; Asians: 2.90; Whites: 3.09
- Asians judging Whites: 4.39; Latinos: 4.46; Asians: 2.90; Whites: 3.25
- Latinos judging Blacks: 3.93; Latinos: 3.57; Asians: 2.74; Whites: 2.87
- Blacks judging Whites: 3.31; Latinos: 3.96; Asians: 3.21; Whites: 3.32
Housing discrimination toward Muslims (post-9/11 context): surge in discriminatory acts toward people of Middle Eastern ancestry; religious discrimination spotlight; Muslim accent as a recognizable but less-established category.
A case involving Pakistani immigrants in San Francisco public housing (Ashan Khan): incident after break-in, Quran desecration, and passport/clothing vandalism; housing authority denied emergency transfer; court report (Lagos 2007) indicates ruling treated as burglary rather than protected transfer.

Summary and implications

HUD National Fair Housing Alliance uses phone audits to detect linguistic profiling linked to aural traits; evidence shows profiling based on accent and dialect affects housing access.
Key empirical patterns: Anglos tend to receive faster responses, broader options, and better financial terms before references or credit checks are reviewed; profiling advantages persist prior to loan checks or credit evaluation.
The investigative process and legal remedies are slow, and fines are often not enough to deter ongoing discrimination.
The primary defense is raising awareness of rights and promoting education about linguistic profiling among the public and policymakers.

Discussion questions and exercises

Watch Equal Housing Opportunity ad illustrating linguistic profiling (Anglo caller with various stigmatized accents).
How would you design a formal investigation into accent-based discrimination in housing telephone interactions? Consider methodology, ethics, and data analysis.
Imagine a friend who has experienced linguistic profiling in housing; outline the steps they would take to pursue a complaint and gather the necessary documents.
Evaluate whether the complaint process is reasonable or unduly complex; reflect on potential barriers to reporting.
Read the suggested readings and discuss whether the tone of the literature leans toward a particular ideological perspective.
Recall or provide examples of linguistic profiling in everyday telephone interactions and describe the reactions and judgments made about the caller.

Visual data references and key figures

Figure 17.1: Diagram illustrating the concept of “Sounds like discrimination” (accent-based profiling over the phone).
Figure 17.2: Population breakdown in five San Francisco-area locales by race/ethnicity and the rate of apartment-appointment confirmations by race/ethnicity.
Table 17.1: Stereotype judgments on intelligence of four groups (Blacks, Latinos, Asians, Whites) by four judging groups (Whites, Asians, Latinos, Blacks).

Real-world relevance and implications

Language as a social marker can have tangible consequences in access to housing, wealth accumulation, and community segregation.
The jurisprudence surrounding fair housing aims to counteract overt and subtle discrimination, but enforcement challenges persist due to hidden practices (e.g., accent-based screening, phone audits).
Ethical considerations include balancing tester methodology with privacy rights and the potential for misinterpretation of language features as proxies for protected characteristics.

Footnotes and clarifications

1 Prop 14/1963 refers to Prop 14/1963 and a later Prop 14; Prop 14 passed in 1964 but is distinguished from 1963.
2 Equal Protection Clause of the 14th Amendment requires equal protection under the law regardless of protected status.
3 HUD stands for the Department of Housing and Urban Development.
4 The HUD/FAHO website: listed above.
5 The question of race/ethnicity from a single utterance is complex and can be affected by racist applications; see Purnell et al. (1999) for experimental details.
6 The testing methodology mirrors legally approved tester practices used by HUD and other agencies (e.g., City of Chicago v. Matchmaker Real Estate Sales Center, 1992; United States v. Youritan Construction Company, 1975).

Linguistic Profiling and Fair Housing

Overview

Historical and legal context

Prohibited actions under federal and state fair housing law

Why discrimination persists

Linguistic profiling as a specific mechanism

Research program and methodology (linguistic profiling)

Experiments: details and findings

Summary and implications

Discussion questions and exercises

Visual data references and key figures

Real-world relevance and implications

Suggested readings and resources

Footnotes and clarifications