Problematic Language Assessment in the US Census

Ana Celia Zentella
UC San Diego

Bonnie Urciuoli
Hamilton C

Laura R Graham
U Iowa

Every ten years the US government undertakes a census. Although many people assume that the purpose of the decennial census is to obtain a total count of the US population, the instrument is designed to assess numerous social dimensions of US society. English-speaking ability and use of other languages are categories that the census now attempts to ascertain through three questions about language: Do you speak a language other than English at home? If so, which language? How well do you speak English: Very well, well, not well, or not at all?

These language-focused questions provide information that policymakers and legislators use to determine resource allocations for education, voting and access to social services and public goods. The questions about language on the US Census are problematic in phrasing and provide misleading information that leads to the reinforcement of social inequalities in contemporary US society.

Questioning the Correctness Ideology
The language questions on the US Census are based on how most Americans conceptualize language. Many think of language as a set of words and grammatical rules that individuals control, and as a system whose primary function is accurate reference. The idea that individual control can be measured in terms of correctness and that insufficient correctness means loss of meaning are prominent within this ideology.

The US Census questions reflect this correctness ideology. They also reflect the assumption that language can “naturally” be measured in discrete levels of correctness—very well, well, not well, not at all—and that individuals understand these levels. Such assumptions ignore the fact that all language use is interactive. What is considered correct in some contexts is irrelevant in others and effective language is as much a function of social relations as of “correct” form.

None of these basic anthropological principles of language as interactive and contextual are reflected in census questionnaires. Given the assumptions underlying its notions of language, the data the census provides are likely to be of limited use. The census data may, in fact, reinforce social inequalities due to the close correlation between assessment of language, assessment of class and assessment of other characteristics seen as foreign and often very naturalized in terms of “race.” This is because prevailing notions of “correctness” correlate with elite forms and the language used by members of the dominant social class.

The Importance of Context
According to the US Census of 2000, approximately 18% of the nation speaks a language other than English at home. This represents an increase of 5% since 1990. The increasing numbers of speakers of other languages in the US is of concern to many public servants, especially educators, health and legal system workers, and social service personnel. Many members of the general public are worried that the role of English as the lingua franca is being threatened, as reflected in the popularity of the English Only movement and the growing number of states that have adopted English Only legislation (29 to date).

In this climate, the public announcements of the US Census Bureau concerning the language data it collects can serve to quiet widespread fears or fan the flames of intolerance. Unfortunately, the Census Bureau uses the levels of English proficiency reported in answer to its question, How well do you speak English: Very well, well, not well, or not at all? as the basis for classifying speakers of languages other than English in ways that reflect misconceived notions about language, and implying that speakers of languages other than English are unable and/or unwilling to participate in our society.

Moreover, the four classifications “very well,” “well,” “not well” and “not at all” are arbitrary. What do these categories mean to individuals attempting to answer the census questions? What is the difference between “well” and “not well”? Might there be an intermediate term, such as “reasonably well”? This is particularly important because the census cut-off for deciding who is “linguistically isolated” is made at “very well.” Why not “well”? Or “reasonably well” if that category existed?

As anthropologists should know, there is no mechanism available by which people who are asked to self-report can consistently interpret categories as they are given on the census. People with the same language capacities often interpret and report their own capacities quite differently. How people see and judge their own English language capacities depends on a variety of factors, including what they use the language for, how likely they feel they may be judged, by whom, according to what standards of correctness, spoken or written, and so on.

Gender, age, years of education and degree of literacy all play important roles in a person’s self-assessment. Some people are unnecessarily modest in judging their own language abilities. Some believe that having an accent, or not being able to recall which English verb tenses are indicated by vowel change rather than an “-ed” suffix, or not being able to recall an English word more quickly than one in their native language mean their speech fits in the category “not well,” for instance.

The US Census classifications mistakenly presuppose that acts of communication take place in a vacuum. Language capacities are not simply properties of individuals. People’s capacity to interact linguistically is a function of several synergetic social factors, such as what they are talking about, how, why, where, how well they know each other, what they can take for granted about each other’s understanding and so on. In short, the sliding scale of “very well” to “not at all” presupposes communication as the act of an individual rather than a social production. In case there is any doubt, individuals need only recall that there are powerful and high-status people who only interact effectively with others like them.

“Isolation” and “Diversity”
Finally, the US Census Bureau’s use of “linguistic isolation,” a category that it adopted in 1990, is especially problematic. Individuals and families are classified as “linguistically isolated” if their household, “is one in which no member 14 years old and over: 1) speaks only English; or 2) speaks a non-English language and speaks English ’very well.’” News media regularly report the percentage of “linguistically isolated” individuals and households without explaining the assumptions that underlie this misleading label.

The majority of speakers of languages other than English speak English “well” and many children under the age of 14 in immigrant families speak only English. Almost all families are in contact with English speakers and are not “linguistically isolated.” By inaccurately suggesting that there are large numbers of “linguistically isolated” households and individuals, the Census Bureau implies that the nation is plagued by language problems and that immigrants are not participating in society. This misinformation fans hostility against immigrant populations, especially working class Latino immigrants. Furthermore, it is worth pointing out the dual standard regarding this categorization, since the 82% of the US population that speaks only English at home is not classified as “linguistically isolated.”

Lamentably, the census also ignores the diverse linguistic strengths of the nation. The language portion of the census contains no questions related to proficiency in languages other than English. In an effort to address its misrepresentations and omissions, the National Council of Teachers of English and the American Association of Applied Linguistics have passed resolutions urging the Census Bureau to change its language questions.

To signal anthropologists’ awareness of ways that the language questions on the US Census may undermine social equalities, the AAA Committee for Human Right’s Language and Social Justice Task group is working to propose a AAA resolution asking the Census Bureau to include a question about proficiency in languages other than English, to stop classifying those who speak English less than “very well,” and to end its use of the “linguistically isolated” classification because the term is inaccurate, discriminatory and promotes an ideology of linguistic superiority that foments linguistic intolerance and conflict.

Ana Celia Zentella (PhD, University of Pennsylvania), an anthropolitical linguist, is nationally recognized for her studies of US Latino varieties of Spanish and English, bilingualism, “Spanglish,” language socialization and language policy. Bonnie Urciuoli is a professor of anthropology at Hamilton College, and has done work on Spanish-English bilingualism, on discourses of race, class and language, and, currently, on the institutional production of diversity discourses in liberal arts education. Laura R Graham, associate professor of anthropology, University of Iowa, is a member and former chair of CfHR. She chairs the CfHR Task Group on Language and Social Justice.