Measurement of Language Use and Language Proficiency, and Literacy and Its Analysis

Siegel, Jacob S.

doi:10.1007/978-3-319-61778-7_5

Jacob S. Siegel²

570 Accesses

Abstract

In this chapter I describe some measures of language use and measures of proficiency in the use of the national language (e.g., English in the United States) by foreign-born residents. This material should be useful in measuring the linguistic resources of a society, advancing the civil rights of minorities, evaluating the progress toward integration and linguistic acculturation of immigrants, developing a more nuanced concept of ethnicity, and analyzing sociolinguistic data obtained in social surveys.

This is a preview of subscription content, log in via an institution to check access.

Access this chapter

Log in via an institution

eBook: USD 16.99; Price excludes VAT (USA)

Softcover Book: USD 109.99; Price excludes VAT (USA)

Hardcover Book: USD 109.99; Price excludes VAT (USA)

Tax calculation will be finalised at checkout

Purchases are for personal use only

Institutional subscriptions

Notes

1.
Some classification issues are: Chinese is more geographic and political than linguistic since it combines Mandarin, Cantonese , and other Chinese languages using Chinese characters ; French is more geographic and political when it includes French Creole and Cajun. Persian should be termed Farsi, the language. Amharic is a Semitic language of the Afro-Asiatic family , not an African (or sub-Saharan African ) language.
2.
The estimation formula for the total population is as follows:
$$ \mathrm{N}++=\left(\mathrm{N}1.1+\mathrm{N}2.1\right)\left(\mathrm{N}1.1+\mathrm{N}1.2\right)/\mathrm{N}1.1 $$
where the symbols refer to the numbers in the 2 × 2 matrix. The equation says that the sum of the first column times the sum of the first row divided by the first entry in the first row yields the total of the four cells, or the total found by the first survey times the total found by the second survey divided by the number found by both surveys yields the total population. (See Siegel 2002, for further information.)
3.
The basis for this dichotomy of responses is partly judgmental, aiming to separate those who are functionally literate in English from those who are not. The English Language Proficiency Study, conducted in the fall of 1982 by the U.S. Census Bureau for the U.S. Department of Education, administered tests of ability to understand English to non-English speaking persons and some English-speaking persons (as a control group) in their homes. This test showed that persons responding “very well” to the census question on English speaking ability had passing levels in the test similar to the English-speaking population which had served as a control group, while persons reporting ability levels of “well” or worse had significantly higher levels of failure. On the other hand, the National Content Test and its Reinterview, conducted by the U.S. Census Bureau in 1986, showed that most people who said they could speak English “well” and nearly all who said they could it speak it “very well” reported that they could read and write English. Incidentally, the National Content Test provides strong corroborative evidence that the English-proficiency question in the census can operationally distinguish different categories of English-speaking ability. On the whole, these finding support the alternative two-category grouping (Kominski 1989)
The U.S. Census Bureau provides local data for “very well” and “less than very well” to the Department of Justice and to the Department of Education in connection with the administration of the Voting Rights Act and the Elementary and Secondary Education Act , respectively, at the request of these agencies. This choice maximizes the coverage of these laws in favor of linguistic minorities and apparently that was its purpose.
4.
The U.S. Census Bureau stopped using the expression “linguistically isolated households ” a few years before the publication of this book, yielding to opposition to the use of the expression from the American Anthropological Association, and began using instead the expression “limited English-speaking households.” The author uses the original expression here because he considers it an appropriate name for the measure. The American Anthropological Association has also objected to the new expression “limited English-speaking households” (Anthropology News, March/April 2014) but the Census Bureau has received support for it from several minority -language groups and plans to continue its use.
5.
Such approximations are adequate, however, for use in regression analysis, along with other variables selected for the regression measuring proficiency in the destination-area language.
6.
The matrix of the ratios/rates must show sufficient age detail at frequent enough intervals for the matrix to be reasonably dense with observations, and the matrix must be organized so that the calendar years are in temporal order and the cohorts are aligned with the age groups in the appropriate calendar years. The data collected on linguistic variables tend to present a problem in this regard since they are likely to be available for only a few years and require smoothing and interpolation. The data may be tabulated in one way in 1 year and in another way in another year and may be irregular because of the sample size. The reader is advised to refer to other publications for guidance in handling the problems of smoothing and interpolating raw data.
7.
The identification problem means that, in a multiple-regression solution relating the dependent variable to the independent variables , no unique set of coefficients can be obtained and an infinite number of solutions give identical fits to the data.
8.
The more recent efforts claim to estimate the three independent effects on the outcome variable with less drastic constraints. The two leading methods proposed for solving the APC problem at present are called the Intrinsic Estimator Model and the Cross-Classified Fixed/Random Effects Model (Frenk et al. 2013; Yang and Land 2006; Yang et al. 2008). O’Brien (2011), Luo (2013), Luo and Warren (2014), and others have challenged these methods on the grounds that they still impose major unprovable constraints on the data.
Luo and Warren (2014) have proposed their own solution, called the APC-Interaction (APC-I) Model, which, they claim, avoids the limitations of the earlier methods. They do not structure the problem in terms of estimating three independent APC effects but, following Ryder’s (1965) concept of a cohort, they consider cohort to be a specific form of the interaction between age effects and period effects . Further, they avoid the assumption that the effects of cohort are constant across the life course. Their APC model expresses the expected value of the outcome variable (e.g., vocabulary score) for age i at the jth period as a function of the value at age i, period j, and cohort k, but cohort k is itself expressed as a special function of the age-by-period interaction. Luo and Warren have applied their method to an APC analysis of trends in vocabulary knowledge in the United States between 1974 and 2012. A summary of their results is presented in Chap. 7.
9.
A literacy rate is a percentage and hence a type of ratio. A rate is also a type of ratio. Technically a percentage is not a rate, which may be defined as a ratio expressing the risk of an event occurring to a population exposed to that risk.
10.
Computing the percent change in such percentages is questionable statistical practice since the base value can vary only between 0 and 100 and therefore could be too small to yield stable results.
11.
This discussion of proximate and isolated illiteracy rates has benefited greatly from the material in Permanyer et al. (2013).

References and Suggested Readings

General

Haug, W. (2009). Ethnic, religious and language groups: Towards a set of rules for data collection and statistical analysis. New York: Open Society Institute.
Google Scholar
Lightfoot, D. (2006). How new languages emerge. Cambridge, UK: Cambridge University Press. See Chapter 8.
Book Google Scholar
Verdoodt, A. F. (1998). The demography of language. In F. Coulmas (Ed.), Handbook of sociolinguistics (pp. 33–43). Oxford: Blackwell Publishing.
Google Scholar