Do some school-age children have no language? Some problems of construct validity in the pre-LAS Espanol

Bilingual Research Journal, Summer 2002 by MacSwan, Jeff, Rolstad, Kellie, Glass, Gene V

Abstract

This article reports the existence of a large group of students identified as "non-nons," Spanish-background school-age children living in the United States who are reported to be non-verbal in both English and Spanish, and brings the validity ofthe"non-non"construct into question. In particular, the authors assess the validity of the PreLanguage Assessment Scales Espanol (Pre-LAS Espanol), an oral language assessment that purports to measure oral native language ability in children ages 4 to 6. A dataset of 38,887 students who took the Pre-LAS Espanol in a large urban school district in 1997 is examined, and questions are raised from internal and external evidence regarding the test's validity. The authors conclude that there are serious concerns regarding the validity of the test, characterize the "non-non" label as an artifact of poor assessment, and recommend that districts and states reconsider current policy requiring or recommending routine oral native language assessment of language minority students.

Introduction

A common belief among teachers, policy makers, and education researchers is that some school-age children have no language. The LA Times reported that 6,800 children in the Los Angeles Unified School District are classified as "non-nons" and said to be "nonverbal in both English and their native language" (Pyle, 1996). Children are so classified as a result of native-language assessment instruments. One such test that is commonly used is the Language Assessment Scales-Espanol (LAS-Espanol) (DeAvila & Duncan, 1990, 1994, 1986a, 1986b), which classifies Spanish-speaking children into the categories "fluent Spanish speaker," "limited Spanish speaker," and "non-Spanish speaker."

In this article, we explore the construct validity of the Pre-LAS Espanol, the Spanish-language version of the LAS intended for use with children ages 4 to 6, by considering both external and internal evidence bearing on its validity.1 Analyses reported here are based on a dataset of 38,887 subjects who took the Pre-LAS Espanol in a large urban school district in California in 1997. By investigating the validity of the test, we further question the legitimacy of the "non-non" label that the test applies to many Spanishbackground English learners.

We begin with a discussion of some possible influences in bilingual education that may underlie current policies in many states that require or recommend assessment of oral native language ability in school-age children, and then turn to a specific analysis of the validity of one such assessment, the Pre-LAS Espanol. We conclude by suggesting that states should reconsider testing policies that routinely assess a language minority child's oral native language ability.

Native Language Assessment Policy

Analysis of program alternatives for English learners has shown that academic instruction in a child's native language is often associated with outcomes superior to those resulting from all-English instruction (Willig, 1985; Ramirez, Pasta, Yuen, Billings, & Ramey, 1991; August & Hakuta, 1998). At least two distinct explanations have been advanced to account for the advantage of native language instruction for minority students. A traditional view is that native language instruction is especially beneficial to language minority children because it makes academic content comprehensible during the years it takes them to learn English well enough to understand all-English instruction (Krashen, 1996).

Another view, which contrasts sharply with this, attributes achievement differences in language minority children to presumed ability differences in children's native oral language. Although this view may be appealing to some, it is extremely difficult to distinguish from classical prescriptivism, in which differences between the language of the educated classes and the language of the unschooled are described in terms of levels of ability, degrees of complexity, or depth of vocabulary. We find such descriptions of children's oral native language to be highly objectionable for empirical and theoretical reasons (for review, see Edelsky, Hudelson, Flores, Barkin, Altweger, & Jilbert, 1983; Martin-Jones & Romaine, 1986; Wiley, 1996; MacSwan, 2000; MacSwan & Rolstad, in press). More narrowly, in bilingual education, the notion of native language ability differences among minority children has been widely used and probably owes its popularity to Cummins' (1976, 1979, 1981, 2000a, 2000b) influential Threshold Hypothesis and BICS/CALP framework. We briefly outline these ideas below with an aim to illuminate current language assessment practices in schools.

Cummins' Threshold Hypothesis originally suggested that "negative cognitive and academic effects. . . result from low levels of competence in both languages" (1979, p. 230), a view that is also known as "semilingualism" (Cummins, 1979) or "limited bilingualism" (Cummins, 1981). The original concern around which the Threshold Hypothesis was developed was a conflict in research findings on the cognitive benefits of bilingualism. Earlier studies had concluded that bilingualism adversely affects cognitive and scholastic progress, while more recent work showed "positive cognitive consequences" for bilinguals. Cummins (1976) pointed out that the studies that found a negative effect were associated with linguistic minorities, where the minority language was being replaced in some sense by the socially dominant one, while the studies that found a positive effect were associated with "additive bilingualism," a situation in which majority-language children acquire a second language. Cummins (1976) hypothesized that linguistic minorities undergo native language loss and that "the level of linguistic competence attained by a bilingual child may mediate the effects of his bilingual learning experiences on cognitive growth" (p. 24). In other words, he attempted to explain the reports of negative effects of bilingualism on "cognitive and scholastic progress" by proposing that the subject population had a low level of linguistic proficiency in its first language? By contrast, children in the "additive" bilingual programs benefited from continued support of their first language in and out of school. As Cummins put it: "Negative cognitive and academic effects are hypothesized to result from low levels of competence in both languages or what Scandinavian researchers (e.g., Hansegard, 196[81; Skutnabb-Kangas & Toukomaa, 1976) have termed 'seingualism' or `double semilingualism"' (1979, p. 230).