Research in bilingual education has made an important impact on bilingual curricula and teaching practices. Unlike many other domains of educational research, research on bilingualism and bilingual education has impacted practitioners, forging a spirit of inclusion and promoting discussion between researchers and teachers. The intense in-service activity of the first decades of federally funded bilingual programs was underway as the field was building theories. By 1981 there appeared to be sufficient consensus in the field that a milestone volume was published by the California State Department of Education, Schooling and Language Minority Students: A Theoretical Framework (1981). The theories, thus made accessible, became a boon to administrators and teacher trainers hungry for explanations of how and why first language instruction would remedy the inefficiency of submersion English instruction for language minority students. As a result of these efforts, no teacher in the state of California is currently el igible to receive a teaching credential without an acquaintance with the work of Cummins and Krashen, for instance. So we might say that certain theories in bilingual education have become institutionalized.
However, the theories themselves may lead to decisions that have negative educational consequences (Edelsky et al. 1983; Martin-Jones and Romaine 1986; Commins and Miramontes 1989; Valadez 1995; MacSwan 1999, 2000). For example, Cummins proposed his famous threshold hypothesis as a way of explaining academic achievement differences in minority language bilinguals (such as Spanish-speaking children learning English in submersion programs in the United States) and majority language bilinguals (such as English-speaking children learning French in immersion programs in Canada) (Cummins 1976, 1979). Cummins posited that certain language minority children have "limited ability in both their languages," a condition he called "semilingualism" (Cummins 1979) or "limited bilingualism" (Cummins 1981).
This perception of students was highlighted in a recent Los Angeles Times article (Pyle 1996). In the schools of the Los Angeles Unified School District there were reported to be 6,800 immigrant students who have been labeled "non-nons," that is, children who allegedly do not know English, Spanish, or any other language. The district's educational response has been to place these children in individual classrooms and provide them with intensive language instruction. At the end of this article, we will address the wisdom or inadvertent harm that may result from such a decision. What may be driving the determination that these children are limited in both languages appears to be an interpretation of Jim Cummins's (1976 1981) threshold hypothesis, in which the lowest threshold of bilingualism is said to include children who have a "low level in both languages." The term that has been applied to these children is "semilingual"; in Los Angeles, they are called "non-nons." In another California district the label is "clinically disfluent."
Almost nowhere, however, is the notion of "non-non" or "semilingual" precisely defined. Cummins (1976, 1981) refers to the condition as one of "low level in both languages," and Toukomaa and Skutnabb-Kangas (1977) refer to it as "semilingualism." The closest we come to an actual definition is that proposed by the philologist Nils Hansegard (1968), cited in Skutnabb-Kangas (1981). He lists the characteristics of "semilingualism" as a deficiency in both (all) languages in any of these categories: (a) size of repertoire of words and phrases; (b) linguistic correctness; (c) degree of automatism; (d) ability to neologize; (e) mastery of cognitive functions of language; and (f) richness of meaning.
We used these criteria to pose our research question:
Do the children bearing the label "semilingual" or "clinically disfluent" have a relatively impoverished knowledge of morphology, make frequent errors in syntax (word order), and appear relatively inexpressive? …