|
|
 | Educational Research in Bilingual Education |
| |
  | |||
| |
Educational Research in Bilingual Education Jim
Cummins University
of Toronto Interpretation
of the voluminous research on bilingual education has been highly controversial
among both academics and policy-makers for more than 25 years.
Clearly the political sensitivity of the issue has contributed to
confusion about what the research is actually saying.
A more fundamental cause of this confusion, however, is the extremely
limited way in which educational researchers have examined the research, and in
particular the quantitative research on this issue. The dominant assumption
among academic opponents and advocates of bilingual education has been that we
can draw policy-relevant conclusions regarding the effectiveness of bilingual
education only from “methodologically acceptable studies.” Typically, these
studies are program evaluations that involve treatment and control groups
compared in such a way that outcome differences can be attributed to the
treatment rather than to extraneous factors. I
argue that this approach represents an appropriate, but extremely limited,
orientation to research in bilingual education. It is limited on two counts: ·
It has
proved virtually impossible to apply rigorous controls to comparisons between
programs due to the myriad human, administrative, and political influences that
impact the implementation of programs over time; ·
Underlying
this approach is the assumption that there is a direct connection between
research and policy: researchers will discover which program alternatives work
best and policy-makers will develop funding and implementation policies based on
these findings. While this straightforward linkage between research and policy
may work well in some spheres (e.g. opinion surveys), it seldom yields clearcut
results in an area as multifaceted as educational research where the treatment
variable of interest (e.g. proportion of first language [L1] instruction) is
intertwined and interacting with hundreds of other variables that will affect
program outcomes. As
I document below, it is not surprising that this dominant research paradigm has
yielded such paltry pickings for policy. The only thing that academic opponents
and advocates of bilingual education seem to agree on with respect to the
policy-related research is that it is of almost universally poor quality (August
& Hakuta, 1997; Greene, 1998; Rossell & Baker, 1996). I
argue in this paper that “poor quality” is in the eye of the beholder.
Viewed through the lens of “methodologically acceptable studies,” it is
possible to find fault with virtually all of the research studies, including
many of those that survived the “rigorous criteria” established by Greene (1978) and Rossell and Baker (1996). However, there is
an enormous amount of relevant and interpretable research, both internationally
and within the United States, that speaks directly to the bilingual education
policy issues. I suggest that the policy issues have remained confused and
contested at least partly because the bulk of the relevant research has been
virtually ignored, both by advocates and opponents of bilingual education. The
relevance of this research is not apparent within the dominant paradigm because
the studies do not conform to the criteria of acceptability within this
paradigm. However, when we examine
this voluminous research from the perspective of an alternative paradigm, its
relevance is immediately apparent. The
alternative paradigm claims that the relevance of research for policy is
mediated through theory. In complex educational and other human organizational
contexts, data or “facts” become relevant for policy purposes only in the
context of a coherent theory. It is
the theory rather than the individual research findings that permits the
generation of predictions about program outcomes under different conditions.
Research findings themselves cannot be directly applied across contexts.
For example, the fact that students in a Spanish-English bilingual
program in New York City performed well academically in relation to grade norms
(Beykont, 1994) tells us very little about whether a similar program might work
with Mexican-American students in San Diego. Yet when certain patterns are
replicated across a wide range of situations, the accumulation of consistent
findings suggests that some stable underlying principle is at work.
This principle can then be stated as a theoretical proposition or
hypothesis from which predictions can be derived and tested. In
contrast to research findings, theories are almost by definition applicable
across contexts. The validity of
any theoretical principle is assessed precisely by how well it can account for
the research findings in a variety of contexts.
If a theory cannot account for a particular set of research findings,
then it is an inadequate or incomplete theory.
Thus, while no individual research finding can “prove” a theory or
confirm a hypothesis, any research finding can disconfirm or refute a theory or
hypothesis. Thus, the criterion of validity for any hypothesis is extremely
stringent: it must be consistent with all of the research data or at
least be able to account for inconsistencies (e.g. poor implementation of a
program). In
this paper, I label the former paradigm the ”Research-Policy” paradigm and
the latter the “Research-Theory-Policy” paradigm.
In order to show the strengths and limitations of each, I examine how
they have been employed in the bilingual education debate and the
policy-relevant findings that each yields.
I conclude that the former paradigm yields largely trivial information
for policy purposes and perpetuates misconceptions regarding bilingual education
that have persisted for almost 30 years. By contrast, the latter paradigm yields considerable
information that has direct relevance to policy and addresses many of the most
contentious issues in the bilingual education debate. The
Research-Policy Paradigm: Debunking the Mythology of “Methodologically
Acceptable Studies” Both
advocates and opponents of bilingual education have largely concurred on the
conditions under which research in general and evaluation studies in particular
can be considered “methodologically acceptable.” For example, Greene (1998),
whose interpretation of research is favorable to bilingual education, and
Rossell and Baker (1996) whose conclusions are highly unfavorable to bilingual
education, agreed on the appropriateness of the following criteria for
designation of studies as “methodologically acceptable:” 1.
Studies
had to compare students in bilingual programs to a control group of similar
students; 2.
The
design had to ensure that initial differences between treatment and control
groups were controlled statistically or through random assignment; 3.
Results
were to be based on standardized test scores in English; 4.
Differences
between the scores of treatment and control groups were to be determined by
means of appropriate statistical tests. The
approaches diverged on several additional points. Greene, for example, focused
only on studies that had been carried out in the United States and which
measured the effects of bilingual education after at least one year of the
treatment whereas Rossell and Baker included international research data in
their review. Rossell and Baker
also categorized programs according to pre-defined labels (structured immersion,
transitional bilingual education, submersion, English-as-a-second-language
[ESL]) whereas Greene adopted a more straightforward categorization of bilingual
education as a program in which all students are taught using at least some of
their native language. An additional difference was that Greene (like Willig,
1985) used meta-analysis to take into account effect sizes in individual studies
whereas Rossell and Baker (like Baker and de Kanter, 1983) simply counted
studies that were favorable or unfavorable to different treatments. Using
their respective criteria of “methodologically acceptable” studies, Rossell
and Baker rejected 228 studies and accepted 72 while Greene could find only 11
studies whose design permitted conclusions to be drawn.
August
and Hakuta (1997) and their colleagues reviewed both the basic and
policy-oriented research on schooling for language-minority children.
Their review is comprehensive, balanced, and useful for researchers and
policy-makers alike. It also offers very appropriate suggestions for improving the
quality of program evaluations. However, it suffers from the same problematic
orientation to research and policy as do the reviews by Greene (1998) and
Rossell and Baker (1996). In interpreting both the basic research and program
evaluations relevant to language learning and bilingual education, the authors
pay only lip-service to the role of theory in mediating the relevance of
research for policy (Fitzgerald & Cummins, in press).
As a result, their conclusions are considerably weaker than they might
have been and also considerably less useful for policy purposes. I
critique these three literature reviews below. Space does not permit a
comprehensive review of these studies and thus my focus is directed towards
identifying the limitations of the research paradigm these authors have adopted. Greene’s
Meta-Analysis Greene
reported that participation in a bilingual program over a period of two years
made a difference of about 1/5 of a standard deviation in achievement. Thus, if
a student in an English-only
program performed at the 26th
percentile at the end of those two years, the bilingual student would be at the
34th percentile. The
problematic nature of this type of meta-analysis can be seen from examining some
of the 11 studies included. One of
the large-scale “methodologically acceptable” studies included in the
analysis (and in Rossell and Baker’s review) was the American Institutes of
Research (AIR) study (Danoff, 1978). This
study aggregated a variety of programs labeled “bilingual education” and
compared them to non-bilingual programs, ignoring the fact that many bilingual
programs are bilingual in name only or involve minimal amounts of L1 instruction
by a non-trained paraprofessional (see, for example, Gandara, 1997). In
addition, as pointed out by August and Hakuta (1997) there was no clear
demarcation between treatment and control groups: Nearly three-quarters of the experimental group had
been in bilingual programs for 2 or more years, and the study measured their
gains in the last few months. Additionally, about two-thirds of the children in
the control group had previously been in a bilingual program; these children did
not represent a control group in the usual sense of the term. Thus the AIR study
did not compare bilingual education with no bilingual education. (p. 140) Krashen
(1999a) has also pointed out that Greene’s analysis included several
“methodologically acceptable” studies in which “bilingual education” was
either not described in any detail or involved minimal use of the L1 (e.g. use
of bilingual paraprofessionals) In
short, Greene’s meta-analysis makes no attempt to test the theoretical
propositions underlying bilingual education or alternative English-only
programs. The apparent rigor involved in reducing the extensive corpus of
bilingual education data to 11 “methodologically acceptable” studies seems
destined to end up in rigor mortis for this approach as the credibility
of even these 11 studies is whittled away. However, this is not because most of
the research is inadequate; the
inadequacy is rather in the lens through which we are examining the research. August
and Hakuta’s National Research Council Report
This
report was intended to provide researchers and policy-makers with a
“state-of-the-art” review of what basic and applied research could tell us
about improving schooling for language-minority students.
One might have expected to find clear and coherent answers to volatile
policy-relevant questions such as: ·
How long
does it take English language learners (ELL) to catch up academically to their
native English-speaking peers? ·
Should
reading instruction be provided initially through bilingual students’ first
language? ·
What
outcomes can be expected from different kinds of bilingual programs? ·
How
important is time-on-task, understood as the amount of instructional time spent
through the target language, on second language academic development?” etc. Although
most of these questions are addressed to some extent in the report, the answers
are at best tentative and not particularly helpful for policy-purposes. This
lack of incisiveness can be attributed to the focus on research in isolation
from theory. Crucial theoretical issues such as the nature of language
proficiency and its relationship to academic development are barely considered
in the report. Similarly, the report fails to address the validity of
the competing hypotheses that have been advanced to support either
English-only programs (the “time-on-task” or maximum exposure to English
hypothesis [e.g. Porter, 1990]) or certain forms of bilingual program (the
“linguistic interdependence” hypothesis [Cummins, 1981]). [1] Consequently,
the first question listed above is barely considered in the report despite the
fact that this has been a central issue to emerge in the context of Proposition
227 in California and the many policy debates leading up to this initiative. Answers
are given to some of the other questions; for example, the report states that
“[t]he degree of children’s native-language proficiency is a strong
predictor of their English-language development” (p. 28). However, the authors
fail to address what this finding means for policy.
As I illustrate below, when this finding is integrated with theory in the
form of the linguistic interdependence hypothesis, it provides an
empirically-testable basis for interpreting the outcomes of bilingual education
programs in widely different sociolinguistic and sociopolitical contexts. Similarly,
the authors’ treatment of the
question of learning to read through a second language correctly identifies the
major pattern of research findings
but then lets the issue drop as though the findings were contradictory and few
policy implications could be drawn: With respect to reading instruction in a second
language, there is remarkably little directly relevant research. Clearly, one of
the major intellectual stimuli to bilingual education programs has been the
belief that initial reading instruction in a language not yet mastered orally to
some reasonable level is too great a cognitive challenge for most learners. The
evidence that better academic outcomes characterize immigrant children who have
had 2 to 3 years of initial schooling (and presumably literacy instruction) in
their native countries (Collier and Thomas, 1989; Skutnabb-Kangas, 1979) is
consistent with the claim that children should first learn to read in a language
they already speak. However, it is clear that many children first learn to read
in a second language without serious negative consequences. (pp. 59-60) By
contrast, I would argue that there is a significant amount of directly relevant
research. Reviews of this research from 25 years ago were able to conclude that
the language of initial literacy instruction is not, in itself, a significant
determinant of academic outcomes (Cummins, 1979; Engel, 1975; Wagner, 1998),
although it may play a role as part of a much broader constellation of variables
related to power relations and identity negotiation (Cummins, 1996). Finally,
overwhelming amounts of data on bilingual education programs internationally
(e.g. Cummins & Corson, 1997) should have permitted the authors to state
definitively that the major theoretical argument underlying the push for
all-English programs is without merit. The
time-on-task (maximum exposure) hypothesis predicts that any form of
bilingual education that reduces the amount of instructional time through the
medium of English will result in academic difficulties in English. Or, as
Rosalie Pedalino Porter (1990) has expressed it: "The evidence of direct
correlation between early, intensive second-language learning and high level of
competence in the second language is inescapable, as is the on-task principle -
that is, the more time spent learning a language, the better you do in it, all
other factors being equal" (p. 119). As a theoretical proposition, this
hypothesis is refuted by the outcomes of countless
bilingual programs evaluated in countries around the globe which demonstrate
that students suffer no adverse effects in their mastery of the majority
language (English in North America) as a result of spending significant
instructional time through the minority language (see Corson, 1993; Cummins,
1996, for reviews). The
policy implication is not that bilingual programs are necessarily
“effective,” or will necessarily succeed better than alternative programs.
Outcomes of any program will depend on a variety of implementation factors;
rather the data and associated theory show clearly that linguistic minority and
linguistic majority students in well-implemented bilingual programs (of various
types) will suffer no adverse consequences as a result of spending instructional
time through both languages. This
type of clear policy-relevant conclusion is not forthcoming from the report, not
because relevant data are lacking, but because the data are viewed through a
very fuzzy lens. The authors’
conclusion seems designed to provide policy-makers on both sides of the
bilingual education issue with what they want to hear: The beneficial effects of native-language instruction
are clearly evident in programs that have been labeled “bilingual
education,” but they also appear in some programs that are labeled
“immersion” (Gersten and Woodward, 1995). There also appear to be benefits
of programs that are labeled “structured immersion” (Baker and de Kanter,
1981; Rossell and Ross, 1986); (p. 147) Theory
is required to bring the data into focus. As
I illustrate below, definitive answers could have been provided to many of the
most contentious policy-related questions in the area of bilingual education had
the authors articulated the major theoretical positions in the literature on
bilingual education and examined the consistency of these positions with the
available data. Rossell
and Baker (1996) The
outcomes of Rossell and Baker’s review of the literature on the educational
effectiveness of bilingual education are clearly stated in the Abstract: The research evidence indicates that, on standardized
achievement tests, transitional bilingual education (TBE) is better than
regular classroom instruction in only 22% of the methodologically acceptable
studies when the outcome is reading, 7% of the studies when the outcome is
language, and 9% of the studies when the outcome is math. TBE is never better
than structured immersion, a special program for limited English proficient
children where the children are in
a self-contained classroom composed solely of English learners, but the
instruction is in English at a pace they can understand. (p. 1) Furthermore,
the comparisons of reading scores between TBE and structured immersion showed
that structured immersion was
superior in 83% of cases and no differences were observed in 17%.
These
conclusions, published in a reputable refereed journal, and apparently based on
rigorous methodological criteria, would cause any policy-maker to question the
merits of transitional bilingual education. Cracks
appear very quickly, however, in the facade of objective rationality that this
review of the literature projects. One problem is immediately obvious: When we
look more closely at the research studies that supposedly demonstrated the
superiority of “structured immersion” over “transitional bilingual
education” it turns out that 90% of these studies are interpreted by
their authors as supporting the effectiveness of bilingual and even trilingual
education. Seven
of the ten studies that Rossell and Baker claim support structured immersion
over TBE were studies of French immersion programs in Canada.
Typically, in these programs English-speaking students are
"immersed" in French (their second language [L2]) in kindergarten and
grade 1 and English (L1) language arts are introduced in grade 2.
The proportion of English instruction increases to about 50% by grades 4
or 5. The closest equivalent to the
program in the United States is dual language immersion which has repeatedly
demonstrated its effectiveness for both majority and minority language students
(e.g. Christian et al., 1997; Dolson & Lindholm, 1995; Thomas & Collier,
1997). Note that, as in the U.S.
dual language programs, Canadian French immersion programs are bilingual
programs, taught by bilingual teachers, and their goal is the development of
bilingualism and biliteracy. Even
at the level of face validity, it seems incongruous that Rossell and Baker use
the success of the Canadian French-English bilingual programs to argue for
monolingual immersion programs taught largely by monolingual teachers with the
goal of developing monolingualism. This
is particularly the case since two of the seven programs they cite as evidence
for monolingual structured immersion were actually trilingual programs
involving instruction in French, English, and Hebrew (Genesee & Lambert,
1983; Genesee, Lambert, & Tucker, 1977). In
addition to these seven French immersion program evaluations, one of the ten
studies (Malherbe, 1946) was an extremely large-scale study of Afrikaans-English
bilingual education in South Africa involving 19,000 students.
The other two were carried out in the United States (Gersten, 1985; Peña-Hughes
& Solis, 1980). The
Peña-Hughes and Solis program (labelled "structured immersion" by
Rossell and Baker) involved an hour of Spanish language arts per day and was
viewed as a form of bilingual education by the director of the program (Willig,
1981/82). I would see the genuine promotion of L1 literacy in this program as
indicating a much more adequate model of bilingual education than the quick-exit
transitional bilingual program to which it was being compared. Gersten's study
involved an extremely small number of Asian-origin students (12 immersion
students in the first cohort and nine bilingual program students, and 16 and
seven in the second cohort) and hardly constitutes an adequate sample upon which
to base national policy. Malherbe's
study concluded that students instructed bilingually did at least as well in
each language as students instructed monolingually despite much less time
through each language. He argues strongly for the benefits of bilingual
education (however, see Krashen, 1999b, for a critique of the design of this
study). In
short, Rossell and Baker’s conclusions are immediately suspect as a result of
the fact that they use the documented success of bilingual and trilingual
programs to argue against bilingual education.
There are many other problems with their literature review which bring
the entire enterprise of basing policy decisions primarily on
“methodologically acceptable” treatment-control group studies into question.
Some of the problems are briefly outlined below (see also Cummins, 1999; Dicker,
1996; Escamilla, 1996; Krashen,
1996) The
criteria for deciding which studies are “methodologically acceptable” are
unclear and are applied in an arbitrary manner. Krashen
(1999b), for example, points out significant design problems with Malherbe’s
study and, as noted above, the AIR study confounds the experimental and control
treatments since both groups experienced some (unknown) forms of bilingual
education. The
labels assigned to different programs are arbitrary and applied in an
inconsistent manner. For example, Rossell and
Baker claim to compare French immersion (structured immersion) programs in
Canada with “transitional bilingual education.” There are no transitional
bilingual education programs in Canada. The El Paso Independent School District
(1987) program was labeled a “Spanish-English dual immersion” program by
Baker in 1992 and a “structured English immersion” program by Baker in 1998,
and a “submersion” program by Rossell and Baker (1996—Appendix C, p. 72).
This program involved a “native language cognitive development”
component of 90 minutes a day at grade 1, gradually
reducing to 60 minutes a day by grade 3 and 30 minutes a day by grade 4 (Gersten
& Woodward, 1995; Krashen, 1996). Limiting the framework of discourse to exclude bilingual programs
designed to promote bilingualism and biliteracy. An additional example of arbitrary labeling
is their treatment of Legaretta’s (1979) kindergarten study.
Originally labeled a “structured immersion” program by Baker and de
Kanter (1983), this study demonstrated the superiority of a 50% Spanish, 50%
English kindergarten program over both English-only and other bilingual program
options with respect to students’ learning of English. Rossell and Baker
(1996) list this study as showing “no difference” between TBE and submersion
(English-only) treatments. Yet the
program option that was significantly better than all others was neither TBE nor
submersion! The consistently positive outcomes of this kind of “Enriched
Education” program (Beykont, 1994; Cloud, Genesee, & Hamayan, in press)
are nowhere represented in Rossell and Baker’s review.
By limiting the framework of discourse to “transitional bilingual
education” versus varieties of English-only programs, they have excluded the
type of dual-language program option endorsed by virtually all applied linguists
and also by some academics who have been highly critical of
bilingual education (Glenn & LaLyre, 1991; Porter, 1990). There
appears, in fact, to be an emerging consensus among advocates and opponents of
“bilingual education” that dual language and other programs that aspire to
bilingualism and biliteracy are effective in developing English academic skills
among both linguistic minority and majority students (Cummins, in press). Rossell
and Baker’s reporting of French immersion data is blatantly inaccurate. In
response to critiques from Kathy Escamilla (1996) and Susan Dicker (1996)
regarding the fact that French immersion programs are fully bilingual in both
goals and implementation, Rossell (1996) pointed out: In the first two years, the program is one of total
immersion, and evaluations conducted at that point are considered to be
evaluations of “structured immersion.”
It is really not important that, in later years, the program becomes
bilingual if the evaluation is being conducted while it is still and always has
been a structured immersion program" (1996, p. 383) The
significance of this point is that the major empirical basis of Rossell and
Baker's entire argument for structured English immersion rests, according to
their own admission, on the performance in French of English-background students
in the first two years of Canadian
French immersion programs. They
interpret this research as follows: Both the middle class and
working class English-speaking students who were immersed in French in
kindergarten and grade one were almost the equal of native French-speaking
students until the curriculum became bilingual in grade two, at which point
their French ability declined and continued to decline as English was increased.
(p. 22) Rossell
and Baker seem oblivious to the fact that at the end of grade one French
immersion students are still at very early stages in their acquisition of
French. Despite good progress in learning French (particularly receptive skills)
during the initial two years of the program, they are still far from native-like
in virtually all aspects of proficiency - speaking, listening, reading, and
writing. Most grade 1 and 2
French immersion students are still incapable of carrying on even an elementary
conversation in French without major errors and insertions of English.
Similarly, it is bizarre to claim, as Baker and Rossell do without even a
citation to back it up, that the French proficiency of grade 6 immersion
students is more poorly developed than that of grade 1 students, and to
attribute this to the fact that L1 instruction has been incorporated in the
program. The
research data show exactly the opposite pattern to that claimed by
Rossell and Baker. Lambert and Tucker (1972), for example, report highly
significant differences between grade 1 immersion and native French-speaking
students on a variety of vocabulary, grammatical and expressive skills in
French, despite the fact that no differences were found in some of the
sub-skills of reading such as word discrimination. By the end of grade four,
however, (after 3 years of English [L1] language arts instruction), the
immersion students had caught up with the French controls in vocabulary
knowledge and listening comprehension, although major differences still remained
in speaking ability. Similarly, in
the United States, the one large-scale "methodologically acceptable"
study that investigated this issue (Ramirez, 1992) found that early-grade
students in "structured immersion" were very far from grade norms in
English even after four years of English-only immersion. In summary, to claim that two years of
immersion in French in kindergarten and grade 1 results in almost native-like
proficiency in French in a context where there is virtually no French exposure
in the environment or in school outside the classroom flies in the face of a
massive amount of research data.
This can be verified by anyone who cares to step into any of the thousands of
grade 1 French immersion classrooms across Canada. In
conclusion, Rossell and Baker’s literature review is characterized by
inaccurate and arbitrary labeling
of programs, inconsistent application of criteria for “methodological
acceptability,” and highly inaccurate interpretation of the results of early
French immersion programs. Ironically, the data they review (both
methodologically “acceptable” and “unacceptable”) does have considerable
relevance for policy purposes, if the interpretive paradigm is changed. The
Research-Theory-Policy Paradigm: Progressive Refinement of Theory to Explain and
Predict Phenomena In most scientific
disciplines, knowledge is generated not by evaluating the effects of particular
treatments under strictly controlled conditions but by observing phenomena,
forming hypotheses to account for the observed phenomena, testing these
hypotheses against additional data, and gradually refining the hypotheses into
more comprehensive theories that have broader explanatory and predictive power.
Take just one example: meteorology or climatology–the understanding and
prediction of weather patterns. What
scientists do to generate knowledge in this discipline (and many others) is to
observe phenomena (e.g. the conditions under which hurricanes
appear) and build up theoretical models that attempt to predict these
phenomena. With further observations they test and refine their predictive
models. There is no control group,
for obvious reasons, yet theory-based predictions are constantly being tested
and refined. In the same way, I
would suggest that a much wider body of research data is both theoretically- and
policy-relevant than typical reviews in the area of bilingual education have
suggested. For example, case
studies of particular programs or evaluations that assess student progress in
relation to grade norms are potentially theoretically-relevant. They become
relevant for theory and policy when their outcomes are assessed in relation to
the predictions derived from particular hypotheses or theoretical frameworks.
The process is as follows: 1.
Establish that the phenomenon is genuine and not an artifact
of measurement or observational procedures (e.g. data collection errors). 2.
Once the phenomenon has been established as genuine, ask
what theoretical constructs can potentially account for the data. 3.
The third stage involves examining hypotheses in relation to
additional data (e.g. designing research to test the hypotheses explicitly). As
noted above, it takes only one contrary finding to refute a theoretical
proposition or cause it to be modified. 4.
The final stage involves continual refinement of hypotheses
and increasing integration into broader theoretical frameworks capable of more
comprehensive explanations and accurate predictions. There is nothing new in any
of this. It reflects, for example,
the process whereby we came to understand the movement of the planets and
countless other scientific phenomena. Why then has this process not been applied
in the recent policy-oriented literature reviews relating to bilingual
education? Had this process been
applied, a much clearer picture of the research findings and their implications
would have emerged. The three literature reviews outlined above would have been
able to communicate to policy-makers and the general public the following
answers to at least some of their central questions: ·
In response to the relatively unsophisticated question,
“Does bilingual education work,” the research shows clearly that successful
bilingual education programs have been implemented in countries around the world
for both linguistic minority and majority students and exactly the same patterns
are observed in well-implemented programs: students do not lose out in their
development of academic skills in
the majority language despite spending a considerable amount of instructional
time through the minority language. This pattern is demonstrated in the vast
majority of the 300 studies listed by Rossell and Baker (1996) as well as in the
broader reviews of literature undertaken by August and Hakuta (1997) and Cummins
and Corson (1997). These data are consistent with predictions derived from the
interdependence hypothesis which suggests that this theoretical construct can be
used as a predictive tool by policy-makers. ·
In response to the question, “Does bilingual education
work better than English-only instructional programs?” no definitive answers
can be given until the term “bilingual education” is defined more precisely.
The trend in much of data is that programs that aspire to develop
bilingualism and biliteracy (Enriched Education programs) show much better
outcomes than English-only or quick-exit transitional bilingual programs that do
not aspire to develop bilingualism and biliteracy. Specific hypotheses (e.g.
regarding the positive effects of bilingualism on cognitive and linguistic
functioning) and more comprehensive theoretical frameworks (e.g. Ovando &
Collier,1998; Cummins, 1996; Lucas, Henze, & Donato, 1990) have been
advanced that are consistent with this trend for Enriched Education programs to
show highly positive outcomes. However, considerably more research is required
to refine these frameworks to take account of the multiple interactions that
occur among variables that contribute to bilingual students’ academic success. ·
In response to the question, “Will students suffer
academically if they are introduced to reading in their second language?” the
research indicates that the language of initial introduction of reading is not,
in itself, a determinant of academic outcomes. The linguistic mismatch
hypothesis therefore has no credibility, as was evident in the 1970s (Engle,
1975; Cummins, 1979). ·
In response to the question, “Will greater amounts of
English instruction (time-on-task) result in greater English achievement?” the
answer is simply, no. The research data overwhelmingly fails to show any
positive relationship between the amount of English instruction in a program and
student outcomes. In arguing for the
theoretical and policy relevance of research findings that report student
outcomes in relation to grade norms without direct control group comparisons, I
am not suggesting that individual studies in isolation provide any definitive
information. Rather the findings of
individual evaluations or research studies represent phenomena that require
explanation. Specific conditions in any particular context may have contributed
to program outcomes such that similar findings are not observed in contexts
where these conditions are absent. What we can state is that a particular set of
findings is consistent or inconsistent with hypothesis X. However, before much
credibility can be placed in the general relevance of hypothesis X, it is
necessary to assess its consistency with a much wider set of findings in
contexts where a variety of other unique conditions may be present.
If the predictions that derive from hypothesis X are confirmed across
these diverse contexts, then the credibility of hypothesis X increases
significantly despite the fact that control group comparisons may not have been
carried out. Let
us take a hypothetical example. Suppose that dual language or two-way bilingual
immersion programs (which usually have between 50% and 90% minority language
instruction in the early grades) were to show consistently the pattern that most
of those that have been evaluated to this date apparently do show: by the end of
elementary school, students from majority language backgrounds develop high
levels of biliteracy skills at no cost to their English (L1) academic
development; students from minority language backgrounds by grade 6 show above
average L1 literacy development and come close to grade norms in English (L2)
academic skills. Let us suppose, hypothetically, that we have 100 such programs
demonstrating this pattern from around the United States and the few programs
that do not demonstrate this pattern can be shown to have been poorly
implemented or not to have followed the prescribed model in some important
respects. Do
these 100 programs demonstrating a consistent pattern of achievement in relation
to grade norms tell us anything that is policy relevant? Rossell and Baker would
say no—these studies are not “methodologically acceptable” because control
groups were not used and results are reported only in relation to grade norms. I
have argued, by contrast, that such a pattern is directly relevant to policy
because it permits us to test certain theoretical predictions against the
research data. Thus, the hypothetical pattern described for both minority and
majority students is clearly inconsistent with the "time-on-task"
hypothesis because students instructed through the minority language for
significant parts of the school day suffered no adverse effects in English
language academic development. These
data would also refute the linguistic mismatch hypothesis since majority
language students whose initial literacy instruction was through Spanish
experienced no long-term difficulty in either English or Spanish literacy
skills. This pattern of data, however, would be consistent with the
interdependence hypothesis which predicts that instruction through a minority
language will result in no adverse consequences for academic development in the
majority language. In conclusion, the
alternative Research-Theory-Policy paradigm conforms more closely to typical
scientific procedures than the virtual elimination of theoretical considerations
in the Research-Policy paradigm that has dominated the recent bilingual
education policy debate. Not
surprisingly, it also yields information for policy that is much more
interpretable and useful. Experimental and quasi-experimental research is an
appropriate approach to inquiry but
by itself is limited in its ability to answer the major policy-related questions
in the education of linguistic minority students, as the inadequacies of the
three literature reviews analyzed above illustrate. Footnotes
1.
The interdependence principle has been stated as follows (Cummins, 1981, p. 29): To the extent that instruction in Lx is
effective in promoting proficiency in Lx, transfer of this proficiency to Ly
will occur provided there is adequate exposure to Ly (either in school or
environment) and adequate motivation to learn Ly. 2.
Rossell
and Baker (1996) do acknowledge the existence of the interdependence hypothesis
but they distort it beyond recognition, attributing to me a “facilitation
theory” (a label which I have never used) which predicts that minority
students taught through their L1 should always perform better in English than
students taught exclusively through English regardless of the conditions or
sociocultural context. This is a very different prediction than that which
derives from the interdependence hypothesis which is that the transfer of
conceptual and linguistic knowledge across languages can compensate for the
significantly reduced instructional time through the majority language (see
Cummins, 1999). References
August, D. & Hakuta, K. (1997). Improving
schooling for language minority children: A research agenda. Washington, DC:
National Academy Press. Baker, K. A.& de Kanter, A. A.
(1981). Effectiveness of bilingual education: A review of the literature. Washington,
DC: U.S. Department of Education. Baker, K. (1992, Winter/Spring). Review
of Forked Tongue. Bilingual Basics, pp. 6-7. Baker, K. (1998). Structured English immersion:
Breakthrough in teaching limited-English-proficient students. Phi Delta
Kappan (November), 199-204. Beykont, Z. F. (1994). Academic
progress of a nondominant group: A
longitudinal study of Puerto Ricans in New York City's late-exit bilingual
programs. Doctoral dissertation presented to the Graduate School of Education of
Harvard University. Christian, D., Montone, C. L., Lindholm,
K. J., & Carranza, I. (1997). Profiles in two-way immersion education. Washington,
DC: Center for Applied Linguistics and Delta Systems. Collier, V. P. & Thomas, W. P.
(1989). How quickly can immigrants become proficient in school English? Journal
of Educational Issues of Language Minority Students, 5, 26-38. Corson, D. (1993). Language,
minority education and gender: Linking
social justice and power. Clevedon, England: Multilingual Matters. Cummins, J. (1979). Linguistic
interdependence and the educational development of bilingual children.
Review of Educational Research, 49, 222-25l. Cummins, J.
(1981). The role of primary
language development in promoting educational success for language minority
students. In California State
Department of Education (Ed.), Schooling and language minority students: A
theoretical framework. (pp. 3-49). Los Angeles: National Dissemination and
Assessment Center. Cummins, J. (1996). Negotiating
identities: Education for empowerment in a diverse society. Los Angeles:
California Association for Bilingual Education. Cummins, J. (1999). Beyond adversarial discourse:
Searching for common ground in the education of bilingual students. In I. A.
Heath & C. J. Serrano (Eds.), Annual editions: Teaching English as a
second language (pp. 204-224). Guildford, CT: Dushkin/McGraw-Hill. Cummins, J. (in press). The ethics of doublethink:
Language rights and the bilingual education debate. TESOL Journal. Cummins, J. & Corson, D. (Eds.). (1998). Bilingual
education. Dordrecht, The Netherlands: Kluwer Academic Publishers. Danoff, M. N. (1978). Evaluation of the impact of
ESEA Title VII Spanish-English bilingual education programs. Technical
Report. Washington, DC: American Institutes for Research. Dicker, S. (1996). RTE Forum: Letters from readers. Research
in the Teaching of English, 30(1), 373-376. Dolson, D. & Lindholm, K. (1995).
World class education for children in California:
A comparison of the two-way bilingual immersion and European Schools model. In
T. Skutnabb-Kangas (Ed.) Multilingualism for all. (pp. 69-102). Lisse:
Swets & Zeitlinger. El Paso Independent School District
(1987). Interim report of the five-year bilingual education pilot 1986-87
school year. El Paso, Tx: Office for Research and Evaluation. Engle, P. (1975). The use of the
vernacular language in education. Bilingual education. (Series No. 2).
Washington, DC: Center for Applied Linguistics. Escamilla, K. (1996). RTE Forum:
Letters from readers. Research in the Teaching of English, 30(1),
371-373. Fitzgerald, J. & Cummins, J. (in
press). Bridging disciplines to critique a national research agenda for
language-minority children’s schooling. Reading Research Quarterly,
34(3). Gandara, P. (1997). Review of
research on instruction of limited English proficient students. Davis:
University of California Linguistic Minority Research Institute Education Policy
Center, Davis Genesee, F., & Lambert, W. (1983).
Trilingual education for majority-language children. Child Development, 54,
105-114. Genesee, F., Lambert,
W., & Tucker, G. (1977). An experiment in trilingual education. Montreal:
McGill University Gersten, R. (1985). Structured immersion for language minority students: Results
of a longitudinal evaluation. Educational
Evaluation and Policy Analysis, 7, 187-196. Gersten, R., & Woodward, J. (1995). A
longitudinal study of transitional and immersion bingual education programs in
one district. The Elementary School Journal, 95(3), 223-239. Glenn, C. L. & LaLyre, I. (1991). Integrated
bilingual education in the USA. In K. Jaspaert & S. Kroon (Eds.), Ethnic
minority languages and education (pp. 37-55). Amsterdam: Swets &
Zeitlinger. Greene, J. (1998). A meta-analysis of the
effectiveness of bilingual education. Claremont, CA: Tomas Rivera Policy
Institute. Krashen, S.D. (1996). Under attack:
The case against bilingual education. Culver City, CA: Language Education
Associates. Krashen, S. D. (1999a). Condemned without a trial:
Bogus arguments against bilingual education. Portsmouth, NH: Heinemann. Krashen, S. D. (1999b). Why Malherbe (1946) is NOT evidence against bilingual
education. Manuscript submitted for publication. Lambert, W.E. & Tucker, G.R. (1972). Bilingual
education of children: The St. Lambert Experiment. Rowley, Mass.:
Newbury House. Legaretta, D. (1979). The effects of program models
on language acquisition by Spanish speaking children. TESOL Quarterly, 13,
521-534. Lucas, T., Henze, R., & Donato, R.
(1990). Promoting the success of Latino language-minority students: an
exploratory study of six high schools. Harvard Educational Review, 60,
315-340. Malherbe, E. G. (1946). The bilingual school.
Johannesburg: Bilingual School Association. Ovando, C. J. & Collier, V. P. (1998). Bilingual
and ESL classrooms: Teaching in
multicultural contexts. Boston:
McGraw-Hill. Peña-Hughes,
E. & Solis, J. (1980). ABCs (unpublished report). McAllen, TX:
McAllen Independent School District. Porter, R. P. (1990). Forked tongue:
The politics of bilingual education. New York: Basic Books. Ramirez, J.D. (1992). Executive
summary. Bilingual Research Journal, 16, 1-62. Rossell, C. H.
(1996). RTE Forum: Letters from readers. Research in the Teaching of English,
30, 376-385. Rossell, C. H. & Baker, K. (1996).
The effectiveness of bilingual education. Research in the Teaching of
English, 30, 7-74. .
Rossell, C. H. & Ross, J. M. (1986). The social science evidence on
bilingual education. Journal of Law and Education, 15(4), 385-419. Skutnabb-Kangas, T. Language in the
process of cultural assimilation and structural incorporation of linguistic
minorities. Arlington, VA: National Clearinghouse for Bilingual Education. Thomas, W. P. & Collier, V. P.
(1997). School effectiveness for language-minority students. Washington,
DC: National Clearinghouse for Bilingual Education. Wagner, D. A. (1998). Putting second
language first: Language and literacy learning in Morrocco. In L. Verhoeven & A. Y. Durgunoglu (Eds.), Literacy
development in a multilingual context: Cross-cultural perspectives. (pp.
169-184). Mahwah, NJ: Lawrence Erlbaum Associates. Willig, A. C. (1981/82). The
effectiveness of bilingual education: Review of a report. NABE Journal,
6, 1-19. Willig, A. C. (1985). A meta-analysis
of selected studies on the effectiveness of bilingual education. Review of
Educational Research, 55(3), 269-317.
| ||||||||||||||||||
