DOCUMENT RESUME

ED 337 034 FL 019 748

AUTHOR Krashen, Stephen D.TITLE Bilingual Education: A Focus on Current Research.

Focus, Occasional Papers in Bilingual Education,Number 3.

INSTITUTION National Clearinghouse fc.r Bilingual Education,Washington, DC.

SPONS AGENCY Office of Bilingual Education and Minority LanguagesAffairs (ED), Washington, DC.

PUB DATE 91CONTRACT 289004001NOTE 18p.PUB TYPE Reports - Descriptive (141) -- Collected Works -

Serials (022)JOURNAL CIT Focus; n3 Spr 1991

EDRS PRICE MF01/PC01 Plus Postage.DEVCRIPTORS *Bilingual Education; Comparative Analysis;

Elementary Secondary Education; *English (SecondLanguage); Immersion Programs; *Language Aoquisition;Language Maintenance; *Language of Instruction;Language Research; *Limited English Speaking; NativeLanguage Instruction; *Second Language Programs

ABSTRACTThe core of the case for bilingual education is that

the principles underlying successful bilingual education emphasizethe same factors that underlie successful language acquisition ingenersa: understanding messages and obtaining comprehensible input;using background knowledge; and the development of literacy. Theimportance of the first language is discussed, and arguments againstbilingual education (including the case for immersion--andsubmersion--programs as alternatives for limited English proficientchildren) are examined and rebutted. An argument is made forstrengthening bilingual education by providing a richer supply ofprimary language reading materials. Sheltered subject matter teachingis also supported as a contributor to second language development andas a valuable part of bilingual education. It is concluded thatproperly organized bilingual programs work. Contains 46 references.(LB)

********************** ***** ********************************************

Reproductions supplied by EDRS are the best that can be madefrom the original document.

*******************Xit**************R***********************************

OcceekmaiPaperInWawa!Edwatlen

Wing 1991Holm 3

Bilingual

Education:ittrA FocusettOn Current

Researchcrz

c.4

ek.

Stephen D. Krashen

U I DEPARTMENT OF Et:WATS:MIO. o( Etkeal110r44 Resspottrt and intotooemeotEDL4CATtONAI RESOURCES

INFORMATfOteCENTER tERCITnd document hei Men reproducedreceived I rOM tne wean or &gem:idly,ondmatewa If

r Mime chances have bun mad to impovrKVOCkleifOn quIilv

Peon of roe* 0Ø'!Onhiiidnlh,sOociimillmf do ooi rmweNsanly,epreaeht hthc,aiOE Rt groiriron or raowy

"PERMISSION TO REPRODUCE THISMATERIAL HAS SEEN GRANTED SY

\41,

TO THE EDUCATIONALRESOURCES

INFORMATION CENTER tERIC)"

National CleannghouseSof Bdingual EducaSion

NCBE BEST COPY MUNI

Itk

Ii

The National Clearinghouse for Bilingual Education(NCBE) is funded by the U.S. Department of Education'sOffice of Bilingual Education and Minority LanguagesAffairs (OBEMLA) and is operated under Contract No.289004001 by The George Washington University'sCenter for the Study of Education and NationalDevelopment, jointly with the Center for AppliedLinguisics The contents of this publication do notnecessarily reflect the views or policies of theDepartment of Education, nor does the mention of tradenames, commercial products, or organizations implyendorsement by the U.S. Government. Readers are free

to duplicate and use these materials in keeping withaccepted publication standards. NCBE requests thatproper credit be given in the event of reproduction.

National Clearinghouse for Bilingual Education1118 22nd Street NWWashington DC 20037

3

2

The Research

SupportingBilingualEducation

The core ofthe case for

bilingual education is that the principles under-lying successful bilingual education are thesame principles that underlie successful lan-guage acquisition in general. These principlesare:

(1) We acquire a second language by under-standing messages, by obtaining comprehen-sible input.

(2) Backgrourrl knowledge can help makesecond language input more comprehensible,and can thus assist in the acquisition of thesecond language.

(3) The development of literacy occurs in thesame way as second language acquisitiondoes. As Goodman (1982) and Smith (1982)have put it, 'We learn to read by reading," bymaking sense of what is on the page. In turn,reading Is the major source of our compe-tence in vocabulary, spelling, writing style, andgrammar. (For reseach evidence supportingthese principles, see Goodman, 1982; Smith,1982; Krashen, 1985a, 1985b, in press.)

The importance of the first language

One of the most salient features of a bilingualeducation program is the use of the first lan-guage as the medium of instruction. The firstlanguage can help in the following ways:

(1) It supplies background knowledge, whichcan make English input more comprehensible.

(2) It enhances the development of basic liter-acy. This is a two-step argument:

a. If we, in fact, learn to read by readinn, itwill be much easier to learn to read in alanguage one already knows, since writtenmaterial in that language will be morecomprehensible.

b. Once you can read, you can read. Thisability transfers to other languages thatmay be acquired.

(3) 11 helps in what we call "advanced liter-acrthe ability to use language, oral andwritten, to solve problems. If students under-stand the composing process in one lan-guage, for example, they will be able to utilizeit in other languages they acquire. (For sup-porting data and detail, see Cummins, 1981;Krashen, 1985a, 1985b, 1990.)

There are other ways the first language canhelp. Research evidence suggests thatadvanced first language development hascognitive advantages (see, e.g., Hakuta,1986), practical advantages (Simon, 1980),and promotes a healthy sense of biculturalism(Cummins, 1981).

Explaining bilingual education:"The Paris argument"

It is not easy to explain the theory underlyingbilingual education to the public. I have hadsome success, however, with the followingexplanation, which I call "The Paris Argu-ment." Pretend that you have just received,and accepted, an attractive job offer in Paris.Your French, however, is limited (you had twoyears of French in high school and onesemester in college, and it was quite a whileago). Before your departure, the companythat is hiring you will send you the followinginformation, in English: What to do when youarrive in Paris, how to get to your hotel, whereand how to find a place to five, where to shop,what kinds of schools are available for yourchildren, how French companies function(how people dress in the office, what timework starts and ends, etc.), and specific infor-mation about the functioning of the companyand your responsibilities.

It would be very useful to get this informationright away in English, rather than getting itgradually, as you acquire French. If you get itright away, the world around you will be muchmore comprehensible, and you will thusacquire French more quickly.

Anyone who agrees with this, in my opinion,agrees with the philosophy underlying bilin-gual education.

Criticisms of Bilingual Education

The arguments most Often heard against bilin-gual education are: (1) it doesn't workresearch on tglingual programs is inconsistentand contradictory, and (2) there is a betteroptionimmersion.

There is a widespread perception thatresearch shows bilingual education is a failureand that students in bilingual education pro-grams do not acquire proficiency in English.What are the facts?

Comparing bilingual achreation andall-English medium programs

First, when *unexamined" bilingual programsare compared to 'submersion" programs (pro-grams using English as the medium of instruc-tion with no special modification for meetingthe linguistic needs of limited English profi-cient (LEP) students), or submersion with pull-out English as a Second Language (whereLEP students are separated from English pro-ficient students for some part of the schoolday in order to receive English as a SecondLanguage (ESL) instruction), there is typicallyno difference in terms of Ervalish languageachievement. ("An unexamined" program issimply one that investigators have labeled"bilinguar without providing additional detailabout how the program was organized. Such

programs may or may not be fully consistentwith the principles discussed earlier.)

Unfortunately, the results of these compari-sons have been interpreted as negative, asshowing that bilingual education doesn't work.Some of the problem is how the results areexpressed. Rossell arxi Ross (1986), forexample, reported the following resutts forEnglish language achievement when transi-tional bilingual education and submersion pro-grams are compared:

number of studies

tansitional bilingual education better

no difference

submersion better

8

14

6

Rossell and Ross concluded that "seventy-one percent (of the studies) show transitionalbilingual eckrcation TBE) programs to be nodifferent or worse" than submersion (p. 399).This is, of course, true-4n 20 out of 28 cases,TBE is no better or worse. But one could alsoconclude that TBE is just as good, if not bet-ter, than submersion programs 79 percent ofthe time (22 out of 28 cases).

Several scholars have pointed out that whenbilingual education is shown to be just aseffective as all-English programs, this is aremarkable result since it means that the chil-dren have acquired just as much English withsignificantly less exposure to English. Thisconfirms the underlying theory of bilingualeducation. Some of the criticsbut not ail--have missed this point entirely (see Rosselland Ross, 1986, p. 407, for this discussion).

3

Table 1

Comparison of"unexamined"transitionalWilngual educationand submersion

programs

4

A recent study(Rosselit inpress) meritsdetaileddiscussionbecause of theimportance

attached to ils conclusions in the popularpress. Rossell compared a program labeledbilingual education and a puli-out ESL pro-gram in the Berkeley (California) UnifiedSchool District. No description of Berkeley'sbilingual education program was provided,other than the fact that it is labeled "bilingualeducation- and that instruction was in Spanish30 to 50 percent of the time. Rossell con-cluded, after a series of regression analyses,that there was no difference between the twoprograms. Interestingly, there is evidence of aslight superiority for bilingual education. In herTable 1 (IDEA Proficiency Test scores for stu-dents in grades K-12), the regression coeffi-cient for participation in bilingual education ispositive and reaches the .10 level for a one-tailed test (Rossell would probably call for atwo-tailed test here, however).

In another analysis Rossell compared Califor-nia Test of Basic Skills (CTBS) scores forbilingual and ESL pull-out students after"reclassification." Rossell rxmcluded thatthese data show no difference between thetwo groups. For each subtest of the CTBS,however, the regression coefficient for partici-pation in bilingual education was positive, andin the case of math, it reached the .05 levelfor a two-tailed test, which Rossell did notindicate.

Rossell also compared Berkeley LEP chil-dren's performance on California Assessmentfor Progress (CAP) tests to performance byLEP children in two districts considered tohave *exemplary" bilingual programs, Frernontand San Jose (Krashen and Biber, 1988).Rossell reported no significant differenceamong the children in the three districts inreading, and reported that the Berkeley stu-dents excelled in math.

There are several problems with this conclu-sion. First, this analysis does not comparegain scores nor does it show how rapidly chil-dren reach norms. It considers LEP childrenas a group. This convarison is only valid if, infact, LEP children in all three districts enteredtheir respective systems at the same level ofcompetence, and if all three districts usedsimilar criteria for exiting children. This maynot be the case. According to Rossell's analy-sis of reclassified children in Berkeley, manychildren scored very well on CTBS longbefore they were exitedin CTBS Reading,for example, children in ESL pull-out scored atthe 33rd percentile two years before reclassifi-cation and at the 54th percentile one yearbefore, while children in bilingual educationwho were reclassified scored at the 35th per-centile two years before reclassification andnear the 60th percentile one year before.CTBS Language data are similar, and scoresin CTBS Math are even highe-, with LEP chil-dren in Berkeley scoring above the 50th per.centile four years before reclassification.Thus, Berkeley scores may look higherbecause some high-scoring children wereretained in these programs longer.

Even if the analysis were a valid oneif chil-dren in all three districts entered at the samelevel and all three districts had equal reclassi-fication criteriaRossi:as results would notnecessarily reflect the quality of bilingual edu-cation in the Berkeley district. The comparisondistricts were chosen becauae their bilingualprograms were exemplary. Yet, the cross-district comparison performed by Rosse II is ofal/LEP children in each district, taken as agroup. A minority of the children in the samplewere in bilingual education. According to Ros-sell's data, only 31 percent of the 108 schoolsstudied had a bilingual education program (33schools). Moreover, it is quite likely that not all

of the LEP children in these 33 schools werein the bilingual program. (Interestingly,according to Rossell's analysis, schools thathad bilingual education reported slightlyhigher CAP scores; the regression coefficient,however, was not significant.)

Several studies have shown that bilingualeducation programs can be as effective as all-English programs. For example, WI llig (1985)performed a meta-analysis on bilingual educa-tion evaluations originally studied by Bakerand de Kanter (1983), and concluded thatthere was a modest overall positive effect forbilingual education, despite the fact that manyof the studies (65 percent) were short-term,lasting one year or less. Wag also found thatbilingual education looked better when theresearch designs used in the studies werebetter. Hero are some examples:

When the comparison group had elementsof bilingual education, there was no differ-ence. But when the comparison group did nothave elements of bilingual education, the bilin-gual group was superior.

When the bilingual program was "unstable"(frequent teacher turn-over, disorganization),the comparison group was better. When thebilingual group was stable, it was superior.

When the comparison group cordained grad-uates of bilingual education, there was no dif-ference. When the comparison group did notcontain graduates of bilingual education thebilingual group was better. As Crawford(1989) put it, Willig showed that on a levelplaying field, bilingual education could out-score the competition.

Recent research shows that when bilingualprograms are set up correctly, they work verywell. In our survey of successful programs in

California (Krashen and Biber, 1988), wefound that etudents in well-designed bilingualprograms consistently outperformed compari-son students, and did very well compared tolocal and national norms, often reachingnational norms between grades three to six.Accorcfing to the view of language acquisitionpresented earlier in this paper, we defined aNve II-designed" program as one that had thefollowing characteristics:

(1) Comprehensible input in English, in theform of high quality ESL classes, and shel-tered subject matter teaching (comprehensi-ble subject matter teaching in the secondlanguage; see below).

(2) Subject matter teaching in the first lan-guage, without translation. This providesbackground knowledge that will make Englishinput more comprehensible.

(3) Literacy development in the first language,which will transfer to the second language.

Our report has been criticized. Imhoff (1990)maintains that the programs in Krashen andBiber (1988) worked because they were in"exenvlary schools that are well-funded,staffed by highly trained and dedicated teach-ers, and compooed of small classes ofselected students" (p. 52). To my knowledge,not all the schools described in our mono-graph were well-funded. The teachers didreceive some extra in-service training in cur-rent theory and methodology, but to say thatthey were more dedicated is not onlyunfounded but is also an insult to teachers inthe comparison groups. Nearly all of the stu-dents in the programs were unselected; thereis no reason to suspect they were differentthan students in comparison groups, andthere is no reason to suspect differences inclass size.

5

6

Bosse II has

alsocriticizedKrashen andBiber (1988)1

pointing outthat one of

the districts we stuclied, Fremont, took otherpositive action for LEP children in their bilin-gual program in addition to bilingual education(preschool, extra English reading, moreparental involvement). While this could meanthat these additional Oft rts were responsiblefor the Fremont children's outstanding perfor-mance, it Is certainly not counterevidence tothe hypothesis that bilirgval education iseffective, a hypothesis that has a great deal ofadditional supporting evidenr J.

Porter (1990a, 1990b), who repeatedly insiststhat bilingual education "just doesn't work"(Porter, 1990a, p. 223), presents severalkinds of arguments. In some cases, she issimply anti-bad bilingual education, attackingpractices that many supporters of bilingualeducation would agree are cwestionable.These Include delaying exposure to writtenEnglish until students reach grade level inSpanish (Porter, 1990a, p. 22), and programsin which teachers are encouraged to code-switch in class (Porter, 1990a, p. 31).

Other attacks are simply unfounded and arenot supported by the data she presents.Porter claims, for examle, that in Boston"several hundred" bilingual education studentshad not learned enough English to be exitedby grade seven (Porter, 1990a, p. 60). Porterdoes not tell us, however, whether this is asmall or large percentage of the total numberof students served (if it is a small percentage,the data would indicate that the program wassuccessful), what kind of bilingual educationwas used, or what the characteristics of thesestudents were.

Another example is her report that only 4000out of 7000 applicants passed an °English-language aptitude test" given by New York'sConsolidated Edison company in 1988. Noneof those who passed, according to Porter,

was a graduate of New York City's bilingualeducation program (Porter, 1990b, p. 24).Without more details, however, it is impossi-ble to draw any conclusions from such astatement. Porter does not tell us, for exam-ple, how many of those who took the testwere ever limited English proficient studentsor how many had even had bilingualeducation.

The Immersion argument

Another popular argument against bilingualeducation is the claim that there is betterwayimmersion. Imhoff (1990) supports thisview: "The language teaching method that isgenerally the fastest, most efficient, and mostetfective is the Ber Utz or immersionmethod. . (p. 50). Imhoff does not presentevidence supporting his view, but could havecited Rosseli and Ross (1986), who claim thatimmersion students outperformed students inbilingual education in English language profi-ciency ill six out of seven studies. Beforereviewing this claim we need to define theterm Immersion." There are, to my knowl-edge, at least four definitions:

(1) Submersion, or "sink or swim." There is nosupport among language education profes-sionals for submersion for LEP children today.

(2) Canadian-style immersion (CSI). As iswell-known, CSI is a program in which middle-class children receive much of their subject-matter instruction through a second language.Efforts are made to make sure the languagethey hear is comprehensible. Children inthese programs learn subject matter success-fully, and acquire a great deal of the secondlanguage.

Consideration of the principles of bilingualeducation presented earlier leads to the

conclusion that CSI is similar, if not identical,to bilingual education. Children in CSI receivecomprehensible input in the second languageand develop literacy and subject matterknoviledge in their first language, both outsideof school and in school. As noted earlier, chil-dren in CSI are typically middle class, and doa considerable amount of reading in Englishoutside of school (suggested by Cummins,1977 and confirmed by Eagon and Cashion,1988). Even in early total immersion pro-grams, a great deal of the curriculum is inEnglish, with English language arts introducedaround grade two. By grade six, half the cur-riculum of early total immersion is in English.Most important, the goal of CSI is bilingual-ism, not the reNacement of one language withanother.

(3) Sheltered subject matter teaching. Shel-tered subject matter teaching was inspired bythe success of CSI. It is subject matter teach-ing done in a second language but madecomprehensible. Research at the universitylevel has confirmed that students in shelteredsubject matter classes acquire impressiveamounts of the second language and teamsubject matter as weil (Edwards, Wesche,Krashen, Clement and Kruidinier, 1983;Lafayette and Buscaglia, 1983; Hauptman,Wesche and Ready, 1988; Stemfeld, 1989).

Sheltered subject matter teaching is not acompetitor to instruction delivered in the firstlanguage, but makes its contribution in a dif-ferent way. As indicated earlier, both shel-tered subject matter teaching and ESLprovide comprehensible input directly, whileteaching in ne first language makes an indi-rect, but powerful contribution by providingbackground knowledge and literacy.

Porter (1990a) approves of sheltered subjectmatter teaching ("content-based language

teaching"), but feels that It cannot be used inbilingual programs that require subjects to betaught in their native language for severalyears" (p. 125). Subject matter instruction andsheltered subject matter teaching can be sen-sibly cornbined, however. In what has beencalled the "Eastman Plan" (Krashen, 1985a,1985b; Krashen and Biter, 1988), beginninglimited English proficient students study coresubject matter subjects in the primary lan-guage. When the students reach the inter-mediate level, subjects mach as math andscience are Oven as sheltered subject matterin English, since these sartjects do not requireas high a level of English language profi-ciency as social studies or language arts.Subjects recpiring more abstract use of lan-guage are still taught in the trst language atthis stage, which will provide additional sub-ject matter knowledge and literacydevelopment

Children with more competence in acwicalicEnglish study math and science in the maistream. At this stage, social studes is tau;htin English using sheltered techniques, withcontinuing literacy development in the firstlanguage. Eventually, all core subjtacts aretaught in English. An example of such a planis presented in Table 2.

Such a plan provides comprehensibleinstruction at every step. It is superior to anall-English program, because it provides con-tinuing development in the first language,which means continuous growth in subjectmatter knowledge and literacy development. Itis better than a program that forces a child tojump from the first language class directly tothe mainstream; with the sheltered class as atransition, the child will acquire a substantialamount of the English academic lhnguageneeded.

9

7

8

Note thataccording tothe plan inTable 2,

childrencontinue onto do

Table 2

Plan combiningESL, shelteredsubject matterteaching, andInstruction Inthe first (home)language

advanced study in their first languagesocialstudies, language arts, or related topics(enrichment). The advantages of advancedfirst language development were listed earlier.This part of the program can take the place ofthe foreign language study option for the for-mer LEP student.

(4) Structured immersion. Also inspired by thesuccess of CSI, structured immersion (SI) issimilar to sheltered subject matter teaching insome ways, but differs in other ways. Asdescribed by Gersten and Woodward (1985),SI has these four characteristics:

1. Comprehensible subject matter instructionto second language acquirers.

2. Use of the first language when necessaryfor explanation, but this is kept to aminimum.

3. Direct instruction of grammar.

4. Pre-teaching of vocabulary.

While the first two characteristics have sup-port in the research literature, there is littleevidence supporting the efficacy of directgrammar instruction (for reviews, seeKrashen, 1984; Hillocks, 1986) and pre-

Level

Beginning

Intermediate

Advanced

Mainstream

Mainstream

art, music, PE

art, music, PE

art, music, PE, math,science

all core subjects

teaching vocabulary has not been found to beconsistently effective (Mezynski, 1983).

Only a few studies of SI have been done. Ger-sten and Woodward (1985) report that chil-dren in SI in Uvalde (Texas) reached the 30thpercentile of the reading comprehensionsubtest of the Metropolitan Achievement Testat the end of grade three. After leaving SI,however, they dropped to the 15th and 16thpercentile in grades five and six (Becker andGersten 1982). While this performance wasbetter than a comparison group, it is still dis-mal. Children at this grade level who havehad proper bilingual education do much better(Krashen and Biber, 1988). (Uvalde childrendid somewhat better on the WRAT readingtest, which emphasizes "decoding skills.")

In a second study of SI, Gersten and Wood-ward (1985) claimed that more LEP childrenin a California school district in SI performedat or above grade level than compafson chil-dren in bilingual education. There were sev-eral serious problems with this study,however. First, no details were provided aboutthe bilingual education program. Second, thesample size was small (28 children in bilingualeducation, 16 in SI). If performance of just afew children had varied slightly, the results of

Sheltered/ESL

ESL

ESL, math, science

ESL, social studies

First Language

all core subjects

language arts, social studies

language arts

enrichment

this study would have looked very different.Third, the study only followed children untilgrade two.

Gersten and Woodward reported that the SIchildren did extremely well In a follow-upstudy, achieving high levels of performance(65th percentile) two years tater, but only twogroups of nine children each were studed,and no comparison with bilingual educationwas made. Sinilarly, SI students studied inGersten (1985) showed very good gains overthe one year they were followed, but no com-parison group was used.

Effectiveness of Immersion programsfor LEP children

We are now ready to return to the claim madeby Rossell and Ross (1986) that immersionwas more effective than bilingual education inEnglish language development in six out ofseven studies. As listed in Rossell and Ross'Table 1 (p. 398), the six studies in whichimmersion was considered superior were:

1. Gersten, 1985;

2. Pella-Hughes and Solis, 1980;

1 Bank and Swain, 1978;4. Bank, Swain and Nwanunobi, 1977;

5. Bruck, Lambert and Tucker, 1977; and

6. Genesee, Lambert and Tucker, 1977.

Study 1 (Gersten, 1985) is the SI studydescribed previously. As noted earlier, it suf-fers from a very small sample size, and nodescription is given of the kind of *bilingualeducation" used for the comparison group.

Study 2 (Pena-Hughes and Solis, 1980) isunpublished, but it is discussed in severalpublished papers. It is a comparison of twoprograms in McAllen (Texas). While Rossell

and Ross label these Immersion" and 'bilin-gual education," Willig (1985, 1987) classifiedthe immersion group as bilingual education,

ng that the Immersion" group had instruc-tion in English in Vie morning and instructionin Spanish reading in the afternoon. In addi-tion, the explicit (pal of the immersion pro-gram was bilingualismdevelopment of bothlanguages.

Also, the group Rossell and Ross label "bilin-gual education" dhi not, apparently, have anideal program. According to an article in theWall Sereet Journal (Schorr, 1983), classes"Were conducted partly in Spanish and partlyin English," suggesting concurrent translation,a method shown to be ineffective (Legarreta,1979). What apparently happened in McAllenis that children in a good bilingual programoutperformed children in a poor bilingualProgram.

The remaining four studies are all studies ofCanadian French immersion. Studies 3 and 4(Balk, Swain and Nwanunobl, 1977; Barikand Swain, 1978) are studies in which earlytotal immersion Is compared to partial immer-sion. In partial immersion, there is less teach-ing in French; from the beginning, somesubjects are taught in English and some inFrench.

Rossell and Ross are not explicit concerningwhy these studies were included, but the ideaseems to be that early total irrunersion is simi-lar to all-English "immersion" for LEP chil-dren, while partial immersion is similar tobilingual education. Since Bank et al., andBatik and Swain show that children in earlytotal immersion acquire more French thanchildren in partial immersion programs,"immersion," it is oancluded, is better thanbilingual education.

9

10

But Canadianearly totalImmersion isnot the sameas anall-Englishimmersion

program for LEP children. In fact, both ver-sions of CSI under consideration here, earlytotal and partial immersion, are quite similar tobilingual education. As noted earlier, much ofthe CSI curriculum is in the first language,English, and children in these programs cometo school with a great deal of literacy develop-ment in the primary language. Since childrenin both programs come to school so well-prepared, it Is reasonable to expect that moroexposure to the second language, French, willresult in more acquisition ot French.

Many LEP students in the United Stales,however, do not come to school with theseadvantages. An WI-second language curricu-lum will be much less comprehensible tothem, even if carefully "sheltered." While shel-tering will clearly help, supplying backgroundknowledge and literacy in the primary lan-guage is a sure way to ensure that instructionin English will be comprehensible. Rosselland Ross are clearly aware of this argument.They point out, in defense of their position,that CSI programs have worked for workingclass raudents as well as middle class stu-dents. A few reports of immersion programsfor working class children have been pub-lished (e.g., Holobrow, Genesee, LambertGastright and Met, 1987). While these chil-dren have done well, evaluations have so farbeen limited to grade two and below. Also, asGenesee (1983) notes, none of these children"can be said to come from destitute or 'hard-core' innw-city areas" (p. 30).

We thus know very little about how well work-ing class children do in second languageimmersion programs and nothing about howwell under-class children would do. What wedo know is that children of lower socioeco-nomic backgrourKi experience less print out-side of school (e.g., Feitelson and Goldstein1986), and that the richness of the printenvironment is related to literacy development

(Krashen, 1985a). We also know that thesechildren do quite well in well-designed bilin-gual programsprograms that provide liter-acy development in the primary language.

Study 5 (Bnick, Lambert and Tucker, 1977)compares children in total immersion (CSI) tonative speakers of French, and thus has nobearing at all on the cpestion of bilingual edu-cation versus immersion. Study 6 (Genesee,Lambert and Tucker, 1977) is listed as anunpublished manuscript. The same authors,however, have published several reports com-paring CSI to "Core French" (see, for cp.:1m-l*, Genesee, Tucker and Lambert, 'In 1.Core French is simply "foreign language in theelementary school," standard foreign lan-guage instruction for about one period perday. If Rossell and Ross are indeed referringto these comparisons, they are, I assume,arguing once again that early total immersionis similar to an all-English Immersion" pro-gram for LEP children in the United States,and, apparently, that core French is similar tobilingual education. Once again, early totalimmersion is similar to bilingual education, notall-English Immersion." Also, it is not hard tosee why Canadian children in early totalimmersion acquire more French than childrenin core French. Children in both programs areequally well-prepared for school, and childrenin early total immersion receive vastly morecomprehensible input in the second language.

There is one more case that warrants discus-sion since it has caused some confusion. Thisis a report that students in the 'bilingualimmersion° program in El Paso (Texas) out-performed transitional bilingual education(TBE) students. AccorcUng to Porter (1990a),in El Paso's bilingual immersion program, Nallsubjects are taught in English, although Span-ish is used occasionally to reinforce a newconcept" (1990a, p. 68). Porter, in fact, in aWashington Post article (Porter, 1990b,

p. 24), refers to bilingual immersion as an'English-language 'immersion' program." ElPaso's bilingual immersion, however, like theMcAllen (Texas) "immersion" program, isclearly bilingual. It contains a "native languagecognitive development" component (NLCD),described by the El Paso Independent SchoolDistrict (1989b) as foliows:

NLCD is taught for 60 to 90 minutes par day.The objective of this component is to developconcepts, literacy, cognition, and critical think-ing skills in Spanish. It is during this period thatinstruction and student4eacher interaction areentirely in Spankih. The more demanding con-tent area concepts are also Introduced duringNCLD, particularly in first grade. (P- 54)

The two programs differed not only in amountof first language use, but also in other Impor-tant ways. The bilingual immersion programemployed the Natural Approach for ESL, awhole language approach to language arts,and sheltered subject matter teaching, whileTBE in El Paso (referred to in the El Pasoreports as SB 477) used a skills-orientedapproach, as described by the El Paso Inde-pendent School District (1987):

It must be understood ti.:4 BIP (bilingualimmersion program) is not an English versionof the SB 477 instructional orogram. SB 477 isbuilt on a philosophy that advocates traditionalconcepts of teaching language . SB 477focuses the child's attention on the details oflanguage such as phonetic sounds and gram-mar rules. (P. 9)

While TBE used some whole language andNatural Approach activities, the most com-monly used materials in TBE were basal textsand workbooks (El Paso Independent SchoolDistrict, 1987, p. 18). According to a 1989report, whole !engage and comprehensibleinput-based methodology had been graduallyintroduced into SB 477 from 1985 to 1987, but

"observations indicate that the changes havenot been fully implemented by SB 477 teach-ers" (El Paso Independent School District,1989a, p. 10).

To summarize, Immersion" in El Paso com-bined instniction in the first language withcomprehensible-input based methodology,similar to the Eastman plan described earlierin the text. The 'bilingual* program (SB 477)used more instruction in the first language,but focused more on skill-building. It appearsthat the bilingual immersion program wasmore consistent with the principles of lan-guage and literacy development and first lan-guage use presented In this paper.

Strengthening Bilingual Education byReading In the Primary Language

Before conckiding, I wish to add my own criti-cism of bilingual education. In my opinion,bilingual programs will not realize their truepotential unless they do a much better job ofproviding a print-rich environment in the pri-mary language. Research Indicates that read-ing, especially free vokmtary reading, is amajor source of both language and literacydevelopment, as well as knowledge. Readingin the primary language will thus providemuch of the "common underlying proficiency"(Cummins, 1981) that helps ensure Englishlanguage development. In addition, a readinghabit in the first language will, most likely,transfer to the second language. Finally, read-ing contributes a great deal to advanced firstlanguage development.

The current situation is not good. The follow-ing excerpt, from informal notes made bySandra Pucci at the University of SouthernCalifornia, illustrates the point.

1 3

12

Ms. Pucci

visiteda book fairat abilingual

/elementarys6hool:

The book fair, an annual event at the school,was going on when I was there, and the kidswere fairly excited, at least the ones whoseparents had given them money to buy a bookor two. I went in to check out the situation, andafter receiving an enthusiastic 'yes' as towhether they were selling books in Spanish,had a look at just what there was. The displayconsisted of eight cases of books of fourshelves each. There was one shelf of Spanishbooks among them. On this shelf were 12 dif-ferent titles, but only two . . appeared to beabove a third grade reading level. There werearound five copies of each book.. . . At thatgiven moment, there were about 20 kids Wingabout, mostly looking at the English shelves

the Spanish books were placed at the bot-tom, and were actually quite hard to get to.

Unfortunately, the situation at this school istypical. The good news is that the sokition isstraightforwarda print-rich environment.

Conclusions

Criticisms of bilingual education, as $.:,ted ear-lier, rest on two assertions: First, it has 'nenclaimed that bilingual programs don't wat-k,and that the evidence for them is inconsistentand contradictory. The research, however, isremarkably consistent. Properly organizedbilingual programs do work, and even "unana-lyzed bilingual programs appear to work atleast as well as all-English programs.

Second, it has been asserted that "immersion"is superior to bilingual education. This has notbeen demonstrated. As we have seen, thereare several definitions of the term "immer-sion." One of them, submersion, is rejected byall professionals as an option for LEP chil-dren. Canadian-style immersion is quite simi-lar to bilingual education. Sheltered subjectmatter teaching makes a different kind of con-tribution to second language development,

and is a valuable part of bilingual education.The research support for structured immer-sion is, at best. mixed. Finahy, much of theresearch claiming to show that immersion issuperior to bilingual education actually con-sists of comparisons of (Afferent kinds of bilin-gual programs or comparisons of differentvarieties of Canadian-style immersion.

Bilingual education can be improved. Butthere is little doubt that bilingual educationworks.

Refetinces

Baker, K., and de Kanter, A. (1983). Federalpolicy and the effectiveness of bilinoraleducation. Lexington, MA: Lexington Books,33-86.

Barik, H., and Swain, M. (1978). Evaluation ofa bilingual education program in Canada: TheElgin study through grade six. Paperpresented at the Colloquium of the SwissInterJniversity Commission for AppliedLinguistics. [ED 174 0733

Barik, H., Swain, M., and Nwanunobi, E.(1977). English-French bilingual education:The Elgin study through grade five. CanadianModern Language Review, 33, 459-475.

Becker, W., and Gersten, R. (1982). Afollow-up of follow through: The later effects ofthe direct instruction model on children in thefilth and sixth grades. American EducationalResearch Journal, 19, 75-92.

Bruck, M., Lambert, W., and Tucker, G. R.(1977). Cognitive consequence of bilingualschooling: The St. Lambert Project throughgrade six. Psycholinguistics, 6, 13-33.

Crawford, J. (1989). Bilingral education:History, politics, theory and practice. Trenton,NJ: Crane Publishing Company.

Cummins, J. (1977). Delaying native readinginstruction in immersion programs: itcautionary note. Canadian Modem LanguageReview, 34, 46-49.

Cummins, J. (1981). The role of primarylarxruage development in promotingeducational success for language minoritystudents. In Callforrga State Department ofEducation (Ed.), Schooling and languageminority students: A theoretical framework.Los Angeles, CA: California State University;Evaluation, Dissemination and AssessmentCenter.

Eagon, R., and Cashion, M. (1988). Secondyear report on a longitudnal study ofspontaneous readng in English by students inearly French immersion classes. CanadianModem Language Review, 44, 523-526.

Edwards, H., Wesche, M., Krashen, S.,Clement, R., and Kruidenier, B. (1984).Second language acquisition throughsubject-matter learning: A study of shelteredpsychology classes at the University ofOttawa. Canadian Modem LarKjuage Review,41, 268-282.

El Paso Independent School District. (1987,July). Interim report of the five-year bilingualeducation pilot 1986-1987 school year. ElPaso, TX: El Paso Independent SchoolDistrict, Office for Research and Evaluation.

El Paso Independent School District. (1989a,January). Bilingual education evaluation: Thefourth year in a longitudinal study. 1987-1988school year. El Paso, TX: El Paso Indepen-dent School District, Office for Research andEvaluation.

El Paso Independent School District. (1939b,August). 3ilingual education evaluation: Thefifth year in a longitudinal study. El Paso, TX:

El Paso Independent School District, Officefor Research and Evaluation.

Feitelson, D., and Goldstein, Z. (1986).Patterns of book ownership and reading toyoung children in Israeli school-oriented andnonschool-oriented families. The ReadingTeacher, 39, 924-930.

Genesee, F. (1983). Bilingual education ofmajority-language children: The immersionexperiments in review. AppliedPsycholingvistics, 4,1-46.

Genesee, F., Tucker, G. R., and Lambert, W.(1978). An experiment in triNngual education.Unpublished manuscript, McGill University,Montréal.

Genesee, F., Tucker, G. R., and Lambert, W.(1978). An experiment in bilingual education:Report 3. The Canadian Modem LanguageReview, 34, 621-643.

Gersten, R. (1985). Structured immersion forlanguage minority students: Results of alongitudinal evaluation. EducationalEvaluation and Policy Ana4esis, 7,187-196

Gersten, R., and Woodward, J. (1985). Acase for structured immersion. EducationalLeadership, 43, 75-79.

Goodman, K. (1982). The selected writings ofKenneth S. Good;ran. London: Routledgeand Kegan Paul.

Hakuta, K. (1986). Mirror of language: Thedebate on bilingualism. New York: BasicBooks.

Hauptman, P., Wesche, M., and Ready, D.(1989). Second language acquisition throughsubject-matter learning: A follow-up study atthe University of Ottawa. Language Learning,38, 433-471.

1 5

Hillocks, G., Jr. (1986). Research on written

corrposition. Urbana, IL: ERIC.

Holobrow, N., Genesee, F., Lambed, W.,Gastright, J., and Met, M. (1987). Effective-ness of padial French immersion for childrenfrom different social class and ethnic back-grounds. Applied Psycholingiistics, 8,137-152.

Imhoff, G. (1990). The position of U.S. Englishon bilingual education. In C. Cazden and C.Snow (Eds.), English plus: Issues in bilingualedvcation. Annals of the American Academyof Political Science (pp. 48-61). NewburyPark, CA: Sage Publications.

Krashen, S. (1984). Writing: Research, theoryand applications. New York: Prentice-Hall.

Krashen, S. (1985a). Inquiries and insights.Hayward, CA: Alemany Press.

Krashen, S. (1985b). The input hypothesis:Issues and implications. New York: Longman.

Krashen, S. (1990). Reading, writing, form,and content. in J. Alatis (Ed.). GeorgetownUniversity Round Table on Languages andLinguistics 1990. Washington, DC: George-

. town University Press.

Krashen, S. (in press). The input hypothesisand some competing hypotheses. In R.Courchene and a) St. John (Eds.), Compre-hension-based language teaching: Currenttrends. Ottawa: University of Ottawa Press.

Krashen, S., and Biber, D. (1988). On course:Bilingual education's success in California.Sacramento, CA: California Association forBilingual Education.

Lafkette, R., and Buscaglia, M. (1983).Students learn language via a civilizationcoursea comparison of second languageclassroom environments. Studies in SecondLanguage Acquisition, 7,323-342.

Legarreta, D. (1979). The effects of programmodels on language acquisition bySpanish-spealdng children. TESOL Quarterly,8, 521-534.

Mezynski, K. (1983). issues concerning theacquisition of knowledge: Effects ofvocabulary training on reacling comprehen-sion. Review of Educational Research, 53.253-279.

Perla-Hughes, E., and Solis, J. (1980). ABCs(unpublished report). McAllen, TX: McAllenIndependent School District.

Porter, R. (1990a). Forked tongue: Thepolitics of bilingual educathan. New York:Basic Books.

Porter, R. (1990b, April 30-May 6). Bilingualeducation hes muted the future for minoritychildren. The Washington Pest NationalWeek4f Edition, pp. 24-25.

Rossell, C. (in press). The effectiveness ofeducational alternatives for limited Englishproficient children. In G. Imhoff (Ed.), Thesocial and cultural context of instruction in twolanguages: From conflict and controversy incooperative reorganizatbn of schools. NewYork: Transaction Books.

Rossell, C., and Ross, J. M. (1986). Thesocial science evidence on bilingualeducation. Journal of Law and Education, 15,385-418.

15

Schorr, B. (1983, November 30). Grade-school project helps Hisparic pupils learnEnglish quickly. The Wall Street Journal.

Simon, P. (1980). The tongue-tied American.New York: Continuum Press.

Smith, F. (1982). Understanding reading.Hillsdale, NJ: Erlbaurn.

Sternfeld, S. (1989). The University of Utah'simmersiontmultiEteracy program: An exampleof an area studies approach to the design offirst-year college foreign language Instruction.Foreign Language Annals, 22, 341-354.

Willig, A. (1985). A meta-analysis of selectedstudies on the effectiveness of bilingualeducation. Review of Educational Research,55,269-317.

Willig, A. (1987). Reply to Baker. Review ofEducational Research, 57,363-376.

About the author

Stephen Krashen is a professor of Nnguisticsat the University of Southern California. He isthe author of several books on languageacquisition, literacy development, andbilingual education, including Principles andPractice in Second Language Acquisition(Prentice Hall, 1982) and On Course:California's Success in Bilingual Education(with D. Biber, California Association forBilingual Education, 1988).

EMBEM1A91-2

The&vegge

wetw A SmINGION DC

ou.Center for AppPed linguistics

1'

1