(2005) A Synthesis of Research On Language of Reading - Slavin & Cheung

A Synthesis of Research on Language of Reading Instruction for English Language Learners Author(s): Robert E.
Slavin and Alan Cheung Source: Review of Educational Research, Vol. 75, No. 2 (Summer, 2005), pp. 247-284 Published by: American Educational Research Association Stable URL: http://www.jstor.org/stable/3516050 . Accessed: 14/02/2014 21:26
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at . http://www.jstor.org/page/info/about/policies/terms.jsp
.
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about JSTOR, please contact support@jstor.org.
American Educational Research Association is collaborating with JSTOR to digitize, preserve and extend access to Review of Educational Research.
http://www.jstor.org
This content downloaded from 128.223.174.86 on Fri, 14 Feb 2014 21:26:11 PM All use subject to JSTOR Terms and Conditions
Reviewof EducationalResearch Summer 2005, Vol. 75, No. 2, pp. 247-284
A Synthesis of Research on Language of Reading Instruction for English Language Learners

Robert E. Slavin Johns Hopkins University Alan Cheung Success for All Foundation
This article reviews experimental studies comparing bilingual and Englishonly reading programs for English language learners. The review method is best-evidence synthesis, which uses a systematic literature search, quantification of outcomes as effect sizes, and extensive discussion of individual studies that meet inclusion standards. A total of 17 studies met the inclusion standards. Among 13 studies focusing on elementary reading for Spanishdominant students, 9favored bilingual approaches on English reading measures, and 4found no differences,for a median effect size of + 0.45. Weighted by sample size, an effect size of +0.33 was computed, which is significantly different from zero (p < .05). One of two studies of heritage languages (French and Choctaw) and two secondary studies favored bilingual approaches. The review concludes that although the number of high-quality studies is small, existing evidence favors bilingual approaches, especially paired bilingual strategies that teach reading in the native language and English at different times each day. However, further research using longitudinal, randomized designs is needed to determine how best to ensure reading success for all English language learners.
KEYWORDS: bilingual education,English language learners,language of instruction, reading,researchreview. The reading education of English language learners(ELLs) has become one of the most importantissues in all of educational policy and practice. As the pace of immigrationto the U.S. and other developed countries has accelerated in recent decades, increasing numbers of children in U.S. schools come from homes in which English is not the primarylanguage spoken. As of 1999, 14 million Americans aged 5-24, or 17% of that age group, spoke a language other than English at home. This is more than twice the number of such individuals in 1979, when only 9% of Americans aged 5-24 spoke a language other than English at home (NCES, 2004). Although many children of immigrantfamilies succeed in reading, too many do not. In particular,Latino and Caribbeanchildren are disproportionatelylikely to perform poorly in reading and in school. As No Child Left Behind and other federal and state policies begin to demand success for all subgroups of children, the reading achievement of English language learners is taking on even more importance. Thousands of schools cannot meet their adequate yearly progress goals, for example, unless their 247
Slavin & Cheung
English language learnersare doing well in reading. More important,American society cannot achieve equal opportunity for all if its schools do not succeed with the children of immigrants. Sixty-five percentof non-English-speaking immigrantsin the United Statesare of Hispanicorigin (NCES, 2004), and this is also one of the fastest growing of all groups.Hispanicshave recentlysurpassedAfricanAmericansas the largestminority group in the United States. Hispanic studentsas a whole, including Englishproficientchildrenin the second generationand beyond, score significantlylower in reading than other students. On the National Assessment of Educational Progress(NAEP; Grigg, Daane,Jin, & Campbell,2003), which excludes children with the lowest levels of English proficiency from testing, only 44% of Latino fourth gradersscored at or above the "basic"level, in comparisonwith 75% of or better, Anglo students.Only 15%of Latinofourthgradersscoredat "proficient" as comparedwith 41% of Anglos. Furthermore, 31%of studentswho speak Spanish at home fail to complete high school, as comparedwith 10%of studentswho speak only English (NCES, 2004). There is considerable controversy, among policymakers, researchers, and educators, about how best to ensure the reading success of English language learners.Many aspects of instructionare importantin the readingsuccess of English language learners, but one question has dominated all others: What is the appropriaterole of the native language in the instruction of English language learners?In the 1970s and 1980s, policies and practice favored bilingual education, in which children were taughtpartiallyor entirely in their native language, andthen transitionedat some point duringthe elementarygrades to English-only instruction.Such programsare still widespread,but from the 1990s to the present, the political tide has turnedagainstbilingual education;California,Arizona, Massachusetts,and other states have enacted policies to greatly curtailbilingual education. Recent federal policies restrict the amount of time that children can be taughtin their native language. Among researchers,the debate between advocates of bilingual and English-only reading instructionhas been fierce, and ideology has often trumpedevidence on both sides of the debate (Hakuta,Butler, & Witt, 2000). This articlereviews researchon the languageof readinginstructionfor English of evilanguagelearnersin an attemptto applyconsistent,well-justifiedstandards dence to drawconclusions aboutthe role of native languagein readinginstruction for these children.The review applies a techniquecalled "best-evidencesynthesis" (Slavin, 1986), which attemptsto use consistent, clear standardsto identify unbiased, meaningfulinformationfrom experimentalstudies and then discusses each qualifying study, computing effect sizes but also describing the context, design, and findings of each study. Best-evidence synthesis closely resembles meta-analysis,but it requiresmore extensive descriptionof key studies.Details of this procedureare describedbelow. The purposeof this review is to examine the evidence on language of instructionin reading programsfor English language learnersto discover how much scientific basis thereis for competingclaims about the effects of bilingual as opposed to English-only programs.Such a review is needed to inform practitioners,policymakers, and researchersabout the current stateof the evidence on this topic and aboutgaps in the knowledgebase in need of furtherscientific investigation. 248
Language of Instruction For many years, researchers,educators, and policymakers have debated the questionof the appropriate languageof readinginstructionfor childrenwho speak other than languages English. Proponentsof bilingualinstructionarguethatwhile childrenarelearningto speakEnglish, they shouldbe taughtto readin theirnative languagefirst,to avoid the failureexperiencethatis likely if childrenare asked to learnboth oral English and English readingat the same time. Programsbased on this philosophytransitionchildrento English-onlyinstructionwhen theirEnglish is sufficientto ensuresuccess, usuallyin thirdor fourthgrade.Alternatively,many bilingualprogramsteach young childrento readboth in theirnative languageand in English, at differenttimes of the day or on alternating days. Thereis a greatdeal of evidence thatchildren'sreadingproficiencyin theirnative languageis a strong predictorof their ultimate English reading performance(Garcia, 2000; Lee & Schallert, 1997; Reese, Garier, Gallimore,& Goldenberg,2000) and that bilingualism itself does not interfere with performancein either language (Yeung, Marsh,& Suliman,2000). Bilingual advocatesalso arguethatwithoutnative language instruction, Englishlanguagelearnersarelikely to lose theirnativelanguage or fail to learnto readin theirnative language,losing skills thatareof proficiency, economic and social value in the world today. Opponentsof bilingual education, on the otherhand, arguethatnative languageinstructioninterfereswith or delays English language development,and relegates childrenwho receive such instruction to a second-class,separate statuswithinthe school and,ultimately,withinsociety. They reason that more time on English reading should translateinto more learning(see Rossell & Baker, 1996). havereached Reviews of the educational outcomesof nativelanguageinstruction that concluded conclusions. In a sharplyconflicting meta-analysis,Willig (1985) bilingual education was more effective than English-only instruction.WongFillmoreand Valadez (1986) came to the same conclusion.However, a review by Rossell and Baker (1996) claimed that most methodologicallyadequatestudies found bilingual education to be no more effective than English-only programs. that Greene(1997) re-analyzedthe studiescited by Rossell and Bakerandreported thetreatments, manyof the studiestheycitedlackedcontrolgroups,mischaracterized or hadotherseriousmethodological flaws.Amongthe studiesthatmet an acceptable standardof methodologicaladequacy,including all of the studies using random assignmentto conditions, Greene found that the evidence favored programsthat madesignificant use of nativelanguageinstruction. AugustandHakuta(1997) concludedthatalthoughresearchgenerallyfavoredbilingualapproaches, the natureof the methodsused andthe populations to which they were appliedwere moreimportantthanthe languageof instruction research on the outcomesof perse. Quantitative bilingualeducationhas diminishedin recentyears, but policy and practiceare still of researchon this topic.The followbeing influencedby conflictinginterpretations to discoverwhatwe can examinethisevidenceto attempt ing sectionssystematically learnfromresearchto guide policies in this controversial arena. English Immersionand Bilingual Programs When a child enters kindergartenor first grade with limited proficiency in English,the school faces a seriousdilemma.How can the child be expectedto learn the skills andcontenttaughtin the earlygradeswhile he or she is learningEnglish? 249
Slavin& Cheung There may be many solutions, but two fundamentalcategories of solutions have predominated: English immersionand bilingual education. English Immersion In immersionstrategies,Englishlanguagelearnersareexpectedto learnin English fromthe beginning,andtheirnativelanguageplays littleor no role in dailyreading lessons. Formalor informalsupportis likely to be given to ELLs to help them cope in an all-Englishclassroom.Such supportmay includehelp from a bilingual aide who providesoccasionaltranslation or explanation,a separateclass in English as a Second Languageto help build oral English skills, or use of a carefulprogression from simplifiedEnglish to full English as children'sskills grow. Teachersof English languagelearnersmay use languagedevelopmentstrategiessuch as total physical response (acting out words) and realia (concrete objects to represent words) to help studentsinternalizenew vocabulary.Teachersmay simplify their languageand teach specific vocabularythatis likely to be unfamiliarto ELLs (see Calder6n, 2001; Carlo et al., 2004). Immersion may involve placing English languagelearnersimmediatelyin classes containingEnglishmonolingualchildren, or it may involve teachingELLsin a separateclass for some time untilthe children are ready to be mainstreamed. These variationsmay well have importancein the outcomes of immersionstrategies,but their key common featureis the exclusive use of Englishtexts, with instruction overwhelminglyor entirelyin English. Many authorshave made distinctionsamong variousforms of immersion.One term often encounteredis "submersion," primarilyused pejoratively to refer to "sink or swim" strategiesin which no special provision is made for the needs of with "structured Englishimmersion," Englishlanguagelearners.This is contrasted which refersto a well-planned,gradualphase-inof English instruction relying initially on simplificationand vocabulary-building strategies.In practice,immersion strategiesare rarelypure types; and in studies of bilingual education,immersion strategies are rarely described beyond their designation as the English-only "controlgroup." Bilingual Education Bilingual education differs fundamentallyfrom English immersion in that it gives English language learners significant amounts of instruction in reading and/or other subjects in their native language. In the United States, the overwhelming majorityof bilingual programsinvolve Spanishbecause of the greater and the greater likelihoodof a criticalmass of studentswho are Spanish-dominant in Chiof materials. There are Portuguese, programs availability Spanish bilingual nese, and other languages, but these are rare. In transitionalbilingual programs, children are taughtto read entirely in their native language throughthe primary grades;they transitionto English readinginstructionsomewherebetween the second and fourthgrades.English oral languageproficiencyis taughtfrom the beginning, and subjectsother than readingmay be taughtin English. But the hallmark of transitional bilingualeducationis the teachingof readingin the nativelanguage for a periodof time. Such programscan be "early-exit" models, with transitionto English completed in second or thirdgrade, or "late-exit"models, in which children may remainthroughoutelementaryschool in native-languageinstructionto ensuretheirmasteryof readingand contentbefore transition(see Ramirez,Pasta, 250
Research on Languageof ReadingInstructionforELLs
Yuen, Billings, & Ramey, 1991). Alternatively,"pairedbilingual"models teach childrento readin both English and theirnative languageat differenttime periods each day or on alternatingdays. Within a few years, the native languagereading instructionmay be discontinued,as childrendevelop the skills to succeed in English. Willig (1985) called this model "alternative becausechildrenare immersion," alternativelyimmersedin native languageand English instruction. also called "duallanguage"or "dualimmersion," Two-way bilingualprograms, in the nativelanguage(usuallySpanish)andin English providereadinginstruction both to ELLs and to English speakers(Calderon& Minaya-Rowe,2003; Howard, Sugarman,& Christian,2003). For the ELLs, a two-way programis like a paired bilingual model, in that they learn to read both in English and in their native languageat differenttimes each day. A special case in bilingual education is the kind of programthat is designed more to preserveor show respect for a given language than to help childrenwho are genuinely struggling with English. For example, Morgan (1971) studied a programin Louisianafor childrenwhose parentsoften spoke Frenchat home but generallyspoke English as well. Such "heritagelanguage"programsare included in this review if the outcome variablein the study is an English readingmeasure. However, these programsshould be thoughtof as addressinga differentproblem fromthataddressedby bilingualor immersionreadinginstruction for childrenwho are limited in English proficiency. Problems of Research on Language of Instruction Researchon the achievementeffects of teachingin the child's nativelanguage, in comparisonwith teaching in English, suffers from a numberof inherentproblems beyond those typical of otherresearchon educationalprograms.First, there are problemsconcerningthe ages of the childreninvolved, the length of time they have been taught in their first language, and the length of time they have been taught in English. For example, imagine that a transitionalbilingual program teaches Spanish-dominant studentsprimarilyin Spanishin GradesK-2 and then graduallytransitionsthem to English by fourthgrade.If this programis compared with an English immersionprogram,at what grade level is it legitimateto assess the childrenin English?Clearly,a test in second gradeis meaningless,as the bilingual childrenhave not yet been taughtto readin English. At the end of thirdgrade, the bilingual studentshave partiallytransitioned,but have they had enough time to become fully proficient?Some would argue that even the end of fourthgrade would be too soon to assess the childrenfairly in such a comparison,as the bilintheirSpanishreadgual childrenneed a reasonabletime periodin which to transfer ing skills to English (see, for example, Hakuta,Butler,& Witt, 2000). A relatedproblemhas to do with pretesting.Imaginethata studyof a K-4 transitional Spanish bilingual programbegan in third grade. What pretest would be An Englishpretestwould understate bilinthe skills of the transitional meaningful? gual students, while a Spanish test would understatethe skills of the English immersionstudents.Forexample, Valladolid(1991) comparedgains from Grades 3 to 5 for childrenwho had been in eitherbilingual or immersionprogramssince These children's"pretest" scores are in fact posttestsof very differkindergarten. ent treatments.Yet studies comparingtransitionalbilingual and immersionprogramsaretypicallytoo brief to have given the studentsin the transitional bilingual 251
Slavin & Cheung
programsenough time to have completed the transitionto English. In addition, many studies begin after students have already been in bilingual or immersion treatments for several years. The studiesthatdo look at 4- or 5-yearparticipations in bilingualor immersion programsare usually retrospective(i.e., researcherssearch records for children who have alreadybeen throughthe program). Retrospectivestudiesalso have characteristicbiases, in thatthey begin with the childrenwho ended up in one program or another.Forexample,childrenwho areremovedfrom a given treatment for systematicreasons,such as Spanish-dominant studentsremovedfromEnglishimmersion becauseof theirlow performance there,can greatlybias a retrospectivestudy, makingthe immersionprogramlook more effective thanit was in reality. Many inherentproblemsrelateto selection bias. Childrenend up in transitional bilingualeducationor English immersionby many processes thatcould be highly studentsmay be consequentialfor the outcomes. For example, Spanish-dominant assigned to Spanishor English instructionon the basis of parentpreferences.Yet parentswho would select English programsare surely differentfrom those who would select Spanishin ways thatmatterfor outcomes. A parentwho selects English may be moreor less committedto education,may be less likely to be planning to returnto a Spanish-speaking country,or may feel morepositive aboutassimilation. Thomas and Collier (2002) reported extremely low scores for Houston studentswhose parentsrefusedto have theirchildrenplaced in eitherbilingualor English as a Second Languageprograms.Are those scores due to relativelypositive effects of bilingual and ESL programs,or are there systematic differences between children whose parents refused bilingual or ESL programsand other children?It is impossible to say, as no pretestscores were reported. Bilingual programsare more likely to exist in schools with very high proportions of English languagelearners,and this is anotherpotentialsource of bias. For example, Ramirez et al. (1991) found that schools using late-exit bilingual proof ELLs thandid early-exitbilingualschools, gramshad much higherproportions and English immersionschools had the smallest proportionof ELLs. This means that whateverthe languageof instruction,childrenin schools with very high proportionsof ELLs are conversingless with native English speakersboth in and out of school thanmightbe the case in an integrated thatuses school andneighborhood English for all studentsbecause its proportionof ELLs is low. Most problematically, individualchildrenmay be assignedto native languageor Englishprograms because of theirperceivedor assessed competence.Native languageinstructionis often seen as an easier, more appropriate placementfor ELLs who are struggling to readin theirfirstlanguage, whereasstudentswho arevery successful readersin their firstlanguageor are felt to have greaterpotentialare placed in English-only as the most classes. This selectionproblemis most vexing at the pointof transition, successful studentsin bilingualprogramsgo throughthe transitionearlierthanthe least successful children. A study comparingbilingual and immersionprograms involving thirdor fourthgradersmay be seriouslybiased by the fact thatthe highwith the result est-achievingbilingualstudentsmay have alreadybeen transitioned, thatthe remainingstudentsare the lowest achievers. A source of bias not unique to studies of bilingual educationbut very importantin this literatureis the "file drawer" problem:the fact that studies showing no differences are less likely to be publishedor to otherwise come to light. This is a 252
Research on Languageof ReadingInstruction for ELLs
particular problemin studies with small sample sizes, which are very unlikely to be publishedif they show no differences.The best antidoteto the file drawerproblem is to search for dissertationsand technical reports,which are more likely to presenttheir dataregardlessof the findings (see Cooper, 1998). Finally, studiesof bilingualeducationoften say too little aboutthe bilingualand immersionprogramsthemselves or the degree or qualityof implementation of the programs.Yet bilingualmodels can vary substantiallyin quality,amountof exposure to English in and out of school, teachers' language facility, time duringthe school day, instructional strategiesunrelatedto languageof instruction,and so on. Because of these inherentmethodologicalproblems,an adequatestudycomparwould (a) randomlyassign a largenumber ing bilingualandimmersionapproaches of childrento be taughtin Englishor theirnativelanguage;(b) pretestthemin their native languagewhen they begin to be taughtdifferentially,either in their native languageor in English (typicallyin kindergarten); (c) follow them long enoughfor the latest-transitioning childrenin the bilingualconditionto have completedtheir transitionto English and have been taughtlong enough in English to make a fair and (d) collect datathroughout the experimentto documentthe treatcomparison; ments received in all conditions.Unfortunately, only a few, very small studies of this kind have ever been carriedout. As a result,the studiesthatcomparebilingual and English-onlyapproaches must be interpreted with greatcaution. Review Methods This section focuses on researchcomparingimmersionand bilingual reading programsappliedwith English languagelearners,with measuresof English reading as the outcomes. The review method,best-evidence synthesis (Slavin, 1986), uses the systematicinclusioncriteriaandeffect size computationstypical of metaanalyses (see Cooper, 1998; Cooper& Hedges, 1994) but discusses the findingsof critical studies in a form more typical of narrativereviews. This strategy is well suited to the literature on readingprogramsfor English language particularly learners,because the studiesare few in numberand are substantivelyand methodit is particularly to learnas much ologically diverse. In such a literature, important as possible from each study, not just to average quantitativeoutcomes and study characteristics. LiteratureSearch Strategy The literaturesearch benefited from the assistance of the federally commissioned NationalLiteracyPanel on the Developmentof LiteracyAmong Language MinorityChildrenand Youth, chairedby Diane August and Timothy Shanahan. The first authorwas initially a memberof the panel but resigned in June 2002 to avoid a 2-year delay in publicationof the presentarticle.The article,however, is independentof the panel's reportand uses differentreview methodsand selection criteria.Research assistants searchedERIC, Psychological Abstracts,and other databases for all studies with the following descriptors: language minority students,Englishlanguage learners,bilingualeducation,bilingualstudents,bilingualism, English as a second language, English immersion,dual language, and two-waybilingual education.Citationsfrom otherreviews and articles were also obtained.In particular, every effort was made to find all studies cited in previous reviews. Fromthis set, we selected studies thatmet the criteriadescribedbelow.
253
Criteriafor Inclusion of The best-evidence synthesis focused on studies that met minimal standards methodologicaladequacyand relevanceto the purposesof the review. These were as follows: Criterion1. The studiescomparedchildrenwho were taughtreadingin bilingual classes with those taughtin English immersionclasses, as definedearlier. Criterion2. Eitherrandomassignmentto conditionswas used, or pretesting or other matchingcriteriaestablishedthe degree of comparabilityof bilingual and immersiongroupsbefore the treatments began. If these matchingvariables were not identical at pretest,the analyses made adjustmentsfor pretestdifferences, or else data permittingsuch adjustmentswere presented.Studies without control groups, such as pretest-posttestcomparisonsor comparisonswith "expected"scores or gains, were excluded. Studies with pretest differences deviationwere excluded.Those with pretestdifferences exceeding one standard hadcarriedout less thanone standard deviationwere includedif the researchers statisticaladjustments. appropriate of A special categoryof studies was rejectedon the basis of the requirement before treatments pretestmeasurement began. Those were studies in which the were alreadyunderway before pretestingor bilingualandimmersionprograms matching.Forexample, Danoff, Coles, McLaughlin,and Reynolds (1978), in a widely cited study, compared 1-yearreading gains in many schools by using or first bilingual or immersionmethods.The treatmentsbegan in kindergarten in to children administered but the were grade, pretests (and, later, posttests) Grades2-6. Because the bilingualchildrenwere taughtprimarilyin theirnative language in K-1 and the immersion children were taught in English, their conpretestsin secondgradewould surelyhave been affectedby theirtreatment dition. Meyer and Feinberg (1992, p. 24) noted the same problemwith reference to the Grades 1-3 componentof the Ramirezet al. (1991) study:"Itis like watching a baseball game beginning in the fifth inning: If you are not told the score from the previous innings, nothing you see can tell you who is winning the game."Similarly,severalstudiestestedchildrenin upperelementaryor secondarygradeswho had experiencedbilingualor immersionprogramsin earlier years. These studies were included if premeasureswere availablefrom before the programsbegan, but in most cases such premeasureswere not reported,so there was no way to know if the groups were equivalent beforehand (e.g., Thomas & Collier, 2002; Curiel,Stenning,& Cooper-Stenning,1980). Criterion3. The subjects were English language learnersin elementaryor secondary schools in English-speaking countries. Studies that mixed ELLs and English monolingual students in a way that did not allow for separate analyses were excluded (e.g., Skoczylas, 1972). Studies of children learning a foreign language were not included. However, Canadianstudies of French immersionhave been widely discussed and are thereforediscussed in a separate section. measuresof EngCriterion4. The dependentvariablesincludedquantitative lish reading performance, such as standardizedtests and informal reading inventories.If treatment-specific measureswere used, they were includedonly if there was evidence that all groups focused equally on the same outcomes. 254
Research on Language of ReadingInstruction for ELLs
Measures of outcomes related to reading, such as language arts, writing, and spelling, were not included. Criterion5. The treatmentdurationwas at least 1 school year. For the reasons discussed earlier,even 1-year studies of transitionalbilingual education are insufficient,because studentstaughtin theirnative languageareunlikely to have transitioned to English.Studiesshorter than 1 yeardo not addressthe question in a meaningfulway. Studies thatpassed an initial screeningfor germanenessto the topic, includingall studies cited by Rossell and Baker (1996) or by Willig (1985), are listed in the Appendix,which indicateswhethereach study was includedand, if not, the main reasonsfor exclusion. Limitations to note thatthe review methodsappliedin this best-evidencesynIt is important thesis have some importantlimitations.First, in requiringmeasurableoutcomes and control groups, the synthesis excludes case studies and qualitative studies. Many such descriptive studies exist and are valuable in suggesting programsor practices that might be effective. Description alone, however, does not indicate how much childrenlearnedin a given programor what they would have learned had they not experiencedthat program.Second, it is possible that a programthat has no effect on readingachievementmeasuresmight neverthelessincrease children's interestin reading or readingbehaviors outside school. However, studies rarely measuresuch outcomes in any systematic or comparativeway, so we can to note thatmany of the studies only speculateaboutthem. Finally, it is important reviewedtook place manyyearsago. Social andpoliticalcontexts, as well as bilingual andimmersionprograms,have changed,so it cannotbe takenfor grantedthat outcomesdescribedherewould applyto outcomesof bilingualandimmersionprogramstoday. Computation of Effect Sizes If possible, effect sizes were computedfor each study. These were computed as the experimentalmean minus the control mean, with the result divided by a pooled standarddeviation. When informationwas lacking, however, effect sizes were estimated using exact t or p values or other well-established estimation methods (see Cooper, 1998; Cooper & Hedges, 1994; Lipsey & Wilson, 2001). For studies that lacked means and standarddeviations and reportedno significant difference between the experimentaland controlgroups and did not indicate the direction of the effect (e.g., Cohen, 1975), an estimated effect size of zero was used. Only English reading measureswere used in determiningeffect sizes, even if othermeasureswere mentionedin the text. No study was excluded solely on the grounds that it did not provide sufficient informationfor computationof an effect size. Data Analysis All datawere enteredinto the betaversionof the Comprehensive Meta-Analysis (Borenstein, 2005) to estimatethe effect sizes of each study,to calculatethe Program overallmeanweightedeffect sizes, andto test whetherthe meanweightedeffect size 255
Slavin & Cheung
was derivedfroma homogeneousset (Q statistic).The weightingfactorwas sample moreto the meanthandid size, so thateffect sizes from largersamplescontributed thosefromsmallersamples.Eachstudycontributed a single effect size to the overall meanweightedeffect size. Forstudiesthathad morethanone independent groupor for each one independent outcome measure,effect sizes were calculatedseparately groupand measure.These effect sizes were then averagedto createone effect size for the study.For longitudinalstudies,the last time-pointwas used to estimatethe overalleffect of the study.Forexample,if a studyfollowed a groupof childrenfrom were used to generatethe Grade1 to Grade5, the outcomemeasuresfor fifthgraders effect sizes.
Previous Quantitative Reviews
The debate about empirical research on language of instruction for English language learners has largely pitted two researchers, Christine Rossell and Keith Baker, against several other reviewers. Rossell and Baker have carried out a series of reviews and critiques arguingthatresearchdoes not supportbilingual education (see Baker & de Kanter, 1981, 1983; Baker, 1987; Rossell, 1990; Rossell & Baker, 1996; Rossell & Ross, 1986). The most comprehensive and recent version of their review was published in 1996. In contrast, Willig (1985) carriedout a meta-analysis and concluded that research favored bilingual education, after controls were introduced for various study characteristics. Other reviewers using narrativemethodshave agreed with Willig (e.g., Wong-Fillmore & Valadez, 1986). Baker (1987) and Rossell and Baker (1996) criticized the Willig (1985) review in detail, and Willig (1987) respondedto the Baker (1987) criticisms. In a review commissionedby the TomasRiveraCenter,JayGreene(1997) carefully reexaminedthe Rossell andBaker(1996) review. WhereasRossell andBaker used a "vote-counting" methodin which they simply countedthe numbersof studies thatfavoredbilingual,immersion,or otherstrategies,Greene(1997) carriedout a meta-analysis in which each study produced one or more effect sizes, the of a standard deviationseparating proportion bilingualandEnglish-onlyprograms. Greenecategorizedonly 11 of the 72 studiescited by Rossell andBakeras methodologically adequate,and among those he calculatedan effect size of +0.21 favoring bilingualover English-onlyapproacheson English readingmeasures.Among five studies using randomassignment,Greene calculatedan effect size of +0.41 on English readingmeasures. As partof this review, we attemptedto obtainevery study reviewed by Rossell and Baker (1996) and by Willig (1985), as well as additionalstudies, and independently reviewed each one against the set of standardsoutlined previously. Consistent with Greene, we found that the Rossell and Baker (1996) review acceptedmany studies that lacked adequatemethodology. The Appendix lists all of the reading studies cited by Rossell and Baker according to categories of methodological adequacyoutlined in this article, which closely follow Greene's categorization.As is apparentfrom the Appendix, only a few of the studies met the most minimal of methodological standards,and most violated the inclusion criteriaestablishedby Rossell and Baker(1996) themselves. We found, however, that most of the 16 studies cited by Willig also did not meet these minimal standards.These are also noted in the Appendix. In itself, this does not mean that the 256
Instruction Research on Language for ELLs of Reading overall conclusions of eitherreview are incorrect,but it does mean that the question of effects of languageof instructionon readingachievementmustbe explored with a different set of studies than the ones synthesized by either Rossell and Baker or Willig. The Rossell and Baker and Willig studies can be categorized as follows (following Greene, 1997): 1. Methodologicallyadequate studies of elementaryreading. These studies comparedEnglishlanguagelearnerswho were taughtto readthroughbilingual strategieswith those who were taughtthroughEnglish-onlystrategies, with randomassignmentor well-documentedmatchingon pretestsor other importantvariables. All of these studies focused on Spanish-dominant students. 2. Methodologicallyadequate studies of heritage language programs. Two studies,one involving Choctawin Mississippi andone involving Frenchin with childrenwho generallyspoke Louisiana,evaluatedbilingualprograms English but were expected to benefit from introductionof their cultural language. 3. Methodologically adequate studies of secondary programs. We put two secondaryschool studies (Covey, 1973; Kaufman, 1968) in this separate category. 4. Canadianstudies of French immersion.Several studies (e.g., Lambert& Tucker, 1972; Genesee & Lambert, 1983) evaluated French immersion programsin Canada.However, because they compared immersion with monolingual English instruction or with brief French-as-a-secondlanguageclasses, these studieswere not evaluationsof bilingualeducation. 5. Studiesin whichthe targetlanguagewas not the societal language.In addiareas tion to Canadianstudies of French immersionin non-Francophone (e.g., Day & Shapson, 1988), there was a study by Ramos, Aguilar, and Sibayan(1967) thatexaminedvariousstrategiesfor teachingEnglishin the Philippines. 6. Studies of outcomes other than reading. A few studies (e.g., Lum, 1971; Legarreta,1979; Pena-Hughes& Solis, 1980) assessed only oral language proficiency,not reading. 7. Studies in which "bilingual" treatmentsinvolved little use of native language reading instruction. A few studies (e.g., EducationalOperations Concepts, 199la, 1991b) evaluatedprogramsthat may be called bilingual but thatin fact made only incidentaluse of the native languageand did not use native languagereadingtexts. were underway. As 8. Studiesin whichpretestingtookplace after treatments notedearlier,manystudies(e.g., Danoff et al., 1978;Rosier & Holm, 1980; Rossell, 1990; Thomas& Collier, 2002; Valladolid, 1991) comparedgains made in bilingual and immersionprogramsafter the programswere well under way. Both the Willig review and the Rossell and Baker review included such studies, and Greene (1997) accepted some of them as "methodologicallyadequate";but we would argue that they add little to the effects of bilingualeducation. understanding 9. Redundantstudies. Rossell and Baker included many studies that were redundant with otherstudiesin theirreview. Forexample, one longitudinal 257
Slavin& Cheung study (El Paso, 1987, 1990, 1992) issued threereportson the same experiment, but it was countedas three separatestudies. Curiel's 1979 dissertation was publishedin 1980, and both versions were counted. 10. Studies with no evidence of initial equality. Several studies either lacked dataon initialachievement,beforetreatments dataindibegan,or presented deviation. cating pretestdifferencesin excess of one standard 11. Studies with no appropriate comparison group. Many of the studies included by Rossell and Baker had no control group. For example, Burkheimer, Conger,Dunteman,Elliott, andMowbray(1989) andGersten (1985) used statistical methods to estimate where children should have been performingand then comparedthatestimatewith theiractualperformance. Rossell and Bakerset standards requiring"a comparisongroup of LEP studentsof the same ethnicityand similarlanguagebackground," yet they included many studies that did not have such comparison groups. Furthermore, manystudiesincludedby Rossell andBakerlackedany information about the initial comparabilityof childrenwho experiencedbilingual or English-only instruction (e.g., Matthews, 1979). These include studies that retroactivelycompared secondary students who had participated in bilingual or English-only programs in elementary schools but failed to obtain measures of early academic ability or performance (e.g., Powers, 1978;Curielet al., 1980). Otherstudiescomparedobviously noncomparablegroups. For example, Rossell (1990) compared 1-year gains of English language learnersin Berkeley, California,who were in Spanishbilingual or English immersionprograms;yet 48% of the ELLs, all in the English immersionprograms,were Asian, whereasall studentsin the Spanishbilingualprogram(32%of the sample)were, of course,Latino. childrenin bilin(1979) comparedSpanish-dominant Similarly,Legarreta childrentaughtin English. gual instructionwith mainly English-dominant children Finally,CarlisleandBeeman (2000) comparedSpanish-dominant in 80% with those in and 20% taught taught 80% in Spanish English English and 20% in Spanish, so there was no English-only comparison group. 12. Brief studies. A few studies cited by Rossell and Baker involved treatment durationsless than 1 year. For the reasons discussed earlier, studies of bilingual education lasting only 10 weeks (Layden, 1972) or 4 months (Balasubramonian,Seelye, & de Weffer, 1973) are clearly not relevant. All but one of these brief studies failed to meet inclusion standardson othercriteriaas well (e.g., they lacked pretestsor had outcomes otherthan reading). ThePresent Review This review carriesout a best-evidencesynthesisof studiescomparingbilingual and English approachesto reading in the elementaryand secondary grades that meet the inclusioncriteriaoutlinedabove. These includethe methodologicallyadequate studies cited in the Willig (1985), Rossell and Baker (1996), and Greene (1997) reviews, as well as other studies located in an exhaustive searchof the litand findings of these studies erature,as describedpreviously. The characteristics are summarizedin Table 1 (p. 260). 258
Studies of Beginning Reading for Spanish-Dominant Students The largestnumberof studiesfocused on teachingreadingto Spanish-dominant studentsin the earlyelementarygrades.Thirteenstudiesof this kind met the inclusion criteria. Threecategoriesof bilingualprogramswere distinguished.The most common among the qualifying studies were studies of pairedbilingual strategies,in which studentswere taughtto readin English andin Spanishat differenttimes of the day, beginning in kindergartenor first grade and continuing through the end of the study. Pairingmay not have begun on the first day of the school year, but if childrenwere being taughtto read in both Spanishand English duringtheir firstyear of readinginstruction,the programwas considereda pairedmodel. A second category involved evaluationsof programsin which childrenwere taughtreadingin Spanishfor 1 year before a transitionto pairedbilingual instruction(English and Spanish). A third category consisted of a single study by Saldate et al. (1985), which did not describe the treatments well enough to permit categorization, model. althoughit seemed to evaluatea transitional In Table 1, the elementary studies of Spanish-dominantchildren are listed studieslisted first. accordingto these treatment categories,with the highest-quality Thatis, randomizedmultiyearstudiesarelisted first,then matchedmultiyearstudies, then matched 1-yearstudies. The studies will be discussed in the same order. Studiesof Paired Bilingual Programs Ten qualifying studies comparedpairedbilingual and English immersionproPuertoRicanchildren grams.Plante(1976) randomlyassigned Spanish-dominant, in a New Haven, Connecticut,elementaryschool with a pairedbilingualmodel or with English-onlyinstruction. Initially,72 childrenwere randomlyassigned,45 to the pairedbilingual groupand 27 to an English-onlycontrolgroup.By the end of the study, 31 childrenremainedin the pairedbilingual group and 22 remainedin the controlgroup.The childrenbegan in kindergarten or firstgrade.The treatment involved a team-teachingarrangement with one native Spanish-speaking teacher and one English-speakingteacher.It was describedas follows: [Spanish-dominant children]aretaughttheirbasic skills, i.e., reading,writsocial studies,andscience,in Spanish.At the sametime,the ing, arithmetic, English-speaking Anglo teacherinitiatesthe teachingof English,beginning withan aural-oral Whenan Englishoralvocabulary is sufficiently approach. in the readingand children,she initiatesinstruction developedin individual of English.Thekey premise in thisinstructional is a conwriting organization with both Spanishand English instruction cept of diagnostic-prescriptive resources being available.(Plante,1976,p. 40) Analyses of pretestscores for the final sample found that the two groups were similar on measures of Spanish and English oral vocabulary.The control group was nonsignificantlyhigheron both measures. Two yearslater,all childrenwere given the Englishform of the Inter-American Test of Readingand the English Metropolitan AchievementTest. Second graders in the pairedbilingual treatmentscored significantlyhigher than control second graders(effect size (ES) = +0.78). The effect size for thirdgraderswas also positive but nonsignificant(ES = +0.26). Total effects were nonsignificant,with a 259
N clL TABLE 1 Languageof reading instruction:Descriptive informationand effect sizesfor qualifyingstudies Intervention Evide N, grade Design, Sample duration characteristics initial e levels Study description Studies of paired bilingual education Plante (1976) Paired bilingual Random assignment, 2 years 55 students, Grades 1-2, 2-3 Spanish-dominant PuertoRican studentsin New Haven, CT
Well ma on Sp oral v but El Engli
Huzar (1973)
Paired bilingual
Random assignment, 2 & 3 years
160 students, Grades 1-2, 1-3
Disadvantaged PuertoRican studentsin PerthAmboy, NJ
Well ma on IQ and in achiev
Campeau etal. (1975), Corpus Christi Ramirez etal. (1991)
Paired bilingual
Matched control, 2 years Matched control, 2 years
171 students, Grades K-l 153 students in four schools, Grades K-l
Spanish-dominant studentsin CorpusChristi, TX Spanish-dominant LEP students
Matched Engli Spani pretes
Paired bilingual
Very we match SES a backg
Campeau et al. (1975), Houston J. R. Maldonado (1977) Alvarez (1975)
Paired bilingual Paired bilingual
Matched control, 3 years Matched control, 5 years
206 students, Grades K-2 126 students, Grades 1-5
Spanish-dominant studentsin Houston, TX Spanish-dominant studentsin six elementary schools in CorpusChristi, TX Spanish-dominant childrenin two schools in Austin, TX
Matche langu and a achie Matched and n years schoo
Paired bilingual
Matched control, 2 years
147 students, Grade2
Matched and in langu profic
Cohen (1975)
Paired bilingual
Matched control, 2-3 years Matched control, 1 year
90 students, Grades K-I, 1-2, 1-3 89 students, GradeK
Spanish-dominant studentsin Redwood City, CA Spanish-dominant studentsin Kingsville, TX
Matched and in langu profic
Campeau et al. (1975), Kingsville, TX
Paired bilingual
Matched and et
TABLE 1 (Continued) Languageof reading instruction:Descriptive informationand effect sizesfor qualifyingstudies Intervention Evide N, grade Design, Sample duration levels characteristics initiale Study description Studiesof paired bilingual education Campeau et al. (1975), SantaFe Paired bilingual Matched control, 1 year 77 students, Grade I Hispanicstudents in SantaFe, NM
Similaro prete PBE >
Studiesof I-year transitionalbilingual education J. A. Maldonado (1994) Campeau et al. (1975), Alice, TX Bilingual, 1-year transition Random assignment, 3 years 20 students, Grades 2-4, 3-5 Spanish-dominant special education studentsin Houston,TX Spanish-dominant studentsin Alice, TX
Bilingual, 1-year transition
Matched control, 2 years
125 students, Grades K-l
Well ma disab langu profic and fa backg Similaro Engli but TB Spani
Studyof bilingual education(unspecified) Saldate et al. (1985) Unspecified Matched control, 3 years 38 students, Grades 1-3 Spanish-dominant studentsin Douglas, AZ
Well ma pretes
Studies of heritage languages Morgan (1971) Paired bilingual Matched control, 1 year 193 students, Grade 1 French-dominant studentsin Lafayette,LA,
Well ma initia abilit
Catholic Schools
MRT
Doebler & Mardis (1980)
Paired bilingual
Matched control, 1 year
63 students, Grade2
Choctaw students in MS
Well ma initial profic
Secondarystudies Covey (1973) Kaufman (1968) Paired bilingual Paired bilingual Random assignment, 1 year Random assignment, 1 and 2 years 200 students, Grade9 139 students, Grade7 Spanish-dominant studentsin Phoenix, AZ Spanish-dominant studentsin New York City
Well ma pretes
InitialC vocab comp scores langu non-la IQ, ag Hoffm biling sched were u covar
t)
a
Note. PBE = Paired Bilingual Education; TBE = Transitional Bilingual Education; El = English Imme
a
Basic Skills; MAT = Metropolitan Achievement Test; SRAAS = Science ResearchAssociates Asse Effect size estimated;datafor exact computationwere not available.
Slavin & Cheung
mean weighted effect size of +0.50. However, these differences did not control for the control group's pretest advantage,so pretest-adjusted differences would have furtherfavoredthe experimentalgroup. On the Metropolitan AchievementTest, total readingscores favoredthe experimental group by 0.4 grade equivalents among second graders and 0.5 grade equivalents among third graders.No standarddeviations or analyses were provided, however. differencesin retentionrates.Only one of the 31 Finally, therewere substantial was children retained in grade, as comparedwith 13 of the 22 (3%) experimental controlchildren(59%).Retentionratesaredetermined by the teachersinvolved on a subjectivebasis, but this is neverthelessan important finding. Not surprisingly, betteron a childrenin the bilingualgroupscoredsubstantially = + 1.02). ES than did the test control (overall Spanishreading group English-only The Plante (1976) study is small, and with only one class in each treatment, teachereffects were completely confoundedwith treatment effects. Yet its use of randomassignmentand a 2-year longitudinaldesign with modest attritionmakes this study an important partof the researchbase on bilingualeducation. Huzar (1973) carried out a randomizedexperiment involving 160 SpanishdominantPuerto Rican children in Perth Amboy, New Jersey. On entry to first grade,the childrenwere assignedto one of two treatments: pairedbilingualor control. The paired bilingual treatmentwas described as follows: "One bilingual teachergave readinginstructionto the class in Spanishfor 45 minutes each day, while the monolingual teacher gave reading instructionin English for the same periodof time each day" (Huzar, 1973, p. 34). In the controlgroup,studentswere taughtonly in Englishfor 45 minutesa day. "All teachingprocedures, qualityof materials,andtime periodsfor readinginstruction were the same, with the exception that the experimental classes received instructionin both Spanishand English, with corresponding textbooks." Therearetwo potentialconfoundsin this study.First,it is unclearwhatthe control group was doing while the experimentalgroup received 45 minutes of daily Spanishreadinginstruction.It may be thatthe experimentalstudentswere receiving more totaltime in reading(Englishplus Spanish).Second, the Englishreading texts used in the two programs classes used a phowere different.The experimental netic program,the Miami LinguisticReaders.Controlstudentsused Scott Foresman's Open Highways series. The 160 study subjectswere assigned at randomto four experimentaland four controlclasses. Two classes of each treatment were at the second- andthird-grade levels, respectively. Third gradershad been in their respective treatmentsfor 3 years, and second gradershad been in theirs for 2 years. MetropolitanReadiness Test first-grade scores were collected from school records;Lorge-Thordike IntelTest scores were obtainedfor thirdgradersand showed no significantdifligence ferences between treatment groupsat eithergradelevel. The posttestwas the Englishreadingtest of the Inter-American Series. For second graders,there were no differences (ES = +0.01). Differences at the thirdgrade level directionallyfavoredthe experimentalgroup (ES = +0.31) but were statisticallysignificantfor boys (ES = +0.44) but not girls (ES = -0.06). The Huzar(1973) and Plante(1976) studies are particularly important, despite takingplace morethana quarter centuryago. Both aremultiyearexperimentsthat, 264
Research on Languageof ReadingInstructionforELLs
because of use of randomassignment,can rule out selection bias as an alternative explanationfor the findings. Both startedwith children in the early elementary grades and followed them for 2 to 3 years. Both used paired bilingual reading instructionby different teachers in Spanish and English, with transitionto allEnglish instructionby second or thirdgrade.The use of both Spanishand English readinginstructioneach day more resemblesthe experienceof Spanish-dominant students in many two-way bilingual programs(see Calder6n & Minaya-Rowe, 2003) thanit does transitional bilingualmodels, which delay Englishreadinguntil second or thirdgrade. One of the most widely cited studiesof bilingualeducationis a longitudinal study studentsin English by Ramirezet al. (1991), which comparedSpanish-dominant immersionschools with those in two forms of bilingualeducation:early exit (transition to Englishin Grades2-4) and late exit (transition to English in Grades5-6). Schools in several districtswere followed over 4 years. Immersionand early-exit studentsin four "two-treatment" schools were well matched,but a groupof "one treatment" schools thatimplementedonly bilingualor immersiontreatments were poorly matched,accordingto the authors.Late-exitstudentswere lower thantheir comparisongroups in socioeconomic status (SES), and their schools had much lower proportions of nativeEnglish speakers.For these reasons,no directcomparisons were madeby the authorsbetweenlate-exitand otherschools. The "two-treatment" comparisonof early-exit transitionalbilingual education and English immersionis the importantcontributionof the Ramirezet al. (1991) study. In the immersionprogram,almost all teacherspeech was in English at all gradelevels. In the early-exitclasses, teacherspeech was 31% Spanishin kindergarten, 29% in first grade, 24% in second grade, 17% in thirdgrade, and 2% in fourthgrade.The early-exitprogramwas describedas follows: In an early-exitprogramthere is some instruction in the child's primary 30 to 60 minutes language, perday.Thisis usuallylimitedto the introduction of initialreading skills.All otherinstruction is in English,withthechild'spriusedonly as a support, for clarification. However,instruction marylanguage in the primary is quicklyphasedout overthe nexttwo yearsso that language all instruction et al., 1991,p. 2) is in English.(Ramirez by gradetwo, virtually Although the Ramirezet al. study is invariablycited as a study of transitional bilingual education, it is in fact a study of paired bilingual education, using the definitionapplied in the presentreview. The authorsnoted that in the early-exit 35.1%of instructional time was spenton English languageartsand kindergartens, 29.9% on Spanish language arts. In first grade the correspondingnumberswere 33.5% (English) and 24.2% (Spanish). In contrast, in the English immersion time was spent on English languageartsin kinderclasses, 63.6% of instructional garten,and 60.2% in firstgrade. of timespenton math,socialstudies,andscience,all in Englishin both Percentages werenearlyidentical conditions. Overallinstructional in thetwo treatment treatments, time was equivalentin the two conditions,so the immersionand early-exitclasses were spendingaboutthe sameamountof time on languagearts,as EnglishandSpanish timein theearly-exit classes. was similar to Englishtimein theimmersion program The longitudinalexperimenthad two stages. Childrenwere pretestedin kindergartenand then posttestedat the end of firstgradeon the English Comprehensive 265
Slavin & Cheung
Test of Basic Skills (CTBS). The experimentaland control students were well matchedon pretests,SES, preschoolexperience,and otherfactors.On first-grade CTBS reading,studentsin the bilingualgroupscored significantlyhigherthanthe immersionstudents(ES = +0.53). The second phase of the study followed studentsfrom spring of first grade to thirdgrade. It is not clear why the kindergarten cohort was not simply followed, althoughthere were serious problems with attritionover the longitudinalstudy. However, the second-phasestudy does not qualify for inclusion in this review, as it uses a pretestgiven long aftertreatments had begun. The Ramirezet al. studywas so important in its time thatthe NationalResearch Council convened a panel in 1991 to review it and a study by Burkheimeret al. (1989). The panel's report(Meyer & Fienberg, 1992) supportedthe conclusions of the Ramirez et al. comparison of the early-exit and immersion programsin Grades K-l. It expressed concern about the second phase on the same criterion as that applied in the presentreview, that the pretestsfor the second phase were given afterthe studentshad alreadybeen in bilingualor immersiontreatmentsfor 2 years. In the mid-1970s, the AmericanInstitutesof Research(AIR) produceda series of reportson bilingualprogramsaroundthe U.S. (Campeau,Roberts,Oscar,Bowers, Austin, & Roberts, 1975). These are of some interest,with one majorcaveat: The AIR researchers were looking for exemplarybilingualprograms.They began with 96 candidatesand ultimately winnowed the list down to 8. Programswere excluded if datawere unavailable,not because they failed to show positive effects of bilingual programs.Nevertheless, these sites were chosen on their reputations for excellence, anda site would clearlybe less likely to submitdataif the datawere not supportiveof bilingualeducation.Also, the Campeauet al. (1975) evaluations were organizedas successive 1-year studies, meaning that pretests after the first treatmentyear (K or 1) are of little value. For reasons described earlier, we includedonly cohortsthatwere pretestedbeforetreatments began.Withthese cautions in mind, the Campeauet al. (1975) studies are describedbelow. A study in CorpusChristi,Texas, evaluateda pairedbilingualprogramin three schools. In kindergarten, the pairedbilingualtreatment involved a 2-hourlanguage and "In both periodalternating English Spanish. practices languages,the patterned stressbasic sentencepatterns andillustratechangesin wordformsandwordorder" the proportion of (Campeauet al., 1975, p. D-59). Over the course of kindergarten decreasedfrom90%to 50%. In firstgrade,the pairedbilingual Spanishinstruction classes had a daily 2-hourEnglish readingand languagearts period and a 1-hour Spanishreadingand languageartsperiod: Phoneticanalysisskillsaretaughtfirstin Spanishbecauseof the highlyphoneticnature of thelanguage.... Englishreadiness skills... aretaught through theHarcourt BraceJovanovich series.Phonetic analysisskillsareintroduced afterthechildhaslearned themthoughhis Spanish lessons.... There reading is a heavy phonicsemphasis,with comprehension skills receivingheavier et al., 1975,p. D-60) emphasisas decodingskillsdevelop.(Campeau Classes in the control groups were "taughtmonolinguallywithout regardfor the stratelanguagedominanceof the children"(p. D-66), but otherwiseinstructional gies were similar. 266
andcontrolclasses were pre- andposttestedeach yearfor 2 years. Experimental For reasons discussed earlier,we did not include comparisonsin which pretests were given after treatmentswere already under way. For this reason, we only focused on the childrenwho were pretestedin kindergarten in the firststudy year and then posttestedat the end of first grade in the second study year. This cohort was well matchedon the StanfordEarlySchool AchievementTest. Therewere no differences at the end of kindergarten,but the paired bilingual students scored significantly higher than controls on the Inter-AmericanEnglish Reading Test (ES = +0.45) at the end of firstgrade.Therewere also substantialdifferenceson the StanfordReadingAssessment. Experimental studentsaverageda gradeequivalent of 2.3, controls 1.8. However, no statisticalcomparisonswere made on the Stanford. The paired bilingual group also scored substantiallybetter than the controlgroupin Spanishreading,of course. A study in Houston also reportedby Campeauet al. (1975) followed three cohortsof studentsin seven pairedbilingual and two English immersionschools. The pairedbilingual programbegan with a Spanish decoding programand then Spanish Laidlaw basals, duringa regularSpanish arts block. Studentswere then taughtEnglish reading, startingwith a transitionalprogramand then continuing with the Harcourt Brace Jovanovichbasal series. Each class had a teacherand an aide. Two control schools, in which Spanish-dominant studentswere taughtonly in English, were selected on the basis of similarityto seven experimentalschools in language,SES, and priorachievement. andcontrolkindergarten studentswere well matchedon the InterExperimental American English Ability Test. At the end of first grade and second grade, the pairedbilingualstudentsscoredsubstantially higherthanthe controlstudents,with effect sizes of +0.69 and +0.54, respectively. Cohen (1975) compared two schools serving many Mexican Americans in RedwoodCity, California.One school was using whatamountsto a two-way bilinstudentsand English-dominantstudents gual program,in that Spanish-dominant were taughtin both Spanish and English. However, from the perspective of the ELLs, the treatmentwas the same as a pairedbilingualmodel. Spanish-dominant studentswere taughtSpanishreadingusing readerssuch as Pepin en PrimerGrado andEnglishreadingusing the phoneticMiamiLinguisticSeries. Spanish-dominant and English-dominant studentswere groupedtogetherfor contentareainstruction but not for reading.Three successive cohorts were comparedat the two schools: GradesK-l, 1-2, and 1-3. In each case, studentswere pretestedandposttestedon a broad range of English reading measures. In all cohorts, Mexican-American students were well matched on English and Spanish pretests. At posttest, there were no significantdifferences, adjustingfor pretests.The data did not allow for computationof effect sizes, so zeros were enteredin Table 1. Maldonado(1977) conducteda 5-year longitudinalstudy on a group of Mexican American children in six elementary schools in Corpus Christi, Texas. The main purposeof this study was to examine how well the bilingual studentswere able to succeed in the regulareducationprogramof the school districtafter they had left the bilingual program.The experimentalgroup consisted of 47 children who had participated in the bilingual programfor 4 consecutive years, from first to fourthgrades.The controlgroupwas comprisedof 79 childrenwho had been in a regular English-only program for the same 4 years and grades. The study 267
Slavin & Cheung
followed the two groupsuntil they reachedfifth grade, 1 year afterthe experimental group left the bilingual program.Studentsin the experimentalgroupenrolled in the Title VII programtitled "Apprendemos En Dos Idioma"(We Learnin Two in which of 2 hours per day in Spanish received "a minimum Languages), they language instructionin the areas of language arts, reading,and mathematicsand social studies" (p. 103). No specific descriptive informationabout the control groupwas providedin the study. No statisticallysignificantdifferenceswere found at any gradeaftercontrolling for first-gradepretests. It is importantto note that teachers in both the bilingual group and the control group were bilingual. However, it is not stated how much these bilingual teachers in the control group used Spanish in their classrooms to stated, help childrenwho were in need for bilingualexplanations.As the researcher "Itis highly possible thatthe controlgroupbilingualteachersmight have used the Spanishlanguage for clarificationof some concepts. This in turnwould not only assist those studentsin the comprehensionof those concepts but at the same time lower the differencebetween the groupsin the areasof mathematicsandreading" (p. 104). A study by Alvarez (1975) followed 147 Mexican-Americanchildren in two schools in Austin, Texas, from first to second grades. Students in the bilingual classrooms and the control classrooms in each school were well matched on SES and initial language proficiency. The instruction programin the bilingual classrooms was described as "a balanced combination of Spanish and English (50 percent/50 percent)" (p. 73). However, children who had very limited English proficiency in these bilingual classrooms were initially taught in Spanish. Reading and language arts were taught in both languages each day for 2 hours by the English-speaking and Spanish-speakingteachers. Oral language in English was taughtfor one-half hour daily to the Spanish-dominantchildren, and oral language in Spanish was taught for one-half hour to the Englishdominant children. Studentsin the controlclassroomshad a curriculum similarto thatof the bilingual classrooms. The only difference was "that the subject matter was taught completely in English;using the same textbooksfor the Englishcomponentof the bilingual classes, but using twice as much time as the balanced combinationof Spanishand English curriculum designed for Mexican studentsin bilingualclassrooms"(Alvarez, 1975, p. 75). The bilingualstudentsscored somewhathigheron the English reading vocabularytest, but the control group scored higher on the English reading comprehension test than the bilingual classes. None of these differenceswere statisticallysignificant. Two of the studies carriedout by Campeauet al. (1975) had 1-yeardurations. The first study was conductedin two low-SES elementaryschools in Kingsville, Texas. Five gradeswere comparedat the two schools: K-4. Only the resultsfrom the kindergarten groups-48 in the bilingualgroupand 41 in the controlgroupafterthe are interpretable because the pretestsfor othergradeswere administered treatments began. All teachers in kindergartenwere bilingual. Instruction time was equally divided between Spanish (50%) and English (50%). Kindergarteners with very limited English were taughtprimarilyin Spanish until their English proficiency reacheda point where they could cope with bilingual instruction. 268
At the end of the study, studentsin both groups were given the same posttests. The mean gains between the two groups were compared. Students in all six kindergartenclasses gained significantly more on the English version of the Inter-American Series thantheir counterparts in the control group,with an effect size of +0.42. Another1-yearstudyin SantaFe, New Mexico, comparedpairedbilingualand immersion programs for Spanish-dominantstudents. Pre- and posttests were as pretestsfor other reportedfor each year, but only first gradewas interpretable, years had already been affected by the treatments.Parentschose to place their children in bilingual or English programs, and apparently parents of higherachieving childrenchose the bilingualgroup,as pretestscores were higherin that group. Studentsin the bilingual classrooms received "a bilingual presentationof all the topics of study in the normalcurriculum" (p. D-16). For example, students were taughtcertainconceptsin Spanishin the morningandwere retaughtthe same lesson in Englishin the afternoon.In addition,studentsin the bilingualclassrooms were grouped by ability during language arts and reading periods. No specific descriptioninformationwas providedaboutthe controlclassrooms.At the end of the study, the bilingual group gained slightly more than the control group in English reading, with an effect size of +0.03. However, the bilingual group also gained more in English reading than the English-only group. No standard deviationswere given, so effect sizes for pretestdifferencesandgains could not be computed. Studies of 1-YearTransitionalBilingual Education J. A. Maldonado (1994) carried out a small, randomized study involving English language learners who were in special education classes in Houston. Twenty students in second and third grades with learning disabilities were randomlyassigned to one of two groups. A bilingual group was taughtmostly in Spanishfor a year, with a 45-minuteESL period.Duringa second year, half of the instructionwas in English, half in Spanish.In a thirdyear, instructionwas only in English. The control group was taught in English all 3 years. Students in both groups received the same amount of reading and language instructionand were taughtby similarteachers. Childrenwere pretestedon the CTBS and then posttestedon the CTBS 3 years later.At pretest,the controlgroupscorednonsignificantly higherthanthe bilingual group,but at posttest the bilingual group scored far higher. Using the means and standarddeviations presentedin the article, the effect size would be +8.33, but using the given values of t and n, the effect size is + 1.67, a more credibleresult. One of the Campeauet al. (1975) studies,in Alice, Texas, comparedSpanishdominantstudentsin bilingualand Englishimmersionprogramsstarting in kinderinvolved teachingkindergartners in garten,for a 2-yearexperiment.The treatment to Englishreadingand were then Spanish.In firstgrade,childrenwere transitioned controlstudents weretaught taughtequalamountsof timein eachlanguage.Matched werecomparable atpretest on Englishmeasures onlyin English.Whilekindergartners of generalability,bilingualstudentsscoredsubstantially higheron a Spanishability test.At posttest(controlling forpretests), scoredsubstantially better bilingualstudents on the Inter-American Englishreadingtest at the end of firstgrade,after2 years of bilingualeducation(ES = +0.49). 269
Studyof UnspecifiedBilingual Education In a poorly specified studyof bilingualeducation,Saldate,Mishra,and Medina (1985) studied62 childrenin an Arizonabordertown who attendedimmersionor model. The to be a transitional bilingualschools. The bilingualtreatment appeared childrenwere individuallymatchedon the PeabodyPictureVocabulary Test in first grade. At the end of second grade, the bilingual studentsscored nonsignificantly lower on the English MetropolitanAchievement Test (MAT) (ES = -0.28) and higher on the SpanishMAT (ES = +0.44). This was to be expected, as they had not yet transitionedto English instruction.At thirdgrade, however, the bilingual outstudents(who had now transitioned to English-onlyinstruction)substantially performedthe immersionstudentsboth in English (ES = +0.89) and in Spanish (ES = +3.01). This study's small size means thatits resultsshould be interpreted cautiously, especially as the number of pairs dropped from 31 to 19 between second and thirdgrades. Studiesof Heritage Languages Two qualifying studies involved languages other than Spanish. These are reviewed separately, not for this reason alone but also because the languages whose use (Frenchin Louisiana,Choctawin Mississippi)are"heritage languages," was intendedas muchto show respectto children'sculturesas to help non-English speakerssucceed in reading. Morgan (1971) carriedout a study of almost 200 children of French-heritage parentsin ruralLouisiana.Existinggroupsof firstgraders,assignedto bilingualor monolingualclasses, were followed for a year. In the bilingual classes, children were taughtin both Frenchand English. The two groups were virtuallyidentical on English tests of mentalabilitiesandreadinessat the beginningof firstgrade.At the end, the childrentaughtin the bilingualclasses scored higher on four English readingmeasures,with a mediandifferenceof +0.26. Differenceswere significant on measuresof word readingand paragraph reading,but not vocabularyor word study skills. It is importantto note, however, that the childrenin this study were English proficient. Their parents may have spoken some French at home, but both experimentaland control studentsscored well at preteston an English mental abilities test. A 1-yearstudyof 63 Choctawsecond gradersin Mississippi compareda bilingual programin Choctaw and English with English-only instruction(Doebler & Mardis, 1980-1981). There were no significantdifferenceson an English reading measure,controllingfor pretests. Studiesof SecondaryReading Two qualifying studies evaluatedprogramsthat introducedSpanish-language instruction to ELLsin the secondarygrades.Both of these usedrandomassignment. ninth Covey (1973) randomlyassigned 200 low-achieving Mexican-American gradersto bilingualor English-onlyclasses. The experimentalinterventionis not describedin any detail, but it clearly involved extensive use of Spanishto supplement English in reading,English, and math.The groups' scores were nearlyidentical at pretest,but at posttest the bilingual studentsscored significantlybetteron the StanfordDiagnostic ReadingTest (ES = +0.72). 270
Research on Language of ReadingInstructionforELLs
Kaufman(1968) evaluateda program in which low-achievingSpanish-speaking seventhgraderswere randomlyassignedto bilingualor English-onlyconditionsin two New Yorkjunior high schools. One school participated in the programfor a and the for 2 other In received three or the students year years. bilingual classes, four periodsof Spanishreadinginstruction each week, while controlswere in art, music, or health education. On standardized tests of reading, there were differencesfavoringthe bilingualclasses in the 2-yearschool andthe nonsignificant 1-yearschool (ES = +0.23 for both).The secondarystudiespointto the possibility that providing native language instructionto low-achieving ELLs in secondary school mayhelp themwith Englishreading.This application is worthyof additional research(also see Klingner& Vaughn,2004). CanadianStudies of French Immersion There are several Canadianstudies (e.g., Lambert& Tucker, 1972) that have role in debatesaboutbilingualeducation.These are studiesof played an important Frenchimmersionprograms, in which English-speaking childrenaretaughtentirely in Frenchin theearlyelementary orprimarily andBaker(1996) emphaRossell years. sized these studies as examples of "structured the approach immersion," English favoredin theirreview. However,Willig (1985) and otherreviewershave excluded them.TheCanadian of thisreviewbecause studiesdo notmeettheinclusionstandards theAnglophone children werelearning a usefulsecondlanguage,notthelanguagefor whichthey wouldbe held accountable in theirlaterschooling.Althoughmanyof the studiestookplacein Montr6al, lived in English-speaking the children neighborhoods and attendedschools in an Englishsystem.The focus of this review is on bilingual education used to help childrensucceedin the languagein whichthey will be taught in the later grades;but the Frenchimmersionchildrenin Canadawere headed to these studies all involved voluntary English secondary schools. Furthermore, in whichparents wantedtheirchildrento lear French,andthe childrenin programs, these studieswere generallyuppermiddleclass, not disadvantaged. Because French immersion programswere voluntary, children who did not thrive in them could be and routinely were returnedto English-only instruction. This means that the children who completed French immersion programs in Canada were self-selected relatively high achievers. Most important, the "bilingual" programsto which Frenchimmersionis comparedwere nothing like bilingualeducationin the United States.At most, childrenreceived 30 to 40 minutes daily of French as a second language, with far less time in French reading instructionthan a U.S. student in a bilingual programwould receive in English duringand aftertransition(see Genesee & Lambert,1983). Yet in many studies, English comparisongroups were not learning French at all. In the widely cited study by Lambertand Tucker(1972), Anglophones in Frenchimmersionclasses were comparedwith Anglophonestaughtonly in English and with Francophones taughtonly in French.Ironically,studies of this kind, cited by Rossell and Baker (1996) as comparisonsof immersionand bilingualeducation,are in fact comparisons of immersionand monolingualeducation.If they existed, Canadianstudies of, say, Spanish speakerslearningFrench in Francophoneschools in Quebec or English in Anglophone schools in the rest of Canadawould be relevant to this review. But studies of voluntary immersion programsas a means to acquiring Frenchas a second languageare only tangentiallyrelevant. 271
Slavin & Cheung
While the Canadianimmersionstudies are not directlyrelevantto the question of the effectiveness of bilingualprogramsfor ELLs learningthe societal language, of the role of they are neverthelessinterestingin gaining a broaderunderstanding native language in foreign language instruction.As a group, these are matched studiesof high methodologicalquality.Quite in contrastto U.S. studies,however, the focus of the Canadian studies is on whether French immersion harms the English languagedevelopmentof native English speakers.It is taken as obvious thatFrenchall day will producemorefacility in Frenchthan30 to 40 minutesdaily in second-languageclasses. Overall, the Canadianstudies paint a consistent picture (Lambert& Tucker, 1972; Lambert & Tucker, 1977; Barik & Swain, 1975; Barik, Swain, & Nwanunobi, 1977; Genessee & Lambert,1985; Day & Shapson, 1988). At least for the overwhelminglymiddle-classstudentsinvolved, Frenchimmersionhad no negative effect on English readingachievement,and it gave studentsfacility in a second language.The relevance to the U.S. situationis in suggesting that similar second-language immersion programs, perhaps including two-way bilingual children, are not likely to harm English reading programsfor English-proficient development.However, the relevanceof these studies to any context in which the childrenof immigrantsare expected to learn the language that they will need to succeed in theirschool and in the largersociety is minimal. Comparisonof Paired Bilingual and TransitionalBilingual Programs As noted earlier,many of the programswith the strongestpositive effects for in which childrenwere Englishlanguagelearnersused a pairedbilingualapproach, in both and their at different times each day native taughtreading English language from the beginning of their schooling. This approachcontrastswith transitional bilingual education (TBE) models, in which children are first taught to read primarilyin theirnative languageand only then transitioned graduallyto Englishonly instruction.Only one study has compared reading outcomes of these two bilingualapproaches. A longitudinalstudyby Gerstenand Woodward(1995) initially favoredpaired over TBE but laterfound the two to be equivalent.This study bilingualinstruction was carriedout with Spanish-dominant ELLs in 10 elementaryschools in El Paso. Five schools used a programin which all subjectswere taughtin Englishbut Spanish instructionwas also provided,for 90 minutesdaily in firstgradeand declining to 30 minutes daily in fourthgrade. The transitionalbilingual programinvolved mostly Spanishinstruction,with 1 hourper day for ESL instruction,with gradual transitionto Englishcompletedin the fourthor fifth grade.The childrenwere well matcheddemographicallyon entry to first grade and scored near zero on a measure of English languageproficiency.In Grades4, 5, 6, and 7, Iowa Tests of Basic Skills were comparedfor the two groups. On Total Reading, the pairedbilingual studentsscoredsignificantlyhigherthanthe transitional bilingualstudentsin fourth grade (ES = +0.31), but the effects diminishedin fifth grade (ES = +0.18) and were very smallin the sixth (ES = +0.06) andseventhgrades(ES = +0.08). Tests of languageand vocabularyshowed similarpatterns.This patternis probablydue to the fact thatthe transitional bilingualstudentshad not completedtheirtransition to English in fourthand fifth grades.When they had done so, by sixth grade,their was nearlyidenticalto thatof the othergroup. readingperformance 272
Research comparingalternativebilingual models is far from conclusive, but to introduce nothing suggests thatit is harmfulto children'sreadingperformance both native languageand English readinginstructionat differenttimes each day. Discussion The most important conclusion from researchcomparingthe relativeeffects of bilingual and immersionprogramsfor English language learnersis that there are far too few high-quality studies of this question. Willig (1985) and Rossell and Baker (1996) agree on very little, but both of these reviews call for randomized, longitudinalevaluationsto producea satisfyinganswerto this criticalquestion.Of course,manywould arguethatrandomizedevaluationsareneededon most importantquestionsof educational practice(see, for example,Mosteller& Boruch,2002; Slavin, 2003); however,in bilingualeducation,they areespecially crucialbecause of the many inherentproblemsof selection bias in this field. Furthermore, this is an area in which longitudinal,multiyearstudies are virtuallymandatory,to track childreninitiallytaughtin theirnativelanguagethroughtheirtransition to English. of this studies are Finally, althoughrandomized,longitudinal sorely needed, topic there are simply too few experimentalstudies of all kinds, including ones with matchedexperimentaland controlgroups. Despite these concerns,however, the existing researchon languageof instruction yields some important lessons at least worthyof furtherstudy.Across 17 qualifying studies of all types of programs, 12 found effects favoring bilingual educationand 5 found no differences.None of the studies found results favoring English immersion. The largest group of studies focused on elementary reading instructionfor students.Nine of 13 studies in this category favored bilingual Spanish-dominant approaches,and four found no differences.The medianeffect size for all 13 studies was +0.45. This effect size is higher than the estimate of +0.21 given by Greene (1997), but Greene did not locate the Campeauet al. (1975) studies that added several positive effect sizes. Also, many of the largestpositive effect sizes were from studieswith very small sample sizes. The mean effect size weighted by students samplesize for the 13 studiesof elementaryreadingfor Spanish-dominant was +0.33. Using proceduresdescribedby Lipsey and Wilson (2001), this effect was found to be significantlydifferentfrom zero, Q = 29.6, p < .05, df = 12. The weighted mean for the threerandomizedstudies was +0.62, Q = 8.53, p < .05. It was surprisingthatmost of the methodologicallyadequatestudies evaluated forms of bilingual educationquite differentfrom those commonly used in recent years. These arepairedbilingualprograms,in which childrenaretaughtto readin English andin theirnativelanguageat differenttimes each day fromthe beginning of their time in school. Another category of programsprovided just 1 year of native-languageinstructionbefore the transitionto English-only reading.Paired bilingual strategies were used in two of the randomized studies (Huzar, 1973; Plante, 1976), and in a study of a 1-yeartransitionalprogram(J. A. Maldonado, 1994). These practicescontrastwith practicesin transitionalbilingual education, in which childrenaretypicallytaughtto readin theirnative languagefromkinderto reading. gartento Grades2, 3, or 4, and then transitioned There are several possible explanationsfor the prevalenceof the pairedbilingual interventions amongthe studiesreviewed.First,most of the studiestook place 273
Slavin & Cheung
in the 1970s, when Title VII was new. At thattime, pairedbilingual models were popular.Second, for reasonsdiscussedearlierin this review, studiesof transitional bilingual educationare very difficult to perform,as they should begin in kindergartenand continuepast the point of transition.A 4-year longitudinalstudywould be requiredto follow children from kindergartento third grade. Allowing for studentmobility, such a study must startwith a large sample to end up with sufficient numbersof students.The U.S. Department of Educationhas recentlyfunded two matchedand one randomizedlongitudinalstudyto evaluatetransitional bilingual education, but before these only the Ramirez et al. (1991) study had the resourcesto carryout an investigationof this kind, and it did not follow a consistent sample from kindergarten to thirdgrade. It is important to note thatmost of the studies thatdid not qualify for inclusion also used pairedbilingual models, not transitional bilingualmodels. A key exception was a series of studies by Thomas and Collier (2002) that followed children who hadbeen in transitional programsbut lackedpretestmeasuresfrombeforethe startof the TBE interventions. Because of the dearthof studies of TBE, it is not currentlypossible to say with confidence whetherpairedbilingual models are more effective than transitional models. Only one study,by GerstenandWoodward(1995), madethis comparison. It found differencesfavoringpairedbilingual strategiesin Grades4 and 5 but not in Grades6 and 7. However, given the supportfor pairedbilingual methods seen in this review, it is worthwhileto speculate about why pairedmethods might be beneficial. Teaching a Spanish-speaking English language learner in Spanish can be expected to establish the alphabeticprinciple, the idea that words are composed of distinct sounds represented by letters (see National Reading Panel, 2000). Early in their reading instruction,children learn to combine letters and sounds into words they know. This process is very difficult if children must form letters and sounds into words they don't know, so it may greatly facilitate phonetic development to learn the alphabeticprinciple in a familiar language ratherthan an unfamiliar one. Once a Spanish-speaking child can confidently decode Spanish text, he or she should be able to make an easy transferto decoding any alphabetic language, such as English, by learning a modest number of new sounds for particulargraphemes (Lindsey, Manis, & Bailey, 2003). Several of the studies of pairedbilingual instructionclearly describeda process of teaching Spanish reading phonetically and then, in a planned manner,transferringthose skills to English decoding. Rather than confusing children, as some have feared, reading instruction in a familiar language may serve as a bridge to success in English, as phonemic awareness, decoding, sound blending, and generic comprehension strategies clearly transfer among languages that use phonetic orthographies, such as Spanish, French, and English (see August, 2002; August, Calder6n, & Carlo, 2001; August & Hakuta, 1997; Durgonolglu, Nagy, & Hancin-Bhatt, 1993; Fitzgerald, 1995; Garcia, 2000; Lee & Schallert, 1997; Lindsey, Manis, & Bailey, 2003). Only two studies of secondaryprogramsmet the inclusion criteria,but both of these were very high quality randomized experiments. Covey (1973) found substantial positive effects of Spanishinstructionfor low-achievingninthgraders; 274
Kaufman(1968) found mixed, but slightly positive, effects of a similarapproach with low-achieving seventh graders. As noted previously, research on language of instruction may suffer from publicationbias: the tendencyfor journalsto publishonly studies thatfind significant differences.However, dissertationsandtechnicalreports(e.g., Covey, 1973; Huzar, 1973; Plante, 1976) less likely to suffer from publicationbias also tended to favor bilingualprograms. Teaching reading in two languages, with appropriate adaptations of the English program for the needs of English language learners, may represent a satisfactory resolution to the acrimonious debates about bilingual education. Proponentsof bilingual educationwant to launchEnglish language learnerswith success while maintainingand valuing the language they speak at home. Opponents are concerned not so much about the use of native language but about delaying the use of English. Paired bilingual models immerse children in both English reading and native language reading at the same time. They are essentially half of a two-way bilingual model; by encouraging English-proficient students to take Spanish reading, any school with a paired bilingual model can readily become a two-way program, offering English-only children a path to early acquisition of a valuable second language (see Calder6n& Minaya-Rowe, 2003; Howardet al., 2003). Languageof instructionmust be seen as only one aspect, however, of instructional programmingfor English language learners.As many previous reviewers have concluded, quality of instruction is at least as importantas language of instruction(for reviews of effective programsand practicesfor ELLs, see August & Hakuta,1997; Fitzgerald,1995; Klingner& Vaughn, 2004; Slavin & Cheung, 2005, in press). Clearly,there is much more thatwe need to know aboutthe role of native language instructionin reading.The researchreviewed in this article may represent the best experimentalstudies currentlyavailable, but better evidence is needed. treatLongitudinal experiments using randomassignmentof studentsto alternative ments are particularlyneeded. Both qualitative and quantitativeresearch are needed to illuminatethe conditionsunderwhich native language instructionmay be beneficial for developing English reading skills, and to explain those effects. Studies that systematically vary programcomponents and studies that combine andqualitativemethodsareneededto morefully explain how various quantitative interventionsaffect the development of reading skills among English language learners.It is time to end the ideological debates and to focus instead on good science, good practice,and sensible policies for childrenwhose success in school means so much to them, to theirfamilies, and to our nation's future. Note This articlewas writtenunderfundingfrom the Instituteof EducationSciences of Education No. R305A040082).However,the views (IES),U.S. Department (Grant do not necessarily reflectIESpositionsor policies. expressed We wouldlike to thankMargarita Diane August,Tim Shanahan, Isabel Calder6n, andBetteChambers for theircomments Beck, JayGreene,NancyMadden, on an earlier draft,andSusanDavis andthe Centerfor AppliedLinguistics for assistancewith the literature search. 275
References Versus of Texas,Austin. Classrooms. Doctoral dissertation, Monolingual University of skills fromSpanishto M., & Carlo,M. (2001). The transfer August,D., Calder6n, NABENews,May/June, 11-12, 42. English:A studyof younglearners.
August, D. (2002). English as a second language instruction: Best practices to support the developmentof literacyfor English language learners. Baltimore: Johns Hopkins August, D., & Hakuta,K. (1997). Improvingschooling for language-minority children: with Cherokee Indian students.Journal of American Indian Education, 21(2), 34-43. Alvarez, J. (1975). Comparison of academic aspirations and achievement in bilingual
Placedat Risk. Center for Research on the Education of Students University,
A researchagenda.Washington, Council. DC:NationalResearch Bacon,H., Kidd,G., & Seaberg,J. (1982). The effectivenessof bilingualinstruction
education. Review of Educational Research, 57, 351-362. Baker, K., & de Kanter, A. (1981). Effectiveness of bilingual education: A review of
of selectedstudiesin the effectivenessof bilingual Baker,K. (1987). A meta-analysis
theliterature. andAnalytic DC:Officeof Technical (Finaldraft report). Washington, of Education. Systems,U.S. Department A. (1983). An answerfromresearch on bilingualeducation. Baker,K., & de Kanter,
American Education, 56, 157-169.
Balasubramonian, K., Seelye, H., & Elizondo de Weffer, R. (1973). Do bilingual of Englishto of Teachers at the 7thAnnualConvention experiment. Paper presented of OtherLanguages, SanJuan,PuertoRico. Speakers evaluationof a large-scaleearly grade Barik,H., & Swain, M. (1975). Three-year Frenchimmersion The Ottawastudy.Language 25(1), 1-30. Learning, program:
education programs inhibit English language achievement? A report on an Illinois
Barik, H., & Swain, M. (1978). Evaluation of a bilingual education program in Canada: The Elgin Study through grade six. Commission interuniversitaireSuisse tion: The Elgin study through grade five. Canadian Modem Language Review, 33,
de linguistique Geneva,Switzerland. appliquee, E. A. (1977). English-French Barik,H., Swain,M., & Nwanunobi, bilingualeduca459-475.
Bates, E., & May B. (1970). The effects of one experimental bilingual program on verbal ability and vocabulary offirst grade pupils. Doctoral dissertation, Texas Tech Borenstein, M. (2005). Comprehensive meta-analysis software. Englewood, NJ:
Lubbock. University,
BioStat. of bilinW. E., & Tucker,G. R. (1977). Cognitiveconsequences Bruck,M., Lambert, 24, 13-33. Project gual schooling:The St. Lambert through gradesix. Linguistics, K. A. G. J., Conger,A. J., Dunteman, G. H., Elliott,B. G., & Mowbray, Burkheimer, students. DC:U.S. Department of Education. Washington, and methodologiesused to teach Spanish-speaking Calderon,M. (2001). Curricula limitedEnglish-proficient studentsto readEnglish.In R.E. Slavin & M. Calderon
(Eds.), Effective Programs for Latino Students. Mahwah, NJ: Erlbaum. Calder6n, M., & Minaya-Rowe, L. (2003). Designing and implementingtwo-way bilin(1989). Effectiveness of services for language-minority limited-English-proficient
Thousand Oaks,CA: Corwin. gual programs. on the Carlisle,J. F., & Beeman,F. F. (2000). The effects of languageof instruction Studies children. of first-grade Scientific Hispanic readingandwritingachievement P. L., Roberts,A., Oscar,H., Bowers,J. E., Austin,M., & Roberts,S. J. Campeau, for Research. Institutes grams.PaloAlto, CA:American 276
(1975). The identification and description of exemplary bilingual education proof Reading, 4(4), 331-353.
39(2), 188-215. Ciriza, F. (1990a). Evalution report of the Preschool Project for Spanish-speaking
Carlo,M. S., August,D., McLaughlin, D., et B., Snow, C. E., Dressier,C., Lippman, al. (2004). Closingthe gap:Addressing the vocabulary needs of Englishlanguage learnersin bilingual and mainstream classrooms.ReadingResearch Quarterly, 1989-90. SanDiego, CA: SanDiego City Schools. Children,
Cohen, A. D. (1975). A sociolinguistic approach to bilingual education. Rowley, MA:
NewburyHousePress. research(3rded.). Thousand Oaks,CA: Sage. Cooper,H. (1998). Synthesizing New York:RussellSageFoundation.
Cooper, H. M., & Hedges, L.V. (Eds.). (1994). The handbook of research synthesis. Covey, D. D. (1973). An analytical study of secondary freshmen bilingual education and its effects on academic achievement and attitudes of Mexican American
students. Doctoraldissertation, ArizonaStateUniversity, Tempe. P. (1980). Achievedreadylevel, selfCuriel,H., Stenning,W., & Cooper-Stenning, to lengthof exposureto bilingualeducation. esteem,andgradesas related Hispanic
Journal of Behavioral Sciences, 2, 389-400. Title VIISpanish/English Bilingual Education Programs. Palo Alto, CA: American
Danoff,M. N., Arias,B. M., & Coles, G. J. (1977). Evaluation of theImpactof ESEA Institutes for Research. D. H., & Reynolds,D. J. (1977).Evaluation Danoff,M. N., Coles, G. J., McLaughlin, Research. D. H., & Reynolds,D. J. (1978). Evaluation Danoff,M. N., Coles, G. J., McLaughlin,
of the impact ofESEA Title VIISpanish/English bilingual education programs: Vol. Day, E. M., & Shapson, S. M. (1988, April). Provincial assessment of early and late French immersion programs in British Columbia, Canada. Paper presented at the de Weffer, R. (1972). Effects offirst language instruction in academic and psychological development of bilingual children. Doctoral dissertation, Illinois Institute of of the impact ofESEA Title VIISpanish/English Bilingual Education Programs: Vol. I. Study Design and Interim Findings. Palo Alto, CA: American Institutes for
3. YearTwoimpactdesigns.Palo Alto, CA:American for Research. Institutes
annualmeetingof the American Research New Orleans. Educational Association,
Technology, Chicago. L. J. (1980-1981). Effectsof a bilingualeducation Doebler,L. K., & Mardis, program for NativeAmerican children. NABEJournal,5(2), 23-28. transfer B. J. (1993). Cross-language A., Nagy, W. E., & Hancin-Bhatt, Durgunoglu,
of phonological awareness. Journal of Educational Psychology, 85(3), 453-465. Educational Operations Concepts, Inc. (1991a). An evaluation of the Title VII ESEA bilingual education program for Hmong and Cambodian students in kindergarten andfirst grade. St. Paul, MN: Author. Educational Operations Concepts, Inc. (1991b). An evaluation of the Title VII ESEA bilingual education program for Hmong and Cambodian students in junior and senior high school. St. Paul, MN: Author. El Paso Independent School District. (1987). Interim report of thefive-year bilingual education pilot, 1986-1987 school year. El Paso, TX: Office for Research and
Evaluation.
year in a longitudinal study.AvailablefromOffice for Researchand Evaluation, 6531 BoeingDrive,El Paso,TX 79925. El PasoIndependent SchoolDistrict. Available evaluation. (1992).Bilingualeducation fromOfficefor Research andEvaluation, 6531 BoeingDrive,El Paso,TX 79925. J. (1995). English-as-a-second-language in the UnitedStates: instruction Fitzgerald,
A research review. Journal of Reading Behavior, 27, 115-152.
El Paso IndependentSchool District. (1990). Bilingual education evaluation: The sixth
277
Slavin & Cheung Garcia, G. (2000). Bilingual children's reading. In M.L. Kamil, P. B. Mosenthal, P. D. Pearson, & R. Barr (Eds.), Handbook of reading research (Vol. 3, pp. 813-834). Mahwah, NJ: Erlbaum. Genesee, F., Lambert, W. E., & Tucker, G. R. (1977). An experiment in trilingual education. Montreal, Canada:McGill University. Genesee, F., & Lambert, W. E. (1983). Trilingual education for majority-language children. Child Development, 54, 105-114. Gersten, R. (1985). Structuredimmersion for language minority students: Results of a longitudinal evaluation. Educational Evaluation and Policy Analysis, 7(3), 187-196. Gersten, R., & Woodward, J. (1995). A longitudinal study of transitional and immersion bilingual education programs in one district. Elementary School Journal, 95(3), 223-239. Greene, J. P. (1997). A meta-analysis of the Rossell & Baker review of bilingual education research. Bilingual Research Journal, 21(2/3). Grigg, W., Daane, M., Jin, Y., & Campbell, J. (2003). The nations's report card: Reading 2002. Washington, DC: U.S. Department of Education. Hakuta, K., Butler, Y. G., & Witt, D. (2000). How long does it take English learners to attain proficiency? (Policy Report 2000-1). Santa Barbara, CA: University of California Linguistic Minority Research Institute. Howard, E. R., Sugarman, J., & Christian, D. (2003). Two-way immersion education: What we know and what we need to know. Baltimore: Johns Hopkins University, Center for Research on the Education of Students Placed at Risk. Huzar,H. (1973). The effects of an English-Spanish primary grade reading program on second and third grade students. Master's thesis, Rutgers University, Camden, NJ. Kaufman, M. (1968). Will instruction in reading Spanish affect ability in reading English? Journal of Reading, 11, 521-527. Klingner, J. K., & Vaughn, S. (2004). Specific strategies for struggling second language readers and writers. In T. L. Jetton & J. A. Dole (Eds.), Adolescent literacy research and practice. New York: Guilford. Lambert,W. E., & Tucker, G. R. (1972). Bilingual education of children: The St. Lambert Experience. Rowley, Quebec, Canada: Newbury House. Layden,R. G. (1972). The relationshipbetween the language of instructionand the development of self-concept, classroom climate, and achievement of Spanish speaking Puerto Rican children. Doctoral dissertation,University of Maryland,College Park. Lee, J., & Schallert, D. L. (1997). The relative contributionof L2 language proficiency and L1 reading ability to L2 reading performance: A test of the threshold hypothesis in an EFL context. TESOLQuarterly, 31, 713-739. Legarreta, D. (1979). The effects of program models on language acquisition by Spanish-speaking children. TESOLQuarterly, 13(4), 521-534. Lindsey, K. A., Manis, F. R., & Bailey, C. E. (2003). Prediction of first-gradereading in Spanish-speaking English-language learners.Journal of Educational Psychology, 95(3), 482-494. Lipsey, M. W., & Wilson, D. B. (2001). Practical meta-analysis. Thousand Oaks, CA: Sage. Lum, J. B. (1971). An effectiveness study of English as a second language (ESL) and Chinese bilingual methods. Doctoral dissertation, University of California, Berkeley. Maldonado, J. A. (1994). Bilingual special education: Specific learning disabilities in language and reading. Journal of Education Issues of Language Minority Students, 14, 127-147. Maldonado, J. R. (1977). The effect of the ESEA Title VIIprogram on the cognitive development of Mexican American students. Unpublished doctoral dissertation, University of Houston, Houston, TX. 278
Research on Languageof ReadingInstruction for ELLs Matthews, T. (1979). An investigation of the effects of background characteristics and special language services on the reading achievement and English fluency of bilingual students. Seattle, WA: Seattle Public Schools, Department of Planning, Research, and Evaluation. Melendez, W. A. (1980). The effect of the language of instruction on the reading achievement of limited English speakers in secondary schools. Doctoral dissertation, Loyola University of Chicago. Meyer, M. M., & Fienberg, S. E. (1992). Assessing evaluation studies: The case of bilingual education strategies. Washington, DC: National Academy of Sciences. Morgan, J. C. (1971). The effects of bilingual instruction of the English language arts achievement of first grade children. Doctoral dissertation, Northwestern State University of Louisiana. Mosteller, F., & Boruch, R. (2002). Evidence matters: Randomized trials in education research. Washington, DC: Brookings Institution Press. National Center for Education Statistics. (2004). Language minorities and their educational and labor market indicators: Recent trends. Washington, DC: U.S. Department of Education. National Reading Panel. (2000). Teaching children to read: An evidence-based assessment of the scientific research literature on reading and its implications for reading instruction. Rockville, MD: National Institute of Child Health and Human Development. Olesini, J. (1971). The effect of bilingual instruction on the achievement of elementary pupils. Unpublished doctoral dissertation, East Texas State University, Commerce. Pena-Hughes, E., & Solis, J. (1980). ABC's: McAllen's Immersion System. McAllen, TX: McAllen Independent School District. Plante, A. J. (1976). A studyof effectivenessof the Connecticut "Pairing" model ofbilingual/biculturaleducation. Hamden,CT: ConnecticutStaff Development Cooperative. Powers, S. (1978). The influence of bilingual instruction on academic achievement and self-esteem of selected Mexican American junior high school students. Doctoral dissertation, University of Arizona, Tucson. Ramirez, J., Pasta, D. J., Yuen, S., Billings, D. K., & Ramey, D. R. (1991). Final report: Longitudinal study of structural immersion strategy, early-exit, and late-exit transitional bilingual education programs for language-minority children. San Mateo, CA: Aguirre International(Report to the U.S. Department of Education). Ramos, M., Aguilar, J. V., & Sibayan, B. F. (1967). The determination and implementation of language policy (Monograph Series 2). Quezon City, The Philippines: Philippine Center for Language Study. Reese, L., Gamier, H., Gallimore, R., & Goldenberg, C. (2000). Longitudinal analysis of the antecedents of emergent Spanish literacy and middle-school English reading achievement of Spanish-speaking students. American Educational Research Journal, 37(3), 633-662. Rosier, P., & Holm, W. (1980). The Rock Point Experience: A longitudinal study of a Navajo school program. Washington, DC: Center for Applied Linguistics. Rossell, C. H. (1990). The effectiveness of educational alternativesfor limited-Englishproficient children. In Imhoff, G. (Ed.), Learning in two languages. New Brunswick, NJ: Transaction Publishers. Rossell, C. H., & Baker, K. (1996). The educationaleffectiveness of bilingual education. Research in the Teaching of English, 30(1), 7-69. Rossell, C., & Ross, J. (1986). The social science evidence on bilingual education. Journal of Law and Education, 15, 385-419. Rothfarb, S. H., Ariza, M. J., & Urrutia, R. (1987). Evaluation of the Bilingual Curriculum Content (BCC) Project: A three-year study, final report. Miami, FL: Office of Educational Accountability, Dade County Public Schools. 279
Slavin & Cheung
andacademic Saldate,M., Mishra,S. P., & Medina,M. (1985). Bilingualinstruction 24-30. Skoczylas,R. V. (1972). An evaluationof some cognitiveandaffectiveaspectsof a Spanishbilingualeducationprogram.Doctoraldissertation,Universityof New Mexico,Albuquerque. and to meta-analytic Slavin, R. E. (1986). Best-evidencesynthesis:An alternative
traditionalreviews. Educational Researcher, 15(9), 5-11. Slavin, R. E., & Cheung, A. (in press). Effective early reading programs for English language learners: Language policy and early literacy education. Greenwich, CT: Slavin, R. E., & Cheung, A. (2005). Effective reading programs for English language achievement: A longitudinal study. Journal of Instructional Psychology, 12(1),
Information Age Publishing.
learners. forData-Driven Reform Center MD:JohnsHopkins Baltimore, University, in Education.
Compton CitySchools. Teschner,R.V. (1990). Adequatemotivationand bilingual education.Southwest

Journal of Instruction, 9(2), 1-42. Thomas, W. P., & Collier, V. P. (2002). A national study of school effectiveness for language minority students' long-term academic achievement. Santa Cruz, CA:
Stem, C. (1975). Final report of the Compton Unified School District's Title VIIbilingual-bicultural project: September 1969 through June 1975. Compton City, CA:
of California on Education, at SantaCruz,Center for Research Diversity, University andExcellence. UnitedStatesInternational Fresno,CA. University,
Valladolid, L. A. (1991). The effects of bilingual education on students' academic achievement as they progress through a bilingual program. Doctoral dissertation, Webb, J. A., Clerc, R. J., & Gavito, A. (1987). Comparison of bilingual and immersion programs using structural modeling. Houston, TX: Houston Independent School education. Review of Educational Research, 55(3), 269-317.
District. of selectedstudieson the effectivenessof bilingual Willig,A. (1985).A meta-analysis and research through meta-analysis Willig, A. (1987). Examining bilingualeducation 363-376. In M.C.WittL., & Valadez,C. (1986). Teachingbilinguallearners. Wong-Fillmore, Some achievement: P. (1988, April).LEPstudent Yap,K. O., Enoki,D. Y., & Ishitani, of the American New Orleans. Educational Research Association, H. W., & Suliman, R. (2000). Cantwo tongueslive in harmony? Yeung,A. E., Marsh, Analysisof the NationalEducation Studyof 1988 (NELS88)longituLongitudinal dinaldataon the maintenance Research American Educational of home language.
Journal, 37(4), 1001-1026. pertinent variables and policy implications. Paper presented at the annual meeting rock (Ed.), Handbook of research on teaching (3rd ed.). New York: Macmillan. narrative review: A response to Baker. Review of Educational Research, 57(3),
Authors ROBERT E. SLAVINis a Principal ResearchScientistat JohnsHopkinsUniversity and Directorof the Centerfor Data-Driven JohnsHopkins Reformin Education, University, 200 W. Towsontown Blvd., Baltimore, MD 21204; e-mail His researchinterestsinclude cooperativelearning,comprerslavin@cddre.org. hensive school reform,policy and practicefor at-riskstudents,Englishlanguage andresearch review. learners, 280
ALAN CHEUNGis a ResearchScientist at the Success for All Foundation,200 W. Towsontown MD 21204; e-mailacheung@successforall.net. Blvd., Baltimore, Startingin August 2005, Dr. Cheung will be an Associate Professor at the of Education, of Educational HongKongInstitute PolicyandAdminisDepartment tration,10 Lo Ping Road,Taipo,Hong Kong, China;e-mailacheung@cddre.org. His researchinterestsincludebilingualeducation,comprehensive school reform, andeffectivereading for privateeducation, programs Englishlanguagelearners.
281
APPENDIX Disposition of studies reviewed Cited by Authors
Remarks
Methodologicallyadequate:Elementaryreading RB RB RB RB and W RB and W RB RB RB RB Alvarez (1975) Bacon et al. (1982) Campeauet al. (1975) Cohen(1975) Doebler & Mardis(1980) Huzar(1973) J. A. Maldonado(1994) J. R. Maldonado(1977) Morgan(1971) Plante(1976) Ramirezet al. (1991) Saldateet al. (1985)
5 separatestudies met criteria
Methodologicallyadequate:Heritage languages RB Doebler & Mardis(1980) Morgan(1971) Choctaw in Mississippi Frenchin Louisiana
Methodologicallyadequate:Secondaryreading RB and W RB and W Covey (1973) Kaufman(1968) Canadianstudies of French immersion RB RB RB RB RB RB RB and W Barik& Swain (1975) Bariket al. (1977) Brucket al. (1977) Day & Shapson(1988) Genesee & Lambert(1983) Genesse et al. (1989) Lambert& Tucker(1972) Studentsnot learning the societal language RB Ramos et al. (1967) LearningEnglish in the Philippines
No readingoutcomes(oral language only) RB and W RB RB RB and W RB and W RB RB 282 Lum (1971) Bates & May (1970) Elizondo de Weffer (1972) (1979) Legarreta Pena-Hughes& Solis (1980) Ciriza(1990) Rothfarbet al. (1987) 6-monthstudy, no pretestdataprovided 4-month study, no readingoutcomes;also preferencefor English languageusage, EI > TBE Kindergarten LanguageAssessment Scales were outcome measure Preschoolonly
APPENDIX (Continued) Cited by Authors
Remarks
Minimaluse of native language instruction RB RB RB RB Becker et al. (1982) EducationalOperations Concepts(199la) EducationalOperations Concepts(1991b) Gersten(1985) Not an evaluationof bilingualprograms No indicationthatK-1 Cambodian,Hmong studentswere taughtin native languages No indicationthatCambodian,Hmong studentsin Grades7-9 were taughtin native languages no bilingual Study of Direct Instruction; comparisongroup
Pretests given after treatmentswere underway RB andW RB RB RB RB and W Danoff,Arias,& Coles (1977) Melendez (1980) Olesini (1971) Rosier & Holm (1980) Rossell (1990) Skoczylas (1972)
RB and W RB RB
Ster (1975) Thomas & Collier (2002) Separatestudies in Maine and Houston Valladolid (1991) Yap, Enoki, & Ishitani(1988) Redundant
Largepretestdifferences;no separate students; analysis for Spanish-dominant childrenin the more English-dominant controlgroup
RB RB RB RB RB and W RB RB RB
Ariza (1988) Barik& Swain (1978) Cohen et al. (1976) Curielet al. (1980) Danoff, Coles, McLaughlin, & Reynolds (1977, 1978) El Paso ISD (1987, 1990) Genesee, Lambert,& Tucker(1977) McConnell (1980a)
Redundantwith Rothfarb(1987) Redundant with Bariket al. (1977) with Cohen (1975) Redundant with Curiel (1979) Redundant Redundant with Danoff, Arias, & Coles (1977) Redundantwith El Paso ISD (1992) Redundant with Genesee et al. (1983) Redundant with McConnell (1980b)
No evidence of initial equality RB RB RB RB and W RB RB Ames & Bicks (1978) Barclay (1969) Bacon et al. (1982) Carsrud& Curtis (1979, 1980) Cottrell(1971) Curiel (1979) Largepretestdifference;mixed gradesand mixed languages Largepretestdifferences;7-monthstudy No measureof early academicability Mixed Spanish-and English-dominant childrenin the analysis Poorly matchedon SES; ANCOVA was used but no pretestdataprovided No measureof early academicability (continued) 283
APPENDIX (Continued) Cited by Authors No evidence of initial equality RB RB RB RB RB RB and W RB RB RB and W El Paso ISD (1992) Layden (1972) Malherbe(1946) Matthews(1979) Powers (1978) Stebbinset al. (1977) Teschner(1990) Vasquez (1990) Zirkel(1972)
Remarks
No measureof early academicability Largepretestdifference in both Spanish and English; 10-week study aboutinitialcomparability Lackedinformation aboutinitialcomparability Lackedinformation No measureof early academicability No measureof early academicability No measureof early academicability No measureof early academicability and Largepretestdifferencesin Hartford no bilingualinstructionin Bridgeport; New Britain,New London
No appropriatecomparisongroup RB RB RB RB RB RB RB Burkheimer et al. (1989) Carlisle& Beeman (2000) de la Garza& Marcella(1985) Lampman(1973) McConnell(1980b) Medina & Escamilla (1992) Moore & Parr(1978) Prewitt-Diaz(1979) Thomas & Collier (1997) Thomas & Collier (2002) Webb, Clert,& Gavito (1987) to expected Comparedactualperformance no real controlgroup performance, Bothgroupswerebilingual(80-20 vs. 20-80) to ComparedSpanish-dominant no pretestdata English-dominant; Mixed Spanish-and English-dominant childrenin the pretestanalysis Comparedto a baseline group;no measure of initial comparability ComparedVietnameseTBE to Hispanic MaintenanceBilingual;no reading outcomes Mixed Spanish-and English-dominant children;also late pretestsfor Grades 1 and 2 17-week study;large pretestdifference No controlgroups Separatestudies in Oregon and Florida lacked controlgroups Scores confoundedwith retentionstatus
Brief studies RB et al. Balasubramonian (1973) 4-month study Unavailable RB and W McSpadden (1979, 1980)
Note. RB = Rossell & Baker, 1996; W = Willig, 1985; TBE = TransitionalBilingual Education;EI = English Immersion;ISD = IndependentSchool District; TBE = transitional bilingualeducation. 284

(2005) A Synthesis of Research On Language of Reading - Slavin & Cheung

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

(2005) A Synthesis of Research On Language of Reading - Slavin & Cheung

Uploaded by

Copyright:

Available Formats

A Synthesis of Research on Language of Reading Instruction for English Language Learners Author(s): Robert E.

Reviewof EducationalResearch Summer 2005, Vol. 75, No. 2, pp. 247-284

A Synthesis of Research on Language of Reading Instruction for English Language Learners

Slavin & Cheung

Research on Languageof ReadingInstructionforELLs

Slavin & Cheung

Research on Languageof ReadingInstruction for ELLs

Research on Language of ReadingInstruction for ELLs

Slavin & Cheung

Well ma on Sp oral v but El Engli

Random assignment, 2 & 3 years

160 students, Grades 1-2, 1-3

Disadvantaged PuertoRican studentsin PerthAmboy, NJ

Well ma on IQ and in achiev

Campeau etal. (1975), Corpus Christi Ramirez etal. (1991)

Matched control, 2 years Matched control, 2 years

Spanish-dominant studentsin CorpusChristi, TX Spanish-dominant LEP students

Matched Engli Spani pretes

Very we match SES a backg

Campeau et al. (1975), Houston J. R. Maldonado (1977) Alvarez (1975)

Paired bilingual Paired bilingual

Matched control, 3 years Matched control, 5 years

206 students, Grades K-2 126 students, Grades 1-5

Matche langu and a achie Matched and n years schoo

Matched control, 2 years

147 students, Grade2

Matched and in langu profic

Matched control, 2-3 years Matched control, 1 year

90 students, Grades K-I, 1-2, 1-3 89 students, GradeK

Spanish-dominant studentsin Redwood City, CA Spanish-dominant studentsin Kingsville, TX

Matched and in langu profic

Campeau et al. (1975), Kingsville, TX

Similaro prete PBE >

Bilingual, 1-year transition

Matched control, 2 years

125 students, Grades K-l

Well ma initia abilit

Doebler & Mardis (1980)

Matched control, 1 year

Well ma initial profic

Slavin & Cheung

Research on Languageof ReadingInstructionforELLs

Slavin & Cheung

Research on Languageof ReadingInstruction for ELLs

Slavin & Cheung

Research on Languageof ReadingInstruction for ELLs

Research on Language of ReadingInstructionforELLs

Slavin & Cheung

Research on Language of ReadingInstruction for ELLs

Slavin & Cheung

Research on Language of ReadingInstruction for ELLs

Placedat Risk. Center for Research on the Education of Students University,

of selectedstudiesin the effectivenessof bilingual Baker,K. (1987). A meta-analysis

Research on Languageof ReadingInstruction for ELLs

Cohen, A. D. (1975). A sociolinguistic approach to bilingual education. Rowley, MA:

3. YearTwoimpactdesigns.Palo Alto, CA:American for Research. Institutes

annualmeetingof the American Research New Orleans. Educational Association,

El Paso IndependentSchool District. (1990). Bilingual education evaluation: The sixth

Slavin & Cheung

Information Age Publishing.

learners. forData-Driven Reform Center MD:JohnsHopkins Baltimore, University, in Education.

Compton CitySchools. Teschner,R.V. (1990). Adequatemotivationand bilingual education.Southwest