VoL 74. No. 2, pp. 194-214.&2008 Council for Exceptional Children.

Exceptional Children

Peer-Assisted Learning Strategies:A "Tier 1 "Approach toPromoting English Learners'Response to Lntervention

KRISTEN L. MCMASTER

SHU-HSUAN KUNG

INSOON HAN

MARISA CAOUniversity uf Minnesota

ABSTRACT:r; This study determined the effectiveness of Kindergarten Peer-Assisted Learning Strategies

(K^PALS)for English Learners (ELs). We compared 20 K-PALS ELs to 20 Control ELs and to 20

K-PALS non-ELs on early reading skill acquisition, using a pretest—posttest control group design

with matched samples. We also compared proportions of ELs unresponsive to K-PALS to ELs unre-

sponsive to traditional instruction. Teachers implemented K-PALS 4 times per week for 18 weeks.

Following intervention, analyses of covariance on posttest measures indicated that K—PALS ELs per-

formed reliably higher than Control EU on measures of phonemic awareness and letter sound recog-

nition, and they performed similarly to K-PALS non-ELs. Findings are discussed in terms of

K—PALS efficacy for ELs in a response-to-intervention framework.

any Engiisti Learners (ELs).struggle CO learn to read,and, as a group, lag far be-hind their native English-speaking peers. According

to the National Center for Education Statistics,in 2005, 27% of fourth-grade ELs performed ator above the basic reading level, with only 7% ator above the proficient level, whereas 67% ofnon-EL.s performed at or above the basic levelwith 34% at or above the proficient level. Severedifficulties in reading can lead to special educa-tion referral (Snow, Burns, & Griffin, 1998); in-

deed, as many as 80% of students with learningdisabilities (LD) struggle primarily with reading(Lyon et al., 2001). How this relates to ELs iscurrently unclear. Disproportionate numbers ofELs are referred to special education, with manyeither over- or under identified as having LD. Dis-proportionate referral rates have been attributed,at least in part, to difficulties in disentanglingpossible learning deficits from limited languageproficiency (McCardle, Mele-McCarthy, Cutting,Leos, & D'Emilio, 2005).

Addressing both the literacy development ofELs and issues surrounding disproportionate

1 9 4 Winter 2008

referral rates of ELs to special education are cur-rent educational priorities. In fact, both the studyof reading interventions and LD identification is-sues were placed high on the agenda for EL-LDresearch by a symposium organized by the U.S.Department of Education and National Institutesof Health in Oc:tober 2003 (U.S. Department ofEducation & U.S. Department of Health andHuman Services). There is an especially urgentcall for effective early intervention approaches tar-geting ELs (e.g., Getsten & Baker, 2000; Slavin& Cheung, 2005), with the view that inappropri-ate special education referral, school failure, andother negative academic outcomes could be pre-vented if strong, evidence-based practices are inplace early on.

G E N E R A L A N D S P E C I A L

E D U C A T I O T ' J R E F O R M S

Contributing to the urgency to ensure that effec-tive early intervention is in place for ELs are cur-rent reforms that place heavy emphasis onevidence-based practice and accountability. Provi-sions of the No Child Left Behind Act (NCLB,2002) stress that schools must work to closeachievement gaps, improving outcomes for alllearners. To achieve this, schools must show thatsubgroups of students, including EI^, are making"adequate yearly progres.s" as measured by state-defined academic achievement tests. Schools thatdo not meet accountability standards may facetough sanctions (Abedi, 2004).

Special education policy is increasingly align-ing with these gtneral education reforms. The re-cent reauthorization of the Individuals WithDisabilities Education Improvement Act (IDEA,2004) emphasize;; early intervention and account-ability, and indicates that disproportionate num-bers of culturally and linguistically diverse studentsreferred to special education must be reduced.Further, IDEA allows local education agencies touse response-to-intervention (RTI) models inplace of discrepancy models for identifying stu-dents with LD. Rl'I involves early identification ofstudents at risk, progress monitoring, andincreasingly intensive intervention for studentswho continue to struggle. Only those studentswho do not make adequate progress despite inter-vention continue on to special education referral.

R E S P O N S E T O I N T E R V E N T I O N :

I M P L I C A T I O N S F O R E L S

An RTI approach is proposed by some (e.g., Ger-ber et al., 2004; Klingner & Edwards, 2006;Linan-Thompson, Vaughn, Prater, & Cirino,2006) as a possible alternative for ELs at risk ofspecial education identification to traditional dis-crepancy models that rely on standardizedachievement and intelligence tests. This is, inpart, because few standardized tests have technicaldata supporting their reliability and validity forELs, and provide little information as to whattypes of instruction will be most beneficial once adisability is diagnosed (Abedi, 2004). Moreover, itoften takes several years for a discrepancy toemerge (Wagner, Francis, & Morris, 2005), whichcan result in persistent academic failure that be-comes increasingly difficult to remediate. RTImay be a less biased, more instructionally relevantapproach for ELs. Moreover, RTI proponents ex-pect that this approach will lead to earlier identifi-cation and intervention, and, in the long run,help decrease proportions of ELs who are inap-propriately referred to special education (Klingner& Edwards).

Special education policy is increasinglyaligning with these general educationreforms.

Yet, some of the same problems associatedwith traditional identification models may alsoemerge from an RTI approach (see Wagner et al.,2005). Eor example, many ELs may performbelow RTI criteria established for non-ELs, lead-ing to allocation of resources for more intensiveinstruction for (again) disproportionate numbersof ELs. Although some may argue that this is nota problem if intensive instruction is beneficial forELs, an important question is whether generalclassroom instruction can be designed to meet theneeds of many ELs, such that more intensive in-terventions are necessary for only the mostseverely struggling students.

In this study, we examined the effectivenessof a classroom-based beginning reading approachto boosting the early reading performance of ELs.First, we provide an overview of the RTI ap-

Exceptiondl Children 19S

proach and review research on early reading inter-vention for ELs. Then, we propose KindergartenPeer-Assisted Learning Strategies (K-PALS; D.Fuchs, Fuchs, Thompson, Al Otaiba, Yen, Yang,et al., 2001) as an evidence-based instructionalapproach that may be beneficial for many ELs be-ginning to learn to read.

E A R L Y R E A D I N G I N T E R V E N T I O N S

F O R E L S : H O W DO T H E Y F I T

W I T H I N A N R T I A P P R O A C H ?

A TIERED MODEL OF INTERVENTION

Two general models of RTI have been describedin the literature to date (see L. S. Fuchs, 2003).One model conceptualizes RTI as response to in-tensive, preventative intervention. In this model,students identified as at risk immediately receivespecialized intervention provided in small groupsby a specialist (e.g., Torgesen et al., 2001; Vel-lutino et al., 1996). Those who continue to per-form at low levels or make very little growth aredeemed unresponsive to intervention and are can-didates for special education. The second model isrooted in general education (L. S. Fuchs, 2003;Speece & Case, 2001) in that high-quality generalclassroom instruction is provided to at-risk stu-dents before the decision is made to implementmore intensive intervention. We focus on this sec-ond model, with the assumption that implement-ing high-quality, evidence-based classroominstruction will meet the needs of many students,thus reducing proportions of students who willrequire more intensive, individualized instruction.

This second RTI model typically includes in-creasingly intensive "tiers" of intervention (e.g.,D. Fuchs & Fuchs, 2006, Vaughn & Linan-Thompson, 2003). Tier I consists of generalclassroom practices that at least refiect sound ped-agogy, and at best consist of evidence-based in-struction implemented with integrity. 1 ier 2 is (a)provided to students for whom Tier 1 is not bene-ficial; (b) is more intensive in that it is typicallyprovided in small groups; (c) includes explicit,skills-based instruction; (d) is conducted morefrequendy or for longer periods; and/or (e) is de-livered by an expert such as a reading specialist.Subsequent tiers are implemented when Tier 2does not result in sufficient progress, are even

more intensive, and may either lead to special ed-ucation referral or are provided within special ed-ucation (D. Fuchs, Mock, Morgan, & Young,2003; Kiingner & Edwards, 2006).

If the early literacy needs of at-risk ELs are tobe addressed within an RTI framework, it is criti-cal that evidence-based instruction is available forELs at each tier (Klingner & Edwards, 2006).Otherwise, it will be difficult to determinewhether ELs who struggle to learn to read do sobecause appropriate instruction is not in place (L.S. Fuchs, 2003; Vaughn, Mathes, Linan-Thomp-son, & Francis, 2005) or because of language orlearning deficits (McCardle et al., 2005). Whereasresearchers have provided evidence of the generaleffectiveness of a number of early reading inter-ventions (see National Reading Panel, 2000), it isinappropriate to assume that these interventionsare effective for all students (Klingner & Ed-wards). Rather, it is critical to examine what typesoF interventions are effective for ELs at risk ofreading failure. Several researchers have begunsuch examinations. Following, we briefiy reviewstudies targeting the early literacy development ofELs, characterizing the instruction provided asTier 1 or Tier 2.

Tier 1 Instruction. As mentioned, in an RTIframework. Tier 1 involves evidence-based readinginstruction provided within general education.Few have investigated outcomes for ELs partici-pating in whole-class reading instruction, but ex-isting evidence is promising. In a large-scale study,D'Angiulli, Siegel, and Maggi (2004) investigatedwhether a districtwide literacy curriculum had adifferential effect on the reading achievement ofELs and non-ELs of varying socioeconomic status(SES). Kindergarten through fifth-grade studentsreceived instruction in phonemic awareness, al-phabetic knowledge, and shared and independentreading and writing. At the lowest and highestends oi the SES distribution, significantly moreELs than non-ELs were proficient or improving toproficiency in word-level reading after participat-ing in the curriculum. In the middle of the SESdistribution, ELs and non-ELs achieved at similarrates. The authors attributed this to the curricu-lum; without it, the trajectories for ELs with thelowest and highest SES were expected to be similarto their non-EL counterparts. Flowever, there was

Winter 2008

no comparison group, which limits conclusionsabout the efficac/ of this curriculum.

Tier 2 Interventions. Many researchers whohave conducted early reading studies with ELshave examined instructional approaches resem-bling Tier 2 interventions (de la Colina, Parker,Hasbrouck, & Lara-Alecio, 2001; Denton, An-thony, Parker, & Hasbrouck, 2004; Gerber et al.,2004; Gunn, Biglan, Smolkowski, bL Ary, 2000;Haager & Windmueller, 2001; Leafstedt,Richards, & Gerber, 2004; Linan-Thompson,Vaughn, Hickm:ui-Davis, & Kouzekanani, 2003;Linan-Thompson et al., 2006; Vaughn, Cirino, etal., 2006; Vaughn, Linan-Thompson, Mathes,Cirino, et al., 2006; Vaughn, Mathes, et al.,2006). These interventions have been supplemen-tary to general classroom instruction, have beendelivered in smill groups by instructors otherthan general classroom teachers, have been imple-mented for 30 to 50 min daily, and have rangedfrom 10 weeks to 8 months. Further, the inter-ventions have incorporated explicit instructiontargeting phonemic awareness, letter-sound cor-respondence, decoding, fluency, and comprehen-sion, all critical beginning reading skills known tobe important for native English speakers whostruggle in reading (e.g.. National Reading Panel,2000). Vaughn and colleagues have also incorpo-rated explicit oral language and vocabulary activi-ties (see Vaughn ct al., 2005).

Across these studies, ELs made reliable gainsin phonemic awareness (Cerber et al., 2004;Linan-Thompson et al., 2003; Vaughn, Cirino, etal., 2006); letter-sound recognition (Vaughn,Cirino, et al., 2006); word attack (Gunn et al.,2000; Vaughn, Cirino, ct al., 2006); word reading(Denton et al., 2004; Leafstedt et al., 2004;Vaughn, Cirino, i;t al., 2006); fluency (de la Col-ina et al., 2001); spelling (Vaughn, Cirino, et al.,2006); and comprehension (de la Colina et al.).ELs reliably outperformed controls in all but twoof the studies, w hich did not include controls(Haager & Windmueller, 2001; Linan-Thompsonet al., 2003). Linm-Thompson et al. (2006) alsodemonstrated thai more ELs who received supple-mental reading intervention in first grade met RTIcriteria than controls and maintained this statusthrough the end of second grade.

Findings of the previous studies suggest thatexplicit, skills-based instruction known to be ef-

fective for many at-risk native English speakersare also beneficial for many at-risk ELs. Thesestudies involved supplementary, small group in-struction resembling Tier 2 intervention. Furtherresearch is needed to determine what types ofgeneral classroom. Tier 1 instruction are effective,reducing the need for more intensive interventionfor ELs. Also, few researchers have directly testedwhether beginning reading interventions are as ef-fective for ELs as for their native English-speakingpeers. It is important to better understandwhether classroom-based reading instruction has adifferential impact on different groups of stu-dents.

K I N D E R G A R T E N P E E R - A S S I S T E D

L E A R N I N G S T R A T E G I E S

One classwide instructional approach that hasbenefited many beginning readers is KindergartenPeer-Assisted Learning Strategies (K-PALS), asupplemental peer-tutoring program developedby researchers at Vanderbilt University (D. Fuchset al., 2001). In K-PALS, higher performingreaders are paired with lower performing readersto practice skills identified as critical for begin-ning reading (e.g.. National Reading Panel,2000), including phonemic awareness,letter-sound recognition, decoding, and Ruency.Results of large-scale experimental studies indicatethat K-PALS has improved beginning readingskills of many low-, average-, and high-perform-ing readers, as well as some children with disabili-ties (e.g., D. Fuchs et al., 2001; D. Fuchs et al.,2002). K-PALS has also been shown to be effec-tive for many students in schools with high pro-portions of minority children and children livingin poverty, as well as in schools with predomi-nantly White, middle-class populations. Teachershave reported that K-PALS is practical, efficient,and fits well with a variety of teaching philoso-phies (McMaster, Fuchs, & Fuchs, 2002).

For the previous reasons, K-PALS appearspromising as a Tier 1 instructional approachwithin an RTI framework; however, as men-tioned, it is important to understand the impactof such approaches for ELs. We believe thatK-PALS may provide effective beginning readinginstruction for ELs for two important reasons.

Exceptional Children

First, K-PALS includes explicit phonemic aware-ness, letter—sound, and decoding instruction,which have been found to be important not onlyfor native English-speaking students but also formany ELs at risk for reading failure (e.g., de laColina et aL, 2001; Denton et al., 2004; Gerberet al., 2004; Gunn et a!., 2000; Vaughn, Cirino,et al., 2006; Vaughn, Linan-Thompson, Pollard-Durodola, Mathes, & Hagan, 2006; Vaughn,Mathes, etal., 2006).

Second, K-PALS includes components rec-ommended for ELs, such as interactive teachingand high levels of student engagement, frequentopportunities for accurate responses, and peer-mediated learning (Gersten & Geva, 2003;Vaughn, Linan-Thompson, et al., 2006). Suchrecommendations align with perspectives of sociallearning theorists, who assume that social interac-tion improves a child's ability to gain and under-stand knowledge (e.g., Vygotsky, 1987), andsecond language-learning theorists, who assumethat ELs develop academic language proficiencyby engaging in tasks that require them to receiveand produce language to learn new concepts (e.g..Cummins, 1980; Long & Porter, 1985). K-PALSprovides a logical context and multiple opportu-nities for such interactions.

The purpose of this study was to determinethe effectiveness of K-PALS (D. Fuchs et al.,2001) implemented with ELs. Two primary re-search questions were addressed: (1) Do ELs whoparticipate in K-PALS outperform control ELs oncritical beginning reading skills? (2) Is K—PALS aseffective for ELs as for non-ELs? A secondary re-search question was whether K-PALS is effectiveas a Tier 1 intervention for ELs, as evidenced bylower proportions of unresponsive students toK-PALS than to traditional classroom instruction.

METHOD

SETTING AND PARTICIPANTS

This study was part of a large-scale randomizedfield trial investigating the "scaling up" of evi-dence-based practice in reading. The large-scalestudy included approximately 1,800 kindergart-ners attending 46 public elementary schools inTennessee, Texas, and Minnesota. Within eachstate, kindergarten classrooms were assigned ran-

domly within school to one of four groups withvarying levels of support: Level 1 (no-PALS Con-trol), Level 2 (1-day workshop). Level 3 (1-dayworkshop plus booster sessions), or Level 4 (1-dayworkshop plus booster sessions plus weekly men-tor visit). The primary purpose of the larger studywas to examine the extent to which these levels ofsupport improved teachers' implementation ofK-PALS and students' beginning reading achieve-ment. The present study differed from the largerstudy in that we focused on the effectiveness ofK-PALS for a subgroup of study participants,specifically ELs in Minnesota; we examined theeffectiveness of K-PALS as a Tier 1 interventionfor ELs; and we did not include an analysis of lev-els of support for this subgroup of students.

The present study was conducted in Min-nesota, which included 8 Control, 12 Level 2, 12Level 3, and 10 Level 4 classrooms in the largerstudy. Five hundred nine kindergartners werestudy participants m these classrooms. Of these,22% were ELs (« = 112), which was slightly lowerthan the overall proportion of EI^ in the partici-pating districts (approximately 30%). The presentstudy included 60 kindergartners selected fromthe 509 Minnesota participants.

Preliminary findings from the large-scalestudy indicated that students who participated inK-PALS outperformed Controls on several mea-sures of beginning reading. Further, accuracy ofK-PALS implementation, or fidelity, was foundto impact student achievement. Level of supportdid not have a consistent impact on studentachievement. Thus, for the purposes of the pre-sent study, we selected students whose teachershad implemented K-PALS with high fidelity(90% or greater) to ensure that participants hadindeed received K-PALS as it was intended to beimplemented (see Procedures for details of how fi-delity was assessed). Fifteen K—PALS teachers metthis criterion. We included all Control classrooms{n = 8) to maximize the number of ELs we couldinclude in the study. Fifty-three percent ofK-PALS teachers and 25% of Control teacherstaught in Title I schools. Fifty-three percent ofK-PALS teachers and 50% of Control teacherstaught in schools meeting adequate yearlyprogress (AYP) standards. None of the schools inthe present or larger study were Reading Firstschools.

Winter 2008

Although we selected teachers with high fi-delity to c-nsurr that participants truly receivedK~PALS instruction, there was the possibility thatthis selection criterion would yield a biased sam-ple. In other words, high-fidelity K-PALS teach-ers might have been higher performing teachers ingeneral. To address this potential problem, wefirst determinal that there were no statisticallysignificant differences between K-PALS and Con-trol teachers on key demographic variables (seeFable 1). Second, we used a Classroom Atmo-sphere Rating Scale (CARS; Wehby, Dodge, &Greenberg, 1993) to assess the quality of the in-structional environment in all classrooms in thelarger study (see Procedures). The correlation be-tween K-PALS fidelity and CARS ratings in thelarger study was r ^ -.39, suggesting that K-PALSfidelity was not strongly related to general in-structional quality. In addition, a comparison ofCARS ratings showed no reliable differences be-tween K-PALS and Control teachers in the pre-sent study (K-PALS, M = 11.08, SD = 3.66;Control,A/=9.l4, SD=2.4l,f[l , 18] =-L23,; '= .23); note that a lower score was associated withmore positive classroom ratings). Based on the.seanalyses, we concluded that the K-PALS teachersselected for thh study were not different fromControls in terms of overall experience or instruc-tional quality.

After identifying classrooms, we selected ELand non-EL participants. In Minnesota, ELs aredefined as students who (a) first learned a lan-guage other than English, usually speak a lan-guage other than English at home, or do not useEnglish as a primary language; and (b) lack theEnglish skills needed to fully participate in classestaught in English (Minnesota Department of Ed-ucation, 2004). Part (a) of the definition is deter-mined by a home language questionnaire given toall students when they first enter a district, andPart (b) is measured by teacher judgment and/orperformance on one of several measures such asthe Student Oral Language Observation Matrix(Minnesota Department of Education, 2003).El^ in our study were included in general educa-tion classrooms with native English-speakingpeers and received pullout EL services. The pri-mary language of instruction was English.

Selection of participants involved a matchingprocedure. In the Control classrooms, 20 ELs had

parental consent to participate in the larger study.We matched 20 K-PALS ELs to the 20 ControlELs based on two pretest measures: Rapid LetterNaming (RLN) and Rapid Letter Sound identifi-cation (RLS). K-PALS and Control ELs wererank-ordered based on these measures. The 20K-PALS ELs whose scores most closely matchedthe 20 Control ELs' scores were selected. Then,the 20 K-PALS ELs were matched to 20 K-PALSnon-ELs using the same procedure. A power anal-ysis with our sample size (A = 60 with 20 stu-dents per cell) indicated adequate power (|3 =.93)to detect a moderate effect size of */= 0.50 at a ^.05.

Table 2 presents student demographic infor-mation, which was collected as part of the largerstudy. Chi square analyses revealed no significantdifferences between K-PALS and Control ELs, orbetween K-PALS ELs and non-ELs, on sex, spe-cial education status, or free or reduced pricelunch. T tests revealed no reliable group differ-ences in terms of number of absences, and no reli-able differences between K-PALS and ControlELs' English proficiency levels. Reliable differ-ences were found between groups on age, suchthat K-PALS ELs were slightly older than Con-trols (5.78 vs. 5.51 years old). Reliable differenceswere also found between EL and non-EL groupson race and how often students received addi-tional reading and language instruction outsidethe classroom. As one might expect, there weremore Hispanic and Asian/Indian ELs than non-ELs. ELs received more reading and languagehelp outside the classroom than did non-ELs.Table 2 includes a breakdown of types of helpreceived.

MEASURES

In the larger investigation, all participants wereadministered a battery of beginning reading mea-sures prior to and immediately following K-PALSimplementation (in October and November andagain in April and May of the study year). Themeasures included three groups of tests: phone-mic awareness, which measured knowledge ofsounds in words; alphabetic, which measured let-ter naming, letter-sound recognition, word read-ing, and spelling skills; and oral reading, whichmeasured reading of connected text. All were

Exceptional Children 1 9 9

TABLE 1

Teacher Demographia

Variable

Sex

Male

Age in years20-2930-3940-4950-39

Race

African AmericanCaucasianHispanicAsian/Pacific Islander

Educational DegreeBS/BA and BS/BA+30MEd/MSandMEd/MS+15MEi:{/MSi-30

Teaching CertificationElementaryElementary and Reading

Teach in a Tide I school

Teach in a school meeting AYP

Years of teaching

Years in current position

Reading in.struction credit hours

Special education credit hours

K-PALS (n = 15)

n

0

554I

01500

654

132

8

8

M

10.63

6.47

3.13

2.00

(%)

(0.0)

(33.3)(33.3)(26.7)

(6.7)

(0.0)(100.0)

(0.0)(0.0)

(40.0)(33.4)(26.7)

(86.7)(13.3)

(53.3)

(53.3)

(SD)

(6-75)

(5.12)

(1.13)

(1.51)

Control (n

n

1

21

41

0

701

42

2

80

2

4

M

11.88

7.88

2.75

2.25

= 8)

(%)

(12.5)

(25.0)(12.5)(50.0)(12.5)

(0.0)(87.5)

(0.0)(12.5)

(50.0)(25.0)(25.0)

(100.0)(0.0)

(25.0)

(50.0)

(SD)

(6.56)

(5.46)

(1.39)

(1.75)

x'1.96

2.01

1.96

4.36

1.17

1.70

0

I

-.42

-.61

.72

-.39

P

.16

.57

.16

.50

.28

.19

1.0

P

.68

.55

.48

.72

Note. K-PALS = Kindergarten Peer-Assisted Ixarning Strategies; BS = Bachelor of Science; BA = Bachelor of Arts;MEd = Master's of Education; MS = Masters of Science; AYP = Adequate Yearly Progress.

given as pre- and posttest measures except forspelling and oral reading, which were given atposttest only.

Phonemic Awareness Measures. A segmenta-tion test hased on the Yopp-Singer test (Yopp,1988) and developed for use in previous PALS re-search (D. Fuchs et al., 2001) was administered.In this test, the examiner says a word andprompts the student to say the sounds in theword. One point is recorded for each phonemesaid correctly. The score is recorded as the numberof correct phonemes expressed in 1 min. A hlend-ing task, also used by Fuchs et al., assessed stu-

dents' ability to blend phonemes into words. For

example, it the examiner says "s-oa-p," the sru-

dent blends the phonemes into "soap." The score

was recorded as the number of words blended

correctly in 1 min.

Alphabetic Measures. A test for RLN was given

to assess rapid letter identification skill. This mea-

sure was developed for use in a previous PALS

study (D. Fuchs et al., 2001) and consisted of

upper and lower case letters presented randomly in

black type on a single sheet of paper. Students

were instructed to name the letters as quickly as

zoo Winter 2008

TABLE 2

Student Demogfaphii's

Variable

SexMale

Special EducationNo IEPHas an (EPIn process

RaceAfrican AmeritanCaucasianHispanicAsian/IndianOther

Receive free or reduced-price lunch

Reading/languagf help'No assistanceReading specialistResource, speechTitle IESL pullout

Age in years

Absences

English proficiemy

K'PALSEU

rn =n

10

200

0

40

871

12

850

16

M

5.78

2.56

3.05

20)

(%)

(50)

(100)(0)(0)

(20)(0)

(40)(35)

(5)

(60)

(40)(25)

(0)(5)

(30)

(SD)

(0.31)

(3.78)

(0.69)

Control ELs(n =

n

12

191

0

1

12

16

0

11

752

15

M

5.51

1.40

3.45

20)

1 yO/

(60)

(95)(5)(0)

(5)(5)

(10)(80)(0)

(55)

(35)(25)(10)

(5)(25)

(SD)

(0.29)

(2.16)

(0.83)

K-PALS non-EU(n =

n

12

18

1

1

11

61

11

12

180

11

0

M

5.79

2.53

20)

(%)

(60)

(90)(5)(5)

(55)(30)

(5)(5)(5)

(60)

(90)(0)(5)(5)(0)

(SD)

(0.35)

(3.44)

lvs.2'

0.40

1.03

10.92*

1.30

2.16

t (

2.84*'

1.04

2.78

lvs.3^

0,40

2.11

19.21"*

0.00

15.85"

38)

' -0.07

0.02

NA

Note. K-PALS = Kindergarten Peer-Assisted Learning Strategies; EL = English learner; IEP = individualized educa-tion plan. ESL - English as a Second language.in the .schools in which this study was conducted, supplementary reading and language help was provided to stu-dents in an English immersion hirmat either in the general classroom or as pullout service. "Group comparison be-tween K-PALS ELs versus Control ELs. 'Croup comparison between K-PALS ELs versus K-PALS non-ELs.*;><.O5. " / x . O l .

they could. The score was recorded as the number was recorded as the number of sounds produced

of letters named correctly in 1 min. correctly in I min.

A test for RLS was given to assess The Word Identification (Word ID) and

letter-sound identification skill. RLS is based on "^"'^ '^"^^'^ ^" ' ^^^^ "^ ' ^ Woodcock Reading, , , , , , , ,,,,,._, Mastery Test-Revised (WRMT-R; Woodcock,

a measure used by Levy and Lysunchuk (lvV7) .1987) were given to measure word recognition

and was developed tor use in a previous PALS . . i- i .n r- • w, , ,T-. I' ' and decoding skills. J>cores on the Word IU and

study (D. Fuchs et al., 2001). All 26 letters of the ^ ^ ^ j ^^^^^^ ^^^^^^^^ . ^^ , ^^ l jgl j ^-^^ ^^^^^

alphabet were presented randomly in black type j^sts of reading, including the Iowa Test of Basic

on a sheet of paper. Students were instructed to Skills Total Reading, the Woodcock-Johnson Test

say the sounds as quickly as they could. The score of Reading Achievement, and the reading subtest

Exceptional ChiUiren SOI

of the Wide Range Achievement Test, with corre-lations of .83 to .92. Internal consistency exceeds.90. The scores for these measures were the num-ber of words read correctly.

The Wechsler Individuai Achievement Test(WIAT; Psychological Corporation. 1992)spelling subtest was administered at posttest only.Students were instructed to write letters andwords on a sheet of paper. The score was thenumber of letters and words written correctly.The WIAT correlates well with other achievementtests (rs = .70s to .80s), and has a test-retest relia-bility coefficient of .94.

Oral Reading Measures. Oral reading mea-sures were administered at posttest only. Studentsread from two passages (Oral Reading A and B)that incorporated sight words and decodablewords introduced in K-PALS lessons. Both sto-ries had a Flesch-Kincaid readability grade level of0.0. Before reading each story, the student was in-structed to read aloud and to try his or her best.As the student read, the examiner marked inser-tions, omissions, substitutions, and mispronunci-ations not caused by speech-related problems ordialects. Omissions and additions of endings wereconsidered errors. Self-corrections were notcounted as errors. Errors were not corrected andno decoding assistance was given. The score wasthe number of words read correctly in 1 min. Al-ternate-form reliability for the larger sample was.95 {p< .001).

READING INSTRUCTION

Core Instruction. All teachers in the largerstudy were asked to report the types of activitiesand materials that comprised their core languagearts instruction. Language arts instrtiction wasgenerally implemented between 90 and 120 minper day. Teachers consistently reported imple-menting "balanced" literacy instruction that in-corporated letter and letter sound recognition andpronunciation; phonemic awareness; "building"words using onsets and rimes (word families);shared and guided reading; whole-class writingand journal writing; comprehension activitiessuch as making predictions, identifying mainideas, and sequencing; and vocabulary develop-ment. Most reading instruction was implementedin a whole-class format followed by individual

seatwork. Teachers generally did not use a singlecurriculum but reported using guided readingbooks, trade books, decodable and pattern booksdownloaded from the Web, teacher-made mateti-als, and letter charts.

K-PALS Instruction. As mentioned, K-PALSwas developed by researchers at Vanderbilt Uni-versity (D. Fuchs et al.. 2001). and was designedto be a supplement to beginning reading instruc-tion in the general education classroom. As partof the larger study, teachers implementedK-PALS for 18 weeks, four times per week, for20 to 30 min each session. In K-PALS, theteacher pairs higher performing readers withlower performing readers. The higher performerin each pair acts as the Coach (tutor) first, andthe lower performer acts as the Reader. The rolesare reciprocal such that during each activity stu-dents take turns being Coach and Reader. Pairschange approximately every 4 weeks. We did notdictate with whom ELs should be paired (e.g.,partners with the same language versus Englishproficient partners); thi.s was left up to teachers.

Teachers prepare their students for K-PALSby modeling the activities during eight traininglessons presented in a whole-class format. Theteacher acts as the Coach and the students are theReaders. Cradually, students take turns assumingthe role of Coach for the whole class. Then, thestudents tutor each other, alternating as Coachand Reader. The teacher circulates among the stu-dent pairs, monitoring their progress and provid-ing corrective feedback.

Two types of activities are included inK-PALS: Sound Play and Sounds and Words.Sound Play is a teacher-directed activity and in-cludes five phonemic awareness games that ad-dress first sounds, rhyming, blending andsegmenting, and ending sounds. Each lesson sheetshows pictures of common animals and objects.For example, the rhyming game shows rows offour pictures (see Figure 1). The first two picturesrhyme. The teacher points to the first two pic-tures and says, "Pan, fan;" then points to theother two pictures and says, "Which one rhymeswith pan and fan, strawberry or ran?" The stu-denrs reply, "Ran." One type of game is presentedfor several lessons, and then another type of gameis introduced. For example, students play the FirstSound Game for several sessions, using a different

Winter 2008

F I G U R E 1

Sample Teacher-Directed Sound Play and Peer-Mediated Decoding Kindergarten Peer-Assisted LeamingStrategies (K-PALS) Lesson

Lesson 42

Lesson 42

k - ^

1

1k

ond

I

u s

i d

e z

P e

is

was

w a

k •

* i

0 J

©

on

a

®

©

@

•

•©

m

k

u

©

was

me

© ©

h 1 m J 1 m T i m

5 a d m

Jim was sad.

a d

©

p a

©

d

©

WhalsouriO'?

Wfiot wofd"?

[?eod i1 slowly

Sing it and-eoaii

hxceptinnal Children 2 O 3

set of pictures each time. Then, students play therhyming game for several sessions, and so on.Sound Play games become increasingly difficult asstudents progress through the K-PALS lessons.

Two types of activities are included inK~PALS: Sound Play and Sounds andWords.

Sounds and Words is made up of four activi-ties that are previewed by the teacher and thenconducted by student pairs. The activities are allprinted on one side of a lesson sheet (see FigureI). The first activity, "What Sound?" displaysrows of letters that students read from left toright. The Coach points to each letter and asks,"What sound?" The Reader says each sound. Starsare interspersed among the letters to prompt theCoach to praise the Reader (e.g., "Great job!").When the Reader makes an error, the Coach says,"Stop, that sound is _. What sound?" The Readersays the sound, and the Coach says, "Good. Readthe line again." At the end of the activity, theCoach marks one of four happy faces printedunder the "What Sound?" section. Then, the stu-dents switch roles and repeat the activity.

After doing "What Sound?" twice, the pairmoves on to "What Word?" which involves read-ing common sight words printed in rows on thelesson sheet. The Coach points to each word andasks, "What word?" The Reader reads the words,and the Coach corrects errors just as in the "WhatSound?" activity. After the Reader reads, theCoach marks a happy face, and the studentsswitch roles and repeat the activity. They thenmove on to "Sound Boxes," in which they readdecodable words comprised of sounds practicedin the lesson. The words are presented in wordfamilies, such as "at," "mat," and "sat." Each letterof a word is in a "sound box." The Coach says,"Read it slowly," and the Reader sounds out theword, pointing to the letter in each box. Then theCoach says, "Sing it and read it." This promptsthe Reader to blend the sounds together, and thenread the word. The Coach corrects errors as in theprevious activities. After the Reader reads, theCoach marks a happy face, and the studentsswitch roles and repeat the activity.

Finally, the students read sentences com-prised of sight words and decodable words prac-ticed in the lesson. The Coach says, "Read thesentence," and corrects any errors as the Readerreads. After reading the .sentence, the students re-peat the Sounds and Words activities, markingmore happy faces each time they complete a sec-tion, until the timer rings (after about 15 min).At the end of the lesson, the students count thehappy faces they have marked, and record thisnumber on point sheets.

Control Instruction. Control teachers con-ducted their regular reading instruction. A re-search assistant observed all Control teachers onceduring a .?O-min structured reading lesson thatthe teacher identified as typical daily instruction.The observed instruction was provided in eitbersmall group or whole-class formats. All teachersimplemented explicit instruction in letter sounds,phonemic awareness skills, letter writing, andword identification. Several Control teachers alsowere observed to incorporate storybook readingand journal writing activities to enhance compre-hension. As mentioned earlier, we used the CARS(Wehby et al., 1993) to determine the quality ofinstructional environment. The instructionalquality in Control classrooms was rated as moder-ate to very high.

PROCEDURES

Test-Administration Training. In October2004, the first author trained examiners to ad-minister all pretest measures in two 1.5-hr ses-sions. Examiners included seven female graduatestudents in Educational Psychology who all hadexperience working with young children inschools. In April 2005, a similar 1.5-hr trainingsession was held to review the pre- and posttestmeasures and introduce the additional posttest-only measures.

Following both pre- and posttest trainings,the first author held individual mock testing ses-sions with each examiner to establish interrateragreement. During this session, the examiner ad-ministered each measure to the first author, whonoted the examiner's accuracy in giving direc-tions, timing, and scoring the tests. The examinerscored each test, and a point-by-point agreementprocedure was used to compare the examiner's

2 0 4 Winter 2008

scores with the first author's scores. Interrateragreement was established by dividing the num-her of agreements by the number of agreementsplus disagreements and multiplying by 100 toyield a percentage. On any measure on which in-terrater agreement was les.s than 90%, the exam-iner was required to practice the measure andredo interrater agreement for that measure. Onaverage, pretests and posttests were administeredand scored with 98% agreement.

Pre- and Posttesting. To administer thepretests, examiners met individually with eachstudent for whom parental consent was obtained.The examiner spent several minutes establishingrapport with the child, and then administered theRLN test. In a second session, the examiner ad-ministered the remaining measures in randomorder to avoid an order effect. At posttest, an ex-aminer who w.is unfamiliar with each student(i.e., had never had contact with the child's cias.s-room during the study, to avoid an experimentereffect) spent several minutes establishing rapport.Then, the RLN test was adminisiered, followedby the remaining tests, which were again given inrandom order.

K-PALS Training and Support In this study,classroom teachers delivered all K-PALS instruc-tion. As part C'f the larger study, all K—PALSteachers attended a 1-day K-PALS workshopconducted by the first author. During this work-shop, teachers learned the purpose and researchbackground of K-PALS. They were then providedwith detailed descriptions and demonstrations ofK-PALS procedures. Each teacher received aK-PALS manual, which was previewed in detail.Teachers were given numerous opportunities topractice K-PALS activities in the roles of bothteacher and students. They were also given theopportunity to pair their students and thinkabout how they would organize their classroomsfor K-PALS. Teachers assigned to Level 3 andLevel 4 also rectived booster sessions and supportfrom K-PALS experts throughout the study.More detail about Level .3 and Level 4 can be ob-tained from the first author.

K-PALS Fidelity. A checklist of teacher andstudent behaviors essential to K-PALS was devel-oped to assess fidelity in the larger study (moreinformation about this checklist may be obtainedfrom the first author). One point was given for

each correct behavior that was observed. Pointswere added and divided by the total number ofpossible points, then multiplied by 100 to yield apercentage. In the larger study, four members ofthe research team conducted fidelity checks of the34 K-PALS classes in December and March.Each observer conducted 10% of her observationswith the first author. Percentage of interrateragreement on the fidelity checklist was establishedby dividing the number of agreements by agree-ments plus disagreements and multiplying by100. Mean interrater agreement was 9L25%. Thefidelity of K-PALS classes selected for this studywas above 90%, with a mean of 95%.

Classroom Atmosphere. A Classroom Atmo-sphere Rating Scale (CARS; Wehby et al., 1993)was used during cla.ssroom observations to deter-mine the quality of the instructional environmentfor all teachers in the larger study. Each classroomwas observed for 30 min during either K-PALSor structured reading instruction (for Controls),and rated on a scale of 1 {very high) to 5 {verylow). Items included levels of student complianceduring structured times, compliance during tran-sitions, adherence to rules, cooperation, interestand engagement, on-task behavior, and tbe degreeto which the environment was supportive of stu-dent behavior. A k)wer total score was associatedwith more positive ratings. Mean interrater agree-ment was 85%. K-PALS and Control teachers inthis study received ratings of moderate to veryhigh; as mentioned, there were no reliable differ-ences between the two groups.

Design and Data Analysis. A pretest-posttestcontroi group design using a matched sample wasused to examine the effectiveness of K-PALS forELs. Matching is a common technique used to se-lect control subjects who are matched with treatedsubjects on background covariates that the re-searcher believes need to be controlled (Gall, Gall,& Borg, 2003). Posttest data were analyzed usingtwo sets of analyses of covariance (ANCOVAs)with pretest RLN as a covariate to compare (a)K-PALS ELs to Control EI^ and (b) K-PALS ELsto K-PALS non-ELs. Pretest RLN was used as acovariate to control for any pretreatment groupdifferences not accounted for by the matchingprocedure. We chose RLN due to its strong initialpredictive validity of future reading achievement(e.g., Torgesen, Wagner, & Rashotte, 1997).

Exceptional ChiUren

To address our secondary research question,we determined proportions of students unrespon-sive to K-PALS and Control instruction. To dothis, we identified students who had scored belowthe 16th percentile (1 SD below the mean of thelarger Minnesota sample) on at least one of thefollowing posttest measures: RLN, RLS, WRMT-R Word ID and Word Attack, and Oral Reading.We chose the 16th percendie as it is currently rec-ommended as a reasonable RTI cutpoint (e.g., D.Fuchs &C Fuchs, 2006). We chose the above mea-sures for the following reasons: RLN for its stronginitial predictive validity of future reading achieve-ment (e.g., Torgesen et al., 1997), RLS for its di-rect relation to skills taught in K-PALS,WRMT-R Word ID and Word Attack to be con-sistent with previous studies examining at-riskreaders' response to intervention (e.g., Vellutino etal., 1996), and oral reading, to be consistent withcurrent RTI recommendations (e.g., D. Fuchs &Fuchs, 2006) and previous research (e.g., Vaughnet al., 2005). Ideally, we wouid have used progress-monitoring data to identify nonresponders; how-ever, this study is based on an extant data set thatdid not include such measures, and so the use ofprogress monitoring data was not possible.

RESU LTS

This study addressed the following primary re-search questions: (a) Do kindergarten ELs whoparticipate in K-PALS outperform Control ELson critical beginning reading skills? and (b) IsK-PALS as effective for ELs as for non-ELs? Toanswer these questions, we conducted ANCOVAson posttest outcomes with pretest RLN as covari-ate. Pre- and posttest means and standard devia-tions {SDs) are provided in Table 3. Results ofANCOVAs on posttest analyses are provided inTable 4.

POSTTEST DIFFERENCES

ANCOVAs for each reading measure are pre-sented in Table 4. All effect sizes reported belowwere calculated by dividing the mean differenceby the pooled SD.

K—PALS ELs Versus Control ELs. To answerour first research question (Do ELs who partici-pate in K-PALS outperform Control ELs on criti-

cal beginning reading skills?), we conducted AN-COVAs comparing K-PALS ELs to Control ELson all posttest measures, using pretest RLN as co-variate. The homogeneity of regression assump-tion was met for these analyses. Effect sizes werecalculated by dividing the mean difference by thepooled SD. K-PALS ELs reliably outperformedControl ELs on Segmentation, F(l,37) = 5.32, p= .05, d= .69; Blending, f{\, 37) = 5A6,p = .03,d^ 0.65; and KLS, F{U 57) = 6.14. p = .02, d =0.58. There were no reliable differences betweenK-PALS ELs and Control ELs on the other mea-sures (see Table 4).

K-PALS ELs Versus K-PALS Non-ELs. To an-swer our second research question (Is K-PALS aseffective for ELs as for non-ELs?), we conductedANCOVAs comparing K-PALS ELs to K-PALSnon-ELs on all posttest measures, using pretestRLN as covariate. The homogeneity of regressionassumption was mer for these analyses. Again, ef-fect sizes were calculated by dividing the mean dif-ference by the pooled SD. There were no reliabledifferences between K-PALS ELs and K-PALSnon-ELs on any of the measures (see Table 4).

RESPONSE TO INTERVENTION

A secondary research question was whether therewould be lower proportions of ELs unresponsiveto K-PALS than to more traditional classroom in-struction. Using the criterion of one SD belowthe larger study sample mean on at least one ofseveral posttest beginning reading measures, weidentified students who were unresponsive to in-tervention. Table 5 presents proportions ofK-PALS ELs, Control ELs, and K-PALS non-ELs who met this criterion at the end of thestudy. Chi square analyses indicated that therewere reliably fewer unresponsive K-PALS ELsthan Control ELs on RLS, x^{\) = 5.63, p = 0.02,and that there were reliably fewer unresponsiveK-PALS ELs than K-PALS non-ELs on RLS,

D I S C U S S I O N

The purpose of this study was to investigate theeffectiveness of K-PALS for ELs beginning toread. Many researchers focusing on early readinginstruction for ELs have employed supplemen-

Wimer 2008

TABLE 3

Means and SDs on Pre- and Posttest Measures by Instruetional Croup and EL Status

Measure

Phonemic awareness

SegmenrationPrePost

BlendingPrePost

AlphabeticRLN

PrePost

PrePost

Word IdencificanionPrePost

Word AtrackPrePost

SpellingPost

Oral readingOral reading A

Post

Oral reading BPost

K-PALS EU(n-

M

3.7518.7

0.909.60

19.3533.60

12.3541.45

0.6510.80

0.055.30

8.90

20.45

18.90

^20)

(SD)

(4.25)(11.72)

(1.59)(7.43)

(16.25)(16.91)

(10.90)(15.98)

(1.23)(11.72)

(0.22)(6.26)

(4.28)

(19.14)

(18.76)

Control EU

M

2.1510.80

0.905.20

19.9535.00

12.1530.95

1.109.55

0.303.85

8.75

16.85

16.85

20)

(SD)

(2.76)(11.29)

(2.45)(6.09)

(19.66)(16.08)

(13.82)(19.92)

(2.77)(11.34)

(0.92)(6.74)

(3.92)

(20.72)

(20.88)

K-PALS Non-EU

rnM

5.2020.35

1.359.50

19.3035.55

12.7538.55

2.609.80

0.904.45

8.05

22.90

23.65

= 20)

(SD)

(8.06)(11.32)

(2.16)(6.40)

(15.96)(21.38)

(11.48)(23.25)

(5.23)(10.82)

(2.34)(5.50)

(4.06)

(26.03)

(28.38)

Note. K-PALS - Kindergarten Peer-Assisted Learning Strategies; EL = English learner; RLN = Rapid LetterNaming; RLS - Rapid Letter Sound.

tary, Tier 2 type interventions with ELs identifiedas at risk of reading failure (de la Colina et al.,2001; Gerber et al., 2004; Gunn et al., 2000;Haager & Windmueller, 2001; Leafstedt et al.,2004; Vaughn, Cirino, et al., 2006; Vaughn,Linan-Thompson, et al., 2006; Vaughn, Mathes,et al., 2006). Few have examined the effects ofclassroom-based instruction, implemented by thegeneral classroom teacher, on the beginning read-ing achievement of ELs. We examined whetherK-PALS could serve as general classroom, "Tier1" instruction that would be beneficial for manyEl^, reserving increasingly intensive interventions

for the most severely struggling beginning readers.

We discuss our findings in terms of our primary

and secondary research questions in the following

section.

Do K-PALS ELs OUTPERFORM CONTROL

ELs ON CRITICAL BEGINNING READING

SKILLS?

In this study, K-PALS ELs reliably outperformedControl ELs on phonemic awareness (blendingand segmenting) and letter-sound recognition,with effect sizes ranging from .58 to .69 for these

Exceptional CJiildren

TA B LE 4

Analyses of Covariance (Posttest)^

Measure

Phonemic awarenessSegmentationBlending

AlphabeticRLNRLSWord IdentificationWord Attack

Spelling

Oral reading

Ota! reading AOral reading B

K-PALS EU

vs. Control ELs

V(U7)

5.32*

5.46*

0.086.14*0.441.300.07

1.070.36

0.690.65

^ . 0 80.58O.U0.220.04

0.180.10

K-PALS EUvs. K-PALS non-EU

W.37)

0.230.00

0.360.340.160.420.85

0.250.66

d"

-0.140.01

-0.100.150.090.140.20

-O.U-0.20

Note. K-PALS = Kindergarten Peer-Assisted Learning Strategies; EL = English learner; RLN = Rapid Letter

Naming; RLS - Rapid Letter Sound.

'Pretest RLN was used as a covariate. ^Effect sizes were calculated by dividing the mean difference by the pooled

SD.> < .05.

TABLE 5

Proportions of K-PALS and Control Students Unresponsive to Lnstruction

Posttest

RLN

RLS

Word Identification

Word Attack

Oral reading A

Oral reading B

K-PALSEU

(n = 20)

n(%)

5(25)

1(5)

1(5)

6(30)

1(5)

1(5)

ControlEU

(n = 20)

n(%)

6(30)

7(35)

2(10)

12(60)

3(15)

3(15)

K-PALS

noH'EU(n = 20)

n{%)

6(30)

6(30)

3(15)

9(45)

2(10)

4(20)

K-PALS EU

vs.

0.13

5.63

0.36

3.64

l .U

l .U

ControlEU

P

0.72

0.02'

0.55

0.06

0.29

0.29

K'PALSEUvs. K-PALS

non-EU

0.13

4.33

1.11

0.96

0.36

2.06

P

0.72

0.05*

0.29

0.33

0.55

0.15

Note. K-PALS = Kindergarten Peet-Assisted Learning Strategies; EL = English learner; RLN = Rapid LetterNaming; RLS = Rapid Letter Sound.> < .05.

Winter 2008

measures. These findings are convergent with pre-vious research demonstrating positive K-PALS ef-fects on the beginning reading performance ofnative English-speaking children (D. Fuchs et al.,2001); are consistent with other early reading re-search conducted with ELs (e.g., de la CoHna cral., 2001; Denton et al., 2004; Gerber et al.,2004; Gunn ec al., 2000; Leafstedt er al., 2004;Linan-Thompson et al., 2006; Vaughn, Cirino etal., 2006; Vaughn, Linan-Thompson et al., 2006;Vaughn, Mathes, er al., 2006); and support theuse of explicit, skills-based insrrucrion aimed atimproving ELs' beginning reading. Phonemicawareness and letter-sound recognition are essen-tial skills for developing reading proficiency (e.g..National Reading Panel, 2000). Postering theseskills in the general classroom tnay serve ro pro-vide at-risk readers, including ELs, with a founda-tion in reading that is critical for continuedschool success.

Although ihe above findings are encouraging,there were no statistically significant differencesbetween K-PALS and Gontrol ELs on theWRMT-R Word ID and Word Attack subtesrs,the WIAT spelling test, or Oral Reading. Thislack of reliable differences perhaps could be at-tributed to lack of sensitivity of the measures thatwere used; they may simply have not detectedsmall bur important differences between the twogroups following K-PALS. It is also possible thatthe stronger phonemic awareness andletter-sound recognition observed in K-PALSELs would eventually lead ro berrer reading per-formance; longitudinal research is needed to ad-dress this question. The relatively small samplesize also limited our power to detect small butpossibly impor':anr effects; indeed, effect sizes forWord Artack and Oral Reading A were positive(.22 and .18, respectively).

On the other hand, perhaps K—PALS wasnot intensive enough ro effect greater perfor-mance on these measures. Future research couldaddress (a) whether K-PALS leads to improvedperformance on more sensitive word-readingmeasures, (b) whether the impact K-PALS has onletter-sound recognition and phonemic awarenessskills evenrually leads to improved performanceon standardized word-reading measures and oralreading, and (c) whether increasing the duration

or intensity of K-PALS leads ro greater readingperformance.

Is K-PALS AS EFFECTIVE FOR ELS

AS FOR NON-ELS?

Results suggest that K-PALS was as effective forELs as for non-ELs in this study. This findingprovides direct evidence thar both ELs and non-ELs can benefit similarly from classroom-basedinstruction, which is important because Tier 1 in-terventions should be effective for diverse groupsof srudents. Yet, there are several considerationsimportant to the interpretation of these results.First, we should note that pretest RLN and RLSscores indicate that our non-EL sample (« = 20)performed slightly below (but wirhin .5 SD oOthe average-achieving students of the non-ELsample in the larger study (« = 397; RLN mean =26.45, SD = 16.34; RLS mean - 15.89, SD =12.97). In other words, by matching non-ELs toour EL sample, we selected from a somewhatlower-per forming segment of the larger sample.Thus, our findings apply to ELs as compared roslightly lower-performing non-EL peers, ratherthan to non-EL peers in general. Second, it is un-clear whether the ELs' pretest RLN and RLSscores reflected their reading skills, their limitedEnglish proficiency, or both. In other words, al-though the ELs and non-ELs were performingsimilarly at the beginning of the study, the rwogroups may have been low performing for differ-ent reasons. Further research should include anexamination of the impact of language profi-ciency on ELs' response to intervention. Third,borh K-PALS and Control ELs received moreourside reading help than K—PALS non-ELs. ELswere pulled our either fot English language in-struction or for additional reading help, whicbcould also account for their similar performanceto non-ELs at the end of the study. Future re-search could include an examination of whetherK-PALS alone is sufficient for both ELs and non-ELs, or whether ELs require outside reading helpin addition to K-PALS to maintain performancelevels similar to their non-EL peers.

Exceptional ChiUrefi ao»

Is K-PALS EFFECTIVE AS A TIER 1INTERVENTION EOR ELS?

In considering "evidence-based" approaches toTier 1 instruction, it is critical to understand forwhom those approaches are evidence-based(Klingner & Edwards, 2006). We further ex-plored this issue by examining whether there werelower proportions of students who were unre-sponsive to K-PALS than to more traditionalclassroom instruction. Results tentatively suggestthat more ELs were responsive to K-PALS thanto traditional instruction. Interestingly, it also ap-peared that more ELs than non-ELs were respon-sive to K-PALS, perhaps because the non-ELswere relatively low-performing (compared to thelarger sample). It is possible that the ELs weremore responsive because tbeir primary difficultieswere relared to language proficiency, which im-proved over time, whereas some of the non-El^'difficulties were related to true reading deficits.

Results tentatively surest that moreELs were responsive to K-PALS thanto traditional instruction.

However, because of the small numbers ofstudents in our sample, whether these findingshave practical significance remains unclear. Forexample, 5% of K-PALS ELs were unresponsivebased on RLS scores, compared ro 35% of Con-trol ELs and 30% of non-ELs. This differenceseems large, but in terms of sheer numbers trans-lates to 1 K-PALS EL, 7 Control ELs, and 6K-PALS non-ELs who were nonresponders. It isimportant to replicate these findings with moresrudents. In addition, we used only posttest per-formance ro judge responsiveness to instruction.However, researchers have suggested that a moreaccurate way of identifying students unresponsiveto intervention is to use a dual discrepancy ap-proach (L. S. Fuchs, 2003; Speece & Case, 2001),which takes into account both level and growrh.We were unable ro collect progress-monitoringdata needed to estimate growth within the re-sources of the larger study. Had we done so, wewould have been able to compare the amount ofgrowth ELs made in K-PALS versus traditional

instruction. This might have yielded differentproportions of unresponsive srudents than thosereported above.

We should also be clear that we do not auto-matically a.ssumc that students who did not meetour RTI criteria are truly unresponsive or thar theycannot benefit from classroom-based instruction.Rather, further work is needed to better under-stand whar needs to be in place for Tier 1 insrruc-tion to be most beneficial for at-risk ELs. Thiswork should include (a) an examination of in-strucrional components that are most beneficialfor ELs, and (b) an examination of instructionalcontexts that are most beneficial for ELs (seeKlingner & Fxiwards, 2006). With respect to in-structional components, K-PALS incorporates anumber of elements suggested to be important forELs (e.g., Gersren & Baker, 2000; Gersten &Geva, 2003; Vaughn, Linan-Thompson, et al.,2006), such as repetition, routines, modeling andpractice, and frequent opportunities to respond ina peer-mediated format. Explicit instruction in vo-cabulary and oral language might further supporrELs' development of early literacy skills (e.g.,Vaughn Linan-Thompson, et al., 2006). An exam-ination of whether explicit in.struction in theseskills can be successfully incorporated intoK-PALS and improve outcomes for ELs would beuseful.

With respect to instructional contexts, Kling-ner and Edwards (2006) suggested that it is alsoimportant to undersrand the impacr specificteacher and classroom variabies have on students'response ro intervention. Gersten, Baker, Haager,and Graves (2005) have addressed this issue byexamining the importance of teacher quality inpredicting outcomes for ELs learning to read.They used an observational instrument to identifyaspects of instruction that differenriaced effectivefrom ineffective teachers, as evidenced by theamount of reading growth made by ELs in theclassrooms they observed. Findings indicated thateffective teachers exhibited a "seamless quality" totheir instruction, whereby one activity flowedsmoothly to the next, with activities appropriatelypaced and designed to be interesting to beginningreaders. Effective teachers also integrated vocabu-lary instruction into reading activities, making themeanings of key words concrete through pictures,demonstrations, and many opportunities for prac-

2 1 O Winter 2008

tice. Writing wss also used frequently, often to ex-plicitly reinforce phonological, word analysis, andvocabulary skills.

Similar observations and analyses to thoseused by Gersten et al. (2005) might be useful forbetter understanding the impact of teacher qual-ity during K-Py\LS for ELs learning to read. Per-haps effective K-PALS reachers integratevocabulary and oral language more explicitly intoK-PALS lesson.s even though those skills are notpart of the K-PALS protocol. Perhaps effectiveK-PALS teachers find ways to seamlessly incorpo-rate K-PALS activities with other reading, lan-guage, and writing activities such that ctiticalskills are more readily generalized. Such informa-tion would shed light on instructional conditionsthat enhance K-PALS for ELs.

LIMITATIONS

There are several limitations to this study. First, inrhe larger srudv, classrooms were assigned ran-domly to K-PALS or Controls, but because thisstudy used extant data from the larger study, wewere unable to assign individual EI^ randomly toK-PALS or Control groups. Whereas our match-ing procedure allowed us to control for pretreat-ment differences, at least in terms of beginningreading skills, a randomized design would havebeen ideal. Seccmd, because of the limited num-ber of ELs in the control group (« = 20), we onlyused approximately 36% of the ELs from thelarger study (n = 40). Moreover, the majority ofthe ELs in the control group were Asian students(80%), whcrea.s K-PALS ELs were distributedacross three major ethnicities (Hispanic 40%,Asian 35%, and African 20%). We chose ro useinirial reading performance level rather than eth-nicity as our matching criterion as we consideredthis more relevant in establishing group equiva-lence; however, we cannot rule out the possibilitythat differences in ethnicities between K-PALSand Control EL^ could have influenced our out-comes. Finally, the generalizabiliry of the studyfindings is limited in that we only included ELsin the Minnesota site of the larger K-PALS study.We did so because EL populations and servicesprovided varied widely from one state to the next;thus, including ELs from other sites would fur-ther complicate our analyses. Furrher research is

needed to replicate our findings with other ELgroups.

IMPLICATIONS EOR RESEARCH

Our results suggest that K-PALS holds promise asan approach for promoting beginning readingskills for ELs. This finding extends other researchfocusing on K-PALS, including the larger study inwhich this study was embedded, as this study in-cludes a specific examination of the efficacy ofK-PALS for ELs. Yet, it is also clear that establish-ing effective classroom-based beginning readinginstruction for ELs is an area in need of further re-search. For example, research is needed to examinethe mediating effects of variables such as homelanguage, language proficiency, and SES on ELs'response to instruction. It is also important to un-derstand better the role of language of instruction.There is some evidence that bilingual or rransi-tional instruction is more effective for ELs thanEnglish-only instruction on reading performance(see Slavin & Cheung, 2005); however, given thatmany urban districts, including those in this study,are comprised of heterogeneous language popula-tions, it may be infeasible to provide bilingual in-struction aligned with each language group.

Our results suggest that K—PALS holdspromise as an approach for promotingbeginning reading skills for ELs.

With respect to specific K-PALS features, fu-ture research could explore ways to improve thisapproach for ELs. For example, researchers couldcompare rhe effects of pairing ELs with otherswho speak the same language versus pairing ELswith native English speakers, or of pairing stu-dents based on English proficiency (e.g., higherEnglish-proficient students paired with lower En-glish-proficient students). Another possible wayto enhance K-PALS features for ELs might be toadd explicit instruction in vocabulary and orallanguage. Finally, it is important to attend to con-textual factors char might infiuence the effective-ness of "evidence-based" instruction (Klingner &Edwards, 2006). Observational methods such asthose employed by Gersten et al. (2005) could

Exceptional Children 211

provide fiarther insight into instructional condi-tions thar impact K-PALS effectiveness.

Another possible way to enhance K-PALS

features for ELs might be to add explicit

instruction in vocabulary and oral

language.

IMPLICATIONS FOR PRACTICE

K-PALS was implemented by classroom teachersin this study, increasing our confidence in its fea-sibility for classroom use. Although 20 to 30 minper session, 4 days per week may seem time-in-tensive for a supplement to core instruction, ourteachers reported that K-PALS fit well with theirexisting language arts programs and required littlepreparation time. Educators interested in imple-menting K-PALS as Tier 1 instruction shouldthtis be encouraged to do so, with the followingconsiderations: Firsr, implementing K-PALS withfidelity is essential, hi this study, we only includedstudents from K-PALS classes with high fidelity.We do not know whether results would be similarif teachers were to modify K-PALS activities sub-stantially.

Second, K-PALS is likely to be most benefi-cial if it is part of a carefully orchestrated begin-ning reading program in which activities areseamlessly integrated, reinforcing oral language,phonologic, decoding, vocabulary, writing, andother skills critical for beginning readers. Finally,as with any evidence-based approach, K-PALSwill not be effective for all students. In this study,several ELs and non-ELs were performing signifi-cantly below average on some measures followingK-PALS. Ir is critical tbat teachers monitor theprogress of students at risk and implement instruc-tional changes when students are not sufficientlyresponding to classroom instruction. It is our hopethat implementing such a process with fidelity willhelp to ensure rhat appropriate classroom instruc-tion is in place for the majority of students, in-cluding ELs, reserving more intense interventionsfor the most severely stru^ing readers.

REFERENCES

Abedi, J. (2004). The No Child Left Behind Act andEnglish language learners: Assessment and accountabil-ity issues. Educational Research, 33, 4-14.

Cummins, J. (1980). The entry and exit fallacy inbilingual education. NABEJournal, 4, 25-29.

D'Angiulli, A., Siegel, L. S., & Maggi, S. (2004). Liter-acy instruction, SES, and word-reading achievement inEnglish-language learnets and children with Hnglish asa first language: A longitudinal study. Leaming Disabil-ities Research and Practice, 19, 202-213.

de la Colina. M. G.. Parker, R. I., Hasbrouck, J. E., &l ara-Alecio, R. (2001). inten.sive intervenrion in read-ing fluency for at-risk beginning Spanish readers. Bilin-gual Research Journal, 25, 503-538.

Denton, C. A., Anthony, J. L., Parker, R., & Has-brouck, J. E. (2004). Effects of two tutoring programson the English reading development of Spanish-En-glish bilingual students. The Elementary School Journal,104(4), 289-305.

Fuch.s, D., & Euchs, L. S. (2006). Introduction ro re-sponse CO intervention: What, why, and how valid is it?Reading Research Qmirterly, 41, 92-')9.

Fuchs. D., Fuchs, L. S., Thompson, A., Al Otaiba, S.,Yen, L, Yang, N., et al. (2001). Is reading important inreading-read!ness programs? A randomized field trialwith teachers as program imp[cmcmers. Journal of Edu-cational Psychology, 93, 251-267.

Fuchs, D., Fuchs, L. S., Thompson, A., Ai Otaiba, S.,Yen, L., Yang, N., et al. (2002). Exploring the impor-tance of reading programs for kindergartners wirh dis-abilities in mainstream classrooms. ExceptionalChildren, 68,295-311.

Fuchs, D., Mock, D., Morgan, P. L., & Young, C. L.(2003). Rcsponsiveness-co-incervention: Defmitions,evidence, and implications fot the learning disabilitiesconstruct. Learning Disabilities, Research, & Practice,18. 157-171.

Fuchs, L. S. (2003). Assessing intervention responsive-ness: Conceptual and technical issues. Learning Disabil-ities Research & Practice, 18> 172-186.

Gall, M. D., Gall, J. R, & Borg, W. R. (2003). Educa-tional research: An introduction (7rh ed.). Boston: Allyn& Bacon.

Gerber, M., Jimenez, T., Leafstedt, J., Villaruz, J.,Richards, C , & English, J. (2004). English reading cf-teccs of small-group intensive intervention in Spanishfor K-1 English learners. Leaming Disabilities Research& Practice. 79, 239-251.

Winter 2008

Gersten, R., & Baker, S. (2000). What we know abouteffective instructional practices for English-languagelearners. Exceptional Children. 66, A'yA-AlO.

Gersten, R., Baker, S., Haager, D., & Graves, A.(2005). Explorin; the role of teacher quality in predict-ing reading outcomes tor first-grade English learners:An observational study. Remedial & Special Education,26, 197-206.

Gersten, R., & Geva, E. (2003, April). Teaching read-ing to early language learners. Educational Leadership,60, 44-49.

Gunn, B., Biglan, A., Smolkowski, K., & Ary. D.(2000). The efficacy of supplemental itisrruccion in de-coding skills for Hispanic and non-Hispanic studentsin early elementary school. Journal of Special Education,34, 90-103.

Haager, D., &i Windmueller, M. P (2001). Early tead-ing intervention for English language learners ar-riskfor learning disabilities: Student and teacher outcomesin an urban schcol. Leaming Disability Quarterly, 24,235-250.

Individuals With Disabilities Education ImprovementAct, Pub. L 108-446 U.S.C. (2004).

Klingner, J. K., &: Edwards, P. A. (2006). Cultural con-siderations with response to intervention models. Read-ing Research Quarterly, 41, 108-117.

Leafkedt, J. M., Richards, C. R., & Gerber, M. M.(2004). Effectiveness of explicit phonological-awarenessinstruction for .ii-risk English learners. Learning Dis-abilities Research und Practice, 19, 252-261.

Levy, B. A., & Lysunchuk, L. (1997). Beginning wordrecognition: Benefits of training by .segmentation andwhole word methods. Scientijic Studies of Reading, 1,359-387.

Linan-Thompson, S.. Vaughn, S., Hickman-Davis, P.,& Kouzekanani, K. (2003). Effectiveness of supple-mental reading instruction for second-grade Flnglishlanguage learners with leading difficulties. ElementarySchool Journal, /03. 221-238.

Linan-Thompson, S., Vaughn, S., Prater, K., & Cirino,R T. (2006). The response to intervention of Englishlanguage learners at risk for reading problems. Journalof Leaming Disabilities, 39, 390-398.

Long, M. H., & Porter, P A. (1985). Group work, in-terlanguage talk, and second language acquisition.TESOL Quarterly, 19, 207-228.

Lyon, G. R., Fletcher, J. M., Shaywitz, S. E., Shaywitz,B. A., Torgesen, J. K., Wood, E B., et al. (2001). Re-thinking learning disabilities. In C. E. Finn, Jr., A. J.Rotherham, & C. R. Kokanson. Jr. (Eds.), Rethinking

special education for a new century (pp. 259-288).Washington, DC: Thomas B. Fordham Foundation.

McCardle, P., Mele-McCarthy, J., Cutting, L, Leos, K.,& D'Emilio, T. (2005). Learning disabilities In Englishlanguage learners: Identifying the issues. Leaming Dis-abilities Research & Practice, 20, 1-5.

McMasrer, K. L., Fuchs, D., & Fuchs, L. S. (2002).Using peer tutoring to prevent early reading failure. InJ. S. Thousand, R. A. Villa, & A. Nevin (Eds.), Creativ-ity and collaborative learning: The practical guide to em-powering students, teachers, and families (pp. 235-246).Baltimore: Paul H. Brookes.

Minnesota Department of F~ducation. (2003). Min-nesota modified student oral language observation matrix.Retrieved May 9, 2006, from http://children.state,mn. us/mde/static/

Minnesota Department of Education. (2004). Min-nesota English Language I'roficiency Standards for EnglishLanguage Learners K-12. Retrieved March 15, 2006,from http://education.state.mn.us/mde/static/002193.pdf

National Center for Education Statistics. (2005). Thenation's report card: 2005 assessment results. RetrievedJanuary 5. 2006, from http://nces.ed.gov/nationsreportcard/nrc/reading_math_200 5 /

National Reading Panel. (2000). Teaching students toread: An evidence-based assessment of the scientific re-search literature on reading and its implications Jor read-ing instruction. Btrthesda, MD: National Institute ofChild Health and Human Development.

No Child Left Behind Act, Pub. L No. 107-110, 115Stat. 1425, 2002 U.S.C. (2002).

Psychological Corporation. (1992). Wechsler individualAchievement Test. San Anronio, TX: The PsychologicalCorporation, Harcourt Brace Jovanovich.

Slavin, R. E., & Cheung, A. (2005). A synthesis of re-search on language of reading instruction for Englishlanguage learners. Review of Educational Research, 75,lAl-l^A.

Snow, C. E., Burns, S., & Griffin, P (1998). Preventingreading difficulties in young children. Washington, DC:National Academy Press.

Speece, D. L., & Case, L. P (2001). Classification incontext: An alternative approach to identifying earlyreading disability. Journal of Educational Psychology, 93,

Torgesen, J. K., Alexander, A. W., Wagner, R. K.,Rashorte, C. A., Voeller, K. K. S., & Conway, T.(2001). Intensive remedial instruction for children withsevere reading disabilities: Immediate and long-rerm

Exceptional Children 2 1 3

outcomes from two instructional approaches.Leaming Disabilities. 34, 33-58, 78.

Torgesen, J. K., Wagner, R. K., & Rashotte, C. A.(1997). Prevention and remediation of severe readingdisabilities: Keeping the end in mind. Scientific Studiesof Reading, 7,217-234.

U.S. Department of Education, U.S. Department ofHealth and Human Setvices. (2003, October). Na-tional symposium on learning disabilities in English lan-

guage learners. Retrieved November 17, 2006, fromhtcp://www. ed.gov/about/off ices/ l is t /osers/products/Id-ell/index, html

Vaughn, S., Cirino, P. T., Linan-Thompson, S.,Mathes, P G., Carlson, C. D., & Hagan, E. C , et al.(2006). Effectiveness of a Spanish intervention and anEnglish intervention for English-language learners atrisk for reading problems. American Educational Re-search JoumaL 43, 449-487.

Vaughn, S., & Linan-Thompson, S, (2003). Group sizeand time allotted to intervention: Effects for studentswith reading difficulties. In B. Foorman (Ed.), Prevent-ing and remediating reading difficulties: Brining scienceto scale (pp. 229-324). Baltimore: York Press.

Vaughn, S., Linan-Thompson, S., Mathes, P. G.,Cirino, P. T, Carlson, C. D., & Pollard-Durodola, S.D. et al. (2006). Effectiveness of Spanish interventionfor first-grade English language learners at risk for read-ing difficulties. Journal of Learning Disabilities, 39,56-73.

Vaughn, S., Linan-Thompson, S., Pollard-Durodola, S.D., Mathes, P G., & Hagan, E. C. (2006). Effective in-terventions for English language learners (Spanish-En-glish) at risk for reading difficulties. In D. K.Dickinson & S. B. Neuman (Eds.), Handbook of earlyliteracy research. Vol. 2. (pp. 185-197). New York: Guil-ford Press.

Vaughn, S., Mathes, P. G., Linan-Thompson, S.,Cirino, P. T., Carlson, C. D., Pollard-Durodola, S. D.,et al. (2006). First-grade English language learners at-risk for reading problems: Effectiveness of an Englishimervendon. Elementary SchoolJoumal, 107, 153-180.

Vaughn, S., Mathes, P. G., Linan-Thompson, S., &Francis, D. J. (2005). Teaching English language learn-ers at risk for reading disabilities to read: Putting re-search into practice. Leaming Disabilities Research &Practice, 20, 58-67.

Vellutino, E R., Scanlon, D. M., Sipay, E. R., Small, S.,Chen, R., Pratt, A., et al. (1996). Cognitive profiles ofdifficult-to-remediare and readily remediated poorreaders: Early intervention as a vehicle for distinguish-ing between cognitive and experiential deficits as basic

causes of specific reading disabiliry. Journal of Educa-tional Psychology, 88, 601-638.

Vygotsky, L. S. (1987). The collected works of L. S.Vygotsky. New York: Plenum.

Wagner, R. K., Francis, D. J.. & Morris, R. D. (2005).Identifying English language learners with learning dis-abiliries: Key challenges and possible approaches.Learning Disabilities Research & Practice, 20, 6-15.

Wehby, J. H., Dodge, K. A., & Greenberg, M. (1993).Classroom Atmosphere Rating Scale. Unpublished techni-

cal manual, Vanderbilr University, Nashville, Tennesee.

Woodcock, R. W. (1987). Woodcock Reading Mastery

Tests-Revised. Circle Pines, MN: American GuidanceService,

Yopp, H. K. (1988). The validity and reliability ofphonemic awareness tests. Reading Research Quarterly,

23, 159-177.

ABOUT THE AUTHORS

KRISTEN L. MCMASTER (CEC MN Federa-tion), Assistant Professor; S H U - H S U A N KUNG

(CEC Taipei Federarion), Doctoral Student:INSOON HAN (CEC MN Federation), DoctoralStudent; and MARISA CAO (CEC MN Federa-tion), Education Specialist, Department of Edu-cational Psychology, Universiry of Minnesota,Minneapolis.

Correspondence concerning this article should beaddressed to Krlsten McMaster, University ofMinnesota, 35OA Education Sciences Building,56 East River Rd., Minneapolis, MN 55455(e-mail: mcmas004@umn.edu).

This research was supported in part hy Crant#R305G0401O4 from the Institute for EducationSciences in the U.S. Department of Education toVanderhilt University, the University of Min-nesota, and the University of Texas-Pan Ameri-can. The arricle does not necessarily reflect theposition or policy of the funding agency, and noofficial endorsement should be inferred.

The authors thank Douglas Fuchs for his helpfulfeedback on an earlier version of this article.

Manuscript received June, 2006; accepted March2007.

214 Winter 2008