MD. SPONS AGENCY - ERIC · Jessie Pollack Maryland Department of Education Baltimore, Maryland. Sally Porter Blair High School Silver Spring, Maryland. Michael Pressley. State University

DOCUMENT RESUME

ED 379 636 CS 012 042

AUTHOR Brown, Rachel; And OthersTITLE A Quasi-Experimental Validation of Transactional

Strategies Instruction with Previously Low-Achieving,Second-Grade Readers. Reading Research Report No.33.

INSTITUTION National Reading Research Center, Athens, GA.;National Reading Research Center, College Park,MD.

SPONS AGENCY Office of Educational Research and Improvement (ED),Washington, DC.

PUB DATE 95

CONTRACT 117A20007NOTE 46p.PUB TYPE Reports Research/Technical (143)

EDRS PRICE MF01/PCO2 Plus Postage.DESCRIPTORS Comparative Analysis; Grade 2; *Instructional

Effectiveness; Primary Education; *ReadingAchievement; Reading Improvement; Reading Research;*Reading Strategies; *Remedial Reading; TeachingMethods

IDENTIFIERS *Students Achieving Independent Learning;*Transactional Strategies Instruction

ABSTRACTA study investigated the effectiveness of the

Students Achieving Independent Learning (SAIL) program, aneducator-developed approach to transactional strategies instruction(TSI). Five groups of six previously low-achieving second-gradestudents received a year of transactional strategies instruction andfive groups of six students received a year of more conventionalreading instruction provided by teachers who were highly regarded byschool district personnel. Each of the 10 groups was housed in adifferent classroom, with each SAIL group matched to a comparisongroup that was close in reading achievement level and matcheddemographically to the school providing the SAIL group. By the end ofthe academic year, there was clear evidence of greater knowledge anduse of strategies by the TSI students, greater acquisition ofinformation from material read in reading group, and superiorperformance on standardized reading tests. Findings suggest a clearvalidation to date of educator-developed transactional strategiesinstruction. (Contains 51 references and five tables of data. Asample SAIL lesson is attached.) (Author/RS)

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

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

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

A Quasi-Experimental Validation ofTransactional Strategies Instructionwith Previously Low-Achieving,Second-Grade Readers

Rachel BrownState University of New York at Buffalo

Michael PressleyState University of New York at Albany

Peggy Van MeterUniversity of Maryland College Park

Ted SchuderMaryland State Department of Education

U R. DEPARTMENT OF EDUCATION0Mce ol Educational Research and Improvement

EDUCATIONAL RESOURCES INFORMATIONCENTER (ERIC)

This document has been reproduced aseceiveo Iron the person or organization

originating it0 Minor changes have been made to improve

reproduction quarry

Points of view or opinions slated in this documen! do not necessarily iePraoecil officialOERI DOSitOn or policy

NationalReading ResearchanterREADING RESEARCH REPORT NO. 33

Winter 1995

2°EST COPY AVAltABLE

NRRCNational Reading Research Center

A Quasi-Experimental Validation of TransactionalStrategies Instruction with Previously Low-Achieving,

Second-Grade Readers

Rachel BrownState University of New York at Buffalo




READING RESEARCH REPORT NO. 33

Winter 1995

The work reported herein is a National Reading Research Project of the University of Georgiaand University of Maryland. It was supported under the Educational Research andDevelopment Centers Program (PR/AWARD NO. I I7A20007) as administered by the Officeof Educational Research and Improvement, U.S. Department of Education. The findings andopinions expressed here do not necessarily reflect the position or policies of the NationalReading Research Center, the Office of Educational Research and Improvement, or the U.S.Department of Education.

3

NRRC NationalReading ResearchCenter

Executive CommitteeDonna E. Alvermann, Co-DirectorUniversity of Georgia

John T. Guthrie, Co-DirectorUniversity of Maryland College Park

James F. Baumarn, Associate DirectorUniversity of Geo gia

Patricia S. Koskinen, Associate DirectorUniversity of Maryland College Park

Nancy B. Mizelle, Acting Associate DirectorUniversity of Georgia

Jamie Lynn Metsala, Interim Associate DirectorUniversity of Maryland College Park

Penny OldfatherUniversity of Georgia

John F. O'FlahavanUniversity of Maryland College Park

James V. HoffmanUniversity of Texas at Austin

Cynthia R. HyndUniversity of Georgia

Robert SerpellUniversity of Maryland Baltimore County

Betty ShockleyClarke County School District, Athens, Georgia

Linda DeGroffUniversity of Georgia

Publications Editors

Research Reports and PerspectivesLinda DeGroff, EditorUniversity of Georgia

James V. Hoffman, Associate EditorUniversity of Texas at Austin

Mariam Jean Dreher, Associate EditorUniversity of Maryland College Park

Instructional ResourcesLee Galda, University of GeorgiaResearch HighlightsWilliam G. HollidayUniversity of Maryland College Park

Policy BriefsJames V. HoffmanUniversity of Texas at Austin

VideosShawn M. Glynn, University of Georgia

NRRC StaffBarbara F. Howard, Office ManagerKathy B. Davis. Senior SecretaryUniversity of Georgia

Barbara A. Neitzey, Administrative AssistantValerie Tyra, AccountantUniversity of Maryland College Park

National Advisory BoardPhyllis W. AldrichSaratoga Warren Board of Cooperative EducationalServices, Saratoga Springs, New York

Arthur N. ApplebeeState University of New York, Albany

Ronald S. BrandtAssociation for Supervision and CurriculumDevelopment

Marsha T. DeLainDelaware Department of Public InstructionCarl A. GrantUniversity of Wisconsin-Madison

Walter KintschUniversity of Colorado at Boulder

Robert L. LinnUniversity of Colorado at Boulder

Luis C. MollUniversity of Arizona

Carol M. SantaSchool District No. 5Kalispell, Montana

Anne P. SweetOffice of Educational Research and Improvement,U.S. Department of Education

Louise Cherry WilkinsonRutgers University

Production EditorKatherine P HutchisonUniversity of Georgia

Dissemination CoordinatorJordana E. RichUniversity of Georgia

Text FormatterAnn Marie VanstoneUniversity of Georgia

NRRC - University of Georgia318 AderholdUniversity of GeorgiaAthens, Georgia 30602-7125(706) 542.3674 Fax: (706) 542-3678INTERNET: [email protected]

NRRC - University of Maryland College Park2102 J. M. Patterson BuildingUniversity of MarylandCollege Park, Maryland 20742(301) 405-8035 Fax: (301) 314-9625INTERNET: NRACQuinail.umd.edu

4

About the National Reading Research Center

The National Reading Research Center (NRRC) isfunded by the Office of Educational Research andImprovement of the U.S. Department of Education toconduct research on reading and reading instruction.The NRRC is operated by a consortium of the Universi-ty of Georgia and the University of Maryland CollegePark in collaboration with researchers at several institu-tions nationwide.

The NRRC's mission is to discover and documentthose conditions in homes, schools, and communitiesthat encourage children to become skilled, enthusiastic,lifelong readers. NRRC researchers are committed toadvancing the development of instructional programssensitive to the cognitive, sociocultural, and motiva-tional factors that affect children's success in reading.NRRC researchers from a variety of disciplines conductstudies with teachers and students from widely diversecultural and socioeconomic backgrounds in pre-kinder-garten through grade 12 classrooms. Research projectsdeal with the influence of family and family-schoolinteractions on the development of literacy; the interac-tion of sociocultural factors and motivation to read; theimpact of literature-based reading programs on readingachievement; the effects of reading strategies instructionon comprehension and critical thinking in literature,science, and history; the influence of i^.evative groupparticipation structures on motivation and learning; thepotential of computer technology to enhance literacy;and the development of methods and standards foralternative literacy assessments.

The NRRC is further committed to the participationof teachers as full partners in its research. . betterunderstanding of how teachers view the development ofliteracy, how they use knowledge from research, andhow they approach change in the classroom is crucial toimproving instruction. To further this understanding,the NRRC conducts school-based research in whichteachers explore their own philosophical and pedagogi-cal orientations and trace their professional growth.

5

Dissemination is an important feature of NRRC activi-ties. Information on NRRC research appears in severalformats. Research Reports communicate the results oforiginal research or synthesize the findings of severallines of inquiry. They are written primarily for re-searchers studying various areas of reading and readinginstruction. The Perspective Series presents a widerange of publications, from calls for research andcommentary on research and practice to first-personaccounts of experiences in schools. InstructimilResources include curriculum materials, instructionalguides, and materials for professional growth, designedprimarily for teachers.

For more information about the NRRC's researchprojects and other activities, or to have your nameadded to the mailing list, please contact:

Donna E. Alvermann, Co-DirectorNational Reading Research Center318 Aderhold HallUniversity of GeorgiaAthens, GA 30602-7125(706) 542-3674

John T. Guthrie, Co-DirectorNational Reading Research Center2102 J. M. Patterson BuildingUniversity of MarylandCollege Park, MD 20742(301) 405-8035

NRRC Editorial Review Board

Patricia AdkinsUniversity of Georgia

Peter AfflerbachUniversity of Maryland College Park

JoBeth AllenUniversity of Georgia

Patty AndersUniversity of Arizona

Tom AndersonUniversity of Illinois at Urbana-Champaign

Harriette ArringtonUniversity of Kentucky

Irene BlumPine Springs Elementary SchoolFalls Church, Virginia

John BorkowskiNotre Dame University

Cynthia BowenBaltimore County Public SchoolsTowson, Maryland

Martha CarrUniversity of Georgia

Suzanne ClewellMontgomery County Public SchoolsRockville, Maryland

Joan ColeyWestern Maryland College

Michelle CommeyrasUnivffcity of Georgia

Linda CooperShaker Heights City SchoolsShaker Heights, Ohio

Karen CostelloConnecticut Department of EducationHartford, Connecticut

Karin DahlOhio State University

Lynne Diaz-RicoCalifornia State University-San

Bernardino

Pamela DunstonClemson University

Jim FloodSan Diego State University

Dana FoxUniversity of Arizona

Linda GambrellUniversity of Maryland College Park

Valerie GarfieldChattahoochee Elementary SchoolCumming, Georgia

Sherrie Gibney-ShermanAthens-Clarke County SchoolsAthens, Georgia

Rachel GrantUniversity of Maryland College Park

Barbara GuzzettiArizona State University

Jane HaughCenter for Developing Learning

PotentialsSilver Spring, Maryland

Beth Ann HerrmannNorthern Arizona University

Kathleen HeubachUniversity of Georgia

Susan HillUniversity of Maryland College Park

Sally Hudson-RossUniversity of Georgia

Cynthia HyndUniversity of Georgia

Robert JimenezUniv of Oregon

Karen oohnsonPennsylvania State University

James KingUniversity of South Florida

Sandra KimbrellWest Hall Middle SchoolOakwood, Georgia

Kate KirbyGwinnett County Public SchoolsLawrenceville, Georgia

Sophie KowzunPrince George's County SchoolsLandover, Maryland

Linda LabboUniversity of Georgia

Rosary LalikVirginia Polytechnic Institute

Michael LawUniversity of Georgia

Sarah McCartheyUniversity of Texas at Austin

Veda McClainUniversity of Georgia

Lisa McFallsUniversity of Georgia

Mike McKennaGeorgia Southern University

Donna MealeyLouisiana State University

Barbara MichaloveFowler Drive Elementary SchoolAthens, Georgia

Akintunde MorakinyoUniversity of Maryland College Park

Lesley MorrowRutgers University

Bruce MurrayUniversity of Georgia

Susan NeumanTemple University

Caroline NoyesUniversity of Georgia

John O'FlahavanUniversity of Maryland College Park

Penny OldfatherUniversity of Georgia

Joan PagnuccoUniversity of Georgia

Barbara PalmerMount Saint Mary's College

Mike PickleGeorgia Southern University

Jessie PollackMaryland Department of EducationBaltimore, Maryland

Sally PorterBlair High SchoolSilver Spring, Maryland

Michael PressleyState University of New York

at Albany

Tom ReevesUniversity of Georgia

Lenore RinglerNew York University

Mary RoeUniversity of Delaware

Nadeen T. RuizCalifornia State University-Sacramento

Rebecca SammonsUniversity of Maryland College Park

Paula SchwanenflugelUniversity of Georgia

Robert SerpellUniversity of Maryland Baltimore

County

Betty ShockleyFowler Drive Elementary SchoolAthens, Georgia

Susan SonnenscheinUniversity of Maryland Baltimore

County

Steve StahlUniversity of Georgia

Anne SweetOffice of Educational Research

and Improvement

Lilting TaoUniversity of Georgia

Ruby ThompsonClark Atlanta University

Louise TomlinsonUniversity of Georgia

Sandy TianarkinStrawberry Knolls Elementary SchoolGaithersburg, Maryland

Sheila ValenciaUniversity of Washington

Bruce VanSledrightUniversity of Maryland College Park

Chris WaltonNorthern Territory UniversityAustralia

Janet WatkinsUniversity of Georgia

Louise WaynantPrince George's County SchoolsUpper Marlboro, Maryland

Priscilla WaynantRolling Terrace Elementary SchoolTakoma Park, Maryland

Dera WeaverAthens AcademyAthens, Georgia

Jane WestAgnes Scott

Steve WhiteUniversity of Georgia

Allen WigfieldUniversity of Maryland College Park

Shelley WongUniversity of Maryland College Park

About the Authors

Rachel Brown is an assistant professor in theDepartment of Counseling and Educational Psy-chology, the State University of New York atBuffalo. Her research interests include text

comprehension, strategies instruction, and

instructional computing.

Michael Pressley is a Professor in the Departmentof Educational Psychology and Statistics, the StateUniversity of New York at Albany. He has studiedstrategies instruction extensively since the 1980sand is an authority in the field of cognitive readingstrategies instruction.

Peggy Van Meter is a doctoral student in theDepartment of Human Development at the Univer-sity of Maryland College Park. Her specializationis in the area of educational psychology. Shecurrently is working as a research assistant for theNational Reading Research Center.

Ted Schuder is a specialist in the Research andDevelopment division of the Maryland StateDepartment of Education. He has developedstrategies-based interventions and conducted

professional development workshops for teacherson strategies instruction.

National Reading Research CenterUniversities of Georgia and MarylandReading Research Report No. 33Winter 1995

A Quasi-Experimental Validation of TransactionalStrategies Instruction with Previously Low-Achieving,

Second-Grade Readers

Rachel Brown

State University of New York at Buffalo




Abstract. Previously low-achieving second-gradestudents received either a year of transactionalstrategies instruction (I'S!) or a year of more con-ventional reading instruction provided by teacherswho were highly regarded by school district person-

nel. By the end of the academic year, there wasclear evidence of greater knowledge and use ofstrategies by the TSI students, greater acquisition of

information from material read in reading group,and superior performance on standardized readingtests. This is the clearest validation to date ofeducator-developed transactional strategies instruc-

tion.

Since Durkin's (1979) seminal finding thatAmerican students receive little instruction

1

about how to comprehend text, there have beenextensive efforts to identify teachable strategiesto increase students' understanding and memo-ry of text. Early on research strategies (for re-views, see Dole, Duffy, Roehler, & Pearson,1991; Pressley, Johnson, Symons, McGold-rick, & Kurita, 1989) tended to focus on in-struction of individual strategies and improve-ments in narrowly defined performances, suchas performance on standardized comprehensiontests following instruction in reading strategies.The typical research method used in thesestudies was to teach one group of students touse a particular cognitive strategy duringreading, while comparison-group students usedtheir own resources to understand text as best

2 Rachel Brown, Michael Pressley, Peggy Van Meter and Ted Schuder

they could. Through this approach, a relativelysmall number of strategiesfor example, visu-alizing ideas in text, summarizing, and self-questioningproved effective in increasingelementary students' comprehension and mem-ory of text. The single-strategy investigationsdemonstrated that when students were underexceptionally strong instructional control (i.e.,they were cold when and where to use a partic-ular strategy), they could use that strategy toimprove comprehension and learning. Butsingle-strategy instruction was rarely observed;there was no evidence of generalized improve-ment in reading.

Based on what is now known about skilledreading, it is not surprising that improvementin reading required more than instruction insingle strategies. During the late 1970s andearly 1980s, a number of analyses of skilledreading were conducted (e.g., Johnston &Afflerbach, 1985; Lytle, 1982; Olshaysky,1976-77; Olson, Mack, & Duffy, 1981; seePressley & Afflerbach, in press, for a summa-ry). It became apparent that skilled reading didnot involve the use of a single potent strategybut rather, the orchestration of a complexrepertoire of cognitive processes. This under-standingthat skilled readers coordinate anumber of strategies while readingpartiallyfueled researchers' efforts to develop instruc-tional interventions that involve teaching multi-ple comprehension strategies (Baker & Brown,1984).

Palincsar and Brown's reciprocal teachingapproach (1984) was one of these researcherdesigned, multiple-strategies interventions. Theresearchers taught students to app'y four strate-gies to expository text as they read: generate

predictions, ask questions, seek clarification,and summarize content. The students usedthese strategies in reading groups, with theadult teacher, gradually turning over responsi-bility for strategic processing as much as possi-ble to students. Palincsar and Brown's idea,consistent with Vygotsky's (e.g., 1978) theoryof socially mediated learning, was that partici-pation in reading-group discussions that in-volve predicting, questioning, seeking clarifi-cation, and summarizing would lead to theinternalization of these processes. In fact, amonth or two of such instruction producesnoticeable improvement in the use of the focalstrategies but only modest improvement onstandardized reading tests (for a review, seeRosenshine & Meister, 1994).

In addition to Palincsar and Brown's study,there were other attempts to teach multiplecomprehension strategies. Some involved pre-senting a large number of strategies quickly;these typically failed to produce improvementsin elementary readers' comprehension (e.g.,Paris & Oka, 1986). Other interventions in-volved more intensive, direct explanation andmodeling of small repertoires of strategies;these approaches generally were more success-ful in improving reading (e.g., Bereiter &Bird, 1985; Collins, 1991; Duffy et al., 1987).

Many educators became aware of research-ers' successes with strategies instruction andbegan to import such instruction into theirclassrooms. It became apparent, however, thatwhen strategies instruction was successfullydeployed in schools, it involved much morethan the type of instruction typically studied inthe well-controlled experiments (e.g., theteaching of single strategies or the quick pre-

NATIONAL READING RESEARCH CENTER, READING RESEARCH REPORT NO. 33

1 0

Transactional Strategies Instruction 3

sentation of many strategies; see Pressley,Goodchild, Fleet, Zajchowski, & Evans,1989). This motivated Pressley and his col-leagues to study the use of comprehensionstrategies instruction in elementary schools (seePressley & El-Dinary, 1993).

After investigating several educator-devel-oped programs, the research group headed byPressley proposed that effective comprehensioninstruction in elementary grades was "transac-tional" in three senses of the term (Pressley,El- Dinary, et al., 1992). First, readers learn tolink text content to prior knowledge, as theyconstruct meaning: this is consistent withRosenblatt's (1978) use of the term. Second,since much of strategies instruction occurs inreading groups, group members use strategiesto construct meaning togethertransactionalactivity in the sense that understanding con-structed by the group differs from the personal-ized interpretations individuals would havegenerated on their own. This is consistent withthe use of the term in organizational psycholo-gy (e.g., Hutchins, 1991). Third, the responsesof all members of the group (including theteacher's instructional decisions and practices)are determined in part by those of others in thegroup. This is "transactional" as the term isused by social development researchers (Bell,1968).

Since the instruction studied in elementaryschool classrooms was transactional in thesethree senses, the research team called it trans-actional strategies instruction (TSI). The short-term goal of TSI is the joint construction ofreasonable interpretations by group members asthey apply strategies to texts. The long-termgoal is the internalization and adaptive use of

strategic processing whenever students encoun-ter demanding text. Both goals are promotedby teaching reading group members to emulateexpert readers' use of comprehension strategieswhen they attempt to understand challengingtext (e.g., Pressley & Afflerbach, in press;Wyatt et al., 1993): Expert readers plan aheadas they read; they are goal-oriented; they com-bine their background knowledge with textcues to create meaning, use a variety of strate-gies (e.g., from seeking the important informa-tion in text to noting details), monitor theircomprehension, attempt to solve their compre-hension problems, and evaluate their under-standing and performance (e.g., by asking Isthe content believable? Is the piece well writ-ten? Am I achieving my goals?). The result isa personalized, interpretive understanding oftext.

At the outset of our work, little was knownabout how students are taught to become moreskillful and independent readers. A variety ofqualitative methods had been used in the de-scriptive studies of TSI (see Pressley, El-Dinary, Gaskins, et al., 1992). These includedethnographies; interviews involving questionsemanating from Pressley, Goodchild et al.'s(1989) tentative description of strategies in-struction; interviews constructed to illuminateobservations made in program classrooms;long-term case studies; and analyses of class-room discourse. Although the TSI programsdiffered in their particulars, there were anumber of common components (Pressley,El- Dinary, et al., 1992):

Strategy instruction is long-term; effec-tive strategies instructors offer it in their


I1


classrooms throughout the school year.The ideal is for high-quality processinstruction to occur across school years.

Teachers explain and model effectivecomprehension strategies. Typically, afew, r "ategies are empha-sized, including ',tried at improv-ing students' memory (e.g., associatingnew information to previously learnedinformation), comprehension (e.g.,constructing mental images to representtext content), and problem solving (e.g.,rereading sections of text that are diffi-cult to understand).

Throughout instruction, the usefulnessof strategies is emphasized; students arereminded frequently about the compre-hension gains that accompany strategyuse. Information about when and wherevarious strategies can be used is com-monly discussed.

Teachers coach students to use strategieson an as-needed basis, providing hints tostudents about potential strategic choicesthey might make. There are many spon-taneous mini-lessons about the use ofparticular strategies.

Both teachers and students demonstratethe use of strategies for one another,thinking aloud as they read. Teachersconsistently model the flexible use ofstrategies; students explain to one anoth-er how they use strategies to processtext.

The strategies are used as a vehicle forcoordinating dialogue about text. Thus,a great deal of discussion of text contentoccurs as teachers interact with stu-dents, reacting to students' use of strate-gies and prompting additional strategicprocessing (see especially Gaskins etal., 1993). In particular, when studentsrelate text to their prior knowledge,construct summaries of text meaning,visualize information covered in a text,and predict what might transpire in astory, they engage in personal interpre-tation of text, with these personal inter-pretations varying from child to childand reading group to reading group(Brown & Coy-Ogan, 1993).

Although the qualitative studies providedin-depth understanding of the nature of TSIprograms, and a variety of informal data attest-ed to the strengths of these programs (e.g.,approximations to controlled comparisonsconducted by school district officials; see

Brown & Pressley, 1994), there were noformal comparisons of students who receivedTSI with students who received more conven-tional instruction. There were several impor-tant challenges to making such comparisons.

One challenge was to determine whatshould be measured. Reading strategies in-struction has tended to focus on gains on oneor a few traditional measures of reading perfor-mance (Pressley, El-Dinary, et al., 1992). Itbecame clear during the qualitative studies,however, that TSI probably affects studentcognition in a number of ways of consequenceto teachers, with both short-term and long-term


12


effects. The cognitive benefits perceived byteachers include greater awareness of readingprocesses, increased interpretation of text,more extensive use of background knowledge,and enhanced comprehension (see Pressley,Schuder, Teachers in the Students AchievingIndependent Learning Program, Bergman, &El-Dinary, 1992)

A second challenge was that many of thepresumed effects of ,uch an intervention appearin the long-termthat is, at a minimum, onlyafter a semester or more of such instruction(see Marks et al., 1993; Pressley, El- Dinary,Gaskins, et al., 1992; Pressley, Schuder, et al.,1992). A credible evaluation, therefore, had tobe long-term. But students often move in andout of schools at a high rate; thus, holdinglarge groups of students together for severalyears was impractical. Our solution was toevaluate one year of TSI, since one year ofintervention was all we believed could becompleted in the participaring district with anintact sample of students.

A third challenge was that the randomassignment of teachers to conditions in such astudy was out of the question. Becoming aneffective transactional strategies teacher takesseveral years (e.g., El-Dinary & Schuder,i993; Pressley, Gaskins, Cunicelli, et al.,1991; Pressley, Schuder, et al., 1992). Thus,we could not take just any group of teachersand randomly assign them to TSI or to compar-ison conditions. Moreover, we could not ran-domly assign accomplished transactional strate-gies teachers to teach some other approach fora year. Our solution was to use a quasi-experi-mental design involving accomplished TSIteachers and other teachers in the same dis-

trictteachers with reputations as excellentreading educators whose instruction followedthe guidelines of the district's regular literacycurriculum.

Before proceeding with a description of theformal methods in our study, we summarizesome of the most important features of theeducator-developed approach to TSI evaluatedhere: the Students Achieving IndependentLearning (SAIL) program (Bergman & Schu-der, 1992). A description of SAIL will permitreaders to understand our expectations in thisquasi-experiment.

SAIL Comprehension Strategies Instruc-tional Program

The purpose of SAIL is to develop indepen-dent, self-regulated readers. The program wasdeveloped over the course of a decade in onemid-Atlantic school district (see Schuder,1993, for a history of SAIL). SAIL studentsare taught to adjust their reading to their spe-cific purpose and to text characteristics (e.g.,Is the material interesting? Does it relate to thereader's prior knowledge? What genre does thetext fit? How difficult is the text?). SAILstudents are instructed to predict upcomingevents (narration) or information (exposition),alter expectations as text unfolds, generate,luestions and interpretations while reading,visualize represented ideas, summarize periodi-cally, and attend selectively to the most impor-tant information. Students are taught to thinkaloud (e.g., Meichenbaum, 1977) as theypractice applying comprehension strategiesduring reading-group instruction. For example,they reveal their thinking to others when they


13


talk about their past experiences in relation totext content. All of these reading processes aretaught through direct explanations provided byteachers, teacher modeling, coaching, andguided practice, both in reading groups andindependently.

SAIL students are a!so taught methods fordealing with difficult wor,ls, including skippingthem, using context clues to determine themeaning of hard-to-decode and unfamiliarwords, and rereading for additional clues tomeaning. They are taught that getting theoverall meaning of text is more important thanunderstanding every word and that sometimes,difficult words can be skipped with little or noloss in meaninga skillful reader practice.

When SAIL instruction occurs in readinggroups, it differs in a number of ways frommore conventional reading-group instruction:(1) Pre-reading discussion of vocabulary iseliminated in favor of discussion of vocabularyin the context of reading. (2) The almost uni-versal classroom practice of asking compre-hension-assessment questions as students readin groups (e.g., Mehan, 1979) is rarely ob-served in TSI (Gaskins, Anderson, Pressley,Cunicelli, & Sat low, 1993). Instead, a teacherencourages students to construct and evaluatean interpretation of text. Teachers can gaugestudents' literal comprehension by having themthink aloud after reading a text segment. (3)There are extended interpretive discussions oftext; these discussions emphasize studentapplication of strategies to text.

Although re ....ling group is an importantinstructional context for SAIL, the teaching ofstrategies extends across the school dayduring whole-class instruction and as teachers

interact individually with their students. Read-ing instruction is also an across-the-curriculumactivity. When students read science or socialstudies texts, they are encouraged to use thesame comprehension and problem-solvingstrategies they use when reading literature. Forexample, they are asked to make predictions,activate their background knowledge, constructmental representations of the text, summarizeimportant information, and clarify confusionsby rereading or using semantic and pictureclues. A sample SAIL lesson is provided in theAppendix. The lesson highlights key featuresdescribed in this section and enables readers tocompare SAIL with other strategies instruction-al approaches (e.g., reciprocal teaching; Pa lin-csar & Brown, 1984).

Hypotheses

Three hypotheses were evaluated here: (1) thatinstruction in SAIL would enhance readingcomprehension as measured by a standardizedtest; (2) that there would be clear indications ofthis improvement after a year of SAIL instruc-tion, and (3) that students would developdeeper, more personalized and iiizerpretiveunderstandings of text after a year of SAILinstruction.

These hypotheses were evaluated withpreviously low-achieving, second-grade stu-dentsa group targeted by SAIL. SAIL wasdesigned originally for elementary students infirst or second grade who were at risk forreading failure. It is intended as a dramaticallyricher and more engaging form of instructionthan the skill-and-drill approaches so oftendelivered to at-risk students kAllington, 1991).


14

ci-qnsactional Strategies Instruction 7

Thus, the evaluation reported here involvedcontrasting the achievement of previously low-achieving, second-grade students who hadparticipated in SAIL with five matched groupsof second-grade students receiving high-quali-ty, but more conventional reading instruction.

METHODS

Participants

Teachers. The five TSI teachers and the fiveteachers in comparison classrooms served inthe same school district in which the SAIL pro-gram was developed. Eight of the teacherstaught second-grade. One SAIL teacher hadfirst- and second-graders in her class; onecomparison teacher had second- and third-graders in her class. All teachers were female.The SAIL teachers had an average of 10.4years of experience in teaching; the comparisonteachers averaged 23.4 years. The five SAILteachers exhausted the pool of second-gradeteachers in the district with extensive experi-ence teaching in the SAIL program (i.e., threeor more years; range = three to six years).The comparison teachers were recommendedby principals and district reading specialists,with nominations of effective teachers based oncriteria such as (1) giving students grade-level-appropriate tasks, (2) providing motivationallearning activities, (3)- using classroom man-agement well to avoid discipline problems, (4)fostering active student involvement in reading,(5) monitoring student understanding andperformance, and (6) fostering academic self --esteem in students. The comparison teacherswere eclectic in their instructional practices,

blending the whole- language tradition favoredin the school district with elements of skill andother traditional forms of reading instruction.The comparison- group teachers had not partic-ipated in any SAIL professional developmentactivities.

All participating teachers were admin-istered DeFord's (1985) Theoretical Orienta-tion to Reading Profile (TORP), a 28-iteminstrument discriminating among teachers whoidentify with phonics, skills, and whole-lan-guage orientations (r = .98). The scoring issuch that those favoring phonics-based readinginstruction score lower than those favoringskills instructionwho, in turn, score lowerthan those identifying with whole language(scores range from 28 to 140). The SAILteachers' mean score was 113 (SD = 9.7), andthe comparison teachers averaged 73 (SD = 7.2),with the SAIL teachers differing significantlyfrom the comparison teachers, dependent t (4)--- 6.24, p < .05. (The dependent t test wascalculated since SAIL and comparison teacherswere subsequently matched on school demo-graphical information and participating stu-dents' fall standardized test performances.When particular items of the TORP were exam-ined, it was clear that the SAIL teachers had amore whole-language orientation than the com-parison teachers, who endorsed phonics andskills more often, smallest I It (4) I = 4.88,p < .05 for any of the three subscales. Thiswas as expected, since SAIL encourages mean-ing- making as the goal of reading and discour-ages teaching of skills in isolation, which isconsistent with whole-language approaches.Informal observations of the comparison teach-ers over the year confirmed that they were


15


more eclectic in their approach to readinginstruction than the SAIL teachers, incorporat-ing a balance of whole-language, phonics-based, and skills-based instruction. Thus, theirmore balanced appraisal of the TORP itemswas consistent with our observations of theirteaching.

At the beginning of the study, the 10participating teachers were also administered a25-item, researcher-constructed questionnairetapping their beliefs about teaching (r = .94;Cronbach's alpha was calculated using partici-pants' responses). Some items were generatedafter observing SAIL and non-SAIL instructorsduring a pilot study conducted the previousyear. Other statements were adaptations ofquestions or Likert-type items developed earlierby Pressley and his associates for interviewstudies (Pressley, Schuder, et al., 1992) and bySAIL program developers for formative evalu-ation purposes. The final version of the inter-view form was reviewed by the second authorand a SAIL program developer to evaluatewhether the items would identify salient differ-ences between SAIL and non-SAIL teachers.

The questions were not direct; subjectsresponded to Likert-type statements (i.e., on astrongly agree to strongly disagree scale). Forexample, teachers who endorse TSI wereexpected to respond affirmatively to statementslike, "The most important message to conveyto students is that reading and thinking areinseparably linked," and, "During instruction,teachers should ask story-related questions thathave no precisely right or wrong answer."SAIL teachers were expected to disagree withitems such as, "Worksheets that enable studentsto practice comprehension skills can be very

useful for low-group students," and, "Duringreading instruction, teachers need to guide stu-dents toward one best interpretation of a sto-ry." The responses were scored so that consis-tency with TSI would result in a low score(maximum score = 120; one item was discard-ed). The scores of the SAIL teachers rangedfrom 25 to 45 on this scale (M = 36.8, SD = 9.5);comparison teachers' scores ranged from 62 to76 (M = 70.8, SD = 5.3), a significant differ-ence, dependent t(4) = 8.84, p < .05. Inshort, there were multiple indicators at theoutset of the study that the SAIL teachers werecommitted to a different approach to teachingthan the conventional teachers and that theSAIL teachers' beliefs about teaching wereconsistent with TSI philosophy.

Students. Student participants were as-signed to second grade but were reading belowgrade level at the beginning of the year. Theywere identified as such through informal test-ing (teacher assessments involving reading ofgraded basal passages and word lists), resultsfrom assessments administered as part of theChapter 1 program, and the previous year'sgrades and reports. Based on these criteria, sixto nine students in each class were identified aspotentially eligible to participate. Unfortu-nately, none of the assessments used by theschool district to classify readers as weak at thebeginning of the year were standardized mea-surements, although there was convergingevidence from informal measures that all

participants experienced at least some difficultyreading beginning-level, second-grade mate-rial.

SAIL and comparison classes in the studywere matched on the basis of student mobility


16


patterns, Chapter 1 status, ethnic and minoritycomposition, size and location, and perfor-mance on standardized tests. A comprehensionsubtest of the Stanford Achievement Test (SAT;Primary 1, Form J; Grade level 1.5-2.5 [ThePsychological Corporation, 1990]) was admin-istered in late November or early December(depending on the class) of the school year tothe six to nine students previously identified ineach class. Administration of this test occurredthen because only then did the teachers feel thatparticipating students could function somewhatindependently at the 1.5 grade level. Unfortu-nately, this required the administration of thetest after SAIL teachers had introduced SAILstrategies, so it was not a perfect pretest.

Fro. those considered eligible, six stu-dents in each of the paired classes were match-ed on the basis of their reading comprehensionscores (n = 60). A pair consisted of one SAILclass and one comparison class (n = 5 pairs).All of the children participating in the studyspoke and comprehended English. In addition,the sample included no children experiencingsevere attentional or behavioral problems.From first to second semester, one SAILstudent and two comparison students in onepair left their classrooms. Back-up studentswere substituted, with no significant differenceoccurring between the newly constituted groupson the fall reading comprehension subtest.

With a maximum raw test score of 40, theSAIL classes in the study averaged 22.20 on thecomprehension subtest of the SAT (SD = 6.85)at the late fall testing, and the comparisonclasses averaged 22.67 (SD = 5.89), a non-significant difference (means per class ana-lyzed), t(4) = 0.59, p > .05. Although not

used for matching, the word skills subtest fromthe same standardized instrument was alsoadministered (maximum score = 36 for thesubtest), SAIL mean = 20.97 (SD = 2.76) andcomparison mean = 21.10 (SD = 3.40),t(4) = 0.10, p > .05. The comparability ofthe paired groups is reflected in their meansand standard deviations on the fall Stanfordreading comprehension subtest (see Table 1).

Although the six children from each class-room are referred to here as a reading group,their instruction varied through the year. First,reading was most often taught in homogenousgroups, although it also occurred during indi-vidualized and whole-class instruction Second,participants did not always remain members ofthe same homogeneous group over the courseof the year (e.g., students who made greatprogress became members of another group).Since the SAIL program was offered to allchildren in the SAIL classrooms and the in-struction in comparison classrooms did notresemble SAIL instruction, variable groupingdid not pose a problem with respect to fidelityof treat,Lient. The six participating children ineach classroom did meet as a homogenousgroup for lessons that were formally analyzed,however. Even so, our use of the term "read-ing group" implies no more than the six target-ed children who received either SAIL or con-ventional instruction daily, always within theirclassrooms, frequently in small groups, andsometimes, as an intact group.

Design

This was an academic-year-long, quasi-ex-perimental study, carried out in 1991-92. The


17


Table 1. Stanford Achievement Test, Comprehension Subtest: Means and Standard Deviations forPaired Groups, Fall Administration

SAIL NOn-SAIL

M SD M SD

20.83 6.94 20.33 7.92

2 19.67 6.47 20.00 8.07

3 15.67 4.63 16.5!.3 6.24

4 21.00 6.23 24.00 6.60

5 33.83 7.28 32.17 6.88

reading achievement of five groups of sixpreviously low-achieving, second-grade stu-dents receiving SAIL instruction was comparedwith the reading achievement of five groups ofsix previously low-achieving, second-gradestudents receiving instruction typical of the dis-trict. Each of the 10 groups was housed in adifferent classroom, with each SAIL groupmatched to a comparison group that was closein reading achievement level at the beginningof the study and matched demographically tothe school providing the SAIL group. Thereading-group meannot individual studentscores was considered the unit of analysis.

A quasi-experimental design was selectedbecause it was impossible to randomly assignteachers to instructional conditions. Althoughwe might have attempted to identify potentialcomparison teachers in the building whereSAIL teachers taught and randomly assignedstudents to those teachers, we chose not to dothis for two reasons: (1) we did not want SAILand non-SAIL participants to communicatewith each other during the study; and (2) wedid not want the identification of potentialteachers limited to the same schools. Instead,we sought the most competent comparison

teachers in the district. Since these teachers didnot serve in the same buildings as the SAILteachers, random assignment of children toteachers was impossible as well. We believethe option we selected was a fair test of SAILrelative to highly regarded, more conventionalreading instruction. Thus, the standardizedtesting (to check for a priori differences) andthe matching of students was used to compen-sate for the lack of randomization. Despite thelimitation of quasi-experimental studies, webelieve this evaluation is stronger than otherclassroom-based studies that do not adopt suchcompensatory measures.

Dependent Measures

The dependent measures are described in theorder in which they were administered duringthe academic year. All measures were adminis-tered by the sam..: researcher; a summary of themeasures appears in Table 2.

Strategies interview. In October and No-vember (i.e., when SAIL instruction was beingintroduced to SAIL students) and in March andApril, a strategies interview was administeredto all students participating in the study. This


18

Tab

le I

. Des

crip

tion

of D

ata

Sour

ces

for

Stud

ents

Dat

a So

urce

Whe

n G

iven

Why

Giv

enD

escr

iptio

n

Stra

tegy

Inte

rvie

w

Stan

dard

ized

Subt

ests

:

Wor

d Sk

ills

Ret

ellin

gQ

uest

ions

Stra

tegy

Inte

rvie

w

Thi

nk-a

loud

Tas

k

Rea

ding

Com

preh

ensi

on

Wor

d Sk

ills

Oct

ober

/N

ovem

ber

To

asse

ss in

itial

ly S

AIL

and

non

-SA

IL C

las-

ses'

dec

lara

tive

and

met

acog

nitiv

e kn

owl-

edge

of

com

preh

ensi

on a

nd p

robl

em-s

olvi

ngst

rate

gies

Nov

embe

r/T

o fo

rm c

ompa

rabl

e SA

IL a

nd n

on-S

AIL

Dec

embe

rSt

anfo

rd A

chie

vem

ent T

est,

Rea

ding

gro

ups

by m

atch

ing

stud

ents

usi

ng r

eadi

ng c

ompr

e-he

nsio

n sc

ores

Nov

embe

r/T

o m

easu

re in

itial

ly S

AIL

and

non

-SA

ILD

ecem

ber

clas

ses

on a

trad

ition

al, s

tand

ardi

zed

and

valid

ated

mea

sure

of

wor

d re

cogn

ition

ski

lls

Mar

a/T

o as

sess

stu

dent

s' r

etel

ling

and

sequ

enci

ngA

pril

of tw

o st

orie

s pr

esen

ted

by e

ach

teac

her

Mar

ch/

To

com

pare

SA

IL a

nd n

on-S

AIL

cla

sses

'A

pril

know

ledg

e of

str

ateg

ies

May

/Ju

ne

May

/Ju

ne

May

/Ju

ne

To

com

pare

SA

IL a

nd n

on-S

AIL

cla

sses

'se

lf-r

egul

ated

use

of

stra

tegi

es d

urin

g st

ory

read

ing;

to d

eter

min

e if

stu

dent

s w

ere

mor

ete

xt-

or r

eade

r-ba

sed

in th

eir

resp

onse

s to

thin

k-al

oud

prom

pts

To

com

pare

SA

IL a

nd n

on-S

AIL

cla

sses

on

a tr

aditi

onal

, sta

ndar

dize

d, a

nd v

alid

ated

mea

sure

of

read

ing

com

preh

ensi

on

To

com

pare

SA

IL a

nd n

on-S

AIL

cla

sses

on

a tr

aditi

onal

, sta

ndar

dize

d, a

nd v

alid

ated

mea

sure

of

wor

d sk

ills

Sem

istr

uctu

red

inte

rvie

w c

onsi

stin

g of

fiv

eba

se q

uest

ions

that

wer

e fo

llow

ed u

p w

ithno

ndir

ectiv

e pr

ompt

s; th

e qu

estio

ns w

ere

adm

inis

tere

d or

ally

and

indi

vidu

ally

to s

tu-

dent

s.

Com

preh

ensi

on S

ubte

st, P

rim

ary

1, F

orm

J

Stan

ford

Ach

ieve

men

t Tes

t, W

ord

Stud

ySk

ills

Subt

est,

Prim

ary

1, F

orm

J

Stud

ents

indi

vidu

ally

wer

e as

ked

cued

and

pict

ure-

cued

ret

ellin

g qu

estio

ns

Stud

ents

wer

e po

sed

the

sam

e fi

ve in

terv

iew

ques

tions

they

wer

e as

ked

in th

e fa

ll du

ring

the

asse

ssm

ent

Stud

ents

wer

e st

oppe

d an

d as

ked

"Wha

t are

you

thin

king

?" a

nd o

ther

non

dire

ctiv

e fo

l-lo

w-u

p pr

ompt

s at

fou

r fi

xed

poin

ts d

urin

gst

ory

read

ing;

stu

dent

s w

ere

ques

tione

din

divi

dual

ly

Stan

ford

Ach

ieve

men

t Tes

t, R

eadi

ng C

om-

preh

ensi

on S

ubte

st, P

rim

ary

1, F

orm

K

Stan

ford

Ach

ieve

men

t Tes

t, W

ord

Stud

ySk

ills

Subt

est,

Prim

ary

1, F

orm

K

20

flr-S

T C

OP

Y A

VA

ILA

BLE


interview tapped students' knowledge aboutstrategies.

Five open-ended questions (adapted fromones used by Duffy et al., 1987) were adminis-tered orally and individually to each participat-ing student:

1. What do good readers do? What makessomeone a good reader?

2. What things do you do before you start toread a story?

3. What do you think about before you read anew story?

4. What do you do when you come to a wordyou do not know?

5. What do you do when you read somethingthat does not make sense?

These questions were presented in a differentorder for each student. If initial student re-sponses were unclear or terse, the researcherprobed for clarifications and/or elaborations.

Story lessons and recall questions. InMarch or April (depending on class schedule),two stories w: re presented to all participatingreading groups. The instruction and interac-tions that occurred during reading were record-ed on videotape and were analyzed to docu-ment differences in instruction in SAIL and incomparison reading groups. After the lesson,each student was asked to retell the story to theresearcher, followed by a task requiring sub-jects to sequence pictures corresponding toevents in the story.

All children in the study read two illustrat-ed stories. "Fox Trot" was a chapter in apopular children's trade book, Fox in Love(Marshall, 1982); "Mushroom in the Rain"(Ginsburg, 1991) was from the Heath ReadingSeries, Book Level 1. The readability for the341-word "Fox Trot" was 2.4; the readabilityfor the 512-word "Mushroom in the Rain" was2.2 (Harris-Jacobson Wide Range ReadabilityFormula; Harris & Sipay, 1985, pp. 656-673).Of the two, "Fox. Trot" was the more challeng-ing story because:. (1) it had a higher readabili-ty level; (2) it required students to make moreinferences about the characters; and (3) it usedvocabulary far less repetitively than "Mush-room in the Rain."

In "Fox Trot," the main character, Fox,decides to enter a dance contest. He asks eachof two friends to be his dance partner, but eachrefuses. They both suggest that Fox ask Raisin,but he is reluctant to do so because she is angrywith him. Nevertheless, he asks and sheagrees. They practice hard and dance welltogether, but on the day of the contest, Raisingets the mumps. Fox returns home and despon-dently sits in front of a blank television screen.Then he decides to teach his little sister thedance steps. They rush to the contest and winsecond prize.

In "Mushroom in the Rain," an ant seeksshelter from a storm. She squeezes herselfunder a small mushroom. A butterfly comes byand asks if he can escape the rain as well, andthe ant allows the butterfly to crowd in. Thencomes a mouse and a bird; crowding under themushroom increases. A rabbit arrives with afox in hot pursuit. The others hide the rabbitunder the mushroom. Once the fox leaves andthe rain stops, the ant asks the others how they


21


managed to fit under the mushroom. A frog,sitting on top of the mushroom asks, "Don'tyou know what happens to a mushroom in therain?" In the version of the story used in thestudy, the answer was not provided to thechildren but was left for them to infer.

These stories were selected because theyprovided ample opportunity for diverse inter-pretations and personal responses. They wereon the school system's approved list and werejudged by the participating teachers as appro-priate for a single lesson for weaker studentslate in the academic year. Three of the matchedpairs of reading groupsone SAIL group andone comparison group to a pairread "Mush-room in the Rain" first; two pairs read "FoxTrot" first.

All decisions about how to teach the storieswere made by the teachers. However, theywere asked to teach each of these stories in onemorning lesson, not to exceed 55 minutes inlength. The mean SAIL lesson lasted 43.40 min-utes (SD = 7.83) and the mean comparison-group lesson lasted 35.50 minutes (SD = 13.34).The lessons were videotaped in order to verifythat teaching in the SAIL groups was differentfrom teaching in the comparison readinggroups.

Approximately two hours after each lesson,each of the six students in the reading groupwas interviewed individually. First, studentswere asked to retell the story:

Pretend that you are asked to tell the story

to other kids in the class who have neverheard the story before. What would youtell them happened in that story? . . . Can

you remember anything else? (Adaptedfrom Golden, 1988)

This was followed by a cued, picture-retellingtask. Students were asked to sequence sixscrambled pictures taken directly from thestory. The students were then informed thatsometimes pictures assist in aiding recall ofstories, and they were asked to use the picturesto prompt recollection of story content.

Think-aloud measures. In May/June,students read a 129-word illustrated Aesop'sfable, "The Dog and His Reflection," selectedfrom a trade book (Miller, 1976). The read-ability was 3.9 (Harris & Sipay, 1985) for thisstory, and thus it was challenging for the stu-dents. Although a text with such a high-read-ability level may have been too taxing for somestudents, the accompanying pictures werehighly informative. In addition, challengingmaterial was needed to trigger the students' useof strategies.

In the story, a dog steals a piece of meatfrom the dinner table. He runs into the woodsand starts to cross a bridge. When he chancesto look down, he sees his reflection in thewater. Thinking his reflection is another dogwith a larger cut of meat, he decides to seizethe other dog's chop. When he opens hismouth, his own piece of meat plunges into thewater. Consequently, the dog ends up withnothing at all.

The students met with the researcherindividually for this task. Students werestopped four times during the reading of thefable and asked to report their thinking. If astudent had difficulty reading a segment, thefirst question posed was, What do you thinkhappened on this page? Otherwise, the onlyquestion posed was, What are you thinking?When students offered unelaborated responsesto these questions, open-ended follow-ups (see


22

14 Racnel Brown, Michael Pressley, Peggy Van Meter and Ted Schuder

Garner, 1988, p. 70) were asked, such as, Canyou tell me more? or Why do you say that?Sometimes an unelaborated comment wasechoed back to the student in the form of aquestion. For example, after a student re-marked that a dog stole a piece of meat fromhis master's table, the researcher might haveasked, What do you think about the fact thatthe dog stole a piece of meat from his master'stable? For every text segment, before thestudent moved on to reading the next segment,the researcher asked, "Is there anything youcould say or do before reading on?"

Stanford Achievement Test subtests. InMay/June, students took the Stanford Achieve-ment Test (The Psychological Corporation,1990), Form K, Reading Comprehension andWord Study Skills subtests. The ReadingComprehension subtest consists of two-sen-tence stories, comprehension questions aboutshort passages, and sentence completion itemsthat form short stories. The Word Skills subtestincludes questions pertaining to structuralanalysis (i.e., compound words, inflectionalendings, contractions) and phonetic analysis(i.e., consonants and vowels). The comprehen-sion test was administered first to all studentsfollowed by the word skills test. The alternate-forms reliability for the full-scale scores ofForms J (administered in the fall) and K was.89.

RESULTS

Each hypothesis reported here was one-tailed.That is, each evaluated whether SAIL instruc-tion produced better performance than thecomparison instruction. Most means appeared

in only one hypothesis test and hence, p < .05was the Type 1 error rate selected for all hy-potheses (see Kirk, 1982, for this and all refer-ences to statistics). For the standardized testdata and strategies inter, *ew data, the simpleeffect of condition was tested in th3 fall, andagain in the spring. The Time (of testing) xCondition interaction was also tested. With eachhypothesis tested at Type 1 error rate p < .05,the overall Type 1 error rate for the analysis ofthe three dependent variables (i.e., the stan-dardized comprehension, standardized vocabu-lary, and strategy interview) did not exceed.15. The same overall Type 1 error rate wouldhave applied for each dependent variable if wehad analyzed the data with a 2 x 2 analysis ofvariance model. Analysis of variance was notconsidered here since the effects it tests werenot precisely matched to the important hypoth-eses in this investigation.

All tests were based on the reading-groupmean as the unit of analysis (i.e., n = 5 groupsfor the SAIL condition and n = 5 ,:--oups forthe comparison condition), since in, ;dual

scores within reading groups were not indepen-dent. Finally, all t tests were dependent t testsbased on the five matched pairs, with one SAILand oile comparison group to a pair; pairingswere determined by the reading groups' fallstandardized comprehension performances. Thus,the cutoff value for determining statisticalsignificance for every comparison reported inthis results section was t(4) = 2.13, p < .05(one-tailed).

Fall/spring strategies interview. The inter-views were designed to determine whetherSAIL and comparison students would differ inthe number of strategies they claimed to use


23


during reading. Two raters scored 20% of theinterviews, with an overall 87% agreement forthe strategies named by students. Differenceswere negotiated and resolved. Only one of thetwo raters scored the remainder of the inter-views.

A strategy was scored as mentioned if itwas named in response to any of the interviewquestions. Any strategies mentioned by stu-dents were recorded, even if they were notstrategies taught in the SAIL program. Thecomprehension strategies mentioned includedthe following:

Predicting: Guessing what will happennext

Verifying: Confirming that a predictionwas supported by text, background knowl-edge, or reason

Visualizing: Constructing a mental pictureof the information contained in the textsegment

Relating prior knowledge/personal experi-ences to text: Making an association be-tween information in the text and infor-mation in the reader's head

Summarizing/retelling: Saying the mostimportant information (summarizing) orrestating in ,ne's own words everythingthat occurred in the text segment just read

Thinking aloud: Verbalizing thoughts andfeeling about text segments just read

Monitoring: Explicitly verbalizing whensomething just read does not make sense

Setting a goal: Deciding a purpose prior toreading, including decisions about bothexpository and narrative texts

Browsing/previewing: Flipping through thestory, glancing at the pictures, or readingthe back cover to get ideas about the story

Skipping: Ignoring a problematic part oftext and reading on

Substituting/guessing: Replacing a difficultpart of text with something else that seemsto make sense and maintains the coherenceof the text segment

Rereading: Returning to a problematic seg-ment of text

Looking back: Looking back in the text forinformation that might help in understand-ing a difficult part of text

Clarifying confusions: Asking a specificquestion to resolve a comprehension prob-lem

Asking someone for help: Asking anotherstudent or the teacher for help with aconfusing section of text

The following strategies for attackingunknown or difficult words were men-tioned:


'1

16 Rachel Brown, Michael. Pressley, Peggy Van Meter and Ted Schuder

Table 3. Means and Standard Deviations for Number of Comprehension and Word-Level StrategiesMentioned in the Fall and Spring Strategies Interviews

Comprehension Strategies

Fall Spring

SAIL Comparison Group SAIL Comparison Group

M SD M SD

0.80 0.45 0.66 0.44

(SAIL not significantly greater, t(4) = 0.65)

M SD M SD

4.20 0.86 1.21 0.40

(SAIL significantly greater, t(4) = 9.73)

Word-Level Strategies

Fall Spring

SAIL Comparison Group SAIL Comparison Group

M SD M SD

2.12 0.74 1.15 0.28


M SD M SD

3.22 0.63 1.60 0.. ;


Skipping: Ignoring a problematic word andreading on

Substituting/guessing: Replacing an un-known word with another word that ap-pears to make sense or that maintains thecoherence of the text

Rereading: Returning to a problematicword in context

Looking back: Looking back in the text forinformation that might help in understand-ing a difficult word

Using picture clues: Looking at pictures inthe story to help determine the meaning ofan unknown word or difficult piece of text

Using word clues: Relying on the surroun-ding text to help decide the meaning of anunknown word or difficult piece of text

Breaking a word into parts: Seeing if thereare recognizable root words, prefixes, orsuffixes contained within the larger word

Sounding out a word: Applying knowledgeof phonics to the decoding of the word

NATIONAL READING RESEARCH CEN PER, READING RESEARCH REPORT NO. 33

25


Table 4. Strategy Interview Analysis: Baseline and Postmeasure Mean Frequencies for ComprehensionStrategies Mentioned or Described During Interviews, by Group

Strategy

Baseline Measure Postmeasure

SAIL M Non-SAIL M SAIL M Non-SAIL M

Predict 0.84 0.37 4.30 1.84Reread 1.67 0.54 2.56 1.34Skip/keep going 0.17 0.57 2.47 0.83Visualize 0.00 0.17 1.77 0.00Ask someone for help 0.51 1.46 1.60 1.00Relate to prior knowledge/

personal experiences 0.34 0.00 1.57 0.17Summarize/retell 0.00 0.00 1.20 0.00Set a goal 0.17 0.00 1.00 0.00Think aloud 0.17 0.00 0.87 0.00Clarify confusions using

text or picture clues 0.i7 0.00 0.83 0.17Substitute/guess 0.00 0.20 0.67 0.51Browse 0.00 0.00 0.67 0.00Verify 0.00 0.00 0.54 0.00Look back 0.00 0.00 0.53 0.00Monitor/say something

doesn't make sense 0.00 0.00 0.40 0.33

Note. Frequencies were calculated by summing class means for each group (SAIL, non-SAIL) on each strategy.

Asking someone for help: Asking anotherstudent or the teacher for help with theconfusing word

The comprehension and word-level strategiesreported are summarized in Table 3. Themeans reported in the table are reading-groupmeans. That is, a mean frequency of strategiesreported for each reading group in the studywas calculated based on individual readinggroup members' reports. The Table 1 meansand standard deviations were based on fivereading-group means. With respect to reports

of comprehension strategies, there was nosignificant advantage for the SAIL students inthe fall, shortly after the program had begun.By spring, however, as expected, the SAILgroups reported many more strategies than thecomparison groups (see Table 4). In thespring, only SAIL students talked about visual-izing, looking back, verifying predictions,thinking alOud, summarizing, setting a goal, orbrowsing. Althoughcomparison-group studentsmentioned predicting, using text or pictureclues to clarify confusions, making connectionsbetween text and their background knowledge


26


Table 5. Strategy interview Analysis: Baseline and Postmeasure Mean Frequencies for Word StrategiesMentioned or Described During Interviews, by Group

Strategy

Baseline Measure

SAIL M Non-SAIL M SAIL M

Postmeasure

Non-SAIL M

Skip/ignore 3.99 0.94 4.33 1.34Substitute/guess 1.67 0.00 2.64 0.17Sound it out 1.83 2.70 2.16 2.85Use picture clues 0.66 0.20 2.10 0.00Reread 0.83 0.00 1.87 0.84Look back 0.00 0.00 1.27 0.17Break into parts/

word clues 0.34 0.17 1.18 0.00Ask someone for help 0.34 1.56 0.34 2.00Skip and return/

use context 1.01 0.20 0.17 0.67

Note. Frequencies were calculated by summing class means for each group (SAIL, non-SAIL) on each strategy.

and experiences, asking someone for help,skipping over confusing parts, and rereading,during the spring interview, the frequency ofsuch reports was always greater for the SAILgroup than for the comparison grew. TheSAIL and comparison groups mentioned moni-toring and guessing approximately equally onthe spring interview.

With respect to word-level strategies, theSAIL students reported more strategies than thecomparison-group participants, even during thefall interview. In the fall, SAIL students men-tioned skipping words, substituting/guessing,using picture or word clues, rereading, andbreaking words into parts descriptively moreoften than did comparison students (see Table5). There was more mention of sounding outwords in the comparison condition in the fall.The introduction to SAIL in the fall monthsprobably accounts for this fall difference inword-level strategies reports. By the spring, all

of thf, word-level strategies were being men-tioned by SAIL students (see Table 5). Incontrast: the word-level strategies mentionedconsistently by more than one student percomparison reading group were skipping anunknown word, sounding it out, rereading,using context clues, and asking someone forhelp.

We also tested whether SAIL studentsmade greater gains in self-reported knowl-edge of strategies over the course of theyear. The one-tailed interaction hypothesistest (e.g., fall-to-spring increase in students'strategies scores by condition) was signifi-cant as expected for both the comprehensionstrategies, t(4) = 7.49, and the word-levelstrategies, 1(4) = 4.30.

In summary, by spring the SAIL studentsreported more comprehension and word-levelstrategies during the interview than did com-parison-group students. That SAIL students


27


were already reporting more word-level strate-

gies in the fall than comparison students proba-

bly reflected the effects of the first month or

two of instruction in the program. Althoughstudents were introduced to a few comprehen-sion strategies prior to the fall interview, such

as predicting and thinking aloud, teachersheavily emphasized the fix-up word strategies

to provide their weaker readers with tools they

could apply to text. By spring, all but twostrategies were mentioned more often in the

SAIL group than in the comparison group. The

exceptions were sounding it out (which wasconsistent with the teaching philosophy of thecomparison teachers) and asking for help witha word (which is difficult to construe as astrategy associated with independence in read-

ing). Most importantly, SAIL students learned

more about comprehension and word-levelstrategies over the year than comparison stu-dents.

Spring Story Lessons

Teaching the lessons. The March/April lessonswere transcribed from the videotape records,

and four raters read the transcriptions. One

rater was a SAIL program developer, and the

other three were graduate students familiar

with TSI and with the SAIL program in partic-

ular. The program developer correctly classi-

fied 9 of the 10 SAIL lessons as consistentwith

the intent and purpose of the SAIL program.This rater was sensitive to whether teachersexplained and modeled strategic processes for

students and encouraged the interpretive con-struction of text meaning through the use ofcomprehension strategies. The program devel-oper looked for evidence that the teachers

thought aloud in their lessons and coached

students to engage text actively (i.e., relate textcontent to prior knowledge as well as applyother strategies as appropriate). He classified

all of the comparison lessons as inconsistent

with the SAIL approach and, in fact, they were

not even close to being consistent with SAIL.

The three graduate students correctly classified

lessons as SAIL or non-SAIL for 59 of the 60

ratings made. Thus, there were clear instruc-

tional differences between the SAIL and non-

SAIL classrooms during the March/April les-

sons.Two raters reviewed the lessons for evi-

dence of strategies teaching, with interrater

agreement of 85 % and disagreements resolvedby discussion. Collapsing across the two les-

sons observed for each teacher, a mean of 9.20

(SD = 1.92) different comprehension strate-gies were observed in the SAIL lessons com-

pared to a mean of 2.00 (SD = 0.71) in the

comparison lessons, t(4) = 7.43. Predicting,

relating text to background knowledge, sum-marizing, and thinking aloud were observed in

all SAIL groups. Only relating to backgroundknowledge was observed in all comparison

groups. In no SAIL group were fewer than

seven of the comprehension strategies taught;

in no comparison group were more than three

observed.On average, again collapsing across each

participating reading groups' two lessons, 4.80

(SD = 0.45) word-level strategies were ob-

served in the SAIL groups and 4.00 (SD = 0.71)

were documented in the comparison reading

groups, t(4) = 4.00. Using semantic context

clues and picture clues were observed in all

SAIL groups; using picture clues and sound-

ing words out were observed in all compari-


28


son classrooms. The range of word-levelstrategies was between 4 and, 5 in the SAILgroups and between 3 and 5 in the comparisongroups.

Thus, one important indicator that theinstruction in the SAIL groups differed fromcomparison instruction was that there was morestrategies instruction in the SAIL groups. Thedifference was much more striking with respectto comprehension strategies, however.

Student recall of lesson stories. Both "FoxTrot" and "Mushroom in the Rain" wereparsed into idea units, a variant of the T-unit(Hunt, 1965). Loosely defined, an idea unit isa segment of written or oral discourse thatconveys meaning, consisting of a verb formwith its associated subject, object, and/ormodifier(s). Length or grammatical structuredo not determine whether a segment is codedas an idea unit; what counts is whether the unitis meaningful. Interrater agreement was calcu-lated for 20 % of the recalled stories. It was89% for classification of the protocols into ideaunits of various types (e.g., literal, interpre-tive)) Differences in classification were re-solved through negotiation.

A first issue addressed was whether SAILstudents recalled more interpretive idea units

'The recall protocols were analyzed using a modifiedanalytic induction approach (Goetz & LeCompte, 1984).

That is, coding categories emerged from analysis of thedata. However, identification of categories also washighly informed by the work of O'Flahavan (1989) andEeds and Wells (1989). In this study, only the results ofthe literal and interpretive analyses were presentedbecause they directly related to the stated hypotheses. The

full categorization scheme and analysis can be four) inBrown (1995).

than comparison students. These responsesreflected students' relating background knowl-edge to text. Interpretive ideas were not ex-plicitly stated in the text or in the pictures butdid not contradict information in the text orpictures. For instance, for the Mushroomstory, "He wanted to be dry," was scored as aninterpretive remark. (The text had said, "Oneday an ant was caught in the rain. 'Where canI hide?' he wondered. He saw a little mush-room peeking out of the ground in a clearingand he hid under it.") Also, the comment, "Butthey tricked him," was scored as an interpre-tive unit for the Mushroom story. (The corres-ponding text was, "How could a rabbit get inhere? Don't you see there isn't any room," saidthe ant. The fox turned up his nose. He flickedhis tail and ran off.") As a third example, onenot corresponding to any exact part of theMushroom story, the remark, "And it was theonly place to keep him dry, " was coded as aninterpretive remark because it was an inferencethat did not contradict anything in the text.

For the Mushroom story, SAIL groupsaveraged 6.12 interpretive units per student(SD = 1.54), which exceeded the correspond-ing figure of 4.44 in the comparison groups(SD = 1.68), t(4) = 3.03. For "Fox Trot,"SAIL groups averaged 5.58 interpretive unitsper student (SD = 1.63), which exceeded thecorresponding figure of 3.78 in the comparisongroups (SD = 1.53), t(4) = 3.13.

In addition to scoring interpretive recall,the literal recall of ideas represented either inthe stories or in the accompanying pictures wasevaluated. For example, one idea unit repre-sented explicitly in the Mushroom story was,"He hid under it." If the student recalled thisidea unit or a paraphrase of it, the student was



scored as having recalled the unit. In "FoxTrot," there was a picture of Carmen andDexter looking through a window, watchingFox dance. One idea unit was scored as re-called if the student reported something like,His friends were looking from the window,watching him dance.

For the Mushroom story, SAIL readinggroups recalled an average of 17.64 (out of amaximum of 79) literal idea units per student(SD = 3.95), which did not exceed literalrecall in the comparison groups who averaged16.48 units (SD = 1.79), t(4) = 1.11. For"Fox Trot," however, SAIL recall (M = 12.26out of a maximum of 59 units; SD = 2.72)exceeded comparison group recall (M = 8.22,SD = 2.88), t(4) = 2.66.

Whether students recalled story events intheir order of occurrence was assessed as well.A student was scored 0 if their retelling unitswere recalled in order, collapsing across bothunprompted retelling and picture-cued retell-ing. A student was penalized one point forevery retelling unit remembered out of se-quence, collapsing act:_,ss the unprompted andpicture-cued retellings. (Again, 20% of thedata were checked for interrater agreement;there was 92 % agreement). For Mushroom, themean SAIL-group mean was 0.35 (SD = 0.27)and the mean-comparison group mean was 0.94(SD = 0.50). For "Fox Trot," the mean SAIL-group mean was 0.41 (SD = 0.18) and the meancomparison-group mean was 0.40 (SD = 0.19).In neither case was SAIL sequencingbetter than comparison-group sequencing,

t(4) I = 1.81 and I t(4) = 0.05, p > .05,respectively.

In summary, SAIL students were signifi-cantly more interpretive in their recalls than the

comparison students, consistent with our ex-pectations. There had not been strong expecta-tions about the literal recall of the stories basedon condition, for we recognized that the com-parison teachers addressed the literal content ofstories very well in their lessons. Even so, thestudents in the SAIL groups recalled moreliteral information than students in the compar-ison groups. The difference favoring the SAILstudents was significant for the more difficultstory. This finding suggests that SAIL studentperformance may exceed comparison perform-ance when texts become challenging. Althoughthere was a trend favoring the SAIL studentsfor memory of the sequence of events for onestory, the sequencing means for the other storywere almost identical.

The SAIL story lessons were longer onaverage than the comparison-group lessons,perhaps accounting in part for why significantresults were obtained with respect to recall ofstories presented during reading group. How-ever, generally, SAIL lessons are longer be-cause negotiating interpretations, explainingand modeling strategies, coaching, thinkingaloud, and selecting and using fix-up strategieswhile reading are time-consuming activities,particularly when they are compared with someactivities in conventional reading lessons (e.g.,answering skill-and-drill and literal comprehen-sion questions). Thus, the total time differencebetween conditions reflects the typical totaltime difference in lessons between SAIL andnon-SAIL instruction.

Spring think-aloud analysis. The think-aloud protocols generated by each student inreaction to Aesop's fable about the dog and hisreflection were transcribed and analyzed usinga modified analytic induction approach (Goetz


30


& LeCompte, 1984). Two raters read throughall of the protocols, independently taking notesand identifying potential categories of reportedreading processes. Through negotiation, atentative set of process categories was identi-fied, with these then applied by both ratersindependently to two protocols, one from aSAIL student and one from a comparison-group student The two raters then met andrefined the categories in light of the difficultiesexperienced in scoring these two protocols.The refined categorization was applied toanother pair of protocols, again independentlyby both raters. The refined categorizationscaptured all of the processes represented inthese protocols, and thus, this set of processeswas used to code all of the think-aloud proto-cols. Thus, protocols were coded using thefollowing three categories: (1) strategy-basedresponding; (2) reader-based responding; or (3)text-based responding.

A response with any indication of strategyuse was coded as "strategy-based." The specif-ic strategies used were also coded using thestrategy definitions from the strategies inter-view, with 89% agreement between two raterson 20% of the protocols on these codings ofspecific strategies. Differences were discussedand subsequently resolved. The mean numberof strategies evidenced by SAIL reading-groupmembers (averaging across all groups) was6.93 (SD = 1.46). The corresponding compari-son-group mean was 3.13 (SD = 1.09). TheSAIL readers applied significantly more strate-gies during the think-aloud task than did thecomparison-group students, t(4) = 11.07. Infact, there was no overlap in the group means,with SAIL -group means ranging from 5.00 to

8.67 strategies used per student and corre-sponding comparison-group means rangingfrom 2.00 to 4.83. All strategies that werescored, except for one (monitoring), wereobserved descriptively more frequently in theSAIL than in the comparison protocols. Thestrategies that occurred in the SAIL condition,from most to least frequent, were as follows:prediction, relating text to prior knowledge,thinking aloud, sunz;;:arizing, using pictureclues, verifying, seeking clarification, monitor-ing, looking back, visualizing, and setting agoal. The corresponding order for the compar-ison condition was: predicting, using pictureclues, verifying, relating text to prior knowl-edge, monitoring, seeking clarification, think-ing aloud, and looking back. Visualizing,

summarizing, and setting a goal never occurredin the comparisongroup think-alou is.

In this example of a strategies-based re-sponse, a student reads a page about the dogrushing out of the house with a piece of meat.The SAIL student (S) begins to talk evenbefore the researcher (R) has asked what thestudent was thinking.

S: I think my prediction is coming out right.

R: Why do you say that?

S: Cuz, cuz I see a bridge over there andwater.

The student spontaneously verifies a pre-diction that he has made earlier by using pic-ture clues. In another example, a differentSAIL student has just finished reading that thedog carried the meat over the bridge. The dog



sees his reflection. Thinking it is another dogwith an even bigger piece of meat, re wants tohave the other dog's meat too:

R: What are you thinking?

S: I'm thinking that he's gonna, uhm, try andget that lamb chop and when he opens hismouth it's gonna fall into the water.

R: Why do you say that?

S: Because it says back here that he's beinggreedy and he wants that, too.

R: But what makes you think he's gonna openhis mouth and it's gonna fall in the water?

S: Well, because, uhm, he has to open hismouth to get the other one.

In this instance, the student makes a pre-diction using both text clues and backgroundknowledge for support.

We also examined whether SAIL or com-parison groups focused more on text- or read-er-based information when they did not re-spond strategically. Responses not classified as"strategies-based" were coded as either "text -based" or "reader-based." (Interrater agreementon 20% of the protocols for classifying text- orreader-based responses was 94% .) Text-basedresponses contained information explicitlystated or pictured in the story. Reader-basedresponses reflected a connection between thestory and a student's prior knowledge, experi-ences, beliefs, or feelings.

Proportions were calculated for each classindicating the relationship of text- and reader-

based responses to the total number of respons-es. From these class proportions, SAIL andcomparison-group means were computed. Themean for reader-based responses for the SAILgroup was .74, SD = .10 (the text-based meanproportion was 1 .74 = .26, SD = .10). Themean proportion of reader-based responses forthe comparison group was .48, SD = 0.15.Thus, the SAIL group produced more reader-based responses than the comparisoh group,t(4) = 3.98. Without exception, all SAILclasses were proportionally more interpretivethan literal in their nonstrategies-based re-sponses. In contrast, only two of five compari-son classes were proportionally more interpre-tive in their responses.

In summary, the SAIL students used strate-gies on their own more than the comparisonstudents. In addition, the results of the think-aloud analysis supported the recall analyses.That is, SAIL students made significantly morereader-based remarks than comparison stu-dents. The SAIL students responded moreinterpretively, as well as personally.

Spring Standardized Test Performance

In May/June, the SAIL students outperformedthe comparison students on the 40-item com-prehension subtest. The mean reading-groupraw score mean in the SAIL condition was34.20 (SD = 2.65); the corresponding compar-ison-group mean was 28.73 (SD = 3.77), t(4)= 4.02 (effect size[es] = 2.54; Cohen, 1988).The SAIL students also outperformed thecomparison students on the 36-item word skillssubtest in the spring, t(4) = 3.98 (es = 2.51):The mean reading-group raw score mean in the


32


SAIL condition was 27.10 (SD = 2.19); thecorresponding comparison-group mean was24.00 (SD = 1.53).

One of the most striking aspects of thespring standardized test data was the muchlower variability within SAIL groups than incomparison groups. (The exacting matching ofthe reading groups in the fall was with respectto both mean performance and variability onstandardized reading comprehension, and thus,there was little difference in SAIL and compar-ison-group variabilities in the fall.) This lowervariability is obvious from examination of thestandard deviations for each matched pair ofreading groups on the spring comprehensionsubtest:

SAIL GroupSD

Comparison GroupSD

#1 4.36 #1 9.28#2 6.26 #2 7.81#3 1.63 #3 9.94#4 3.72 #4 8.12#5 2.40 #5 4.22

The same trend occurred in the word skillsdata, although it was not quite as pronounced:

SAIL GroupSD

Comparison GroupSD

41 4.05 #1 5.75#2 4.26 #2 5.24#3 2.80 #3 6.01#4 2.43 #4 5.73#5 5.79 #5 5.60

We believed that an especially strongdemonstration of the efficacy of the SAILprogram would be greater gains on standard-ized measures over the course of the academicyear in SAIL versus the comparison condition(see the summary in the methods section of thefall data). Thus, we tested the size of the fall-to-spring increase in raw scores in the SAILgroups versus the comparison groups. Thisone-tailed interaction hypothesis test was sig-nificant as anticipated for the comprehensionsubtest, t(4) = 3.70 (es = 2.34). The word skillssubtest was significant as well, t(4) = 5.41(es = 3.42).

In one of the matched pairs, there weresome perfect scores on the comprehensionposttest. For this. pair of reading groups, aversion of the next level of the Stanford com-prehension subtest (Primary 2, Form J) wasadministered. Consistent with the analysesreported earlier, the SAIL-group mean wasgreater than the matched comparison-groupmean, and the SAIL group standard deviationwas lower than the comparison-group standarddeviation: SAIL mean = 29.8, SD = 5.42;comparison-group mean = 21.8, SD = 10.17.

In summary, by academic year's enu, thesecond-grade SAIL students clearly outper-formed the comparison-group students, withgreater improvement on the standardizedmeasures over the course of the academic yearin the SAIL condition. Gains in comprehensionwere expected because, more than mythingelse, SAIL is intended to increase students'understanding of text. The analogous effects onstudents' word skills performance was more ofa surprise, for we knew that all teachers,regardless of condition, taught phonics and

NATIONAL READING RESEARCH CENTFR, READING RESEARCH REPORT NO. 33

33


word-attack skills, albeit at different times ofday (e.g., integrated into various content areas)and in different ways (e.g., addressed in theform of worksheets, mini-lessons, etc.).

DISCUSSION

We observed many differences in instructionbetween SAIL and comparison classes through-out the 1991-92 school year. The differenceswere apparent in the two lessons that wereanalyzed in the spring. Neither a SAIL pro-gram developer nor several graduate students

who were familiar with TSI had difficultydiscriminating between transcripts of SAIL andnon-SAIL lessons. One important differencehighlighted in the analysis of the spring lessonswas that discussion of strategies was muchmore prominent in the SAIL groups than incomparison reading groups. That the differenc-es in teaching were so clear bolsters our confi-

dence in this study as a valid assessment of theefficacy of SAIL with at-risk, second-gradechildren.

SAIL had positive short-term and long-term effects. In the short-term, students ac-quired more information from stories read inreading group and developed a richer, morepersonalized understanding of the stories.Whether the focus is on the amount of literalinformation recalled from stories covered inreading group or student interpretations of thetexts read, these data indicate superior perfor-mance by SAIL students versus the comparisonstudents. We infer that SAIL students learnmore daily from their reading-group lessons

than do students receiving more conventionalinstruction.

SAIL had long-term effects as well. Con-sistent with our expectations, the SAIL studentswere much more conversant about strategicprocesses by the end of the year than werecomparison students. Consistent with our beliefthat such strategic awareness was developed bythe SAIL curricular experiences, strategicawareness increased over the school yearsignificantly more in the SAIL than in thecomparison students. SAIL students also usedstrategies more than did the comparison stu-dents, as reflected by their self-reported cogni-tive processing as they read Aesop's fable atthe end of the year. The standardized testperformances of the SAIL students also weresuperior to the comparison students at the endof the year. Most critically, there was signifi-cantly greater improvement on standardizedmeasures of reading comprehension from fallto spring in the SAIL versus the comparisonclassrooms. In short, all measurements ofstudent reading achievement reported hereconverged on the conclusion that a year ofSAIL instruction improves the reading ofat-risk, second-grade students.

The study reported here is the strongestformal evidence to date that TSI improves thereading of elementary students. There weremany elements controlled in this study thatvaried freely in more informal comparisons ofSAIL and alternative instruction such as thosegenerated by the school district where theprogram was developed: (1) The student partic-ipants were carefully matched in this investiga-tion so that there was no striking difference intheir standardized reading achievement at theoutset of the study. (2) The teachers were care-fully selected. From years of observing and


3 4


interviewing committed SAIL teachers, weknew that they were excellent teachers in gen-eral, who offer rich language arts experiencesfor their students. Thus, it was imperative thata compelling evaluation of SAIL be in compar-ison with excellent, conventional grade-2 in-struction. Accordingly, we sought comparisonteachers who were highly regarded by districtadministrators and reading consultants in adistrict that has garnered numerous nationalawards for excellence in instruction. (3) Thelessons analyzed in the TSI and comparisongroups involved the groups processing thesame stories. (4) The same dependent measureswere administered by the same tester so thatmeasurement experiences were equivalent forparticipants.

Another strength of this validation was thatit relied on multiple assessments of students'reading. Despite current criticisms of standard-ized tests, we administered one because stan-dardized tests traditionally have been used toevaluate gains in students' reading perform-ance. However, we included assessments ofchildren's memories and interpretations ofstories read in class because we felt they re-flected better the day-to-day comprehensiondemands on students than do standardizedmeasures. Although thinking-aloud measuresare far from perfect indicators of thinking(Ericsson & Simon, 1980), the assessments ofchildren's thinking as they read Aesop's fablearguably tapped more directly the thinkingprocesses of the children than did the standard-ized assessments. Thus, we gave students notonly a traditional measure of reading believedto be less sensitive to changes in strategicprocessing but also more process-oriented

measures designed to assess strategic process-ing. Taken together, the results gave a morecomplete picture of students' reading capabili-ties than if either type of assessment had beengiven alone.

Are the outcomes reported here generallysignificant beyond the specifics of the SAILprogram? As we argued earlier (see also Press-ley, El-Dinary, Gaskins, et al., 1992), SAIL isan example of reading comprehension strate-gies instruction as adapted by educators. Long-term, direct explanation and the scaffoldedpractice of a manageable repertoire of powerfulcomprehension strategies is an approach repli-cated in a number of settings (see Pressley &El-Dinary, 1993, for examples). The practicehas raced ahead of the science, however, withthe educator-developed adaptations being moreambitious in scope, more complex, and ulti-mately very different from the researcher-validated interventions (e.g., reciprocal teach-ing) that inspired the educator efforts. There isa very real need to evaluate such adaptations,for there is no guarantee that the strategiesinstruction validated in basic research studies iseffective once it is translated and transformeddramatically by educators.

The research reported here contrasts withtrue experimental research on strategies in-struction in a number of ways. First, the inter-vention studied here had several componentsand was nonanalytical with respect to compo-nents of the intervention. In other words, wedid not attempt to isolate the unique effect ofeach SAIL component on students' readingperformance. In contrast, basic strategiesinstruction research typically has been muchmore analytical. Therefore, from a true exper-


35


imental stance, a multicomponent analysismight introduce confounding. However, wecan defend the evaluation of an entire TSIpackage for two reasons. For one, it is the unitof instruction that interests us. We wanted toknow whether or not an instructional packageas a whole works, so a study evaluating thatwhole relative to other, comparable instructionmade sense. Furthermore, we were interestedin studying naturalistic, classroom-based in-struction. In real-life classrooms, multipletreatment interventions naturally occur.

Second, the program of research thatincludes this study is a blend of qualitative andquantitative research. In contrast, most basicstudies of strategies have been only quantitativein nature. We are certain that the quantitativestudy reported here would have been impossi-ble without the three years of qualitative re-search leading up to it. At a minimum, thatqualitative research affected the selection ofdependent measure; and the decision to studyonly accomplished SAIL teachers (see Press-ley, Schuder, et al., 1992). More generally, itmade obvious to us the scope of an investiga-tion necessary to evaluate TSI so that thetreatment was not compromised by the evalua-tion.

Third, most basic strategies res rch isdesigned and conducted by researchers. Wheneducators have participated in basic studies, ithas been as delivery agents only. In the pro-gram of TSI research, researchers, programdevelopers, and teachers have combined theirtalents to produce a body of research thatrealistically depicts TSI and evaluates it fairly.As the study was designed and as it unfolded,school-based educators were consulted fre-

quently about the appropriateness of potentialdependent measures and operations of thestudy. The result has been a much more com-plete and compelling set of descriptions of TSI,and now we have a thorough appraisal of theimpact of one TSI program on second-grade at-risk readers.

One potential alternative interpretation ofthis study might be that the results observedhere reflect the effects of holistic teaching,since SAIL teachers embrace the values ofwhole language, and SAIL is rich in authenticliterature experiences. In fact, the results inthis study support claims made in previousqualitative studies of TSI that whole-language-based instruction and TSI share several attrib-utes (Pressley, El-Dinary, et al., 1992; Press-ley, Schuder, et al., 1992). However, TSI isnot synonymous with whole-language-basedinstruction. Although SAIL teachers identifiedthemselves more closely with whole-languagethan with skills or phonics approaches on theTORP, they did not unequivocally endorsewhole-language instruction (see the TORPresults). Instead, they perceived some keydistinctions between TSI and typical whole-language instruction. These differences can besummarized as follows:

There are . . . some important differ-ences between whole language andtransactional strategies instruction, notthe least of which is that whole lan-guage is more psycholinguistic andphilosophical . . . than cognitive. Trans-actional strategies instruction is direct-ly interventionist, emphasizing model-ing, direct explaining, and coaching of


3 6


cognitive processes. In contrast, somewhole-language theorists stronglyfavor discovery, noninterventionistlearning. (Pressley, El-Dinary, et al.,1992, p. 528)

These differences, and specifically theexplicitness of instruction, may account forwhy SAIL students outperformed non -SAILstudents on the standardized word-attack andreading comprehension measures. We highlightthe interventionist role of TSI teachers becausewhole-language-based instruction, unlike SAILinstruction, has not been shown to have a strik-ing impact on the standardized reading perfor-mance of at-risk students (Stahl, McKenna, &Pagnucco, in press; Stahl & Miller, 1989).

As we close this report, we must express aregret. The cost of a study such as this one isgreat, enough that we doubt there will be alarge number of comparative evaluations ofTSI. (We have one other currently in prog-ress.) It will not be possible to map out theeffects of such instruction with a variety oftypes of readers, at different grade levels, orwith a wide range of TSI options. We haveopted instead for depth with respect to descrip-tion and for evaluation of TSI. We believe thatthe combination of this report with other re-ports of primary-level TSI (see Pressley &El-Dinary, 1993; Pressley, El-Dinary, Gas-kins, et al., 1992) provides a much fullerportrait of comprehension strategies instructionthan existed before we decided to learn fromour educator colleagueseducators who wereattempting to use basic research to improveeducation.

Author Note. We are grateful for the input of anumber of University of Maryland and school-basedcollaborators, including Pam El-Dinary, Jan Berg-man, Laura Barden, Marsha York, and the tenteachers who so graciously permitted this researchinto their classrooms during 1991-92.

Correspondence regarding this article can bedirected to the first author at the Department ofCounseling and Educational Psychology, StateUniversity of New York at Buffalo, Amherst, NY14260.

REFERENCES

Allington, R. L. (1991). The legacy of "Slow itdown and make it concrete." In J. Zutell & S.McCormick (Eds.), Learner factors/teacherfactors: Issues in literacy research and instruc-tion: Fortieth yearbook of the National ReadingConference (pp. 19-29). Chicago: NationalReading Conference.

Baker, L., & Brown, A. L. (1984). Metacognitiveskills and reading. In P. D. Pearson, R. Barr,M. L. Kamil, & P. Mosenthal (Eds.), Handbookof reading research (pp. 353-394). New York:Longman.

Bell, R. Q. (1968). A reinterpretation of the direc-tion of effects in studies of socialization. Psycho-logical Review, 75, 81-95.

Bereiter, C., & Bird, M. (1985). Use of thinkingaloud in identification and teaching of readingcomprehension strategies. Cognition and In-struction, 2, 131-156.

Bergman, J., & Schuder, R. T. (1992). Teachingat-risk elementary school students to read strate-gically. Educational Leadership, 50(4), 19-23.

Brown, R. (1995). A quasi-experimental validationstudy of strategies-based instruction for low-achieving, primary-level readers. Unpublished



doctoral dissertation, University of Maryland atCollege Park.

Brown, R., & Coy-Ogan, L. (1993). The evolutionof transactional strategies instruction in oneteacher's classroom. Elementary School Journal,94, 221-233.

Brown, R., & Pressley, M. (1994). Self-regulatedreading and getting meaning from text: Thetransactional strategies instruction model and itsongoing evaluation. In D. Schunk & B. Zimm-erman (Eds.), Self-regulation of learning andperformance: Issues and educational applications(pp. 155-179). Hillsdale, NJ: Erlbaum.

Cohen, J. (1988). Statistical power analysis for thebehavioral sciences (2nd ed.). Hillsdale, NJ:Erlbaum.

Collins, C. (1991). Reading instruction that increas-es thinking abilities. Journal of Reading, 34,510-516.

Dole, J. A., Duffy, G. G., Roehler, L. R., &Pearson, P. D. (1991). Moving from the old tothe new: Research on reading comprehensioninstruction. Review of Educational Research, 61,239-264.

Duffy, G. 0., Roehler, L. R., Sivan, E., Radcliffe,G., Book, C., Meloth, M. S., Vavrus, L. G.,Wesselman, R., Putnam, J., & Bassiri, D.(1987). Effects of explaining the reasoningassociated with using reading strategies. ReadingResearch Quarterly, 22,347-368.

DeFord, D. (1985). Theoretical orientation toreading instruction. Reading Research Quarterly,

20, 351-367.Durkin, D. (1979). What classroom observations

reveal about reading comprehension instruction.Reading Research Quarterly, 14, 481-538.

Eeds, M., & Wells, D. (1989). Grand conversa-tions: An exploration of meaning construction inliterature study groups. Research in the Teachingof English, 23(1), 4-29.

El-Dinary, P. B., & Schuder, R. T. (1993). Seventeachers' acceptance of transactional strategiesinstruction during their first year using it. Ele-mentary School Journal, 94, 207-219.

Ericsson, K. A., & Simon, H. A. (1980). Verbalreports as data. Psychological Review, 87, 215 --251.

Garner, R. (1988). Verbal report data on cognitiveand metacognitive strategies. In C. E. Weinstein,E. T. Goetz, & P. A. Alexander (Eds.), Learn-ing and study strategies: Issues in assessment,instruction, and evaluation (pp. 63-76). SanDiego: Academic Press.

Gaskins, I. W., Anderson, R. C., Pressley, M.,Cunicelli, E. A., & Satlow, E. (1993). Sixteachers' dialogue during cognitive processinstruction. Elementary School Journal, 93(3),277-304.

Ginsburg, M. (1991). Mushroom in the rain. In D.Alvermann, C. A., Bridge, B. A. Schmidt, L.W. Searfoss, P. Winograd, & S. G. Paris (Eds.),My best bear hug (pp. 144-154). Lexington,MA: D. C. Heath.

Goetz, J. P., & LeCompte, M. D. (1984). Ethno-graphy and qualitative design in educational res-earch. San Diego, CA: Academic Press.

Golden, J. M., (1988). The construction of a liter-ary text in a story-reading lesson. In J. L. Green& J. 0. Harker, (Eds.), Multiple perspectiveanalyses of classroom discourse (pp. 71-106).Norwood, New Jersey: Ablex.

Harris, A. J., & Sipay, E. R. (1985). How toincrease reading ability: A guide to developmen-tal and remedial methods. New York Longman.

Hunt, K. W., (1965). Grammatical structureswritten at three grade levels. NCTE ResearchReport No. 3. Champaign, IL: National Councilof the Teachers of English.

Hutchins, E. (1991). The social organization ofdistributed cognition. In L. Resnick, J. M.


38


Levine, & S. D. Teasley (Eds.), Perspectives onsocially shared cognition (pp. 283-307). Wash-ington, DC: American Psychological Associa-tion.

Johnston, P., & Afflerbach, P. (1985). The processof constructing main ideas from text. Cognitionand Instruction, 2, 207-232.

Kirk, R. E. (1982). Experimental design (2nd ed.).Monterey, CA: Brooks/Cole.

Lytle, S. L. (1982). Exploring comprehension style:A study of twelfth-grade readers' transactionswith text. Doctoral dissertation: University ofPennsylvania (University Microfilms No. 82-27292).

Marshall, E. (1982). Fox in love. NY: Penguin.Marks, M. B., Pressley, M., Coley, J. D., Craig,

S., Gardner, R., DePinto, T., & Rose, W. (1993).Three teachers' adaptations of reciprocal teachingin comparison to traditional reciprocal teaching.Elementary School Journal, 94. 267-283.

Mehan, H. (1979). Learning lessons: Social organi-zation in the classroom. Cambridge: HarvardUniversity Press.

Meichenbaum, D. (1977). Cognitive-behaviormodification: An integrative approach. NewYork: Plenum Press.

Miller, .1. P. (1976). Tales from Aesop. NY: Ran-dom House.

O'Flahavan, J. F. (1989). Second graders' social,intellectual, and affective development in variedgroup discussions about literature: An explo-ration of participation structure. Unpublisheddoctoral dissertation. Urbana: University of Illi-nois.

Palincsar, A. S., & Brown, A. L. (1984). Recipro-cal teaching of comprehension-fostering andcomprehension-monitoring activities. Cognitionand Instruction, 1, 117-175.

Paris, S., & Oka, E. R. (1986). Children's readingstrategies, metacognition, and motivation. De-velopmental Review, 6,25-56.

Olshaysky, J. E. (1976-77). Reading as problem-solving: An investigation of strategies. ReadingResearch Quarterly, 12, 654-674.

Olson, G. M., Mack, R. L., & Duffy, S. A. (1981).Cognitive aspects of genre. Poetics, 10, 283-315 .

Pressley, M.. & Afflerbach, P. (in press). Verbalprotocols of reading: The nature of constructivelyresponsive reading. Hillsdale, NJ: Erlbaum.

Pressley, M., & EI-Dinary, P. B. (Guest editors)(1993). Special issue on strategy instruction.Elementary School Journal, 94(2).

Pressley, M., El-Dinary, P. B., Gaskins, I., Schu-der, T., Bergman, J. L., Almasi, J., Brown, R.(1992). Beyond direct explanation: Transactionalinstruction of reading comprehension strategies.Elementary School Journal, 92, 513 -555.

Pressley, M., Gaskins, I. W., Cunicelli, E. A.,Burdick, N. J., Schaub-Matt, M., Lee, D. S., &Powell, N. (1991). Strategy instruction at Bench-mark School: A faculty interview study. Learn-ing Disability Quarterly, 14, 19-48.

Pressley, M., Goodchild, R., Fleet, J., Zajchowski,R., & Evans, E. D. (1989). The challenges ofclassroom strategy instruction. ElementarySchool Journal, 89, 301 -342.

Pressley, M., Johnson, C. J., Symons, S., McGol-drick, J. A., & Kurita, J. A. (1989). Strategiesthat improve children's memory and comprehen-sion of text. Elementary School Journal, 90,3-32.

Pressley, M., Schuder, T., Teachers in the StudentsAchieving Learning Program, Bergman, J. L., &El-Dinary, P. B. (1992). A researcher-educatorcollaborative interview study of transactionalcomprehension strategies instruction. Journal ofEducational Psychology, 84, 231 -246.

Rosenblatt, L. M. (1978). The reader, the text, thepoem: The transactional theory of literary work.Carbondale: Southern Illinois University Press.


39


Rosenshine, B., & Meister, C. (1994). Reciprocalteaching: A review of the research. Review ofEducational Research, 64(4), 479-530.

Schuder, R. T. (1993). The genesis of transactionalstrategies instruction in a reading program for at-risk students. Elementary School Journal, 94,183-200.

Stahl, S. A., McKenna, M. C., & Pagnucco, J. R.(in press). The effect of whole language instruc-tion: An update and reappraisal. EducationalPsychologist.

Stahl, S. A., & Miller, P. D. (1989). Whole lan-guage and language experience approaches forbeginning reading: A quantitative research syn-thesis. Review of Educational Research, 59,87-116.

The Psychological Corporation (1990). Stanfordachievement test series: Technical data report(8th ed.). Harcourt Brace JaVanovich.

Vygotsky, L. S. (1978). Mind in society: Thedevelopment of higher psychological processes(M. Cole, V. John-Steiner, S. Scriber, & E.Souberman, Eds. and Trans.). Cambridge:Harvard University Press.

Wyatt, D., Pressley, M., EI-Dinary, P. B., Stein,S., Evans, P., & Brown, R. (1993). Comprehen-sion strategies, worth and credibility monitoring,and evaluations: Cold and hot cognition when ex-perts read professional articles that are importantto them. Learning and Individual Differences, S,49-72.

APPENDIX

A Sample SAIL Lesson

The following lesson was presented by oneSAIL teacher to student participants. Thelesson focuses on the story, "Mushroom in theRain" (Ginsburg, 1991; see Story Lessons and

Recall Measures in the methods section for asynopsis of the story). SAIL teachers differsomewhat in their instructional practices asthey tailor instruction to meet the needs of theirstudents. Also, their emphasis on specific SAILcomponents varies as the year progresses andas students become more competent readers.However, the lesson highlights key featuresthat are typical to SAIL instruction.

The SAIL teacher (T) began her "Mushroom inthe Rain" lesson by reviewing what goodreaders do:

T: Well, today's reading group is almost at theend of the second grade. And we have spentthe whole year learning so much about beinggood readers, and you have become so muchbetter readers than you were at the beginningof second grade; it's incredible. And I think thereason why you've become such good readersis because of what you've learned this year.Let's just summarize what we've learned aboutbeing a good reader this year, being an expertreaderwhat you have learned that you knownow that you didn't know before the secondgrade. Let's talk.

Taking her cue, the students named and de-scribed the various problem-solving and com-prehension strategies they had used throughoutthe year and talked about how those strategieshelped them. As students began to talk aboutvisualizing, the teacher gave a personal exam-ple of her use of the visualizing strategy:

T: Do you know what I do when I'm reading? Itry, as I'm reading a novel, to make a picturein my mind of the events that are aking place.


40


If the story takes place in the setting of awoods, I try to visualize the woods in mybrain, and I try to visualize what's happening,and it helps me remember when I want tosummarize, when I want to look back, and tryto think what's happened so far. I try to thinkwith my brain, but also use my visualizingstrategy to picture, oh yes, this is what hap-pened first, yes this is what happened second,and that helps me remember.

After reviewing individual strategies, theteacher brought up the importance of the flexi-ble use of a repertoire of strategies. She in-formed students that different strategies wereused for different purposes and applied todifferent text types. The teacher addressed thisissue directly when she asked students:

T: How do you know which one to pull out? Howdo you know? A carpenter doesn't pull out ahammer when he wants to screw in a screw.You have to make those decisions don't you?Now as expert readers you have to make deci-sions which strategy to use, which one willwork. Do you want to make a prediction at theend of the story necessarily? Do you want touse a fix-up strategy if you know all the wordsand you understand what the story's about? Doyou want to visualize? If there's a picture rightthere and, gee, that's exactly what you'rethinking is happening in the story, it looks justlike what you're imagining? How do you knowwhen to use a connection . . . a think-aloudstrategy? How do you know?

After discussing their answers, the teacherreminded students to apply their repertoire ofstrategies when reading, thereby emphasizingstudent choice and control.

T: Today, the story we're going to readI wantyou to try to use your, now that you're expertsecond-grade readers, use your bag of strategytricks and select the one that's going to helpyou as you read. Can you use more than oneduring a reading time?

S: Yes.

T: Absolutely. But you are now in control, youare now readers in control, you have your ownbags, you have your own tricks, your ownstrategies, your own tools to help you under-stand what you read and now you are the boss.You choose what strategy is going to help youunderstand the story.

Following the strategy review, the teachermentioned that in addition to using all thestrategies, she wanted students to focus onvisualizing. She then started the lesson byreading the title and first page and modelingher thinking processes for students.

T: One day, well, I, let's see, the title's "Mush-room in the Rain." I know what mushroomsare because we've been studying plants, andwe saw a filmstrip just day before yesterday.

(She then reads the first page.)

T: Well, I know what a clearing is, it's where, inthe woods, where they've taken down a lot oftrees, or it's an open space where there aren'ta lot of tall bushes and trees and things likethat. Well, I guess for an ant, though, thatcould be pretty small. I'm visualizing a clear-ing as a grown-up would be. It would be a bigplace without trees. But 1'11 bet a clearing for alittle tiny ant would just be a place where there


41


are maybe bits of leaves to make shadows, butjust maybe, just grass, just growing around.And so he's looking, he's looking for some-thing to act sort of like a . . .

S: . . . cover.

S: . . . umbrella.

T: . . . like an umbrella, like an umbrella, to keephim, to keep him dry until the rain stops. (Shecontinues to read: "He sat there waiting for therain to stop, but the rain came down and camedown.")

T: You know, that's happened to me, what I'mpicturing. I'm visualizing when I was strandedhere at school, without my car, with my bag ofschool papers to grade, waiting under theoverhang, that was almost like my umbrella,waiting for the rain to slow down a little bit, sothat I could walk home to my house. Gosh,sometimes it seems like forever till the rainstops. Has that ever happened to you?

S: I have a prediction, urn, this is gonna be like,urn, like "The Mitten" one, like urn these urn,all these insects are gonna try to come in.

T: You think so?

S: Yeah.

T: What made you think that?

S: Well, I see another insect.

T: And so . . .

S: . . Hrnmm . .

T: You made that connection? Well, we'll see ifyou're right. Do you want to take charge now,S6? Are you ready to take charge. I would likeyou to. Will you try today?

In the preceding section, the teacher mod-els for students her use of visualizing andmaking connections between story events andher personal experiences. She then encouragesstudents, in context, to think of similar situa-tions when they were stranded in the rain. Astudent, without prompting, makes a predic-tion, comparing the current story with a storythe class read in the middle of the year. Theteacher asks the student to provide some sup-port, and the student alludes to a picture clue.Finally, the teacher turns control over to astudent, who serves as lesson leader for theday.

Students take turns reading story segments.When students come to a word they do notknow, they often use one of their fix-up (e.g.,problem-solving) strategies spontaneously.However, sometimes, the teacher cues stu-dents, as in the following example:

T: Use a strategy.

S: (rereading) ". . . went, then, wait, dry thanwet."

T: Does that make sense?

S: (The child nods and continues reading.)

After reading a segment, a student eitherspontaneously thinks aloud or is prompted todo so. Thinking aloud takes the form of sum-marizing story content, venturing a prediction,


42


or offering an interpretation. Often, otherstudents, either on their own, or prompted bythe teacher or the student leader, offer theiropinions. For example, after a child thinksaloud, the student leader asks:

S: Anyone else want to read?

But first, a student returns to the topic of "TheMitten":

S: I want to predict something. Well, I made aconnection. Well, urn, this is reminding me of"The Mitten" when urn, when lots of animalswere trying to get into one mitten, one littlemitten, and then it exploded.

As students read, they continue to make aconnection between "The Mitten" and "Mush-room in the Rain." Students and teacher discussdifferent versions of the story. Then, studentscontinue to think aloud and discuss each textsegment as it is read. To stimulate discussion atone point the teacher asks:

T: Can you visualize what it would be like, to bethat, to be those characters? That's what I tryto do when I'm reading. I try to put myself inthe character's place and it helps me understandhow they're feeling.

During one text segment discussion, a stu-dent observes an emerging patternthe animalsentering the mushroom are getting increasinglybigger. As students volunteer connections,opinions, or predictions like the preceding one,the teacher encourages students to keep them inmind as the group reads on.

As students continue to read, they talkabout the various animals entering the mush-

room. Extended discussion ensues when therabbit appears on the scene. Students recognizethat the rabbit's desire to get under the mush-room differs from the motives of other ani-mals. One student recalls a directly relevantpersonal experience about seeing a pet rabbitchased by a fox. Another student predicts thatthe fox will say he's wet so that he can enterthe mushroom to get the rabbit. Still anotherstudent connects the episode to a personalexperience:

S: Prediction, well, not a prediction, but whenev-er my sister was at urn, Camp Sunshine, a fox,whenever they were staying over in a tent, afox came up to their urn, tent, right up in front,and then they, I don't remember what they didtogether, and then the fox ran away into thewoods.

T: So you've made a connection, how does thatrelate to the story? What are you thinking?

S: Well . . .

T: What do you think might happen?

S: Probably the fox's um, just like, the fox isprobably after the, like they're in a tent kind oflike . . .

T: Okay, that was your connection, the tent was,kind of like the mushroom?

S: (The student concurs.)

With the various predictions raised by stu-dents, the group decides to take a vote aboutwhether the fox will or will not get the rabbit.They decide to read on to verify their predic-tions.


43


A student is selected to read the next seg-ment. The child experiences great difficultyreading the page, and frequently skips overunknown words. The teacher suggests that thegroup reread the last page together. One stu-dent agrees that group reading is a good ideabecause what the other student read, "doesn'tmake sense. " Another student elaborates:"Doesn't really make sense because he skippeda few words." The teacher reinforces, seizingan opportunity to interject some explicit in-struction about what good readers do:

T: "Well, let's . . . why don't we, why don't weread it. I think that's a good, I think that's agood idea. I do that. I do that. I go back andreread when I don't understand. That's whatgood readers do. Let's go back to the top of thepage then . . " (and they read the page togeth-er).

The students then discuss how the otheranimals are protecting the rabbit and where therabbit is located. One student suggests that therabbit is hiding in the middie of the otheranimals. Another student, pointing to a pictureclue, shows the rabbit on the far side of themushroom, away from the fox. Then onestudent observes:

S: Urn, that, um, when it says mushroom, it's likeit's a room, it's like a house, and then a room.

T: Oh, it's like, right now, is it acting like theirhouse in a way? It is a shelter, isn't it? It is ashelter.

S: A "mush-room."

S: Like an igloo.

S: With all these people in it, but it's kind of likea room, but lots of people in it, so, a room,but, people are mushed in there.

T: Are they kind of smooshed all together?

S: Yeah, "mush-room."

In this case, two students picked up theparts of the word "mushroom"; one made aparallel between the mushroom serving as aroom to hold the animals; the other childfocused on the fact that the animals were"mushed" together.

A short time later, a student had difficultyreading:

S: I can't figure out what he did with his tail, butall . . .

T: Okay, so what can you do, cause you've gotsome strategies you can use.

S: Skip it and go on.

T: Why don't you try that?

S: (The student skips and continues reading untilthe end of the passage.)

The student then thinks aloud, summariz-ing and expressing her ideas about why the foxcouldn't find the rabbit. The teacher does notgo back to discuss the word, "flicked," that thestudent missed. Instead, she focuses on wheth-er the student got the gist of the passage.

As the lesson proceeds, students continueto discuss their views and offer new or modi-fied predictions. Throughout these discussions,the teacher rephrases students' comments, asks


4 1


for elaborations, or initiates a discussion witha question. However, she does not present herown interpretation of story events.

The group decides to read the last pages ofthe story together. They come to the questionthe ant poses about how all the animals couldhave fit under one small mushroom. One childsuggests that the mushroom grew because " . .

. the sun came out and it got plenty of water togrow a little more." Other students concur thatthe mushroom grew, offering their support.But one student says that the animals caused themushroom to stretch. So the teacher asks: "Doyou think they stretched it or do you think thatit grew?" Most agree that the mushroom grew.

The discussion then returns to the story,"The Mitten." The teacher and students discusssimilarities and differences between the twostories (e.g. , The mitten popped and mushroomstayed just fine.). The teacher praises thestudents for making a connection between thetwo stories.

The lesson concludes with an evaluation ofstrategy use and ample praise for student use ofstrategies during the lesson:

T: Well, what I'm so impressed with is the factthat you chose strategies to help you understandand I heard some strategies that you used thatI didn't help you (with). In the beginning ofsecond grade, I had to say, "Okay, today we'reall going to make predictions; today we're allgoing to summarize; today we're all going totry to visualize."

S: I remember that.

tired. And I don't have to do that anymorebecause now you're the bosses of your reading.You choose the strategies that help you under-stand. And I am very proud of your thinking.I hope you carry all these strategies in yourstrategy bag, which is sort of imaginary, isn'tit? Sort of imaginary. I hope you carry them tothird grade and to fourth grade, and to fifthgrade, and the rest of your life because they'llalways help you. They help me. And if theyhelp me, and I'm a grown-up, they certainlyare gonna help you every step along the way.Give yourselves pats on the back for doingsuch a super, super thinking job. You got moreout of this story I think than I did when I readit! Good job!

T: Do you remember that? Do you rememberwhen I, we practiced and practiced and prac-


45

NRRC NationalReading ResearchCenter318 Aderhold, University of Georgia, Athens, Georgia 30602-71252102 J. M. Patterson Building, University of Maryland, College Park MD 20742

4 6 BEST COPY AVAILABLE

Documents

MD. SPONS AGENCY - ERIC · Jessie Pollack Maryland Department of Education Baltimore, Maryland. Sally Porter Blair High School Silver Spring, Maryland. Michael Pressley. State University