Sie sind auf Seite 1von 18

Activating Background Knowledge in Reading Comprehension Assessment Author(s): Deborah Wells Rowe and Lawrence Rayford Reviewed work(s):

Source: Reading Research Quarterly, Vol. 22, No. 2 (Spring, 1987), pp. 160-176 Published by: International Reading Association Stable URL: http://www.jstor.org/stable/747663 . Accessed: 18/04/2012 03:32
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at . http://www.jstor.org/page/info/about/policies/terms.jsp JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about JSTOR, please contact support@jstor.org.

International Reading Association is collaborating with JSTOR to digitize, preserve and extend access to Reading Research Quarterly.

http://www.jstor.org

Activating background knowledgein assessment readingcomprehension


DEBORAH WELLS ROWE
Vanderbilt University

RAYFORD LAWRENCE
RMCResearch Corporation

THIS STUDY was an investigationof readers'activationof backgroundknowledgein response to prepassagepurpose questions selected from the readingcomprehensionsection of the research edition of the MetropolitanAchievementTests (1985). A total of 74 students from Grades 1, 6, and 10 were shown three purposequestionsfrom appropriate levels of the MAT and asked to make predictionsabout the content of relatedpassages. Interviewswere transcribed and analyzedfor the relationshipof the responsesto the contentof the purposequestions andfor the extentof schemaelaboration.Studentsat all gradelevels producedresponses that were relatedto the informationin the purposequestionor to unstatedstory schema categories. At Grades 6 and 10, patternsof schema elaborationwere found to be significantly relatedto the purposequestionsused, whereasGrade 1 studentsmaintaineda similarpattern across all purposequestions. The results indicatethat a broad age range of studentscan use purpose questions as cues to activate backgroundknowledge, but that all purposequestions are not equallyeffective in performingthis cuing function. Topicfamiliarity, amountof information presented, and the presence of genre clues are suggested as text features affecting schema activation.

de texte Reactiverle savoir acquis en comprdhension CETTE RECHERCHE a 6tudie des lecteurs h r6activerleur savoir acquis h partirde r6ponses '

des questions sur l'intentiond'un passage pos6e avant la lecture du passage. Ces questions sont extraitesdes Metropolitan Achievement Tests(MAT, 1985), edition des chercheurs,dans la section portantsur la comprehensionen lecture. Un total de 74 6tudiantsde premiere, sixieme et dixieme ann6eont 6t6 choisis et soumis a trois de ces questionsde niveauxappropri6s prises dans le MAT;on leur a ensuitedemand6de fairedes predictionsquantau contenu des passages concern6s. On a transcritles entrevuesafin d'analyserla relationentre les reponses obtenueset le contenudes questionspr6sent6es,de meme que l'envergured'61aboration des sch6mas. Les e61ves de tous les niveaux ont fourni des r6ponses reli6es a l'information contenuedans les questionsou reli6es ' des categoriesde representations (sch6des mas) de l'histoirenon mentionnees.En sixieme et dixieme annie, les plans d'61aboration sch6mas6taientsignificativementrelies a la questionutilis6e alors qu'enpremiereannie, les pas malgr6le changementde question. Les r6sultatsder6ponsesne variaientpratiquement montrentque les e61vesappartenant a des categories d'age assez variees peuventutiliser de telles questions comme signal de reactivationdu savoir acquis mais que les questions ne remplissentpas toutes aussi efficacementcette fonction. On suppose que les caract6ristiques du texte telles que la familiarit6du sujet, la quantit6d'information pr6sent6e,et la presence d'indicesde genre peuventinfluencerla concr6tisation du schema.

160

READING RESEARCH QUARTERLY * Spring 1987

XXII/2

La activaci6n del conocimientoprevio en la evaluaci6nde comprensi6nde lectura


ESTE ESTUDIO investig6 la activaci6n del conocimiento previo de los lectores en respuestaa preguntassobre el prop6sito de los pasajes hechas antes de leerlos. Las preguntasfueron seleccionadas de la secci6n de comprensi6nde lecturade la edici6n de investigaci6nde las AchievementTests: MAT, 1985). A pruebasde aprovechamiento Metropolitan (Metropolitan un total de 74 estudiantesde los grados 1, 6 y 10 se les ensefiarontres preguntasde niveles apropiadosdel MAT y se les pidi6 que hicieran predicciones sobre el contenido de pasajes relacionados.Se transcribieron y analizaronlas entrevistas paraver la relaci6nde las respuestas al contenidode las preguntasy paraver la extensi6nde la elaboraci6nde esquemas. Los estudiantesde todos los niveles produjeronrespuestasque se relacionabana la informaci6n en la preguntasobre el prop6sitoa categoriasno mencionadasdel esquemade la historia. En los grados 6 y 10 se encontr6 que los patronesde elaboraci6nde esquemas estabansignificativamente relacionados a la preguntausada, mientras que en los estudiantes de primer gradono hubo variaci6nsignificativaa lo largo de sus respuestasa las preguntas.Los resultados indicanque un amplio rangode los estudiantespueden usar preguntassobre el prop6sito de un pasaje como sefiales para activar conocimiento anterior,pero que todas las preguntas sobre el prop6sitono son igualmenteefectivas en desempefiar esta funci6nguiadora.La familiaridad del tema, la cantidad de informaci6n presentaday la presencia de las claves de de texto que afectanla instanciaci6nde esquemas. g6nero se sugierencomo caracteristicas

des Vorbildungswissen in Leseverstdindnis-Einschatzung Aktivierung

DIESE STUDIE befa8t sich mit der Aktivierung des Vorbildungswissens von Lesern im Hinblick auf Vorabschnitt-Zweckfragen, die ausgesuchtwurdenaus der Verstdindnis-Sektion der Forschungsausgabe der Metropolitan Achievement Tests(MAT, 1985). Im ganzen wurden 74 Schfilernaus der 1., 6. und 10. Klasse drei Zweckfragendes ihnenentsprechenden Grades des MAT vorgelegt, und sie wurden angehalten,fiber den Inhaltder beziiglichen Episoden zu machen. Die Befragungenwurdenaufgenommenund analysiertauf die BeVorhersagen ziiglichkeitder Antwortenauf den Inhaltder Zweckfragenund auf das AusmaBvon SchemaAusarbeitung.Schiiler aller Klassen gaben Antworten,die sich auf die Informationin den In der 6. und Zweckfragenbezogen oder auf unerwdihnte Geschichtenschema-Kategorien. 10. Klasse stellte sich heraus, daBdie Beispiele von Schema-Ausarbeitung sich stark auf die Zweckfragenbezogen, wohingegenSchiler der 1. Klasse in ihren Antwortenauf Zweckfragen nicht sonderlich abwichen. Die Resultateweisen daraufhin, daBSchiiler verschiedener Klassen Zweckfragenbenutzen, um Vorbildungswissen zu aktivieren,jedoch daB nicht alle mit dem Thema, Zweckfragengleich wirksamsind mit ihrerHinweis-Funktion.Vertrautheit der Anteil an gebotener Informationund Anwesenheitvon Art-Hinweisenwerden als TextMerkmalevorgeschlagen,um die Schema-Vergegenwdirtigung zu erleichtern.

Authors of a number of recent reviews of the status of reading research (Harste, 1985; Pearson, 1985) have suggested that there is a strongtrendfor researchersto adopta theory of readingthat acknowledgesthe active role of the readerin constructingmeanings from texts. As Pearsonputs it, "Wenow view text as a sort of blueprintfor meaning, a set of tracks or clues that the reader uses as s/he builds a model of what the text means" (1985, p. 726). For the most part, such a trend means that researchers
Activatingbackground knowledge

have moved away from information-transfer theories and towardschema-theoretic models of the reading process. According to these newer theories, the use of backgroundknowledge is a key activity in comprehension.

ActivatingBackgroundKnowledge During Assessment


Readingcomprehensiontests constituteone type of reading task where the activation of
161

ROWE & RAYFORD

background knowledge is particularly important-and also particularly difficult, because tests are usually composed of a series of brief, unrelatedtexts that proceed linearly, with little regardfor the orderof content. In orderto comprehend the series of passages presented on readingtests, studentsmust activate a different network of backgroundknowledge as they encountereach new text. This variationin content creates conditions that are quite different from most "real-world"reading tasks. Despite the growing consensus among researchers about the active role of the readerand the importance of backgroundknowledge in the reading comprehensionprocess, schema theory has had little impact on the reading comprehensiontests commonly used in schools (Johnston, 1983). However, one recent attempt to apply schema-theoreticfindings to test design is the inclusion of a purpose question before each reading comprehensionpassage on the Metropolitan Achievement Tests (MAT6; Prescott, Balow, Hogan, & Farr, 1985). Reportedly,the publishers of the tests of General Educational Development(GED) will make a similar addition on the forthcomingedition of that instrument. The authorsof these tests suggest thatthe inclusion of a high-level purpose question before each reading comprehensionpassage may be one way of providingcues to help test-takers activate appropriate background knowledge. Specifically, the authorsof the MAT6 have operationallydefinedpurposequestionsas prepassage questions that focus on higher order informationcontained in or inferred from the passages they precede. Accordingto the developers (Tone, 1984), purpose questions may be written to help readers (a) focus on a central point (e.g., "Willthe North Americanwolf survive?"), (b) challenge passage content (e.g., "Did these people really see unidentifiedflying objects?"),or (c) make an evaluativejudgment of its content (e.g., "How good a friend would Alex make?").All items following the passage are relateddirectly or indirectlyto the purpose question. The authorsof the MAT6 suggest that their reasonsfor includingpurposequestionsare twofold. First, purposequestionsshouldsuggestper162

spectives readers can use for organizing their to passagecontent.The global thinkingin relation schemataneededto "getinto"the passageshould be activated as theybegin to read.Second,the inclusion of purposequestionsmay servea motivational function, building interestin the passage form personalreasonsfor and helpingtest-takers reading,over and above the obvious purposeof well."Purposequestionsareproposedas "scoring one means of providingboth the motivationand the textualcues studentsneed to makemaximum use of their existing readingstrategiesand abilities. Textual Cues for Schema Activation From a schema-theoreticperspective, prepassage cues such as purpose questions should help readers select appropriate schemata to serve as the basis for text comprehension.Such cues take on added importance in light of research indicatingthat the concepts presentedin the text will be rememberedas a functionof the schema in which they were initially encoded by the reader (Anderson, Reynolds, Schallert, & Goetz, 1977; Bransford& McCarrell, 1974). The idea of inserting questions at various points in educationaltexts to aid comprehension is not new. There is a considerablehistoryof research in the area of adjunctquestions, though many of the studies in the late 1960s and early 1970s (Frase, 1967; Rothkopf, 1966; Rothkopf & Billington, 1975; Rothkopf & Bisbicos, 1967; Rothkopf& Kaplan, 1972) were based on behaviorist rather than schema-theoretic models of reading. In these early studies, researchersoften investigatedthe effects of question position on retentionof both directly and indirectlyrelatedinformation.The overallconclusion from this research is that postpassage questions facilitate both direct and indirect learning, whereas prepassage questions facilitate only direct, question-specificlearning.This generalizationhas frequentlybeen cited in arguments against the prepassage placement of questions (e.g., Boyd, 1973; Faw & Waller, 1976; Rickards, 1979; Swenson & Kulhavy, 1974). However, as Watts and Anderson (1971) point out, the studies from which these concluXXII/2

* Spring 1987 READING RESEARCH QUARTERLY

sions were drawnprimarilyinvestigatedthe effects of literal-level questions, requiring only recall of isolated facts. Because, in their opinion, "most importanteducational goals do not entail verbatimrecall"(p. 387), Wattsand Anderson suggest thatmore attentionshouldbe devoted to investigatinghigh-level questions in an effort to promote deeper processing. Though fewer in number,some studieshave investigated high-level adjunct questions. Both Watts and Anderson (1971) and Rickards and DiVesta (1974) found evidence that high-level questions positioned after a passage helped students to build an overallrepresentation of the text and to recall informationnot tappedby the questions. Even more importantto the discussion of purpose questions as techniques for activating background knowledgeare two recentinvestigations of high-levelprepassagequestions(Wilhite, 1983, 1984). In both studies, questions were writtento directreaders' attention to the top-level structure of the text. Wilhite organizational (1983) initially found that the-use of high-level prepassagequestioninghelped studentsactivate an overall schema for the passage and allowed for its instantiation with specifics from the text. In the second study (1984), in which Wilhite used a cued recall ratherthan a free recall measure, the facilitativeeffects of higher level questions were less pronounced, affecting only the high-ability students. In explaining the differences betweenthe high- and low-abilitysubjects' responsesto the two types of questions, Wilhite suggestedthat high-abilityreadersmay be more able to focus their attentionon informationat a level in the organizational structure of particular the text. The discrepanciesbetweenthe findings of the two studiesindicatethatthe facilitativeeffect of prepassagequestionsmay be temperedby such factors as the demandsof the comprehension assessment measure and the ability of the readers. Thoughthe amountof researchevidence on the effect of high-level prepassagequestions is limited, it generally supports a facilitative-or at least a neutral-effect on comprehension. There is certainly no evidence that high-level prepassage questions have a detrimentaleffect

on overall comprehension, as might be concluded if only studies of factual-levelquestions were examined. Instead, prepassage questions that direct the reader'sattentionto the top-level structure of a text seem promising for aiding comprehensionof both global and specific information. The existing evidence on the effects of high-level prepassage questions is, however, subject to a number of important limitations (Rowe, 1986). One of the most glaring is the lack of research describing the types of backgroundknowledgeactivatedin responseto purpose questions before the related passages are read. Most researchershave inferredthe effect of prepassage textual cues by examining students' postreadingresponses; the most popular research techniques have included analyses of retellingsand responsesto researcher-generated questions. When researchers (Langer, 1980, 1981, 1985; Langer& Nicolich, 1981) have examined the natureand organizationof students' initial responses to prepassagecues, they have found evidence that the responses vary in the amountand type of background knowledgeactivated. Clearly, more direct investigationof the initial stages of schema activation is needed if we are to understandhow readers make use of prepassagecues. Two other pieces of evidence now lacking are the effects of various types of high-level questions, and the effects of prepassage questions for students of different ages, histories. abilities, and instructional The presentstudywas designedas an initial attemptto addresssome of these limitations.Becauseprepassage arecurrently questions being included in tests used for assessment as well as those used for instruction,more informationis neededaboutreaders' responsesto these text features. This information,and the theoreticalinsights it generates, are needed to direct the of instructional and assessmentmadevelopment terialsthatmakemaximum use of the schema-activationpotential of prepassage questions. Thus, our major aim in this study was to describethe natureand organizationof students' initial responses to purpose questions selected from the researchedition of the MAT(1985). In

Activating background knowledge

ROWE & RAYFORD

163

addition, because the limited researchon high- TableI Purposequestionsused in the level prepassagequestioninghas raisedthe posinterviews sibility that different groups of students might react differently, we sought to make tentative Grade 1 comparisons of the responses of different age 1. How can you cross a street safely? did the neighborsplanttrees? 2. groups. Specifically, we set out to investigate 3. Why Whatwas Tom's joke? four majorquestions: 1. Do studentsactivatebackgroundknowlGrade6 4. Why would people eat weeds? edge in responseto purposequestions? 5. How does this writerget away from the city? 2. How is that backgroundknowledge re- 6. Why was Socratesimportant? lated to the information presented in the purGrade 10 pose question? 7. How good were Jules Verne's predictions? 3. Are there differences in the types of re- 8. Why doesn'tPhiliphave friends? sponses generated for different purpose ques- 9. How are these scientiststryingto readyour mind? tions? from wereselected the Note: These Intermediate, 4. Are there differences in the ways elementary,middle, and high school students respond to purposequestions?
Primary, questions editionof the Metropolitan andAdvanced levels of the research
Achievement Tests(1985).

Method

Subjects were selected from two schools Participants in a small-city school district in south-central Indiana. The 24 first-grade and the 28 sixthgrade students were members of intact classes in an elementaryschool, whereas the 22 tenthgrade students were selected from several classes by high school administrators. We asked the school administrators to providea cross-section of the academicabilities of studentsin their Interview Procedures Interview procedures were first piloted school. Most of the studentswere from white, with a group of 10th-gradestudentsnot particimiddle-incomefamilies. pating in the study. We followed a semi-strucMaterials tured interview guide containing the major for this studyweredrawnfromthe questions to be asked and specifying the types Materials andAdvancedlevels of the of follow-up questionsto be used. However,we Primary, Elementary, research edition (a preliminaryversion devel- adjustedquestion phrasingand pacing accordoped for field review and comment)of the Met- ing to the responsesof individualstudents.The ropolitanAchievementTest (1985). In order to order of presentationof the three questionswas at each gradelevel. investigatethe types of backgroundknowledge counterbalanced Each participant was individually interstudentsactivatein responseto adjunctquestions used on comprehensiontests, we randomlyse- viewed in a quiet location. Interviews were lected three purposequestionsfrom each of the audiotapedand usually lasted approximately15 threetest levels. When we examinedthese ques- minutes-roughly 5 minutes per purpose questions in relationto theirpassages, we determined tion. Studentswere informedthattheir task was

thatthey required studentsto inferanswersby inacross sentencesand parategratinginformation Each graphs. questionwas removedand attached to an unlined3" x 5" (7.5 cm x 12.5 cm) card. Three stimulusquestionsseemed an appropriate numberfor providingadequatecontentvariability, but not too many to promotefatigue among the younger studentsparticipating in the study. Table1 presentsthe purposequestionsused with the threegradelevels.

164

* Spring1987 READING RESEARCH QUARTERLY

XXII/2

not to read the passages or answer the test Table2 Sampletranscriptof a 10th-grade aboutpassage coninterview items, but insteadto "guess" tent. Studentswere briefly shown a copy of the test booklet appropriatefor their grade level, Researcher (R): O.K. That'spretty good. Now let's do another question. Wouldyou please read this and were informed of the general nature and question, and then tell me what you think functionof purposequestions;thatis, they were the story is going to be about. told that purpose questions had been added to Student(S): How are (pause) these scientists trying to this new test to help test-takersbegin thinking read your mind? aboutthe passages before they readthem. Next, R: Right. How are these scientists trying to read your mind? What do you think the students were shown both a purpose question passage is going to be about? isolated on a card, and the correspondingtest S: (1) About some scientists who are reading We to the locapage. briefly pointed prepassage / or (2) trying to readyour mind./ tion of the question and explained that the iso- R: O.K. Can you tell me some more? Well, it's about (3) mind readers lated purpose question had been cut from a S: (4) who are some scientistpeople similar test booklet. After this brief exposureto (5) who work in laboratories/ and the context in which the purpose question was (6) readminds and things. (pause) used, the booklet was removedand the students' (7) I guess I don'treally know. / attentionwas focused on the isolated question. O.K. That'sgood. Trythinkingaboutsome R: more things. Respondentswere asked to read each pur- S: (8) They are mind readers (9) who read and make about the pose question predictions minds./ passage that accompaniedit in the test booklet. R: O.K. Now think aboutif you were takinga Pilot interviewsindicatedthata numberof firstreadingtest and the passage had this question in frontof it...what would the passage grade children needed help in reading the sebe about? lected purpose questions, so we supplied S: Well, it would be about (10) some men unknown words when necessary. Sixth- and who made up ways to read/ (pause) 10th-grade students, however, were asked to (11) they read minds, (12) get into your head / and (13) find out about you, (14) read the questions on their own. For all grade about the way you think / and (15) what inlevels, we avoidedsupplyingany substantive think about / and like that. And it you formationabout the purpose questions, and inwould tell you about (16) how they could stead used neutralelaborationprobes (Gordon, do it / and (17) could not do it / and (18) 1980) to encourage students to make predicwhy it's illegal (19) for all of us to do it / but (20) the Governmentcan. / And (21) tions and extend their responses. names / and (22) the men who are doing it When participantshad completed their re/ and (23) why./ to this initial we used sponses predictiontask, R: Tell me some more... two other types of probing questions. Students S: (24) So they can find out things aboutit./ were askedto put themselvesin the position of a (25) My aunthad mind readers./ She did? test-takerand to describe what they would "try R: Yeah. (26) She said they attachedsome litto find out"during reading if they had encoun- S: tle dart / or (27) something to her head tered that purpose question on a test. Finally, (28) when she was in the hospital / and then (29) she had it (30) even after it was they were asked to assume a test-taking perpaid for.../ (etc.) spective and to guess about the content of the thebeginning of a codedpredicate questions that might appear at the end of each Note: (#) marks proposition / marksthe close of a coded proposition,or separatescontest passage. These two interview questions nectives not coded (and, but, or, etc.) were includedto encouragestudentsto use their knowledge of test-taking to provide a context for understandingthe nature of the interview task, and for activatingbackgroundknowledge presented in Table 2 illustrates the questions relatedto the purpose questions. The transcript used to elicit studentpredictions.

Activating background knowledge

ROWE & RAYFORD

165

Data Analysis Because the purpose of this study was to present an initial description of students' responses to purpose questions, it was important that the descriptive and explanatorycategories used in data analysis be grounded in the data ratherthanbeing specified a priori. Forthis reamethoddescribed son, the constantcomparative by Glaser and Strauss (1967) and Lincoln and Guba (1985) was used for data analysis. Because this procedurerequirescontinualcategorical review and refinement, the two phases of analysis it eventually produced were, in practice, interdependent.Specifically, the categories generated during Phase 1 focused on the relationship between the content of students' predictionsand the contentof the purposequestions. As these categories were refined, a second and overlapping set of categoriesdescribing the extent of schema elaborationemerged. The Phase 2 analysis, therefore,necessitateda reexaminationof the data in order to systematically apply a second set of categories reflecting the complexity of the students' responses. PHASE 1 Analytic Procedures All interviews were transcribed and then organizedby grade, student,and purposequestion. All of the students' responseswere then divided into units by identifying predicate propositions (Feathers, 1984; Harste & Feathers, 1979; Kintsch& van Dijk, 1978; Turner& Greene, 1977). Following the suggestions of Lincoln and Guba (1985), each predicateproposition was placed on a separatecard, and these were then sorted into content categories. We each completedthis process for all students'responses to one purpose question, and then met to compare and refine categories. This process was repeated for students' responses to each purpose question. Through discussion, we eventually refined content categories and reached agreement regarding their defining properties.The rules thus derivedwere then applied to each of the nine purpose questions, in anotherpass throughthe dataset.
166

Resultsand Discussion
Content Analysis The Phase 1 analysis of students'responses to the selected purpose questions resulted in three majortypes of categories:contentcategoand miscues. ries, metastatements, Contentcategories. As seen in Table3, the majorityof propositionswere placed in content categories labeled with one of the word concepts (Turner& Greene, 1977) expressedin the purpose question. That is, when each purpose question was analyzed into its propositionsfollowing the procedure of Harste and Feathers (1979), there were categories of responsescorrespondingto the predicaterelationand each of its arguments. In addition, there were categories related to argumentsfilling case roles implied or requiredby the predicate,but unstated in the text. The categorization of responses for the sixth-grade purpose question "How does this writer get away from the city?" was typical. When written as a propositionaltext base, this question appears as a series of four propositions: PURPOSE Howdoes this writer QUESTION: get awayfromthecity? 1. (GETAWAY, A: Writer, S: City,G: $) 2. (REFERENCE: WRITER, $) 3. (MANNER: $, (1)) 4. (DESCRIBE, TEST-TAKER, (3))
(NOTE: A

= required case, not present in surface structureof the text # = a proposition earlier in the text

= agent; S = Source; G = goal

base

> = inferred proposition

As can be seen in Table4, four of the categories generatedby the constant comparativemethod correspond to (1) the predicate relation GET AWAY and its connective propositionof condition, MANNER; and the argumentsof (2) the agent, (3) the source, and (4) the goal. The correspondingcategories were, in turn, labeled (1) "aboutthe manner of getting away," (2) "about the writer," (3) "aboutthe city"(i.e., the writer's point of origin), and (4) "aboutthe writer'sdestination." Note that the goal-the writer'sdestiXXII/2

* Spring 1987 READING RESEARCH QUARTERLY

Table3 Phase 1 analysis: Contentcategories generatedfor each purposequestion


Title of Category No. of Propositions

Question 1: How can you cross a street safely? About safety rules for crossing 61 20 About crossing the street Question2: Whydid the neighborsplant trees? 4 About neighbors 23 About trees 16 About plantingtrees About reasonsfor plantingtrees 55 Question3: Whatwas Tom's joke? 8 Aboutjoke-telling Aboutjokes 47 14 Aboutjoke-listeners 7 About reactionsto jokes Question4: Whywouldpeople eat weeds? 46 About weeds 28 About eating weeds 45 About people who eat weeds 45 About reasonsfor eating weeds 9 About where people eat weeds Question5: How does this writerget awayfrom the city? 56 About the writer 20 About the mannerof getting away 65 About the reasonfor getting away 6 About the city (point of origin) 11 About the writer'sdestination Question6: Whywas Socratesimportant? 68 About Socrates(as person) 34 About Socrates(as non-person) Question 7: How good wereJules Verne's predictions? 47 About predictions 40 About Jules Verne 22 About the qualityof predictions About the moralof story 6 Question8: Whydoesn'tPhilip havefriends ? 66 About Philip 43 About reasonsfor no friends About the moralof story 6 Question9: How are these scientists tryingto readyour mind? 24 About the mindreaders/scientists About the person whose mind is read 9 About mindreading 68

nation- is absent from the surface structureof the purpose question, but is logically required In order to GET by the concept GET AWAY. the writermust have both a point of oriAWAY, gin (source) and a destination(goal). One other content category was generated for this purpose question: "why the writer gets away."Although this category relates to an unActivating background knowledge

stated propositionof causality (i.e., CAUSALITY: $, (1)), its inclusion can be explained in relation to Stein and Glenn's (1979) story schema categories (see Table 3). According to their system, the four previouslydiscussed categories relate to the setting ("aboutthe writer," "aboutthe city"), the attempt ("aboutthe manner of getting away"), and the consequence the writer'sdestination").The fifth con("about content category, "why the writer gets away," tains responses related to Stein and Glenn's story schema categories of initiating event and internal response. Thus, students'responses to this purpose question included informationdirectly relatedto the purposequestion'spredicate and arguments, as well as informationnot explicit in the text of the purposequestion but related to an important story category. This patternholds true for the contentcategories generatedfor each of the nine purpose questions: In each case, students' responses were first groupedinto categories relatedto the concepts explicitly presented in the purpose question. The other content categories generated related to important slots in the story schema as describedby Stein and Glenn (1979). Table5 illustrateshow the sample protocol was assigned to contentcategories. In this analysis, the numberand type of categories seemed to be affected by the amountof informationpresentedin the purpose question. This was especially apparentin the case of the sixth-gradepurpose question "Whywas Socrates important?" This questioncontainsthe predicate proposition(1) (EXIST, SOCRATES),the qualifying proposition(2) (QUALITY OF, SOCRATES,IMPORTANT),and the causal proposition (3) (CAUSALITY: $, (2)). A fourth proposition relates the question to the testing situation: (4) (DESCRIBE, TEST-TAKER, all the semantic (3)). In termsof story grammar, informationprovidedby this question relatesto the setting. The only other clue that might be used for schema activationis the unstatedcause of Socrates' importance. Thus, the purpose question provides the name of the main topic and an expectation that causal information should be sought during reading. The categorization of the sixth-gradestudents'responses reflected this limited amountof information:This
167

ROWE & RAYFORD

Table4 Relationsof contentcategoriesto wordconcepts, case roles, and story schema categories
Contentcategories About the mannerof getting away About the writer About the city (point of origin) About the writer's destination Why the writer went away Wordconcepts in purposequestion GET AWAY, MANNER OF GETTINGAWAY WRITER CITY (implied) -Case roles expressed/ implied Predicaterelation Conditional Agent Source Goal -Story schema categories Attempt Setting Setting Consequence InitiatingEvent, InternalResponse

wasPurpose doesthiswriter thecity?" forthesecategories Note:Thestimulus 5, "How Question getawayfrom question

Table5 Classificationof a 10th-grade transcriptinto contentand schema elaborationcategories


Contentcategory About the scientists Phase 1 Analysis Propositions (1), (2), (3), (4), (6), (8), (9), (11), (20), (21), (22), (23), (24) (25), (28), (29), (30) (10), (12), (13), (14), (15), (16), (17), (18), (19), (26), (27) (7)

About those who have their minds read About the process of mind reading About the respondents'personalprocessing or test-takingbehaviors

Schema elaboration category Restatements Generalstatements Specific statements

Phase 2 Analysis Propositions (1), (2), (4), (6), (9), (16) (3), (8), (10), (17), (22), (23), (24) (5), (12), (13), (14), (18), (19), (20), (25), (27), (28), (29), (30) (7) (11), (21), (15), (26),

Metastatements

2. Purpose Note: Thefulltranscript is presented inTable 9, Question

"Howare these scientists trying to read your mind?"is the stimulus for these responses.

purpose question generated only one content category,entitled"aboutSocrates." Metastatements. In addition to responses that could be categorizedaccordingto their relation to the substantiveinformationin the purpose questions, some students at each grade level producedmetastatements. Kintschand van Dijk (1978) obtained similar interview responses, which they characterized as "comments on the structure, the content, or the schema of the text"(p. 42). They furthernoted that their protocols often containedother types of metacomments expressing the respondents' opinions or attitudes. took a In the present study,metastatements numberof forms. Some reflected students'metacognitivebeliefs abouttheir abilityto perform the predictiontask (e.g., "No one can say with Others reflected this little bit of information"). their beliefs aboutthe strategiesneeded to complete the task (e.g., "Ican'ttell without reading the story"),or aboutthe resultsof their comprehension monitoring (e.g., "I don't really know who Jules Verne is"). Still anothertype of metastatement was producedin responseto the interview question "Whatwould you try to find In responseto this out as you readthe passage?" probe, several studentsproducedmetalinguistic comments, such as "themain idea,""topicsthey talk about,"and "how to outline it." Although occurredat all three grade levmetastatements els and in response to all nine purpose ques-

168

* Spring 1987 READING RESEARCH QUARTERLY

XXII/2

tions, they constitutedonly a small percentage of the responses. Uncorrected miscues. A third type of response comprised respondents' uncorrected miscues. Such responses were relatively infrequent on seven of the nine purpose questions, but when they occurred, they sometimes led students to activate schemata that were unrelated to the purposequestion. An example is the sixth-grade respondent who substituted water for writer in the question "How did this writer get away from the city?" In this case, the student activated a schema related to transporting water("Ifpeople'sout [of water], they run it in a jug and take it to people"). Responses to the purpose questions for which miscues were common ("Whywas Socrates important?" and "How good were Jules Verne'spredictions?")demonstratethat the effect of miscues on the activationof appropriate schematamay depend on other factors, such as the total amount of semantic informationcontained in the question. In response to both of these purposequestions, studentstendedto miscue on the name of the protagonist.However,as discussed above, the question about Socrates contained much less informationthan the other questions selected for this study.When students were unable to pronouncethe name, they were left with few clues to suggest which schemata would be needed for comprehending the up-, coming passage. In fact, less thanhalf of the respondents ascribed animate status to Socrates. Of those who did indicate that Socrates was a person, only one placed him as an ancient Greek philosopher. The others respondedwith inaccurate information ranging from the sugto the gestion that Socrates was an "architect" that "he was a who did guess guy somethingfor his country." Other studentsactivated schemata related to the inanimate subjects of "soccer," and "secrets?'." Because activation "sauerkraut," of appropriate schematadependedprimarilyon correctly decoding the name Socrates, uncorrected miscues had a detrimentaleffect on students' initial efforts to bring to mind informationrelatedto the theme of the passage. It seems, however,thatmiscues may be less misleading if purpose questions contain a num-

ber of cues to the semantic content of the passage. In reading the purpose question "How good were Jules Verne'spredictions?"a large percentageof the tenth-gradestudents failed to pronounceJules Verne'sname correctly (often Nevidentifying the main characteras "Julie"). ertheless, they were able to instantiatea great deal of informationrelated to predictions and the people who make them. As in the preceding example, only a few studentswere able to activate specific informationaboutJules Verneas a historical figure. However, students were able to activate schematarelatedto other aspects of the question. Thus, uncorrectedmiscues may affect schema activation differently,depending on the total amount of semantic information availablein the purposequestion. Summary The resultsof the initialstageof dataanalysis that helps to anprovidedescriptiveinformation swer the first two researchquestions:(a) Do studentsactivatebackground knowledgein response to purposequestions,and (b) how is thatinformation relatedto the informationpresentedin the purposequestion?First, underthe interviewconditions of this study, studentsat all grade levels were able to use purpose questions to activate about background knowledgeto makepredictions passagecontent,thoughsome questionsprovided moreeffectivecues thanothers.Second,the overall resultsof Phase 1 categorization indicated that the students' to (a) concepts responseswererelated presented in the purpose questions, (b) story schema"slots" implied,but not addressed,in the surfacestructures of the purposequestions,and it appeared (c) the interviewtask itself. Further, that the type of contentinformation activatedin unfamilresponseto purposequestions containing iar conceptsor topics dependedto a greatextent on the number of familiar conceptsalso presented in the purpose question. Uncorrectedmiscues seemedto have a less disruptive influenceon the activationof appropriate schemataif the purpose questionprovideda numberof familiarconcepts whichrespondents coulduse as the basis for makaboutthe passage. ing predictions The Phase 1 analysis also led to the observation that, within the content categories, indi-

Activating background knowledge

& RAYFORD ROWE

169

results. At this point, a Pearsonproduct-moment correlationexceeding r = .94 was obtainedfor our classificationdecisionsacrossthe entiredata set of 2,624 propositions. All disagreements resolved were subsequently aboutcategorization by discussion. The categories identifiedby this procedure were appliedto the responsedata, and numbers of responses falling into each schema elaboration category were tallied for each purpose question. The resultingcontingencytables were then examinedto see if therewere differencesin the patternsof responsesgeneratedfor different purpose questions. Because we observed that some questionsseemed to generatemore of one type of responsethanothers, a chi-squaretest of independencewas used at each grade level to determinewhether the schema elaborationlevels were truly independentof the purposequestions. Finally, as a means of tentatively PHASE2 addressingthe fourth research question, about differences in response patternsbetween grade Procedures Analytic In Phase 2, the purposeof the dataanalysis levels, the contingency tables for the three was to generateand refine a second set of cate- groups were examined visually. More formal was prevented by differences in the gories which would describe the extent of analysis schema elaborationrepresentedby each of the purpose questions, and in the amountof assistance in decoding given at the threegradelevels. predicate propositions identified during Phase 1. This type of analysis requiredthateach proposition be viewed in the contextof the interview from which it was drawn, in order to see what Resultsand Discussion part it played in the student'sinitial envisionment (Langer, 1985) of the test passage. To developthese categories,we workedover Schema Elaboration Analysis The categories generated to describe the the course of several months to devise possible classification schemes that would describe the extent of schema elaboration were not unlike varying degrees of responsespecificity.For this those generated in previous investigations was again divided into (Crafton, 1983). Threeof the four responselevanalysis, each transcript predicate propositions. We independently re- els reflected varying degrees of content specito generatecategoriesthat ficity and were directly relatedto the extent to viewedeach transcript of the which students were willing (or able) to make describedthe extentof schemaelaboration one verbal content predictions and to elaborate on from data the After sorting propositions. their passage envisionments. The fourth level, to these categories, purposequestionaccording we reviewedour coding decisions and set tenta- however,was not directlyrelatedto content, but tive guidelines for furthercategorization.This instead contained students' metacognitive reprocess was repeateduntil one questionat each flections, as described in the Phase 1 analysis. gradelevel had been classified. At this point, the Table5 illustratesthe classificationof the proprefined categories were applied in a final pass ositions in the sample transcriptfirst into conthrough the entire data set. Once again, we tent categories (Phase 1), and then into schema worked independently, and met to compare elaborationcategories (Phase 2).
170 * Spring 1987 READING RESEARCH QUARTERLY XXII/2

vidual data units could be characterized in anotherway. Thatis, they seemed to vary in the extentto which the studentshad chosen to move past global schemata to produce predictions containing specific, detailed scenarios. It was further observed from the students'metastatements that they found predictionmore difficult in response to some purpose questions than to others. Together,these observationsled to the working hypothesis that the extent of schema elaborationin the responsesmight vary by purpose question. To test the accuracy of this hypothesis, we undertook a second phase of analysis in which we recoded the datato reflect a second set of categoriesand examinedthe distribution of these categories across purpose questions.

Restatements.The first schema elaboration category,restatements,was composed of propositions which either restatedor paraphrased the purpose question. An example of a restatement for the purpose question "Howare these scientists trying to read your mind?"is "aboutsome scientists who are reading. . .your mind" (see Table 2). Students occasionally provided only restatements,but more frequentlyused this type of responseto begin their story predictions. General statements. The second schema elaboration category, general statements, was differentiated from a restatementbecause the studentprovidedsome informationnot directly stated in the text of the purpose question, but this informationremainedat a global level similar to that of the purposequestionitself. An example of a general statement drawn from the same student'stranscript is "about some men who made up ways to read [minds]." In terms of schema theory, such responses might be described as "slots" waiting to be filled during reading. Specific statements. The third category, specific statements,comprised responses offering new informationsuch as names, dates, examples, descriptions,and other informationnot contained in the purpose question. One of the propositions in Table 5 that was coded in this category reads "[scientists]who work in laboratories."In this case, the student provided specific informationabout the schema activatedto interpretthe reference to scientists in the purpose question. Metastatements.The fourthcategory of responses was composed of metastatements. These responses did not contain prediction inawareformation,but insteadreflectedstudents' ness of their own processing capabilitiesand of the demandsof the interview situation. The Table 2 transcript contains only one metastatement, "I guess I really don't know,"but other transcriptscontained statementsthat were even more specific: "I don't know anything about or "Youhave to read the story Soecrates [sic]9" to know what this is about." Tests of Independence Accordingto resultsof the chi-squaretests, the hypothesis of independence between
Activatingbackgroundknowledge

schema elaboration levels and purpose questions shouldbe rejectedat the 6th- (X2= 74.65, df = 6, p < .001) and 10th-gradelevels (X2= 37.85, df = 6, p < .001), but should be accepted for the 1st-gradestudents (x2 = 12.24, df = 6, p s .10). Observedand expected values for each grade level are given in Table 6. Because we did not enter the study with a priori predictionsaboutthe distributionof the schema elaborationcategories, expectedvalues for each cell were calculatedby multiplyingthe row frequency by the column frequency and dividing by the total numberof responses (Hays, 1981). Response Patterns Inspection of the first-gradedata indicates a similar thatthis sample of childrenmaintained pattern of responses across the three purpose questions. Overall, they produceda large number of specific responses to each purpose question, and a smaller number of restatements, general statements, and metastatements. This finding confirmed our observations that firstgrade subjectstendedto tell stories when asked to predictpassage content. Quite often they provided very detailed accounts relatedto the purpose question. For example, when one first-grade respondent was given the question "How can you cross a street safely?"she took nearly five minutesto recountthree instancesin Table 6 Observed (and expected) frequencies for schema elaborationcategories
SchemaElaborationCategory General RePurpose question statement statement 1 2 3 17 (22.3) 24(20.5) 21(19.3) Specific Metastatement statement

Grade 1 6 (14.0) 223 (208.8) 12 (12.9) 15 (12.9) 188 (191.8) 10(11.9) 18 (12.1) 170 (180.5) 14 (11.8) 181(151.9) 180(170.3) 96(134.8) 20 (32.9) 18(36.9) 61(29.2)

4 5 6 7 8 9

Grade6 52 (47.9) 102 (122.3) 45(53.7) 155(137.1) 47 (42.5) 111(108.5)

Grade 10 33 (39.4) 112 (95.7) 101 (118.1) 27 (19.8) 43(42.5) 72(103.4) 165 (127.6) 14 (21.4) 45(39.1) 109 (95.0) 97(117.3) 20(19.7)

&RAYFORD ROWE

171

which she had attemptedto cross busy streets with her mother,a friend, and her two brothers. Personal narrativeswere not the only type of stories generatedby the first-gradechildren in response to the predictiontask. Several students respondedwith stories told in "booklanguage," complete with dialogue carriers, and deliveredin word-by-word intonationcharacteristic of many beginning reading groups. Some of the first-grade children demonstratedjust how strong the influence of instructionwas on their responses to the interviewtask when they inserted familiar characters and events from their basal readers into their story predictions. In general, the first-gradestudentsrespondedto all three purpose questions by telling detailed stories, and most of the propositions in these stories were categorizedas specific statements. The type of response made by 6th- and 10th-grade students, on the other hand, was found to be dependenton the purpose question used. Inspectionof the 6th-gradedata indicated that studentresponses on Purpose Questions 4 and 5 followed similarpatterns,with the greatest number of responses falling in the general and specific categories, a moderatenumber in the restatementcategory, and relatively few in the metastatement category. However, Purpose Question 6, "Why was Socrates important?" showed a very differentpattern.Here the numbers of restatements and general statements were close to the expectedvalues, but the number of specific statementsobserved was much lower than expected. The reverse was true for in responseto PurposeQuestion metastatements 6; that is, the observed number of metastatements far exceeded the expectation. Thus, Question 6 seemed to be particularlyimportant in establishing the dependence of the schema elaboration patterns on the purpose questions used as stimuli. Inspection of the 10th-grade data reveals that Purpose Question 10, "Whydoesn't Philip have friends?" had a similarimpacton the overall chi-squarevalue for this group. Tenth-grade studentsgenerallyrespondedwith similar numbers of general and specific statements,except in the case of Question 10, where specific statements were more than twice as numerous as
172

general ones. Thus, in this interview task, the characterof the purpose question had a significant impacton the extent of schemaelaboration in 10th- and 6th-graderespondents' initial envisionmentsof the test passages. These resultsallow some tentativeobservations about similarities and differences across the three grade levels. Overall, the two older groups' responses to the prediction task were similar, in that they tendedto producea considerable number of general and specific statements for each purpose question, althoughthe natureof the questions sometimes affected this pattern. These older groups provided general statementsoutlining topics that might be covered in the test passage, as well as some possible scenarios that could be included. First-gradestudentsalso activatedrelatedschemata, but their responses were more frequently imaginative or personal narrativesrather than general topic statements. Relations between Purpose Question Format, Background Knowledge, and Response Patterns The most importantfinding of the Phase 2 analysis concernsthe associationof the 6th- and 10th-gradestudents'responseswith the purpose questions used. It is clear from the patternsof responsesobservedthatthe topic and amountof informationcontained in the purpose question affected the extent of schema elaborationin the responses of these older students. Close examinationof the protocolsindicatesa complex relation betweenthe amountof informationpresent in the purpose question and the students'backgroundknowledgeaboutthe topic. This complexity is demonstrated by a comparison of the responses of the 6th-grade students to Question 6, "Why was Socrates important?"and those of the 10th-grade students to the question "How good were Jules Verne'spredictions?" In both cases, the number of miscues on the propername and the number of metastatements concerningthe identityof the subject of the passage indicatedthat many students had little accessible informationaboutthe persons named. Based on this observation it mightbe predictedthatthe patternsof responses
XXII/2

* Spring 1987 READING RESEARCH QUARTERLY

for these two questions would differ from the pattern of responses for questions containing more familiar content. This was not the case, however.The usual 6th-gradepatternof schema elaboration was disrupted for the "Socrates" question, but the 10th-gradestudents'responses to the "Jules Verne"question were similar to those for the questionsabout"mind-reading." Examinationof the protocols indicatedthat the 10th-gradestudentshad enough information available from cues in the purpose question to infer that Jules Verne was a person, and to instantiateschematarelated to the act of predicting. The presence of two capitalizedwords may have been important in evoking a "name" schema. Not one student suggested that Jules Verne might be an object. The 6th-grade students, on the other hand, were left with few clues to passage contentif they could not immediately activate schemata related to Socrates. The purpose question provided no other clues on which they might base their predictions. Thus, it seems that familiaritywith the specific topic presented in the purpose question is not the only factor affecting students' responses. The extent of schema elaboration in students' initial predictions about an upcoming passage may also be affectedby the formatand the number of clues to passage content included in the purpose question. Support for this hypothesis comes from Langer's (1985) research on the natureof the envisionmentsreadersconstructas they read test passages. She suggests that three sources of textual information simultaneously influence readers as they develop their envisionments of a passage: (a) the textual cues to genre, (b) the content that is introduced throughoutthe text, and (c) the linguistic material of the text itself, including such featuresas syntax and vocabulary. Evidence for the interactionof background knowledgeand the form and contentof purpose questions is also providedby the patternof responses of the 10th-gradestudentsto the question "Why doesn't Philip have friends?"It was this question that was most influential in producing a significant association between the purpose questions and the schema elaboration categories for this grade level. Here, students
Activatingbackground knowledge

produced more specific scenarios than expected, and fewer general statements. Once again, it is possible to hypothesize that backgroundinformationand question format played an importantrole in the types of responses observed. The friendship topic introducedin this question seemed to activate a whole range of familiarscenarios from the students' lives. As one of the respondentsnoted, "Everyone knows a guy like Philip." And because the question was not writtento specify a particular Philip, the studentswere free to create their own narratives - some personal, some In because the wording addition, imaginative. of the question seemed to indicate that the test ratherthan an passage might containa narrative expository text, students may have felt more freedom in making specific predictions about the contentof the text. Thoughthe 1st-gradechildrentendedto respond similarly to the three purpose questions at their grade level, the same interaction between backgroundknowledge and text features may partiallyaccount for their strong tendency toward personal and imaginarynarratives. All of the purpose questions selected from the primary level of the MAT6 contained concepts which should have been familiar to most sixyear-old children in the region of the country where this study was conducted. Crossing streets, planting trees, and telling jokes are all subjectslikely to be in their background experience, and because of the help we gave them in readingthe questions, miscues were not a problem for this youngest group. In addition, these questions were written to suggest that the upratherthan coming passage wouldbe a narrative an expository passage. It is possible that this combinationof purpose question featurestends to evoke specific schematawhetherthe respondents are 1st- or 10th-gradestudents.

GENERALCONCLUSIONS
This study was designed as an initial exploration of how readersat threegradelevels activate schemata in response to prepassage purpose
173

ROWE & RAYFORD

questions on reading comprehensiontests. Because test-makershave suggested that purpose questionsmay act as cues to passage content, it seemed importantto explore the initial envisionments students form after reading purpose questions, and to examine the relationbetween their responsesand the purposequestions. The resultsof this studyindicatethata broad canuse purposequestions as age rangeof students cues to activatebackground knowledge,and thus generallysupporttheir value as prepassagecontent cues. However, a secondfindingof this study is thatpurposequestionsmay differconsiderably in theirvalue as cues, depending on how they are written.It appearsthat the type of envisionment which is formedin responseto purposequestions is a complex transactionbetween the reader's availbackground knowledgeandthe information able in the purpose question itself. Though the methodologyemployedin this exploratory study does not allow us to drawfirm conclusionsabout the natureof this reader/text our obtransaction, servationssuggest that readerfactorssuch as familiarity with the topics introduced, and text factors such as the amountof informationprovided or the presenceof clues to the genre of the upcomingpassage,deservefurther investigation. Of the purposequestionsused in this study, those that presented familiar topics and provided several pieces of informationseemed to evoke more elaboratedschemata.Althoughthis study was not designed to determine whether the specificity of activated schemata in initial envisionmentspositively or negatively affected comprehension, one hypothesis is that as long as readersremainopen to revision of their initial predictions, the activation of specific scenarios might be helpful. By virtue of their detail, these scenarios should activate a whole networkof relatedschematathat might be helpful in comprehending the passage. In this study, when purpose questions presented unfamiliar topics and little substantiveinformation, readers used the strategy of limiting themselves to global predictions, which indicates that they probably received little help from this type of purpose question. Therefore, although there will always be individual differences between students' envisionments due to differences in
174

backgroundknowledge and cognitive processing, the findings of this study suggest that test developersshouldconsiderthe text factorsmentioned above if they wish to enhance the cuing functionof purposequestions. The conclusions that can be drawn about test-takers' responsesto purposequestionsfrom this studyare subjectto a numberof limitations. First, and most important,this study involved an interviewtask in which studentswere asked to talk about the local envisionmentsthey constructed after reading the purpose questions. Whereas this may be an effective way to find out what studentsthinkwhen they focus specifically on the purposequestion, it does not necessarily reflectthe way readerswouldmakeuse of these questions in a testing situation. Under usual test conditions, in which purpose question, passage, and items are all availablefor inspection, there is no guarantee that students would readthe purposequestionfirst, or stop to develop theirpredictionsaboutthe passage. Because the procedureused in this study was designed to get as much information about activated schemata as possible, the resulting protocols may reflect a more developed envisionment than would be expected under the usual testing conditions. In addition, because readers usually encounter purpose questions and comprehensiontest passages together, it is likely that the initial envisionmentsconstructed after reading the purpose question are altered almost immediately as students begin to read the rest of the text. Whereasthis study provides insight into readers'initial activation of backgroundknowledge, of particularimportancefor further research is observation of how testtakersactuallybuild on these initialunderstandings as they read the passage and answer questions in a testing situation. A second limitationof this study relates to its attempt to make some observations about differences between groups of older and younger students. Although the similaritiesand differences between the responses of the three age groups are intriguing,cross-gradecomparisons based on data from this study must be considered extremelytentative.There are a numberof factors, such as differentialamounts of experiXXII/2

* Spring 1987 READING RESEARCH QUARTERLY

ence with school tasks and with differenttypes edge in both assessment and instructional of texts, which almost certainly affected stu- contexts. dents' responses to the interview task. Thus, more work is needed before conclusions can be REFERENCES drawn about the differences between older and ANDERSON, R.C., REYNOLDS, R.E., SCHALLERT, D.L., & E.T. (1977). Frameworksfor comprehending GOETZ, younger students' responses to purpose quesdiscourse. AmericanEducationalResearchJournal, 14, tions. In future studies, it will be importantto 367-382. gather more information on students' under- BOYD, W.M.(1973). Repeatingquestions in prose learning. Journal of EducationalPsychology, 64, 31-38. standingsof the researchsettingand its relationBRANSFORD, J.D., & MCCARRELL, N.S. (1974). A sketch of a It will to their other school experiences. ship cognitive approach to comprehension:Some thoughts also be importantto examine the purpose quesabout understandingwhat it means to comprehend. In tions and passages at each level of the test for W. Weimer& D. Palermo (Eds.), Cognitionand symbolic processes (pp. 189-229). Hillsdale, NJ: Erlbaum. differences with regard to topic familiarity, L.K. (1983). Learningfrom reading: What hapamount of information presented, and genre CRAFTON, pens when studentsgenerate their own backgroundinclues. formation?Journal of Reading,26, 586-593. This study is only a first step in providing FAW, H.W., & WALLER, G. (1976). Mathemagenicbehaviors reader-based analysis of the comprehension and efficiency in learningfrom prose materials:Review, critique, and recommendations.Reviewof Educational processes in which studentsengage as they enResearch, 46, 691-720. counter purpose questions on comprehension FEATHERS, K. (1984). Semantic features of text, their interthe tests. Although it has highlighted imporaction and influence on reading behavior Unpublished tance of the transactionbetween readers'backdoctoraldissertation,IndianaUniversity,Bloomington. ground knowledge and features of purpose FRASE, L.T. (1967). Learning from prose material: Length of passage, knowledge of results, and position of quesquestions, futurestudies are needed to describe tions. Journal of EducationalPsychology, 58, 266-272. the differenttypes of relationshipsbetweenpurGLASER, B., & STRAUSS, A. (1967). The discovery of pose questions and passages, and then to invesgroundedtheory.Chicago: Aldine. R. (1980). Interviewing: tigate the effects of different types of GORDON, Strategy,techniques,and tactics. Homewood, IL: Dorsey. question-passage relationships on students' J.C. (1985). Portraitof a new paradigm:Reading comprehension and test performance. An im- HARSTE, research. In A. Crismore (Ed.), Landcomprehension the effect of portantnext step in understanding scapes: A state-of-the-artassessment of reading comadding purpose questions to comprehension prehension research, 1974-1984 (pp. 12:1-24). tests is to follow students'developing envisionBloomington: Indiana University, LanguageEducation Department. ments as they read the purpose question, read & FEATHERS, K. (1979). A propositionalanalthe passage, and finally, answer the test ques- HARSTE, J.C., ysis of Freddie Miller, Scientist. Bloomington:Indiana of that intions. The results this study suggest University,LanguageEducationDepartment. terview techniques may be effective in HAYS,W.L. (1981). Statistics (3rd ed.). New York: Holt, Rinehart& Winston. gathering detailed informationon students'unP.H. (1983). Reading comprehension assessderstandingsof test passages and thatthe gener- JOHNSTON, ment: A cognitive basis. Newark, DE: International ation of descriptivecategories groundedin this ReadingAssociation. data may be an effective method for analyzing KINTSCH, W., & VAN DIJK, T.A. (1978). Towarda model of text comprehensionand production.Psychological Rethis qualitative data. Such studies should inview, 85, 363-394. of the cognitive proccrease our understanding esses involved in answering different types of LANGER, J.A. (1980). Relation between levels of prior knowledge and the organization of recall. In M.L. questions and, more specifically, provideinforKamil & A.J. Moe (Eds.), Perspectivesin reading remation on the effects of prepassage purpose search and instruction. Twenty-ninth yearbook of the National ReadingConference(pp. 8-33). Clemson, SC: questions on the comprehension strategies of NationalReadingConference. readers of differentages and abilities. Such inLANGER, J.A. (1981). From theory to practice: A prereadformation is needed to develop guidelines for ing plan. Journal of Reading, 25, 152-156. in that are effective writing prequestions help- LANGER,J.A. (1985). Levels of questioning: An alternative knowlview. ReadingResearchQuarterly,20, 586-602. ing readersactivaterelevantbackground
& RAYFORD ROWE 175

Activating background knowledge

LANGER, J.A., & NICOLICH, M.

(1981). Priorknowledgeand its relationshipto comprehension.Journal of Reading Behavior, 13, 373-379. Beverly Hills: Sage.

D.W.(1986). Does research supportthe use of purROWE, tests?Journal pose questionson readingcomprehension of EducationalMeasurement,23, 43-55.
STEIN, N.L., & GLENN, C.G. (1979).

LINCOLN,Y.S., & GUBA, E.G. (1985). Naturalistic inquiry. P.D. (1985). Changing the face of reading comPEARSON,

An analysis

of story

prehensioninstruction. The Reading Teacher,38, 724738.


PRESCOTT, G.A., BALOW, I.H., HOGAN, T.P., & FARR, R.C.

(1985). Metropolitan Achievement Tests. Cleveland, OH: The Psychological Corporation.


PRESCOTT, G.A., BALOW, I.H., HOGAN, T.P., & FARR, R.C.

AchievementTests, ResearchEdi(1985). Metropolitan tion. Cleveland,OH: The Psychological Corporation.


J.P. (1979). RICKARDS, Adjunct postquestions in text: A

comprehensionin elementaryschool children. In R.O. Freedle (Ed.), Advances in discourseprocesses: Vol2. New directions. Norwood, NJ: Ablex. SWENSON, I., & KULHAVY, R.W. (1974). Adjunct questions and the comprehensionof prose by children.Journalof EducationalPsychology, 66, 212-215. ? Mimeois a purpose question B. (1984, July). What TONE, graphedpaper.Cleveland,OH: The Psychological Corporation.
TURNER, A., & GREENE, E.

(1977). The constructionand use of a propositional text base (Tech. Rep. No. 63).
Boulder, CO: Institute for the Study of Intellectual Be-

critical review of methodsand processes. Reviewof EducationalResearch, 49, 181-196. RICKARDS, J.P., & DIVESTA, E.J. (1974). Type and frequency of questions in processing textual material. Journal of EducationalPsychology, 66, 354-362.
E.Z. (1966). Learning from written instructive ROTHKOPF,

havior.
WATTS, G.H., & ANDERSON, R.C. (1971). Effects of three

types of inserted questions on learning from prose. Journal of EducationalPsychology, 63, 387-394.
WILHITE,S.C. (1983). Prepassage questions: The influence of structural importance. Journal of Educational Psy-

material:An explorationof the controlof inspectionbehavior by test-like events. American Educational Research Journal, 3, 241-249.
M.J. (1975). Relevance and ROTHKOPF, E.Z., & BILLINGTON,

chology, 75, 234-244.

WILHITE,S.C. (1984). Hierarchical importance of pre-passage questions: Effects of cued recall. Journal of Read-

similarity of text elements to descriptions of learning goals. Journal of EducationalPsychology, 67, 745-750. E.E. (1967). Selective facilitaROTHKOPF, E.Z., & BISBICOS, tive effects of interspersedquestions on learning from written materials. Journal of Educational Psychology, 58, 56-61. ROTHKOPF, E.Z., & KAPLAN, R. (1972). Explorationof the effect of density and specificity of instructionalobjectives on learningfrom text. Journal of EducationalPsychology, 63, 293-302.

ing Behavior, 16, 41-59.

Footnote The authors would like to acknowledgethe contribution of the MAT Fellowship Program at Indiana University, sponsoredby The Psychological Corporation,in providing supportfor this study.

176

READING RESEARCH QUARTERLY

* Spring 1987

XXII/2

Das könnte Ihnen auch gefallen