DOCUMENT RESUME

ED 024 543 24 RE 001 489

By- t to, Wayne; Barrett, Thomas C.Iwo Studies of Children's Ability to Formulate and State a Literal Main Idea in Reading. Report from the

Reading Project.Wisconsin Univ., Madison. Research and Development Center for Cognitive Learning.Spans Agency-Office of Education (DHEW), Washington, D.C. Bureau of Research.

Repor t No- WRDCCL- TR 57Bureau No- BR- 5- 0216Pub Date Jun 68Contract- OEC- 5-10-154Note- 32p.EDRS Price MF-$0.25 HC-11.70Descriptors-Concept Formation, *Elementary School Students, *Reading Comprehension, *Reading Processes,

*Reading Research, Reading SkillsThe two reported studies examined children's approaches to and success in

conceptualizing a literal main idea in reading. The first study examined elementary

pupils ability to formulate a main idea for brief, carefully controlled paragraphswritten with one specific but unstated main idea. The study revealed that although

subjects' grade placement and paragraph readability were critical factors in

determining response quality, the children's main idea responses were generally of low

quality. In the second study second- and fifth-grade students were asked toformulate hypotheses about the main idea after each successive sentence of aparagraph was presented. This study revealed that relatively few subiects weresuccessful in formulating a high level main idea statement and that children may haveno clear conception of what a main idea ought to be. It was suggested thatsystematic teaching designed to channel pupils energies in formulating "am ideastatements would yield worthwhile results. Background information, methodologyemployed, and paragraphs used in the study are included. (RT)

Technical Report No. 57

TWO STUDIES OF CHILDREN'S ABILITY TO FORMULATE AND STATEA LITERAL MAIN IDEA IN READING

By Wayne Otto and Thomas C. Barrett

Report from the Reading ProjectWayne Otto and Karl R. Koenke, Principal Investigators

Wisconsin Research and DevelopmentCenter for Cognitive Learning

and theLaboratory for Research in Basic Skills

The University of WisconsinMadison, Wisconsin

June 1968

U. S. DEPARTMENT OF HEALTH, EDUCATION& WELFARE

OFFICE OF EDUCATION

THIS DOCUMENT HAS BEEN REPRODUCED EXACTLY AS RECEIVED FROM THEPERSON OR ORGANIZATION ORIGINATING IT. POINTS OF ViEW OR OPINIONSPOSITION OR POLICY,

STATED DO NOT NECESSARILY REPRESENT OFFICIAL OFFICE OF EDUCATION

The research reported herein was performed pursuant to a contract with the United States Office of

Education, Department of Health, Education, and Welfare, under the provisions of, the CooperativeResearch Program.

Center No. C-03 / Contract OE 5-10-154

PREFACE

This technical report is a joint product of the reading project of theResearch and Development Center for Cognitive Learning and the Laboratoryfor Research in Basic Skills of the University of Wisconsin. The readingproject is within Program 2, Processes and Programs of Instruction, at theCenter. The purpose of the program is to improve educational practicethrough the application of knowledge to instructional problems within disci-plines, such as reading.

This report deals with two studies which investigate the ability ofyoung children to formulate and state the main idea in a paragraph. Both

studies deal with this general problem. Study I examines this ability acrossgrade levels (grades 1-6); Study II is more explicit and deals with theemergence of this ability in a sentence by sentence presentation of eachparagraph for children of Grades 2 and 5.

This report is a good example of the research being done on instructionalprocesses at the University of Wisconsin. It demonstrates methodologicalprocesses for investigating instructional problems. It presents descriptiveinformation about school children and how they respond in to importantinstructional tasks. And, it illustrates the caliber of research which cancooperatively be undertaken by two research groups at the University ofWisconsin.

Thomas A. RombergDirector of Program 2

I

,

CONTENTS

List of Tables and Figures

Abstract

70t,

Page

vii

ix

I. Introduction 1

Background of the Studies 1

Preparation for the Specific Studies 2

Study I 4

Method 4

Results 7

Discussion 8

III. Study II 10

Method 10

Results 11

Discussion 15

Appendixes 16

References 25

v-4 r

ACKNOWLEDGMENTS

A number of people have contributed materially to the planning, imple-mentation and reporting of the studies reported here.

Professor Theodore Harris provided the initial impetus and thoughtfulguidance through all of the planning and execution stages. Professors JohnKean, James Macdonald and G. Lawrence Rarick contributed their time,ideas, and guidance as the studies were designed and piloted. Their con-tribution has been substantial and they are responsible for much of whateverstrengths these studies have.

Several research assistants also participated actively and productivelyin the planning and execution of the studies. Specifically, Jenny Armstronghandled much of the data analysis, Karl Koenke coordinated the data gatheringand was mainly responsible for the main idea rating scale, Judy Stein andBob Jerrolds gathered and organized data , and Susan Tatham put together thefirst draft of this report. Elvira Benter, secretary to the Laboratory for Researchin Basic Skills, kept the entire operation moving smoothly.

The elementary school children who participated as subjects were drawnfrom area schools. The cooperation of central office and building personnel inmaking subjects available and in assisting with the implementation of datagathering efforts is gratefully acknowledged.

Support for two research assistants was provided by the Research Com-mittee of the University of Wisconsin Graduate School. This assistance, too,is gratefully acknowledged.

LIST OF TABLES AND FIGURES

Table Page

1 Analysis of Variance of Main Idea Statements for Two Typesof Paragraphs, Two Styles of Paragraphs, and Six GradeLevels of Subjects 8

2 Mean Ratings of Main Idea Statements Across Grade LevelsIrrespective of Type of Style of Paragraphs Read 8

3 Frequency of Mean Response Rating by Sex and ParagraphType for Primary and Intermediate Subjects 9

4 Intercorrelations of judges Ratings 1 1

Recurring Response Patterns Across Grades Two and Five 1 2

6 Across Grade Frequency of Response Patterns with One orMore Category Six Ratings 1 2

7 Percent of Response Patterns with a Category of Response(C6) in Four Descriptive Categories 1 3

8 Recurring Response Patterns by Grade 13

9 Within Grade Frequency of Response Patterns with One orMore Category Six Ratings 1 4

1 0 Percent of Response Patterns with a Category 6 Response(C6) in Four Descriptive Categories by Grade Level 15

Figure

1 Mean Rating Given to Each Response: Second Grade 1 4

2 Mean Rating Given to Each Response: Fifth Grade 1 4

vii

,

41071.011..

ABSTRACT

The two studies reported here were the first in a series devoted to the

examination of children's approaches to and success in conceptualizing a

literal main idea in reading. Because existing studies include neither con-

cise methodological guidelines nor definitive descriptive data regardingchildren's ability to synthesize and state a main idea, it was necessary tofocus simultaneously upon the development of an operation approach e.g. ,operational definition of main idea, appropriate reading materials, directionsto subjects , a method for evaluating responsesand the collection of des-criptive data. A substantial portion of this report, then, is devoted to meth-odological matters.

The first study was devoted to the examination of elementary pupils'ability to formulate a main idea for brief, carefully controlled paragraphs

written with one specific but unstated main idea. The salient finding was

that although subjects' grade placement and paragraph readability werecritical factors in determining response quality, the children's main idearesponses were generally of low quality as evaluated by the rating scaledeveloped for use in the study. In the second study, subjects were asked

to formulate hypotheses about the main idea after each successive sentenceof a paragraph was presented. The purpose was to examine response patternsin the hope that the descriptive information yielded would lead to clarification

of the cognitive processes involved. On the basis of the pattern analysis,

four descriptive categories of patterns were identified. The categories arediscussed in terms of tentative implications for teaching and research.

ix

INTRODUCTION

The general purpose of the two studies re-ported here was to establish a framework forthe further study of children's approaches tothe task of conceptualizing the main idea inreading. The focus was upon (a) methodologicalconcernsi.e. , devising and testing materials,procedures and scaling methods appropriate foruse with elementary school children and (b)descriptive data i.e. , analyses of children'sactual responses when they tackled the taskof conceptualizing the main ideas inherent inshort, contrived paragraphs.

BACKGROUND OF THE STUDIES

Research on children's ability to read form-ain ideas has generally demonstrated thedifficulty of this task. Alderman (1926) foundthat the lack of ability.to select central thoughtsof a paragraph and organize them logicallyaccording to the writer's purpose was a majorfactor in preventing children from scoring "upto a standard" in comprehension. Keneally(1939) gave a series of controlled passages tosixth graders and found that while 65% couldsupply minor ideas in an outline, only 10%could write original headlines or topics forparagraphs. These results can be questionedon the basis of the adequacy of the preliminarydirections and the criteria used in evaluatingresponses , but the existence of such questionspoints up some of the difficulties in conductingresearch in the area.

Thorndike (1917) analyzed the results of atest given to 200 sixth graders who were di-rected to read a "simple paragraph" and respondto questions following it. The first question,which asked for the "general topic" of the pas-sage, elicited diverse and often irrelevant re-sponses. The investigator concluded that somewords in the directions were overpotent, lead-ing subjects to focus on irrelevant details ofthe task. Furthermore, the appropriateness ofthe passage with respect to its.vocabulary and

structure is highly questionable, as is the As-sumption that a clear central thought was offered.The questions raised by the study point directlyto the need to provide children with unambiguousdirections and appropriate materials.

In an effort to improve reading tests in whichability to conceptualize .a main idea was ignoredor inadequately handled, Woody (1923) con-structed a series of paragraphs graduated indifficulty. Each one was followed by five state-ments in which a thought or partial thought men-tioned in the paragraph was included. By takingstatements almost directly from the text, Woodyhoped to test the subjects' ability to weigh theworth of one statement against the other. Thetest was administered to 1,500 children in thefourth through eighth grades , and on the basisof his data Woody concluded that ability to se-lect a main idea is not well developed in chil-dren, although it does increase across the grades.One confounding factor in his study was thequality of the distractors for each item. Appar-ently it was .difficult to select four equally goodstatements from the short paragraphs used.

Attempts were made in two studies to deter-mine the skills involved in reading for the mainidea. McCullough's (1957) subjects were first,second, and fourth graders to whom she admin-istered the Ginn readiness tests appropriate toeach grade level. Each test included two ques-tions regarding the main idea of the materials.Correlations of the different aspects of compre-hension showed a positive relationship amongthem, the common factor being the reader's"fact-getting" ability. Broening (1941) analyzedreading abilities of a group of secondary schoolstudents and concluded that ability to grasp themain idea in reading is based upon the followingthree skills: (a) noting key words and topic sen-tence clues; (b) differentiating between mainpoints and supporting details; and (c) knowingthe meaning of relational words such as "but,""therefore," and "consequently." The reportdoes not make clear how the author determined

1

the presence of these skills. How, for ex-ample, were key words defined? Perhapskey words for one reader are not the same foranother.

Taken together, the studies that have dealtwith the ability to locate main ideas in readingpresent a clouded and confused picture of theexact nature of the task required of the sub-jects. Methodology has generally been ques-tionable due to overly complex directions,inappropriate materials, and/or an undefinedrationale for including and evaluating severaldiverse tasks. There is a need for researchin which (a) the specific behaviors sought areclearly defined, (b) materials that encouragethese behavioss are provided, and (c) appro-priate methods for evaluating the responsesobtained are utilized.

Furthermore, it seems reasonable to expectthe responses of children at different gradelevels to vary in the degree to which they ap-proximate the necessary synthesis requiredfor an adequate main idea statement. In viewof the present lack of knowledge about chil-dren's actual performance, there is no empiri-cal basis for devising instruction to developskill in this area. Thus, the feeling was thatqualitative analyses of children's main idearesponses might reveal developmental patternsin their cognitive approaches to the task.Assuming such patterns do exist, by knowingthem we would be in a better position to sug-gest improvements in instructional proceduresat specific grade levels.

With these things in mind, the present in-vestigators undertook the two exploratorystudies reported here. The preliminary planfor the studies follows; specifics are givenlater, in the report of each study.

The decision was to focus in Study I uponthe description and qualitative analysis ofchildren's responses when they were requiredto formulate their own main ideas for a selec-tion. The expectation was that analysis acrossgrade levels would yield developmental infor-mation with implications for devising instruc-tional programs. An additional problem forthis study was the preparation of (a) carefullycontrolled materials that elicit unambiguousmain ideas, (b) unambiguous directions fortackling the task, and (c) an instrument forscaling the diverse responses anticipated.

The intent was to use the methods and ma-terials developed in Study I to examine moreclosely, in Study II, the specific point atwhich a reader is able to formulate a mainidea. Here the procedure would be to presentmaterials of paragraph length a line at a timeand to have the subjects verbalize their emeringmain idea formulation at the end of each line.

2

PREPARATION FOR THE SPECIFIC STUDIES

Lack of precedents made it necessary to de-velop and try out materials and procedures be-fore they were used in data gathering situations.

The first step was to develop an operationaldefinition of "main idea" that would provide (a)a consensual framework for the implementationof the studies, (b) guidelines with respect tostructure of the experimental materials to bedeveloped, and (c) guidelines for the develop-ment of a scale to be used in scoring responses.In view of these requirements, it was finallyagreed that an adequate main idea statementwould include two elements: (a) reference toa general topic, as illustrated in the statement"Birds build nests," and (b) a restriction placedon the general topic by reference to the specificcontent of the passage, e.g., "Birds buildnests / in different places." Consequently, acombination of general topic and specific refer-ent was considered a .correct response whenchildren's main idea statements were subse-quently evaluated. This operational definitionis the referent when the term "main idea" isused in the discussion that follows.

The second step was to construct readingmaterials, appropriate for use at the first-through sixth-grade levels, that contained animplicit but unambiguous main idea. Expositoryrather than narrative passages were constructedbecause pilot studies showed that narrative ma-terials tended to evoke divergent rather than con-vergent responses. To further encourage con-vergent responding, the decision was to writeshort paragraphs with a single unequivocal cen-tral thought. As they were constructed, theparagraphs were piloted with adult subjects toget assurance that each one did, indeed, con-tain just one main idea. Three experimentalparagraphs were finally produced.

The subject matter of the paragraphs wasdrawn from the common curriculum area of "na-ture" to assure reasonable appeal across grad.,levels. The procedure was to start with an ac-ceptable main idea statement and then to writea paragraph of four simple sentences, each ofwhich would contain an element of the mainidea, in which the main idea was implicit butnot stated. The first version of each paragraphwas written at the first-grade difficulty leveland subsequent versions were written at in-creasingly more difficult levels, from second-through sixth-grade. The Spache ReadabilityFormula (1953) was used in writing the materialsfor grades 1-3, and the Dale-Chall ReadabilityFormula (1948) was used for Grades 4-6. Sen-tence complexity was manipulated by increasingthe use of phrases and clauses at the upperlevels. The materials were pilot tested several

times with children at different grade levels.As expected, some of the words in the first-grade paragraphs seemed to pose special dif-ficulties despite the control imposed by thereadability formula. In these cases, easierwords were substituted. At the conclusion ofthe pilot testing, the materials appeared to begraded appropriately for each level and to sat-isfy the requirements of the operational con-trols.

Contrary to the practice in much of the ex-isting research, where the procedure was tohave subjects select responses in a multiple-choice situation, the fact that each paragraphcontained a main idea that was implicit butnot stated required that the subjects formulatetheir own main idea statement. The feelingwas that this requirement would not only elimi-nate the possibility of guessing but also yieldmore information about children's behavior asthey attempt to formulate a main idea.

The third step was to devise a method forscoring the responses for purposes of descrip-tion and analysis. The approach was to developa scale that could be used to place responsesinto descriptive categories in order to quantifyresponse quality. Initially an attempt wasmade to categorize responses gathered in pilotstudies on the basis of response generality.The result was a scale that comprised threecategories: (1) responses that were too gen-eral, (2) responses that were too specific,and (3) responses that appropriately combinedthe two elements of the main idea. It soonbecame evident, however, that within each ofthese categories the responses were not homo-geneous and that further subdivisions were re-quired: After several revisions, a scale withseven categories from "no response" at Level0 to "one main idea element correct" at Level4, to "both elements correct" at Level 6wasdevised. Using this final scale, independent

judges were able to categorize all of the re-sponses with a fair degree of agreement.

The final step was to produce a set of di-rections that would (a) provide sufficient infor-mation to enable each subject to understand theexact nature of the task, (b) provide no infor-mation that would amount to training for thespecific task at hand, and (c) be reasonablysuccinct. First, a pilot study was conductedto find out what children actually thought whenthey were asked to find the main idea of a pas-sage. The diversity of responses obtainedunderscored the fact that most children do nothave a clear conception of what a main idea isor how to go about finding one. Specifically,the use of the term "main idea" seemed to im-pose a rigid set approach to the task for manychildren. Therefore, the term was completelyeliminated from the directions and materialspresented to the subjects. After some furtherpiloting, five sets of directions which variedin length and amount of information providedwere devised and tested. An analysis of theresponses evoked by each set of directions ledto acceptance of the shortest, most concise set.Furthermore, it seemed clear that a brief warmuptask for establishing rapport with each subjectbefore the directions were given would be de-sirable. The warmup task selected, after sev-eral possibilities were piloted, was one in whichthe subject was asked to compose and read fourshort simple sentences. A serendipitous sideeffect was that this introduced the subject tothe idea of responding in sentences to a groupof related sentences.

Thus, the paragraphs, directions, and scor-ing scale were devised, tested, and revised inview of the constraints imposed by the opera-tional definition of main idea and feedback frompilot testing. More specific details are givenwhere appropriate in the following reports of thetwo studies.

II

STUDY I

The general purpose of Study I was to ex-amine the ability of pupils in the latter partof Grades 1-6 to synthesize and state themain idea or brief, carefully controlled para-graphs which were written with one specificbut implied main idea. More specifically, theseven hypotheses that follow were tested.

Hi Subjects' main idea response ratingsdo not differ for paragraphs written at a basallevel (Type II) and at grade level (Type I).

H2 Subjects' main idea response ratingsdo not differ for paragraphs written with gen-eral (Style A) and specific referents (Style B).

H3 Subjects' main idea response ratingsdo not differ by grade level.

H4 No interaction between type and styleof paragraphs is reflected by main idea re-sponse ratings.

H5 No interaction between the type of para-graph and grade level is reflected by main idearesponse ratings .

H6 No interaction between the style ofparagraph and grade level is reflected by mainidea response ratings.

H7 No interaction among type of paragraph,style of paragraph, and grade level is reflectedby main idea response ratings.

METHOD

Subjects

Two hundred eighty-eight children fromGrades 1-6 in an urban,Wisconsin schoolsystem were chosen to serve as subjects.The sample was drawn from three schools lo-cated in diverse but representative socio-economic areas of the city.

Initially, 75 childrenfrom each grade levelwere designated at random. The appropriateclassroom teachers were then asked to identifythose children who, in their judgment, couldread a sample test selection. The assumption

4

was that classroom teachers would be reasonablyaccurate in making such judgments regardingtheir pupils' reading ability. From this pool ofpupils who were expected to have no difficultywith the mechanics of reading, 24 boys and 24girls from each grade level were chosen at ran-dom to participate in the study.

Materials

Paragraphs. When an extensive search failedto yield published materials that met the criterionof well-structured paragraphs containing oneimplicit main idea, the decision was made toconstruct paragraphs. The basic problem thenwas to provide an appropriate reading task forfirst- through sixth-grade subjects. To accom-plish this and to permit subsequent comparisonsof responses across grade levels, the procedurewas to establish general content and a specificmain idea for base level, i.e. first-grade, para-graphs and then to manipulate readability levelacross grades by application of existing reada-bility formulas. Support for the notion that asingle main idea would be appropriate for allsubjects was derived from pilot studies in whichit was clear that upper grade children (1) haddifficulty in synthesizing the elements requiredfor a simple main idea statement, and (2) madeno systematic attempt to go beyond the literalstatement of main ideas.

The approach, then, was to devise main ideasthat could be (1) expressed in first-grade vocabu-lary, and (2) developed in a series of four com-parable and related, but independent, sentences,four sentences having been found to be adequatefor the development of a simple main idea.First, main ideas were formulated from the gen-eral area of "nature," a sufficiently universalcurriculum-interest area to insure at least abase level of meaningful vocabulary among firstgraders; and eventually three main idea state-ments were selected: (A) Animals help thefarmer in different ways; (B) Birds build nests

in different places; (C) Animals use claws fordifferent things. Second, the operational de-cision was to work with a structural formulathat prescribed that four subjects of Lhe samegeneral classificationone for each sentencebe combined with predicates involving dif-ferent tasks or functions.

Each paragraph was initially developed atthe first-grade readability level in accord withthe following guidelines: (1) In line with theaccepted operational definition of the mainidea, each sentence provided information abouta subject performing an act (the general topic)and about the specific nature of the action(the restrictive element). Thus, the main idea"Birds build nests in different places" wasdeveloped in four sentences, each of whichincluded a reference to birds as the subjectand to a specific place where birds build nests.(2) Each sentence was independent except forpronoun antecedents. (3) Each sentence con-tained a single idea that contributed to themain idea.

Because the degree to which children arerequired to synthesize is apt to vary in class-room materials, the decision was to considerthe effect of two constructions of the subjectiveelement of the sentences used. Accordingly,alternate forms of each sentence designatedStyle A and Style B in the study were written.In Style A, the subject of each sentence in aparagraph was the name of the general classbeing discussed, so synthesis was requiredonly by the predicate element. In Style B, thesubject of each sentence was the name of aspecific member of the class being discussed,so both the subject and predicate elementsrequired synthesis. Development of the mainidea "Birds build nests in different places"with Style A and Style B sentences is illustratedin the following schema.

Subject elements

Concept given

Once the base, or first-grade, paragraphswere written, the sentences in each paragraphwere expanded in terms of vocabulary difficultyand/or sentence length to bring the readabilitylevel up to subsequent grade levels, but everyattempt was made to keep the kernal thought ofeach sentence intact. The Spache ReadabilityFormula (1953) for grades 1-3 and the Dale-Chall Readability Formula (1948) for grades 4-6provided guidelines for controlling readability.Due to limitations of the formulae, however, itwas necessary to impose additional controls.Wherever possible, for example, words from theStone List (1957) were used for the primarygrades; but when adherence to the list was notpracticable, phonetidally regular words wereused. Furthermore, care was taken to keepsentence length within reasonable limits.Finally, linguistic structure was arbitrarilyregulated across grades by expanding the sen-tences with additional phrases and clauses.The procedure was arbitrary because the exist-ing formulae make no provision for the controlof this stylistic variable. The structure of thetest paragraphs is illustrated by the followingschema for Paragraph C, which was essentiallyidentical to the other paragraphs except for spe-cific sentence order.

Grade Level

Sentence 1 2 3 4 5 6

1

2

3

4

PPPP

PPPP

PPPP

CPPP

CPCP

The first grade paragraph comprised foursimple sentences with no more than a singlephrase (P) as a modifier; and at subsequent

STYLE A

Some birds (Example 1)Many birds (Example 2)Some birds (Example 3)A few birds (Example 4)

Subject elements

(Example 1)(Example 2)(Example 3)(Example 4)

STYLE B

Robins (Example 1)Bluejays (Example 2)Ducks (Example 3)Woodpeckers (Example 4)

Predicate elementsbuild nests under a roof.like nests in trees.even make nests in tall grass.make nests inside wood fence posts.

Predicate elements

may build nests under a roof.like nests in trees.make nests in tall grass.make nests inside wood fence posts.

5

grade levels additional modifying phrases (P)and dependent clauses (C) were added asshown in the schema.

Because there was no basis for predictinghow much the readability controls would affectthe subjects' ability to recognize and formulatethe main idea of each paragraph, the decisionwas made to have subjects at each grade levelread either base level, first-grade (Type II)paragraphs or paragraphs written for their gradelevel (Type I). The strategy, then, was tocompare the responses to base level and tograde level paragraphs in order to determinewhether the readability level of the paragraphshad a significant impact upon success in formu-lating a literal main idea statement.

Scale Value

6

5

6

4

3

All materials were written in an expositorystyle, for pilot studies had indicated that nar-rative materials evoked divergent rather thanthe desired convergent responses. Thp actualparagraphs written are given in Appendix A.

Response Scale. As already pointed out, thefinal scale employed for rating' main idea re-sponses was developed after a number of pilotscales had been tried and found to be inadequate.The final version, too, was tried out in a pilotstudy, and there was consensus among the in-.vestigators that it was adequate for the presentresearch task. The final seven point scale in-cluded the categories given in the schema thatfollows:

Category DescriptionBoth elements correctly stated.*

One element correctly stated, the other too generally or too specifically stated.e.g. Where birds like to build nests.

How different animals help the farmer.How animals use their claws.What animals use their claws for.Robins , bluejays, ducks , and woodpeckers build nests in different placesHorses , dogs, cats, and cows help the farmer in different ways.Lions, tigers , bears, and cats use their claws for different things.

One element correctly stated.e.g. Animals that help the farmer on the farm.

Animals put nests in different places.Animals use claws.

Irrelevant or incorrect material plus one element correctly stated OR cne elementcorrectly stated and the other too general or specific OR both elements correctly

stated.e.g. How birds make nests.

All the animals help the farmer in the summertime.How animals do and do not help the farmer.Where most birds build nests.

2 One or both elements too generally stated.e.g. Birds or Nests.

Animals or Claws.About animals in the woods.Where birds live.Animals on the farm and what they do.

1 One or both elements too generally or specifically stated plus irrelevant or incorrectmaterial OR one or both elements too specifically stated OR only irrelevant orincorrect material.

e.g. Animals have sharp claws.Birds hide their nests.Reread paragraph or a single sentence.How safe the farmer keeps the farm.

0 No response.

* Synonyms of the verb and of the adjective in the final prepositional phrase are acceptable.

Procedure

Directions. Pilot studies revealed the needfor a brief warmup task to permit the establish-ment of rapport between examiner and subjectand to establish a set among the subjects torespond in complete sentences. The warmuptask was simply to compose and read backfour simple sentences . See Appendix B for theactual directions.

The test directions required the subject toread a paragraph silently while thinking about"what all the sentences together say," aphrase intended to direct him toward a synthesisof all the elements present rather than selectionof a single specific thought. The subject waspermitted to ask for any words which he didnot know. Then he was told to "make up justone sentence in your own words that says whatall the sentences tell you." The directionswere partially repeated between paragraphs(see Appendix B).

Testing. Within each grade, equal numbersof boys and girls were assigned to read para-graphs of each type and style. Each subjectread three test paragraphs, randomly ordered,of a single type and style. Three graduatestudents, each an active participant in theprior pilot studies and the production of mate-rials, served as examiners. One examinerworked in each of the three schools from whichsubjects were selected.

The 288 subjects were tested individuallyin a private room provided by the school. Atesting session began with a warmup task, fol-lowed by oral directions given by the examiner.Upon completion of the first paragraph, thesubject was asked for his response to the mainidea task. The response was written downverbatim by the examiner on a specially pre-pared answer form (see Appendix C). The ex-aminer accepted what the subject said withoutcomment unless there was need for clarificationof the referent given e.g. , "What do youmean by 'they'?" The procedure for readingthe next two paragraphs was identical. Eachparagraph was completed before going on to thenext one. The. entire task took approximatelyeight minutes.

Scoring Responses. Each subject's main idearesponses were coded and typed on mastersheets to eliminate possible bias in judging.Four judges, each of whom had had experiencein developing and working with the ResponseScale, scored the 864 responses independently.If at least two judges did not agree, the re-sponses were returned for reconsideration andrescaling by each judge. Final interjudgeagreement was .79, which was consideredadequate for the requirements of the study.

The mean of the 12 scale values given to thethree responses by four judges comprised thesubject's final score. That is, each of the fourjudges rated each subject's response to Para-graphs A, B, and C, and the mean of the result-ant 12 ratings was the subject's score. Thismean score was used in the analyses of the data.

Design and Analysis of the Data

A 2 x 2 x 6 completely crossed analysis ofvariance design, which contained 24 cells with12 subjects in each cell, was used in the study.The three independent variables were: (1) Typeof paragraph, where Type I was written at gradelevel and Type II was written at first-grade levelin difficulty; (2) Style of paragraph, where StyleA required synthesis of the predicate only andStyle .B required synthesis of both the subjectand predicate; and (3) Grade level of the sub-jects. Although equal numbers of boys and girlsserved as subjects, sex was not considered asan independent variable in the analysis for tworeasons: first, preliminary investigations re-vealed no systematic sex differences in responses;and, second, there was little promise of differ-ential prescriptions by sex if a difference wereto be found. The soundness of this decision isdemonstrated in the discussion of results.

The dependent variable was the responserating for the main idea statements of each sub-ject on a scale that ranged from 0-6.

RESULTS

On the basis of the analysis of variance sum-marized in Table 1 there was reason to rejectHypotheses 1that there would be no differencesdue to paragraph typeand 3that there wouldbe no differences attributable to subject's gradeplacement; but Hypothesis 2 that there wouldbe no differences due to paragraph stylewasaccepted.

Hypothesis 1 was rejected because signifi-cant differences were found when the mean rat-ings of main idea statements produced by sub-jects who read Type I paragraphs were.comparedwith the mean ratings of statements by subjectswho read Type II paragraphs across Style andGrade Level. Type I paragraphs were thosewritten at each of the six grade levels accordingto the Spache and Dale-Chall Readability Formu-las, while the Type II paragraphs were writtenat first-grade level or base level according tothe Spache formula. Subjects who read baselevel paragraphs (mean = 3.8) produced mainidea statements that were rated higher thanthose produced by subjects who read grade levelparagraphs (mean = 3.3) .

Analysis of Variance of Main Idea Statementsfor Two Types of Paragraphs , Two Styles of

Paragraphs, and Six Grade Levels of Subjects

Source ofVariation df MS

Type 1 17.552812 14. 04*Style 1 2.257812 1.81Grades 5 16.546451 13. 23*Type xStyle 1 1.039201 < 1.00

Type xGrades 5 .891729 < 1.00

Style xGrades 5 2.535063 2.02

Type xStyle xGrade 5 .336285 < 1 .00

Error 264 1.249069

*2. < .05

In the case of Hypothesis 3, a significantdifference in mean ratings of main idea state-ments was found across the six grade levelswhen Type and Style of paragraphs were notconsidered. As shown in Table 2, there wasa consistent increase in the mean ratings ofmain ideas given by subjects in Grades 1-6.Scheffg post hoc tests revealed that with theexception of the fourth- and fifth-grade andfifth- and sixth-grade comparisons, the differ-ences between all other grade level meanswere significant at the .05 level.

Table 2

Mean Ratings of Main Idea Statements AcrossGrade Levels Irrespective of Type or Style of

Paragraphs Read (N = 48 at each grade)

Grade Levels

2 3 4 5 6

MeanRatings 2.71 3.12 3.52 3.96 4.05 4.20

Hypothesis 2 was accepted because therewas no significant relationship between thestyle of the paragraph and the mean rating ofthe main idea statements. In other words, thesubjects' ability to formulate the main idea ofa paragraph was not affected by the generalityor specificity of the subjects of specific sen-tences. This was true across Grade Levels andacross Types of paragraphs.

8

Hypotheses 4-7 dealt with the two-way in-teractions of Type x Style, Type x Grades, andStyle x Grades and the three-way interactionof Type x Style x Grades inherent in the design.As shown in Table 1, none of these interactionswas found to .be significant. In other words,there was no significant relationship betweenthe mean ratings of main idea statements andcombinations of Type and Style of paragraphs,Type of paragraphs and Grade, Style of para-graphs and Grade, or Type and Style of para-graphs and Grade.

The data in Table 3 are offered for descriptivepurposes. Response frequencies for the sevencategories of the rating scale are given by gradelevel blocks, paragraph type, and sex. Thelack. of a systematic sex difference is clear,which supports the earlier decision not to con-sider sex in the analysis of variance. The shifttoward higher response-category values withbase level (Type II) paragraphs is also clear.Most interesting is the shift toward higher re-sponse-category values from primary to inter-mediate grades. This is apparent with both baselevel (Type II) and grade level (Type I) para-graphs, which seems to suggest that the ade-quacy of main idea responses is not simply tiedto the readability level of materials.

DISCUSSION

The central purpose of Study I was to examinethe ability of pupils in the latter part of Grades1-6 to synthesize and state the main ideas ofbrief, carefully controlled paragraphs whichwere written with one specific but implied mainidea. Two types and two styles of paragraphswere employed for this purpose. Before present-ing the conclusions and implications of thestudy, it seems appropriate to discuss some ofthe basic limitations inherent in the study itself.

An important limitation derives from the factthat inherent in the operational definition of amain idea in this study were the assumptionsthat (a) the optimal main idea statement is asentence, not a topic or phrase; (b) the optimalmain idea statement includes the general topicof the passage and the specific restrictions ofthe passage; and (c) the general topic portionof the main idea statement is more importantthan the specific portion of the main idea state-ment. These assumptions were implicit in anumber of judgments made by the investigators.The accepted definition also influenced the de-velopment of the main idea rating scale utilizedin the study. For example, the general topicportion of the main idea statement was deemedto be more important than the specific portionof the main idea statement. In other words, theinvestigators made this value judgment which,

Table 3Frequency of Mean Response Ratings by Sex and Paragraph Type for

Primary and Intermediate Subjects

Primary(Grades 1-3)

Type I

Type II

BoysGirls

BoysGirls

Response Category

0 1 2 3 4 5 6

10 6 11 6 4 0

10 10 11 4

4 8 10 10 3

5 12 9 7 3

Intermediate(Grades 4-6)

Type I

Type II

BoysGirls

BoysGirls

2 6 14 7 5

2 4 11 12 7

3 4 3 16 11

1 3 4 15 10

1

1

1

1

in turn, influenced the magnitude of the ratingthat any one main idea statement received.Finally, the paragraphs constructed for theinvestigation were carefully controlled in termsof content, readability and conceptual struc-ture, and length. The obvious limitation isthat the paragraphs were not necessarily rep-resentative of reading materials generallyavailable to children. Whether generallyacceptable "representative" materials canever be found is , of course, a moot question.

With these limitations in mind several con-clusions and implications can be drawn fromthe results of the investigation.

1. Since the mean ratings of the main ideastatements were signif Itly higher when baselevel paragraphs (Type II) rather than at gradelevel paragraphs (Type I) were used acrossgrade levels , it would seem that, in general,the easier the materials are in terms of reada-bility the more adequate children's main ideastatements will be. In other words, the shorterthe sentences and the easier the words in aselection, the less energy the child has toexert in decoding words.and deriving literalmeanings and the more energy he can exert insynthesizing the ideas in a selection and infer-ring, the main idea. The implication seemsclear: instructional programs designed toteach children to formulate main ideas shouldemploy very easy materials.

2. The significant differences among gradesin mean ratings of main idea statements andthe fact that the mean ratings consistentlyincreased in magnitude from Grade 1 throughGrade 6 suggests that the ability to formulatemain idea statements is developmenial in na-ture. However, the facts that the practicaldifferences in mean ratings were not greater

than they were from grade to grade, and thateven at the sixth-grade level the mean ratingwas not particularly high in terms of the scaleused, raise questions about the effectivenessof the instruction c'e -oted to the developmentof this basic comprehe :. sion ability during theelementary school years. It would seem,therefore, that pupils should be given practicein inferring main ideas beginning in the primarygrades and that such practice should be con-tinued over a period of years.

3. Since the style of the paragraphs utilizeddid not significantly influence the mean ratingsof main idea statements made by subjects in thisinvestigation while the level of difficulty of theparagraphs did, there is some evidence that thesemantic structure of materials may not be ascrucial to pupils' formulation of main ideas asthe number of difficult words and the length ofsentences in a selection. This does not meanthat further work on the relationship betweenthe semantic structure of materials and theability of children to state main ideas shouldnot be considered, but it does seem to suggestthat the readability level of the materials couldprofitably be given foremost consideration instudies of this nature.

4. Finally, the fact that there is a paucityof research dealing with main idea comprehen-sion abilities suggests that what has been donein this investigation should be viewed as aninitial attempt in this area. Although the meth-odology of the study was carefully developed, itundoubtedly can be improved upon in the future.Furthermore, the impact of different cognitivestyles and abilities upon performance with amain idea task must be examined before sound,individualized instructional programs can bedevised.

9

III

STUDY II

Study H was undertaken to study the emer-gence of main idea statements as children inGrades 2 and 5 responded to a sentence bysentence presentation of a brief paragraph withan implicit main idea. The purpose was toexamine the characteristics of these responsepatterns both within and across grade levels.The hope was that the descriptive informationyielded by such an analysis of primary andintermediate children's approaches to the mainidea task would lead to clarification of thecognitive processes involved.

Answers to three specific questions weresought:1. Are there general patterns of responding assubjects attempt to formulate a main idea state-ment after reading each subsequent sentencein a paragraph?2. Are there characteristic response patternsamong subjects who are successful in formu-lating a high level (Category 6) main ideastatement?3. Do the response patterns of second andfifth graders differ?

METHOD

Subjects

Sixty second- and 60 fifth-grade pupilsserved as subjects. They were drawn fromthe same urban Wisconsin school system usedin Study I, but three different schools, whichwere presumed to be representative of thecity's socioeconomic areas, were used.

The population comprised all of the secondand fifth graders in the three schools. As inStudy I, teachers were asked to identify thepupils who, in their judgment, were able toread a sample test selection. From this groupof children who presumably would have no dif-ficulties with the mechanics of the readingtask, 30 boys and 30 girls from each gradelevel were randomly chosen to serve as subjectsfor the study.

10

Materials

The directions and paragraphs used in StudyI were modified or restricted for Study II in thefollowing ways: (1) only paragraph B was used,the feeling being that sufficient data would bederived from a single paragraph; (2) to increasethe response opportunities, two sentenceswritten according to the procedures establishedin Study Iwere added to the paragraph, makinga total of six; (3) the paragraph was Type I,first-grade level of difficulty, and Style A, re-quiring synthes'is of the predicate only, to makethe reading task as simple and straightforwardas possible; (4) the six sentences were writtenon separate lines so that one sentence at a -dinecould be exposed to the reader; and (5) the di-rections required the subject to give a main idearesponse after each successive sentence waspresented. See Appendix D for the paragraphand Appendix E for specific directions used inStudy II.

The response scale described in Study I wasused to guide the scoring of the main idea state-ments.

Procedure

DireCtions. Except for the modifications indi-cated..above, the directions for Study I were used.

Testing. Equal numbers of boys and girls withineach grade read the Type I, Style A version ofParagraph B. Three graduate students who hadparticipated in Study I were the examiners. Oneexaminer worked in each school from which thesubjects were drawn. The 120 subjects wereindividually tested in a room provided by theschool. As in Study I, the testing session beganwith the warmup task and was followed by oraldirections from the examiner. The examiner thenheld a card over the test paragraph so that onlythe first sentence was exposed. The subjectwas directed to read the sentence and tell theexaminer what he thought "all the sentencestogether will say." After his response was

written down verbatim by the examiner on aprepared form (Appendix F), the second sen-tence was uncovered. This procedure wasfollowed until all six sentences had been ex-posed; then the subject was asked to statewhat all the sentences together said.

Scoring Responses. The scoring method out-lined in Study I was used in rating the sixresponses from .each subject. In the presentstudy, however, responses were not returnedto the judges for rescaling when no two judgesagreed. There was no single mean score foreach subject as the intent was to examine theentire response pattern.

Design and Analysis of the Data

Identical material and directions were pre-sented to each subject; the independent variable,then, was grade level, with equal numbers ofsubjects from the second and the fifth grades:Although equal numbers of boys and girls servedas subjects, there was no attempt to examinethe data by sex because the results of Study Iindicated no systematic difference in boys' andgirls' main idea responses. Thus there were,in effect, six dependent variables: the responseratings for the main idea statements evoked bythe exposure of each successive sentence inthe six-sentence experimental paragraph.

The basic intent in Study II was to gatherdescriptive information by examining the re-sponse patterns that appeared most frequentlywithin and across grades and by examining theresponse patterns of subjects who at somepoint gave a response that was rated.six, thehighest rating on the scale. To organize thedata for this purpose, the numerical ratings ofeach subjects' six responses were punched ona card, and cards were grouped by patternthrough a process of sequential card sorting.This procedure was followed with each of thefour judges' ratings. In addition, means andstandard deviations for the rated values ofeach subsequent main idea statement withinand across grades were computed for eachjudge.

RESULTS

Because the task of presenting four distinctsets of findingsthe responses of each subjectwere categorized separately by each of fourjudges seemed unwieldy, the decision wasmade to determine whether a single, most rep-resentative judge might be identified with aview toward reporting the results in terms ofhis ratings. Consequently, interjudge consensuswas determined by examining the four ratingsgiven to an arbitrarily chosen sentence. Inter-judge correlations computed for these ratings

are presented in Table 4. The ratings of Judge3 were most highly correlated with the ratingsof the other three judges; thus, the results ofStudy II are reported in terms of the ratings as-signed by Judge 3. The results are presentedin relation to the three general questions posedfor this study.

Table 4

Intercorrelations of judges' Ratings

1 2

Judge

3 4

1 .64 .75 .612 .85 .643 ONO =al .774

General Response Patterns

Although the expectation was that severaldistinct response patterns would emerge fromthe data, preliminary examination of the patternsseemed to reveal about as many patterns as in-dividual subjects. Upon more careful examina-tion, however, it was observed that much of theinconsistency appeared to be attributable towide, and apparently random, variations in re-sponses to the first two sentences within eachpattern. In view of the fact that responses tothe first two sentences necessarily had to beextremely tentative, almost random, main ideastatements, the decision was made to ignorethe responses to the first two sentences and tolook carefully at the patterns that emerged whenthe responses to the last four sentences wereexamined.

The results of the analysis are summarizedin Table 5, where the frequency of occurrenceof each pattern that recurred among the second-and fifth-grade subjects' res-ponses is given.Although the overlap among response patternsis limited, two general observations seem to beworthy of mention. First, the recurring patternsreveal a general tendency to perseverate. Thatis, many of the subjects continued to give iden-tical responses or responses of the same qualityover the last four sentences of the paragraph. c

This tendency in certain instances (e.g. , Pat-terns 5, 6, and 10) appears to have precludedthe attainment of a high scale score by the timethe entire paragraph had been exposed. Second,the most frequently occurring patterns were Pat-tern 15, four consecutive ratings of 5, and Pat-tern 19, four consecutive ratings of 6. Subjectswho exhibited these patterns had succeededearly in formulating highly acceptable main idearesponses and, once having arrived at a high

11

level response, did no further shifting. Per-severation was obviously not a detriment inthe case of Pattern 19.

Table 5

Recurring Response PatternsAcross Grades 2 and 5

Pattern

1. 11112. 11153. 11164. 11665. 12226. 22227. 23338. 311 69. 3311

10. 333311. 44441 2. 535513. 553514. 555315. 55551 6. 555617. 611118. 666519. 6666

Total

Frequency

32

2

2

2

5

2

2

2

62

2

2

2

2222

2

1 2

76

To sum up, a diversity of response patternswas apparent even among the 76 subjects whoresponded with recurring patterns. The processof formulating a main idea statement as addi-tional information becomes available is highlyidiosyncratic. The single generalization thatemerges as the result of pattern analyses seemsto be that individual subjects tend to perse-verate at a response level that is attained quiteearly in the sequence; that is, after a hypothe-sis has been stated on the basis of limited in-formation there is little tendency to change,even when more information becomes available.

Response Patterns with a Category 6 Rating

The frequencies of all response patterns inwhich at least one main idea statement wasplaced in Category 6 of the response ratingscale are given in Table 6. All six sentencesare included in order to provide maximum infor-mation about the response patterns of thesesubjects who were successful in formulatinghigh level main idea statements. Only 40 sub-jects, or a third of the total number, ever gavea Category 6 main idea response; and in 13

1 2

instances the Category 6 response was not thefinal response. Furthermore, there was no con-sensual pattern of responses. The individualsubjects apparently arrived at their high levelresponses in personal, idiosyncratic ways.

Table 6

Across Grades Frequency of Response Patternswith One or More Category Six Ratings

Pattern Frequency

1. 0331162. 1111163. 1111664. 111 6665. 11 66666. 1333667. 1561118. 1566659. 163163

10. 16666511. 1 666661 2. 21111613. 22516514. 24666615. 26611116. 26633617. 26666618. 3111 6619. 31 633320. 31666621. 33311622. 33363123. 33566624. 33666625. 35566526. 36511327. 36666628. 41 261129. 4331 6130. 43355631. 45555632. 46133133. 46634634. 46666635. 55366636. 55555637. 666666

Total 40

The data summarized in Table 6 were reex-amined in order to determine (a) the point atwhich the Category 6 response first appearedin the pattern, and (b) the extent to which theCategory 6 response, once started, was main-

tained in the remaining responses. To sum-marize this reexamination, the patterns weregrouped into four types:

1. Category 6 response (C 6) given, but notas final response.2. C 6 attained early, not held, but returnedto as final response.3. C 6 attained before last response and heldthrough final response .4. C 6 attained on last response only.The results are given in Table 7, where fourgeneral observations can be made.

Table 7

Percent of Response Patterns with a Category6 Response (C 6) in Four Descriptive Types

(N = 40)

Type DescriptionPercent(acrossgrades)

1 C 6 response given, but notas final response 3 2.5

2 C 6 attained early, not held,but returned to as final response 5.0

3 C 6 attained before last responseand held through final response 39.0

4 C 6 attained on last responseonly 23 .5

1. A surprising number of subjects (32.5percent in Type 1 on Table 7) apparently didnot recognize the adequacy of their early Cate-gory 6 response and moved to less acceptablemain idea statements as succeeding sentencesof the paragraph were shown.

2. Few subjects (5 percent in Type 2) wereerratic in giving high level main idea state-ments. It appears that when subjects com-pleted the task with a C 6 response they eitherstated it early and maintained it (Type 3) orarrived at it only after reading the entire para-graph (Type 4).

3. A substantial number of subjects (39percent in Type 3) attained a high level mainidea statement early in the sequence and heldit .

4. Subjects in Type 4 either were unableto arrive at a C 6 main idea statement anysooner than the last response or deliberatelyheld back in their responding until all the in-formation was available. Clarification of indi-vidual strategies might lead to implications forinstruction.

The salient generalization on the basis ofthe Category 6 pattern analyses seems to be

that even the subjects who are successful ingiving a high level main idea response tacklethe task in a rather idiosyncratic manner.

Response Patterns of Second and Fifth Graders

Frequencies of recurring response patternsfor the last four responses are given by gradein Table 8. Despite the generally small numberof repeated patterns within grade, two markedbetween-grade differences are demonstrated.

Table 8

Recurring Response Patterns by Grade

Pattern Frequency

Grade 2 Grade 5

1. nu 3 0

2. 1115 2 0

3. 1116 2 0

4. 1166 2 0

5. 1 222 2 0

6. 2222 3 2

7. 2333 2 0

8. 3116 2 0

9. 3311 1 1

10. 3333 3 3

11. 4444 1 1

12. 5355 2 0

13. 5535 2 014. 5553 1 1

15. 5555 6 1 6

1 6. 5556 1 1

17. 6111 0 2

18. 6665 0 2

19. 6666 1 11

Totals 36 40

1. The preponderance of the low ratingsoccurred in the second-grade patterns, whilefifth graders' responses generally receivedhigher ratings. This finding is, of course, asexpected and it is consistent with the resultsof Study I.

2. There were fewer clusters of consensualresponse patterns among the second graders.That is, the second-grade subjects were fairlyevenly distributed across the nineteen responsepatterns; whereas, the majority of the fifthgraders clustered around two high level patterns,15 and 19.

Returning to a consideration of all 1 20 sub-jects, the mean ratings given to each of the sixsentences are given for Grades 2 and 5, respec-tively, in Figures 1 and 2. Two salient generali-

13

zations based upon both figures are that (a)the second graders' mean ratings are lowerthan the fifth graders' and (b) the secondgraders' performance across sentences is er-ratic, while the plot of the fifth graders' per-formance shows positive acceleration acrosstrials in the manner of a classic learning curve.

4

3

2

1

1 2 3 4 5 6

Sentence

Figure 1. Mean Rating Given to Each Response:Second Grade

1 2 3 4 5 6

Sentence

Figure 2. Mean Rating Given to Each Response:Fifth Grade

Frequencies of patterns in which there wasat least one Category 6 response are summarizedby grade level in Table 9. Fifth graders hadabout twice as many patterns with Category 6

ratings as second graders. When the patternswith Category 6 responses were grouped accord-ing to time of emergence and maintenance ofCategory 6 responses, distinct differencesbetween grade levels became apparent. Thegroupings are summarized in Table 10, whichreveals two major between-grade differences.First, while just under half of the second-grade

14

Table 9

Within Grade Frequency of Response Patternswith One or More Category Six Ratings

Grade 2Pattern Frequency

1. 033116 1

2. 111116 1

3. 111166 1

4. 21111 6 1

5. 225165 1

6. 311166 1

7. 316333 1

8. 33311 6 1

9. 41 2611 1

10. 4331 61 1

11. 433556 1

1 2. 455556 1

13. 466346 1

14. 466666 1

Total 14

Grade 5Pattern Frequency

15. 111666 1

16. 116666 1

17. 133366 1

18. 156111 1

19. 156665 1

20. 1 63163 1

21. 166665 1

22. 166666 2

23. 246666 1

24. 266111 1

25. 266336 1

26. 266666 1

27. 316666 1

28. 333631 1

29. 335666 1

30. 336666 2

31. 355665 1

32. 365113 1

33. 366666 1

34. 461331 1

35. 466666 1

36. 553666 1

37. 555556 1

38. 666666 1

Total 26

patterns contained a Category 6 rating only onthe last response, this was true of only a very

Table 10

Percent of Response Patterns with a Category6 Response (C 6) in Four Descriptive Types

by Grade Level

Type Description Percent

Grade 2 Grade 5

1 C 6 response given ,but not as finalres ponse 29

2 C 6 attained early,not held, but returnedto as final response 7

3 C 6 attained beforelast response and heldthrough final response 21

C 6 attained on lastresponse only 43

35

4

57

4

small portion of fifth-grade patterns . The fifthgraders , then, generally arrived at a Category6 response earlier in the sequence. Secondand this is in line with the first generalizationmore than half of the fifth graders arrived ata Category 6 response early in the sequenceand maintained it, whereas only about one-fifth of the second graders exhibited such apattern.

DISCUSSION

The limitations pointed out in the discussionof Study I also hold for the present discussion.

With the presentation sequence employed inStudy II, the subjects tended to respond moreor less at random to the first two sentences.In the absence of contextual or other relevantcues, random responding to extremely limitedinformation would, of course, be expected.Then, as more information becomes available,the expectation would be for increasingly moreadequate main idea statements to occur. Ingeneral such a trend was observed , at leastamong fifth graders, but the number of subjectswho arrived at the highest level main ideastatements was not great. One implicationappears to be that children might profit frombeing encouraged to formulate a hypothesisabout the main idea of a reading selection very

early in the reading sequence and to continueto revise the hypothesis as long as more rele-vant information is forthcoming. In view ofthe perseveration tendencies demonstrated,emphasis upon the latter appears to be particu-larly desirable.

Considering the entire response patterns ofthe present subjects , consensual response pat-terns were not demonstrated. This lack can beattributed largely to the random responses tothe first two sentences. However, when thepatterns were limited to the last four responsesgiven, several distinct characteristics werenoted. Relatively few of the subjects (i.e. ,

two-fifths of the fifth graders and one-fifth ofthe second graders) were successful in formu-lating a high level (Category 6) main idea state-ment; and among these subjects there were nopredominant response patterns. A surprisingobservation was that a number of children whodid arrive at a Category 6 response at somepoint before the end of the sequence changed itto an inferior response. The suggestion seemsto be that children may have no clear conceptionof what a main idea ought to be. If this is so,then teachers need to help them to develop cri-teria for deciding (a) what a good main ideastatement is , (b) how it is recognized, and (c)how it is formulated.

Despite the diversity of response patternsamong children who managed to formulate aCategory 6 main idea statement while readinga paragraph, the patterns were categorized ac-cording to the point at which the statement wasmade and the degree to which it was maintainedfor the remainder of the task . The four descrip-tive categories devised for this purpose may beuseful in permitting teachers to classify pupilsas to their need for greater accuracy, efficiency,and/or confidence in reading for the main idea.

The more erratic and less successful attemptsof second graders to formulate main idea state-ments may be attributable to either or both ofthe following: (a) general developmental lacksthat preclude high level conceptualizing; and(b) classroom procedures in the early gradesthat do not provide as much guidance in readingfor main ideas as in the upper grades. In eithercase, systematic teaching designed to channelpupils' energies as they tackle the task of formu-lating a main idea statement would probablyyield worthwhile results.

15

APPENDIX A

PARAGRAPHS, STUDY I

PARAGRAPH A, STYLE A

FIRST GRADE

Cats help the farmer keep mice from his corn.A horse helps the farmer work.Cows give milk to the farmer.A dog helps the farmer watch the barnyard.

SECOND GRADE

Cats help the farmer keep his corn safe fromhungry mice.

A horse helps the farmer plow in the spring.Cows give milk to the farmer.A dog helps the farmer watch over the barnyard.

THIRD GRADE

Cats help the farmer keep his corn safe fromgreedy and hungry mice.

A horse helps the farmer plow his fields in the

s pringtime .Cows give milk to the farmer and his large

family.A dog helps the farmer guard the whole barn-

yard when it is night.

FOURTH GRADE

Cats help the farmer keep his corn safe fromgreedy and hungry mice.

A horse helps the farmer plow his wheat fieldswhen spring comes.

Cows give milk to the farmer and his largefamily of six.

A dog helps the farmer guard the whole barn-yard when it is night.

FIFTH GRADE

Cats help the farmer keep his corn safe when-ever greedy and hungry mice try stealinggrain.

16

A horse helps the farmer plow his wheat fieldswhen spring comes.

Cows give milk to the farmer and his largefamily of six.

A dog helps the farmer guard the entire barnyardwhen it is night and people are asleep.

SIXTH GRADE

Cats help the farmer keep his valuable corn safewhenever greedy, hungry mice try stealinggrain.

A horse helps the farmer plow his wheat fieldswhen spring comes.

Cows give very tasty milk when the farmer andhis large family of six need it.

A dog helps the farmer guard the entire barnyardparticularly carefully when it is night andpeople are asleep.

PARAGRAPH A, STYLE B

FIRST GRADE

Some animals help the farmer keep mice from

his corn.Others help the farmer work.Many animals give milk to the farmer.Some animals help the farmer watch the barnyard.

SECOND GRADE

Some animals help the farmer keep corn safefrom hungry mice.

Others help the farmer plow in the spring.Many animals give milk to the faruer.Some animals help the farmer watch over the..

barnyard.

THIRD GRADE

Some animals help the farmer keep his cornsafe from greedy and hungry mice.

GPO 801.-23-4

Others help the farmer plow his fields in thes pringtime .

Many animals give milk to the farmer and hislarge family.

Some animals help the farmer guard the barn-yard when it is night.

FOURTH GRADE

Some animals help the farmer keep his cornsafe from greedy and hungry mice.

Others help the farmer plow his wheat fieldswhen spring comes.

Many animals give milk to the farmer and hislarge family of six.

Some animals help the farmer guard the barn-yard when it is night.

FIFTH GRADE

Some animals help the farmer keep his cornsafe whenever greedy, hungry mice trystealing grain.

Others help the farmer plow his wheat fieldswhen spring comes.

Many animals give milk to the farmer and hislarge family of six.

Some animals help the farmer guard the entirebarnyard when it is night and people areasleep.

SIXTH GRADE

Some animals help the farmer keep his valuablecorn safe whenever greedy, hungry mice trystealing grain.

Others help the farmer plow his wheat fieldswhen spring comes .

Many animals give tasty milk when the farmerand his large family of six need it.

Some animals help the farmer guard the entirebarnyard particularly carefully when it isnight and people are asleep.

PARAGRAPH B, STYLE A

FIRST GRADE

Robins may build nests under a roof.Bluejays like nests in trees.Ducks make nests in tall grass.Woodpeckers make nests inside wood fence

posts.

SECOND GRADE

Robins may build nests under the roofs ofhouses and barns.

Bluejays like their nests in trees.Ducks make nests in the tall grass near other

duck nests .Woodpeckers make nests inside old wood fence

posts.

THIRD GRADE

Robins may build their nests under the roofs ofhouses and barns.

Bluejays like their nests in trees that have manybig branches.

Ducks, however, carefully make their nests inthe wild rice, high weeds, or tall marsh grassnear other duck nests.

Woodpeckers sometimes make nests inside oldwood fence posts.

FOURTH GRADE

Robins may build their nests under the roofs ofhouses, garages, and barns.

Bluejays like their nests in broad trees that havebig branches.

Ducks, however, very carefully make their nestsin the wild rice, high weeds , or tall grassnear other duck nests.

Woodpeckers sometimes make nests that arequite soft and comfortable inside old woodenfence posts.

FIFTH GRADE

Robins may build their nests under the roofs ofhouses, garages, and barns.

Bluejays like their nests in broad trees that havebig branches.

Ducks, however, carefully make nests in wildrice, high weeds, or tall marsh grass thatmay contain many duck and other wild lifehomes.

Woodpeckers sometimes make nests that arequite soft and comfortable inside old woodenfence posts.

SIXTH GRADE

Robins may build their nests under house, garage,and barn roofs where they overhang the building.

Bluejays like their nests in leafy trees that havebig branches .

Ducks, however, carefully make nests in thewild rice, high weeds, or tall marsh grassthat may contain many duck and other wildlife homes.

Woodpeckers sometimes make nests that arequite soft and comfortable inside old woodenfence posts.

17

PARAGRAPH B, STYLE B

FIRST GRADE

Some birds build nests under the roof.Many birds like nests in trees.Some even make nests in tall grass.A few birds make nests inside wood fence posts.

SECOND GRADE

Some birds build nests under roofs of housesand garages and barns.

Many birds like their nests in trees.Some even make nests in tall grass near other

bird nests.A few birds make nests inside wooden fence

posts.

THIRD GRADE

Some birds build their nests under the roofs ofhouses and garages and barns.

Many birds like nests in trees that have bigbranches.

Some, however, carefully make their nests inthe wild rice, high weeds, or tall marshgrass near other bird nests.

A few birds make nests inside old wooden fenceposts.

FOURTH GRADE

Some birds build their nests under the roofsof houses, garages , and barns.

Many birds like nests in broad trees that havebig branches.

Some, however, very carefully make their nestsin the wild rice, high weeds, or tall grassnear other bird nests.

A few birds make nests that are soft and com-fortable inside old wooden fence posts.

FIFTH GRADE

some birds build their nests under the roofsof houses, garages, and barns.

Many birds like nests in broad trees that havebig branches.

Some, however, carefully make nests in wildrice, high weeds, or tall marsh grass thatmay contain many bird and other wild lifehomes .

A few birds make nests that are soft and com-fortable inside old wooden fence posts.

SIXTH GRADE

Some birds build their nests under house, garage,and barn roofs where they overhang the building.

18

Many birds like nests in leafy trees that havebig branches.

Some, however, carefully make nests in thewild rice, high weeds , or tall marsh grassthat may contain many bird and other wildlife homes.

A few birds make nests that are soft and com-fortable inside old wooden fence posts.

PARAGRAPH C, STYLE A

FIRST GRADE

Lions use claws to hold their food.Bears have claws for digging.Cats' claws help them climb trees quickly.Tigers use strong claws for killing.

SECOND GRADE

Lions use claws to hold their food.Bears have long claws for digging.Cats' claws help them to climb tall trees in a

hurry.Tigers use strong claws for fighting in the woods.

THIRD GRADE

Lions use claws when they hold and eat theirfreshly caught food.

Bears have long claws for digging up many dif-ferent roots and bugs.

Cats' claws help them to climb the most diffi-cult trees in a hurry.

Tigers use their strong claws for fighting theirmany enemies in the woods.

FOURTH GRADE

Lions use claws when they hold and eat theirfreshly caught food.

Bears have long claws for digging up differentkinds of roots and insects.

Cats' claws help them to climb the most diffi-cult trees in a hurry.

Tigers use their strong claws when they attackand fight their jungle enemies.

FIFTH GRADE

Lions use claws when they hold and eat theirfreshly caught food.

Bears have long claws that help them to dig upvarious roots and insects.

Cats' claws help them to climb even the mostdifficult trees in a hurry.

Tigers use their strong, powerful claws whenthey attack, fight, and frequently kill theirjungle enemies.

SIXTH GRADE Many animals use claws for fighting their manyenemies in the woods.

Lions use claws when they grasp and eat theirfreshly caught food.

Bears have long claws that help them to digup and tear apart various roots and insects.

Cats' claws help them if they are forced toclimb even the most challenging trees veryquickly.

Tigers use their strong, powerful claws whenthey attack, fight, and frequently kill theirjungle enemies.

PARAGRAPH C, STYLE B

FIRST GRADE

Some animals use claws to hold food.A few hav,e claws for digging.Some animals' claws help them climb trees.Many animals use claws for killing.

SECOND GRADE

Some animals use claws to hold food.A few have claws for digging.Some animals' claws help them climb tall trees

in a hurry.Many animals use claws for fighting in the

woods .

THIRD GRADE

Some animals use claws when they hold andeat their freshly caught food.

A few have long claws for digging up differentroots and bugs.

Some animals' claws help them climb the mostdiff:' :ult trees in a hurry.

FOURTH GRADE

Some animals use claws when they hold and eatfreshly caught food.

A few have long claws for digging up differentkinds of roots and insects.

Some animals' claws help them climb the mostdifficult trees in a hurry.

Many animals use claws when they attack andfight their jungle enemies.

FIFTH GRADE

Some animals use claws when they hold and eatfreshly caught food.

A few have long claws that help them dig upvarious roots and insects.

Some animals' claws help them climb even themost difficult trees in a hurry.

Many animals use their powerful claws whenthey attack , fight, and frequently kill theirjungle enemies.

SIXTH GRADE

Some animals use claws when they grasp andeat freshly caught food.

A few have long claws that help them dig upand tear apart various roots and insects.

Some animals' claws help them if they areforced to climb even the most challengingtrees quickly.

Many animals use their powerful claws whenthey attack, fight, and frequently kill theirjungle enemies.

Warmup:

Hello, there. My name isWhat's your name? Good.did you notice that I used a sentence to tellyou my name? /Read and fill in name./ Yousaid your name was . Can youmake up a sentence just like mine to tell meyour name? Good. I have your sentencewritten here too. /Read and fill in name./Suppose I say, "My teacher's name is Mr.Brown." /Show sentence as you say this./Can you make up a sentence like mine to tellme your teacher's name?Very good. I'll write that down in here./Write name in./ Now we have four sentencesto read. Will you read each one back to me?

APPENDIX B

DIRECTIONS, STUDY I

Task:

Very good. Now I have some more sentencesfor you to read, but we're going to do some-thing a little different. This time, as you read,think about what all the sentences togethersay. When you finish reading, make up justone sentence in your own words that says whatall the sentences tell you. You can read thissilently to yourself. If you do not knowany of the words, ask me and I will tell you./Give task materials to S. If S reads aloud,allow him to do so./

20

Posttask:

What do all of these sentences together tellyou? You may look back at the page if you wish./Note response.//Pause 10 seconds. If S does not answer, re-peat question./

Additional question:

/If S gives a subject referent that is not clear(They, Some, etc.) point to the word in thewritten response (e.g. "They") and ask:/ Whatdo you mean by ? /Note re-sponse./

Second and third tasks:

Now let's do the same thing on another page.Would you read this and think about what allthe sentences say? /Give task materials to S./What do all of these sentences together tellyou? /Note response.//Pause 10 seconds; if S does not answer, re-peat question.//If S gives a subject referent that is not clear,point to the word in the written response andask:/What do you mean by ? /Noteres ponse . /

. . ... -

APPENDIX C

ANSWER FORM, STUDY I

Name Examiner

School Task

Grade

Sex

If S gives a subject referent that is not clear (they, some, etc.), point to the word in the writtensentence (e.g. "they") and ask: "What do you mean by 'they' ?"

Task

"What do you mean by 'they'?"

Task

21

22

APPENDIX D

PARAGRAPH, STUDY II

Some birds build nests under the roof.

Many birds like nests in trees.

Some even make nests in tall grass.

A few birds make nests inside wood fence posts.

Other birds build nests in bushes.

Many birds build nests in bird houses.

Warmup:

Hello, there. My name isWhat's your name? Good.di'd you notice that I used a sentence to tellyou my name? READ AND FILL IN NAME.You said your name was . Canyou make up a sentence just like mine to tellme your name? Good. I have your sentencewritten here too. READ AND FILL IN NAME.Suppose I say, "My teacher's name is Mr.Brown." SHOW SENTENCE AS YOU SAY THIS.Can you make up a sentence like mine to tellme your teacher's name? Very good. I'llwrite that down in here. WRITE NAME IN.Now we have four sentences to read. Willyou read each one back to me?

APPENDIX E

DIRECTIONS, STUDY II

Task:

Reading is like solving a puzzle. First we findout what one sentence tells us. As we readmore sentences, we try to figure out what allthe sentences together tell us.Today I want you to try to solve a reading puz-zle. We will use a group of 6 short sentences,but to make this a puzzle I will show you justone sentence at a time. Like this. / DEMON-STRATE ON PRETASK / After each sentence I

want you to tell me what you think all the sen-tences together will say even before you seeall the sentences.As you read more sentences you may or may notchange your ideas. But that isn't important.It is important to try to figure out what all thesentences together will say even before you seeall the sentences. Here is the first sentencefor you to read. You can read it silently toyourself. If you do not know any of the words,ask me and I will tell you.

S READS SENTENCE ---

Now that you've read this, make up a sentencein your own words to tell me what you thinkall the sentences together will tell you.****All right. Here is another sentence to read.

S READS SENTENCE ---

USE FOR ALL BUT LAST SENTENCE:Now that you've read these, make up a sen-tence in your own words to tell me what youthink all the sentences together will tell you.

[This last paragraph would be used after eachnew reading. For the last sentence, changewording to "all the sentences together tellyou."]

Name

School

Grade and Sex

Standard task:

Examiner

APPENDIX F

ANSWER FORM, STUDY II

What do you think all the sentences togetherwill say? Tell me in a sentence in yourown words. /Note response./ Repeat questionafter 10 seconds if S does not answer./ Con-firm each response with neutral "all right."/

[Omit "will" for sentence 6./ Omit "Tell me ina sentence..." if S gives sentence response./If S gives a subject referent that is not clear(they, some, etc.) point to the word in the writ-ten answer (e.g. , "they") and ask: "What doyou mean by 'they' ?"]

24

Responses:1

1

1-2

1-3

1-4

1-5

1-6

REFERENCES

Alderman, G. H. Improving comprehension insilent reading. Journal of Educational Re-search, January, 1926, 13, 11-21.

Broening, A. M. Abilities which contribute toeffective reading. Education, 1941, 62,11-17.

Dale, E., & Chall, J. A formula for predictingreadability. Educational Research Bulletin,January, 1948, 27, 11-20, 28.

Keneally, K. G. A study of the relative orderof difficulty of several types of study skills.Master's thesis , Boston University, 1939.

Macdonald, J. B., Harris, T. L., & Rarick,G. L. An experimental study of the groupversus the one-to-one instructional rela-tionship in first grade basal reading program.University of Wisconsin, School of Educa-tion, 1966.

GPO 809-283-2

McCullough, C. Response of elementary schoolchildren to common types of reading compre-hension questions. Journal of EducationalResearch, September, 1957, 51, 65-70.

Spache, G. A new readability formula for pri-mary grade reading materials. ElementarySchool Journal, March, 1953, 53, 410-13.

Stone, C. R. Measuring difficulty of primaryreading material: A constructive criticism ofSpache's measure. Elementary School Journal,October, 1957, 57, 36-41.

Thorndike, E. L. Reading as reasoning: A studyof mistakes in paragraph reading. Journal ofEducational Research, June, 1917, 8, 323-32.

Woody, C. Measurement of a new phase ofreading. Journal of Educational Research,November, 1923, 8, 315-26.