ED 024 540By- Koenke, KarlThe Effects of a Content-Relevant Picture on the Comprehension of the Main Idea of a Paragraph. ReportFrom the Reading Project.

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

Repor t No- WRDCCL- TR-56Bureau No- BR-5-0216Pub Date Jun 68Contract- OEC- 5-10-154Note- 41p.EDRS Price MF-$0.25 NC-T.2.15Descriptors-Grade 3, Grade 6, Pictorial Stimuli, *Readability, *Reading Comprehension, *Reading Instruction,*Reading Research, *Visual Learning, Visual Perception

The importance of content-relevant pictures in the comprehension of the mainidea of a paragraph was investigated. Various reading conditions were constructedwhich contained a paragraph, a picture, or both, each with three types ofinstructions. In addition, the effects of paragraph readability, student gradeplacement (grade 3 and 6), and sex were examined. An analysis of variance of themain idea responses to the pictures showed that in general there were no significantdifferences between boys and girls or between third and sixth graders. Responses toone picture were better than rèspohses to the other tiwo, but only for sixth-gradesubjects. The analysis of variance of ratings of the responses to the paragraphswith/without pictures and directions did not affect the adequacy of main idearesponses. Although both readability and grade effects were significant, post hocanalyses showed that reading basic paragraphs led to significantly better main idearesponses only among the sixth graders and that the significantly higher boys' meanaccounted for the difference. (Author/WB)

DOCUMENT RESUME

24 RE 001 486

A

A

A

=C.

Technical Report No. 56

THE EFFECTS OF A CONTENT-RELEVANT PICTURE ON THE COMPREHENSION

OF THE MAIN IDEA OF A PARAGRAPH

By Karl Koenke

Report from the Reading ProjectWayne Otto and Karl 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 & WELFAREOFFICE OF EDUCATION

THIS DOCUMENT HAS BEEN REPRODUCED EXACTLY AS RECEIVED FROM THE(PERSON OR ORGANIZATION ORIGINATING IT. POINTS OF VIEW OR OPINION&STATED DO NOT NECESSARILY REPRESENT OFFICIAL OFFICE OF EDUCATIONPOSITION OR POLICY.

The research reported herein was performed pursuant .to a contract with the United States Office ofEducation, Department of Health, Education, and Welfare, under the provisions of the Cooperative

Research Program.

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

PREFACE

This Technical Report is a product of the reading project within Program

2, Processes and Programs of Instruction, of the Wisconsin R & D Center forCognitive Learning. The overall purpose of the program is to improve educa-tional practice through the application of knowledge to instructional problems

within disciplines, such as reading.The study reported in this document was designed to gain information

about the importance of content-relevant pictures to the comprehension of themain idea in a paragraph to be read. Various reading conditions were con-structed which contained a paragraph, a picture, or both, each with threetypes of instructions. In addition, the effects of readability af the paragraphs ,grade placement of the students (third and sixth grades), and sex were exam-ined.

This report illustrates process related research and contributes to the

understanding of cognitive learning within instructional systems.

T . A. RombergDirector of Program 2

CONTENTS

List of Tables and Figures

Abstract

Page

vii

ix

Problem 1

Method 9

Results and Discussion 17

Summary, Limitations and Conclusions , and Implications 23

Appendix A: Pretask 27

Appendix 13: Directions 28

Appendix C: Materials 29

Appendix D: Response Sheet 32

References 33

LIST OF TABLES AND FIGURES

Table Page

1 Readability Data for All Paragraphs According to Main Idea 12

2 The Number of Phrases and Clauses in Each Sentence ofthe Paragraphs, Grades 1-6 13

3 Sentence Location of Phrases and Clauses in Paragraphs.1, 3, and 6 of Main Ideas A, B, and C 13

4 Frequency of Main Idea Ratings for Each Picture 17

5 Analysis of Variance of the Picture Response Ratings 17

6 Mean Main Idea Response Ratings for Subjects ViewingPictures 18

7 Ordered Means and All Pair-Wise Gaps: Pictures Main Effect 18

8 Mean Main Idea Ratings: Pictures x Grade 18

9 Ordered Means and All Pair-Wise Gaps: Picture x Grade 18

10 Mean Main Idea Scores of 192 Subjects Reading Paragraphs 19

11 Analysis of Variance of the Mean Main Idea Response Ratings 19

12 Mean Main Idea Ratings: Grade x Sex 20

13 .Ordered Means and All Pair-Wise Gaps: Grade x SexInteraction 20

14 Ordered Means and All Pair-Wise Gaps: Readability Levelsx Grades Interaction 20

Figure

1 The Cognitive Functions Used in Attaining the Concepts ina Main Idea 11

2 The Second Main Idea Scale 14

3 The Third Main Idea Scale 14

vii

.1+ .

f9:

FigurePage

4 The Fourth Main Idea Scale 15

5 Mean Main Idea Ratings Obtained from Three Pictures 18

6 Mean Main Idea Scores of Boys and Girls at Grades 3and 6 20

7 Mean Main Idea Scores for Grades 3 and 6 with Paragraphsat Two Readability Levels 21

8 The Fifth Main Idea Scale 26

viii

,

ABSTRACT

The study was designed to investigate the effects of content-relevantpictures upon main idea responses of third- and sixth-grade students . Therewere two general questions: (1) To what extent can pictures carry a mainidea message? (2) To what extent can content-relevant pictures reinforce oraid in the comprehension of the main idea of a paragraph?

Subjects in the five conditions were asked to either (1) view three pic-tures, (2) read three paragraphs, (3) read three paragraphs with pictures,(4) read three paragraphs with pictures with directions to attend to the pic-ture, or (5) read three paragraphs with pictures with an explanation of therelationship of the picture to the paragraph. One half of the subjects in thetreatments using written material saw the basic paragraph, while the otherhalf saw materials judged to be their attained grade in readability.

The tabulation of responses to the pictures showed that the responsesgenerally received low ratings on a seven7point main idea scale. An analysisof variance of the main idea responses to the pictures showed that in generalthere were no significant differences between the responses of boys and girlsor third and sixth graders. Responses to one picture were better than responsesto the other two, but only for sixth-grade subjects. The analysis of varianceof ratings of the responses to the paragraphs with/without pictures and directionsshowed that the addition of the pictures and directions did not affect the adequacyof the main idea responses. Although both Readability and Grade effects weresignificant, post hoc analyses showed that reading basic paragraphs led to sig-nificantly better main idea responses only among the sixth graders and that thesignificantly higher boys' mean accounted for the difference.

THE PROBLEM

Past investigations of the effect of picturesupon reading comprehension have typically beenjustified by the fact that educators did not knowthe effects of the large number of pictures inschool books upon children's reading compre-hension. This justification for the investiga-tion of the picture-reading comprehension rela-tionship is still valid, for reviewers (Spaulding,1955; Vernon, 1954; and Weintraub, 1966a) ofstudies of this relationship have agreed that thedubious quality of the studies and their con-flicting results make it impossible to base con-clusions about the relationship between pic-tures and reading comprehension upon research.The same writers have made two suggestionsfor further research concerning the effects ofpictures upon reading comprehension. (1) Therelationship between the pictures and the writtenmaterial should be described. (2) The relation-ship of pictures to specific types of comprehen-sion should be considered.

Therefore, in the present study of the effectsof pictures upon reading comprehension, thepictures are related to the written materiais inthat an artist drew the pictures to illustrate thesame main idea message carried by the para-graphs , and the pictures are related to a spe-cific type of comprehension in that only onetype of main idea message was considered inthe writing of the paragraphs, the drawing ofthe pictures, and the comprehension testing.Within these limitations and because empiricaldata were not available, answers to two ques-tions were sought. (1) To what extent can pic-tures carry a main idea message? (2) To whatextent can content-relevant pictures reinforceor aid in the comprehension of the main idea ofa paragraph?

REVIEW OF THE RELATED RESEARCH

The work of earlier researchers which servesas the rationale for the present study of theeffects of content-relevant pictures on the com-

prehension of the main idea of a paragraph isthat which was focused on the picture-readingcomprehension relationship. The intensivereview of this research comprises two parts:(1) studies of the effects of pictures on thereading comprehension of elementary schoolpupils , and (2) studies of the effects oftures on the reading comprehension of secondaryschool pupils.

Studies In The Elementary School

The most frequently cited investigations cfthe effects of pictures upon reading compre-hension was done by Miller (1938). The pur-pose of his study was to determine whetherchildren in Grades 1-3 who read a basal setof primary readers with the accompanying illus-trations scored higher on a comprehension testthan children who read the same material with-out the pictures. Six hundred subjects fromfifteen classrooms in three Springfield, Illinois,elementary schools were involved. Half of thesubjects within each grade level read theirbasal reader with the pictures covered whilethe other half had textbooks as they came fromthe publisher. Each classroom contained bothexperimental groups matched on reading testscores. Teaching methods were not described,but assurance was given that the groups receivedidentical treatment. The comprehension mea-sure was the difference between pre- and post-test scores over individual stories and the dif-ference between posttest scores across gradesand treatments. Gains in gross reading abilitywere measured by alternate forms of a standard-ized achievement test given at the beginningand the close of the semester.

The reading comprehension tests for the indi-vidual stories seemed unique, and one wonderswhy Miller chose to operationalize reading com-prehension in the following manner:

Children were required to choose, from agroup of words , a word spoken by the teacher;

1

to select a phrase from two phrases whenone was spoken by the teacher; to crossout an extraneous word from a group of threewords; to complete sentences after readinga paragraph; and to put in proper sequencethe happening recorded in a paragraph to beread [Miller, 1938, p. 678].The following factors may have confounded

Miller's results: (1) the relationship of pictureto questions was not considered; (2) the reSUltsof the comparison across grades on individualstories was confounded since the story content,he pictures, the readability levels, and thequestions on the tests varied while the secondmeasure, the standardized test, might not havebeen sensitive enough for the withrn-semestercomparison of reading achievemetit which Millerwanted; (3) the experimental groups sharedteacher, space, and time; and one wondershow many subjects actually did not see mostor all of the pictures or hear another subject'sresponse to a teacher's question concerning arelevant portion of the content of a picture.Since the groups were matched on reading ability,it is possible that both groups were present ineach reading circle or group and were taughtat the same time. Considering these problems,Miller's results are not surprising; no signifi-cant differences were found between the pictureand the nonpicture groups on the standardizedreading achievement tests or on the comprehen-sion tests covering the individual stories.

The only other investigator to concern him-self iiith pictures and written materials whichappear in a basal series was Weintraub (1960),who limited his investigation to the Sheldonbasal series and second-grade subjects. Hemeasured comprehension with a ten questionmultiple-choice test on each of three differentstories in the book. Weintraub stated thatthese tests measured comprehension of the mainpoints of the story. No further definition of thecomprehension measure was given, but inspec-tion of the questions showed that they dealtwith the movements, descriptions, and reasonsfor action of the characters in the stories.

Weintraub reported that the subjects readingtext alone scored significantly higher on hismultiple-choice tests, but a check by the pres-ent investigator revealed a subtraction errorbetween means compared in the Duncan RangeTest. Actually, the means of the groups readingtext and text with pictures were not significantlydifferent (p. 44). Furthermore, althoughWeintraub concluded that poor readers readbest without pictures while good readers werenot affected by presentation differences (ex-cluding pictures alone), a check of the actualstatistics (p. 53) showed that the Duncan RangeTest did not show a significant difference be-

2

tween treatmentstext alone and picture withtext for poor readers. Weintraub's conclusionwas correct on the basis of a gross comparisonof means; but the difference, according to hisown choice of a post hoc test, was not signifi-cant.

It should be noted that each story was intro-duced, but the form was not described in detailand Weintraub states that the introductionsvaried (p. 32). He also listed four limitationsof his study: (1) the texts were from one pub-lisher; (2) the pictures were taken as theyappeared, and, therefore, no controls wereplaced upon the picture function; (3) the popu-lation was upper-middle class only; and (4)there was no demonstrated or theoretical rela-tionship between the questions and specificaspects of reading comprehension.

Two studies of the picture-reading compre-hension relationship used pictorial and writtenmaterial which approximated basal readers,but there the similarity ended (Strang, 1941;Halbert, 1943). Halbert's study was uniquein that the sample of 234 subjects was selectedmostly from one-room schools in Jackson County,Kentucky, which was 100% rural accordingto the U. S. census of 1941. In fact, Halbertmentioned that the largest urban center had apopulation of 175. The subjects ranged in agefrom 8 1/2 to 12, but their grade placementcould not be considered as typical becausereading achievement scores reported showedthe sample to be generally below the nationalnorm.

The pictures and reading materials in theHalbert study were created for a project in whichan attempt was made to disseminate dietaryinformation through supplementary basal readers.Halbert described the written materials as hav-ing a reading level, a style, and a vocabularytypical of basal readers of the period. No spe-cific material or picture controls were presentei,except that the children in the pictures were tobe dressed as the readers were accustomed todress. The pictures were rather general, e.g,a boy looking at the rain through a window, aboy placing a worm on a fish hook for a girl.Halbert states that the subjects were not toldto view the pictures or to consider their con-tent. The subjects were tested individuallyand an informal warmup task consisted of askingthe subject his name and other personal infor-mation. Comprehension was measured by thenumber of ideas stated by the subject afterreading the material orally. The choice ofmeasure was based upon the inability of thesubjects to communicate well in an essay situa-tion and the investigator's acceptance of astudy which showed multiple-choice questionswere troublesome. The three conditions which

were tested were the usual picture and the text,the text alone, and the picture alone. The re-sponses were classified as relevant or irrele-vant. Operationally, the former were sentencescontaining a summary or a description of adetail, an event, or a principle. The latterwere statements not related to the written ma-terial or the illustrator's stated topic for thepicture.

The statistics were descriptive and tests ofsignificance were not performed on the data.No provisions were made for consideration ofpossible interactions of readability of text,ability of the child, sex, and treatment.

Halbert noted in discussing her results thatthe three different pictures evoked differentnumbers of relevant responses, and that thehighest number of relevant responses evokedby a picture and text combination included thepicture which evoked the greatest number ofrelevant ideas. No comment was made con-cerning the text. The picture and text formatswere associated with the greatest number ofrelevant responses. Pictures alone evoked agreater number of responses than text alone,but most were irrelevant. Good and averagereaders gave fewer total responses than poorreaders, but the good and average readers gavemore relevant responses than the poor readers.

Halbert concluded, "To the extent that mem-ory for ideas is a measure of comprehension,to that extent pictures contribute to the compre-hension of reading materials [p. 57]." It mightbe added that this observation was probablyaccurate within the limitations of sample,materials , and use of statistical procedures.

Whereas Halbert (1943) used a rather uniquepopulation, subject matter, and comprehensionmeasure, Strang's study (1941) of gains andlosses in concepts as indicated by pupil readingscores after the addition of illustrations toreading material was not marked by these uniquefeatures. Strang described his population asthe entire fourth, fifth, and sixth grades of asingle school in Philadelphia. Furthermore,the achievement of the subjects and the socialstatus of the families were above the norm.The materials were basal reader-like in appear-ance, and, although no controls were defined,the materials had been submitted to severaleditors and authors of children's literature forcriticism and general validation. Style, length,and content varied from passage to passage.Each passage was followed by four multiple-choice questions. The first questions dealtwith the main idea as a title to the story, whilethe other three asked for identification of spe-.cific facts. The pictures were 2" x 3" blackand white photographs or drawings mounted inthe upper right hand corner of the page. The

relevance of picture to passage or question wasnot stated by Strang. Furthermore, inspectionof the material by the present investigator didnot reveal a pattern or relationship.

The subjects read the unillustrated storiesfirst, but the grade levels were split with halfreading stories A, B, and C without picturesand the other half reading stories D, E, and Fwithout pictures. Several weeks later the groupsread the three stories they had not read previous-ly, this time with pictures. The directions con-sisted of sample exercises.

The gains and losses were measured relativeto the individual's score on the unillustratedmaterial. Although means for various groupswere given and compared, tests of statisticalprobability were not used. The individual'sscore was a summation of the scores of thethree stories under each condition. No hypothe-ses were stated, and the resUlts were difficultto interpret, but Strang did conclude that therewere more gains than losses from the unillus-trated to the illustrated reading conditions atall grade levels. The multiple-choice testswere presumed to be measures of concepts,but there was not a stated definition of the

.term "concept." Operationally, Strang mea-sured main idea and specific fact in a multiple-choice test without reliability estimates. Thesubjects represented a special though not in-frequently described type of population, but thepicture-text-test relationship was not explained.There also appeared to be no way to separatethe results from practice effect, since the illus-trated condition was always the second in orderof presentation.

Brief reports of three studies by Goodykoontz(1936) revealed that she did not seem concernedwith establishing a relationship between thematerials her subjects read and classroom ma-terials, as was the case with the previous in-.vestigators. In all three studies the subjectsread a twelve page booklet about cork, but nofurther information about the materials wasgiven. In the first study she tested 294 sub-jects in Grades 6-8 with thirteen questionsabout the written text and thirteen questionsabout the pictures. No directions were givento the subjects to look at the pictures accom-panying the written text. Since Goodykoontzdid not have separate groups reading the ma-terial with and without pictures, she asked thesubjects to tell her whether they had looked atthe pictures during the reading. She found that6 percent of the subjects had not looked at thepictures, 25 percent had looked at the picturesfor fun after reading all the text, and 50 percenthad looked at the pictures as they came to them.The results of the testing showed no differencesbetween groups on the questions concerning Lhe

3

text, but of the thirteen questions concerningthe pictures the subjects looking at the pictureshad a median score which was two successfully,answered questions better than those who didnot view the pictures.

In the second study in the series Goodykoontzdirected "nearly ninety" sixth-grade students toanswer thirteen questions concerning the pic-tures in the booklet about cork while looking atthe pictures. The questions were described as"main point," but no example was given. Sincethe group had a median of eight correct ,Goodykooniz concluded that the pictures didadd extra information which the students candiscover.

In the last study by Goodykoontz' (1936) thesame "nearly ninety" subjects read a section ofthe booklet about cork which they had not seen,then looked at two pictures which they had notseen and finally listed all the information whichthe pictures added to what they learned from thetext. Goodykoontz found that the subjects'median number of bits of information seen whichrelated to the previously read text was two perpicture. The writing of this result left thepresent investigator with the feeling thatGoodykoontz was disappointed with the fewbits of information seen in each picture.

Since Goodykoontz was one of the first toinvestigate the picture-reading comprehensionrelationship and the reports were brief, thefollowina comments concerning methodologyshould be viewed in that light. Goodykoontzdid not describe her sample, question format,and specific aspect of comprehension testedor give the exact sample size. Furthermore,the characteristics of the material and the pic-tures were not given, and she did not use sta-tistical probability in the first of the studieswhere it would have facilitated interpretationof the results.

Lewernz (1929), like Goodykoontz (1936),used pictures which carried more informationthan the written text and tested for this extrainformation. Lewernz compared the readingachievement scores of a group that read andsaw pictures with those of a group that onlyread the material. The subjects were 359 eighth-grade students from five Los Angeles junior highschools. The materials were eight still photo-graphs from a movie and accompanying historicalbackground about Columbus. One group wasdirected to study the pictures because the testwas based upon them, but the directions to theother group were not mentioned. Example ques-tions from the true-false test concerned thecolor of a wall, the construction of a chair,and the type of jewelry worn by men. The per-centage difference favored the group who sawthe pictures and the material. This was expected

4

because, as Lewernz stated, the questionswere based upon the content of the pictures.

The group that saw the pictures obtainedscores averaging 15% higher than thosenot viewing pictures, but the number ofquestions , the use of probability, the reliabilityand the content of the test, and the difficultyof the material were not mentioned. It wouldseem that what Lewernz actually found was thatif eighth-grade students were asked questionsabout details of a picture that they had notseen, they could not answer the questions withas great a degree of success as those studentswho saw the picture.

Studies In Secondary Schools

While investigations of Miller (1938),Weintraul. (1960), Halbert (1943), Strang (1941),Goodykoontz (1936), and Lewernz (1929) werelimited to elementary school populations,studies by Burdick (1959), Kambly (1943),Vandermeer (1950), and Davis , Hicks, andBowers (1966) extended the investigation ofthe effects of pictures on reading comprehen-sion to the high school level.

Burdick (1959) investigated the relativeabilities of 313 subjects in Grades 7-12 toread cross-section and perspective cutawaydrawings that might appear in science textbooks.In contrast to Goodykoontz and Lewernz, Burdickused the pictures to reinforce the information inthe written text rather than to carry more infor-mation. He found no significant differencesbetween mean comprehension scores of thegroups that read with and without illustrations,but there was a significant grade effect in theanalysis of variance. Burdick did not presenta table of means or use a post hoc test, so thebasis for his conclusion that ability to read ma-terial with and without cross-sections and per-spective cutaway drawings is learned and in-creases through grade levels cannot be checked.Since the same written material was used at allsix grade levels, it might not approximate atypical reading task. While the pictures usedby Burdick were described as representing thosefound in a list of selected textbooks, the char-acteristics of the written portion were not given.

Vandermeer (1950) also had both elementaryand secondary school students involved in hisstudy, but in contrast to Burdick (1959) onlytwo grade levels eighth and eleventh wereused. Vandermeer found no significant differ-ences within each grade level between a groupthat saw a commercially produced historicalfilmstrip and heard a monologue and a groupthat read the monologue silently while theteacher read it orally. While the subjects wererandomly assigned to treatments and one teacher

taught both methods, the possibility of an inter-action between grade and instructional methodwas not considered and the relationship of themultiple-choice test to the monologue and thepictured information was not described. Sinceneither group actually read textbook materialor read silently without the benefit of a teacher'soral presentation of the same material and therelationship between the effects of pictures ina filmstrip and pictures in a text upon compre-hension is unknown, it is not possible to gen-eralize to the effects of pictures in texts uponreading comprehension of historical material.

In what may be the first published study ofits kind, Kambly (1943) compared the mean gainscores from pre- and posttests on a generalscience unit of ninth-grade students who readtheir science material either with or withoutthe textbook pictures. Each group comprisedthree classrooms with both groups receiving thesame demonstrations. The test of group meansshowed a difference favoring the subjects read-ing the text accompanied by pictures. In thesame article Kambly reported results favoringthe text with picture group when the materialsand test concerned biology. In both cases therelationship of the textbook pictures to the textwas not given and the population, test content,and text format were not described.

In the most recent American study of theeffects of illustrations on reading comprehen-sion, Davis, Hicks, and Bowers (1966) investi-gated the comprehension of chronological rela-tionships as a function of the addition of timelines to a historical passage. Although timelines are graphs , not pictures , this study isincluded because a single element of compre-hension was studied. The sample consistedof 81 eleventh- and twelfth-grade students fromseveral world history classes in the high schoolattached to a North Carolina teacher-traininginstitution. Twenty-seven subjects were ran-domly assigned to read a.specially constructed700 word passage about the acquisition of thePanama Canal with a correct time line, withouta time line, and with an incorrectly drawn timeline. Davis, Hicks, and Bowers characterizedthe time line as a reinforcer because it presentedthe same information as the written text and wastherefore a repetition of information for the goodreader and a source of information only for thepoor reader. The subjects were told to readand study the narrative for ten minutes, but nomention was made of the time line in the dir-ections. Comprehension of chronological rela-tionships was measured by a twelve item mul-tiple-choice and ordering test with a split-halfreliability of .74. IQ measures were availableand were incorporated in the design as an after-thought. The condition "correct-time-line"

was significantly better than condition "no-time-line," but "incorrect-time-line" did notdiffer from either. As expected, the high IQ(1 20 plus) group was the most successful andno interaction was observed.

An assumption that students enrolled in aspecial high school are usually above averagein scholastic aptitude seems to be borne outby the-IQ scores Davis, Hicks, and Bowersused to delimit their three IQ classificationsunder 109, 110-119, and 120 plus. It may beconcluded then that the results are limited toan above average group, such as one may findin selected suburban areas.

In contrast to all the preceding investiga-tions, Vernon (1953, 1954) reported a series ofstudies of the effects of pictures on the readingcomprehension of English school children.Although Vernon placed many controls upon herinvestigations that were not found in the workpreviously reviewed in this section, a basicproblem is that the population was English andin all cases part of the sample was older thaneleven and a half, at which time the examina-tion which segregates the college-bound fromthe terminal students is taken. Therefore, thesimilarity of Vernon's samples from severalspecial types of schools to any U. S. popula-tion is unclear.

The first two studies reported by Vernon(1953) were exploratory in nature. The questionasked in the first study was: "Do pictures helpor hinder.the acquisition of knowledge from thetext?" The subjects were 14 boys and 21 girls,ages 16 to 18, in attendance at an Englishgrammar school. The materials were two arti-cles of 700-800 words. Article A dealt withthe causes and cure of TB, while article B dealtwith causes of illness in young children. Therewere two versions of each article: A1 and B1were written in "popular" style and accompaniedby photographs characterized as "striking" byVernon; A? and B2 were written "more objectively"and were accompanied by an unspecified numberof graphs. Ten minutes were allowed for readingand study of the material. Each subject wasfirst asked to recall orally what the article wasabout and then was asked to state what mightbe done to reduce the danger of TB or childhoodillnesses. The scores for the oral reports werethe number of "major points" recalled, the num-ber of details recalled, and the coherence andconsistency of the report which was graded Athrough D.

In the second of the two exploratory studies,Vernon investigated the effects of pictorial illus-tration on the amount remembered. The subjectswere 24 boys and girls , ages 15 and 1 6, froma secondary modern school. The materials wereversions A1 and B1 with and without pictures.

5

The task and the scoring were the same as inthe first study.

The results of both studies were of the samenature. No significant differences were foundin the percentage of major points or detailsrecalled between material treatments. Also,the type of presentation did not affect the con-sistency and coherence of the responses.However, five major points in the text directlyillustrated by pictures were remembered betterthan those same points unillustrated by pic-tures:

On the basis of the two preceding studiesVernon concluded, " ...although the picturesmay weight some part at the expense of others,their effect upon a coherent recall of the wholewas neither favorable or unfavorable [p. 183]."

In a second series of experiments , Vernon(1954) emphasized the instructional aspects ofthe picture-text relationship. In these studiesshe used children with an age range of 10 to 13years. The first set of materials consisted oftwo selections of 755 and 940 words dealingwith collecting and printing of news and thepassage of a ship into a harbor and its un-loading. Each selection was read under twoconditions: (1) with eight full-page picturesfacing the text to which they related, and (2)with eight small black-and-white line drawingsof some items mentioned in the text. Afterreading and studying the selection for ten min-utes, the subject was asked six general ques-tions "emphasizing the sequence of events."Vernon stated that where the text was difficultand the pictures were of help, they were notused. The second set of materials consistedof three passages pertaining to bridges, wheels,and furniture. The language was "simplified"and the length in all cases was 300 words.Each selection was presented orally and ac-companied by (a) ten pictures viewed in se-quence, (b) three or four simple outline draw-ings on a single card, or (c) four photographson a single card. The experimenter who readthe material either presented or pointed to theappropriate picture when related context wasread. The task here was to recall as much aspossible of a historical sequence of eventsor cause and effect. The boys did significantlybetter than the girls, and since the IQs werenot significantly different, Vernon attributedthe girls' lack of success to their lack of in-terest. There were no significant differencesamong the treatments of the material as mea-sured by the comprehension test.

Vernon (1954) concluded on the basis of thepreceding studies that "the pictorial illustra-tions in general had little effect upon the under-standing and recall of verbal material [p. 177]."

6

SCOPE OF THE PROBLEM

The dubious quality and the conflicting re-sults of previous investigations of the effectsof pictures on reading comprehension have beennoted here and in two earlier reviews (Vernon,1954; Weintraub, 1966a). One of the reviewers(Weintraub, 1966a) also concluded on the basisof his observations that final statements con-cerning the picture-reading comprehension re-lationship could not be drawn from the existingresearch. This conclusion seems to be sub-stantiated by the present investigator's sum-mary of observations about the methods and theresults of existing research: (1) The statisticalprocedures and methodology used in earlier in-vestigations were less than adequate exceptfor the studies of Vernon (1954) and Davis,Hicks, and Bowers (1966). (2) Atypical popu-lations were sampled by Strang (1941), Halbert(1943), Vernon (1953, 1954), Weintraub (19 60),and Davis , Hicks, and Bowers (19 66). (3) Themeasure of comprehension was unique in thestudies of Miller (1938) and Halbert (1943).(4) There was an inadequate description ofpopulation, comprehension measure, or materialsin most of the studies. (5) The results and con-clusions differ from study to study, even wheresome of the procedures were approximately thesame. For example, Lewernz (1929) andGoodykoontz (1936) apparently used pictureswhich did not contain the same material as thewritten text, but Lewernz' results favored thepicture with text and Goodykoontz' resultsshowed no differences between text-and-picturesand text alone. Furthermore, Miller (1938) ,Strang (1941), Weintraub (1960) used materialswhich were of the basal reader type, but Millerfound no differences between text-with-picturesarid text, Strang found text-with-pictures supe-rior, and Weintraub concluded in favor of textalone.

The reviewers of the studies of the effectsof pictures upon reading comprehension (Vernon,1954; Weintraub, 1966a) have also recommendedmeeting two conditions not present in most ofthe existing work. The recommendations were(1) to describe the relationship between thepicture and the written material with which itis associated and (2) not to work with globalcomprehension measures, rather to deal witha specific aspect of comprehension and itsrelationship to pictures.

The investigator's concern for the recom-mendations of Vernon and Weintraub is shownin the two general questions which served asthe foci of the study: (1) To what extent canpictures carry a main idea message? (2) Towhat extent can content-relevant pictures rein-

force or aid in the comprehension of the mainidea of a paragraph?

The particular type of reading comprehen-sion, reading for the main idea of a paragraph,was chosen as a focus of the present study forseveral reasons. (1) The "main idea" has notbeen operationally defined or considered sepa-rately in any of the existing studies, althoughStrang (1941), Halvert (1943), Vernon (1953),and Weintraub (1960), have either attemptedto measure major points or included questionsconcerning titles and main thoughts in.a totalmeasure. (2) Tightly controlled materials whichpermit an operational definition of the main ideawere available through-the Laboratory for Re-search in Basic Skills, University of Wisconsin.(3) Authorities (Russell, 1961; Harris , 1961;Gray, 1960), a factor analysig by Davis (1944) ,and a survey by Broening (1941) have identifiedreading for the main idea as an important skill,a separate skill, and a commonly used skill.

Within the framework of the two major ques-tions , the effects of direction to look at thepicture, readability of the written materials ,and sex and grade level of the subjects werealso examined. The impact of direction to:look at the picture when it did accompany thetext was exaMined because Goodykoontz (1936)found that her subjects did not all habituallylook at the pictures and Miller (1938) concludedthat direction was necessary if learning frompictures was to take place. Readability formulaecontrols were placed on the written materialssince this seemed the best way to provide forgeneralization and replication. This is in con-trast to all previous work. In fact, onlyWeintraub (1960) and Strang (1941) have pro-vided enough information about the written ma-terial to allow replication. The decision to'have materials rated as first- and attained-gradereadability levels read at both third and sixthgrades was based on the lack of research uponthe effects of level of reading difficulty on useof pictures as comprehension aids. Only twoinvestigators have commented on this possibleinfluence. (1) Vernon (1954) made informalobservations of her subjects and concludedthey did not use the pictures as aids when thematerial was difficult. (2) Weintraub (1960)found that 13 "poor" readers read equally wellwith and without pictures accompanying thetext.

The sample composition was controlled asto the number of boys and girls because theconclusion of two recent reviews of sex differ-ences in reading comprehension (Weintraub,1966b; Davis and Slobodian, 1967) was thatalthough girls are generally thought to be betterreaders, existing results are mixed and may

depend on the specific measure or aspect ofcomprehension. Furthermore, Vernon (1953)and Lewernz (1929) found that boys make betteruse of pictures in some cases. Two gradelevels, three and six, were chosen becauseonly three studieS of the picture-reading com-prehension relatiohship have been found whichdeal with a primary grade population (Halbert,1943; Miller, 1938; Weintraub, 1960) and onlytwo studies have used a middle elementaryschool population (Goodykoontz, 1936; Strang,1941). Grade 3 was chosen because (a) it rep-resents a part of the primary grade populationand (b) the skill of reading for a main idea is .

not generally considered an appropriate com-prehension task for first and second graders.Grade 6 was chosen to represent the middleelementary grades because this is the lastgrade level at which many children receiveformal instruction in reading, and knowledgeof their success upon completing "formal readinginstruction" might yield implications for instruc-tion in the junior high school. Furthermore,the three-year span between third and sixthgrades should give any growth in children'sability to comprehend the materials and thepictures a chance to appear.

HYPOTHESES

The focus of the present study was stated intwo general questions. (1) To what extent canpictures carry a main idea message ? (2) Towhat extent can content-relevant pictures rein-force or aid in the comprehension of the mainidea of a paragraph? These questions were theresult of the investigator's observations con-cerning the existing research of the effects ofpicturas upon reading comprehension and recom-mendations for further research by Vernon (1954)

and Weintraub (1966a).The first general question of this study was

broadly conceived and considered, but the sec-cond general question was stated as the firstworking hypothesis, with the five other hypothe-ses necessitated by the inclusion of materialand organismic variables. The direction ofhypothesis reflects the investigator's interpre-tation of the existing research previously pre-sented.1. The addition of a content-relevant pictureto written text enhances subjects' main idearesponses.2. Subjects who are directed to view a content-relevant picture before reading give better mainidea responses than subjects not so directed.

3. Subjects who are told the purpose of a con-tent-relevant picture and told to examine the

7

picture before reading give better main idearesponses than subjects not told the purposeof the picture and subjects to whom the pic-ture is not mentioned.4. Main idea responses to paragraphs writ-ten as simply as possible, yet consistentwith a main idea, are better than main idearesponses given to paragraphs rated as

8

appropriate for the subject's attained gradelevel, i.e., third or sixth.5. Pictures have a greater effect upon boys'main idea responses than girls' main ideares pons es .

6. Sixth graders give better main idea re-sponses than third graders.

II

METHOD

The subjects involved in this study wereasked to state the main idea of each of threeplctures, or of each of three paragraphs, or ofeach of three paragraphs when accompanied bya content relevant picture. Since the effect ofdirection to view a paragraph and a picture asan integrated presentation was also of concern,three different sets of instructions for subjectsviewing paragraphs with pictures were used.Thus, the various reading conditions were con-tent relevant pictures in isolation, paragraphsalone, paragraphs and pictures with no direc-tion to view the picture, paragraphs and pic-tures with minimum direction to view the pic-ture, and paragraphs and pictures with maximumdirection to view the pictures. The other vari-ables were readability of the paragraphs (basicand equal to the reader's grade placement),grade placement of the subject (third and sixthgrades), and sex of the subf t.

SUBJECTS

The sample comprised 240 subjects 60boys and 60 girls from third grade and likenumbers from sixth grade. The subjects, whowere selected at random from among the thirdand sixth grade students attending nine publicelementary schools in Janesville, Wisconsin,met the following criteria: (a) they had neverbeen enrolled in a special class, and (b) theyhad never repeated a grade.

The city of raneSville, with a population ofapproximately 34,000, is located in the extremesouth-central portion of Wisconsin. Since theschool district is not limited by the city bound-aries, the nine schools which comprised thepopulation of this study do not represent thewhole Janesville district; rather, the schoolssampled enrolled all the children living withinthe Janesville city limits and attending a publicelementary school.

TREATMENT CONDITIONS

The various conditions of material and in-struction which preceded the elicitation of aresponse to be rated as to its degree of similar-ity to a model main idea are described in thissection. The common warmup task and the in-struction peculiar to each of the five treatmentsare specifically discussr..d.

Warmup Task

As previously stated, the warmup task wasthe same for all subjects, irrespective of treat-ment. The objective of the warmup task wasto help the subject become accustomed to re-sponding in sentence form to the investigator.This was deemed essential since the responsedesired from the subject after he saw the ex-perimental materials was a sentence.

The warmup task was merely a formalizingand standardizing of the initial contact betweenthe investigator and the subject. The subjectwas greeted; the investigator stated his firstname in a sentence; the subject was askedhis name; this response was reinforced verbally;and then the investigator informed the subjectthat he (the investigator) had used a sentenceto tell his name. At this time the investigatorwrote the sentence, "My name is Karl," andasked the subject for a sentence of an identicalnature containing the subject's name. The cor-rect verbal response was copied by the investi-gator and reinforced verbally, after which theinvestigator showed the subject a second sen-tence, "My teacher's name is Mr. Brown."The subject was asked to state the name of histeacher in a sentence similar to the one beforehim, and the correct response was written downand reinforced verbally. Specific directionsfor the task were then.introduced.

Warmup procedures, designated "Pretask ,"are given in Appendix A. Although the taskmay seem too easy for sixth-grade Students,

9

this did not prove to be the case in pilot trialsof preliminary drafts of the directions or in thefinal study.

Manipulation Of Materials And Directions

The five combinations of material format anddirection to use it were as follows:1. One group saw only the pictures (Pic). Theywere asked to "think about what the whole pic-ture shows" and to "make up just ,one sentencein your own words that says what the wholepicture tells you." Each subject saw threepictures in a preassigned randomized order.The instruction between pictures was a singlesentence which restated the task. The picturewas not taken away before the subject responded.

2 & 3. The group of subjects who saw only thethree paragraphs (Text) and the group of subjectswho saw the paragraphs with their appropriatecontent relevant pictures (PT nod) received thesame directions. This meant that the PT nodgroup did not receive direction to view thepicture,. After the standard warmup task, thesubjects in both groups were told to read theparagraph and "think about what all the sen-tences together say." The subjects were toldto read silently and ask for help if they did notknow any of the words . The material remainedvisible while the subjects responded. Since thesubjects had three paragraphs to read, a stand-ard task-directing sentence was used betweenparagraphs.4. The fourth group of subjects saw the samepictures and paragraphs as PT nod, but minimumdirections were given to view the picture care-fully (PT min). As in the other conditions , thematerial was visible to the subjects until theyanswered, and the subjects received a sentenceof direction between the paragraphs.

5. The fifth group saw the pictures and theparagraphs under conditions of maximum direc-tion (PT max), which meant that the subjectswere told that a picture was described by aparagraph. Furthermore, they were directed tostudy the picture in order to find out what the"whole picture" showed. Finally, they weredirected to read a paragraph and state the mainidea of the picture and the text. Procedureswith regard to providing unknown words , re-moval of materials, and interjecting the stand-ard sentences between the three tasks were thesame as in the previous sets of directions .

In the directions to the subject (Appendix B)the term "main idea" was not used because itbecame clear during the piloting of the prelimi-nary drafts that the third-grade students did not

10

have a grasp of the meaning of the term. Thismay be the result (a) of a common format forexercises pertaining to the statement of a mainidea where the student is asked to "give a nameor title to the story" without any specific men-tion of the term "main idea," or (b) of formalteaching of the skill of reading to formulate themain idea or to find the topic sentence atGrades 4, 5, and 6.

Pilot studies also showed that statementssuch as "Tell me what the story is about"evoked summaries of the entire selection. Thedirections to the subject, therefore, includedeither one or both of the following statements:"Think about what the whole picture shows"and/or "Think about what all the sentencestogether say." Directions were then given tothe subject to state a sentence in his own wordsconcerning the material he saw.

Testing

The testing of the 240 subjects involved inthe present study took place during the twoweek period from May 8 to May 19, 1967. Oneday was given to the testing of all the subjectsenrolled in each of seven schools. Two othersmaller schools which furnished fewer numbersof subjects were visited on a single day. Allsubjects were tested individually by the in-vestigator in a small room in the school whichthey attended. Since the subjects had beenscreened as to exceptional class placementand failure to advance regularly, the only prob-lem was to arrange for testing of those subjectswho were absent on the day the investigatorvisited their school. It was possible to testall the absentees during the final two days ofthe two week testing period.

DESCRIPTION OF MATERIALS

The pictures, paragraphs, model main ideastatements , and their assumed relationshipsare described in this section.

Pictures

The three 4" x 6" black and white ink draw-ings (Appendix C) were done by a professionalillustrator with experience in illustrating edu-cational materials. The illustrator was givencopies of the main ideas and the paragraphs-whioh had first-, third-, and sixth-graderatings according to the readability formulaeand was asked to draw several pictures illus-trating each of the three main ideas. It wasrecommended that the information from eachof the four sentences developed as specific

examples for a main idea be integrated into onepicture. This was done for all sample picturesof the three main ideas. The illustrator andthe investigator then chose a best picture foreach main idea from the three or four available.

The picture illustrating the main idea "Ani-mals help farmers in different ways" is pictureA. Picture B illustrates "Birds build nests indifferent places." Picture C illustrates "Ani-mals use claws for different things." They arepresented in this order in Appendix C.

Paragraphs

The description of the paragraphs used inthis study entails the description of the struc-ture of the main ideas, their relationship to theparagraphs, and the structure of the paragraphs.The paragraphs are reproduced in Appendix C.

The structure of the main ideas. The three mainidea statements which follow were devel-oped in four-sentence paragraphs for Grades1-6 by the personnel of the Laboratory for Re-search in Basic Skills, University of Wisconsin.

Each of the main ideas is six words in lengthand contains a class noun for the subject, atransitive verb, a direct object and a preposi-tional phrase of three words preposition,adjective, and noun. The length and the struc-ture of the main idea sentence were delimited(1) by the acceptance of T. L. Harris' model ofa main idea,' and (2) by the content.

Harris reasoned that a model main idea sen-tence should include two elements: (1) a state-ment of the general topic cova:ed, and ( 2) a

1 Unpublished paper entitled "Notes on Control-ling the Ideational Structure of Paragraphs,"1965.

General TopicSentence Synthesize Recognize

1

2

3

4

restrictive statement derived from the specificcontent. Thus, each of the three main ideascomprise two main elements.

(A)

(B)

(C)

General TopicAnimals help thefarmerBirds build nestsAnimals use claws

Specific Restriction

in different ways.in different places.for different things.

Relationship of main idea to paragraph. Thesubject matter of the main ideas and thefour-sentence paragraphs were dictated in partby the words listed on the Stone list of 769easy words (Stone, 1957), which is a revisedvocabulary measure used in the Spache reada-bility formula for primary grade material(Spache, 1953). This list provides some guid-ance regarding words known by first-gradechildren, and an attempt was made to writebasal paragraphs at the first-grade level.

The following description of the formulationof a main idea of a paragraph further demon-strates the relationship of the content of theparagraph to the content of the main idea.

The formulation of the main idea of theparagraph can be viewed as a process of syn-thesizing four examples into one main idea orprinciple (Gagne, 1965). Figure 1 is a sche-matic representation of the operation of a readerwithin the framework of a first grade paragraphwritten to evoke Main Idea B (Appendix C). Asshown in Figure 1, the reader synthesizes fourreferents to conceptualize portions of the firstand second elements of the main idea and alsorecognizes the remaining words as those whichappear in most of the sentences of the para-graph. Davis (1966, p. 254) specifically placessynthesizing of class nouns (birds, animals)and relationships (different places, differentways) within the same type of concept learn-

Specific RestrictionRecognize Synthesize

Robins may build nests

Bluejays

Ducks

Woodpeckers

Main Idea Birds

like nests

make nests

make nests / inside wood fence posts.

under

in

in

a roof.

trees.

tall grass.

build nestsmakelike

in different places.

Fig. 1. 'The Cognitive Functions Used in Attaining the Concepts in a Main Idea

11

ing categorizing. Assuming this to be true,the conclusion is that although the generaltopic may be hypothetically more importantthan the specific restriction of the paragraph,cognitively the two elements are formulated insimilar ways

Structure of the paragraphs. As previously men-tioned, the three paragraphs developed foreach of the three main ideas were four sen-tences long. This length was held constanteven though within a set of three paragraphscarrying the same main idea the readabilitylevels varied from the simplest form consistentwith the main idea and sentence structure con-trols placed upon the material to a sixth-gradedifficulty level. Besides the three basic para-graphs, one for each main idea, readabilityratings showed one paragraph for each mainidea to be at third-grade diffibulty and oneparagraph for each main idea to be at sixth-grade difficulty. The Spache readability formula(Spache, 1953; Stone, 1 957) was used to ratethe basic paragraphs and the third-grade para-graphs, but the Dale-Chall readability formula(Dale and Chall, 1948a, 1948b; Klare, 195 2)was used to rate the sixth-grade paragraphs.The use of two formulae was necessary becauseno commonly used readability formula developedfor use on educational materials can be used tojudge both third- and sixth-grade material.

Table 1

The difficulty level of the paragraphs wasmanipulated by increasing the length of thesentences and adding more words not appearingon the list of easy words associated with theformula. Complete readability information isgiven in Table 1, and the paragraphs are re-produced in Appendix C.

The three structural controls placed on theparagraphs, other than the selection of thetopics from the Stone word list (1957), werethose measured directly by readability formulae,i.e. vocabulary and sentence length, and oneextra control, internal structure of sentences.The final control was decided upon by the staffof the Laboratory for Research in Basic Skillsbecause some control of sentence structure wasfelt necessary and because it would facilitatedescription of the material. Since data-werenot available to support a progressive orderingof sentence complexity to coincide with thereadability levels associated with Grades 1-6,an arbitrary manipulation of the number ofphrases and clauses across grade levels wasused. The manipulation shown in Table 2 wasthought to (a) give sentence structure some ofthe developmental aspects of the vocabularyand sentence length variables which readabilityformulae measure and (b) clarify the placementof phrases and clauses in the materials used inthis study (Table 3). The paragraphs used are

Readability Data for all the Paragraphs According to Main Idea

Readability

Factors

Main Idea A Main Idea B Main Idea C

Grades Grades Grades

1 3 6 1 3 6 1 3 6

No. of Sentences 4 4 4 4 4 4 4 4 4

No. of Words 25 50 63 25 50 63 25 50 63

x Sentence Length 6.5 12.5 15.75 6.25 12.5 15.75 6.25 12.5 15.75

No. of "Hard" Words a1 5 5 2 5 5 1 5 5

Percentage "Hard" Words 4 10 7.9 8 10 7.9 4 10 7.9

Readability Scoresb 2.1 3.5 5.69 2.4 3.5 5.69 2.1 3.5 5.69

a "Hard" words for paragraphs for Grades 1 and 3 are those not appearing on Stone's list (1957)which is used as the vocabulary control in the Spache formula. For the sixth-grade paragraphs"hard" words were those not on the Dale list, which serves as the vocabulary control for theDale-Chall formula.

bReadability scores for paragraphs for Grades 1 and 6 are expressed in grades and are computedfrom the Spache formula. The readability scores for the sixth-grade paragraphs are Dale-Challscores which place each paragraph in the upper 1/3 of the score range for Grades 4-6.

1 2

Table 2

The Number of Phrases and Clauses in EachSentence of the Paragraphs, Grades 1-6

Sentence

GRADE

2 3 4

1

2

3

4

pa

PPpp

PPpppc

pppppcPc

5 6

pp pcb

pc pcpc pcpc pc

a "p" denotes a prepositionEd, infinitive,gerund, or participial phrase."c" denotes an adjective or adverb clause.

designated as for Grades 1, 3, and 6 in Tables2 and 3. The following assumptions about therelative difficulty of prepositional phrases andsubordinate clauses were the basis for the or-dering. (a) A prepositional phrase does notadd as much to the complexity of a sentenceas a subordinate clause does. (b) A sentencewith only one prepositional phrase is easier toread than a sentence with one clause or twoprepositional phrases . (c) A two-phrase sen-tence is not as difficult to read as a sentencewith a phrase and a clause, but is more diffi-cult than a fsingle-phrase sentence. (d) A sen-tence containing a phrase and a clause is moredifficult to read than a sentence without a clause.

While Table 2 shows the relationship of phraseto clause as it was conceived, Table 3 shows the

Table 3

actual phrase-clause ordering within each sen-tence of the nine paragraphs used in this study.

DESIGN, SCORING, AND ANALYSIS

Design

Although it would seem that a single factorialdesign would encompass all variations of themajor factors, there are several reasons forseparate consideration of the data yielded bythe subjects who saw pictures in isolation.(a) It was not the intent of the present studyto compare the relative abilities of picturesand written materials to evoke main idea state-ments; rather, the intent was to investigatethe possibility of differential effects on mainidea statements made after the reading of writ-ten materials with content relevant pictures andwritten materials without pictures. (b) Thefo.cus concerning the pictures was upon theportion of the main idea that pictures evoked,and that could not be indicated by statisticaltechniques which are appropriately used in afactorial design. (c) Finally, the assumptionof homogeneity of variance could not be metwhen the data from the pictures treatment wascompared with the data from the other fourreading conditions. The test used in this in-stance was Hartley's F Max Test (Winer, 1962,p. 92).

With the preceding rationale in minds, twofactorial designs were incorporated into this.study.

Sentence Location of Phrases and Clauses in Paragraphsfor Grades 1, 3, and 6 of Main Ideas A, B, and C

Main Ideas

Sentence

A

Grades

1 3

Grades Grades

6

1

2

3

4

a bpp pc

pp pc

cp

pc pc

1 3 6 1 3 6

P PP Pc p pc pc

p pc pc p p cp

P PP Pc p p cp

p p cp p pp pc

a "p" denotes phrase

"c" denotes clause

13

A4x2x2x 2 completely crossed factorialdesign was used to check the hypotheses con-cerning (a) the effects of pictures and direc-tions, i.e. , text alone, pictures with text andno direction, pictures and text with minimaldirection to view the picture, and pictures andtext with maximal direction as to the relation-ship between pictures and the text; (b) theeffects of readability levels of the material,i.e. , basic or subject's grade placement; (c)

the effects of the subject's grade placement,i.e. , third or sixth grade; and (d) the effectsof sex. Since six subjects of the same sexand grade placement were randomly assignedto each cell and there were 32 cells, a totalof 192 subjects was used in this design. Eachsubject saw three Pictures, or three paragraphs,or three paragraphs each with the appropriatecontent relevant picture. The order of presenta-tion was randomized, but the readability anddirection conditions were held constant acrossthe three replications for each subject.

To find to what extent the pictures alonecould carry the main idea message, the 144responses of 12 boys and 12 girls from the thirdgrade and like numbers from the sixth grade,i.e. , 48 subjects , who saw only the picturewere tabulated so that the kinds of main idearesponses evoked by the pictures were apparent.A 3 x 2 x 2 factorial design with repeated mea-sure of the first factor was employed to answerquestions concerning the differential effects ofthe three pictures , grade placement, and sexon the main idea responses.

Scale

The 7-point scale which was used to ratethe main idea responses in the present studywas the fourth scale developed by the personnelof the Laboratory for Research in Basic Skills,University of Wisconsin. The first scale wasbasically hypothetical and preceded attemptsto write materials with readability formulae dif-ficulty ranging from first through sixth gradewhile holding the main idea constant. The otherscales were developed after the material com-plexity was set and pilot studies were run.Three assumptions were reflected in the con-struction of the scales. (1) The optimal mainidea statement is a sentence, not a topic orphrase. (2) The optimal main idea contains thegeneral topic of the passage and the specificrestrictions of the passage. (3) The generaltopic portion of the main idea statement is hy-pothetically more important than, but not opera-tionally different from, the specific portion ofthe main idea statement.

The first main idea scale ranked main idearesponses in terms of the elements of the main

14

idea: (1) specific restriction; (2) general topic;(3) partially general-specific, i.e. , both ele-ments stated but not adequately or else not inproper relation to eaCh other; or (4) fully general-specific, i.e. , correct. In terms of a main ideaused in the present study, "Birds build nests indifferent places," the specific restriction isthe prepositional phrase "in different places,"while the general topic is "birds build nests ."Since it became clear during pilot studies thatsome children would not respond and some wouldgive wrong answers, a second scale, shown inFigure 2, was developed. The terms "relevant"and "irrelevant" mean "correct" and "incorrect"in actual practice.

Category Desci_laism

No responseIrrelevant general topic and/or

irrelevant specific restrictionRelevant specific restriction onlyRelevant general topic onlyIrrelevant general topic onlyRelevant general topic and

irrelevant specific restrictionRelevant general topic and relevant

specific restriction

ScaleValue

0

1

2

3

4

5

6

Fig. 2. The Second Main Idea Scale

The third scale was developed in responseto the judges' feelings that the second scalewas too elaborate. They had been scaling theresponses of first graders for some time andhad forgotten that older children gave more com-plex answers. The third scale did work on thefirst-grade responses , but was not important forthat reason. It was important because in it theoperational equivalence of the two elements ofthe main idea statement was recognized (Fig-ure 3.)

Category Description

No responseNo elements relevantOne element relevantBoth elements relevant

ScaleValue

01

23

Fig. 3. The Third Main Idea Scale

The last scale to be developed prior to thisstudy is shown in Figure 4. This scale was de-veloped after the judges and the investigatortried out complete sets of materials and direc-tions across six grade levels. It was then

ScaleValue

6

5

Category Description

Both 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 places.Horses , dogs, cats, and cows help the farmer in different ways.Lions, tigers, bears and cats use their claws for different things.

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

Animals put nests in different places.Animals use claws.

3 Irrelevant or incorrect material plus one element correctly stated OR one element

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

1 One or both elements too generally or specificallystated plus irrelevant or incorrectmaterial OR one or both elements too specifically stated OR only irrelevant or

incorrect material.e.g. Animals have sharp claws.

Birds hide their nests.Re-read 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.

Fig. 4. The Fourth Main Idea Scale and Example Responses

realized that the younger children, upon whoseresponses Scale 3 was based, had given moreconcise answers than were being noted as aresult of the final set of directions and materials.

The more elaborate fourth main idea scalediffered from the first two scales in that re-sponses were not ranked according to whichelement of the main idea was mentioned, andit was similar to the third scale in that themain idea elements were operationally equated.Main idea responses were ranked on the fourthmain idea scale by the degree of synthesizingwithin a main idea sentence. Optimal valuewas placed upon the complete main idea sen-tence; next were ranked sentences which did

not contain all the synthesized.material; thencame general phrases or titles which childrenmay be accustomed to make; and finally cameincorrect main idea statements, garbled phrases,and nonsynthesized responses.

Scoring

The main idea responses were recorded bythe investigator on specially constructed re-sponse sheets (Appendix D) which containedinformation as to the subject's name, grade,sex, treatment, and the pre-assigned random-ized order of the three sets of materials.

15

The main idea responses were independentlyrated by three experienced judges. The investi-gator did not participate in the rating of theresponses. All the main idea responses werecoded, scrambled, and typed so that the judgescould not know the treatment or have any infor-mation about the subject. The pages containingthe responses were randomly ordered for eachof the judges. At least two of the three judgesgave the same rating to all but 25 of the 740responses. This meant that two judges agreedon 96.7% of the main idea response ratings.The high interjudge agreement visas importantbecause that rating given by two or more judgeswas the rating assigned to the response andwas the basis of the subject's score. The sub-ject's score was the sum of the ratings givento his three main idea responses. It was pos-sible for a subject's score across the threemain ideas to range from 0-18, since eachmain idea rating ranged from 0-6. The 25 re-sponses which did not receive a common ratingwere resubmitted to the judges for scaling.After the second independent scaling, threeresponses still had not received common ratingsby two judges. These were discussed by thejudges and a consensus was reached.

Analyses

A4x 2x 2x 2 analysis of variance (fixedmodel) was used to check the hypotheses con-cerning (1) the effects of pictures and directions,i.e. , text alone, pictures with test and nodirection, pictures and text with minimal direc-tion to view the picture, and pictures and text

16

with maximal direction as to the relationshipbetween pictures and the text; (2) the effectsof readability levels of the material, i.e., firstgrade or subject's attained grade; (3) the effectsof grade placement, i.e. , third or sixth grade;and (4) the effects of sex. There were sixreplicates of the basic design, i.e. , 192 sub-jects. The score for each subject was the sumof the ratings given each of his three main idearesponses. As previously mentioned, a mainidea response was considered rated when twojudges gave it the same scale value, and therating then was that scale value.

To find to what extent the pictures alonecould carry the main idea message, the 144responses of the 48 subjects who saw only thepicture were separated from the, responses ofthe other 192 subjects. A 2 x 2 x 3 analysisof variance with repeated measure of the lastfactor (Pictures) was used. The other maineffects were Grade.and Sex. There were twelveresponse ratings per cell.

A method of post hoc analysis described byWiner (1962, pp. 85-89) and called by him the"Tukey (a) test" was used where significanteffects were found. This test was chosen be-cause Winer (p. 89) recommended it as appli-cable in a broad class of situations and assimple to apply. It was the present investi-gator's interpretation of Winer's statement,"In making tests on differences between allpossible pairs of means it (Scheffe method)will yield too few significant results," (p. 89)that the Scheffe eliminated too many possibili-ties for the appearance of truly significant dif-ferences.

III

RESULTS AND DISCUSSION

The discussion of results comprises threesections: (1) the analyses of the main idearesponses to the pictures only, (2) the analysesof the main idea responses to the reading con-ditions , and (3) consideration of the hypothesesstated in Chapter I in view of the data.

ANALYSES OF RESPONSES TO PICTURES

Scale ratings of the main idea responsesgiven by subjects who saw only the picture aregiven in Table 4 for descriptive purposes. Eachof the 48 subjects responded to all three pic-tures, so there are 48 rated main ideas for eachpicture. The responses are grouped at the lowerend of the scale. Main ideas placed in Category1 were either (a) too generally or specificallystated and contained irrelevant or incorrectmaterial, (b) too specifically stated, or (c)incorrect or irrelevant. Inspection of the re-sponses in Category 1 showed that the 119responses comprised 36 incorrect responses;36 overly general statements with extra mate-rial, e.g., birds in the spring; 35 overly spe-cific responses with extra material, e.g. , lion,tiger, bear, and cat are hungry; and 1 2 overlyspecific answers, e.g. , lion, tigers, bears ,and cats.

Table 4

Frequency of Main Idea Ratingsfor each Picture

Scale Ratings

Pictures 0 1 2 3 4 5 6

A 0 39 9 0 0 0 0

0 37 4 2 2 3 0

0 43 5 0 0 0 0

Total 0 119 18 2 2 3 0

Since the investigator was able to make dis-criminations among types of responses withinCategory 1, further refining of the scale, atleast for responses to pictures, is necessary.

To determine the effect of specific pictures,grade, and sex upon the main idea responses ,a repeated measures analysis of variance wasrun. The analysis is summarized in Table 5.The assumption of homogeneity of variance wasconfirmed with Hartley's F max test (Winer,1962, p. 9 2). Comparisons of the largest andsmallest variances were made within all maineffects. None of the variance ratios were sig-nificant at the .05 level.

Table 5

Analysis of Varianceof the Picture Response Ratings

Source df MS

Grades (G) 1 1.36 2.43Sex (S) 1 .11 <1

G x S 1 .25 <1

Subj. w. groupserror (between) 44 .56

Pictures (P) 2 2.59 5.08*G x P 2 2.09 4.1 0*S x P 2 .26 <1

GxSxP 2 .06 <1

P x subj. w. groupserror (within) 88 .51

* < .05

As shown in Table 5, there was no signifi-cant Grade or Sex effect, but the Pictures maineffect and the Grade x Picture interaction weresignificant. Inspection of the means presentedin Table 6 reveals that the responses evoked byPicture B differed from those evoked by PicturesA and C.

17

Table 6

Mean Main Idea Response Ratingsfor Subjects Viewing Pictures

Pictures Grade Sex

A B C Three Six Boys Girls

1.19 1.54 1.10 1.18 1.38 1.25 1.31

The post hoc comparison of response ratingmeans for the three pictures using the Tukey(a) test (Winer, 1962, p. 87), summarized inTable 7, revealed that the mean ratings of re-sponses to pictures A and C did not differ sig-nificantly, but both differed from the mean ofpicture B. As shown in Table 4, the responsesto B were rated as high as Category 5-oneelement correct plus the second element eithertoo generally or specifically stated- whichwould account for the greater mean rating.

Table 7

Ordered Means and All Pair-Wise Gaps:Pictures Main Effect

PictureMeans

Picture Means

1.10A

1.19 1.54

C-1.10 .09 .44*A-1.19 .35*B-1.54 IlOO

*R. < .05

The nature of the significant Grade x Pictureinteraction is clarified in Table 8, in whichmeans are presented by Grade and by Picture.

Table 8

Mean Main Idea Ratings: Picture x Grade

Grade

3

Pictures

A

1.171.21

1.21 1.181.88 1.04

The sixth grade subjects who saw picture Bhave a higher mean response rating than anyother group. The marked divergence of thesixth graders' responses to picture B is illus-trated in Figure 5.18

2.00

1.90

1.80

1.70

1.60

1.50

1.40

1.30

1.20

1.10

1.00

3rd Grade it

---- 6th Grade,/\

Pictures

Fig. 5. Mean Main Idea Ratings Obtained fromThree Pictures

The post hoc comparison of the six meansusing Tukey's test (Winer, 1962, p. 87), sum-marized in Table 9, confirmed the fact that theonly mean significantly different from any otheris that of the sixth-grade subjects who viewedpicture B. Therefore, the significant main ef-fect of Pictures , mentioned previously, is notirrespective of grade of subject, but is withinGrade 6 only.

Table 9

Ordered Means and All Pair-Wise Gaps:Picture x Grade

Means C-6 A-3 C-3 B-3 A-6 B-6

1.04 1.17 1.18 1.21 1.21 1.88

1.04 .13 .14 .17 .17 .84*1.17 ---- .01 .04 .04 .71*1.18 .03 .03 .70*1.21 .69*1.21 .69*1.88

* 2. < .05

An inspection of the pictures in Appendix Cshowed that picture B's main idea, "Birds buildnests in different places," is the only one illus-trated through a differential placement of objects.

It could be said that the spacing meant some-thing in picture B, but not in pictures A and C.

ANALYSIS OF RESPONSES TO TEXT

The main effects in the analysis of varianceof the main idea responses of subjects who readthe paragraphs were: reading conditions, reada-bility levels, grades, and sex. The assumptionof homogeneity of variance was confirmed withHartley's F max test (Winer, 1962, p. 92).Comparisons were made within each main effect.None of the variance ratios approached signifi-cance at the .05 level.

The means of the main idea scores are givenby major variable in Table 10. The analysis ofvariance, summarized in Table 11, revealed nosignificant differences among the reading con-ditions- i.e. (1) text without picture (Text),(2) picture with text but no direction to use the

Table 10

picture (PT nod), (3) picture with text and mini-mal direction to look at the picture (PT min),and (4) picture with text and direction as to thesupposed relationship between picture and text(PT max). Sex, too, had no significant effect.

The readability levels and grade effects weresignificant at the .01 and .001 levels respec-tively. Since both effects contained but onepair of means a post hoc test was unnecessary.Inspection of Table 10 shows that the mean ofthe subjects who read first-grade level materials(Easy) is higher than the means of subjects whoread materials at their grade level. Table 10also shows that the mean of the sixth-gradesubjects is higher than that of the third-gradesubjects.

Two interactions were also significant:grade x sex (2. < .05) and readability level xsex (2. > .05 < .10). Inspection of the means

Mean Main Idea Scores of 192 Subjects Reading Paragraphs

Reading Conditions

Text PTnod PTmin

9.96 10.83 9.48

PTmax

9.71

Readability

Easy Grade

10.91 9.08

Grade Sex

Three Six Boys Girls

8.66 11.33 9.95 10.04

Table 11

Analysis of Variance of the Main Idea Response Ratings

Source

Reading ConditionsReadability LevelsGradeSex

(RC)(RL)

(G)(5)

RC x RLRC x GRC x SRL x GRL x SG x SRC x RL x GRC x RL x SRCxG xSRLxG xSRCxRLxGxS

Error Within

df MS

3

1

1

16.84159.51344.01

1.1210.60***22.86* ***

1 .42 < 1

3 8.56 < 13 8.56 < 1

3 18.92 1.261 45.05 2.99*1 11.51 < 1

1 81.38 5.41**

3 17.05 1.133 .23 <13 27.76 1.841 4.05 < 1

3 16.98 1.13

160 15.05

* < .10** 2. < .05

*** p. < .01****2. < .001

19

of the main idea scores for grade x sex givenin Table 12 reveals that the boys have the low-est mean (third grade) and the highest mean(sixth grade). The slopes plotted in Figure 6serve to clarify the nature of the interaction.

Table 12

Mean Main Idea Ratings: Grade by Sex

Sex

Grade Boys

3 7.966 11.94

3 6

Grades

Table 13

Ordered Means and All Pair-Wise Gaps:Grade x Sex Interaction

Means by Means by Grade & SexGradeand Sex 3-Boys 3-Girls 6-Girls 6-Boys

7.96 9.35 10.73 11.94

3-Boys7.96

Girls 3-Girls9.35

9.35 6-Girls10.73 10.73

6-Boys11.94

Fig. 6. Mean Main Idea Scores of Boys andGirls at Grades 3 and 6. Ordered Means and All Pair-Wise Gaps:

Readability Levels x Grades Interaction

WIP 1.39 2.77* 3.98*

,. MD amp 1.38 2.59*

1 . 21

* 2. < .05_

should be greater for sixth than third grade.Table 14, which contains the means of thethird- and sixth-grade subjects according toreadability of material, and Figure 7, whichillustrates the differences, show this. Table14 also shows that the sixth-grade subjectsreading first-grade material have the highestmean and the only mean to differ significantlyfrom any of the other three. Therefore, thesixth-grade subjects read the first-grade para-graphs significantly better than the at-grade-level paragraphs , but the third graders did not.

Table 14

Post hoc comparisons of the relevant meansby the Tukey (a) test, summarized in Table 13,show that the sixth-grade boys' mean is signifi-cantly different from both boys' and girls' meansin third grade, but not different from that of thesixth-grade girls. The sixth-grade girls' meanis not different from the sixth-grade boys or'the third-grade girls' mean. Only the third-grade boys' mean differs significantly from thatof the sixth-grade girls'.

The readability x grade interaction was feltto have practical significance because it aidsin the interpretation of the significant readabilitylevels effect. As mentioned previously, thematerials considered under the treatment heading

Means byGradeRead ability

Means by Grade and Readability

3-Gr8:23

3-Easy 6-Gr9.08 9.94

6-Easy12.73

3-Gr8. 23

3-Easy9.08

6-Gr9.94

6-Easy

,MM. .11. .85 1.71

.86

4.50*

3.65*

2.79*

grade levels were not of equal difficulty, rather 12. 73they were rated as at the attained grade of thesubjectthree or six. Furthermore, the rela- * < .05tive difference between the readability levelsof materials read within grade was not the same. CONSIDERATION OF QUESTIONS AND HYPOTHESES

A first-grade paragraph is easier to a sixth graderthan a third grader; therefore, the differences The results of the analyses are summa-between the means according to material read rized in relation to the two questions and

4.6.

GPO 80111-2(125

1341/4 3rd Grade

1 2 ---- 6th Grade

11

s. 10N*

c/)9

8

EasyReadability

GradeLevels

Fig. 7. Mean Main Idea Scores for Grades 3and 6 with Paragraphs at Two Reada-bility Levels

six hypotheses which served as the bases ofthe present study.

The first general question was, "To whatextent can pictures carry a main idea message?"The answer to this question was sought by meansof a tabulation by category of main idea responsesof subjects viewing only the pictures. The re-sponses to the pictures received generally lowratings as main ideas. In fact, 119 of 1 44 pos-sible responses were rated as Category 1 (Table4), which was defined asOne or both elements too generally or specificallystated plus irrelevant or incorrect material ORone or both elements too specifically stated OR

only irrelevant or incorrect material.e.g. Animals have sharp claws.

Birds hide their nests.Re-read paragraph or a single sentence.How safe the farmer keeps the farm.

(Fig. 4)

The second general question was, "To whatextent can content relevant pictures reinforceor aid in the comprehension of the main idea ofa paragraph?" The answer to this question,which was restated as the first of six workinghypotheses , was sought by means of an analy-sis of variance of main idea response ratings.The analysis of variance is summarized in Table11.

Hypothesis I: The addition of a content relevantPicture to written text enhances subjects' mainidea responses.

The analysis of variance summarized in Table11 shows that the Reading Conditions (RC) maineffect was not significant. Since the onlymethodological difference between two of thefour RC was the presence or absence of pic-tures, Hypothesis 1 is rejected within the limi-tations of the materials used in the present study.

Hypothesis 2: Subjects who are directed toview a content relevant picture before readinggive better main idea responses than subjectsnot so directed.

As previously mentioned, the RC main effectwas not significant. This means that the dif-ference between the means of the groups withand without direction to view the picture wasnot significant. Hypothesis 2 is, therefore,rejected within the limitations of the study.

Hypothesis 3: Subjects who are told the pur-pose of a content relevant picture and told toexamine it before reading give better main idearesponses than subjects not told the purposeof the picture and subjects to whom the pictureis not mentioned.

Hypothesis 3 is also rejected within thelimitations of the materials used in *the studybecause.there was no significant RC main effect.That is, the only methodological differenceamong three RC was direction, and the meanof the group receiving the most thorough direc-tion concerning the picture did not differ signifi-cantly from the means of the other two groups.

Hypothesis 4: Main idea responses to para-graphs written as simply as possible, yet con-sistent with the main idea, are better than mainidea responses given to paragraphs rated asappropriate for the subject's attained gradelevel, i.e., third or sixth.

Hypothesis 4 is accepted conditionally be-cause both the readability levels (RL) maineffect and the RL x Grade interaction were sig-nificant (Table 11). The acceptability of thehypothesis is limited to sixth-grade subjectssince the RL x Grade interaction and subsequentpost hoc tests (Table 1 4, Fig. 7) show thatonly the responses of the sixth graders readingbasic material were better than their counter-parts who read grade level material.

Hypothesis 5: Pictures have a greater effect onboys' main idea responses than on girls' mainidea responses .

Neither the analysis of variance of ratingsof responses to pictures in isolation, summa-rized in Table 5, nor the analysis of varianceof ratings of responses to paragraphs with con-tent relevant pictures, summarized in Table 11,showed a significant Sex main effect. There-fore, Hypothesis 5 is rejected within the limi-tations of the materials used in the presentstudy.Hypothesis 6: Sixth graders give better mainidea responses than third graders.

Both of the analyses of variance, i.e. of thepicture response ratings and of the picture withparagraph response ratings, had a significantGrades main effect, but in both analyses a sig-

21

nificant interaction established limitations .The analysis of response ratings to pictureshad a significant grade x picture interaction(Table 9) which showed that only the pictureabout birds elicited significantly better mainidea responses from sixth graders than fromthird graders . The analysis of response rat-ings of paragraphs with pictures had a signifi-cant grade x sex interaction (Table 13) whichshowed that only the sixth-grade boys' mean

22

response rating was significantly better thantheir third-grade counterparts.

Therefore, Hypothesis 6 is accepted withthe following limitations: (a) when respondingto pictures, sixth graders make better mainidea statements to some pictures; and (b) whenresponding to pictures with paragraphs , sixth-grade boys respond better than third-grade boys.These statements are limited by the materialsand the methodology used in the present study.

IV

SUMMARY, LIMITATIONS AND CONCLUSIONS, AND IMPLICATIONS

SUMMARY

This study was designed to investigate theeffects of content relevant pictures upon themain idea responses of third- and sixth-gradestudents. The pictures were drawn to representthe main idea of the paragraphs , and the para-graphs were written Within limits imposed byreadability formulae.

Sixty boys and sixty girls from the thirdgrades and like numbers from the sixth gradesof the nine public elementary schools in Janes-ville, Wisconsin, were randomly chosen fromamong all pupils not in special classes.

In the following five treatments , to whichsubjects were randomly assigned by grade andsex, the subject was asked for the main ideacarried by the materials: (1) after viewing acontent relevant picture in isolation; (2) afterreading a paragraph without any pictorial illus-tration of its main idea; (3) after reading a para-graph accompanied by an illustration of its mainidea, but without directions to attend to thepicture; (4) after reading a paragraph accom-panied by an illustration of its main idea, withdirections to attend to the picture; and (5) afterreading a paragraph accompanied by an illustra-tion of its main idea with directions to attendto the picture because it illustrated the mainidea and was, therefore, an aid to comprehen-sion of the rnain idea. One half of each of theboys and girls frorn Grades 3 and 6 within thefour treatrnents using written materials (Treat-ments 2, 3, 4, 5) saw paragraphs rated at afirst-grade difficulty level by the Spache formulawhile the other half of the subjects saw mate-rials judged as at their attained grade level indifficulty by the Spache (Grade 3) or the Dale-Chall (Grade 6) readability formula.

Although the material varied as to form andreadability level, only three main ideas wereused, and all were seen in random order byevery subject. Each main idea was developedin a picture and in a basic four-sentence para-

graph plus paragraphs written at third- andsixth-grade difficulty levels. The pictureswere 4" x 6" black and white drawings judgedby the investigator and the artist as the bestof several illustrations of each main idea doneby the artist. Operationally, each main ideastatement in this study has two elements ageneral topic and a specific restriction andrequires the same type of conceptualizing inboth elements.

All subjects were tested individually in theirrespective schools by the investigator. Thedirections and warmup task were standard forall subjects. Responses were recorded verba-tim, randomized, and rated by three experi-enced judges who used a 7-point scale.

The following analyses were performed:(a) The frequency of responses to the picturesin isolation was tabulated within each categoryon the scale. (b) The scale values of the threemain idea responses from each of 48 subjectsviewing the pictures alone were submitted toa 2 x 2 x 3 sex, grade, picturesrepeatedmeasures analysis of variance. (c) The sumof the three main idea ratings for each of the192 subjects reading the text with or withoutpictures was submitted to a 4x 2x 2x 2reading conditions , readability level, grade,and sex analysis of variance.

The salient results of the analyses can bebriefly summarized. (1) The tabulation of re-sponses to pictures in isolation showed thatthe responses had generally received low rat-ings on the main idea response scale. (2) Theanalysis of variance of the main idea responsesto pictures in isolation showed that generallythere were no significant differences betweenthe responses of boys and girls and of subjectsin Grades 3 and 6. Responses to one picturewere better than responses to the other two,but only for sixth-grade subjects. (3) Theanalysis of variance of the responses to theparagraphs with or without pictures and direc-tions showed that the addition of the pictures

23

and directions did not affect the adequacy ofthe main idea responses. Although both reada-bility and grade effects were significant, onthe basis of post hoc analyses it was suggestedthat reading easy materials led to significantlybetter main idea responses only among the sixthgraders and that the boys alone accounted forthe difference. In fact, there were no signifi-cant differences between the responses of girlsin third and sixth grades or between the responsesof boys and girls within each grade.

LIMITATIONS AND CONCLUSIONS

The generality of the conclusions is limitedby the following factors. (1) The paragraphsare short, expository in style, specificallycontrolled as to the number of words, sentences,and referents to the class nouns in the mainidea sentence, and specifically controlled asto the vocabulary level, content, and grossreadability score. (2) The pictures were drawnto illustrate the main idea of a paragraph. (3)The structure of the main idea sentence and thescale used to judge it are specific to the study.(4) The main idea is not explicitly stated in theparagraphs; therefore, the results are not ap-plicable to paragraphs with explicitly statedmain ideas. (5) Subjects' reading level wasnot controlled, so no comparisons can be madeto specific segments of the population as de-fined by reading level. (6) Only third and sixthgrades were sampled. (7) The sample is froma Midwestern city of 35,000 which is not knownto contain the extreme socioeconomic groupings.

The results of the analyses warrant the fol-lowing tentative conclusions:

1.. Pictorial representations of a main ideado not necessarily evoke adequate main idearesponses.

2. The addition to an expository paragraphof a content relevant picture similar to thoseused in this study does not enhance the scaleratings of elicited main idea responses. Fur-thermore, directions to use the picture as anaid to comprehension of the main idea do notenhance the scale ratings of the main idea re-sponses evoked.

3. In general, responses evoked by picturesare not affected by the sex or grade placementof the subject, but sixth graders' responsesevoked by a picture expressing the main ideathrough the placement objects were better than(a) third graders' responses irrespective ofpicture and (b) sixth graders' responses to theother two pictures.

4. When the pictures used for this studywere presented alone, very low order attemptsat stating a main idea were evoked. In view

24

of this fact, and the defensible expectationthat content relevant pictures and directionsto peruse them ought to enhance performance,an inescapable conclusion is that perhapsthe inadequacy of the pictures was responsiblefor the present negative results. If illustra-tions capable of evoking higher order main ideastatements had been devised, quite differentresults might have been obtained.

5. Although some investigations of the roleof pictures in reading comprehension have shownsex differences favoring boys , the present datashowed no significant differences between theresponses of boys and girls within the samegrade.

6. Within the limitations of the expositorystyle and brevity of the paragraphs used in thisstudy, varying the readability level of the para-graphs from that equal to the reader's attainedgrade to that designated here as appropriatefor a first-grade student enhances the mainidea statements of sixth-grade students , butnot third-grade students.

7. Further elaboration of the main ideascale, particularly among the lower valuedcategories, seems to be essential becausefurther classification of low level responseswas necessary.

IMPLICATIONS

Implication For Instruction

The failure of the pictures in this study toaid in the comprehension of the main idea im-plies that pictures in textbooks should not beexpected to aid to any great extent in the com-prehension of a main idea of a paragraph. Thepreceding statement is made because it seemsillogical to expect textbook pictures, whichmay be drawn for many other reasons, to beaids in comprehension of a main idea whenpictures drawn to represent a main idea cannotaffect comprehension. The fact that picturesdo serve other purposes is seen in Vernon's(1954) finding that specific facts pictoriallyrepresented were recalled relatively more timesthan those facts not pictorially represented.Furthermore, Whipple (1953), among others,has found that pictures increase the interestappeal of the book, and, in general, teachersuse pictures to introduce stories and new words.

Implications For Research

Implications for further research can befound with regard to subjects and the natureand evaluation of the main idea responses.

A need for investigation of the picture andcomprehension relationship among poor readerscan be said to exist because there are no stud-ies in the area upon which to base conclusions.There .are two studies in the literature whichdeal with the poor reader (Halbert, 1943; Wein-traub, 1960) and the picture and reading com-prehension relationship, but both must be viewedwithin limitations. Among other things, Halbertused a sample of impoverished Kentucky hillpeople, and Weintraub used the lowest 10% ofa superior-achieving group of second gradersas his poor reader sample. Another reason forinstigating research on poor readers is that thematerials in most previous studies have notbeen so difficult that a reader was forced toseek outside help. Only Vernon (1954) notedinformally that she had chosen materials whichwere too hard for some students, and she didnot develop this area of investigation.

The material used in the study was exposi-tory, short, well-structured, and did not con-tain a stated main idea. Since these are notthe characteristics of all textbooks and basalreaders that third- and sixth-grade childrenread, investigation of the effects of picturesupon the comprehension of the main idea oflong, rambling, narrative material with statedmain ideas might yield significantly differentfindings.

The pictures used in the present study wereconventional black-and-white drawings meantto illustrate and therefore reinforce what wasin the written passage. These very limitationsshould be studied. There are no studies avail-able at this time which focus on the relationshipof color to comprehension or on the role of con-ventionality of size or format of the illustrationto comprehension. It can be speculated thatcolor will influence the extraction of a spe-cific fact from a picture and hence increase thelikelihood of its being understood in the text,but the possible relationship of color to themain idea is not so clear. Furthermore, it ispossible that pictures which represent a portionof the main idea or other relationship among thesentences of a paragraph would be of value.Certainly, it is evident that comic strips cancarry messages without verbal materials, butthe effects of a similar format on reading cpm-prehension is not a matter of record.

Even though the main ideas used in the pres-ent study are of similar construction, it was

discovered that the illustrations are different.One picture which carried a main idea messagethrough the spacing of objects was found. Sincethis proved to be the best stimulus, further de-velopment and testing of main ideas or principleswhich can be expressed through the placementof the objects in a picture seems justified. Thepictures from the present study might be re-drawn to maximize the spacing of the objectsin an attempt to evoke responses which wouldreceive higher scale ratings. If this were pos-sible, the suggestion that pictures evokinghigher order main ideas might be the key to thesuccessful use of pictures as aids to compre-hension of the main idea could be investigated.

The response mode in the present study wasnot the common multiple-choice form; instead,personally formulated statements were required.Since a rather exhaustive search of the litera-ture produced only one study (Vernon, 1962) ofthe multiple-choice and essay examinationrelationship, it was concluded that a lack ofempirical data about the relationship of theresponse formats exists. Also there is no workwith elementary or secondary school children,since Vernon used college students. If statinga main idea is not the same thing as recognizingit in a multiple-choice examination, and statinga main idea iS more like an independent.studytask, then no reading achievement examinationtests a skill which is probably more like thatused in independent studying. Suppoil for thisposition can be found in Thomas C. Barrett'sincorporation of reading comprehension skillsinto a cognitive function hierarchy. 2 He pre-sented recognition and formulation of main ideasas different skills.

Since further classification of responsesevoked by the pictures was necessary, a scalewas constructed which makes finer discrimina-tions among responses possible (Figure 8). Thescale, which has twelve categories instead ofseven, also does not make a judgment as to therelative values of overgeneralizing or becomingtoo specific. It is suggested that this scalemay be of value in further investigations of thecharacteristics of main idea responses elicitedfrom various populations of readers under vary-ing conditions of material and direction.

2Unpublished paper entitled "Taxonomy ofCognitive and Affective Dimensions of ReadingComprehension," 1966.

25

Category DescriptionValue

No response.0

Irrelevant or incorrect material.1

One element too generally or specifically stated plus irrelevant or incorrect data. 2

One element too generally or specifically stated. 3

Two elements too generally or specifically stated plus incorrect or irrelevant material. 4

Two elements too generally or specifically stated. 5

One element correctly stated plus irrelevant or incorrect data. 6

One element correctly stated.7

One element correctly stated plus one element too generally or specifically statedplus irrelevant or incorrect data. 8

One element correctly stated plus one element too generally or specifically stated. 9

Two elements correctly stated plus irrelevant or incorrect data. 10

Two elements correctly stated.11

26

Fig. 8. The Fifth Main Idea Scale

DIRECTIONS

A PPENDIX A

PR ETASK

Hello, thcrc. My name isWhat's your name? Good.did you noticc that I used a sentence to tellyou my name? READ AND FILL IN NAME.You said your name wasyou make up a sentence just likeme your name? Good. I have your sentence

. Canminc to tell

written here too. READ AND FILL IN NAME.Suppose I say, "My teacher's name is Mr.Brown." SHOW SENTENCE AS YOU SAY IT.Can you make up a sentence like mine to tellmc 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?

RESPONSE SHEETS

My name is

My name is

My teacher's name is Mr. Brown.

My teacher's name is

APPENDIX B

DIRECTIONS

PICTURE

Very good. Now we're going to do somethinga little different. I have some pictures for youto look at. As you look at each one think aboutwhat the whole picture says. When you finishlooking, make up just one sentence in your ownwords that says what the whole picture tells you.GIVE MATERIALS TO S.

S LOOKS AT PICTUREWhat does this whole picture tell you? Youmay look back at the page if you wish. IF NORESPONSE IN 10 SECONDS, REPEAT QUESTION.Now let's do the same thing on another page.Would you look at this and think about what thewhole picture says ?

PT Nod

Very good. Now I have some more sentencesfor you to read, but we're going to do somethinga little different. This time, as you read, thinkabout what all the sentences together say.When you finish reading, make up just onesentence in your own words that says what allthe 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 MATERIALS.

S READS MATERIAL

What do all of these sentences together tellyou? You may look back at the page if youwish. IF NO RESPONSE IN 10 SECONDS,REPEAT. Now let's do the same thing on anotherpage. Would you read this and think about whatall the sentences say? GIVE MATERIALS.

PT Min

Very good. Now I have some more sentencesfor you to read, but we're going to do some-

28

thing a little different. This time, before youread, look carefully at the picture. Then, asyou read, think about what all the sentencestogether say. When you finish reading, makeup just one sentence in your own words thatsays what all the sentences tell you. You canread this silently to yourself. If you donot know any of the words, ask me and I willtell you. GIVE MATERIALS.

S READS MATERIAL

What do all of these sentences together tellyou? You may look back at the page if youwish. IF NO RESPONSE IN 10 SECONDS,REPEAT. Now let's do the same thing onanother page. Would you look at the picture,then read this and think about what all thesentences say?

PT Max

Very good. Now we're going to do somethinga little different, I have some pictures withsentences that tell about them. Before youread, look at the picture and think what thewhole picture shows. Then, as you read,think about what all the sentences togethersay. When you finish reading make up justone sentence in your own words that says whatthe whole picture and all the sentences to-gether say. You can read this silently toyourself. If you do not know any of the words,ask me and I will tell you. GIVE MATERIALS.

S READS MATERIAL

What does all this together tell you? You maylook back at the page if you wish. IF NO RE-SPONSE IN 10 SECONDS, REPEAT. Now let'sdo the same thing on another page. Would youlook at the picture, think what it all says,and then read and think about what all the sen-tences say?

GPO 8011-202.4

APPENDIX C

MATERIALS

PICTURE A

Horses help farmers to work. Cats keep farmers' corn safe from mice.Cows help farmers by giving milk. Dogs help farmers to watch the barnyard.

Horses help the farmers to plow their wheat and corn fields in the spring-time. Cats help the farmers to keep their corn safe from greedy and hungrymice. Cows give milk to the farmer and his large family. Dogs help farmersby guarding the whole barnyard when it is night.

Horses help farmers to plow their wheat and corn fields when springcomes. Cats help the farmers to keep their valuable corn safe whenevergreedy, hungry mice try stealing the grain. Cows give very tasty milk whenthe farmer and his large family of six need it. Dogs help the farmers byguarding the entire barnyard particularly carefully when the people are asleep.

29

30

PICTURE B

Robins may build nests under a roof. Bluejays like nests in trees. Ducksmake nests in tall grass. Woodpeckers make nests inSide wood fence posts.

Robins build their nests under the roofs of houses and barns. Bluejayslike nests in trees that have many big branches. Ducks, however, carefullymake their nests in the wild rice, high weeds, or tall marsh grass near other

duck nests. Woodpeckers sometimes make nests inside old wood fence posts.

Robins build their nests under house and barn roofs where they overhang

the building. Bluejays like nests in leafy trees that have big branches. Ducks,however, carefully make nests in wild rice, high weeds, or tall marsh grassthat may contain many duck and other wildlife homes. Woodpeckers sometimesmake nests that are quite soft and comfortable inside old wooden fence posts.

PICTURE C

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

Lions use their claws to hold the food that they are eating. Bears havelong claws for digging up many different roots and insects. Cats' claws helpthem to climb the most difficult trees in a hurry. Tigers use their strong clawsfor fighting their many enemies in the woods.

Lions use their claws to grasp the food that they have just caught. Bearshave very long claws that help them to dig up and tear apart various roots andinsects. Cats' claws help them if they are forced to climb even the mostchallenging trees very quickly. Tigers use their strong, powerful claws tofight their jungle enemies whom they frequently attack.

Name

School

Grade

Sex

APPENDIX D

RESPONSE SHEET

Examiner

Task

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

"What do you mean by 'they'?"

Task

REFERENCES

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

Burdick, J. G. A study of cross-section draw-ings as technical illustrations in high schoolscience textbooks. Unpublished doctoraldissertation, Syracuse University, 1959.

Dale, E., & Chall, J. S. A formula for pre-dicting readability. Educational ResearchBulletin, 1948, 27, 11-20. (a).

Dale, E., & Chall, J. S. A formula for pre-dicting readability: Instructions, Educa-tional Research Bulletin, 1948, 27, 37-54.(b)

Davis , F. B. Fundamental factors of compre-hension in reading. Psychometrika, 1944,9, 185-197.

Davis, G. A. A note on two basic forms ofconcepts and concept learning. Journal ofPsychology, 1966, 62, 249-254.

Davis, 0. L. , Hicks, L. C., & Bowers, N. D.The usefulness of time lines in learningchronological relationships in text materials.The Journal of Experimental Education, 1966,3, 22-25.

Davis , 0. L., & Slobodian, J. J. Teacherbehavior toward boys and girls during firstgrade reading instruction. American Educa-tional Research Journal, 1967, 4, 261-269.

Gagne, R. M. The conditions of learning.New York: Holt, Rinehart and Winston, 1965.

Goodykoontz, B. The relation of pictures toreading comprehension. Elementary EnglishReview, 1936, 13, 125-130.

Gray, W. S. Reading. In C. W. Harris (Ed.)Encyclopedia of Educational Research. NewYork: Macmillan, 1960. Pp. 1086-1135.

Halbert, M. An experimental study of chil-dren's understanding of instructional mate-rials. Bulletin of School Service, 1943,15(4), 7-60.

Harris , A. J. How to increase reading ability.New York: David McKay, 1961.

Kambly, P. E. Science textbook illustrations.Science Education, 1943, 27, 17-19.

Klare, G. R. A table for rapid determinationof Dale-Chall readability scores. Educa-tional Research Bulletin, 1952, 31, 43-47.

Lewernz, A. S. Some results of a visual edu-cation class in junior high school socialstudies taught with the aid of flat pictures.Educational Research Bulletin, 1929, 84-16.

Miller, W. A. Reading with and without pic-tures. Elementary School Journal, 1938,39, 676-682.

Russell, D. H. Children learn to read. Bos-ton: Ginn, 1961.

Spache, G. A new readability formula for pri-mary-grade reading materials. ElementarySchool Journal, 1953, 53, 410-413.

Spaulding, S. Research on pictorial illustra-tion. Audio-Visual Communication Review,1955, 3, 35-45.

Stone, C. R. Measuring difficulty of primaryreading material: a constructive criticismof Spache's measure. Elementary SchoolJournal, 1957, 57, 36-41.

Strang, A. M. A study of gains and losses inconcepts as indicated by pupils' readingscores after the addition of illustrations tothe reading matter. Unpublished doctoraldissertation, Temple University, 1941.

Vandermeer, A. W. Relative contributions tofactual learning of the pictorial and verbalelements of a filmstrip. The School Review,1950, 58, 84-89.

Vernon, M. D. The value of pictorial illustra-tion. British Journal of Educational Psychol-ogy, 1953, 23, 180-187.

Vernon, M. D. The instruction of children bypictorial illustration. British Journal ofEducational Psychology, 1954, 24, 171-179.

Vernon, P. E. The determinants of reading com-prehension. Educational and PsychologicalMeasurement, 1962, 22, 269-286.

33

Weintraub, S. A. The effect of pictures onthe comprehension of a second grade basalreader. Unpublished doctoral dissertation,University of Illinois, 1960.

Weintraub, S. A. Illustrations for beginningreading.67. (a)

Weintraub, S

34

Reading Teacher, 1966, 20, 61_

. A. Sex differences in reading

achievement. Reading Teacher, 1966, 20,155-165. (b)

Whipple, G. Appraisal of the interest appealof illustrations. Elementary School Journal,1953, 53, 262-269.

Winer, B. J. Statistical principles in experi-mental design. New York: McGraw-Hill,1962.

GPO 809-282-2