DOCUMENT RESUME

ED 101 318 CS 001 614

AUTHOR Levin, Joel R.TITLE What Have We Learned About Maximizing What Children

Learn? Theoretical Paper No. 49.INSTITUTION Wisconsin Univ., Madison. Research and Development

Center for Cognitive Learning.SPONS AGENCY National Inst. of Education (DHEW), Washington,

D.C.REPORT NO WRDCCL-TP-49PUB DATE Jul 74CONTRACT NE-C-00-3-0065"NOTE- 31p.; Report from the Project on Children's Learning

and Development

EDRS PRICE MF-$0.76 HC-$1.95 PLUS POSTAGEDESCRIPTORS *Cognitive Processes; Comprehension; *Educational

Research; Elementary Education; Elementary SchoolStudents; Learning Processes; *Paired AssociateLearning; *Pictorial Stimuli; *Visual Stimuli

ABSTRACTThis summary of a program If research in children's

learning describes the effects of pictures and visual imagery onchildren's cognitive performance. The' role of individual differencesis highlighted throughout the paper, as are variables whichpotentially limit the effects discussed. Some of the conclusions werethat in a large number of learning tasks children seem to learnbetter when the materials are presented in pictures than when thematerials are presented in words; there appear to be limitations onthe effectiveness of pictures in terms of both learner and taskdifferences; subject-generated organizational strategies (visualimagery in particular) greatly facilitate associative learning; andthe benefits from an imagery strategy have also been obtained ontasks requiring comprehension of a passage. Areas for furtherinvestigation are suggested. (RB)

U.S DEPARTMENT OF HEALTHEDUCATION A WELFARENATIONAL INSTITUTE OF

EDUC: TIONTHIS DOCUMENT HAS BEEN REPRODUCED EXACTLY AS RECEIVED FROMTHE PERSON OR ORGANIZATION ORIGINATING IT POINTS OF VIEW OR OPINIONS

STATED 00 NOT NECESSARILY REPRESENT OFFICIAL NATIONAL INSTITUTE OFEDUCATION POSITION OR POLICY

.Theoretical Paper No. 49

WHAT HAVE WE LEARNED ABOUTMAXIMIZING WHAT CHILDREN LEARN?

by

?oel R. Levin

Report from the Project onChildren's Learning and Development

Wisconsin Research and DevelopmentCenter for Cognitive LearningThe University of Wisconsin

Madison, Wisconsin

July 1974

QD 1974 - The Regents of the University or Wisconsin System for the Wisconsin

Research and Development Center for Cognitive Learning,

Copyright is claimed only during the period of development, test, and evaluation,

unless authorization is received from the National Institute of Education to

claim copyright on the final materials. For the current copyright status, contact

either the copyright proprietor or the National Institute of Education.

Published by the Wisconsin Research and Development Center for Cognitive Learning,

suppIrted in part as a research and development center by funds from the National

Institute of Education, Department of Health, Education, and Welfare, The opinions

expressed herein do not necessarily reflect the position or policy of the National

Institute of Education and no official endorsement by that agency should be inferred,

Center Contract No. NE-C.00.3.0065

Li

HEST COPY AVAILABLE

Statement of Focus

Individually Guided Education (IGE) is a new comprehensive system ofelementary education, The following components of the IGE system are invarying stages of development and implementation: a new organization forinstruction and related administrative arrangements; a model of instructionalprograming for the individual student; and curriculum components in prereading,reading, mathematics, motivation, and environmental education. The develop-ment of other curriculum components, of a system for managing instruction bycompiler, and of instructional strategies is needed to complete the system.Continuing programmatic research is required to provide a sound knowledgebase for the component,: under development and for improved second generationcomponents, Finally, systematic implementation is essential so that the prod-ucts will function properly in the IGE schools.

The Center plans and carries out the research, development, and imple-mentation components of its ICE program in this sequence: (1) identify theneeds and delimit the component problem area; (2) assess the possible con-straintsfinancial resources and availability of staff; (3) formulate generalplans and specific procedures for solving the problems; (4) secure and allo-cate human and material resources to carey out the plans; (5) provide foreffective communication among personnel and ef,Acient management of activi-ties and resources; and (6) evaluate the effectiveness of each activity andits contribution to the total program and correct any difficulties through feed-back mechanisms and appropriate management techniques,

A self-renewing system of elementary education is projected in eachparticipating elementary school, i.e., one which is less dependent on externalsources for direction and is more responsive to the needs of the children attend-ing each particular school, In the IGE schools, Center-developed and othercurriculum products compatible with the Center's instructional programing modelwill lead to higher student achievement and self-direction in learning and inconduct and also to higher morale and job satisfaction among educational per-sonnel, Each developmental product makes its unique contribution to IGE asit is implemented in the schools. The various research components add to theknowledge of Center practitioners, developers, and theorists

4

Acknowledgments

I am grateful to Sharon Stevens for her thoughtful editorialsuggestions regarding this paper.

iV

Table of Contents

Page

Acknowledgments iv

List of Tables and Figures vii

Abs tract ix

Introduction 1

II. Pictures as Learning Materials 3

Evidence for Picture Superiority 3The Role of Individual Differences: An Empirically Derived

Representation 4

Pictures and the Comprehension of Prose

A Footnote to the Benefits of Pictures in Learning 7

III. Visual Imagery as an Organizational Strategy 9

The Development of a Visual Imagery Strategy 10

Some Further Evidence for the Age-Imagery Relationship -:12

Visual Imagery and the Comprehension of Prose 13

A Footnote to the Benefits of Visual Imagery in Learning 16

IV. Inferred Cognitive Strategies: A Proposed Approach and SomeIllustrations 19

V. Summary 23

References 25

6

List of Tables and Figures

Table

1 Correspondence Between Initial and SubsequentLearner-Type Classifications

Page

6

Figure

1 A conceptualization of individual differences asrelated to the potential effectiveness of a cognitivestrategy 5

2 Reading comprehension of two types of learnersunder different instructional conditions. 15

A proposed approach to the investigation of inferredcognitive processes. 20

7

Abstract

A summarization of a program of research in children's learn-ing is presented. Major topics include the effects of pictures andvisual imagery on children's cognitive performance. The role ofindividual differences is highlighted throughout the paper, as arevariables which potentially limit the effects discussed. Contin-ued investigation into the nature and development of cognitiveprocesses is indicated.

8 ix

Introduction

As a concerned year-in, year-out con-sumer of educational research, you are en-titled to ask what we have learned aboutmaximizing what children learn. As a con-cerned year-in, year-out conductor ofeducational research, I am compelled toanswer. If what we have learned aboutmaximizing what children learn were "little"or "nothing," I would not have chosen toinitiate this venture. Happily, however, inrecent years queries into the realm of chil-dren's learning have proven extremely fruit-ful, in terms of both currently accessibleknowledge and promises of things to come..So let us continue.

The present paper does not pretend toencompass a complete summarization of thechildren's learning literature (for earlierreviews, see Goulet, 1968, and Keppel, 1964).Nor does it pretend to bear directly on educa-tional policy or curricular decision making(lee Glaser, 1972, and Rohwer, 1972).Re,ther, in this paper I shall provide an up-to-date account of our research program at

the Wisconsin Research and DevelopmentCenter for Cognitive Learning. Where appro-priate, I will incorporate our findings into thelarger body of children's learning literature,as was done in an initial report of our re-search (Levin, 1972b).

In the evolution of our research program,we--as others before us--have come to realizethat the task of maximizing what children learncan generally be accomplished through the useof techniques which concretize what is to belearned. That is, we have opted for methodsand materials which capitalize"on children'sprevious encounters with their environmentsand which provide a closer approximation tothose environments than can be providedthrough more abstract representations. Spe-cifically, our efforts toward concretizationhave involved pictures, both as learning ma-terials and as the principal ingredient inlearner-initiated cognitive strategies. Whatfollows, then, is essentially a case for pic-tures in children's learning, with occasionalcaveats where necessary.

1

IIPictures As Learning Materials

One of the more ubiquitous findings inthe literature is that pictorially representedobjects are more memorable than their asso-ciated verbal labels. This, of course, comesas no surprise to the visual shapers of ourculture and commercial society who are wellaware that the form, the features, lnd eventhe phenomenal fool who "ate the wholething" all contribute in their own pictorialway to the effectiveness of a communication.Yet despite the obvious contributions of theseresidents of Madison Avenue (and more re-cently, of our children's friends on SesameStreet), the puzzle of pictures continues tofascinate educational researchers at variouslevels of scientific investigation. Let usstart by considering some empirical studiesin which pictures and words have been com-pared.

Evidence for Picture Superiority

In 1967, Roger Shepard discovered thatwhile adults have an unusually large capacityfor storing verbal material, it is substantiallysmaller than their capacity for storing pic-tures. Shepard examined performance onthree recognition memory tasks, each con-sisting of approximately 600 stimuli, andfound that while previously exposed wordsand sentences were recognized with about89% accuracy, pictures were recognized withabout 97% accuracy; the pictorial level de-creased to the verbal level only after a one-week interval between presentation and test-ing. An even more impressive demonstrationof pictorial recognition memory capacity hasbeen provided by Standing, Conezio, andHaber (1970), who reported over 90% accu-racy for more than 2,500 pictures, even withdelays of three days between presentationand testing.

Although such evidence of seeming"unlimited memory" for pictures may capture

one's imagination, it does not address itselfto the issue of concern here, namely, memorydifferences between pictures and words.Across-material comparisons in the Shepard(1967) data may be made only in an offhandway since (1) the data were obtained fromdifferent subject populations in sequentiallyconducted experiments and (2) he word andpicture lists were not identical either in lengthor in content (the former consisting of Englishnouns and adjectives and the latter consistingof pictures taken primarily from magazine ad-vertisements). However, subsequent researchwith adults (e.g., Paivio & Csapo, 1969) inwhich recognition memory. for line drawings offamiliar objects (e.g., a clock, atouse, apiano) was better than that for the verballabels of the same objects permits a moredirect inference.

Precisely the same conclusions arereached when the children's learning literatureis examined. That is to say, recognitionmemory for pictures is extremely high, evenfor preschoolers (e.g., Brown & Scott, 1971),and direct picture-word comparisons revealthe superiority of pictures (e.g., Corsini,Jacobus, & Leonard, 1969).

Our own research conducted primarilywith elementary school children has uncoveredthe same consistent trend, both in learningtasks that do not require the subject to repro-duce the previously studied items (i.e., tasksthat demand item recognition) and in thosethat do (i.e., tasks that demand item recall).This research includes a reading comprehen-sion study in which subjects had to acquirethe gist of a fictitious passage either by read-ing a regular printed version or by looking ata specially constructed cartoon-like sequenceof pictures that told the story (Levin, 1973).The finding that slightly more questions basedon the passage were correctly answered by"picture" subjects than by "print" subjects isinteresting and of potential educational signi-ficance. Duriiig our discussion of individual

0

3

differences in the use of learning materialsand learning strategies which follows we willreturn to this point.

The Role of Individual Differences:An Empirically Derived Representation

Although the picture-over-word effecthas been found to generalize across a widevariety of populations as represented by suchsubject variables as age, sex, IQ, and SES/race (cf. Rohwer & Levin, 1971), we havenoted in a number of studies that the magni-tude of the effect varies reliably with certainof these variables. For example, we haveobserved several instances in which the rela-tive superiority of pictures to words has in-creased with age from childhood through ado-lescence (see also Ghatala & Levin, 1973,and Reese, 1970). As an illustration, in oneof our studies (Levin, Davidson, Wolff, &Citron, 1973) across two different methods ofassessing associative learning (recognitionand recall), we found that younger children(second graders) correctly associated anaverage of 14% of the pictorial items and 12%of the verbal items--a difference of 2%. Forolder children (fifth graders), however, thisdifference increased to 15%; these childrencorrectly associated 28% of the pictures and13% of the words. While the age populationsand experimental tasks sampled across stud-ies are not sufficiently comparable to providea strict confirmation of an age by picture-word interaction (specifically, an increasingdifference between picture and word learningwith increasing age), the trends abstractedfrom a composite of several studies are con-sistent with this suggestion.

Picture-word differences appear to in-teract with other subject characteristics aswell, much in the manner that I have de-scribed elsewhere (Levin, 1972a) and whichwill be expanded upon here. Figure 1 depictsan admittedly oversimplified conceptualizationof individual differences in learning as a func-tion of variations in methods or materials.

As will be seen, both the "Population"and "Method/Materials" labels in Figure 1can be adapted to characterize a number oflearning outcomes that have been noted inour.own research and in that of others. Fornow, in order to concretize the intent ofFigure 1, let us assume that the Populationvariable represents a gross breakdown of theacademic accomplishments of students in aparticular classroom; that is, students fromPopulation I are "good" learners, studentsfrom Population II are "average" learners,and students from Population III are "poor"

4

learners, where individual students are looselyallocated to one of the three populations, letus say, on the basis of a teacher's long-termassessment of in-class achievement. Let usassume further that we wish to determine eachstudent's mastery of a particular instructionallesson.

This is where the Method/Materialsvariable comes in. Suppose that the studentswere taught the lesson in a mildly disorganized,uninspiring fashion (due either to a fault of theteacher or the textbook or to some combina-tion). If we assess mastery following such"impoverished" instruction, which might beMethod A in Figure 1, we will note that onlythe good learners (from Population I) succeed- -in a sense, they will have learned in spite oftho instruction. In contrast, the average andpoor learners (from Populations II and III) willfail; at least some of them will not havelearned because 91 the instruction.

On the other hand, suppose that thelesson were presented to the students in ahighly organized, effective manner (MethodB). The good learners will still succeed,probably to an even greater degree than before.However, the clarity of the instruction willnow enable those students (from Population II)who would otherwise fail (under Method A) alsoto succeed. At the same time, some of thenonlearners under Method A (those from Popu-lation III) will continue to fail even with opti-mal instruction--they will fail la spite of. theimproved instruction. (This is not to say thatPopulation III students can never be taught thelesson successfully, but rather that the vari-able considered here, viz., a poor versus agood presentation to a group of students, maynot be sufficient to effect mastery for them.)

What is important in this example isthat for a large number of students the proba-bility of mastery is closely related to thequality of instruction received. With poorinstruction many children who would other-wise succeed will fail. It would, therefore,appear to be a worthwhile endeavor to identifyinstructional techniques which maximize thesuccess ratio for a given group of students.Following my (Levin, 1972a) distinction be-tween external and internal variables whichaffect learning, we will be concerned with iden-tifying student-initiated cognitive saategieswhich enable learning to occur despite a lackof optimal instructional methods and materials.

But first let us look at a few experimentsdealing with picture-word differences in learn-ing from which the representation in Figure 1was itself modeled. In an earlier study con-ducted by Rohwer, Ammon, Suzuki, and Levin(1971) concerned with learning differencesamong certain age, sex, and socioeconomic

BE

ST

CO

PY

AV

AIL

AB

LE

Hig

hP

opul

atio

n I

Pop

ulat

ion

II

A

Pop

ulat

ion

111

S

Met

hod

or M

ater

ials

B

Figu

re 1

. A c

once

ptua

lizat

ion

of in

divi

dual

dif

fere

nces

as

rela

ted

to th

e po

tent

ial e

ffec

tiven

ess

of a

cogn

itive

str

ateg

y.(A

con

nect

ed li

ne g

raph

is u

tiliz

ed h

ere

sim

ply

to e

mph

asiz

e th

e in

-te

ract

ion

of in

tere

st th

at is

dis

cuss

ed th

roug

hout

the

text

.

groups, we (Levin, Rohwer, & Cleary, 1971)nosed that even within these dumcgraphicallydefined groups there appeared to be reliabletndividual differeicces in learning from pic-tures and words. In particular, subjects whowere classified as exhibiting either relativelylarge or relatively small picture-word differ-ences tended to be similarly reclassified on aparallel form of the task administered twodays later. This was found to be true in tenof the twelve samples studied; the effect wasstatistically significant in six of them.

Following up this notion, Levin, Divine-Hawkins, Kerst, and Guttmann (1974) discoveredthat with such a task, the vast majority of sub-jects--in this case, fourth graders--could beclassified into three learner types: subjectswho learn relatively well from both picturesand words (Hi P, Fit W); subjects who learnrelatively poorly from both pictures and words(Lo P, Lo W); and subjects who learn rela-tively well from pictures but relatively poorlyfrom words (Hi P, Lo W). It is interesting topoint out that, in accordance with the researchfindings reported thus far, virtually all sub-jects learned picture items better than worditems. The evidence for this was (1) consis-tently superior picture recall within the Hi 13,Hi W and Lo P, Lo W groups and (2) a verysmall number of subjects classified as Lo P,Hi W subjects (a result which led to oureventual decision that the Lo P, Hi W cate-gory is not a learner-type classification ofpractical importance).

6

To investigate the stability of the threelearner types identified, a parallel form ofthe learning task was administered to the sub-jects on the clay following their initial clasSi-fication. 'these tasks, unlike those of ourearlier study , were group administered. Theresults of the study are summarized in Table1; it may be seen that 30 out of 41 initialclassifications (nearly 75%) were confirmedby the second assessment. While such datado not speak to the long -tIrm stability oflearner types nor to their concomitant charac-teristics , they should encourage future inves-tigations into an area which has been beset byserious Methodological and substantive diffi-culties (cf. Levin, Rohwer, & Cleary, 1971).A preliminary effort in this direction was madein a second experiment by Levin, Divine-Hawkins, Kerst, and Guttmann (1974), whichwill be described in a later section.

For now, it is worth noting that in thetwo experiments reported by Levin, Divine-Hawkins , Kerst, and Guttmann (1974) , 36Hi P, Hi W subjects, 40 Lo P, Lo W subjects,and 22 Hi P, Lo W subjects were identified byinitial classification procedures. On the basisof these data it certainly appears that theability to learn from pictures and the ability tolearn from words are highly correlated in ourtask (since most of the classifications are ofthe Hi 13, Hi W type, or the Lo P, Lo W type).However, it is also apparent that for a good20% of the children, the ability to learn isrelated to the type of materials presented.

TABLE 1CORRESPONDENCE BETWEEN INITIAL AND

SUBSEQUENT LEARNER-TYPE CLASSIFICATIONS

,S=1==s=1:=111=12201=111=111,

e q Hi P, Hi W

Lo P, Lo W

Hi P, Lo W

Hi P, Hi W

9

2

1

Day 1

Lo P, Lo Wa

1

14

2

a Three additional subjects, initially classified as Lo 13, Lo Wwere later classifiek.. as Lo P, Hi W.

.13

Hi P, Lo W

That is, the Hl P, Lo W children functionlike poor learners (Lo P, Lo W children) whenwords are presented but like good learners(Hi P, Hi W children) when pictures are pre-sented,

This situation is represented in Figure 1by the seeming inability of subjects from Pop-ulation II (here, Hi P, Lo W children) to per-form successfully when materials of one kind(words) are presented and their startlinglydifferent performance when materials ofanother kind (pictures) are presented. In theexp.riment just reported, for example, whenword pairs were presented the Hi P, Lo Wsubjects correctly recalled an average of 16%,as compared to 14% by Lo P, Lo W (PopulationIII) subjects and 50% by Hi P, Hi W (Popula-tion I) subjects. On the other hand, whenpicture pairs were presented, the mean per-formance of Hi P, Lo W subjects (61% cor-rect) clearly surpassed that of Lo P, Lo Wsubjects (33% correct) and rivaled that ofHi P, Hi W subjects (66% correct). SinceFigure I was derived from a composite ofseveral experiments, it does not portray thisresult exactly, but it surely approximates it.

Pictures and the Comprehension of Prose

I now wish to focus on two recent find-ings which further substantiate the essentialsof Figure 1 with regard to individual differen-ces in learning from pictures and words. Un-like the experiments just discussed, in thesestudies the criterion task was not one ofsimply associating lists of words or pictures.Rather, subjects were presented prose pas-sages which resembled those encountered intheir school readers and were subsequentlyasked questions about the factual materialcontained in the passages.

In one experiment, Rohwer and Matz(in press) compared the comprehension skillsof middle-class white and lower-class blackfourth graders. Passages were presented tothe subjects aurally by means of a tape re-corder. For one of the experimental conditions-(Aural/Print), subjects were simultaneouslyshown the printed version of the passage, inmuch the same way as in the elementaryschool classroom when students read alongwith the teacher or with another student.1

1 In a different context, some researchfrom our laboratory indicates that this com-monly used instructional format may not be aterribly effective one (Kaplan, 1971; Levin,Horvitz, & Kaplan, 1971).

For the other condition (Aural/Pictures), in-stead of "reading along with Mitch," subjectswere shown a series of line drawings that in-corporated the details of the aural passage.Not surprisingly, under the school-like Aural/Print presentation, middle-class whitesanswered substantially more questions aboutthe passage correctly (an average of 82%) thandid lower-class blacks (an average of 58%).However, in the Aural/Pictures condition,this difference was markedly diminished; inthe spirit of the Populations (I versus II) byMaterials interaction of Figure 1, middle-classwhites averaged 90% correctly and lower-classblacks averaged 81% correctly.

Levin (1973) wanted to see whether thesame type of interaction could be detectedwithin a relatively homogeneous SES/racialpopulation. Subjects consisted of predomi-nantly lower mittile-class white fourth graderswho.were classified as good or poor readerson the basis of standardized reading test per-formance and reading group assignment withinthe school. Passages were presented to sub-jects either as a sequence of printed sentencesor as a sequence of pictorial (cartoon-like)representations; no aural accompaniment wasprovided in either case. Although the inter-action of interest was statistically nonsigni-ficant, performance differences between goodand poor readers were found to be slightly lesswhen the pictorial version of the passage waspresented (averages of 85% correct and 75%correct: respectively--a 10% difference) thanwhen the printed version was presented (81%correct and 67% correct, respectively - -a 14%difference).

Possible reasons that this effect was notlarger are discussed by Levin (1973) and in-clude the speculation that poorer readers maynot have attempted to translate the pictorialmessage into verbal form or may not have beensuccessful In translating them. Unlike theRohwer and Matz (in press) study, where anaural description accompanied the pictures, inthis experiment the pictures by themselveswere sufficient to tell the story. The sugges-tion that a dual verbal/pictorial representationof text may be required was indicated by theperformance of subjects in a third condition ofthe Levin (1973) study. However, these datawill be examined in a later section, where arationale for considering learner-initiatedcognitive strategies is presented.

A Footnote to the Benefitsof Pictures in Learning

In a number of experiments that havefocused on the learning and use of concepts by

7

subjects, pictures have been found to be notonly nonfacilitative in comparison to wordsbut frequently inferior. For example, Run-quist and Hutt (1961) found that conceptsselected to represent a series of stimulusinstances (e.g the concept "soft" for theinstances lad, NE, pillow, and moccasin)were acquired more rapidly when the stimuliwere words than when they were pictures.Just how is this curious reversal of picture-word differences--based on what we havelearned till now -- explained? Runquist andHutt 'suggest (as does Hollenberg, 1970, inanother context discussed later) that the vastnumber of unique perceptual details associ-ated with pictures may impede the formationor retrieval of broader, more abstractly de-fined concepts. Support for the notion thatirrelevant perceptual details may contributeto such "conceptual blindness" was obtainedin the Runquist and Hutt study, where theword-over-picture effect was diminished whenpictures were used which perceptually high-lighted the intended concept (e.g. , for theconcept "soft," the ed was made to appear"soft and billowy").`

Additional support for this view may beextracted from an experiment by Deno (1968),in which subjects were presented an associ-ative-learning task consisting of either famil-iar picture or word stimuli paired with 12 un-familiar Japanese words. In one version ofthe task the stimuli were 3 different instan-ces from each of 4 conceptual categories(buildings, clothing, furniture, andanimals), and in the other they were 12 un-related items. According to verbal-learningtheory, performance in the conceptually re-lated list should suffer relative to that in theunrelated list since the conceptually similaritems should produce intralist interference.The result of concern here is that while thiswas indeed the case when the stimuli werepresented as words, there was little inter-ference when the stimuli were presented as

2 Further evidence comes from Wohl-will's (1968) research utilizing Piagetianclass-inclusion tasks, where mastery isassumed to be synchronous with the emer-gence of concrete operations in the child.Wohlwill has found that verbally presentedclass-inclusion problems are answered moreaccurately than pictorially presented prob-lems but that the difference can be reducedby presenting the pictures in ways that helpto break down their perceptually biasingfeatures.

8

pictures. This result is consistent with thenotion that the conceptual categories werenot elicited as readily in the picture conditionwhere the perceptual details of the picturesdominated the subjects' attention (althoughother interpretations are possible). We haverecently obtained a similar result in a studydealing with item recognition rather thanitem recall (Levin, Bourne, Yaroush, Ghatala,DeRose, & Hanson, in press).

It is worth noting that the "detraction"explanation also fits a large number of studiesin which it has been found that children's ini-tial word decoding is retarded by the inclusionof pictures (cf. Samuels, 1970). That is, theaddition of a picture to represent a to-be-decoded word is thought to pull subjects' at-tention toward the picture and away from thecritical letter configurations. This notion hasreceived some support by Lippman and Shanahan(1973), who observed that when pictures were .

physically integrated with the letters them-selves (e.g. , striped letters resembling candycanes to represent the word candy), word learn-ing was facilitated.

To conclude this section, let me reiter-ate that in the vast number of learning studiesthat I have come across in the psychologicalliterature, there I^ precious :.ttle evidence torefute the claim that p1 ;tares are learned betterthan words.3 One m.able exception to thisstatement consists of a class of experiments,some of which were noted above, dealing withthe acquisition and utilization of concepts and/or semantic categories (not to be confused,however, with situations in which picturesmay be used in coniunction with (or as "ad-juncts" to) verbal descriptions--a techniquewhich has frequently been found to facilitatethe understanding of abstract concepts as well).Such a counterexample is an intriguing oneand should do much to further our understand-ing of the basic processes or mechanismsunderlying picture-word differences in avariety of learning tasks. Efforts in thisdirection have been initiated by a host of in-vestigators, and although a recounting ofcurrent theoretical positions is beyond thescope of this paper, they will be dealt within a forthcoming paper by Elizabeth Chatala,Larry Wilder, and myself.

A f)

3 However, this statement may be re-stricted to normal rates of stimulus presenta-tion, since Paivio and Csapo (1969, 1971)have found that pictures are serially recalledmore poorly than words when they are presen-ted too rapidly for verbal labeling to occur.

IIIVisual Imagery As An Organizational Strategy

While one way to maximize what chil-dren learn is to use pictures, another is toembed the materials in a meaningful contextor organization in which learning, can occur.For example, it will take most children sev-eral trials to recall a 14-item list of wordssuch as rat., jog, street, bowl, mak, chair,

in their correct serial order. However,this becomes virtually a one-trial prepositionwhen the items are presented as: The grey=jumped over the 12 land crossed thestreet to find the bowl of cold raja under thechair . (Levin & Rohwer, 1968). AsRohwer (1967) has aptly noted, in such situ-ations performance is greatly enhanced byLido to the to-be-remembered stimuli, afinding which prompts the seemingly para-doxical inference that the more there is tolearn, the better the learning. Thul, in thisexample the number of words to be processedin the first case (14) is substantially lessthan when the sentence context is added (52);yet the learning in the first case is substan-tially worse. Of course, this finding posesno paradox at all for students of organiza-tional processes in memory who view thedifficulty of a task in terms of associationaland contextual factors rather than in terms ofindividual items gar m.

One can also apply the notion of or-ganization to the data considered in theprevious section. Recall that a consistentfinding is that pictures are more easily asso-ciated than words. However, the magnitudeof this effect is typically not nearly as greatas that produced when the experimenter pro-vides an organizational context for the paireditems, Thus, the sentence context, "The catjumped over the jag., " has a marked facilita-tive effect on the learning of the pair, =-Lost, as does a pictorial context in which acat is depicted jumping over a log. Examplesof the potency of such context effects inassociative learning for both younger andolder subjects, as well as for subjects from

different social class and ability strata, maybe found in Rohwer (1967).

While others have embraced strategiesin which the experimenter provides a sentenceor pictorial context to facilitate learning, ourpreference has been to provide learners with astrategy for generating their own organizations.This distinction between imposing an organi-zation on learners and inducing an organiza-tion in learners has been discussed elsewhere(Levin, 1972b), as has our rationale for opt-ing for the latter approach (Levin, 1972a),

Essentially, our argument is that most"real world" learning situations are less thanoptimally structured in that they require activemental transformations and elaborations on thepart of the learner in order for him to processthe to-be-learned material effectively. Ac-cordingly, efficient learning will occur onlyif .the learner is equipped with organizationalstrategies that will free him from the typicallyimpoverished quality of incoming stimuli. Byhypothesizing about the nature of such strat-egies and then by testing these hypotheses insamples of presumed effective and ineffectivelearners (e.g. , older versus younger subjects)we should be able both to document theirpsychological reality and to determine whetheror not it is experimentally plusible to in-struct others in their usage.'

We have begun to launch an attack inthis direction. For instance, in an associative-learning task, rather than "spoon organizing"the pairs for subjects by providing them with

4 This is not to say that an equally con-vincing case cannot be made for the "imposed"approach, where variations in stimulus mater-ials (e.g. , semantic and syntactic properties)may be precisely defined and carefully con-trolled and their effect on subsequeht learningnoted.

i69

an effective sentence or pictorial context,we have instructed subjects to generateappropriate sentences--and especially,dynamic visual images--for themselves,Thus, when the items cat-kg appear, sub-jects are instructed to think up a sentenceor to make up a vivid picture in their mindwhich relates the pair members to one anotherin some meaningful way. The results of suchinstructions have been astounding in thatlearning gains have been observed whichrival--and in some cases surpass-those ob-tained from supposedly optimally defined(i.e., experimenter-imposed) organizations.In the following sections, then, I will discusscertain of our investigations into visualimagery as an organizational strategy.

The Development of aVisual Imagery Strategy

One of the more intriguing findingsthat we have uncovered is that the ability tobenefit from an experimenter-induceci organi-zational strategy is, in large part, a functionof the cognitive maturity of the learner. Thus,it has been noted that while children youngerthan six or seven years of age have little dif-ficulty in using an experimenter-imposedstrategy effectively, they are typically un-successful when requested to generate sucha strategy on their own. And the develop-mental disparity between "using" and "gen-

erating" appears greater for visual imagerythan for sentence production, as indicated bysome recent evidence refuting Jensen andRohwer's (1965) claim that young children areunable to generate appropriate mediating sen-tences in an associative-learning task(McCabe, 1973; McCabe, Levin, & Wolff,1974).

But let us concentrate here exclusivelyon the development of visual imagery. Intheir most recent work on the topic, Plagetand Inhelder (1971) distinguish between twogeneral classes of visual imagery in the child.The first, rearodugtive imagery, is ontoge-netically more primitive in that it emerges inthe child's preoperational years, Being staticin nature, it is typified by internalized copiesor imitations of external events. The second,anticipatory imagery, does not emerge until alater stage of development--in particular,the onset of operations at about age seven oreight. This type of imagery is dynamic innature in that with it the child is capable ofemploying unique mental operations (e.g. ,spatial rotations and transformations) tostatic or physically removed events,

10

17

We may distinguish between reproduc-tive and anticipatory imagery in terms of ourassociative-learning paradigm, The preopera-tional child is capable of imagining only aperceptual copy of the to-be-associated items,whereas the operational child is capable ofmanipulating these images in novel ways(e.g. , conjuring up an imaginal interactionbetween the paired items). If a thematic in-teraction is the basis for optimal associative-learning performance, then it would not besurprising to find that an operational childwho is successfully employing an imagerystrategy learns more quickly than either apreoperational child who is unsuccessfullyemploying one or an operational child who isnot employing one. This prediction wastested and further elaborated in a series ofstudies conducted by Peter Wolff, myself,and others.

In one experiment we found that bothsecond and fifth graders benefited from aninduced visual imagery strategy (i.e., "makeup a picture in your head of the two things in-each pair doing something together") in anassociative-learning task (Levin, Davidson,Wolff, & Citron, 1973). On the other hand,Montague (1970) had previously reported thather sample of first grade inner-city childrendid not benefit from Such a strategy. Thus,indirect support for the proposed anticipatoryimagery deficit in younger children might beinferred.

However, in order to make a direct testof this proposition (i.e., by utilizing the sametask and materials with subjects from a geo-graphically and demographically similar loca-tion), Wolff and Levin (1972) compared theperformance of third graders and kindergartners,assuming that the former possessed well-developed anticipatory imagery while the lat-ter did not. The outcome was precisely asanticipated: in the third-grade sample, in-stuctions to employ an imagery strategy facili-tated performance relative to a poninagerycontrol condition (77% correct versus 32%correct, respectively); however, in the kinder-garten sample a nonsignificant difference.(41% versus 30%) was observed, Apart fromthese performance data, the children's atti-tudes during imagery generation (i.e., holdingtheir heads motionless, keeping their eyesclosed or gazing upward, and obviously con-centrating) and their subsequent verbal reports,lent credence to the conclusion that thirdgraders could indeed comply with the imageryrequest while kindergartners could not,

But this is not the whole story. Onefundamental assumption of Piagetian theory isthat operational thought grows out of the early

sensorimotor activity of the child, In thepresent context, visual imagery is regardedas internalized motor activity in that itoriginates in the child's early play and imi-tation, which later become internalized. Ifthe child's overt motor activity does in factprovide the basis from which covert imageryevolves, then it is not unreasonable toAssume that children approaching the opera-tional stage (i,e, our kindergartners) couldproduce an external motoric representationwhich in turn might meliate the formation ofan internalized imaginal representation.

In the Wolff and Levin (1972) studyjust reported, the stimulus materials con-sisted of children's toys , While in the firstpart of the study the child was instructed togenerate a visual interaction between thepaired toys internally , in the third experi-mental condition he was permitted to gener-ate the visual tnteraction externally by actu-ally manipulating the toys. Clearly this wasno problem even for the kindergartners, andit was clear that the interactions they pro-duced had memorial consequences as evi-denced by the increased learning in thiscondition (64% correct), surpassing by farthe kindergarten control condition (30%) andapproximating the imagery condition perform-ance of third graders (77%).

While it is tempting to interpret theseresults in terms of the overt motor activityeliciting covert visual imagery, a morestraightforward interpretation is that subjectsin the manipulation condition actually wereprovided with an imooseci interaction in thatnot only were they generating an interaction(a process), but they were able to view theresult of this activity (a product) as well.Since it is well known that imposed visualinteractions facilitate associative learning(as also found in the Wolff-Levin study whenthe experimenter actually created the inter-action), it could be argued that theseinteractionsand not the motor activitypreceding them--were the proximate causesfor performance differences between the man-ipulation and the imagery conditions in thekindergarten sample.

To resolve this problem, Wolff andLevin (1972) conducted a second experimentin which some children were permitted tomanipulate the object pairs but were not per-mitted to see the resultant interactions sincethe manipulations took place behind a curtainWhich shielded the toys from the children'sview. The learning of the manipulation-instructed subjects was 58% better than thatof the imagery-instructed subjects who werepermitted to hold onto the toys through thecurtain but not allowed to manipulate them.

This finding compares favorAbly with the 55%facilitation figure in the first experiment wherevisual inspection of the interaction wasallowed, As further evidence of the none ssen-tiality of visual feedback to the motor effect,in a subsequent study Wolff, Levin, andLongobardi (1972) indepencientl} varied thevisual and tactual components of the motoractivity, It was found that relative to appro-priate control conditions, the quality of thesubject-produced interactions and subsequentlearning were not related to the presence orabsence of visual feedback.

These experiments suggest that anti-cipatory visual imagery constitutes a usefulorganizational strategy in associative-learningtasks. While older children can benefit di-rectly from covert imagery instructions,younger children appear unable to do so. Forthem, however, it is possible to induce covertvisual imagery through overt motor activity;when this is done, the learning of youngerchildren approximates that of older childrengiven simple imagery instructions.5 (Notethat these induced imagery findings can beeasily incorporated into the representation inFigure 1 which will be done in a latersection.)

Furthermore, a study by Varley, Levin,Severson, and Wolff (1974) indicates that eventhough young children do not appear to benefitfrom covert imagery instructions or Ag (i.e.,in the absence of concurrent motor involve-ment), they can be trained to do so. Essen-tially, it was found that after prolonged prac-tice involving motor manipulations kinder-canners successfully accommodated imageryinstructions (with no concurrent motor activity)on a subsequent task, in comparison to chil-dren who were given covert imagery practicewithout accompanying motor activity. Thus,the possibility of eliciting anticipatory imageryin young children via motor training is promis-ing.6 Clearly, the long-term benefits of such

5 The notions of "mediational," "pro-duction," and "control" deficiencies (Flavell,1970; Kendleri 1972) are applicable and dif-ferentially evident throughout this research,although they will not be pursued here.

6 But just as there appears to be ,a lowerage limit--of about seven years--in the abil-ity to benefit from a simple imagery instruc-tion, we have recently discovered that motorinducement of visual imagery may also have alower age limit - -of about five years (Levin,McCabe, & Bender, in press).

11

efforts, as well as their breadth of transfer,deserve careful study.

Some Further Evidence for theAge-Imagery Relationship

In the preceding section we propo.f.L.dthat the ability to benefit from simple %flatletimagery instructions in associative leavingdevelops in children beyond the age of Lkor seven. In this section, however, we ex-amine evidence that suggests that the abilitydoes not become fully realized until a laterstage of developmentsomething once againinterpreted in terms of Piagetian theory.

Let us begin with the empirical obser-vation that while imagery instructions havebeen found to facilitate the learning ofadults (cf. Bower, 1972), this has not beentrue as consistently for children--even forthose well into the concrete-operationalstage. Although it is difficult to compareresults across studies (where subject andtask characteristics vary considerably),Levin (1972b) has observed that in experi-ments where imagery instructions have failedto improve children's associative learning,the stimuli have typically consisted of verbalmaterials (i.e. , printed or aural noun pairs).To pursue this observation further, we con-duc ad two experiments with 11- and 12-year-olds ind found that although the childrenbenefited from an imagery strategy when thestimuli consisted of line drawings, they didnot when the stimuli consisted of printedwords (Levin & Kaplan, 1972). And in afollow-up to this study, Eoff and Rohwer (1972)detected a developmental shift within theelementary school grades: initially subjectscould not employ an imagery strategy eitherwith pictures or with words; at a later age,they could employ the strategy with picturesbut not with words; finally, they could employit with both.

On the basis of these results, it maybe tentatively concluded that the effective-ness of an imagery strategy (especially forchildren at the elementary school age) de-pends on the concreteness of the materialsto be organized. That is, just as Paivio(1971) has reported that imagery generationproceeds more slowly with abstract nouns(e.g. , truth, democracy) in adults, it isreasonable to suppose that children findimagery generation from words a relativelydifficult task. Levin (1972b) has presenteda flow chart depicting the hypothesized stepsrequired when transforming either picture orword pairs into an interactive image; withwords (less concrete) the number of trans-

12

formations involved in the encoding and de-coding phases is seen to be greater than withpictures (more concrete).

Thus, it may be presumed that eventhough children benefit from an organizationalimagery strategy under certain conditions(e.g., when objects or pictures are the stim-ulus materials), they may not under others(e.g., when the stimuli are words). How-ever, as was stated previously vis -a -vis thediffering task and subject characteristics ofthe studies investigated, this generalizationshould be interpreted loosely at present, foralthough Horvitz and Levin (1972) found thatthird graders could not effectively utilizeimagery instructions with verbal materials,Levin, Davidson, Wolff, and Citron (1973)found that second graders could when thesematerials were embedded in a single list con-taining both word and picture pairs. It isclear that a more systematic investigation ofthe "stimulus concreteness" phenomenonneeds to be rxinducted and its limiting para-meters specified.

The same type of systematic documenta-tion needs to be assembled for a finding thatwe have repeatedly noted in our research,exemplified here by the data of Kerst andLevin (19 73). In that experiment, fourth andfifth graders were instructed to generate visualimagery to facilitate their learning of a list ofpaired pictures. As would be anticipated onthe basis of the research reviewed so far,the performance of the induced imagery sub-jects was by far superior to that of controlsubjects and, in fact, equalled that of sub-jects who were shown experimenter-imposedpictorial interactions. But what is of parti-cular interest here is that the distribution ofthe induced imagery group was considerablymore variable than the distributions of theother two groups.

On the basis of the Kerst and Levin(1973) data it was concluded that (1) somebut not all children benefit from a self-generated imagery strategy (as evidenced bythe greater variability of the induced imagerydistribution), but (2) for those who do, per-formance surpasses that of subjects givenexperimerter-generated organizations (asevidenced by the slightly greater number ofextremely high scores in the induced imagerydistribution than in the imposed imagerydistribution) . We have interpreted this latterresult as beim', consistent with the plethoraof American, Genevan, and Soviet data demon-strating importance of the organism's activerole in the learning process. That is, we be-lieve that optimal learning occurs when thesubject interacts with his environment in ameaningful way, and clearly our own research

19

with induced organizational strategies sup---rorts this position. Moreover, in accordance

with "active participation" theorists, thereare associative-learning data to suggest thatgiven a well-organized event, the memorialconsequences are superior when the eventhas been constructed by the subject himselfthan when it has merely been presented tohim (e.g., Bobrow & Bower, 1969; Wolff,Levin, & Longobardi, 1974).

Let us now attempt to account for. thefinding that imagery instructions seem toproduce a variable effect with children, asreflected by performance both within a taskas in the Kerst and Levin (1973) study andfrom one type of material to another (e.g. ,from pictures to words) as in the Levin andKaplan (1972) study. To do this, we willcapitalize on the Piagetian belief that theconcrete-operational thought of elementaryschool children evolves to the formal-operational thought of adolescents (at about11 to 13 years of age). It is only at this laterstage of development that the child becomescompletely free from the stimulus-boundednessof his earlier concrete-operational thought inthe sense of being able to exploit symbolicrepresentations more fully and to deal withabstractions more proficiently.

The elementary school grades encompasschildren representing a wide range of cogni-tive-developmental levels; from transitionalpreopetational/concrete-operational childrento transitional concrete-operational/formal-operational children. And, based on our ownobservations and those of others, within anyparticular grade level at a given school(especially in the middle elementary grades,grades 3 to 5) virtually this entire develop-mental spectrum is represented. Given thatcertain task demands (e.g., imagery genera-tion) are related to the cognitive-developmen-tal maturity of the learner (see, for example,Levin, Labouvie-Vief, & Urberg, in press), itis, therefore, not surprising that increasedperformance variability would result becausesome subjects are able to comply successfullywhile others are not. On the other hand, whenthe task demands are sufficiently obscure (asin an imposed organization condition), perfor-mance tends to be less variable (even whenceiling and floor effects are ruled out). Itshould also be noted here that the same vari-ability argument may be applied to Rohwer's(1972) model where subjects exhibit, duringthe adolescent years, an increasing propensityfor employing organizational strategies spon-taneously (i.e. , as they would have to do tosucceed in a control condition).

Finally, the Kerst and Levin (1973) datahave implications for the "active participation"

hypothesis mentioned earlier. That inducedimagery led to performance which was compar-able in mean level to, yet considerably morevariable than, that of performance in the im-posed imagery condition is clearly compatiblewith the hypothesis, as long as individual dif-ferences are considered. That is to say, ifthe poor performance of certain induced imag -ery subjects stems mainly from their being at astage of cognitive development which pre-cludes their benefiting from the strategy, itwould be expected that had the task been ad-ministered a year or two later, their perform-ance would have increased dramatically,thereby creating a mean increase and variancereduction relative to the performance of im-posed imagery subjects. Unfortunately, sucha "wait till next year" hypothesis has yet toreceive empirical verification, as does thesimpler cross-sectional hypothesis that animagery strategy will produce increasinglyless variable performance from the elementaryschool years into adolescence.

Visual Imagery and theComprehension of Prose

Just as it has been demonstrated thatexperimenter-imposed pictures can serve asadjunct aids to facilitate prose learning (i.e. ,Rohwer & Matz, in press), we have found thatexperimenter-induced visual imagery is simi-larly facilitative. Thus, the effects of a visualimagery strategy that were discussed in thecontext of associative-learning tasks seem togeneralize to more complex forms of information-processing as well. For example, our imagery/comprehension research has been conductedwith children in the upper elementary gradeswhere, as noted in the previous section, visualimagery has been found to be a process acces-sible to many children. For children at thisstage of development, simple examples andinstructions to "think up pictures in your headwhile you are reading (or listening to) thestory about what the people are doing, whatthey look like. . ." seem to improve per-formance on a subsequent test of factual re-call. And, in accordance with the motor-imagery relationship suggested by the Wolff-Levin studies, we (in collaboration with AlanLesgold at the University of Pittsburgh'sLearning Research and Development Center)are in the process of exploiting this relation-ship with regard to the improvement of compre-hension skills in young (preoperational) chil-dren.

Similarly, some of the individual differ-ence variables (apart from age) which wereidentified earlier have proven to be worth

2013

studying in d comprehension context. Recallthe experiment in which we classified chil-dren on the basis of how well they learnedfrom pictures (Hi P versus Lo P). Given thevalidity of these classifications, it is notunreasonable to assume that the Lo P sub-ject, like the child who has not yet acquiredanticipatory imagery, probably experiencessome difficulty in pictorially representingstimulus materials and/or mentally manipu-lating them to enhance learning. Whatevercauses Lo P subjects to have difficulty inlearning from externally presented picturesshould also operate to reduce the effective-ness of an internalized pictorial strategy.

In a second experiment of the Levin,Divine-Hawkins, Kerst, and Guttmann (1974)study, we presented Hi P and Lo P subjects aprose passage to be read either in the presenceor in the absence of visual imagery instruc-tions, As may be seen in Figure 2, while theperformance of the Hi P subjects (includingthat of both Hi P, Hi W and Hi P, Lo Wsubjects) increased under imagery instruc-tions, the performance of Lo P subjects didnot, and in fact, decreased. (Perhaps theaddition of an unfamiliar strategy complicateda task on which Lo P subjects have learned tosucceed through the use of other cognitiveprocesses.) At any rate, even though therewas an overall increase in performance whenvisual imagery was induced, the importantpoint here is that not all subjects appearedto benefit from such a strategy.

Precisely the same conclusion isreached when we examine in more detailanother reading study reported earlier (Levin,1973). It will be remembered that differen-ces between good readers and poor readerstended to be smaller when children viewed asequence of pictures than when they read theprinted text. Obviously for children whosecomprehension difficulties may be traced tovocabulary or word-decoding problems ,.adifferent representation of text (in this case,via pictures) should be helpful. But asWieuer and Cromer (1967) have argued, manychildren with adequate vocabulary and decod-ing skills may also be labeled "poor readers"if they have not learned to combine wordsand phrases effectively to derive meaningfrom them. These authors have referred topoor readers of the first type (i.e. , thosewith vocabulary and/or decoding problems)as Deficit Poor readers and to those of thesecond type (those with organizational prob-lems) as Difference Poor readers. Undernormal reading conditions, Difference Poorreaders are assumed to behave differentlyfrom good readers in that they do not spon-taneously summon up effective organi-

14

zational stragegies to help them comprehendwhat they are reading.

Accordingly, we speculated that inducingDifference Poor readers to employ an imageryorganizational strategy while reading wouldimprove their comprehension of the passage.On the other hand, inducing such a strategyin Deficit Poor readers should not prove help-ful since they are lacking the vocabulary/decoding skills which are prerequisite to re-ceiving assistance through organization.Good and poor readers were identified throughstandardized reading achievement scores,teachers' ratings, and reading-track place-ment within the school. Within the poorreader category, Difference and Deficit sub-jects were identified in terms of their respec-tive standardized reading Vocabulary andComprehension subtest scores. That is, theComprehension scores of both types of poorreaders were low, but the Vocabulary scoresof Deficit Poor readers were substantiallylower than those of Difference Poor readers.

As in the Levin, Divine-Hawkins, Kerstand Guttmann (1974) study, half the subjectsreceived imagery instructions before readinga passage and half read it under normal condi-tions. Although imagery instructions werefacilitative overall (84% correct versus 72%correct), the predicted interaction emerged.When the Difference Poor readers were givenimagery instructions, they correctly answeredmore questions than their nonimagery counter-parts (a difference of 26%); Deficit Poorreaders given imagery instructions exhibitedno improvement (actually a difference of 2%in favor of the nonimagery group). And, inaccordance with the comparative performancesof Population I and Population II subjects inFigure 1, while under normal reading condi-tions the good readers by far surpassed theDifference Poor readers (81% versus 60% cor-rect), when the Difference Poor readers wereprovided with an organizational strategy theirperformance (86% correct) approximated (inthis case, slightly exceeded) that of goodreaders reading under normal conditions.

The results of these investigationsaccord well with predictions, in that they(1) document the possibility of improvingcomprehension through the implementation ofa self-generated organizational strategy and(2) suggest that the potential benefits fromsuch a strategy are closely tied to the pre-requisite skills of the user. With regard tothe second point, this is not to say that weshould give up on children who do not seemto profit from simple imagery instructions andregard them as doomed to failure. Quite tothe contrary, it is of especial importance to

100

90 80 70 60 50

BE

ST

CO

PY

AV

AIL

AB

LE

Hi P

Lo P

rI

Con

trol

Inst

ruct

ions

Imag

ery

Figu

re 2

. Rea

ding

com

preh

ensi

on o

f tw

o ty

pes

of le

arne

rs u

nder

dif

fere

ntin

stru

ctio

nal c

ondi

tions

.(A

con

nect

ed li

ne g

raph

is u

tiliz

ed h

ere

sim

ply

to e

mph

asiz

e th

e in

tera

ctio

n of

inte

rest

that

is d

iscu

ssed

thro

ugho

ut th

e te

xt.)

determine whether Deficit Poor readers willbenefit from imagery instructions if vocabu-lary and decoding demands are eliminatedand whether Hi P, Lo W children will benefitfrom alternative types of organizationalstrategies (e.g., verbal paraphrase) )r fromsubject-generated supplements to an imagerystrategy (e.g., relevant motor activity) inmuch the same way that the learning of "pre-imagery" children has been facilitated bysentence production (Levin, McCabe, &Bender, in press; McCabe et al., 1974)and motor involvement (Varley et al., 1974;Wolff & Levin, 1972).

A Footnote to the Benefits ofVisual Imagery in Learning

Thus far we have presented evidencecompatible with the assertion that a visualimagery strategy facilitates learning andcomprehension. But, as with our earlierdiscussion of pictorial facilitation, we mustask how universal this assertion is. We havealready argued that the success of an imagerystrategy depends on the capabilities of theuser, and we have identified the individual'scognitive-developmental level as a particu-larly relevant subject variable. Similarly,we will now show that the unique features ofa particular task may also moderate thestrategy's effectiveness,

For example, Bower (1970) has demon-strated that imagery p_er.gg does not facili-tate associative learning, To do this, heinstructed subjects to form an image of thepair members one next to the other (i.e. ,

not interacting). He found that performancewas no better than that of subjects given rote-rehearsal instructions and appreciably worsethan that of subjects given interactive imageryinstructions. Thus, it may be safely con-cluded on the basis of these and other datathat if the imagery is devoid of an organiza-tional component, associative learning isnot enhanced. Similar conclusions havebeen reached about strategies which wouldand would not be expected to facilitate othertypes of learning (cf. Levin & Rohwer, 1968;Rowe & Paivio, 1971).

Unfortunately, knowledge of thesenominal task characteristics is not sufficientto permit speculation on whether or not aparticular strategy will be effective. Rather,procedural factors and the nature of the stimu-lus materials provided in the task also needto be considered. With regard to proceduralfactors, we mentioned previously that theusual picture-over-wo effect in seriallearning vanishes when the presentation rate

16

is too rapid for the pictures to be implicitlylabeled, even though they are perceived(see Footnote 3, Chapter II) . The same maybe assumed to be true of the imagery strategyeffect in learning and comprehension taskswhen a fast rate of presentation is employed(cf. Levin & Divine-Hawkins, 1974; Paivio,1971). Now let us show how the effective-ness of an imagery strategy may depend onthe stimulus materials provided.

Earlier it was stated that it may beeasier for children to utilize an interactiveimagery strategy in associative learning whenthe stimulus materials are pictures than whenti.ey are words. This was assumed to be re-lated to the differing transformational com-plexities associated with each. For differentreasons, imaging the referent of individualwords for subsequent recall should be moreeffective than simply pronouncing the word(Paivio & Csapo, 1973), whereas pronouncingthe label of a picture or object should be moreeffective than imaging it (Davies, 1972;Paivio & Csapo, 1973). Aspects of this inter-action have recently been discussed by Levin,Ghatala, De Rose, Wilder and Norton (in press)in the content of a discrimination learningtask where recall is not required.

Such observations clearly limit thepotential benefits accruing to visual imageryas a cognitive strategy. So do observationsthat indicate that the magnitude and directionof imagery effects vary across tasks whichostensibly reqdire different cognitive pro-cesses. A case in point is a study by Hollen-berg (1970). Recall the earlier suggestionthat while experimenter-provided picturesfacilitate certain types of learning, they mayinterfere with others. In particular, they maybe assumed to be a deterrent to concept ac-quisition/utilization if the perceptual charac-teristics of particular exemplars prevent thelearner from retrieving the more abstract con-ceptual information required. Hollenbergreached the same conclusions when comparingthe learning skills of children identified ashigh and low imagers on the basis of selectedtests for imagery ability. Her finding of majorconcern here is that even though high-imagerychildren outperformed low-imagery children onan associative-learning task, they were sub-sequently less able to transfer this knowledgeto a conceptual-learning task involving thesame items.

Considering Hollenberg's (1970) dataalongside the comparable results producedfrom picture-word comparisons discussed pre-viously, suggests that what is true of "pic-tures" when defined as externally provided(imposed) representations may also hold for"pictures" when defined as internally gener-

23

lted (induced) representations, With referenceto our own research, the finding that childrenwho learn well from pictures can benefit froman imagery strategy on a comprehension taskwhile children who do not learn well frompictures cannot (cf. Figure 2) certainlyserves to corroborate this notion. The find-

@10 40.40114

ing also serves as a basis for inferring thatsimilar cognitive processes are evoked whenpictures are perceived and images are gener-ated (Levin & Divine-Hawkins, 1974, andPaivio, 1971; but see Anderson & Bower, 1973,and Pylyshyn, 1973). This paper will be con-cluded with a discussion of that inference.

24

17

IVInferred Cognitive Strategies:

A Proposed Approach and Some Illustrations

Throughout this paper I have distin-guished between imposed and induced cog-nitive strategies; a desired mental process ispresumably evoked directly in the case of theformer (via experimenter-provided materials)and inArectly in the case of the latter (viaexperimenter-suggested strategies). How-ever, when investigating a covert processsuch as visual imagery, one of the naggingquestions that must be addressed is whethersuch procedures evoke covert mental proc-esses and, if so, whether these are in factthe processes desired. Although some in-vestigators have taken advantage of subjectreports, typically answers to the first part ofthe question are inferred on the basis of per-formance differences between subjects pre-sented with a strategy and those not presentedwith one (e.g., imagery instructed versuscontrol subjects); and answers to the secondpart are inferred on the basis of performancedifferences between subjects presented withone strategy and those presented with anotheror on the basis of performance differencesobtained with different tasks or different setsof materials.

It should be noted that such indirectmeasures of a cognitive process are abhor-rent to some who believe that only throughdirect, visible methods (like "peeking intothe brain" by means of electrode implanta-tions and associated EEG readouts) are infer-ences about cognitive processes warranted.However, in the absence of refinements ofsuch techniques (where "noise" and vari-ability are the order of the day), others con-sider indirect measures of a cognitive processto be acceptable, if not preferable, at present.It should also be mentioned that questionsdealing with the psychological reality of acognitive process like visual imagery encom-pass far-reaching theoretical issues. Becauseof the essentially applied focus of this parr,only scant attention will be paid to them here.

However, the reader is referred elsewhere fora discussion of some of these issues (cf.Bower, 1972; Neisser, 1967; Paivio, 1971;Pylyshyn, 1973; Sheehan, 1972).

To illustrate the matter of inferring thata cognitive process such as visual imagery isoperating, I have presented a prototypic para-digm and idealized data set (Figure 3). InFigure 3, two "tasks" are represented; in thisdiscussion, "task" refers loosely to proceduralor stimulus-material variations within a task,as well as to the literal meaning of nominallydifferent tasks. Various imposed or inducedstrategies are labeled A, B, and 0 in thefigure; A and B represent experimental strate-gies and 0 denotes a control strategy.7 Withreference to our question about the existenceand nature of covert mental processes, differ-ences between experimental (A or B) and con-trol (0) performance may be taken as evidencethat subjects presented with a strategy havedone something differently from subjects notpresented with a strategy, while performancedifferences between Strategy A and Strategy Bwithin tasks (or, preferably, in interactionwith tasks) may be taken as evidence that thetwo strategies have evoked different cognitiveprocesses.

A sentence-learning study recently com-pleted by Joseph Guttmann, Ann McCabe, andmyself illustrates a within-task manipulation,while a discrimination-learning study by Levin,Ghats la, Wilder, and Inzer (1973) illustrates adifferential strategy effect from one task to thenext. In the sentence-learning study, sixthgraders were required to remember the content

7 As will be seen shortly, however,this variable can also incorporate strategydifferences as inferred from identifiable dif-ferences among learners.

2519

8

20

Performance

26

of 20 aurally presented sentences under basi-cally three types of rehearsal strategies:Imagery, in which subjects were instructed toimage the content of each sentence as it waspresented; Repetition, in which subjects wereinstructed to repeat each sentence threetimes as it was presented; and Control, inwhich no rehearsal strategy was suggested.The mean prompted recall of subjects in t zerespective conditions is adequately repre-sented by the Task 1 profile in Figure 3.That is, Imager: organization (A) was superi-or to (either uvert or covert) Repetition (B),with Control (0) intermediate to the two.This pattern suggests that subjects in con-ditions A and B were doing something differ-ent from the control subjects (or possiblythat some control subjects were spontane-ously employing Strategy A and othersStrategy B?), but more importantly, it showsthat the two strategies resulted in differentlevels of performance. The result is con-sistent with the data of Bower (1972) andRohwer (1972) which show that filling thestudy interval with an irrelevant or ineffec-tual rote-rehearsal strategy is interfering,relative to both a semantic-organizationalstrategy such as the imagery instruction hereand a no-strategy control where at least some(and especially older) subjects are presum-ably ac opting an effective semantic-organi-zational strategy. What is of additionalinterest here, however, is that the negativeeffect of repetition holds (albeit not asstrongly) for subjects instructed to repeatthe sentences silently Ikthemselves as wellas for those told to repeat them aloud, indi-cating that when subjects wore instructed todo different covert things (imagery versusrepetition), they apparently complied faith-fully with the instructions.

Now let us incorporate Task 2 ofFigure 3 in order to approximate the resultsof the Levin, Ghatala , Wilder, and Inzer(1973) discrimination-learning study. Beforedoing so, however, a word about the discrim-ination-learning paradigm is in order. In it,the subject must learn which item in each ofseveral pairs has been arbitrarily designated"correct" by the experimenter. Rowe andPaivio (1971) have demonstrated that an ef-fective rehearsal strategy in this paradigmis one that provides rehearsal of the "correct"item, thereby increasing its familiarity(Ekstrand, Wallace, & Underwood, 1966).In the Levin, Ghatala , Wilder and Inzer(1973) study, the stimulus materials con-sisted of printed nouns and the subjects wererequested to employ one of three rehearsal

BEST COPY AVAILABLE

strategies during the informative feedbackinterval: Imagery, in which subjects wereinstructed to form an image in their mind ofeach "correct" item's referent; Verbal, inwhich subjects were instructed to pronouncealoud each "correct" item; and Control, inwhich subjects were left to their own devices.Included in the stimulus materials were pairsof homonyms, e.g., sun-son (Task 1 inFigure 3), and pairs of synonyms, e.g.,boy-lad (Task 2); it was predicted that ifsubjects were actually adopting the specifiedstrategies, a different pattern of strategy ef-fectiveness would emerge on each task. Inparticular, Imagery (A) should constitute arelatively more effective strategy than Verbal(B) when the subject is discriminating betweenhomonyms (similar-sounding items), whereasthe reverse should be true for synonyms(similar-looking items). This is because thelack of a relevant discriminative cue mightretard increased familiarity with the "correct"item. The results of two experiments essen-tially confirmed the predictions--depicted bythe idealized representation in Figure 3.

Experiments such as these are importantin that when the same strategy can be shownto produce different effects with differentmaterials, inferences are strengthened aboutthe psychological reality of the processesassumed to be involved. Other studies inwhich we have attempted (not always success-fully) to produce differential effects by manip-ulating procedures and materials are those ofDavidson and Levin (1973), Horvitz and Levin(1972), Levin and Divine-Hawkins (1974), andLevin and Horvitz (1971).

An alternative inferential approachwould incorporate subjects who are assumedto employ (or benefit from) different strategiesinto the ABO distinction of Figure 3. Hollen-berg's (1970) study and our study in whichHi P and Lo P subjects responded differentlyto visual imagery instructions are illustrativeof this approach. However, one of the mostclever illustrations is by Paivio and Okovita(1971), who found that subjects deficient invisual imagery (blind subjects) did not benefitfrom materials rich in visual imagery-arousalas sighted subjects do. On the other hand,they did benefit from materials rich in audi-tory imagery-arousal whereas sighted subjectsdid not. This utterly fascinating and fertileresearch area dealing with cognitive processesdeserves continued investigative efforts, asthe wealth of information contained thereinpromises to be of value to both the psychol-ogist and the educator.

2721

Summary

What have we learned about maximizingwhat children learn? Let us attempt to sum-marize the highlights.

1. In a large number of learning taskschildren seem to learn better when the ma-terials are pictures than when the materialsare words. This has been noted not merelyfor simple item-learning tasks but also fortasks which demand comprehension of factsand relationships.

2. There appear to be limitations to thethe effectiveness of pictures in terms of bothlearner and task differences. With regard tothe learner, we have found that certain chil-dren learn appreciably better from picturesthan from words and, for these children, thetype of materials presented during learninglargely determines whether they will resemblegood or poor learners. With regard totask differences, in tasks demanding theformation or utilization of abstract concep-tual categories, pictures tend not to be facili-tative and may even impede performance.

3. Subject-generated organizationalstrategies (visual imagery in particular)greatly facilitate associative learning. Theability to generate effective imaginal organi-zations appears to be closely related to thecognitive-developmental level of the learner.However, even children below the necessarylevel may be induced to generate imagery bymeans of appropriate auxiliary activity ortraining. The ability to generate effectiveverbal organizations appears to be develop-mentally an earlier process than the abilityto generate effective imaginal organizations.This is interesting in that it suggests that(1) even young children can benefit from an

induced organizational strategy if it is anappropriate one and (2) covert verbal andimaginal processes may be differentiated inthe young child.

4. Benefits from an imagery strategyhave also been obtained on tasks requiringcomprehension of a passage, but once againthere are identifiable learner and task charac-teristics that are useful in predicting the ef-fectiveness of the strategy. With regard tothe former, subjects lacking prerequisite word-decoding skills and those who do not learnwell from pictures do not seem to generateeffective imaginal organizations. With regardto the latter, throughout this research we areconstantly reminded that the potential effec-tiveness of a strategy is dependent on thenature of the task and materials. What con-stitutes an appropriate strategy in one contextmay be completely inappropriate in another,and it is up to the experimenter (and, morepractically, the "cognitively mature" learner)to recognize which strategy will be helpful when.

5. We have also learned that a greatdeal remains to be understood about the ex-ternal and internal mechanisms responsiblefor the phenomena that have been observed.For example: Why are pictures generallyeasier to learn than words? What is accom-plished by imaginal or verbal organizationsthat make them effective learning strategies?What other individual difference variables arerelated to the ability to profit from a particularstrategy? While the picture is far from com-plete, one cannot help but imagine the glitter-ing psychological treasures buried just beyondthe next horizon.

28

23

References

Anderson, J. R. , & Bower, G. H. Humanassociative memory. Washington, D. C.:V. H. Winston & Sons, 1973.

Bobrow, S. A., & Bower, G. H. Comprehen-sion and recall of sentences. journal ofExperimental Psychology, 1969, 31,455-461.

Bower, G. H. Imagery as a relationalorganizer in associative learning.Journal of Verbal Learning and Verbal,Behavior, 1970, 9, 529-533.

Bower, G. H. Mental imagery and associa-tive learning. In L. Gregg (Ed.),Coanition in learning and memory.New York: John Wiley & Sons, 1972.

Brown, A. L. & Scott, S. Recognitionmemory for pictures in preschool children.Journal of Experimental Child Psychology,1971, 11, 401-412.

Corsini, D. A., Jacobus, K. A., &Leonard, D. Recognition memory of pre-school children for pictures and words.Psvchonomic Solenge, 1969, 192-193.

Davidson, R. E., & Levin, J. R. Imagery intransfer. Technical Report No. 249.Madison: Wisconsin Research andDevelopment Center for CognitiveLearning, 1973.

Davies, G. M. Quantitative and qualitativeaspects of memory for picture stimuli.Journal of Experimental Child Psychology,1972, ja, 382-393.

Deno, S. L. Effects of words and picturesas stimuli in learning language equiva-lents. Journal of Educational Psycholoav,1968, 59, 202-2n6.

Ekstrand, B. R., Wallace, W. P., &Underwood, B. J. A frequency theory ofverbal-discrimination learning. Psycho-logical Review, 1966, 12, 566-578.

Eoff, J. E., & Rohwer, W. D. , Jr.A developmental study of imagery instruc-tions in noun-pair learning. Paper pre-sented at the annual meeting of the Ameri-can Educational Research Association,Chicago, April 1972.

BEST COPY AVAILABLE

Flavell, J. H. Developmental studies ofmediated memory. In L. Lipsitt & H.Reese (Eds.), Advances in_child_develoD-rnent and behavior. Vol. 5. htnIN York:Academic Press, 1970.

Ghatala, E. S., & Levin, J. R. Developmentaldifferences in frequency judgments of wordsand pictures. Journal of Experimental ChildPsychology, 1973, 495-507.

Glaser, R. Individuals and learning: The newattitudes. gducational Researcher, 1172,I (6), 5-12.

Goulet, L. R. Verbal learning in children:Implications for developmental research.alatigkaical Bulletin, 1968, §9, 359-376.

Hollenberg, C. K. Functions of visual imageryin the learning and concept formation ofchildren. Child Development, 1970, 41,1003-1015.

Horvitz, J. M., & Levin, J. R. Semantic andimaginal factors in learning as related toage. Journal of Exoerimental_Child Psy-gholoay, 1972, LI, 11-20,

Jensen, A. R., & Rohwer, W. D., Jr.Syntactical mediation of serial and paired-associate learning as a function of age.Child Development, 1965, , 601-608.

Kaplan, S. A. Elaboration and reading achieve-ment as factors in children's learning oftext materials. Technical Report No. 180.Madison: Wisconsin Research and Devel-opment Center for Cognitive Learning, 1971.

Kendler, T. S. An ontogeny of mediationaldeficiency. Child Development, 1972, .13. ,

1 - 1 7

Keppel, G. Verbal learning in children.Psychological _Bulletin, 1964, fa, 63-80,

Kerst, S. , & Levin, J. R. A comparison ofexperimenter-provided and subject-generated strategies in children's paired-associate learning. Journal of EducationalPsYcholoav, 1973, 300-303.

Levin, J. R. Comprehending what we read:An outsider looks in. Journal of ReadingBehavior, 1972, i., 18-28. (a)

29

25

BEST COPY AVAILABLE

Levin, J. R, When is a picture worth athousand words? In Issue_s_Inlraageryand learning: Four papers. TheoreticalPaper No. 36. Madison: WisconsinResearch and Development Center forCognitive Learning, 1972. (b)

Levin, J. R. Inducing comprehension inpoor readers: A test of a recent model,Journal of Educational Psychology, 1973,_61, 19-24.

Levin, J. R., Bourne, L. E., Jr., Yarotish,R. A., Gitatala, E. S., De Rose, T. M.,& Hanson, V. Picture-word differencesand conceptual frequency juwgments.Technical Report No. 323. Madison:Wisconsin Research and Development Cen-ter for Cognitive Learning, in press.

Levin, J. R. Davidson, R. E. f WOW, P.& Citron, M. A comparison of inducedimagery and sentence strategies inchildren's paired-associate learning.Journal of_Educational Psychology, 1973,64, 306-309.

Levin, J. R. , & Divine-Hawkins, P.Visual imagery as a prose-learning pro-cess. 'canal of Reading - Behavior, 1974,§, 23-30.

Levin, J. R., Divine-Hawkins, P. ,Kerst, S., & Guttmann, J. Individual dif-ferences in learning from pictures andwords: The development and applicationof an instrument. Journal of EducationalPsychology, 1974, .61, 296-303.

Levin, I. R., Ghatala, E. S., DeRose,T. M. , Wilder, L. & Norton, R. W.A further comparison of imagery andvocalization strategies in children'sdiscrimination learning. journal of.Educational Psychology, in press.

Levin, J. R., Ghatala, E. S. , Wilder, L.,& InzeL , E. Imagery and vocalizationstrategies in children's verbal discrimina-tion learning. Journal of EducationalPsychology, 1973, la, 85-90.

Levin, I. R., & Horvitz, J. M. The meaningof paired associates. laurnaLof Educa-tional Psychology, 1971, §2., 209-214.

Levin, J. R., Horvitz, J. M. , & Kaplan,S. A. Verbal facilitation of paired-associate learning: A limited generaliza-tion? journal of Educational Psychology,1971, j, 439-444.

Levin, J. R., & Kaplan, S. A. Imaginalfacilitation of paired-associate learning:A limited generalization? journal ofgducational Psychology, 1972, .61,429-432.

Levin, J. R., Labouvie Viet, G. , & Urberg,K. A. Cognitive abilities and strategiesin children's paired-associate learning.Technical Report No. 315. Madison:

26

Wisconsin Research and DevelopmentCenter for Cognitive Learning, in press.

Levin, J. R., McCabe, A, E., & Bender,B. G. A note on imagery-inducing motoractivity in young children. Child Devel-opment, in press.

Levin, J. R., & Rohwer, W. D., Jr.Verbal organization and the facilitationof serial learning. journal of EducationalPsycholoaT, 1968, 186-190.

Levin, J. R., Rohwer, W. D., Jr. , &Cleary, T. A. Individual differences inthe learning of verbally and pictoriallypresented paired associates. American;ducational Research Journal, 1971, 6,11-26.

Lippman, M. Z., & Shanahan, M. W.Pictorial facilitation of paired-associatelearning: Implications for vocabularytraining. Journal of Educational PsYchol-fay, 1973, fa , 216-222.

McCabe, A. E. An investigation of the inter-action of motor activity and sentence pro-duction in young children. TechnicalReport No. 207. Madison: WisconsinResearch and Development Center forCognitive Learning, 1973.

McCabe, A. E., Levin, J. R., & Wolff, P.The role of overt activity in children'ssentence production. journal of Experi-mental Child Psvcholoay, 1974, 12,107-114.

Montague, R. B. The effect of mediationalinstructions on associative skills of firstgrade innercity children. Paper presentedat the annual meeting of the American Edu-cational Research Association, Minneapolis,March 1970.

Neisser, U. Cognitive psychology.New York: Appleton-Century-Crofts, 1967.

Paivio, A. imagery and verbal Processes.New York: Holt, Rinehart and Winston,1971.

Paivio, A. , & Csapo, K. Concrete-image andverbal memory codes, Journal of Experi-mental Psychology, 1969, QQ, 279-285.

Paivio, A. , & Csapo, K. Short-term sequen-tial memory for pictures and words.psvchonomic Science., 1971, 2A., 50-51.

Paivio, A. , & Csapo, K. Picture superiorityin free recall: Imagery or dual coding?CognitiYILPsvchologv, 1973,1, 176-206.

Paivio, A. , & Okovita , H. W. Word imagerymodalities and associative learning inblind and sighted riubjects. Journal ofVerbal Learning and Verbal Behavior, 1971,IQ, 506-510.

Piaget, J. , & Inhelder, B, Mental imagery in,the child. New York: Basic Books, 1971.

Pylyshyn, Z. W. What the mind's eye tellsthe mind's brain: A critique of mental

0100 110,-40M11

imagery. Laysjh2iggical Bu Ilgtin, 1973,31, 1-24.

Reese, H. W. (Chm,), Imagery in children'slearning: A symposium. PsychologicalBulletin, 1970, U, 383-421.

Rohwer, W. D. , Jr. Social class differencesin the role of linguistic struct.vres in oaired-associate learning: Elaboration and learningproficiency. Final report on basic researchProject No, 5-0605, U. S. Office of Edu-cation, 1967.

Rohwer, W. D., Jr. Derisive research:A means for answering fundamental ques-tions about instruction. educationalHeaearcher, 1972, .1 (7), 5-1I.

Rohwer, W, D. , Jr., Ammon, M. S.,Suzuki, N., & Levin, J. R. Populationdifferences and learning proficiency.Tournal of Educational Psychology, 1971,12,,, 1-14.

Rohwer, W. ID., Jr., & Levin, J. R.Elaboration preferences and differences inlearning proficiency. In J. Hellmuth (Ed.),Coanitiva Studies. Vol. 2. New York:Brunner/Mazel, 1971.

Rohwer, W. D. , Jr., S. Matz, R. Improvingaural comprehension in white and in blackchildren: Pictures versus print. Journal,o Experimental Child Psychology, in press.

Rowe, E. J., & Paivio, A. Imagery andrepetition instructions in verbal discrim-ination and incidental paired-associatelearning. Tournal of Verbal Learning andVerbal Behavior, 1971, 10, 668-672.

Runquist, W. N., & Hutt, V. H. Verbalconcept learning in high school studentswith pictorial and verbal representationof stimuli. lournal_oflaucational_Psy-chology, 1961, §2, 108-111.

Samuels, S. J. Effects of pictures on learn-ing to read, comprehension and attitudes.Jleview of Educational Research, 1970,AQ., 397-407.

0,0 0,-406.11

BEST COPY AVAILABLE

Sheehan, P. W, (Ed.) , Ttlg funotion_andnature of imagery, New York: AcademicPress, 1972.

Shepard, R. N. Recognition memory for words,sentences, and pictures, Tournal of VerbalLearning and Verbal Bei)ayior,, 1967, 6,156-163.

Standing, L. , Conezio, J., & Haber, R. N.Perception and memory for pictures: Single-trial learning of 2500 visual stimuli. ay,-gharignici12kard2, 1970, Li, 73-74,

Varley, W. H. , Levin, J. R., Severson, R. A,Wolff, P. Training imagery production in

young children through motor involvement.Iournal of Education 1 Psvcholoay. 1974,in, 262-266.

Wiener, M., & Cromer, W. Reading andreading difficulty: A conceptual analysis.Harvard Educational Review, 1967, 37,620-643.

Wilder, L., & Levin, J. R. A developmentalstudy of pronouncing responses in thediscrimination learning of words and pic-tures. jmund of Experimental ChildPsYchologv, 1973, 278-286.

Wohlwill, J. F. Responses to class-inclusionquestions for verbally and pictorially pre-sented items. Child Development. 1968,

449-465.Wolff, P. , & Levin, J. R. The role of overt

activity in children's imagery production.Child Development, 1972, ja., 537-547.

Wolff, P., Levin, J. R., & Longobardi, E. T.Motoric mediation in children's paired-associate learning: Effects of visual andtactual contact. Journal of ExperimentalChild Psvcholoav, 1972, 176-183.

Wolff, P. , Levin, J. R., & Longobardi, E. T.Activity and children's learning. ChildDevelopment, 1974, 91., 221-223.

27