Scandinavian Journal of Psychology, 1992, 33, 193-21 1
Evaluating young children’s cognitive capacities through computer versus hand drawings JOHN OLSEN Padagogisk Psykologisk Rddgivning, Elsinore, Denmark
Olsen, J. ( 1992). Evaluating young children’s cognitive capacities through computer versus hand drawings. Scmdimavian Journal of Psychology, 33, 193-21 1. Young normal and handicapped children, aged 3 to 6 years, were taught to draw a Scene of a house, garden and a sky with a computer drawing program that uses icons and is operated by a mouse, The drawings were rated by a team of experts on a 7-category scale. The children’s computer- and hand- produced drawings were compared with one another and with results on cognitive, visual and fine motor tests. The computer drawing program made it possible for the children to accurately draw closed shapes, to get instant feedback on the adequacy of the drawing, and to make corrections with ease. It was hypothesized that these features would compensate for the young children’s limitations in such cognitive skills, as memory, concentration, planning and accomplishment, as well as their weak motor skills. In addition, it was hypothesized that traditional cognitive ratings of hand drawings Tay underestimate young children’s intellectual ability, because drawing by hand demands motor skills and memory, concentration and planning skills that are more developed than that actually shown by young children. To test the latter hypothesis, the children completed a training program in using a computer to make drawings. The results show that cognitive processes such as planning, analysis and synthesis can be investigated by means of a computer drawing program in a way not possible using traditional pencil and paper drawings. It can be said that the method used here made it possible to measure cognitive abilities “under the Boor” of what is ordinarily possible by means of traditionally hand drawings. Key words: Developmental psychology, computer systems, cognitive ability.
John Olsen, Paahgogirk Psykologisk Rridgivning, Birkednlsvej 27, 3000 Elsinore, Denmark
Children’s drawings have been studied for about 100 years to understand what they reveal about the development of children’s intelligence, language, personality and emotions. These studies have dealt with hand drawings, but this paper will demonstrate that as the computer becomes a familiar tool for children, researchers have been provided with a new method for the study of children’s psychological processes. Hams (1963) conducted a thorough review of the research investigating children’s drawings from 1888 to 1963. He found that although children draw different kinds of objects (e.g. houses; BalIard, 1913), most of the studies had focused on the child‘s freehand drawing of a person. He also reported that most studies focused on school-age children, although drawings of 3 to 6 year old children have also been studied (Hams, 1963; Selfe, 1983; Lark-Horovitz et al., 1973; Colomb, 1974; and Mortensen, 1984). Hams concluded that studies clearly demonstrated that the drawing process changes as the child matures. A similar conclusion was drawn by Lark-Horovitz et al. (1973) who found a universal course of The research was conducted while the author was a visiting researcher in 1988 at Trace Research & Development Center, Univeristy of Wisconsin, Madison. I gratefully acknowledge the Center’s generous support during all phases of my work. I thank Elizabeth Doll (University of Colarado-Denver) and Hans Vejleskov (The Royal Danish School of Education Studies) for comments on a draft of this article.
194
J. Olsen
Scand J Psycho1 33 (1992)
development in their comparisons of child art from Western and non-Western cultures. Most researchers in the field would agree with Luquet’s description of these ‘universal’ stages, paraphrased here from Selfe (1983): (1) The pre-schematic stage or the stage of “synthetic incapacity” from 2.5-5 years approximately. This is the stage of the child’s earliest scribbles and gradual attempts to draw forms that represent some object of his experience, although he fails in positioning, in co-ordination of lines and in general to synthesize elements.
Most of the children in this study are in this age group while only few of them are in the beginning of the next age group (6 to 9 years).
(2) The schematic stage or the stage of “intellectual realism.” This is the stage when the child draws “what he knows rather than what he sees”, and none of the children are in the last age group (3) The stage of “visual realism” usually attained after age 10 years when the child can choose to draw relatively true-to-life representations of visual scenes from one fixed view point.
-
According to Selfe (1983), most researchers in the field have found cognitive development to be at the core of the drawing process. In spite of this agreement, the explanations of which psychological processes are involved and how they are involved differ considerably among main schools of thought. Hams (1963) in his review of the literature suggests that a child‘s drawing of an object is an index of his conception of that object. This is the basis of the Goodenough-Harris drawing test, that: “. . . the child’s drawing of any object will reveal the discriminations he has made about that object as belonging to a class, i.e., as a concept. In particular, it is hypothesized that his concept of a frequently experienced object, such as a human being, becomes a useful index to the growing complexity of his concepts generally.” (Hams, 1963). Although most studies have focused on the drawing of a man, Hams points out that other objects, like houses, are often drawn spontaneously, as already shown by Ballard (1913). Hams (1963) also points out that as long as the child draws a known object the product will reflect the child’s concept and cognitive level. Eng (1931, 1957), in agreement with Harris’s position, held that children’s drawings are schematic because their concepts are more concrete, less differentiated and abstract, but also because their drawing techniques are limited. Children produce more realistic drawings as their concepts become clearer and more differentiated, ,and as their skill with the medium increases. Further she explains the child’s defective sense of proportion and placement as being due to their feeble Gpadty for synthesis: “They cannot, when occupied with drawing the details, retain a grasp of the whole. The parts are drawn one by one without taking the total effect into account.” And (p.187): “The child’s drawings testify to an imagination which is active and lively, although weak and planless. It flits in chance associations from idea to idea, with only slight hold on objective reality. Later. . , follow a leading idea and to create an organic whole in its drawing.” Although the concept of intelligence and the acceptance of the influence of the environment on cognitive development have changed since Eng, her point of view is still the main basis for interpretation of children’s hand drawings as cognitive measures as in the Goodenough-Harris drawing test. Alternatively Goodnow (1977) analyses the drawing process in terms of the child’s use of units such as straight lines, scribbled lines, circles, squares, triangles or Latin crosses. She
Scand f Psycho1 33 (1992)
Computer versus hand drawings
considers how the child chooses units, and how he combines them. She also notes that the result depends on the sequences of choosing and combining units and where to start on the paper. She points out that the starting point on the paper often makes a difference between success and failure, and that each step affects later possible steps. She states that omissions are not always caused by lack of higher level concepts but may be a result of the technical difficulties children experience when combining certain units. It might simply seem too difficult for the child to try. It is acknowledged by researchers such as Eng (1931) that the very medium of drawing imposes constraints on which results are possible, just as opportunity to practice, exposure to teaching, level of motor skills and the nature of the drawing task do. Colomb (1974) argues that the child’s representation of a person clearly varies as a function of the task (such as drawing, modeling, sketching to dictation, completing a drawing and construct a drawing from a puzzle) and the medium employed (such as pencil-paper, water color, modeling dough and concrete materials). Goodenough (cited in Hams, 1963) showing the positive effect of training on a first-grade child’s drawing. It is well known that motor skills and coordination are developed not only through practice but also through maturity. Selfe describes the process of hand drawings as very much similar to writing: a global, intuitive idea, an image has to be produced through a linear process of sequences before the final product, here a drawing, can be perceived by others in a global, intuitive way. A11 of this presumes that the child is drawing by hand. Taken together, prior researchers have suggested that the adequacy of the child’s hand drawing will be iduenced by the sophistication of the child’s concepts, the child’s motor coordination, the child’s ability to plan and impose structure on the drawing act, the child’s mastery of the medium, and the child‘s sense of proportion and placement. The lack of motor coordination in young children accounts for their imprecision in making straight lines and sharp angles, but does not by itself account for their lack of proportion and the difficulty they experience in the placement of parts relative to the whole. Drawing by computer is different from drawing by hand because it lessens or even eliminates the impact of some of these iduences. Consider this example: When the child is asked to draw a house with a pencil or marker he, presumably begins with an inner picture of a house. He must also have some kind of plan for where to start on the paper, which part and unit to start with, and how to combine these units into a house. Suppose he tries to make a square as an outline for the house. After each line and angle is drawn, his choices among possible solutions for the “class of houses” will decrease drastically. Moreover his opportunities for compensating for former errors with later solutions is lessened. Not only is it necessary to draw most of the outline of the square before seeing if the end result will look like a house or not, but it is almost impossible to correct any errors and try over again. Even when using a pencil instead of a colored marker, a young child’s use of an eraser could damage the paper or at least leave a big gray smudge on the outline. Continuing with the drawing and placing windows, a door, a roof, a chimney, trees, a bush, a sun and a cloud demands additional cognitive planning to insure, for instance, that the child does not run out of space on the paper and that he will be able to position parts relative to one other in an appropriate way. For instance, the child must know where to place the door once the two windows are drawn, and where to place the cloud once the house, garden and the sun are drawn. When drawing by computer, the child uses the rectangle icon as a unit. He moves the icon by moving a ‘‘mouse”-Controlled arrow on the screen. The arrow moves when the
195
196
J. Olsen
&and J Psycho1 33 ( 1992)
(5 File Edit lioodies Font FontSize Style
a I
bd C
I
Fig. 1. Picture of a computer screen with a drawing program menu. The four rectangles represent steps in construction of an outline for house.
mouse moves, and picks up or drops the icon when a button on the mouse is clicked. Fig. 1- shows a picture of a computer screen with the icons that a child has been using to make an outline for a house. To pick up the rectangle, he moves the arrow to the rectangle icon and clicks, which highlights the icon. Then he releases the button and moves the arrow to where he wants to start the outline, somewhere in the middle area of the screen. To drop the rectangle there, he again clicks the button. Then he keeps pressing the button while moving the mouse diagonally across the screen, and something completely different from using pencil or marker happens: a whole rectangle develops instantly and completely with four straight, parallel lines and right angles. The child might stop the movement at point a still pressing the button to judge if he wants the outline looking that way. He decides he wants it wider and moves the mouse horizontally to b, and while he moves, the rectangle widens instantly and fully into a new rectangle. He decides at b, that the outline is too thin, so he moves vertically to c. There he judges that it is now too thick, so he moves the mouse vertically to d. While he moves the rectangle shrinks. At d, he is satisfied with his outline, he releases the button and the outline is done. Fig. 1 gives a distorted, static picture of how the operations are actually perceived by the child. On the screen, only one of the squares or rectangles is visible to the child at a time, and each changes instantly with movement of the mouse. A special feature with this drawing program is that when the mouse button is released the shape drawn is instantly filled with whatever color has been chosen from the color menu at the bottom of the screen.
Scand J hychol 33 (1992)
Computer vemw hand drawings
Constructing a window is done in the same way and with the same icon. If the child wants another color he moves the arrow to click on the wanted color icon, and then moves to where he wants to start the window. By pressing and moving the mouse, he can change the size of the window until he is satisfied with its proportion relative to the house’s outline and his internal image of “house”. This drawing process is described in such detail because these special features of the computer drawing program make the computer drawing process completely different from hand drawings. Using these computer features, drawing is not done by sequentially combining lower order units (lines, angles), and the child does not need to wait until most of the construction is finished to judge it. The units used on the computer are of a higher order than those used by pencil; they are wholes in and of themselves. The child doesn’t use lines and angles in a complex sequence to make a rectangle, but uses a rectangle from the first movement. He does not need to go through several steps of construction to see the end result; it is shown instantly and completely. He doesn’t need to start again on a new sheet of paper to change the outline, he just moves the mouse in another direction. If a mistake is made, he doesn’t have to live with it, make a wild association and reformulate the task, or give up. He can just move back again or use UNDO (a special order in the drawing program). In this way, the child can experiment with each visual gestalt as one single variable at a time. During training, the examiner can point out exactly what is the important aspect of that gestalt and describe it verbally while the child is concretely manipulating it. In this way the child can explore different sizes and types of quadrangles (rectangles and squares) which would take days to make on paper with a pencil or marker. Further, the quadrangle on the screen is not distorted by curved, shaky lines and “round” angles, so the concept of the shape is illustrated concretely and precisely. The use of these features of the software makes it possible to investigate when distortions of proportion and placements in children’s hand drawings are due to constraints of cognition and when they are due to the children’s use of tools that are inappropriately &cult for them to manipulate. This experiment will make use of such drawing software to investigate how children from about three to six years draw a scene with a house, a garden and a sky by hand and by use of a computer. It assumes, that the children’s cognitive level will be expressed as precisely through their drawing of a house as through their drawing of a person.
Hypothesis The hypothesis proposed in this experiment is that: interpretation of paper and pencil or marker drawings might considerably underestimate a child’s actual level of cognition. This will be especially true for normal young children age three to five years who are working at Luquet’s preschematic stage of drawing ability. METHOD Equipment used For hand drawings the children used Crayola color markers and white bond paper, and in the Frostig test they used a J. R. Moon 600 pencil, both of which were familiar to them in their classroom. The computer drawing program used was a commercial software package called Paintworks P h , operated using the mouse on an Apple IIgs Personal Computer equipped with a color monitor.
Subjects The children were chosen from the Early Childhood and Kindergarten classes of the Waisman Center on Mental Retardation and Human Development, and were in three age class groups: Group 1: 10 children
197
198
J. Olsen
Scand J Psycho1 33 (1992)
from the so-called Cloud class, age three (mean age 3.4 year); Group 2: 13 children from the so-called Sunshine class, ages four to five (mean age 4.7 year); and Group 3: 13 childen from Kindergarten, ages five to six (mean age 5.6 year). All enrolled children were signed up for the experiment voluntarily by their parents and all who signed participated. The Early Waisman Childhood and Kindergarten program is specially designed to be a site for child development research, and it integrates normal children with children having cognitive and/or physical impairments. Three children from the Kindergarten and three children from the Sunshine class received special education in their classes. The impairments ranged from minor speech problems to more serious syndromes such as Downs and Willy syndromes. Most of the children’s parents were researchers at the University of Wisconsin, Madison. All the children knew a computer from their classes, and the teaching in the classes were excellent. Thus all the subjects were very well stimulated. and cannot be regarded a random sample from a representative population; however, this fact is of no relevance to the hypothesis being tested. Only three of the children from the oldest group, the Kindergarten class knew how to operate the mouse in a computer drawing program and none of them knew how to use the special drawing features used in this experiment. Because none of the children had worked with this hardware or the software before, a method was developed to teach the children to use the computer and to make computer drawings. The two youngest children from the Cloud class were excluded after the third session because they could not follow the instructions for operating the mouse, but their results are included in the data set for the first 3 sessions. Two children differedvery much from the other children in their approach to drawing, which yielded results different From the overall pattern. One child from the Kindergarten class got caught very easily in his phantasies about rockets. Almost everything gave him associations that led him to see what he was doing as related to rockets. So, not surprisingly, his free drawings at the end of the experiment couldn’t be recognized by the raters as being a house. A child from the Cloud class showed a similar tendency: almost every kind of shape gave him associations of being caught in a trap, so that he redefined the task to get out of that trap. The data from both of these children are included in the analysis.
Procedure Before selection of the subjccts, the examiner participated as a visitor in the classrooms, so that the children could become familiar with him. The subjects were tested and trained individually by the examiner over a period of about three weeks in a room familiar to them near their classrooms. Four premeasures of the children’s ability in skills related to drawing were administered as a measure of the children’s cognitive ability: 1) Raven Coloured Progressive Matrices (Raven, 1962) assessing perceptual reasoning, 2) a set of informal tasks using concrete materials to determine their basic skills in counting to eight, and understanding of “most” and number conservation, and 3) The Sentences subtest from the widely used Wechsla Preschool and Primary Scale of Intelligence (Weschler, 1966). The Sentence Span subtest assesses ability to repeat sentences verbatim and was used as a measure of the child‘s short-term memory. (Some of the children had the Sentence Span test after the first but before the fourth session of the experiment because of the late decision to include this test). As a measure of the children’s he-motor skills 4) the Eye-hand coordination subtest (PartI) of the Frostig Development Test of Visual Perception (Frostig, 1966) was used. This subtest assesses skill in drawing various kinds of continuous lines within boundaries and from point to point. The children were also given some informal tests of determining the color (red, yellow, blue, green); the number of objects up to eight; the set with most objects (4 compared with 6); to arrange in sequence five pieces of wood of different height and to realize the conservation of numbers. Age norm scores for all, except the informal pretests are shown in Table 1. Though the age norms for these tests are very old and probably show too high age-norm scorn compared with todays children, it is evident that this sample of children is far from being representative. This fact had to be held in mind when generalizing from these data.
Sessions Training and experimental task of each session had the following content: Session 1: Introduction to the computer (about 10 min); Session 2: Learning and testing technical skills (about 30 min); Session 3: Learning and testing technical skills (about 20 min) followed by an introduction to drawing a house, sun, sunbeams and a cloud h s t by tracing a predrawn scene, and then by drawing to dictation (about 10 min); Session 4 Learning and testing technical skills (about 10 min) followed by training in drawing
Scand J Psycho1 33 (1992)
Computer versus hand drawings
a house and a sky by tracing, by dictation and by free drawing (about 10 min). Then the children got an introduction to draw trees and a bush by tracing and by dictation (about 5 min) followed by free drawing of a scene with a house, garden and a sky (about 5 rnin); Session 5: Training and exploring drawing abstractly (about 10 rnin), drawing scenes (about 15 min) and drawing free (about 5 min); Session 6: Making a house, a tree and a man from a puzzle by using the computer (about 15 min) and then drawing a scene to dictation of the computer (about 3 min); Session 7: Make a house, a tree and a man from a puzzle by using paper pieces equivalent to the computer versions (about 15 min) followed by drawing a scene to dictation by hand drawing (about 3 min); Session 8: Testing computer operation skills (about 30 min) and Session 9: Testing computer operation skills (continued) (about 20 min) followed by drawing a scene first on the computer (about 5 min) and finally by hand (about 5 min). The pre-test and session 1 were run together and took about one hour, while the other sessions each took about half an hour. Half of the children had session 7 before session 6. All together the time of the sessions was used in the following way: Instructions and training in technical computer operation skills (about 50 min), testing computer operation skills (about 85 min), teaching how to draw a scene on the computer (about 45 rnin), testing how to make a scene by puzzle and to dictation (computer version: about I S rnin and paper version: about 20 rnin), testing drawing a scene (computer version: about 10 rnin and paper version: about 5 rnin). The tasks were tried out with three pilot children, one from each class, before the start of the experiment. The results from these three children are not included in the data analysis. The highly structured sequence of 9 sessions took about a month to complete with all children. All sessions took about 30 min except the first session, which took 15 rnin. The tasks were presented by the examiner guided by written instructions, and the standardized sessions to ensure that, as far as possible, all children were presented in the same way to the same kind of stimuli. The children performed three types of tasks. The first kind of task was a technical one about using the computer drawing program. They were taught the basic mouse operations of pointing to an icon, clicking on it, moving it out in the working area on the screen, and operating it by dragging. (Dragging means to keep pressing the bottom and move the mouse, whereby the child can draw lines on the screen). Most of the exercises with the technical tasks consisted of teaching the basic mouse operations to ensure that the children’s skills in operating the mouse were on the same level as their skills using a marker. These instructions in using the computer drawing program was given in sessions 1, 2, 3, 4, 5 and 6 and took all together about 50 min. The second type of task was to learn to make a drawing of a scene on the computer, see Category 7 , the model in Fig. 2. This was done by tracing a model of a scene, by drawing by dictation and by free drawings on the screen. This teaching took place in sessions 3, 4,and 5 and took all together about 45 min. The third type of task was testing the children’s skills. Their technical skills in using the computer drawing program were tested in sessions 2, 3, 4, 8 and 9 and took all together about 85 min. Their skills in making a scene either from a puzzle, by dictation or by free drawing both on the computer and with paper and pencil were tested in sessions 5, 6, 7 . 8 and 9 and took all together about 50 min. The data analysed in this article describe the children’s results in making a free drawing of a house, garden and a sky using both a computer and pencil and paper.
Measures The children made a hand drawing of a house in the first session. At the end of the fifth session as well as in the 9th and last session they were told to make a computer drawing of a scene consisting of a house, garden and a sky. After having finished the work with the computer in the last session the children were asked to make a hand drawing of the scene. Also in the last session, a computer version of the Frostig (part I) paper version was given to measure the precision with which the children drew lines by use of the mouse and the pcncil icon on the computer screen. The results of this computer version of the Frostig test were scored by the Same criteria as the paper version and the results of the two versions are compared to estimate whether it was more difficult to make lines with precision by use of pencil and paper or by use of a mouse and a pencil icon on a computer screen. A rating scale was developed to evaluate and compare computer and hand drawings (Appendix). The scale was constructed on the basis of the evaluations and discussions of an initial group of four raters: three psychologists and a research specialist in the field of Communication Disorders, who rated all the drawings. The scale describes seven levels of drawing ability and included examples of each category from the computer drawings, see Fig. 2. Drawings were evaluated relative to what could be expected of a normal, ten-year-old child. The age-norm of ten years was chosen as the criteria for the maximum score in order to prevent the raters from making
199
200
J. Olsen
Scand J Psycho1 33 (1992)
The scale.
Category 1. child number 6
Category2,child number 1
Categoryt child number 2
Catsgwp6,cbildnumbst 29
CateeW7 3. child number 9
-
Categury 5, child number 21
cateqorp7, child number
n
The euamplesan redrawn from the orighat scanned drapinos and hwe thorn in their 40% size without colors. Fig. 2 The scale. The categories shown are the examples on the scale illustrating the lower criteria of the categories.
Scand J Psycho1 33 (1992)
Computer versus hand drawings
too many allowances for what is typical of young children’s abilities to draw. The main focus of the evaluation was on the proportion of the house and placement of the elements in relation to the whole. If the rater was in doubt he/she was told to “follow his/her intuitive impression”. The raters were instructed in the criteria defining the scale about five minutes before they began rating, and during the rating they could read the guidelines. The hand and computer drawings of the scenes were rated using this seven-point scale by six raters including two persons (R3 and R4) from the initial group and four new raters. Analyses were conducted using these six raters, which were the examiner, R3; the staff member from the Communication Disorder Section, R4; an engineer, R6; an occupational therapist, R2; and two speech therapists, R5 and R7. All raters worked at the Trace Research and Development Center and all were without special experience in evaluating drawings. All six raters rated all four drawings independently for all children, except R2 who did not rate the first free computer drawing made in session 5.
RESULTS In this section the results of four analyses are described. First, the pre-test measures of some psychomotoric skills necessary for operating a computer are described and compared with the norm samples from the test manuals. Second, the differences among the realism between the children’s hand and computer drawings are examined. Then the data are examined to see if they fit to a Rasch model, and the Drawing Score Category Scale as well as the children’s overall drawing abilities and the quality of the realism for the hand and the computer drawings are analysed using this Rasch measurement model. Finally, it is described how well the children did on a paper version compared to a computer version of one of the pre-tests, the Frostig Eye-Hand Coordination Subtest (part I), when they used a drawing tool on the computer which works like a normal pencil. Pretest
Table 1 presents the mean scores of all children on the three pretest measures having standardized age norms: The Sentence Span Subtest; the Raven Coloured Progressive Matrices and the Frostig Eye-Hand Coordination Subtest. When the mean scores are compared for the normal children between the classes it is seen for all three tests, that the Cloud class scores lower than the Sunshine class and the Kindergarten class, while the mean scores for the Sunshine class and the Kindergarten class are very close to each other. One way Anova analysis gives significant differences ( p < 0.01) between the mean scores for Raven and the Frostig (part I) and ( p c 0.05) for sentence span among Cloud versus Sunshine and Kindergarten classes. No mean score differences between Sunshine and Kindergarten classes is significant ( p > 0.05). When age norms are compared with the children’s living ages it is seen, that the children in this study, except for the special education children, got higher scores than comparable mean scores for children in the reference population. All of the children above 4 years of age except two had mastered the basic concepts of color, numbers to 8, conservation of number and senation. Among the youngest children in the Cloud class, only three of the children had acquired all these skills, the two whose participation was discontinid had not mastered any of these skills, while the rest had mastered some and were unsure of others of these skills. Interrater reliability
The six raters used the same 7 point rating scale to score all four drawings, hand as well as computer drawings. Correlations among scores for all four drawings among all pairs of the six raters are all significantly different from zero ( p
Evaluating young children’s cognitive capacities through computer versus hand drawings JOHN OLSEN Padagogisk Psykologisk Rddgivning, Elsinore, Denmark
Olsen, J. ( 1992). Evaluating young children’s cognitive capacities through computer versus hand drawings. Scmdimavian Journal of Psychology, 33, 193-21 1. Young normal and handicapped children, aged 3 to 6 years, were taught to draw a Scene of a house, garden and a sky with a computer drawing program that uses icons and is operated by a mouse, The drawings were rated by a team of experts on a 7-category scale. The children’s computer- and hand- produced drawings were compared with one another and with results on cognitive, visual and fine motor tests. The computer drawing program made it possible for the children to accurately draw closed shapes, to get instant feedback on the adequacy of the drawing, and to make corrections with ease. It was hypothesized that these features would compensate for the young children’s limitations in such cognitive skills, as memory, concentration, planning and accomplishment, as well as their weak motor skills. In addition, it was hypothesized that traditional cognitive ratings of hand drawings Tay underestimate young children’s intellectual ability, because drawing by hand demands motor skills and memory, concentration and planning skills that are more developed than that actually shown by young children. To test the latter hypothesis, the children completed a training program in using a computer to make drawings. The results show that cognitive processes such as planning, analysis and synthesis can be investigated by means of a computer drawing program in a way not possible using traditional pencil and paper drawings. It can be said that the method used here made it possible to measure cognitive abilities “under the Boor” of what is ordinarily possible by means of traditionally hand drawings. Key words: Developmental psychology, computer systems, cognitive ability.
John Olsen, Paahgogirk Psykologisk Rridgivning, Birkednlsvej 27, 3000 Elsinore, Denmark
Children’s drawings have been studied for about 100 years to understand what they reveal about the development of children’s intelligence, language, personality and emotions. These studies have dealt with hand drawings, but this paper will demonstrate that as the computer becomes a familiar tool for children, researchers have been provided with a new method for the study of children’s psychological processes. Hams (1963) conducted a thorough review of the research investigating children’s drawings from 1888 to 1963. He found that although children draw different kinds of objects (e.g. houses; BalIard, 1913), most of the studies had focused on the child‘s freehand drawing of a person. He also reported that most studies focused on school-age children, although drawings of 3 to 6 year old children have also been studied (Hams, 1963; Selfe, 1983; Lark-Horovitz et al., 1973; Colomb, 1974; and Mortensen, 1984). Hams concluded that studies clearly demonstrated that the drawing process changes as the child matures. A similar conclusion was drawn by Lark-Horovitz et al. (1973) who found a universal course of The research was conducted while the author was a visiting researcher in 1988 at Trace Research & Development Center, Univeristy of Wisconsin, Madison. I gratefully acknowledge the Center’s generous support during all phases of my work. I thank Elizabeth Doll (University of Colarado-Denver) and Hans Vejleskov (The Royal Danish School of Education Studies) for comments on a draft of this article.
194
J. Olsen
Scand J Psycho1 33 (1992)
development in their comparisons of child art from Western and non-Western cultures. Most researchers in the field would agree with Luquet’s description of these ‘universal’ stages, paraphrased here from Selfe (1983): (1) The pre-schematic stage or the stage of “synthetic incapacity” from 2.5-5 years approximately. This is the stage of the child’s earliest scribbles and gradual attempts to draw forms that represent some object of his experience, although he fails in positioning, in co-ordination of lines and in general to synthesize elements.
Most of the children in this study are in this age group while only few of them are in the beginning of the next age group (6 to 9 years).
(2) The schematic stage or the stage of “intellectual realism.” This is the stage when the child draws “what he knows rather than what he sees”, and none of the children are in the last age group (3) The stage of “visual realism” usually attained after age 10 years when the child can choose to draw relatively true-to-life representations of visual scenes from one fixed view point.
-
According to Selfe (1983), most researchers in the field have found cognitive development to be at the core of the drawing process. In spite of this agreement, the explanations of which psychological processes are involved and how they are involved differ considerably among main schools of thought. Hams (1963) in his review of the literature suggests that a child‘s drawing of an object is an index of his conception of that object. This is the basis of the Goodenough-Harris drawing test, that: “. . . the child’s drawing of any object will reveal the discriminations he has made about that object as belonging to a class, i.e., as a concept. In particular, it is hypothesized that his concept of a frequently experienced object, such as a human being, becomes a useful index to the growing complexity of his concepts generally.” (Hams, 1963). Although most studies have focused on the drawing of a man, Hams points out that other objects, like houses, are often drawn spontaneously, as already shown by Ballard (1913). Hams (1963) also points out that as long as the child draws a known object the product will reflect the child’s concept and cognitive level. Eng (1931, 1957), in agreement with Harris’s position, held that children’s drawings are schematic because their concepts are more concrete, less differentiated and abstract, but also because their drawing techniques are limited. Children produce more realistic drawings as their concepts become clearer and more differentiated, ,and as their skill with the medium increases. Further she explains the child’s defective sense of proportion and placement as being due to their feeble Gpadty for synthesis: “They cannot, when occupied with drawing the details, retain a grasp of the whole. The parts are drawn one by one without taking the total effect into account.” And (p.187): “The child’s drawings testify to an imagination which is active and lively, although weak and planless. It flits in chance associations from idea to idea, with only slight hold on objective reality. Later. . , follow a leading idea and to create an organic whole in its drawing.” Although the concept of intelligence and the acceptance of the influence of the environment on cognitive development have changed since Eng, her point of view is still the main basis for interpretation of children’s hand drawings as cognitive measures as in the Goodenough-Harris drawing test. Alternatively Goodnow (1977) analyses the drawing process in terms of the child’s use of units such as straight lines, scribbled lines, circles, squares, triangles or Latin crosses. She
Scand f Psycho1 33 (1992)
Computer versus hand drawings
considers how the child chooses units, and how he combines them. She also notes that the result depends on the sequences of choosing and combining units and where to start on the paper. She points out that the starting point on the paper often makes a difference between success and failure, and that each step affects later possible steps. She states that omissions are not always caused by lack of higher level concepts but may be a result of the technical difficulties children experience when combining certain units. It might simply seem too difficult for the child to try. It is acknowledged by researchers such as Eng (1931) that the very medium of drawing imposes constraints on which results are possible, just as opportunity to practice, exposure to teaching, level of motor skills and the nature of the drawing task do. Colomb (1974) argues that the child’s representation of a person clearly varies as a function of the task (such as drawing, modeling, sketching to dictation, completing a drawing and construct a drawing from a puzzle) and the medium employed (such as pencil-paper, water color, modeling dough and concrete materials). Goodenough (cited in Hams, 1963) showing the positive effect of training on a first-grade child’s drawing. It is well known that motor skills and coordination are developed not only through practice but also through maturity. Selfe describes the process of hand drawings as very much similar to writing: a global, intuitive idea, an image has to be produced through a linear process of sequences before the final product, here a drawing, can be perceived by others in a global, intuitive way. A11 of this presumes that the child is drawing by hand. Taken together, prior researchers have suggested that the adequacy of the child’s hand drawing will be iduenced by the sophistication of the child’s concepts, the child’s motor coordination, the child’s ability to plan and impose structure on the drawing act, the child’s mastery of the medium, and the child‘s sense of proportion and placement. The lack of motor coordination in young children accounts for their imprecision in making straight lines and sharp angles, but does not by itself account for their lack of proportion and the difficulty they experience in the placement of parts relative to the whole. Drawing by computer is different from drawing by hand because it lessens or even eliminates the impact of some of these iduences. Consider this example: When the child is asked to draw a house with a pencil or marker he, presumably begins with an inner picture of a house. He must also have some kind of plan for where to start on the paper, which part and unit to start with, and how to combine these units into a house. Suppose he tries to make a square as an outline for the house. After each line and angle is drawn, his choices among possible solutions for the “class of houses” will decrease drastically. Moreover his opportunities for compensating for former errors with later solutions is lessened. Not only is it necessary to draw most of the outline of the square before seeing if the end result will look like a house or not, but it is almost impossible to correct any errors and try over again. Even when using a pencil instead of a colored marker, a young child’s use of an eraser could damage the paper or at least leave a big gray smudge on the outline. Continuing with the drawing and placing windows, a door, a roof, a chimney, trees, a bush, a sun and a cloud demands additional cognitive planning to insure, for instance, that the child does not run out of space on the paper and that he will be able to position parts relative to one other in an appropriate way. For instance, the child must know where to place the door once the two windows are drawn, and where to place the cloud once the house, garden and the sun are drawn. When drawing by computer, the child uses the rectangle icon as a unit. He moves the icon by moving a ‘‘mouse”-Controlled arrow on the screen. The arrow moves when the
195
196
J. Olsen
&and J Psycho1 33 ( 1992)
(5 File Edit lioodies Font FontSize Style
a I
bd C
I
Fig. 1. Picture of a computer screen with a drawing program menu. The four rectangles represent steps in construction of an outline for house.
mouse moves, and picks up or drops the icon when a button on the mouse is clicked. Fig. 1- shows a picture of a computer screen with the icons that a child has been using to make an outline for a house. To pick up the rectangle, he moves the arrow to the rectangle icon and clicks, which highlights the icon. Then he releases the button and moves the arrow to where he wants to start the outline, somewhere in the middle area of the screen. To drop the rectangle there, he again clicks the button. Then he keeps pressing the button while moving the mouse diagonally across the screen, and something completely different from using pencil or marker happens: a whole rectangle develops instantly and completely with four straight, parallel lines and right angles. The child might stop the movement at point a still pressing the button to judge if he wants the outline looking that way. He decides he wants it wider and moves the mouse horizontally to b, and while he moves, the rectangle widens instantly and fully into a new rectangle. He decides at b, that the outline is too thin, so he moves vertically to c. There he judges that it is now too thick, so he moves the mouse vertically to d. While he moves the rectangle shrinks. At d, he is satisfied with his outline, he releases the button and the outline is done. Fig. 1 gives a distorted, static picture of how the operations are actually perceived by the child. On the screen, only one of the squares or rectangles is visible to the child at a time, and each changes instantly with movement of the mouse. A special feature with this drawing program is that when the mouse button is released the shape drawn is instantly filled with whatever color has been chosen from the color menu at the bottom of the screen.
Scand J hychol 33 (1992)
Computer vemw hand drawings
Constructing a window is done in the same way and with the same icon. If the child wants another color he moves the arrow to click on the wanted color icon, and then moves to where he wants to start the window. By pressing and moving the mouse, he can change the size of the window until he is satisfied with its proportion relative to the house’s outline and his internal image of “house”. This drawing process is described in such detail because these special features of the computer drawing program make the computer drawing process completely different from hand drawings. Using these computer features, drawing is not done by sequentially combining lower order units (lines, angles), and the child does not need to wait until most of the construction is finished to judge it. The units used on the computer are of a higher order than those used by pencil; they are wholes in and of themselves. The child doesn’t use lines and angles in a complex sequence to make a rectangle, but uses a rectangle from the first movement. He does not need to go through several steps of construction to see the end result; it is shown instantly and completely. He doesn’t need to start again on a new sheet of paper to change the outline, he just moves the mouse in another direction. If a mistake is made, he doesn’t have to live with it, make a wild association and reformulate the task, or give up. He can just move back again or use UNDO (a special order in the drawing program). In this way, the child can experiment with each visual gestalt as one single variable at a time. During training, the examiner can point out exactly what is the important aspect of that gestalt and describe it verbally while the child is concretely manipulating it. In this way the child can explore different sizes and types of quadrangles (rectangles and squares) which would take days to make on paper with a pencil or marker. Further, the quadrangle on the screen is not distorted by curved, shaky lines and “round” angles, so the concept of the shape is illustrated concretely and precisely. The use of these features of the software makes it possible to investigate when distortions of proportion and placements in children’s hand drawings are due to constraints of cognition and when they are due to the children’s use of tools that are inappropriately &cult for them to manipulate. This experiment will make use of such drawing software to investigate how children from about three to six years draw a scene with a house, a garden and a sky by hand and by use of a computer. It assumes, that the children’s cognitive level will be expressed as precisely through their drawing of a house as through their drawing of a person.
Hypothesis The hypothesis proposed in this experiment is that: interpretation of paper and pencil or marker drawings might considerably underestimate a child’s actual level of cognition. This will be especially true for normal young children age three to five years who are working at Luquet’s preschematic stage of drawing ability. METHOD Equipment used For hand drawings the children used Crayola color markers and white bond paper, and in the Frostig test they used a J. R. Moon 600 pencil, both of which were familiar to them in their classroom. The computer drawing program used was a commercial software package called Paintworks P h , operated using the mouse on an Apple IIgs Personal Computer equipped with a color monitor.
Subjects The children were chosen from the Early Childhood and Kindergarten classes of the Waisman Center on Mental Retardation and Human Development, and were in three age class groups: Group 1: 10 children
197
198
J. Olsen
Scand J Psycho1 33 (1992)
from the so-called Cloud class, age three (mean age 3.4 year); Group 2: 13 children from the so-called Sunshine class, ages four to five (mean age 4.7 year); and Group 3: 13 childen from Kindergarten, ages five to six (mean age 5.6 year). All enrolled children were signed up for the experiment voluntarily by their parents and all who signed participated. The Early Waisman Childhood and Kindergarten program is specially designed to be a site for child development research, and it integrates normal children with children having cognitive and/or physical impairments. Three children from the Kindergarten and three children from the Sunshine class received special education in their classes. The impairments ranged from minor speech problems to more serious syndromes such as Downs and Willy syndromes. Most of the children’s parents were researchers at the University of Wisconsin, Madison. All the children knew a computer from their classes, and the teaching in the classes were excellent. Thus all the subjects were very well stimulated. and cannot be regarded a random sample from a representative population; however, this fact is of no relevance to the hypothesis being tested. Only three of the children from the oldest group, the Kindergarten class knew how to operate the mouse in a computer drawing program and none of them knew how to use the special drawing features used in this experiment. Because none of the children had worked with this hardware or the software before, a method was developed to teach the children to use the computer and to make computer drawings. The two youngest children from the Cloud class were excluded after the third session because they could not follow the instructions for operating the mouse, but their results are included in the data set for the first 3 sessions. Two children differedvery much from the other children in their approach to drawing, which yielded results different From the overall pattern. One child from the Kindergarten class got caught very easily in his phantasies about rockets. Almost everything gave him associations that led him to see what he was doing as related to rockets. So, not surprisingly, his free drawings at the end of the experiment couldn’t be recognized by the raters as being a house. A child from the Cloud class showed a similar tendency: almost every kind of shape gave him associations of being caught in a trap, so that he redefined the task to get out of that trap. The data from both of these children are included in the analysis.
Procedure Before selection of the subjccts, the examiner participated as a visitor in the classrooms, so that the children could become familiar with him. The subjects were tested and trained individually by the examiner over a period of about three weeks in a room familiar to them near their classrooms. Four premeasures of the children’s ability in skills related to drawing were administered as a measure of the children’s cognitive ability: 1) Raven Coloured Progressive Matrices (Raven, 1962) assessing perceptual reasoning, 2) a set of informal tasks using concrete materials to determine their basic skills in counting to eight, and understanding of “most” and number conservation, and 3) The Sentences subtest from the widely used Wechsla Preschool and Primary Scale of Intelligence (Weschler, 1966). The Sentence Span subtest assesses ability to repeat sentences verbatim and was used as a measure of the child‘s short-term memory. (Some of the children had the Sentence Span test after the first but before the fourth session of the experiment because of the late decision to include this test). As a measure of the children’s he-motor skills 4) the Eye-hand coordination subtest (PartI) of the Frostig Development Test of Visual Perception (Frostig, 1966) was used. This subtest assesses skill in drawing various kinds of continuous lines within boundaries and from point to point. The children were also given some informal tests of determining the color (red, yellow, blue, green); the number of objects up to eight; the set with most objects (4 compared with 6); to arrange in sequence five pieces of wood of different height and to realize the conservation of numbers. Age norm scores for all, except the informal pretests are shown in Table 1. Though the age norms for these tests are very old and probably show too high age-norm scorn compared with todays children, it is evident that this sample of children is far from being representative. This fact had to be held in mind when generalizing from these data.
Sessions Training and experimental task of each session had the following content: Session 1: Introduction to the computer (about 10 min); Session 2: Learning and testing technical skills (about 30 min); Session 3: Learning and testing technical skills (about 20 min) followed by an introduction to drawing a house, sun, sunbeams and a cloud h s t by tracing a predrawn scene, and then by drawing to dictation (about 10 min); Session 4 Learning and testing technical skills (about 10 min) followed by training in drawing
Scand J Psycho1 33 (1992)
Computer versus hand drawings
a house and a sky by tracing, by dictation and by free drawing (about 10 min). Then the children got an introduction to draw trees and a bush by tracing and by dictation (about 5 min) followed by free drawing of a scene with a house, garden and a sky (about 5 rnin); Session 5: Training and exploring drawing abstractly (about 10 rnin), drawing scenes (about 15 min) and drawing free (about 5 min); Session 6: Making a house, a tree and a man from a puzzle by using the computer (about 15 min) and then drawing a scene to dictation of the computer (about 3 min); Session 7: Make a house, a tree and a man from a puzzle by using paper pieces equivalent to the computer versions (about 15 min) followed by drawing a scene to dictation by hand drawing (about 3 min); Session 8: Testing computer operation skills (about 30 min) and Session 9: Testing computer operation skills (continued) (about 20 min) followed by drawing a scene first on the computer (about 5 min) and finally by hand (about 5 min). The pre-test and session 1 were run together and took about one hour, while the other sessions each took about half an hour. Half of the children had session 7 before session 6. All together the time of the sessions was used in the following way: Instructions and training in technical computer operation skills (about 50 min), testing computer operation skills (about 85 min), teaching how to draw a scene on the computer (about 45 rnin), testing how to make a scene by puzzle and to dictation (computer version: about I S rnin and paper version: about 20 rnin), testing drawing a scene (computer version: about 10 rnin and paper version: about 5 rnin). The tasks were tried out with three pilot children, one from each class, before the start of the experiment. The results from these three children are not included in the data analysis. The highly structured sequence of 9 sessions took about a month to complete with all children. All sessions took about 30 min except the first session, which took 15 rnin. The tasks were presented by the examiner guided by written instructions, and the standardized sessions to ensure that, as far as possible, all children were presented in the same way to the same kind of stimuli. The children performed three types of tasks. The first kind of task was a technical one about using the computer drawing program. They were taught the basic mouse operations of pointing to an icon, clicking on it, moving it out in the working area on the screen, and operating it by dragging. (Dragging means to keep pressing the bottom and move the mouse, whereby the child can draw lines on the screen). Most of the exercises with the technical tasks consisted of teaching the basic mouse operations to ensure that the children’s skills in operating the mouse were on the same level as their skills using a marker. These instructions in using the computer drawing program was given in sessions 1, 2, 3, 4, 5 and 6 and took all together about 50 min. The second type of task was to learn to make a drawing of a scene on the computer, see Category 7 , the model in Fig. 2. This was done by tracing a model of a scene, by drawing by dictation and by free drawings on the screen. This teaching took place in sessions 3, 4,and 5 and took all together about 45 min. The third type of task was testing the children’s skills. Their technical skills in using the computer drawing program were tested in sessions 2, 3, 4, 8 and 9 and took all together about 85 min. Their skills in making a scene either from a puzzle, by dictation or by free drawing both on the computer and with paper and pencil were tested in sessions 5, 6, 7 . 8 and 9 and took all together about 50 min. The data analysed in this article describe the children’s results in making a free drawing of a house, garden and a sky using both a computer and pencil and paper.
Measures The children made a hand drawing of a house in the first session. At the end of the fifth session as well as in the 9th and last session they were told to make a computer drawing of a scene consisting of a house, garden and a sky. After having finished the work with the computer in the last session the children were asked to make a hand drawing of the scene. Also in the last session, a computer version of the Frostig (part I) paper version was given to measure the precision with which the children drew lines by use of the mouse and the pcncil icon on the computer screen. The results of this computer version of the Frostig test were scored by the Same criteria as the paper version and the results of the two versions are compared to estimate whether it was more difficult to make lines with precision by use of pencil and paper or by use of a mouse and a pencil icon on a computer screen. A rating scale was developed to evaluate and compare computer and hand drawings (Appendix). The scale was constructed on the basis of the evaluations and discussions of an initial group of four raters: three psychologists and a research specialist in the field of Communication Disorders, who rated all the drawings. The scale describes seven levels of drawing ability and included examples of each category from the computer drawings, see Fig. 2. Drawings were evaluated relative to what could be expected of a normal, ten-year-old child. The age-norm of ten years was chosen as the criteria for the maximum score in order to prevent the raters from making
199
200
J. Olsen
Scand J Psycho1 33 (1992)
The scale.
Category 1. child number 6
Category2,child number 1
Categoryt child number 2
Catsgwp6,cbildnumbst 29
CateeW7 3. child number 9
-
Categury 5, child number 21
cateqorp7, child number
n
The euamplesan redrawn from the orighat scanned drapinos and hwe thorn in their 40% size without colors. Fig. 2 The scale. The categories shown are the examples on the scale illustrating the lower criteria of the categories.
Scand J Psycho1 33 (1992)
Computer versus hand drawings
too many allowances for what is typical of young children’s abilities to draw. The main focus of the evaluation was on the proportion of the house and placement of the elements in relation to the whole. If the rater was in doubt he/she was told to “follow his/her intuitive impression”. The raters were instructed in the criteria defining the scale about five minutes before they began rating, and during the rating they could read the guidelines. The hand and computer drawings of the scenes were rated using this seven-point scale by six raters including two persons (R3 and R4) from the initial group and four new raters. Analyses were conducted using these six raters, which were the examiner, R3; the staff member from the Communication Disorder Section, R4; an engineer, R6; an occupational therapist, R2; and two speech therapists, R5 and R7. All raters worked at the Trace Research and Development Center and all were without special experience in evaluating drawings. All six raters rated all four drawings independently for all children, except R2 who did not rate the first free computer drawing made in session 5.
RESULTS In this section the results of four analyses are described. First, the pre-test measures of some psychomotoric skills necessary for operating a computer are described and compared with the norm samples from the test manuals. Second, the differences among the realism between the children’s hand and computer drawings are examined. Then the data are examined to see if they fit to a Rasch model, and the Drawing Score Category Scale as well as the children’s overall drawing abilities and the quality of the realism for the hand and the computer drawings are analysed using this Rasch measurement model. Finally, it is described how well the children did on a paper version compared to a computer version of one of the pre-tests, the Frostig Eye-Hand Coordination Subtest (part I), when they used a drawing tool on the computer which works like a normal pencil. Pretest
Table 1 presents the mean scores of all children on the three pretest measures having standardized age norms: The Sentence Span Subtest; the Raven Coloured Progressive Matrices and the Frostig Eye-Hand Coordination Subtest. When the mean scores are compared for the normal children between the classes it is seen for all three tests, that the Cloud class scores lower than the Sunshine class and the Kindergarten class, while the mean scores for the Sunshine class and the Kindergarten class are very close to each other. One way Anova analysis gives significant differences ( p < 0.01) between the mean scores for Raven and the Frostig (part I) and ( p c 0.05) for sentence span among Cloud versus Sunshine and Kindergarten classes. No mean score differences between Sunshine and Kindergarten classes is significant ( p > 0.05). When age norms are compared with the children’s living ages it is seen, that the children in this study, except for the special education children, got higher scores than comparable mean scores for children in the reference population. All of the children above 4 years of age except two had mastered the basic concepts of color, numbers to 8, conservation of number and senation. Among the youngest children in the Cloud class, only three of the children had acquired all these skills, the two whose participation was discontinid had not mastered any of these skills, while the rest had mastered some and were unsure of others of these skills. Interrater reliability
The six raters used the same 7 point rating scale to score all four drawings, hand as well as computer drawings. Correlations among scores for all four drawings among all pairs of the six raters are all significantly different from zero ( p
