Expressive Strategies in Drawing are Related to Age and Topic

ORI GIN AL PA PER

Expressive Strategies in Drawing are Related to Ageand Topic

Delphine Picard Æ Claire Brechet Æ Rene Baldy

Published online: 21 August 2007� Springer Science+Business Media, LLC 2007

Abstract In drawing, psychological mood can be denoted in a direct way (i.e., ‘‘liter-

ally’’) through facial expression cues (e.g., a frowning face denotes sadness in a direct

way), but it can also be connoted in an indirect way (i.e., ‘‘non-literally’’) through figu-

rative or non-figurative cues. This study examines how child and adult drawers selectively

use literal and non-literal expressive strategies in accordance with the nature of the topic

being depicted. In a between-subject design, 120 participants produced drawings of either a

person or a house, in one of three versions: baseline, happy, and sad. The results indicated

that drawers preferentially used literal expressive strategies for the person and non-literal

strategies for the house. There was an increasing tendency between 7 and 11 years of age

to express the drawn person’s mood non-literally in addition to literally. The positive

correlation obtained between representational and expressive drawing ability suggests that

enrichment of drawers’ graphic repertoire enhances their ability to draw expressively.

Implications for clinical and educational practitioners are discussed.

Keywords Expressive strategies � Children’s drawings � Psychological mood

Introduction

Drawing is a public instrument of representation (Freeman 1980) and a powerful nonverbal

tool for conveying emotions or psychological moods observed in humans or various other

topics (Gombrich 1972). To date, much of psychologists’ interest in children’s drawings

has been in the representational aspects of drawing behavior (e.g., Cox 1992; Freeman

1980). In contrast, children’s expressive drawings remain an often neglected area of

research, although there seems to be a recent growing interest in this topic (e.g., Burkitt

D. Picard (&) � C. Brechet � R. BaldyJE 2687 Department of Psychology, University of Montpellier III, Route de Mende 34 199,Montpellier, Francee-mail: [email protected]; [email protected]

123

J Nonverbal Behav (2007) 31:243–257DOI 10.1007/s10919-007-0035-5

et al. 2003a, b, 2004; Jolley et al. 2004). Indeed, the rigorous study of expressive drawings

both has a theoretical value (e.g., with respect to Parsons’ developmental stages of art

understanding, 1987), and has potential applications for clinical and educational practi-

tioners (e.g., with respect to the interpretation of children’s drawings used in some clinical

settings). The present study examines how child and adult drawers selectively use a set of

potential expressive strategies in accordance with the nature of the topic being depicted.

Psychological mood can be portrayed in a limited number of ways in a drawing

(Goodman 1968). It can be represented in a direct way (i.e., ‘‘literally’’) through facial

expression cues. Sadness, for instance, is portrayed literally through a frowning face.

Psychological mood can also be depicted in an indirect way (i.e., ‘‘non-literally’’). In this

case, the emotional significance of a drawing has to be inferred by the observer: what is

represented does not denote literally an emotion, but connotes, in a more subtle and con-

textualized way, an emotion. There are two main non-literal styles of communication, called

‘‘content’’ and ‘‘abstract’’ (see also Ives 1984; Jolley et al. 2004). In content style, psy-

chological mood is portrayed non-literally through figurative cues. For instance, rain and

clouds, a broken down house, or a wilted flower, are figurative cues that potentially connote

sadness (in addition to what they denote literally). In abstract style, psychological mood is

portrayed non-literally through non-figurative cues relative to the physical properties of

color, line, or size of a drawing. For instance, dark colors, drooping lines, or small size

objects are non-figurative cues that potentially convey a negative emotion, such as sadness.

Parsons (1987) has suggested that children at around age 7 are essentially interested in

the literal qualities of art (i.e., the objects represented and their physical characteristics).

Starting at the age of 10–12, children’s principal concern becomes the expressive qualities

of art (i.e., how a drawing or a painting can express mood through content and abstract

cues). Studies using picture-matching tasks (Carothers and Gardner 1979; Jolley and

Thomas 1995) confirmed that it was not until the age of 11 that children spontaneously

used expressed mood as a criterion for matching pictures. Below the age of 11, children

either matched pictures on a random basis, or according to subject matter.

Studies using drawing tasks have provided evidence of the development in children of

the expressive strategies used to represent mood in drawings. Ives (1984) asked 4- to 16-

year-olds to produce happy, sad, and angry drawings of a tree and of a non-representational

topic (a line). This author scored the drawings on their expressive qualities and found that

the children obtained good performance from age 7 on. The strategies used to express

mood varied as a function of age: the young children mainly depicted mood in an abstract

way (e.g., an undulating line was intended to express sadness) or used literal expressions

(e.g., a smiling face on the tree). Starting at age 7, the use of literal expressions decreased

greatly, while the use of content cues (e.g., a tree with fruit on a sunny day) slowly began

to develop. Interestingly, the strategies varied as a function of the topic: in tree drawings,

the participants divided their responses among strategies that relied on literal, content, and

abstract cues. In line drawings (non-representational topic), on the other hand, the par-

ticipants relied primarily on the use of abstract expression.

Further studies have quantified the use of both abstract and content cues in children’s

expressive drawings. Winston et al. (1995) asked 6-, 9-, and 12-year-olds to produce a

happy and a sad tree. The authors scored the expressive drawings for the presence of

content cues (e.g., personification, weather) and abstract cues (e.g., size, color). They found

an increase with age in the number of both content and abstract cues. Jolley et al. (2004)

showed that the use of content cues developed earlier than that of abstract cues. In their

study, children ages 4, 6, 9, and 12 were asked to produce a happy and a sad drawing of a

house and a tree. The initial development of the children’s expressive drawings was due to

244 J Nonverbal Behav (2007) 31:243–257

123

an increase in the use of content cues, with abstract cues (size, color, line) developing more

slowly.

However, different studies have demonstrated that, under very specific conditions,

children as young as age 4 are able to use abstract size and color cues symbolically. When

the task required drawing from a model or coloring a picture, and the children had no other

means of expressing mood than using size or color, then they performed well (Burkitt et al.

2003a, b). Burkitt et al. (2004) demonstrated that children could systematically alter their

use of color to fit with the affective characterization of a topic in a context where they drew

in the absence of a model. In this specific context, however, children were only permitted

to use one color for a given drawing. Additionally, Burkitt and Newell (2005) reported that

the color use could vary as a function of the type of human figure (adult vs. child) that 6–

8 years-old children were asked to portray.

The current literature suggests that from about age 7 children have a set of potential

strategies at their disposal in order to express psychological mood in their drawings. As

children grew older, they show an increasing tendency to make use of non-literal styles of

communication in their free drawings. However, children’s use of expressive cues may

vary greatly according to the nature of the task and to the nature of the topic being

depicted. The present study was designed to advance our understanding of children’s

expressive drawing abilities through examining the impact of topic, age, and gender on the

use of literal, and non-literal (content, abstract) expressive strategies in a free drawing

context.

No studies so far have been devoted to comparing children’s expressive abilities in

human and non-human representational topics, at least not in a free drawing context. In the

present study, we examined how children select expressive strategies to depict psycho-

logical mood (happy, sad) in drawings of the human figure and the house. Children

commonly make drawings of persons and houses, and the same set of expressive strategies

is potentially available, but not necessarily suitable, to portray mood in drawings of both

topics (e.g., a literal depiction of emotion would certainly be best suited to the human

topic). It is therefore important to assess strategy use in relation to human and non-human

stimuli, in order to evaluate how much flexible children of different ages are in their use of

expressive strategies.

The children observed in this study were 7 and 11 years old, two keys ages according to

Parsons’ (1987) stages of art understanding. We also observed an adult control group. Our

first hypothesis stated that the ability to draw expressively would increase between the ages

of 7 and 11, and would be easier to achieve for the person than for the house. A second

hypothesis assumed that the strategies used to depict mood should vary according to the

topic drawn: non-literal strategies involving abstract cues (e.g., size, color, line) and

content cues (e.g., weather cues, objects) should be observed more frequently for drawing a

house than for drawing a person. In contrast, literal strategies based on facial expression

cues should be more frequent for expressively depicting a person than a house. A third

hypothesis stated that between age 7 and age 11, the use of non-literal (content and/or

abstract) strategies should become more prevalent for the house drawing, and should add to

that of a literal strategy for the person drawing.

Person and house drawings have been the focus of psychological tests assessing various

aspects of children’s cognitive development (e.g., Barrouillet et al. 1994; Goodenough

1926; Harris 1963), and personality or emotional state (e.g., Buch 1948; Koppitz 1968;

Machover 1949). Thomas and Jolley (1998) suggested that drawings are inaccurate and

unreliable as personality or state assessments, but can be influenced by a child’s emotional

attitude toward the topic drawn. Assessing strategy use in relation to person and house

J Nonverbal Behav (2007) 31:243–257 245

123

drawings can provide further insights into the validity and reliability of the clinical use of

children’s free drawings of these topics (see also Burkitt et al. 2003a, b, 2004). Moreover,

it may be relevant to assess the relationship between children’s ability to portray mood in

person and house drawings, and their level of graphic development for these topics. With

age, children’s graphic repertoire increases (e.g., a higher level of detail is attained).

Improvements in representational drawing could enhance expressive drawings in children

if the two skills develop concurrently. We expected to find positive associations between

children’s scores on graphic development and their ability to portray mood in drawings.

In addition to age and topic, gender was a target factor in the present study. None of the

studies reviewed here mentioned variations in expressive drawing abilities relative to

gender. However, it was not possible to determine whether gender effects were not

reported because sex was not a significant factor, or because sex effects were not assessed.

There is abundant evidence that females are more accurate decoders of nonverbal affective

cues than males are (for a review see Hall 1978, 1984; McClure 2000). We therefore

explored whether sex was responsible for differences in the ability to draw expressively

and/or in the type of strategy that drawers used.

Method

Participants

One hundred twenty individuals volunteered to participate. They included 40 children age

7–8 years (M = 7 years 5 months, SD = 3 months, range = 6;11 to 7;11, 20 girls and

20 boys), 40 children age 11–12 years (M = 11 years 7 months, SD = 3 months,

range = 11;0 to 11;11, 20 girls and 20 boys), and 40 adults (M = 24 years, SD = 2 years,

range = 20 to 28, 20 women and 20 men). The children were from public schools in a

middle-class area of a southern French city. None of the children were ahead or behind in

school and none suffered from a known psychomotor drawing or handwriting disorder.

They were observed with written parental consent. The adults were university students who

had no specific education or experience in art or design.

Materials

The materials consisted of individual sheets of white paper (size = 21 · 29.7 cm). Three

graphic spaces (8 · 8 cm boxes) were drawn side-by-side on the paper, which was laid out

in the landscape direction. The leftmost space was devoted to the first drawing, the middle

space to the second drawing, and the rightmost space to the third drawing. We used a

regular pencil (HB) and nine colored pencils (black, brown, red, orange, pink, yellow, blue,

green, and purple).

Design and Procedure

The participants were observed individually, in a quiet room in their school (for the

children), or at the university (for the adults). They produced drawings of a person or a

house. We used a between-subjects design in which half the participants in each age group

(10 males, 10 females) were assigned to the person-drawing condition, and the other half

246 J Nonverbal Behav (2007) 31:243–257

123

to the house-drawing condition. This type of design was chosen to ensure that the par-

ticipants would not use the same strategy for the two topics simply because they would

replicate the procedure used for the first topic while drawing the second one.

Each participant first drew a person or a house (baseline drawing task) after hearing the

following instructions: ‘‘Here you have a sheet of paper and pencils. Can you draw a

person (a house) here?’’ Participants produced the drawing in the leftmost graphic space on

the paper. They used whatever pencils they wished. If they asked the experimenter which

pencil(s) they should take, the experimenter answered ‘‘It’s up to you.’’ There was no time

limit on the baseline drawing task. This task was intended to activate the usual procedure

for drawing a person (or a house) before the participants drew the objects in a happy or sad

mood (mood portraying task). The baseline drawing was used to derive graphic devel-

opment scores (see Scoring below). In addition, the baseline drawings allowed us to

objectively detect any changes (e.g., size increase) that a drawer made in order to depict

mood. Some authors (e.g., Ives 1984; Jolley et al. 2004; Winston et al. 1995) did not use

baseline drawings, and the children were directly introduced to the mood representation

task. By contrast, Burkitt et al. (2003a, b) always used a baseline drawing task to determine

whether children changed the size or the color of their drawings in the expressive con-

ditions. In the present study, we used baseline drawings to improve the coding accuracy of

the drawings obtained in the mood-portraying task.

Once a participant had completed the first (baseline) drawing, the experimenter asked

him/her to draw a happy and a sad version of the object (mood-portraying task). The happy

and sad versions were drawn in a counterbalanced order, i.e., in each age group, half the

participants drew the happy object and then the sad object, while the other half drew the

two versions in the reverse order. We used the following oral instructions: ‘‘Now imagine

that the person (house) is very happy. I want you to draw the person (house) so we can see

that he (it) is a very happy person (house)’’. We used the same type of instructions for the

sad version. Note that brief instructions are strong enough to induce the desired mood

toward the topics (see Burkitt and Barnett 2006). Participants produced the second and

third requested drawings successively on the sheet of paper, using the middle and rightmost

graphic spaces for each drawing, respectively. The baseline drawing was visible when a

participant drew the happy and sad versions of a topic. As in the baseline drawing task,

participants used the pencils freely in their drawings. There was no time limit on the mood-

portraying task. The full session lasted about 15 min per participant. We thanked and

complimented all participants on their drawings.

Scoring

Success at Depicting Emotion

Two judges worked independently to determine whether each drawing visibly expressed a

happy, neutral, or sad emotion, irrespective of how the mood was expressed. The judges

were unaware of the age of the drawers and the instructions given for each drawing. They

evaluated the drawings separately, one by one and in random order. One day later, the two

judges replicated their evaluations. Intra-judge consistency was nearly perfect

(M = 99.6%). Inter-judge agreement for each age group ranged between 88.3 and 90%

(M = 89.4%). Disagreements concerned cases for which one judge rated a drawing as

happy or sad whereas the second judge rated it as neutral. In these cases, we considered the

drawing to have a neutral expression.

J Nonverbal Behav (2007) 31:243–257 247

123

Strategies Used

We then looked solely at the expressive drawings (i.e., drawings of happy and sad objects

that the judges considered as such) to analyze the strategies used to express mood. A

preliminary analysis of the data showed that mood could be expressed through a literal

strategy (hereafter abbreviated L), a non-literal content strategy (NLC), a non-literal

abstract strategy (NLA), or some combination of the three. Table 1 lists the graphic cues

pertaining to each strategy. Two judges scored each individual expressive drawing for the

strategy(ies) used to depict mood. A drawer was considered to have employed a strategy

when he/she used at least one graphic cue from the strategy. The two judges performed the

coding together, using the individual baseline drawings as reference drawings. They settled

the very few cases of disagreement (< 5%) by discussion. Note that 95% of the baseline

drawings of the house and 65% of the baseline drawings of the person appeared emo-

tionally neutral to the judges. Some of the baseline person drawings, mostly produced by

the youngest children (age 7), visibly expressed happiness. The fact that some children

produced a smiling person in the baseline task did not, however, hinder their ability to

depict happiness when asked to do (e.g., the person displayed an accentuated smile).

Figures 1–3 provide illustrations of expressive drawings produced with each strategy.

Mood-depiction Score

In addition to scoring the drawings on the strategy used to express mood, we attributed

each drawing a mood-depiction score on a scale ranging from 0 to 6, with higher scores

indicating more complex mood depictions. Among the three major strategies (L, NLC, and

Table 1 Strategies used to depict mood, respective cues, and description

Strategy Cues Description

Literal (L) Facial expression Presence of open eyes and/or a smiling mouth (Happy).Presence of tears and/or a frowning mouth (Sad).

Non-literalcontent

Weather Presence of the sun (Happy). Presence of clouds and/or rain (Sad).

(NLC) Objects Presence of objects like people, flowers, gifts, etc. (Happy).Presence of objects like wilted flowers, spiders, a brokendown car, etc. (Sad).

Body position(person only)

Hands of the person drawn raised up (Happy). Hands of the persondrawn close to the body and/or shoulders down (Sad).

State of object(house only)

Duplication of features, like windows, doors, etc. (Happy).Reduction of the house to its core features (body-roof)and/or broken down house (Sad).

Non-literalabstract(NLA)

Size Increase of size from at least 1/6 of its height and/or surface area(Happy). Decrease of size from at least 1/6 of its height and/orsurface area (Sad).

Color Inclusion of at least three new colors (Happy). Disappearanceof color or sole use of black (Sad).

Line Presence of smooth and curved lines (Happy). Presenceof undulating lines (Sad).

Note: The description of the cues used is meaningful when an individual expressive drawing is comparedwith its individual baseline drawing

248 J Nonverbal Behav (2007) 31:243–257

123

NLA), strategy L is clearly the simplest expressive strategy: it depicts mood through

concrete elements in a direct manner. Strategy NLC is more complex than strategy L in

that it depicts mood through figurative components but in an indirect manner. Strategy

NLA is the most complex expressive strategy of these three because it depicts mood in a

purely abstract manner. Note that the complexity of a strategy is inversely related to the

strength of cues to emotion, and to how obvious the emotional inference is. The strategies

for portraying mood were ordered and scored as follows: no strategy (score 0), strategy L

(score 1), strategy NLC (score 2), strategy NLA (score 3), strategy L-NLC (score

1 + 2 = 3), strategy L-NLA (score 1 + 3 = 4), strategy NLC-NLA (score 2 + 3 = 5), and

strategy L-NLC-NLA (score 1 + 2 + 3 = 6). A score of 3 could imply the sole use of

strategy NLA as well as the combined used of strategies L and NLC. However, the

ambiguity of this score did not pose a problem insofar as participants almost never used

strategy NLA alone (see Table 2).

Graphic Development Scores

Finally, we scored the baseline drawings of the person on a 51-point scale, as described by

Goodenough (1926). The baseline drawings of the house were scored on a 22-point scale,

as described by Barrouillet et al. (1994). Two raters performed the coding of the baseline

drawings. They had specific training in the use of Goodenough and Barrouillet et al.’s

scales. Inter-rater reliability was 99%.

Fig. 1 Samples of expressive drawings produced with strategies L (literal), NLC (non-literal content), andNLA (non-literal abstract). Each expressive drawing is presented together with its individual baselinedrawing; sex and age group of the drawers are mentioned, and expressive cues are given in parentheses

J Nonverbal Behav (2007) 31:243–257 249

123

Fig. 2 Samples of expressive drawings produced with strategies L-NLC (literal and non-literal content),and L-NLA (literal and non-literal abstract). Each expressive drawing is presented together with itsindividual baseline drawing; sex and age group of the drawers are mentioned, and expressive cues are givenin parentheses

Fig. 3 Samples of expressive drawings produced with strategies NLC-NLA (non-literal content and non-literal abstract), and L-NLC-NLA (literal, non-literal content, and non-literal abstract). Each expressivedrawing is presented together with its individual baseline drawing; sex and age group of the drawers arementioned, and expressive cues are given in parentheses

250 J Nonverbal Behav (2007) 31:243–257

123

Analysis

The main dependent measures were success or failure at depicting the requested emotion in

the drawing, and the presence or absence of each strategy (and strategy combination) used

to depict mood. Each of these dependent measures was put into a 3 · 2 · 2 · 2 mixed

analysis of variance (ANOVA) with age (3), sex (2), and topic (2) as between-subject

factors and mood (2) as a within-subject factor. Although it may seem unconventional to

use parametric tests with binary data, some authors (e.g., Greer and Dunlap 1997; Lunney

1970) have shown that ANOVAs can be performed with such data under certain conditions

(e.g., the number of degrees of freedom of the error term must be above 40), which were

satisfied here. Two additional dependent measures were the mood-depiction scores

(maximum 6) and the graphic development scores for the person drawing (highest score

51) and for the house drawing (highest score 22). The relationships between mood-

depiction scores and graphic-development scores were assessed using Pearson’s product-

moment correlation. We used an alpha level of .05 for all statistical tests. Effects that did

not reach significance at p inferior or equal to .05 are not reported in the following section.

Results

Success at Depicting Emotion

Table 2 shows the mean percentage of expressive drawings by strategy and total expressive

drawings, by topic and age group. Our first hypothesis stated that the ability to draw

expressively would increase between the ages of 7 and 11, and would be easier to achieve

for the person than for the house. The results (see Table 2, bottom row) confirmed our

expectations. Expressive drawings were produced more frequently for the person (79.2%)

than for the house (51.7%). A main effect for topic was obtained, F (1, 108) = 19.80,

p < .01, with a medium effect size, gp2 = .15. A main effect for age was found, F (2,

108) = 2.86, p < .05, with a small effect size, gp2 = .05. Planned comparisons indicated that

the percentage of expressive drawings was higher at age 11 (71%) than at age 7 (55%)

Table 2 Mean percentage of expressive drawings by strategy(ies) and total expressive drawings, by topicand age group

Strategy(ies) Person House

Age 7 Age 11 Adult Mean Age 7 Age 11 Adult Mean

L 42.5 30 40 37.5 22.5 10 15 15.8

NLC – – – – 5 2.5 17.5 8.3

NLA – – – – – 5 – 1.7

L-NLC 10 32.5 25 22.5 2.5 - 2.5 1.7

L-NLA 7.5 10 2.5 6.7 2.5 5 2.5 3.3

NLC-NLA – – 5 1.7 10 35 15 20

L-NLC-NLA 5 12.5 15 10.8 2.5 – – 0.8

Total expressive 65 85 87.5 79.2 45 57.5 52.5 51.7

Note: L = literal, NLC = non-literal content, NLA = non-literal abstract, L-NLC = literal and non-literalcontent, L-NLA = literal and non-literal abstract, NLC-NLA = non-literal content and non-literal abstract,L-NLC-NLA = literal, non-literal content, and non-literal abstract

J Nonverbal Behav (2007) 31:243–257 251

123

(p < .05), but did not differ significantly between the 11-year-olds and the adults (70%)

(p > .05).

Strategies by Topic

Our second hypothesis stated that the strategies used to depict mood would vary according

to the topic drawn. In line with this hypothesis, the ANOVAs revealed significant effects of

the topic on the prevalence of strategies L, F (1, 108) = 9.69, p < .01, gp2 = .08, NLC, F (1,

108) = 9.78, p < .01, gp2 = .08, L-NLC, F (1, 108) = 25.22, p < .01, gp

2 = .19, NLC-NLA,

F (1, 108) = 23.41, p < .01, gp2 = .18, and L-NLC-NLA, F (1, 108) = 10.28, p < .01,

gp2 = .09. All of these main effects had a small or moderate effect size. The results (see

Table 2, Mean columns) showed that participants more frequently employed strategies L,

L-NLC, and L-NLC-NLA for drawing the person (L: 37.5%, L-NLC: 22.5%, L-NLC-

NLA: 10.8%) than the house (L: 15.8%, L-NLC: 1.7%, L-NLC-NLA: 0.8%). In contrast,

they used strategies NLC and NLC-NLA specifically to depict mood in the house drawing

(NLC: 8.3%, NLC-NLA: 20%), but not or did so very little in the person drawing (NLC:

0%, NLC-NLA: 1.7%). As a whole, these results fit in with our predictions: literal strat-

egies were more frequent for expressively depicting a person than a house, whereas non-

literal strategies were observed more frequently for drawing a house than a person.

Strategies by Age

Our third hypothesis assumed that between age 7 and 11, the use of non-literal (content

and/or abstract) strategies should become more prevalent for the house drawing, and

should add to that of a literal strategy for the person drawing. The results (see Table 2)

showed that the use of strategies NLC and NLC-NLA became more prevalent with

increasing age in the house drawing. In the person drawing, we noted that the use of a

literal strategy remained rather constant across ages, while the use of strategy L-NLC

increased with age. These observations were supported by significant age · topic inter-

action effects on the use of strategy NLC, F (2, 108) = 3.03, p < .05, gp2 = .05, strategy

NLC-NLA, F (2, 108) = 4.83, p < .01, gp2 = .08, and strategy L-NLC, F (2, 108) = 3.06,

p < .05, gp2 = .05. Post hoc analyses on these interactions (Tukey tests) revealed that

strategies NLC and NLC-NLA were age-related for the house drawing, the former being

more frequent in adults (17.5%) than in children (3.75%) and the latter being more frequent

at age 11 (35%) than at age 7 (10%). Strategy L-NLC was age-related for the person

drawing, being more frequent at age 11 (32.5%) than at age 7 (10%). These results are in

line with our expectations: for the house drawing, non-literal content strategies developed

with increasing age; for the person drawing, non-literal content strategies combined with

literal strategies from age 7 on.

Looking at exactly which content cues (weather, objects, body position, and state of the

object) participants used more and more frequently as they grew older to portray mood in

each topic, we found that, for the person drawing, modifications of body position essen-

tially accounted for the increase in strategy L-NLC between ages 7 and 11 (see also Fig. 2).

For the house drawing, changes in the state of the object mainly accounted for the increase

of strategy NLC between age 11 and adulthood (see also Fig. 1). For the abstract cues

(size, color, line), participants more frequently employed both size and color to depict

mood in their house drawing between the ages of 7 and 11, while they rarely used lines to

express mood.

252 J Nonverbal Behav (2007) 31:243–257

123

Relations between Mood-Depiction Scores and Graphic-Development Scores

Table 3 shows the mean scores on mood-depiction and graphic development obtained for

each age group and topic. Scores on graphic development fell in the normal range for both

topics. Both mood-depiction scores and graphic-development scores increased significantly

between age 7 and 11 (all ps < .01). We expected to find positive associations between

children’s scores on graphic development and their ability to portray mood in drawings.

For all ages and topics pooled, there was a significant, positive correlation between the

graphic-development scores and the mood-depiction scores for the happy drawings,

r = .28, F (1, 118) = 9.83, p < .01, as well as for the sad drawings, r = .23, F (1,

118) = 6.65, p < .05. As expected, higher levels of graphic development were associated

with more complex mood depictions. However, when the correlation analyses were con-

ducted separately for each topic, the relationships were only significant for the person

(r = .25 and .37, for the happy and sad drawings respectively, ps < .05), not for the house

(r = .17 and .11 for the happy and sad drawing respectively, ps > .05).

Sex Differences

Exploring whether sex was responsible for differences in the ability to draw expressively

and/or in the type of strategy that drawers used, we observed significant gender-related

differences in the use of two expressive strategies. A main effect for sex was found on the

use of strategies L, F (1, 108) = 4.64, p < .05, gp2 = .04, and L-NLC-NLA, F (1,

108) = 4.57, p < .05, gp2 = .04. These sex effects indicated that male participants used

strategy L less frequently than female participants did (males: M = 19.2% vs. females:

M = 34.2%), but they employed the combined strategy L-NLC-NLA more frequently than

females did (males: M = 9.2% vs. females: M = 2.5%).

We assessed whether the girls’ greater use of literal cues meant that they were providing

enriched depictions of emotional expressions, which potentially led to the production of

more effective drawings in female than in male drawers. Two adult judges (one male and

one female) were asked to sort the 156 expressive drawings into two numerously equiv-

alent groups that differed in their effectiveness in conveying an emotional expression (the

two drawing groups were called the ‘‘weakly expressive’’ and ‘‘strongly expressive’’

groups). The judges were blind to the sex of the drawer. Among the 156 drawings, 82 were

produced by females, and 74 by males. It was suspected that the sex of the drawers would

bias the distribution of the drawings: a larger proportion of girls’ drawings would be

included in the ‘‘strongly expressive’’ group in comparison with boys’ drawings. The

Table 3 Mean scores on mood depiction and graphic development, by topic and age group

Score Person House

Age 7 Age 11 Adult Age 7 Age 11 Adult

Mood depiction 1.32 2.42 2.40 1.15 2.25 1.42

(1.35) (1.76) (1.85) (1.67) (2.34) (1.73)

Graphic development 16.05 19.95 22.50 10.45 12.85 13.50

(6.08) (8.59) (8.13) (2.52) (1.57) (3.17)

Note: The maximum score on mood depiction was 6. The maximum scores on graphic development were 51for the person, and 22 for the house. Standard deviations are given in parentheses

J Nonverbal Behav (2007) 31:243–257 253

123

results showed that this was not the case. The distribution of male and female drawings in

the ‘‘strongly expressive’’ group (44 female drawings and 34 male drawings) and ‘‘weakly

expressive’’ group (38 female drawings and 40 male drawings) did not depart significantly

from a distribution that would be expected by chance, v2 (1, N = 156) = .93, p > .05.

Discussion

The present study was designed to examine the impact of topic, age, and gender on the use

of literal and non-literal (content, abstract) expressive strategies in a free drawing context.

Our main hypothesis assumed that the nature of the drawn topic (person vs. house) would

have a strong influence on the way the child and adult drawers select expressive strategies.

Age-related changes in the use of expressive strategies were also forecasted for each topic

separately. Additionally, gender effects were assessed.

Our study provides clear evidence of the impact of the topic being drawn on partici-

pants’ ability to draw expressively. Both children and adults were found to depict

psychological mood more easily in a human than in an inanimate object such as a house.

Importantly, our study provides clear evidence of the selective use of different expressive

strategies that depend on the drawn topic in free drawing conditions. As expected, drawers

preferentially used literal strategies (specific facial expressions) to depict psychological

mood in the person drawing but not for the house. Non-literal (content and abstract)

strategies were combined with literal cues in the expressive person drawing, but these two

strategies were never used alone for the human topic. By contrast, the sole use of a content

strategy and the combination of content and abstract strategies were confined to the

depiction of mood in the house drawing. Ives (1984) had already noted that the use of

expressive strategies varied according to whether the topic was representational (e.g., a

tree) or non-representational (e.g., a line). Our results suggest that this differential use of

expressive strategies also applies to representational topics, depending on whether they are

human or inanimate.

From a developmental perspective, and in line with the literature (e.g., Ives 1984; Jolley

et al. 2004; Winston et al. 1995), the present experiment confirms that the ability to draw

expressively develops with age, being higher at age 11 and in adults than at age 7. Inter-

estingly, the expressive strategies used by drawers were related to their age for each topic

separately. We observed an increasing tendency between the ages of 7 and 11 to express

mood through non-literal content cues in the person drawing rather than just literally. A

closer look at our data revealed that, in addition to relying on facial expressions to depict

mood at a literal level, the older children were inclined to use body position in their

drawings of the person to depict mood. Body positions and movements are relevant emotion

cues (Coulson 2004). It has been established that children are sensitive to these cues. For

instance, Boone and Cunningham (1998) showed that, between the ages of 4 and 8, children

demonstrated an increasing ability to decode the emotional significance of bodily move-

ments made by artistic dancers. The use of body position as a cue for emotion in a static

display (drawing) reflects high level representational and expressive abilities in children.

For the house drawing, we noted an increasing tendency between the ages of 7 and 11 to use

non-literal (content and abstract) cues. A closer look at our data indicated that increasing

age led to a greater reliance on the state of the house as a typical way of expressing mood on

this topic. Altering the state of the house (e.g., a broken down house to express sadness) is

indicative of high expressive abilities in drawers. Surprisingly, for the expressive house

drawing, we noted a greater sole use of content cues in adults as compared to children.

254 J Nonverbal Behav (2007) 31:243–257

123

Jolley and collaborators (2004) found that the development of expressive drawings

between 4 and 12 years of age was initially due to an increase in the use of content cues,

while the use of abstract cues developed more slowly. It may seem odd that the sole use of

a content strategy was much more characteristic of the drawing behavior of adult partic-

ipants than that of children. One possible explanation for this phenomenon is that adults,

unlike children, are much more economical in their drawing style, and hence prefer a

single (content) to a combined non-literal strategy (content and abstract) when drawing

expressively. The addition of abstract cues to content ones in a drawing would essentially

enhance the probability that an observer will decode the expressed mood unambiguously.

Another explanation is that our adult group failed to rely on the sole use of abstract cues to

convey emotion in their drawings of the house because this ability might be mastered only

by a specific part of the adult population, such as artists or expert drawers. Our control

group had no specific education or experience in art and design, and could therefore have

suffered from a lack of knowledge or mastery of abstract drawing techniques (see also Lin

and Thomas 2002).

We noted a relationship between drawers’ ability to depict mood on standard topics and

their level of graphic development on these topics. In our view, representational drawing

and expressive drawing develop concurrently, insofar as enrichment of the graphic rep-

ertoire participates in the ability to draw expressively. For instance, drawing a happy

person with a big smile and hands raised up on a sunny day requires the availability of

sufficiently elaborated facial expressions in the child’s graphic repertoire, as well as

mastery of the contour line technique to depict certain body positions (see Goodnow 1978).

The fact that the relationship was found to be significant here for the person—but not for

the house—could be related to the fact that children spontaneously depict emotion (in

general, happiness) more often in drawings of a person than in drawings of a house. As a

consequence, the baseline task and mood-portraying task were more alike for the person

than for the house.

Our study revealed interesting sex differences in the use of expressive strategies in

drawing behavior: compared to male drawers, female drawers used the literal strategy more

often and the combined literal-non-literal strategy less often. As a whole, however, the

females did not produce a greater number of expressive drawings than the males. We also

failed to find sharp qualitative differences in the expressive strength of the drawings

produced by the two sexes. We know that boys and girls exhibit differences in the ways

they depict the human figure (Cox 1992; Koppitz 1968; Sitton and Light 1992), and in the

sex of their human figures (Silver 1996). There is also abundant evidence that females,

including children in our age range, are more accurate decoders of nonverbal affective cues

than males are (Hall 1978, 1984; McClure 2000). In the area of expressive drawings,

however, our study is the first study that reports a gender-related preference in the sole use

of facial expression cues to depict psychological mood.

Before concluding, let us consider the implications our study has with respect to Par-

sons’ stages of art understanding, and for clinical and educational practitioners. From a

theoretical point of view, our findings globally fit with Parsons’ (1987) stages of art

understanding: between 7 and 11 years of age, children’s depictions of mood evolve from

using literal to using non-literal expressive strategies. However, the differential use of

expressive strategies with age was not as clear-cut as a stage-like model. It depended

greatly on the topic drawn. In our view, the development of our understanding about art

can be better accounted for by phases than stages (in the Piagetian sense of the term). In

evaluating Parsons’ suggestion of a stage-like development of art understanding, Lin

and Thomas (2002) drew the conclusion that child and adult responses to popular art are

J Nonverbal Behav (2007) 31:243–257 255

123

more consistent with a branching model than a stage-like model of development. These

authors found that individual experience greatly influenced the development of art

understanding.

From a practical standpoint, the findings of our study provide no empirical support for

the view that size and color can be interpreted as a relevant index of the drawer’s emotional

attitude toward a freely drawn object. In free drawing conditions, we found that drawers

rarely employed size and color as cues for emotion (see also Jolley et al. 2004). The

availability of other (literal and non-literal content) drawing strategies overrode the use of

different sizes and colors. Giving size or color an emotional significance in freehand

drawing is therefore irrelevant, if not erroneous (specifically when there is no available

baseline drawing). This does not mean, however, that size and color can never be used as a

relevant index of a child’s emotional attitude toward the object depicted. As Burkitt et al.

(2003a, b) demonstrated, size or color can reveal mood in standard drawings (e.g., person,

house) when the child draws from a simple pre-drawn standard model or completes a pre-

drawn model with one color pen. Thus, practitioners interested in measuring the drawer’s

emotional attitude toward a topic should stop using freehand drawing contexts, and take

advantage instead of a copying or coloring task.

To conclude, we have to point out some of the limitations of our study. First, the age

range studied here is too small to draw general conclusions about the development of

expressive drawing strategies from to childhood to adulthood, and our cross-sectional

design limits the generalization regarding age effects. Second, our study was confined to an

examination of expressive strategies across two topics, using a between-subjects design. A

within-subjects design would be appropriate as a follow-up study to better understand

individual differences and to assess whether participants switched strategies across the

topics. Finally, our sex findings call for future research on the use of facial expression cues

in female and male drawers. A relevant avenue for future studies would be to assess

differences between males’ and females’ ability to represent a set of basic emotions (e.g.,

happy, sad, angry, surprised, afraid, disgusted) in their free drawings of the human figure.

We would anticipate that females are more accurate drawers of facial cues for emotions

than males are.

References

Barrouillet, P., Fayol, M., & Chevrot, C. (1994). Le dessin d’une maison, construction d’une echelle dedeveloppement. L’Annee Psychologique, 94, 81–98.

Boone, R. T., & Cunningham, J. G. (1998). Children’s decoding of emotion in expressive body movement:The development of cue attunement. Developmental Psychology, 34, 1007–1016.

Buch, J. N. (1948). The HTP test. Journal of Clinical Psychology, 4, 151–159.Burkitt, E., & Barnett, N. (2006). The effects of brief and elaborate mood induction procedures on the size of

young children’s drawings. Educational Psychology, 26, 93–108.Burkitt, E., Barrett, M., & Davis, A. (2003a). The effect of affective characterization on the size of

children’s drawings. British Journal of Developmental Psychology, 21, 565–584.Burkitt, E., Barrett, M., & Davis, A. (2003b). Children’s colour choices for completing drawings of

affectively characterized topics. Journal of Child Psychology and Psychiatry, 44, 445–455.Burkitt, E., Barrett, M., & Davis, A. (2004). The effect of affective characterizations on the use of size

and colour in drawings produced by children in the absence of a model. Educational Psychology, 24,315–343.

Burkitt, E., & Newell, T. (2005). Effects of human figure type on children’s use of colour to depict sadnessand happiness. International Journal of Art Therapy, 10, 1–8.

Carothers, T., & Gardner, H. (1979). When children’s drawings become art: The emergence of aestheticproduction and perception. Developmental Psychology, 15, 570–580.

256 J Nonverbal Behav (2007) 31:243–257

123

Coulson, M. (2004). Attributing emotions to static body postures: Recognition accuracy, confusions, andviewpoint dependence. Journal of Nonverbal Behavior, 28, 117–139.

Cox, M. V. (1992). Children’s drawings. Harmondsworth: Penguin.Freeman, N. H. (1980). Strategies of representation in young children. London: Academic Press.Gombrich, E. H. (1972). The visual image. Scientific American, 227, 82–96.Goodenough, F. L. (1926). Measurement of intelligence by drawings. New York: World Book Co.Goodman, N. (1968). Languages of art. Indianapolis: Bobbs Merrill.Goodnow, J. J. (1978). Visible thinking: Cognitive aspects of change in drawings. Child Development, 49,

637–641.Greer, T., & Dunlap, W. P. (1997). Analysis of variance with ipsative measures. Psychological Methods, 2,

200–207.Hall, J. A. (1978). Gender effects in decoding nonverbal cues. Psychological Bulletin, 85, 845–857.Hall, J. A. (1984). Nonverbal sex differences: Communication accuracy and expressive style. Baltimore:

Johns Hopkins University Press.Harris, D. B. (1963). Children’s drawings as measures of intellectual maturity. New York: Harcourt, Brace

& World.Ives, S. W. (1984). The development of expressivity in drawing. British Journal of Educational Psychology,

54, 152–159.Jolley, R. P., & Thomas, G. V. (1995). Children’s sensitivity to metaphorical expression of mood in line

drawings. British Journal of Developmental Psychology, 13, 335–346.Jolley, R. P., Fenn, K., & Jones, L. (2004). The development of children’s expressive drawing. British

Journal of Developmental Psychology, 22, 545–567.Koppitz, E. M. (1968). Psychological evaluation of children’s human figure drawings. London: Grune and

Stratton.Lin, S. F., & Thomas, G. V. (2002). Development of understanding of popular graphic art: A study of

everyday aesthetics in children, adolescents, and young adults. International Journal of BehavioralDevelopment, 26, 278–287.

Lunney, G. H. (1970). Using analysis of variance with a dichotomous dependent variable: An empiricalstudy. Journal of Educational Measurement, 7, 239–263.

Machover, K. (1949). Personality projection in the drawings of the human figure. Springfield, IL: Thomas.McClure, E. B. (2000). A meta-analytic review of sex differences in facial expression processing and their

development in infants, children and adolescents. Psychological Bulletin, 126, 424–453.Parsons, M. J. (1987). How we understand art: A cognitive developmental account of aesthetic experience.

Cambridge: Cambridge University Press.Silver, R. A. (1996). Silver drawing test of cognition and emotion. Sarasota, FL: Albin Press.Sitton, R., & Light, P. (1992). Drawing to differentiate: Flexibility in young children’s human figure

drawings. British Journal of Developmental Psychology, 10, 25–33.Thomas, G. V., & Jolley, R. P. (1998). Drawing conclusions: A re-examination of empirical and conceptual

bases for psychological evaluation of children from their drawings. British Journal of Clinical Psy-chology, 37, 127–139.

Winston, A. S., Kenyon, B., Stewardson, J., & Lepine, T. (1995). Children’s sensitivity to expression ofemotion in drawings. Visual Art Research, 21, 1–15.

J Nonverbal Behav (2007) 31:243–257 257

123

Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.