Anda di halaman 1dari 11

Child Development, September/October 2010, Volume 81, Number 5, Pages 1356–1366

Cognitive Flexibility in Drawings of Bilingual Children


Esther Adi-Japha, Jennie Berberich-Artzi, and Afaf Libnawi
Bar-Ilan University

A. Karmiloff-Smith’s (1990) task of drawing a nonexistent object is considered to be a measure of cognitive flexi-
bility. The notion of earlier emergence of cognitive flexibility in bilingual children motivated the current
researchers to request 4- and 5-year-old English–Hebrew and Arabic–Hebrew bilingual children and their mono-
lingual peers to draw a flower and a house that do not exist (N = 80). Bilinguals exhibited a significantly higher
rate of interrepresentational flexibility in their drawings (e.g., ‘‘a giraffe flower,’’ ‘‘a chair-house,’’ found in 28 of
54 drawings), whereas the level of complex intrarepresentational change was similar across groups.
Interrepresentational drawings were previously reported only for children older than 7 years. The specific mech-
anisms by which bilinguals’ language experience may lead to interrepresentational flexibility are discussed.

As children develop, they learn to adjust to chang- representational aspects. The task allows for few
ing demands and priorities, to consider something distinct modification categories that involve the
from a fresh or different perspective, to switch relaxation of procedural and representational con-
between perspectives, and to ‘‘think outside the straints (Karmiloff-Smith, 1990; Picard & Vinter,
box.’’ These behaviors are referred to as ‘‘cognitive 2007). The current study compared the performance
flexibility’’ and build on inhibitory control mecha- of monolingual and bilingual children whose con-
nisms and working memory, which are together trol over specific inhibitory control mechanisms
regarded as core executive skills critical for cogni- was shown to differ (Carlson & Meltzoff, 2008;
tive development (Diamond, Barnett, Thomas, & Martin-Rhee & Bialystok, 2008).
Munro, 2007). Current research shows that early In the drawing of nonexistent objects task chil-
bilingualism leads to a developmental advantage in dren are asked to draw ‘‘an X that does not exist,’’
nonlinguistic tasks that necessitate cognitive flexi- such as an X they invent, that they have never seen
bility (e.g., Bialystok & Senman, 2004; Carlson & before, a strange X, an X with something funny or
Meltzoff, 2008). This advantage exists as early as odd about it (Karmiloff-Smith, 1990; Spensley &
age 4 (e.g., Bialystok, 1999; Bialystok & Martin, Taylor, 1999). Karmiloff-Smith (1990) showed that
2004) and is maintained in adulthood (Bialystok, in response to a request to draw a man ‘‘who does
Craik, & Ryan, 2006; Costa, Hernandez, & Sebas- not exist’’ (and the same for an animal and a
tian-Galles, 2008). house), 4- to 6-year-olds tended to make size,
For over half a century, drawing has been used shape, or deletion changes to elements within their
to study cognitive development (Arnheim, 1956; drawings, whereas 8- to 10-year-olds exchanged the
Freeman, 1995; Golomb, 1974, 1992; Luquet, 1927; position of elements, added extra elements from
Piaget & Inhelder, 1948 ⁄ 1967). Among the variety the same representation (e.g., a man with two
of drawing tasks that tap into executive skills, heads), or included cross-category insertions that
Karmiloff-Smith’s (1990) task of drawing nonexis- involved synthesizing two different representations
tent objects is considered to be a measure of cog- (e.g., a house with wings).
nitive flexibility (Spensley & Taylor, 1999), or Different studies that replicated Karmiloff-
representational flexibility (Picard & Vinter, 1999, Smith’s (1990) original findings (Berti & Freeman,
2006, 2007), present at that age. Children’s skilled 1997; Spensley & Taylor, 1999; Zhi, Thomas, & Rob-
graphic production implicates procedural as well as inson, 1997) verified a classification of these modifi-
cations into two main types of representational
flexibilities: An interrepresentational flexibility
The authors thank Professor Norman Freeman and Dr. Gil
Deisendruck for comments on earlier versions of this manuscript.
Correspondence concerning this article should be addressed to
Esther Adi-Japha, School of Education, Bar-Ilan University,  2010 The Authors
Ramat-Gan 52900, Israel. Electronic mail may be sent to Child Development  2010 Society for Research in Child Development, Inc.
japhae@mail.biu.ac.il. All rights reserved. 0009-3920/2010/8105-0003
Cognitive Flexibility in Drawings of Bilingual Children 1357

observed when a link between components from 4-year-old bilingual children verified their advan-
different categories appears in one drawing (as dis- tage in tasks that tap cognitive flexibility (Bialystok,
played by cross-category insertions), which 1999; Bialystok & Martin, 2004). In our pilot studies,
emerges at around 8 years, and an intrarepresenta- we saw that a man, a house, or an animal were too
tional flexibility connected to changes seen in the difficult for children aged around 4, whereas a
components of a given graphic representation (as flower was reasonably drawn by most children of
displayed by the other types of modifications), seen that age. At the age of 5 years we saw that although
already in 4- and 5-year-olds. a house was reasonably drawn, an animal was still
Although 4- and 5-year-olds typically did not too difficult to draw, and when these children were
spontaneously produce interrepresentational changes requested to draw a nonexistent man, some irrele-
or complex intrarepresentational changes (i.e., posi- vant responses (e.g., a dead man) emerged. In
tion or orientation change and same-category inser- Study 1, 4-year-old English–Hebrew bilinguals and
tion), they did so when explicitly requested: for their monolingual Hebrew-speaking peers were
example, when asked to draw an animal-man (i.e., therefore asked to draw a flower and a flower that
a man who is both an animal and a man) and a does not exist. In Study 2, 5-year-old Arabic–
house-man (Berti & Freeman, 1997); when asked to Hebrew bilinguals and their monolingual Hebrew-
draw a house with wings or a mushroom with speaking peers were also asked to draw a house
arms (Picard & Vinter, 2007); when given an earlier and a house that does not exist. The Arabic–
specific example, for example, a man with wings Hebrew bilinguals were tested as a second bilin-
instead of arms when the task was to ‘‘draw a man gual group and to control for the possibility that
with some parts being replaced with parts of an the different direction of writing (right-to-left in
animal’’ (Spensley & Taylor, 1999); or when shown both Arabic and Hebrew, as opposed to English)
a picture they could follow (i.e., a woman with two may contribute to the grapho-motor flexibility in
heads; Zhi et al., 1997). Furthermore, in Spensley the English–Hebrew bilinguals’ production. If
and Taylor’s (1999) experiment, 4- to 6-year-old indeed bilinguals are more successful in drawing
children succeeded in drawing examples that were nonexistent objects, this will comprise further sup-
specified verbally, but they were unable to utilize port for the notion that flexibility in the linguistic
this ability independently. These studies suggest a domain may facilitate flexibility in other nonlin-
developmental constraint: Children who are less guistic domains.
than 7 years of age cannot respond to the task
‘‘draw an X that does not exist’’ with complex rep-
resentational change without being primed by an
Study 1
explicit example that they can follow carefully, or
without being instructed to produce a specific type Method
of insertion.
Participants
Further experiments verified that complex intra-
representational as well as interrepresentational Twenty-five monolingual children (12 boys and
modifications were associated with procedural flex- 13 girls) and 25 bilingual children (13 boys and 12
ibility (Hollis & Low, 2005; Picard & Vinter, 1999, girls) participated in this study. The participants
2007). However, the notion that interrepresentation- were 4-year-olds (range = 42–60 months, M = 49.76,
al flexibility crucially requires the relaxation of a SD = 3.79 months and M = 51.12, SD = 4.75 months
constraint of independence between representations for the monolingual and bilingual children, respec-
may suggest different pathways to cognitive flexi- tively). The monolingual children attended a pri-
bility for complex intrarepresentational and inter- vate Hebrew kindergarten. The bilingual children
representational change in terms of the inhibitory attended a private English–Hebrew bilingual kin-
control processes implied. dergarten. The kindergartens were in the same
In this study, drawing of nonexistent objects neighborhood, which is located in an upper-
made by 4-year-old English–Hebrew and 5-year-old middle-class Israeli suburb, and operated for at
Arabic–Hebrew bilingual children were compared least 9 hr a day.
with those made by their Hebrew-speaking mono- For the bilingual group, kindergarten teacher
lingual peers. Representational innovations made reports ensured that the children participating in
by 4-year-olds are of special interest because at the study had acquired their English from home. In
this age children just master their representational this group, either both parents were native English
skills (Thomas & Silk, 1989). Previous studies of speakers who spoke both English and Hebrew at
1358 Adi-Japha, Berberich-Artzi, and Libnawi

home and in the community, or one parent spoke elements, insertion of new (same-category) ele-
English and the other spoke Hebrew. ments, change in element shape or size, whole-
shape changes, position or orientation changes, and
cross-category insertions. Cohen’s j coefficients for
Procedure
interrater agreement were above .9 (p < .001) in all
Children were observed individually in a quiet change categories. Disagreements were settled by
room at their kindergarten. Testing took place in discussion. As in the original study, the categories
1–2 sessions (session time did not exceed 20 min). were not mutually exclusive.
Bilingual children had an additional session for the The drawing was classified as showing change if
English vocabulary test. Summer drawing always the nonexistent flower violated ‘‘flowerness.’’ Simi-
preceded the nonexistent flower task (see next). The lar flower and nonexistent flower, a bigger and
experiment took place in 2 successive years with smaller flower, or a flower and a different flower
similar results. were classified as not showing change, even if the
verbal response indicated some difference (e.g., this
flower does not exist because ‘‘something shines on
Measures
it’’ or ‘‘because it is green, in make-belief’’).
Peabody Picture Vocabulary Test–III. The Peabody The drawing complexity scale. The Kellogg Scale
Picture Vocabulary Test–III (PPVT–III; Dunn & was used to score the complexity of the drawings
Dunn, 1997) is a test of receptive vocabulary. Bilin- of the summer, flower, and nonexistent flower.
gual children were tested using the English PPVT– This scale has seven figural prerepresentational
IIIA. Following the general procedures of Bialystok complexity levels (Kellogg, 1970; Golomb, 1992):
(1988), the monolingual and bilingual children were (1) scribbling; (2) placing scribbling marks in pat-
tested on an expert translation of the PPVT–IIIB terns; (3) scribble constellation that constitutes
developed at Bar-Ilan University (see also Bialy- emergent diagrams in the shapes of circles, ovals, tri-
stok, Shenfield, & Codd, 2000). Different forms of angles, rectangles, or crosses; (4) six outlined fig-
the test were used such that there was no overlap ures comprise the stage of diagrams—the rectangle,
in the specific vocabulary tested. Testing continued the oval, the triangle, the Greek cross, the diagonal
until the child made 8 errors out of a block of 12 tri- cross, and the odd shape that is closed but irregu-
als. Raw scores are reported. lar; (5) shapes that incorporate two diagrams are
The visual reception test. This test was taken from termed combines; (6) shapes that incorporate three
the Illinois Test of Psycholinguistic Abilities (ITPA; or more diagrams are termed aggregates; and (7)
Kirk, Mccarthy, & Kirk, 1968) and measures the complex graphic formulas joined into figures that
ability to infer meaning from pictures. During the are devoid of pictorial intention or likeness to an
test, the child is presented with a picture for 3 s. object. Three more levels were added because the
On the following page, the child is requested to tasks used in this study explicitly requested draw-
identify one of four pictures, which resembles the ings that resemble a known object (representations
image on the previous page. corresponding to summer or flower) that incorpo-
Drawing summer. The children were asked to rate a relatively simple scheme (at least at the
draw the summer. This drawing was used to test complexity of a combine, as in item 5). These lev-
their free drawing skills. Drawing complexity was els include: (8) a recognizable figure composed of
measured using the Kellogg Scale (see next). two line objects (e.g., in the flower drawing a cen-
‘‘Draw a flower that does not exist’’ test. The chil- ter and petals; a stalk and an outline of a flower or
dren were first asked to draw a flower. After com- circle); (9) a recognizable figure composed of three
pleting their drawing, they were requested to draw line objects (e.g., in the flower drawing a stalk, a
‘‘a flower that does not exist.’’ Several phrasings center, and petals); and (10) a recognizable figure
were used to enable the children to understand the that includes complex graphic formulas. For the
task: A flower you invent, you have never seen summer drawing, the most complex element in the
before, a strange flower, with something fun- drawing was scored. The two independent judges
ny ⁄ odd, something make-believe, pretend. After who scored the drawings for the change categories
drawing the nonexistent flower, the children were also scored the drawings for complexity. The
asked to verbalize why such a flower does not exist. scores ranged between 2 and 10. Interjudge
Following the procedure developed by Karmiloff- agreement was, on average, .96. Cohen’s j coeffi-
Smith (1990), two independent judges scored the cients for interrater agreement on a categorical
categories of changes as no change, deletion of scale were, on average .73 (p < .001), and weighted
Cognitive Flexibility in Drawings of Bilingual Children 1359

j coefficients for an ordinal scale were, on average —for deletion, v2(1) = 6.35, p < .05, and for cross-
.92 (p < .001). Kellogg Scale scores of the three category insertions, v2(1) = 11.53, p < .001—
drawing tasks were significantly correlated, r(50) > although deletion results should be interpreted
.75, p < .001, and significantly correlated with age with some caution given that multiple chi-square
r(50) > .51, p < .001. tests were performed without making an adjust-
ment in the alpha level. The 13 cross-category inser-
tions made by the bilingual children consisted
Results and Discussion
of modified flowers that included human features
Monolingual and bilingual children performed (7 children) and those that included animal features
similarly on the PPVT–III test, and had similar (6 children).
visual reception and drawing complexity scores As can be seen from the figures, children’s verbal
(Table 1). Furthermore, the bilingual group had response helped explain the drawings’ content
comparable language skills in Hebrew and in Eng- (Adi-Japha, Levin, & Solomon, 1998). The verbal
lish, t(24) = 1.73, ns. Although similar language description matched production in 43 of the 50
skills in monolingual and bilingual young children children who participated in the study (22 monol-
are not common, similar findings were reported in inguals and 21 bilinguals). One child did not
other studies (e.g., Davidson, Jergovic, Imami, & respond. Overall, the drawing descriptions of 13
Theodos, 1997). bilingual versus 2 monolingual children were of a
As in the original study, we used six change cat- cross-category type (e.g., Figure 1a) or reflected a
egories (see Method), with a slight difference: cross-category insertion (e.g., Figure 1b), v2(1) =
Changes in the position of flower elements did not 11.53, p < .001, whereas the response of 12 monolin-
appear. Change categories were not mutually exclu- gual versus 3 bilingual children reflected a deletion,
sive (Karmiloff-Smith, 1990). Drawings were classi- v2(1) = 7.71, p < .01.
fied as having 0–4 category changes with most About half of the monolingual children but only
children (45 ⁄ 50) producing 0–2 changes (Picard & one fourth of the bilinguals exhibited deletion as a
Vinter, 2007). significant change. This difference was reflected in
Table 2 presents the number of children in each the drawings’ complexity scores. A comparison of
of the ‘‘object that does not exist’’ categories of the Kellogg Scale scores revealed that among the
change (Karmiloff-Smith, 1990). Monolinguals’ monolinguals, the regular flower was drawn with
most frequent change was deletion (Figure 1), a higher level of complexity (Wilcoxon signed
whereas bilinguals’ most frequent change was rank test, Z = 2.12, p < .04), whereas no such dif-
cross-category insertions (Figure 2). Differences ference was observed for the bilinguals (Z = 1.52,
between groups in these categories were significant p > .1).

Table 1
Language and Drawing Skills in Monolingual and Bilingual Children

Study 1 Study 2

Monolinguals Bilinguals Monolinguals Bilinguals

M SD M SD t M SD M SD t

PPVT–IIIB (Hebrew) 58.16 13.48 59.48 16.51 0.31 73.60 25.09 50.13 14.72 3.14**
PPVT–IIIA 50.80 16.16 44.93 16.29
Visual reception 12.83 4.11 12.64 4.31 0.16 14.63 5.92 15.53 6.09 0.18
Digit span — — — — 6.60 3.08 6.73 1.75 0.15
Drawing complexity scores
Drawing summer 6.96 1.66 7.14 1.87 1.26 — — — —
Regular flower 6.86 2.62 7.58 2.54 0.78 7.67 2.06 8.36 1.60 1.00
Nonexistent flower 6.24 2.32 7.02 2.16 1.29 6.73 3.26 7.79 2.32 0.99
Regular house — — — — 7.67 2.67 8.07 1.97 0.52
Nonexistent house — — — — 7.73 1.83 7.21 2.51 0.63

Note. PPVT = Peabody Picture Vocabulary Test.


**p < .01.
1360 Adi-Japha, Berberich-Artzi, and Libnawi

Table 2
Production of the Strange Flower Task of 4-Year-Old English–Hebrew Bilinguals and Hebrew Monolinguals

Study 1 Study 2

Flower Flower House

Mono. Bi. Mono. Bi. Mono. Bi.


n = 25 n = 25 v2 n = 15 n = 15 n = 15 n = 14 Z

Succeeded 16 21 2.60 11 13 9 12 1.49


Deletion 13 6 6.35* 5 4 5 5 0.48
Insertion 2 2 0.00 1 2 0 1 0.60
Element’s shape 6 2 2.38 3 3 3 4 0.23
Whole shape 2 6 2.38 3 1 3 5 0.55
Orientation ⁄ position 2 2 0.00 0 0 2 2 0.00
Cross-category insertion 2 13 11.53*** 2 7 3 8 2.96**

Note. In Study 2, the Wilcoxon test was used; Z statistics were computed for 29 of 30 children who had data on both drawings.
Mono. = monolingual children; Bi. = bilingual children.
*p < .05. **p < .01. ***p < .001.

(a) (b) (c)

Girl 3;11 Girl 3;8 Girl 4;2

(d) (e) (f)

Boy 4;0 Girl 4;1 Boy 4;1

(g) (h) (i)

Girl 4;8 Boy 4;0 Girl 4;2

Figure 1. Deletion drawings by monolingual children.


Note. The child’s verbal response and (in parentheses) the Kellogg Scale scores for regular and strange flowers are presented for each
drawing. (a) ‘‘It’s missing leaves’’ (9; 6). (b) ‘‘This is not real a flower because it has only one petal’’ (7; 4.5). (c) Didn’t answer (9; 6).
(d) ‘‘This is not real a flower because it doesn’t have a stalk’’ (4.5; 4). (e) ‘‘This is not real a flower because it doesn’t have roots’’ (9; 8).
(f) ‘‘This is not real a flower; it doesn’t have a stalk’’ (9; 8). (g) ‘‘This is not real a flower because it doesn’t have the center’’ (8; 8).
(h) ‘‘This is a flower in the wind, it doesn’t have leaves’’ (8; 8). (i) ‘‘A flower which is a heart.’’ Deletion (of stalk) is a subcategory in
this example (9; 9).

The findings of Study 1 indicate that the 4-year- exhibited significantly more changes that involved
old English–Hebrew bilingual children and their deletion of elements within their drawing; bilingual
monolingual Hebrew-speaking peers were compa- children inserted significantly more features of
rable on a set of background measures. In spite other representational categories in their drawing
of this similarity, the children produced differ- of a flower. The results of Study 1 support the pos-
ent types of modifications: Monolingual children sibility that 4-year-old bilingual children display a
Cognitive Flexibility in Drawings of Bilingual Children 1361

(a) (b) (c)

Girl 4;6 Boy 4;2 Boy 5;0

(e) (F)
(d)

Girl 4;4 Boy 4;2 Boy 3;9

(g) (h) (i)

Boy 4;1 Boy 3;8 Girl 4;7

Figure 2. Cross-category insertion drawings of bilingual (English–Hebrew) children.


Note. The child’s verbal response and (in parentheses) the Kellogg Scale scores for regular and strange flowers are presented for each
drawing. Responses in italics were originally given in English. (a) This is a giraffe flower (9; 9.5). (b) This is a flower with a tail (9; 9). (c)
This is a camel flower (8; 8.5). (d) This is a lion flower with lots of hair. It has a lot of tails and shoes (9; 10). (e) This is a silly flower with arms
and legs (6; 4). (f) ‘‘This is not a real flower because it has arms and legs’’ (9; 6). (g) This flower has teeth; it doesn’t want to brush them’’
(8.5; 9). (h) ‘‘This is a flower that knows how to smile’’ (6; 6). (i) ‘‘This is a flower with hands’’ (9; 9).

different pattern of changes than their monolingual families, and had parents who had higher educa-
peers, and that this difference is related to the lar- tion experiences. The children attended Hebrew-
ger portion of cross-category insertions seen in their speaking kindergartens for at least 9 hr a day.
drawings. The bilingual children came from Muslim fami-
Hebrew and English differ in direction of writing: lies. Bilingual participants attended a Hebrew-
Hebrew is (read and) written from right to left. This speaking kindergarten for not less than 2 years, at
may lead to greater graphic flexibility by English– least from their 2nd year of life. These children
Hebrew bilinguals. To control for this possibility, in came from an Arab town and attended a kindergar-
Study 2 Arabic–Hebrew bilingual children were ten in one of the nearby kibbutzim. Before their
tested as a second bilingual group that share writing child entered kindergarten, the parents spoke with
characteristics, including direction of writing, with their children in Arabic. These children can there-
their monolingual Hebrew-speaking peers. fore be classified as early sequential bilinguals (Kan
& Kohnert, 2008). After their child entered kinder-
garten, at least one parent (usually the mother)
spoke to the child in Hebrew. The parents’ Hebrew
Study 2
level was sufficiently high to do so, partly because
Method of their experiences in higher education, where the
language of instruction is Hebrew, and experience
Participants
at their work place.
Fifteen monolingual (6 boys and 9 girls) and 15
bilingual children (6 boys and 9 girls) participated
Procedure
in this study. The participants were 5-year-olds
(range = 42–74 months, M = 61.00, SD = 9.47 and All children were observed individually in a
M = 61.87, SD = 11.14 months for the monolingual quiet room at their kindergarten. Testing took place
and bilingual children, respectively). The children in 2–3 sessions (session time did not exceed
participating in the study came from centrally 20 min). Arabic and Hebrew vocabulary tests were
located areas in Israel, came from middle-class administered in separate sessions.
1362 Adi-Japha, Berberich-Artzi, and Libnawi

that preschool bilingual children who first experi-


Measures
ence one language, but early on experience the sec-
The PPVT–IIIA was translated into Arabic by an ond language, may have similar vocabularies in
expert. The drawing summer assessment was not both languages (Kan & Kohnert, 2008).
used (as a result of a ceiling effect), and a forward Six change categories were used for both the
digit span test was added to ensure compatibility flower and the house drawings, with a slight differ-
of groups. ence from Study 1: A change in the position or ori-
Forward digit span. This test was used as an entation of elements was exhibited only in
assessment of short-term memory. The children nonexistent houses. Drawings were classified as
were presented with a random string of digits and having 0–4 category changes with most children
were required to repeat the string back in the same (28 ⁄ 30 and 23 ⁄ 30 for the flower and house, respec-
order. The strings began with two digits and tively) producing 0–2 changes (Picard & Vinter,
increased by one digit every two trials. One point 2007). Overall, the children responded similarly to
was awarded for every correctly reproduced the nonexistent flower and house tasks: McNemar
sequence, producing the possibility for two points tests applied to each of the change categories
for each string length. Testing ended when the revealed consistent modifications (p > .13); Wilco-
child incorrectly reproduced both sequences at a xon tests showed that the overall number of change
given string length. categories (Z = .98, ns), as well as drawing com-
‘‘Draw a flower ⁄ house that does not exist’’ test. The plexities (Z = .99, 1.06, ns, for the regular and non-
procedure detailed in Study 1 was used. The same existent objects, respectively) were similar.
phrasing was used to describe a flower and house Table 2 presents the number of children in each
that do not exist. The flower and house tasks were of the change categories (Karmiloff-Smith, 1990).
administered with 1–2 weeks separation, half of the Compared with Study 1, children’s modifications
children began with a house, and half with a were distributed more evenly across the modifica-
flower. As a result of the absence of one child, only tion categories. Nevertheless, bilingual children
14 bilingual children completed the house drawing. produced significantly more cross-category inser-
Cohen’s j coefficients for interrater agreement were tions in the nonexistent flower as well as in the
above .92 (p < .001) in all change categories. nonexistent house drawing (Figure 3). Eleven bilin-
The drawing complexity scale. The complexity gual children produced at least one cross-category
scoring procedure detailed in Study 1 was used drawing compared with only four monolingual
except for Levels 8 and 9. Level 8 was applied children, v2(1) = 6.53, p < .02. In addition to human
when the house included either a door or window. (5 ⁄ 15 drawings) and animal (3 ⁄ 15 drawings) fea-
Level 9 was applied when both appeared. Scores tures added to the flower and house, in Study 2
ranged between 2 and 10. Interjudge agreement bilingual children replaced the flower (3 ⁄ 15 draw-
was, on average, .94. Cohen’s j coefficients for ings) and house (4 ⁄ 15 drawings) with objects (e.g.,
interrater agreement on a categorical scale were, on a flower-door, a ball-house). Except for 2 children
average .76 (p < .001), and weighted j coefficients (1 monolingual and 1 bilingual) who did not pro-
for an ordinal scale were, on average .93 (p < .001). vide a verbal response, children’s verbal responses
Kellogg Scale scores of drawing tasks were matched drawn modifications, and differences
significantly correlated, r(28) = .68, p < .001, and between the two groups in cross-category verbal
significantly correlated with age r(29,30) > .52, responses paralleled differences in cross-category
p < .001. drawn modifications.
The findings of Study 2 indicate that the 5-year-
old Arabic–Hebrew bilingual children and their
Results and Discussion
monolingual Hebrew-speaking peers were compa-
Monolingual and bilingual children performed rable on a set of background measures but differed
similarly on the forward digit span test, and had in their language skills. In spite of this difference,
similar visual reception and drawing complexity and similar to the findings of Study 1, bilinguals
scores (Table 1). However, monolinguals obtained exhibited significantly more cross-category inser-
significantly higher scores than their bilingual peers tions in their drawings of nonexistent objects than
on the PPVT–III, t(28)=3.14, p < .01. A paired- their monolinguals peers. In Study 2, the children
sample t test indicated that the bilingual group had were requested to draw two nonexistent objects: a
comparable language skills in Arabic and in flower and a house. The results showed a consistent
Hebrew, t(14) = 1.49, ns. Other studies have shown pattern of production and modifications in the
Cognitive Flexibility in Drawings of Bilingual Children 1363

(a) (b) (c)

Boy 5;11 Girl 4;0 Girl 5;10

(d) (e) (f)

Boy 3;6 Boy 5;8 Boy 6;2

(g) (h) (i)

Girl 4;6 Boy 5;8 Boy 4;10

Figure 3. Cross-category insertion drawings of bilingual (Arabic–Hebrew) children.


Note. The child’s verbal response and (in parentheses) the Kellogg Scale scores for regular and strange flowers are presented for each
drawing. Responses in italics were originally given in Arabic. (a) ‘‘This is a tree flower’’ (9; 9). (b) This flower is a door (9; 9). (c) ‘‘A
butterfly flower’’ (9; 8.5). (d) ‘‘This is a kite flower’’ (5; 6). (e) This is a girl flower (9; 9). (f) ‘‘A robot house’’ (9; 10). (g) ‘‘This is a chair
house’’ (9; 8). (h) This is a ball which is a house (10; 9). (i) ‘‘This is an orange flower’’ (8; 8).

drawings of these objects. This may be related to bilinguals’ language experiences may advance the
the fact that both these objects are frequently pro- former, but not on the latter type of representa-
duced by 5-year-olds. Furthermore, the rate of dele- tional flexibility.
tions of object features was similar across the Two samples of bilingual children participated
groups, mainly because of a reduction in the rate of in the study: 4-year-old English–Hebrew simulta-
deletions produced by monolingual children. This neous bilingual children and 5-year-old Arabic–
reduction, albeit nonsignificant (Z = 1.93, in the Hebrew early sequential bilingual children. As in
comparison of two nonexistent flower drawing previous studies, the atypical performance of the
tasks made by monolinguals in Studies 1 and 2), bilingual children was apparent already at age 4.
may be related to the older age of the participants Differences between groups were qualitative as
in Study 2. well as quantitative and were probably not the
result of the specific language groups compared, or
specific object used in the task. Although the find-
ing that 4- to 6-year-olds do not spontaneously pro-
General Discussion
duce cross-category insertions was replicated in
Studies of representational flexibility consistently several studies, and the advantage of 4-year-old as
replicate the finding of a developmental con- well as the 5-year-old bilinguals in introducing
straint: Children less than 7 years do not respond cross-category insertions was robust, future studies
to the task ‘‘draw an X that does not exist’’ with a are needed to replicate the current results in a more
complex intrarepresentational or interrepresenta- heterogeneous sample of language backgrounds.
tional innovation. The findings of this study indi- Why would bilinguals produce interrepresenta-
cate that 4- and 5-year-old bilingual children can tional innovations earlier than their monolingual
produce a variety of interrepresentational modifi- peers? Previous studies have shown that bilinguals
cations (i.e., cross-category insertions) in response exhibit better comprehension of figures or objects
to the experimenter’s request to draw a flower (a that comprise interrepresentational cues (Bialystok
house) that does not exist. These results under- & Martin, 2004; Bialystok & Senman, 2004;
score the difference between interrepresentational Bialystok & Shapero, 2005). For example, bilinguals
and intrarepresentational changes and suggest that have demonstrated superior performance on an
1364 Adi-Japha, Berberich-Artzi, and Libnawi

ambiguous figure task when required to identify demands) but not in the latter (impulse control;
the alternative image in a reversible figure (e.g., rat- Carlson & Meltzoff, 2008; Martin-Rhee & Bialystok,
man). It is possible that these children also have a 2008). This may explain why bilinguals excelled
superior ability to create (or at least imagine) fig- only in the interrepresentational category of
ures that convey two representational meanings change. Drawing is a complex skill involving bio-
and that may enhance the frequency of drawings mechanical, fine-motor, perceptual, cognitive, and
depicting interrepresentational flexibility. The other developmental competencies that do not vary
results of several studies support the notion that by language (Braswell & Rosengren, 2008; Freeman
bilinguals simultaneously keep their two language & Adi-Japha, 2008; Morra, 2005; van Sommers,
systems active and switch between them (e.g., Her- 1984). It is possible that these factors minimize the
nandez, Dapretto, Mazziotta, & Bookheimer, 2001; differences in representational flexibility that are
van Heuven, Schriefers, Dijkstra, & Hagoort, 2008). expected in bilinguals.
Bilingual children spontaneously switch from one Previous research has shown that bilingual chil-
language to another in case of a lexical gap, creat- dren excel in cognitive flexibility tasks requiring a
ing mixed-language sentences (e.g., Lindholm & response to bivalent perceptual cues (e.g., Bialystok
Padilla, 1978; Nicoladis & Secco, 2000). An early & Martin, 2004; Carlson & Meltzoff, 2008). This
experience in mixing representational systems may study extends the scope of these results to an imag-
as well enhance the frequency of interrepresenta- inative cognitive flexibility task from a symbolic
tional modifications. production domain (i.e., drawing), enabling an
Several studies suggest that advantages in execu- unlimited set of responses. Our findings suggest
tive functions in bilinguals are specific rather than that the cognitive advantage of bilinguals is limited
general (Carlson & Meltzoff, 2008; Martin-Rhee & rather than general: Bilinguals produced more
Bialystok, 2008). In line with these findings, the interrepresentational (but not intrarepresentational)
results of Studies 1 and 2 suggest that bilinguals modifications to their drawings of nonexistent
did not demonstrate an overall flexibility advan- object. The findings of this study provide further
tage. Such an advantage would have been mani- support for the notion that flexibility in the linguis-
fested if a higher rate of position and orientation tic domain may facilitate flexibility in other nonlin-
changes, same-category insertions, and cross- guistic domains demonstrating the effects of
category insertions would have been found in their experience (and in particular bilingualism) on chil-
drawings. Rather, bilinguals’ performance was dif- dren’s development (Emmorey, Luk, Pyers, &
ferent from that of their monolingual peers in just Bialystok, 2008; Karmiloff-Smith, 2009). Representa-
one (of three) categories: Bilingual children pro- tional as well as procedural aspects of grapho-
duced more cross-category insertions. motor skills (e.g., drawing and writing, and in
Although drawing order was not systematically general any motor skill) improve with experience
recorded, it was noticed that most bilingual chil- (e.g., Adi-Japha & Freeman, 2000, 2001; Adi-Japha,
dren either added the cross-category item ⁄ s at the Karni, Parnes, Loewenschuss, & Vakil, 2008;
end of the drawing procedure (Figures 2b,d,g–i and Adi-Japha et al., 1998; Adi-Japha et al., 2007). Sym-
3d,f) or from first stroke made a new, blended, fig- bolic representation consists of other domains such
ure (Figure 2a,c,f and most drawings in Figure 3). as play or sculpture. Experience in symbolic
For that to occur, children need to develop inhibi- domains is considered essential for cognitive devel-
tory control mechanisms allowing them to manage opment and the nurturing of executive skills (Dia-
two conflicting demands (e.g., that the depicted fig- mond et al., 2007). Symbolic domains may provide
ure would be recognizable both as a flower and as a bilingual children with a natural environment for
giraffe). The production of position and orientation presenting and enhancing their relative advantage.
changes and same-category insertions typically
involves graphic modifications that are executed in
the middle of the drawing sequence (the child
begins to draw, executes the graphic modification, References
and then goes on to finish the drawing as usual; Adi-Japha, E., & Freeman, N. H. (2000). Regulation and
Hollis & Low, 2005; Zhi et al., 1997). Graphic inter- division of labor between cognitive systems controlling
ruptions in the middle of the drawing procedure action. Cognition, 76, 1–11.
necessitate impulse control. Recent studies suggest Adi-Japha, E., & Freeman, N. H. (2001). Development of
that bilinguals have an advantage in the former differentiation between writing and drawing systems.
inhibitory control mechanism (managing conflicting Developmental Psychology, 37, 101–114.
Cognitive Flexibility in Drawings of Bilingual Children 1365

Adi-Japha, E., Karni, A., Parnes, A., Loewenschuss, I., & Diamond, A., Barnett, W. S., Thomas, J., & Munro, S.
Vakil, E. (2008). A shift in task routines during the (2007). Preschool program improves cognitive control.
learning of a motor skill: Group averaged data may Science, 318, 1387–1388.
mask critical phases in individuals’ acquisition of Dunn, L. M., & Dunn, L. M. (1997). Peabody Picture Vocab-
skilled performance. Journal of Experimental Psychology: ulary Test–III. Circle Pines, MN: American Guidance
Learning Memory and Cognition, 34, 1544–1551. Service.
Adi-Japha, E., Landau, Y., Frenkel, L., Teicher, M., Gross- Emmorey, K., Luk, G., Pyers, J. E., & Bialystok, E. (2008).
Tzur, V., & Shalev, R. S. (2007). ADHD and dysgraphia: The source of enhanced cognitive control in bilinguals:
Underlying mechanisms. Cortex, 43, 700–709. Evidence from bimodal bilinguals. Psychological Science,
Adi-Japha, E., Levin, I., & Solomon, S. (1998). Emergence 19, 1201–1206.
of representation in drawing: The relation between Freeman, N. H. (1995). The emergence of a framework
kinematic and referential aspects. Cognitive Develop- theory of pictorial reasoning. In C. Lange-Kuttner & G.
ment, 13, 23–49. V. Thomas (Eds.), Drawing and looking: Theoretical
Arnheim, R. (1956). Art and visual perception: A psychology approaches to pictorial representation in children (pp. 135–
of the creative eye. London: Faber & Faber. 146). London: Harvester.
Berti, A. E., & Freeman, N. H. (1997). Representational Freeman, N. H., & Adi-Japha, E. (2008). Pictorial inten-
change in resources for pictorial innovations: A three tion, action and interpretation. In V. Lange-Küttner
component analysis. Cognitive Development, 12, 501– & A. Vinter (Eds.), Drawing and the non-verbal mind
522. (pp. 104–120). Cambridge, UK: Cambridge University
Bialystok, E. (1988). Levels of bilingualism and levels of lin- Press.
guistic awareness. Developmental Psychology, 24, 560–567. Golomb, C. (1974). Young children’s sculpture and drawing.
Bialystok, E. (1999). Cognitive complexity and attentional Cambridge, MA: Harvard University.
control in the bilingual mind. Child Development, 70, Golomb, C. (1992). The child’s creation of a pictorial world.
636–644. Berkeley: University of California Press.
Bialystok, E., Craik, F. I. M., & Ryan, J. (2006). Executive Hernandez, A. E., Dapretto, M., Mazziotta, J., & Bookhei-
control in a modified anti-saccade task: Effects of aging mer, S. (2001). Language switching and language repre-
and bilingualism. Journal of Experimental Psychology: sentation in Spanish–English bilinguals: An fMRI
Learning, Memory, and Cognition, 32, 1341–1354. study. Neuroimage, 14, 510–520.
Bialystok, E., & Martin, M. M. (2004). Attention and inhi- Hollis, S., & Low, J. (2005). Karmiloff-Smith’s RRM dis-
bition in bilingual children: Evidence from the dimen- tinction between adjunctions and redescriptions: It’s
sional change card sort task. Developmental Science, 7, about time (and children’s drawings). British Journal of
325–339. Developmental Psychology, 23, 623–644.
Bialystok, E., & Senman, L. (2004). Executive processes in Kan, P. F., & Kohnert, K. (2008). Fast mapping by bilin-
appearance–reality tasks: The role of inhibition of atten- gual preschool children. Journal of Child Language, 35,
tion and symbolic representation. Child Development, 75, 495–514.
562–579. Karmiloff-Smith, A. (1990). Constraints on representa-
Bialystok, E., & Shapero, D. (2005). Ambiguous benefits: tional change: Evidence from children’s drawing. Cog-
The effect of bilingualism on reversing ambiguous fig- nition, 34, 57–83.
ures. Developmental Science, 8, 595–604. Karmiloff-Smith, A. (2009). Nativism versus neurocon-
Bialystok, E., Shenfield, T., & Codd, J. (2000). Languages, structivism: Rethinking the study of developmental
scripts, and the environment: Factors in developing disorders. Developmental Psychology, 45, 56–63.
concepts of print. Developmental Psychology, 36, 66–76. Kellogg, R. (1970). Analyzing children’s art. Palo Alto, CA:
Braswell, G., & Rosengren, K. (2008). The interaction of National Press Books.
biomechanical and cognitive constraints in the pro- Kirk, S. A., Mccarthy, J. J., & Kirk, W. (1968). The Illinois
duction of children’s drawing. In V. Lange-Küttner Test of Psycholinguistic Abilities (Rev. ed.). Urbana, IL:
& A. Vinter (Eds.), Drawing and the non-verbal mind Institute for Research in Exceptional Children.
(pp. 123–138). Cambridge, UK: Cambridge University Lindholm, K., & Padilla, A. (1978). Language mixing
Press. in bilingual children. Journal of Child Language, 5,
Carlson, S. M., & Meltzoff, A. N. (2008). Bilingual experi- 327–335.
ence and executive functioning in young children. Luquet, G. H. (1927). Le dessin enfantin [Children’s draw-
Developmental Psychology, 11, 282–298. ing]. Paris: Alcan.
Costa, A., Hernandez, M., & Sebastian-Galles, N. (2008). Martin-Rhee, M. M., & Bialystok, E. (2008). The develop-
Bilingualism aids conflict resolution: Evidence from the ment of two types of inhibitory control in monolingual
ANT task. Cognition, 106, 59–86. and bilingual children. Bilingualism: Language and Cog-
Davidson, D., Jergovic, D., Imami, Z., & Theodos, V. nition, 11, 81–93.
(1997). Monolingual and bilingual children’s use of the Morra, S. (2005). Cognitive aspects of change in drawings:
mutual exclusivity constraint. Journal of Child Language, A non-Piagetian theoretical account. British Journal of
24, 3–24. Developmental Psychology, 23, 317–341.
1366 Adi-Japha, Berberich-Artzi, and Libnawi

Nicoladis, E., & Secco, G. (2000). Productive vocabulary Spensley, F., & Taylor, J. (1999). The development of cog-
and language choice. First Language, 20, 3–28. nitive flexibility: Evidence from children’s drawings.
Piaget, J., & Inhelder, B. (1967). The child’s conception of Human Development, 42, 300–324.
space (F. J. Langdon & J. L. Lunzer, Trans.). New York: Thomas, G. V., & Silk, A. M. J. (1989). An introduction to
Norton. (Original work published 1948) the psychology of children’s drawings. New York: Har-
Picard, D., & Vinter, A. (1999). Representational flexibility vester Wheatsheaf.
in children’s drawings: Effects of age and verbal van Heuven, W. J. B., Schriefers, H., Dijkstra, T., & Hag-
instructions. British Journal of Developmental Psychology, oort, P. (2008). Language conflict in the bilingual brain.
17, 605–622. Cerebral Cortex, 18, 2706–2716.
Picard, D., & Vinter, A. (2006). Decomposing and connect- van Sommers, P. (1984). Drawing and cognition: Descriptive
ing object representations in 5- to 9-year-old children’s and experimental studies of graphic production processes.
drawing behaviour. British Journal of Developmental Cambridge, UK: Cambridge University Press.
Psychology, 24, 529–545. Zhi, Z., Thomas, G. V., & Robinson, E. J. (1997). Constraints
Picard, D., & Vinter, A. (2007). Relationships between on representational changes: Drawing a man with two
procedural rigidity and interrepresentational change in heads. British Journal of Developmental Psychology, 15,
children’s drawing behavior. Child Development, 78, 275–290.
522–541.

Anda mungkin juga menyukai