Bilingualism and Cognitive Development

Author(s): Anita D. Ianco-Worrall

Source: Child Development, Vol. 43, No. 4 (Dec., 1972), pp. 1390-1400
Published by: Wiley on behalf of the Society for Research in Child Development
Stable URL: http://www.jstor.org/stable/1127524
Accessed: 08-02-2016 14:33 UTC

REFERENCES
Linked references are available on JSTOR for this article:
http://www.jstor.org/stable/1127524?seq=1&cid=pdf-reference#references_tab_contents

You may need to log in to JSTOR to access the linked references.

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at http://www.jstor.org/page/
info/about/policies/terms.jsp

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content
in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship.
For more information about JSTOR, please contact [email protected].

Society for Research in Child Development and Wiley are collaborating with JSTOR to digitize, preserve and extend access to
Child Development.

http://www.jstor.org

This content downloaded from 192.236.36.29 on Mon, 08 Feb 2016 14:33:26 UTC
All use subject to JSTOR Terms and Conditions
 BILINGUALISMAND COGNITIVEDEVELOPMENT

ANITAD. IANCO-WORRALL

Johannesburg, South Africa

IANCO-WORRALL, ANITA D. Bilingualism and Cognitive Development.

CHILD DEVELOPMENT, 1972, 43, 1390-1400. Limited to one definition of
bilingualism, namely, dual-language acquisition, in a one-person, one-lan-
guage home environment, experiments were designed to test Leopold's ob-
servations on the earlier separation of word sound from word meaning by
bilingual compared to matched unilingual children. Attention to meaning
or to sound of words was tested with the semantic and phonetic preference
test, a two-choice test in which similarity between words could be inter-
preted on the basis of shared meaning or shared acoustic properties. The
notion that bilingualism leads to the earlier realization of the arbitrary nature
of name-object relationship was tested with the questioning technique de-
scribed by Vygotsky. This called for the explanation of names, whether
names can be interchanged and, when names are interchanged in play,
whether the attributes of the objects change along with their names. The
results support Leopold's observation on the earlier separation of sound and
meaning by bilingual children.

Although traditionally the effect of bilingualism has been evaluated in

terms of scores on standardized intelligence tests, a more recent approach
has been to investigate its effects on more specific manifestations of cog-
nitive development (Anisfeld 1964; John 1970; Peal & Lambert 1962).
One of the earliest contributors to this approach has been Leopold
(1939-1949). His detailed account of the development of his bilingual

This paper is based upon a doctoral dissertation submitted to Cornell Uni-

versity. The author is indebted to Professor Harry Levin for his invaluable guid-
ance and assistance. Mrs. Erika Theron ably assisted by translating materials
into Afrikaans and conducting all testing in that language. Author's address:
Institute for Child Guidance, Rand Afrikaans University, 55 Kingsway, Auckland
Park, Johannesburg,South Africa.
[Child Development, 1972, 43, 1390-1400. @ 1972 by the Society for Research in Child Develop-
ment, Inc. All rights reserved.]

This content downloaded from 192.236.36.29 on Mon, 08 Feb 2016 14:33:26 UTC
All use subject to JSTOR Terms and Conditions
 ANITAD. IANCO-WORRALL

child, brought up from the very beginning in a one-person, one-language

home environment, contained numerous entries on what he considered to
be favorable effects of bilingualism on the child's mental development. One
of the most remarkable effects of bilingualism, Leopold wrote, was "a
noticeable looseness of the link between the phonetic word and its mean-
ing" (1961, p. 358). His daughter, Hildegard, from a very early age, could
render the same story freely in both languages. When memorizing rhymes
or songs she would often destroy the rhyme with her own insertions of
meaningfully related vocabulary. She readily accepted new names for objects
already denoted in one language and asked to be given the name in the
second, or even a third, unfamiliar language.
The possible implication of these observations have intrigued many
a student of bilingualism. Anisfeld (1964) and Peal and Lambert (1962)
used them to support their views regarding the favorable effects of bilin-
gualism on the development of cognitive processes. There is no empirical
evidence, however, that Leopold's findings can be generalized to other
bilingual children brought up under circumstances similar to or different
than those of Hildegarde's. Furthermore few studies to date have been
sufficiently rigorous in providing an adequate definition of the childhood
bilingualism under investigation. Given our knowledge of the importance
of early experience on the development of cognitive processes, the circum-
stances under which the two languages were acquired have to be con-
sidered: whether from infancy, in preschool, or grade school; at home in
a one-person, one-language acquisition context, or in a setting where the
two languages are used interchangeably. A different, albeit overlapping,
constellation of variables would have to be defined for each type of
dual-language acquisition, and this in turn related to aspects of cognitive
development.
Assuming that Leopold's observations point to a faster rate of develop-
ment in specific areas of cognitive functioning, a first step in testing this
hypothesis would be to find empirical support for these observations. The
present study is part of a larger project which had as its aim to test the
above hypothesis. This paper reports on two experiments aimed at testing
Leopold's observations on the earlier separation of word sound from word
meaning by bilingual children. Attention to meaning or to sound was tested
with the semantic or phonetic preference test, a two-choice test on which
similarity between words could be interpreted on the basis of shared mean-
ing or shared acoustic properties. Exploratory studies have shown that on
this test semantic preference increases with age. The notion that bilingual-
ism leads to the earlier realization of the arbitrary nature of name-object
relationship was tested with the questioning technique described by Vygot-
sky (1962).
Bilingualism, in this study, is defined as dual acquisition of language
in a one-person, one-language home environment. All bilingual children

1391

This content downloaded from 192.236.36.29 on Mon, 08 Feb 2016 14:33:26 UTC
All use subject to JSTOR Terms and Conditions
CHILD DEVELOPMENT

used in this iesearch have been brought up under circumstances similar

to those described by Leopold. Not being a longitudinal study, however, the
age groups selected were determined by such considerations, among others,
as a subject's responsiveness to the demands of a structured test situation.
Leopold had ample opportunities to make his observations at earlier stages
in his child's development.

METHOD

Subjects
The Ss were 30 Afrikaans-English bilinguals, 14 girls and 16 boys, aged
4-6 and 7-9 years. They were drawn from nursery schools and grades 2
and 3, respectively. Each bilingual was paired to two unilingual children,
one Afrikaans speaking, the other English speaking, matched on intelligence,
age, sex, school grade, and social class. The setting was Pretoria, South
Africa.

Selection Materials

Bilingualism and unilingualism were ascertained through mother inter-

views, a test of bilingualism, and teachers' reports.

Mother Interviews

Bilingual sample.-In the course of a home interview the mother was

asked to supply information on the socioeconomic and language background
of her family. The first part of the questionnaire dealt with
background
information and was administered verbally to the mother. The second part
dealt with details on the language interaction in the family. The mother
was asked to read the instructions and then to underline on a
four-point
scale the word most applicable in her child's case. The first part dealt with
how often English was being used in the family, the second
part with how
often Afrikaans was being used. The mother's interaction was determined
first: how often she spoke to the child in English/Afrikaans, how often
the child replied in that language when she spoke to him, and whether the
conversation then continued in that language. The child was then considered
in relation to mother: how often he addressed the mother in
English/Afri-
kaans, how often she replied to the child in that language, and whether
the conversation continued in that language. Father-child language inter-
action was similarly ascertained, as were the child's interactions with sib-
lings and other members of the family. Further information was obtained
on the child's language experience outside the home, in school, playground,
and so on.

1392

This content downloaded from 192.236.36.29 on Mon, 08 Feb 2016 14:33:26 UTC
All use subject to JSTOR Terms and Conditions
 ANITAD. IANCO-WORRALL

The final phase of the interview was taped. It consisted of an informal

conversation with the mother on her impressions and observations concern-
ing the child's bilingual experience.
Unilingual sample.-The task of drawing a unilingual sample proved
very time-consuming because of the predominantly bilingual nature of a
city like Pretoria. Mothers were therefore telephoned. They were asked to
describe the extent to which the child had been exposed to the second lan-
guage; whether the family used only one language exclusively; whether
the child had opportunities to interact with others who spoke the second
language at home, at nursery school, in the playground, in church, and so
on. This interview proved to be straightforward as South Africans are very
conscious of the second language. Mothers seemed quite certain of whether
or not their children had been exposed to the second language and whether
or not they functioned in it.

Test of Bilingualism

For a rough indication of a S's competence in the two languages, equiv-

alent forms in English and Afrikaans were adapted from the vocabulary
subtest of an old South African intelligence test for children. The test
tapped basic skills in production, comprehension, and reproduction of
language, and was not a measure of the degree of competence in that lan-
guage. Although it was administered in the selection of both bilingual and
unilingual Ss, it was most useful in selecting the unilingual sample. Being
so simple, any S showing difficulty in responding to the test was obviously
unable to function in that language.
Bilingual Ss were tested with both English and Afrikaans versions in
two different sessions, several days apart. The order in which the two ver-
sions were administered was random. The Afrikaans test was administered
by a native speaker of that language.
Unilingual Ss were tested with both language versions in one session.
Both Es were present. The S was presented with the language version of
least familiarity. If he showed difficulty in responding, the alternate lan-
guage was used. Complete or almost complete failure in the second language
and complete or almost complete success in the first was used as one
criterion for unilingualism.
Teacher's reports.-The way in which S functioned in one or both
languages was informally discussed with the class teacher.

Procedure

The two experiments were carried out in one session. Afrikaans test-
ing was conducted by a native speaker of that language. Bilinguals were
tested in both languages, in separate sessions several days apart.

1393

This content downloaded from 192.236.36.29 on Mon, 08 Feb 2016 14:33:26 UTC
All use subject to JSTOR Terms and Conditions
CHILD DEVELOPMENT

EXPERIMENT
I

Materials
The semantic and phonetic preference test consisted of eight verbally
presented, one-syllable sets of words. There were two language versions.
The words in the two languages were not made to correspond. All words
were drawn from infant readers.
Each set was made up of three words. One word was the standard
word, the other two were choice words. One choice word was phonetically
related to the standard-the two words shared the same sound in word-
initial position. The second choice word was semantically related to the
standard. The S was told: "I have three words: cap, can, and hat. Which
is more like cap, can or hat?"
The eight sets of words were presented three times, the order between
the sets as well as the position of the choice words within the sets was random.
In the total of 24 sets presented, phonetic and semantic choices appeared
in set-final position 50% of the time.

EXPERIMENT
II

Vygotsky (1962) asked his young S to explain names of things, whether

names of things may be interchanged, and when names are being inter-
changed in play whether the attributes of the objects change along with
their names.
From this brief outline described in Vygotsky, we developed a short
questionnaire, with equal versions in both languages. There were three
parts to this test:
Part I called for an explanation of six names: dog, cow, chair, jam,
book, and water. The S was asked, "Why is a dog called 'dog'?"
Part II called for whether or not names may be interchanged. There
were three pairs of names: dog and cow, chair and jam, book and water.
The S was asked, "Could you call a dog 'cow' and a cow 'dog'?"
In the course of an exploratory study we made an interesting observa-
tion not mentioned by Vygotsky. Older children, like the younger ones,
denied that names may be interchanged. The former felt that names may
not be interchanged at will, that there are social and linguistic conventions
that may not be broken. It was therefore necessary to add another question,
namely, whether or not, in principle, one could interchange names of things,
thus: "Suppose you were making up names for things, could you then call
a cow 'dog' and a dog 'cow'?" The answer, among older children in-
variably was, "Yes."
Part III called for the interchange of names in play, thus: "Let us
play a game. Let us call a dog 'cow.' " This was then followed by two ques-

1394

This content downloaded from 192.236.36.29 on Mon, 08 Feb 2016 14:33:26 UTC
All use subject to JSTOR Terms and Conditions
 ANITA D. IANCO-WORRALL

tions: "Does this 'cow' have horns?" and "Does this 'cow' give milk?" Two
other pairs of names were also interchanged, thus a chair was called 'jam'
and a book was called 'water.' These were each followed by two questions,
thus: "Can you eat this 'jam'?"; "Can you sit on this 'jam'?"; "Can you
drink this 'water'?" and "Can you read this 'water'?"

RESULTS

Experiment I
Table 1 presents the proportion of semantic and phonetic choices as
a function of linguality, age, and language of test. The difference between
semantic and phonetic preference was tested for significance by the Wil-

TABLE 1
GROUP CHOICES AND INDIVIDUAL PREFERENCES FOR SEMANTIC
AND PHONETIC DIMENSIONS

PROPORTION OF

Group Choices IndividualPreferences

TESTANDAGE GROUP No. Ss Semantic Phonetic Semantic Phonetic Neither
English:
4-6 ...... Unilinguals 12 .46 .54 .08 .25 .67
Bilinguals 13 .68 .32 .54 .08 .38
7-9 ...... Unilinguals 16 .60 .40 .38 .25 .37
Bilinguals 17 .67 .33 .59 .24 .17
Afrikaans:
4-6 ...... Unilinguals 13 .49 .51 0 .08 .92
Bilinguals 13 .60 .40 .46 .08 .46
7-9 ...... Unilinguals 17 .64 .36 .65 .29 .06
Bilinguals 17 .66 .34 .59 .24 .17

coxon signed-ranks test for paired replicates (Siegel 1956). A similar analysis
of the difference in semantic choices between bilinguals and unilinguals
showed a significant difference in favor of bilinguals on the English test
(p < .05 on a two-tailed test). There were no other significant differences
in semantic choices between the groups.
Our exploratory studies led us to expect a strong position bias among
younger Ss. The tendency had been to repeat the last word heard in the
three-word sequence. In the present study, the Afrikaans-speaking nursery
school unilinguals showed the strongest position bias. Similarly, the con-
sistency of choice across the three presentations of the eight sets of words
was low for these Ss. As these limitations would be most strongly reflected
in the group choice analyses, an individual preference analysis was then
undertaken.

1395

This content downloaded from 192.236.36.29 on Mon, 08 Feb 2016 14:33:26 UTC
All use subject to JSTOR Terms and Conditions
CHILDDEVELOPMENT

A S was considered unidimensional in preference if one dimension was

chosen at least twice as often as the other. Thus, if at least 66 % of his
choices were semantic, S was classified in the semantic category. If at least
66% of his choices were phonetic, S was classified into the phonetic cate-
gory. If neither dimension was preferred more than 66% of the time, S
was classified into the neither category. A S showing a strong position bias
would then fall into the neither category. All Ss classified into one of the
two preference categories showed an intertrial agreement which exceeded
66%.
Table 1 also presents the proportion of Ss in each group classified into
one of the three categories of choice. There were no differences in the
way
bilingual Ss were classified in the two language tests.
As the expected cell frequencies were too small to meet the
assump-
tions of the X2 test, the binomial test, suggested by
Siegel (1956), where
1/2 (A + D) is less than 5, was used to test the difference between the

groups. In the younger, nursery school group, a significantly higher propor-

tion of bilinguals fell into the semantic category of choice,
English test,
p = .035; Afrikaans test, p = .016. In the older group, there were no
significant differences between bilinguals and unilinguals in either language
test. The phonetic preference category did not
distinguish between the
groups.
The semantic group choice data was tested for age trend by Kendal -
(Siegel 1956). In the unilingual groups this trend significantly increased
with age, English unilinguals, p = < .04; Afrikaans
unilinguals, p = < .01.
There was no significant age trend in semantic choice
among bilinguals on
either language test.

Experiment II
Part I.-The Ss' explanations of names fell into seven
categories. These
categories or types of explanations correspond by and large to the "language
frames" which children used in Bruner and Olver's (1963)
grouping experi-
ment. Explanations fell into one or more of the
following categories: ignor-
ance, arbitrary justification, perceptible attributes, extrinsic functional
attributes, intrinsic functional attributes, social convention, and abstract
justification. There were no differences between bilinguals and unilinguals
in the types of explanations they offered. For all
groups the majority of
answers fell within the explanation by attributes
categories. For example:
"A cow is called 'cow' because you milk it."
Part II.-The Ss' responses to the two questions
dealing with inter-
change of names were combined. The combined response classified him into
one of the following four categories: no-no, no-yes,
yes-yes, and yes-no. Only
Ss consistent in the type of combined
response they offered across the three
pairs of words were thus classified. A fifth category, inconsistent, was used

1396

This content downloaded from 192.236.36.29 on Mon, 08 Feb 2016 14:33:26 UTC
All use subject to JSTOR Terms and Conditions
 ANITAD. IANCO-WORRALL

TABLE 2
PROPORTION OF SS CLASSIFIED INTO THE FIVE COMBINED-RESPONSE CATEGORIES:a
BY LINGUALITY AND AGE LEVEL

4-6-YEAR-OLDS 7-9-YEAR-OLDS

COMBINED
RESPONSE English Afrikaans English Afrikaans
CATEGORY Unilinguals Bilingualsb Unilinguals Unilinguals Bilinugalsb Unilinguals
No-no ....... .08 .23 .38 .31 .23 .71
No-yes ....... .08 .38 .08 .38 .59 ...
Yes-yes ...... .08 .08 .. ...
Yes-no ....... ... ...
Inconsistent .. .76 .31 .54 .31 .18 .29
a The two types of questions were: "Could you call a cow 'dog' and a dog 'cow'?" and "Suppose
you were making up names for things, could you then call a cow 'dog' and a dog 'cow'?"
b Bilinguals first test response, regardlessof language.

to group the inconsistent Ss. Table 2 presents the proportion of Ss grouped

into the five combined-response categories by linguality and age level.
As bilingual Ss were presented with both language versions of this inter-
view, and as the game part (or Part III of the test administered in a previous
session) would influence a S's response to Part II of the test in the subse-
quent language version, only a bilingual's first test response was considered,
regardless of the language in which it had been administered.
As can be seen in table 2, the majority of Ss' combined responses fell
into the no-no, no-yes, or inconsistent categories. Since the no-yes category
corresponded to the response combination obtained from high school stu-
dents in our exploratory study, the difference between the proportion of
bilingual and unilingual Ss thus classified was tested for significance by the
Cochran Q (Siegel 1956). In the younger age group the difference ap-
proached significance, 12 matched sets, Q (2) = 5.33, p = between .05 and
.10. In the older age group the difference between the three groups was
significant, 16 matched sets, Q(2) = 11.45, p = < .01. At both age levels
the highest proportion of Ss to be classified into the no-yes combined-response
category was in the bilingual group. Although the great majority of unilin-
gual Ss replied that names of objects could not be interchanged, the major-
ity of bilingual Ss agreed that in principle this could be done.
Part III.-The Ss were asked to interchange names of three pairs of
objects, each "game" being followed by two questions. A maximum score of
6 could be obtained if all answers were correct, that is, if the attributes of
the object were retained although the name was changed to the alternate
object. Thus to the game "Let us call a dog 'cow,' does this 'cow' have horns?"
the correct answer was "No." This was generally followed by S volunteering
the information "because it is really a dog."
Table 3 presents mean number of correct responses and standard devia-
tions for each group as a function of linguality and age level. It also shows

1397

This content downloaded from 192.236.36.29 on Mon, 08 Feb 2016 14:33:26 UTC
All use subject to JSTOR Terms and Conditions
CHILD DEVELOPMENT

bilingual Ss' first test response, regardless of the language in which the test
was administered. The proportion of Ss consistently correct across all three
"games" was 50% for the younger age group and 88% for the older age
group. Although the ability to interchange the names of objects in play im-
proved with age, there were no differences between bilinguals and unilin-
guals at either age level.

TABLE 3
CORRECT RESPONSES TO THE GAME: BY LINGUALITY AND AGE LEVEL

4-6-YEAR-OLDS 7-9-YEAR-OLDS
TESTANDGROUP N Mean SD N Mean SD
English:
Unilinguals ......... 12 4.75 1.42 16 5.88 0.35
Bilinguals .......... 13 4.92 1.60 17 5.94 0.24
Afrikaans:
Unilinguals ......... 13 4.62 1.89 17 5.82 0.39
Bilinguals .......... 13 4.53 1.61 17 5.82 0.52
First test:
Bilinguals .......... 13 4.31 2.06 17 5.76 .57

DISCUSSION

Leopold's observed separation of word sound from word meaning in

young bilinguals found experimental support in this study. Of the young,
4-6 year-old bilinguals, 54% consistently chose to interpret similarity be-
tween words in terms of the semantic dimension. Of the unilingual groups
of the same age, not one Afrikaans speaker and only one English speaker
showed similar choice behavior. Whereas the age trend analysis as well as
the individual preference analysis showed that semantic preference increased
as a function of age in the unilingual groups, as far as our two age
groups
are concerned the bilingual group did not show a similar progression. The
conclusion we draw is that bilinguals, brought up in a one-person, one-lan-
guage home environment, reach a stage in semantic development, as mea-
sured by our test, some 2-3 years earlier than their unilingual peers. A high
percentage of these bilingual youngsters perceived relationship between
words in terms of their symbolic rather than their acoustic
properties, a
finding not unlike Kolers's (1966) that bilingual college students found a
linguistically mixed text as informative as a unilingual text. Here too it was
the symbolic or conceptual relationship between the words that was attended
to.
The application of the Vygotsky technique led to some interesting results.
It was found that the notion that names are arbitrarily assigned to objects

1398

This content downloaded from 192.236.36.29 on Mon, 08 Feb 2016 14:33:26 UTC
All use subject to JSTOR Terms and Conditions
 ANITA D. IANCO-WORRALL

develops later than the ability to separate the qualities of objects from their
names. There was little difference in the way bilinguals and unilinguals
separated the names from the objects in play. Bilinguals of both age groups
excelled over unilinguals, not where names are conceived to be aspects of
things but where the question of whether names can be interchanged re-
quired the formulated concept that names are arbitrarily assigned to things.
What insights do these findings give us into the development of basic
cognitive structures of bilingual children? What aspects of cognitive func-
tioning does bilingualism, as defined in our study, in fact influence? Two
tests of psychological functioning, namely, intended and actual classification
of objects as studied by Inhelder and Piaget (1964) and optional shift be-
havior (Kendler, Kendler, & Learnard 1962), did not differentiate between
our bilinguals and unilinguals at either age level. These tests were not
chosen at random. They were selected on the basis of our assumptions of
how bilingual language learning could influence these modes of functioning.
We conclude that speculations on the especific effects of bilingualism on
the development of basic cognitive functioning are premature. In this we
are in agreement with Macnamara (1970). We feel that much more care-
ful studying of symbolic development, such as began in the present study,
research into the nature of the translation process and the skills developed
in that process, as well as into the process of keeping two languages distinct
from a very early age, should precede assumptions as to how bilingualism
affects thought processes and their development. Although bilingualism and
its effect on intelligence has been studied since the early 1920s, we have
only recently begun to carefully define our variables, take a close look at
the nature of the bilingual event, consider ways in which the experience of
bilingual children may differ from those of unilingual children and how
these experiences may influence the course of cognitive development.
That Leopold's astute observations have found empirical support, we think,
is a major breakthrough in the study of childhood bilingualism, but the
road ahead remains arduous and hazardous.

REFERENCES
Anisfeld, E. A comparison of the cognitive functioning of monolinguals and bi-
linguals. Unpublished doctoral dissertation, McGill University, 1964.
Bruner, J. S., & Oliver, R. R. The development of equivalent transformationsin
children. In J. C. Wright & J. Kagan (Eds.), Basic cognitive processes in
children. Monographs of the Society for Research in Child Development,
1963, 28 (2, Serial No. 86).
Inhelder, B., & Piaget, J. The early growth of logic in the child. London: Rout-
ledge & Kegan Paul, 1964.
John, V. Cognitive development in the bilingual child. In J. E. Alatis (Ed.),
Monograph Series on Languages and Linguistics. Georgetown University
School of Languages and Linguistics, 1970, 23, 59-67.

1399

This content downloaded from 192.236.36.29 on Mon, 08 Feb 2016 14:33:26 UTC
All use subject to JSTOR Terms and Conditions
CHILDDEVELOPMENT

Kendler, T. S.; Kendler, H. H.; & Learnard, B. Mediated responses to size and
brightness as a function of age. American Journal of Psychology, 1962, 75,
571-586.
Kolers, P.A. Reading and talking bilingually. American Journal of Psychology,
1966, 79, 357-376.
Leopold, W. F. Speech Development of a bilingual child. Evanston, Ill.: North-
western University Press, 1939-1949. 4 vols.
Leopold, W. F. Patterning in children's language learning. In S. Saporta (Ed.),
Psycholinguistics. New York: Holt, Rinehart & Winston, 1961.
Macnamara,J. Bilingualism and thought. In J. E. Alatis (Ed.), Monograph Series
on Languages and Linguistics. Georgetown University School of Languages
and Linguistics, 1970, 23, 25-40.
Peal, E., & Lambert, W. E. The relation of bilingualism to intelligence. Psycho-
logical Monographs,1962, 76 (27, Whole No. 546).
Siegel, S. Non-parametric statistics. New York: McGraw-Hill, 1956.
Vygotsky, L. S. Thought and language. Boston: M.I.T. Press, 1962.

1400

This content downloaded from 192.236.36.29 on Mon, 08 Feb 2016 14:33:26 UTC
All use subject to JSTOR Terms and Conditions