Word Magic Revisited: Monolingual and Bilingual Children's Understanding of the Word-

Object Relationship
Author(s): Tamar Rosenblum and Steven A. Pinker
Source: Child Development, Vol. 54, No. 3 (Jun., 1983), pp. 773-780
Published by: Wiley on behalf of the Society for Research in Child Development
Stable URL: https://www.jstor.org/stable/1130064
Accessed: 10-02-2019 21:10 UTC

REFERENCES
Linked references are available on JSTOR for this article:
https://www.jstor.org/stable/1130064?seq=1&cid=pdf-reference#references_tab_contents
You may need to log in to JSTOR to access the linked references.

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].

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at
https://about.jstor.org/terms

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

This content downloaded from 128.103.224.7 on Sun, 10 Feb 2019 21:10:12 UTC
All use subject to https://about.jstor.org/terms
Word Magic Revisited: Monolingual and
Bilingual Children's Understanding of the
Word-Object Relationship

Tamar Rosenblum

Harvard University

Steven A. Pinker

Massachusetts Institute of Technology

ROSENBLUM, TAMAR, and PINKEI, STEVEN A. Word Magic Revisited: Monolingual an

Children's Understanding of the Word-Object Relationship. CmLD DEVELOPMENT
773-780. We examined the claims that preschool children regard an object's nam
inseparable from its intrinsic properties and that bilingual children relinquish th
earlier than monolingual children. 12 monolingual and 12 bilingual children of equiv
nonverbal intelligence, talkativeness, and willingness to consider counterfactual situa
asked whether various objects could be renamed and then asked to identify object
sense names and names for other objects. Virtually all the children performed the r
tasks without error, and the monolinguals and bilinguals were equally likely to assen
naming objects. However, monolinguals were more likely to mention an object's p
when justifying an answer, whereas bilinguals were more likely to mention the soci
of the renaming act. These findings suggest that neither bilingual nor monolingual
are necessarily subject to "word magic"; rather, monolinguals have learned that an o
have more than 1 name by virtue of its various attributes, whereas bilingual chil
learned, in addition, that an object can have more than 1 name by virtue of the
social contexts in which its name is used.

One of the more intriguing claims that If preschool children have difficulty ap-
have been made about children's lack of preciating that words and objects are related
by convention only, it would be natural to
metalinguistic awareness is that they cannot
expect that bilingual children, who know two
grasp the arbitrary nature of the w6rd-object
relation. Piaget (1929) observed that pre-
words for most objects, would come to ap-
preciate this fact at a younger age. Indeed,
school children answer questions about why
objects have the names they do as if they
diary studies (Leopold, 1949; Slobin, 1978)
believed that the word is "magically" contain
con- reports of bilingual children making
nected to the object or that the word precocious
is an observations about words and their
intrinsic property of the object, like itsproperties
size (though, of course, these may have
or color (a phenomenon he called "nominal been precocious children in general). Obser-
vations of groups of monolingual and bi-
realism"). In the same vein, Vygotsky (1962)
lingual children have shown that bilinguals
noted that children give idiosyncratic answers
to questions calling on them to manipulateoften are more willing to refer to an object
names, such as "Can you call this dog by a anonsense or unconventional name (e.g.,
cow?" and "Can this 'cow' give milk?" Ben-Ze'ev,
(e.g., 1977; Feldman & Shen, 1971;
they would assent to the latter). Ianco-Worrall, 1972) or to articulate the
We are grateful to the children, parents, and child-care center staff who participated in
this study and especially to Karen Silverston and Nancy Josephson of the Harvard Law School
child-care center and to Mildred Axelrad of the Temple Emanuel Kindergarten in Brookline,
Massachusetts. In addition, we thank Dante Cichetti and his research group for the use of their
intelligence tests and Roger Brown, Catherine Snow, Susan Pinker, I. M. Schlesinger, Ellen
Markman, Sophia Cohen, John Flavell, Linda Siegel, and two anonymous reviewers for their
comments and advice. This research was supported by funds from the Department of Psychol-
ogy and social relations at Harvard University and by NSF grants BNS 8114916 and BNS
8209450 awarded to the second author. Requests for reprints should be sent to Steven A.
Pinker, Department of Psychology, E10, Massachusetts Institute of Technology, Cambridge,
Massachusetts 02139.
[Child Development, 1983, 54, 773-780. @ 1983 by the Society for Research in Child Development, Inc.
All rights reserved. 0009-3920/83/5403-0025$01.001

This content downloaded from 128.103.224.7 on Sun, 10 Feb 2019 21:10:12 UTC
All use subject to https://about.jstor.org/terms
774 Child Development
principle that words and objects are related may be brighter, higher in socioeconomic
by arbitrary convention (Cummins, 1978). status, more likely to entertain counterfac-
However, these differences in answers to tuals, or simply more talkative in an experi-
questions have not always translated into mental setting (in which case they might give
behavioral differences (e.g., Cummins, 1978; "better" answers to open-ended questions
lanco-Worrall, 1972), and occasionally no simply because they are willing to proffer an
differences have been found in tasks of un-extended answer, whereas shyer children
derstanding the word-object relation (e.g.,would answer in monosyllables, if at all).
Pinker, 1979; Sandoval, 1976). Interestingly, in each of the studies reporting
differences between monolingual and bilin-
In any case, we would like to argue that gual children, one or more of these factors
neither hypothesis-that preschoolers believe were not equated (e.g., intelligence for Feld-
words to be intrinsic properties of their refer- men & Shen, 1971; SES for lanco-Worrall,
ents and that bilingual children are disabused 1972; talkativeness and comprehension of
of this notion earlier, when they learn two counterfactuals in all the studies); and in the
names for an object-receives unambiguousstudies that failed to find statistically signifi-
support from these data, even if the data cant differences (Pinker, 1979; Sandoval,
were perfectly consistent. We see three prob- 1976), both intelligence and SES were
lems of interpretation concerning these studies.equated.

1. Children's interpretation of the ques- 3. Interpreting differences-semanticver-

tions: There is reason to believe that chil-
sus pragmatic knowledge: It is not really true
dren's answers to the experimenter's questionsthat monolingual children hear a single name
in Piagetian tasks may not always reflect theirfor an object; consider, for example, Sparky,
cognitive abilities accurately (see Siegel,
dog, poodle, pet, animal, and living thing.
1978). This could affect the interpretationWhat differentiates the bilinguals' experience
of children's metalinguistic judgments in sev-from the monolinguals' is that, when a mono-
eral ways. First, preschoolers do not seem tolingual child hears a new word for an object
understand fully the words for linguistic en-(synonyms aside), the new word will differ
tities and relations such as word, say, means,semantically from the known name (e.g.,
sounds like, and so on (Slobin, 1978). Fur- animal refers to any animate creature, where-
thermore, children (Osherson & Markman,as dog refers only to those with a muzzle, fur,
1975) and adults in preliterate cultures (Cole a bark, etc.). In contrast, when a bilingual
& Scribner, 1974) often answer logical ques- child hears a new word for an object, often
tions or hypothetical questions as if the the new word will differ in that it is used
questions referred to empirical states of thein a different social context (e.g., in school
world rather than to the necessity, possibil-vs. at home; when speaking to mother rather
ity, or impossibility of the premise itself. Thusthan to father). Thus, monolinguals may learn
the somewhat bizarre question "Can you call that an object is referred to by two or more
a cow a tree?" may be interpreted as the farnames by virtue of its individuality and its
more plausible "Is the English word for aphysical properties, whereas bilingual chil-
cow 'tree'?" (see Markman, 1976); therefore, dren learn that an object is referred to by
the children's answers in the negative maytwo or more names by virtue of those attri-
not be as anomalous as they would first ap- butes as well as its use in different discourse
pear. This might be so because children arecontexts. As a result, in an experimental set-
incapable of entertaining counterfactual in- ting, bilinguals might be more likely to agree
formation of any sort or because they chooseto dubbing an object with a new name pre-
not to interpret the question in its hypotheti-cisely because the experiment is a different
cal sense owing to the bizarreness of thatsocial setting from that in which the regular
interpretation in the discourse context (see name applies; the differences would be in the
Grice, 1975). On this interpretation, if thechildren's knowledge about when and why
children were "set" to attend to counter-
an object can have a second name, rather
factual information in that context, they might
than whether it can have a second name or
indeed answer the name-switching question
any other possible misconceptions about nom-
in an adult-like way.
inal realism or word magic. Simply obtaining
a child's assent or dissent to switching names
2. Variables confounded with bilingual-
(e.g.,
ism: Bilingual children may differ from mono- Ben-Ze'ev, 1977; Feldman & Shen,
1971) will not settle this issue; at the very
linguals in more ways than simply knowing
least, one would need the children's rationales
another language. In any given sample they

This content downloaded from 128.103.224.7 on Sun, 10 Feb 2019 21:10:12 UTC
All use subject to https://about.jstor.org/terms
 Rosenblum and Pinker 775

for their assent or dissent to see if they re- Nonverbal intelligence was evaluated by
ferred to an object's properties, the social two performance tests from the Wechsler
setting, or both. Preschool and Primary Scale of Intelligence
(Wechsler, 1963). The "Animal House" sub-
This study was designed to address these
test, which requires the child to associate one
three objections to the conclusion that mono-
of four colored pegs with each of four animal
lingual, but not bilingual, children are sus-
pictures, was chosen because it is a visual
ceptible to nominal realism. To address the
analog to the name-learning and usage tasks.
objection about children's interpretation of
The mean for the monolingual children on
the question, we presented the children with
this subtest was 12.8; for the bilingual chil-
a nonlinguistic example of a counterfactual dren, it was 13.1, t(22) = .30, N.S. In ad-
scenario and assessed whether they could
dition, the Block Design subtest, which re-
entertain and comprehend the scenario. This
quires the child to arrange blocks so as to
serves three purposes. First, if all the children
reproduce geometric patterns from models
fail this test, we could attribute any subse-
and pictures, was chosen because it is rela-
quent general failure in a name-switching
tively language-independent and, unlike Ani-
task to an inability to consider counterfac-
mal House, does not have a parallel in the
tuals rather than to a metalinguistic deficit
interview procedure. The means for the mono-
per se. Second, if bilinguals could pass the
lingual and bilingual children were 14.0 and
test but monolinguals could not, any subse-
14.4, respectively, t(22) = 1.27, N.S.
quent differences between groups in the meta-
linguistic task could be attributed to this All subjects were children of profession-
general cognitive difference. Third, if both als or university affiliates and were enrolled
groups would ordinarily balk at name switch- in English-language child-care centers in the
ing but would agree to it only if "set" to Boston area. All of the bilinguals were chil-
attend to counterfactuals seriously, this test dren of either Israeli Hebrew-speaking fami-
would provide the appropriate set and, if all lies who had moved to the United States but
children passed it, would give us more confi- still spoke Hebrew in the home or (in two
dence that any subsequent failures in the of the 12 cases) American English-speaking
name-switching task reflected metalinguistic families who returned from an extended stay
deficits.
in Israel with their Israeli-born Hebrew-speak-
In addition, we ensured that our groups ing child. One of these children spoke He-
were equivalent in education and socioeco- brew at home; the other spoke English. Their
data were comparable to those of other chil-
nomic status by using Hebrew-English bilin-
gual children of Israeli emigrants (who, un- dren in the bilingual group. The monolinguals
like many bilingual groups, tend to be above were also all children of professionals; four
average in parental education and SES); we of the 12 were known to be Jewish.
also assessed the children's nonverbal IQ, Bilingual subjects were selected accord-
their vocabulary development, and their gen- ing to three criteria. First, each of the bilin-
eral talkativeness. Finally, to assess whether gual subjects used both languages daily in
differences are in semantic or pragmatic different settings: Hebrew with family at
knowledge, we recorded the children's rea- home, and English with peers in the day-care
sons for their answers on several name-ma- center. We sought a reasonable balance be-
nipulation tasks rather than simply the answer tween proficiency in the two languages, which
itself.
we assessed by speaking with the child in
both languages and interviewing the parents
Method about the child's language history. Finally,
if a child satisfied both these criteria, the
Subjects Peabody Picture Vocabulary Test (PPVT)
The subjects of the study were 12 En- was administered in both its English standard-
glish monolingual and 12 Hebrew-English ized form and in a Hebrew nonstandardized
bilingual preschoolers of equivalent age, sex, translation. For the bilingual children, half
nonverbal intelligence, and socioeconomic sta- the PPVT items were administered in one
tus. The subjects ranged in age from 4-0 to language, then the other half were adminis-
5-10; the monolingual group's average age tered in the other language; which half set
was 4-9, and the bilingual group's average was given in which language was counter-
age was 5-1, t(22) = 1.25, N.S. In each balanced across subjects. The first language
group, seven subjects were girls and five were
tested was always the language in which the
boys. interview was conducted (see below). To

This content downloaded from 128.103.224.7 on Sun, 10 Feb 2019 21:10:12 UTC
All use subject to https://about.jstor.org/terms
776 Child Development
tabulate the PPVT scores based on half the their hands [demonstrates], and rides tricycles
test questions, the basal and ceiling itemsto work." The experimenter then offered the
were established as usual, but twice the num-
puppet to the child, who then assumed the
ber of correct responses between the two role of Mr. Blue. The experimenter prompted
the child to have Mr. Blue repeat his story
points were added to the basal score to com-
in the first person. At this point the experi-
pensate for the halved number of test items.
We chose a 50-point difference between menter ceased prompting and asked the child
English and Hebrew PPVT scores as the maxi-three questions: "Is your hair the color of the
mum allowable disparity for inclusion in the
earth or of the sky?" (correct answer: sky);
"When you go for a walk, do you put your
bilingual group. Using these criteria, we se-
lected 12 children from the 16 originally
shoes on the part of your body that has
tested. fingers or that has toes?" (correct answer:
fingers); and "Do the grown-ups ride to work
Monolinguals scored significantly higher
on something that has three wheels or that
than the bilinguals on the English PPVT-- has four wheels?" (correct answer: three).
monolingual X = 116.5, bilingual X = 89.1, The order of mention of the two alternatives
t(22) = 4.75, p < .001. The Hebrew scores
in each question was counterbalanced across
cannot be compared directly with the English
children. Note that the form of these ques-
scores because the translated version is not
standardized. Lower PPVT scores were ex- tions is similar to the form of questions on
the name-switching tasks: first the child is
pected for the bilinguals; their vocabulary
asked to accept some counterfactual; then he
development in each language tends to lag
or she is asked to deduce some implication
somewhat behind that of their monolingualof the counterfactual.
counterparts until the middle school-age years,
presumably because the bilingual child has Test of volubility.-Each child was first
asked to name five common toys (boat,
exposure to fewer words or a reduced amount
of exposure to any word in a given language
giraffe, horse, table, truck), then the experi-
than the monolingual has (see Ben Ze'ev, menter asked two questions: "Which one do
1977; McLaughlin, 1978). you like best?" and "Which two of these are
most alike?" The number of content words
Procedure
in the child's response (nouns, adjectives, and
Each child was interviewed individually
main verbs other than to be, which has no
by the first author in a quiet room. All the
present tense form in Hebrew) were recorded.
monolingual children were tested in day-care
Because these content words correspond one
centers; four of the bilinguals were tested
to one in English and Hebrew, this measure
in day-care centers, and eight in the child's
may be used to assess whether monolingual
home. Each session lasted approximately 40
and bilingual children are equally voluble or
min and was tape-recorded. For the bilin-
shy in the experimental setting.
gual children, the choice of language in which
the interview was to be conducted was left Name-manipulation test.-The experi-
to the subject. All four bilinguals interviewed
menter began the interview itself by address-
in the day-care center chose to speak English,ing two hypothetical questions to the child-
and six of the eight children tested in their"Can you call this table a shig?" and "Can
homes chose to speak Hebrew; therefore, you
it call this boat a cow?"-and then asked
turned out that half of the bilinguals were
the child to justify his or her responses.
tested in English, and half of them were
tested in Hebrew. The experimenter next administered three
"renaming" tasks. First, the experimenter said
After getting acquainted with the child,"Let's call this table a shig," referred to the
the experimenter administered the test of object by that name in two requests ("Hand
comprehension of counterfactuals, the test of
me the shig" and "Put the shig next to the
volubility, and the name-manipulation test
truck"), and asked for a justification with
(which had five parts), followed by the the question, "Why can we now call the table
WPPSI Animal House subtest, the PPVT, anda shig?"
the WPPSI Block Design subtest.
Second, the child was taught a different
Counterfactual test.-The experimenter
name, but one already existing in the child's
put on a blue-haired puppet and explained
language, for another of the objects on the
that "Mr. Blue" was from a "faraway coun-
table (e.g., cow for boat), responded to re-
try, where everyone has blue hair [turns pup-
quests as before, and, in addition, answered
pet around to show its blue hair], walks on
questions about the object's characteristics

This content downloaded from 128.103.224.7 on Sun, 10 Feb 2019 21:10:12 UTC
All use subject to https://about.jstor.org/terms
 Rosenblum and Pinker 777

("Does this 'cow' have legs or a smokestack?"; by a nonsense name or by a different name
"Does it walk or does it sail?"; order of alter- was acceptable or unacceptable. The number
natives was counterbalanced). As before, the of bilinguals and monolinguals who agreed
child was asked why the renaming was or and disagreed with the possibility of calling
was not possible. a table a shig was identical (seven out of 12,
X2[1] = 0). Seven bilinguals and four mono-
Third, the child learned to call an object
linguals assented to calling a boat a cow, a
by the name of another object on the same
table (e.g., truck for giraffe). As before, the nonsignificant difference, X2 (1) = 1.51, N.S.
These results show that neither group was
child was asked to manipulate the object as
significantly more influenced than the other
referred to by its new name, was asked
by so-called word magic.
whether it had wheels or eyes and whether
it drives or eats, and was asked why the re- There were also no differences between
naming was or was not admissible. In addi- the monolingual and bilingual groups' perfor-
tion, in this case the child was asked, "If mance on all three of the name-learning tasks.
that's a truck, what is this?" A larger butAll 24 subjects successfully responded to com-
otherwise identical toy giraffe was shown,mands to manipulate the objects when they
giving the children the opportunity to gener-were referred to by a nonsense name (shig
alize from the name they had learned in the for table), a different name (cow for boat),
interview framework to another instance of and a switched name (truck for giraffe, when
the same object. a truck was also present). All children chose
the correct property for an object over a prop-
Results erty associated with the referent of its new
name, except for one monolingual child and
Comprehension of Counterfactuals one bilingual child when asked about the ob-
All children tested answered the three ject with the switched name. Only on the
third task-the generalization from one in-
questions correctly with ease, with the excep-
tion of the question, "Does he put his shoes stance of a newly learned switched name to
on the part of his body that has fingers or another ("If this is a truck what is this?")-
that has toes?" Two bilingual and one mono-did some of the subjects have difficulty. Four
lingual subject (out of the 24) answered thisof the monolinguals and three of the bilin-
question incorrectly. In view of the success guals treated the giraffe as a second instance
of the other 21 children, and of these chil- of the learned name truck, but the remaining
dren with the other two items, it seems rea- children did not make that generalization and
sonable to conclude that our children, both
called it a giraffe. None of these small dif-
ferences between monolinguals and bilinguals
mono- and bilingual, had little difficulty pro-
cessing information introduced in counterfac-was statistically significant.
tual situations.
Although all of the subjects performed
Control for Volubility similarly in their ability to give the correct
There was no significant difference be-answers in the name-manipulation tasks, the
tween the mean scores of the two groups intwo groups differed significantly in their ex-
volubility. The mean number of content wordsplanations for their performance when re-
for both questions for the monolinguals wassponding to the two questions "Why can/
12.3, and for the bilinguals was 10.5, t(22)can't you call this X a 'Y'?" or the three ques-
= 1.11, N.S. tions "Why can/ can't we now call this X a
On the basis of these two controls and 'Y'?" following the name changes. The sub-
jects' reasoning about the possibility of name
the measures reported in the Subjects section,
manipulation was assessed in the following
it can be concluded that any differences in
the performance of the two groups on the way. We excluded an putative explanation
name manipulations are unlikely to result that simply mentioneX the name of the ob-
from differences in the groups' socioeconomic
ject (e.g., "You can't call it a shig because
status, ability to accept counterfactual situ- it's a table") or that made no sense (e.g.,
ations, nonverbal intelligence, or volubility, "You can't call it a shig because it's nothing,"
but from the group's knowledge of a second or "It's not a shig, it's a pig, oink! oinkl").
language per se. If a child could give an interpretable reason
for his or her answer, it was then assigned
Name-Manipulation Test to one of two subcategories that we found,
The first two questions were hypotheti- post hoc, to account for most of the reasons:
cal inquiries into whether calling an object reasons based on objects' attributes ("because

This content downloaded from 128.103.224.7 on Sun, 10 Feb 2019 21:10:12 UTC
All use subject to https://about.jstor.org/terms
778 Child Development
they're both green"/!"because it doesn't have across the interview questions. Out of a pos-
green on it," or "because it has four legs"); sible score of +5 (all five answers were rea-
or reasons based on the experimental, situ- soned on the basis of context) to -5 (all five
ational, or personal context in which the name answers were reasoned on the basis of attri-
transfer was acceptable (e.g., "because it's butes), the monolingual mean score was
in our game," or "because you told me and -1.17, and the bilingual mean score was
I know... I know what it is and you know 1.17, t(22) = 2.05, one-tailed p <.05. This
what it is," or "because it's not true in life," difference remains significant even when the
or "because the kids [downstairs] wouldn't scores are adjusted for age in an analysis of
know . . . because we didn't tell them"). In covariance, t (21) = 1.88, one-tailed p < .05.
addition, in the last task (calling an object
by the name of another on the table), five of Discussion
the children (four monolinguals and one bi-
Our results show that middle-class pre-
lingual) gave reasons based neither on the
context nor attributes but on the ambiguity schoolers are not necessarily seduced by word
that would arise from having "two trucks." magic, do not necessarily treat an object's
name as inseparable from its attributes, and
These responses were assigned to a third cate-
do not necessarily fail to realize that an ob-
gory and excluded from the analysis described
below. ject can be given a new arbitrary name. All
24 of our subjects were able to identify a
The Hebrew responses were all trans- familiar object when it was referred to by
lated into English, and all the responses were a new, nonsense name and by a name ordi-
typed on slips of paper, which were given to narily used for a different object; 22 of the
a naive rater, unlabeled as to source and in a 24 were even able to understand a new name
random order. The rater then assigned them that referred to another object currently in
to "context" and "attribute" categories, agree- view. These children were also able, in simi-
ing in 95% of the cases with the first author's lar proportions, to answer questions about a
judgments. The second author, unaware of renamed object's attributes without imbuing
the group from which the child came, scored the object with the properties of the new
the responses on the remaining 5% of the name's conventional referent. Even in the
answers. hypothetical questions demanding more re-
flection on the part of a child ("Can you call
Two findings emerged from this this analysis.
table a shig/ cow?"), which children al-
First, in all five tasks, bilinguals offered more
legedly deny, about half of our subjects re-
reasons for their behavior than monolinguals.
sponded "yes" (58% and 46% for the nonsense
More interestingly, the monolingual name andandthebi-new name, respectively). And
lingual children differed in the types of rea-
it is possible that the children who answered
sons they gave: monolinguals referred to the in the negative may have
these questions
attributes of the object more often done than
so nottobecause of word magic but for
the context of the naming exercise in all outlined
the reasons five in the introduction-
questions, whereas bilinguals referred to the
that is, the question may have been construed
context more often than to the attributes in
as being about the object's true name in
four of the five questions. This English.
interaction
(In fact, the ideal adult response to
can be tested in two ways. First,thea child
question can
would be something li e "You
be categorized as "context" if thecould,
majority of but that's not its name,"
I suppose,
his or her reasons were based on reference
and if one was speaking to a person learning
to the context and "attribute" if the majority
English, even that response would be inap-
of reasons were based on the object's attri-
propriate. In an informal survey of six Stan-
butes. Nine of the 12 monolinguals based
ford students and staff, we found that all six
their responses predominantly on the object's
thought that the "correct" answer to the ques-
characteristics (categories were tied
tion is for
"no.") two
children), whereas eight of the 12 bilinguals
based their responses predominantly Weon cannot
thebe certain why our results
differ from previous ones. It could be, as we
context, x2 (1) = 5.04, p < .05. Second, one
can perform a t test on the mean proposed
scoresin the introduction, that our ques-
of
tions
each group, where each subject wasabout hypothetical situations set the
scored
on the number of reasons based on context child to entertain questions and requests con-
(out of five possible) minus the number of cerning counterfactuals; alternatively, our
reasons based on attributes (also out of five) children could have been more precocious

This content downloaded from 128.103.224.7 on Sun, 10 Feb 2019 21:10:12 UTC
All use subject to https://about.jstor.org/terms
 Rosenblum and Pinker 779

(they were all children of professionals), or Sparky may lack; animal may be used as
our procedure more congenial to the child, name for a class of creatures possessing a st
than those used in the past. wider class of attributes; and so on [see
Our second conclusion is that there is Anglin, 1977; Carey, 1978; Clark, 1973; Mac-
little evidence for a difference in awareness namara, 1982]). As a result, a monolingual
child might come to the conclusion that any
of the word-object relation between mono- object can have a variety of names, each
lingual and bilingual children when con-
being appropriate to the object only when
founding factors such as nonverbal intelli-
certain of its properties, the ones specified
gence, SES, volubility, and willingness to en-
tertain counterfactuals are controlled. Our two by the word's meaning, are to be emphasized.
In fact, such "attribute" responses are not
groups were equally likely to assent to thenecessarily wrong or silly; the following hypo-
possibility of calling an object by a new, non-
thetical adult dialogues, for example, have a
sense name; were equally able to learn new
nonsense, different, and switched names for
plausible ring to them: "Can you call garlic
a vegetable?"/"Yes, because it is an edible
an object; were equally able to identify an plant/No, because it is too small and pungent
object's intrinsic properties despite a distract-
to be eaten alone"; or "Can you call this com-
ing name change; and were equally unlikely
puter an engine?"/"Yes, because it is mechan-
to generalize a new name to a similar object.
ical and runs automatically/No, because it
Although there may be small quantitative dif- doesn't move."
ferences between mono- and bilingual chil-
dren that our measures, many of which were Bilingual children clearly must share this
at ceiling, were incapable of detecting, there experience with monolinguals; in fact, they
does not seem to be a dramatic difference in
provided similar numbers of answers referring
susceptibility to word magic, with monolin- to attributes as did the monolingual children
gual children utterly incapable of considering(1.4 vs. 1.7 reasons). However, the far great-
alternatives to the conventional name for an
er number of reasons they offered referring
object. to the naming context (2.5 vs. .5 reasons)
probably reflects their experience with a dif-
The third conclusion from this study is
that there indeed does seem to be a differ- ferent set of contingencies about names. A
bilingual child must learn that he or she can
ence between monolingual and bilingual chil-
give a dog the name dog when addressing
dren's awareness of the word-object relation,
a certain person or type of person or when
even if that difference is not that one group
in a particular situation, but that that beast
is aware of the nature of the relationship and
must be given the name chien when address-
the other group is not. Rather, it is that the
ing other people or in a different setting. As
two groups articulate different aspects of that
a result, a bilingual child may realize that
relationship. Monolingual children are more
objects can have two different names by vir-
likely to refer to the physical properties of tue
an either of various subsets of its attributes
object when justifying or denying a new name
or by virtue of the various social contexts
for it (e.g., you can call the giraffe a truck
in which the object is named. The justifica-
because it has four legs and the hooves look
tions the children gave for renaming an ob-
like wheels, or you can't call a boat a cow
ject in this study reflect just that knowledge.
because it doesn't have legs), whereas bi-
lingual children were more likely to refer to
the social context of naming or the sharedReferences
knowledge that results (e.g., you can call it
a cow "because it's in our game"). These Anglin, J. M. Word, object, and conceptual de-
responses are perfectly consistent with what velopment. New York: Norton, 1977.
the child must deduce about word meanings Ben Ze'ev, S. The influence of bilingualism on cog-
when he or she learns words. For a mono- nitive strategy and cognitive development.
lingual child, objects are given several names, Child Development, 1977, 48, 1009-1018.
which are distinguished by virtue of the prop- Carey, S. The child as word learner. In M. Halle,
erties of the objects in their extensions (e.g., J. Bresnan, & G. Miller (Eds.), Linguistic
Sparky may be used to refer only to the dog theory and psychological reality. Cambridge,
with a specific appearance, behavior, and Mass.: MIT Press, 1978.
domicile; dog may be used as a name for anyClark, E. V. What's in a word? On the child's
creature within a far wider range of appear- acquisition of semantics in his first language.
ance and behavior, including attributes that In T. Moore (Ed.), Cognitive development

This content downloaded from 128.103.224.7 on Sun, 10 Feb 2019 21:10:12 UTC
All use subject to https://about.jstor.org/terms
780 Child Development
and the acquisition of language. New York: referent differentiation. Child Development,
Academic Press, 1973. 1976, 47, 742-749.
Cole, M., & Scribner, S. Culture & thought: A Osherson, D., & Markman, E. M. Language and
psychological introduction. New York: Wiley, the ability to evaluate contradictions and
1974. tautologies. Cognition, 1975, 3, 213-226.
Cummins, J. Bilingualism and the development ofPiaget, J. The child's conception of the world.
metalinguistic awareness. Journal of Cross- London: Routledge & Kegan Paul, 1929.
Cultural Psychology, 1978, 9, 131-149. Pinker, S. M. Bilingual and monolingual children's
Feldman, C., & Shen, M. Some language-related awareness of words. Unpublished bachelor's
cognitive advantages of bilingual five-year- thesis, McGill University, 1979.
olds. Journal of Genetic Psychology, 1971,Sandoval, J. Aspects of cognitive development in
118, 235-244. the bilingual: An exploratory study of word-
Grice, H. P. Logic and conversation. In P. Cole object separation. Unpublished master's the-
& J. L. Morgan (Eds.), Syntax and seman- sis, UCLA, 1976.
tics. (Vol. 3): Speech acts. New York: Semi-Siegel, L. Is the young child really preoperational?
nar, 1975. In L. S. Siegel & C. J. Brainerd (Eds.),
lanco-Worrall, A. D. Bilingualism and cognitive Alternatives to Piaget: Critical essays on the
development. Child Development, 1972, 43, theory. New York: Academic Press, 1978.
1390-1400.
Slobin, D. A case study of early language aware-
Leopold, W. F. Speech development of a bilin- ness of a bilingual. In A. Sinclair, R. J. Jar-
gual child. (Vol. 4): Diary from age two. vella, & W. J. M. Levelt (Eds.), The child's
Evanston, Ill.: Northwestern University conception of language. New York: Springer-
Press, 1949. Verlag, 1978.
McLaughlin, B. Second language acquisition in Vygotsky, L. S. Thought and language. Cam-
childhood. Hillsdale, N.J.: Erlbaum, 1978. bridge, Mass.: MIT Press, 1962.
Macnamara, J. Names for things. Cambridge, Wechsler. D. Manual for the Wechsler Preschool
Mass.: MIT Press, 1982. and Primary Scale of Intelligence. New York:
Markman, E. M. Children's difficulty with word- Psychological Corp., 1963.

This content downloaded from 128.103.224.7 on Sun, 10 Feb 2019 21:10:12 UTC
All use subject to https://about.jstor.org/terms