Copyright 1995 by the American Psychological Association, Inc.

Journal of Educational Psychology " 0022-0663/95/53.00

1995, Vol. 87, No. 2, 241-252

Identity and the Language of the Classroom: Investigating the

Impact of Heritage Versus Second Language Instruction on
Personal and Collective Self-Esteem

Stephen C. Wright Donald M. Taylor

University of California, Santa Cruz McGill University

The connection between heritage language instruction and self-esteem was investigated.
Participants were Inuit, White, and mixed-heritage (Inuit-White) children living in a subarc-
tic community. Testing occurred before and after their 1st year in a heritage language or a 2nd
language program. Children from all 3 groups who were educated in their heritage language
showed a substantial increase in their personal self-esteem, whereas Inuit and mixed-heritage
children educated in a 2nd language did not. Among the Inuit, Inuttitut instruction was
associated with positive regard for the ingroup, whereas English or French instruction
was associated with preference for the White outgroup. The present findings support claims
that early heritage language education can have a positive effect on the personal and
collective self-esteem of minority language students—a benefit not provided by 2nd language
instruction.

In the present study, we investigated the differential ef- dian Native children. In residential schools in both Canada
fects of early education in the heritage language versus early and the United States, children were verbally berated, pub-
immersion in a second language on the child's personal and licly humiliated, and even physically punished for using
collective self-esteem. The question of language of instruc- their heritage language (Freeman, 1978; Jordon, 1988; Pel-
tion has been the center of considerable public and aca- letier, 1970; Platero, 1975). In fact, eradication of the
demic debate. Traditional thinking held that early entry into child's heritage language and identity was an explicit goal
English language education would speed the minority lan- of many residential and missionary schools (Cummins,
guage student's transition into the majority culture and 1990; Tschantz, 1980). Although the tactics may have mel-
improve his or her chances of competing in the mainstream lowed, we continue to find considerable support for the
society. In many cases, zealous supporters of this assimila- assimilationist-"English only" position.
tionist perspective set out to replace the student's heritage In contrast to this view is the movement toward early
languages with the dominant language. Some of the clearest heritage language education. Many of these programs in-
and most dramatic examples of this can be found in North volve the use of a bilingual classroom, in which the child's
America's history of educating Native American and Cana- heritage language and the dominant second language are
both used. Less frequently, early instruction is offered ex-
Stephen C. Wright, Psychology Department, University of clusively in the child's heritage language.
California, Santa Cruz; Donald M. Taylor, Psychology Depart- The controversy over heritage language education has
ment, McGill University, Montreal, Quebec, Canada. involved political and practical as well as pedagogical is-
This research was supported by a grant from the Kativik School sues. Among the pedagogical issues are the child's general
Board and the Social Sciences and Humanities Research Council academic achievement, the pace of second language acqui-
of Canada. sition, maintenance of heritage language fluency, develop-
We thank the following people for their contributions to this ment of heritage language literacy, and the ease of transition
research project: the community, all the children who took part in
our study, and the Education Committee and staff at Jaanimmarik into the school environment (see August & Garcia, 1988;
School. We especially acknowledge the effort and time given by Crawford, 1989; Cummins, 1989; Cummins & Swain,
Principal Marguerite Kopiec. We also thank the testers: Michelle 1986; Genesee, 1987; Taylor, Meynard, & Rheault, 1977).
Auroy, Claudette Baron, Gaston Cote, Mary Elijassiapik, Annie The enormous variability in the types of programs and in the
Kudluk, Judy McArthur, and Sue McNicol. Their many hours of groups of students involved in heritage language education
hard work are greatly appreciated. Finally, we offer special thanks has posed difficulties for researchers in this area. Despite
to Doris Winkler and Mary Aitchison who provide continuous,
invaluable assistance with our work in Arctic Quebec and to Karen
these challenges and the opposition of English-only advo-
Ruggiero who has made important contributions to this research. cates, evidence indicates that minority language speakers
Correspondence concerning this article should be addressed to benefit from early instruction in their heritage language. At
Stephen C. Wright, Psychology Department, Clark Kerr Hall, present, three conclusions appear warranted. First, using
University of California, Santa Cruz, California 95064. Electronic children's heritage language in the school has a positive
mail may be sent via Internet to [email protected]. impact on their subsequent abilities in the heritage lan-
241
 242 STEPHEN C. WRIGHT AND DONALD M. TAYLOR

guage. It appears that the school can play a role in the sense of who she or he is includes at least two components
maintenance and enhancement of first-language skills (see or levels: the personal identity and the social or collective
Crawford, 1989). Second, heritage language instruction identity.
need not result in academic impairment. Minority students Personal identity involves those aspects of the self that
in programs that use their heritage language have, in some make the individual unique—personal attributes, skills, and
cases, shown faster academic progress (for reviews, see experiences. It is an evaluation of this level of identity that
Crawford, 1989; Cummins, 1989). Third, the common as- is tapped by most of the traditional work on self-esteem. For
sumption that the use of the heritage language will nega- this reason, we might appropriately think of most of the
tively affect the acquisition of English skills is clearly false. standard measures of self-esteem as measuring personal
In fact, there is evidence that heritage language instruction self-esteem. The other level of identity, the collective self,
may result in better performance in English in the long run involves those aspects of the individual that connect her or
(see Willig, 1985). him with others—group memberships. Just as there are
As the support for heritage language education grows, dimensions of personal self-esteem (see Marsh, Craven, &
some authors have claimed in the strongest terms that Debus, 1991; Marsh & Shavelson, 1985), there are likely to
greater use of heritage languages in school is a necessary be numerous group memberships that contribute to one's
remedy for present patterns of school failure among minor- collective self-esteem. Examples of these could be one's
ity students (Cummins, 1989, 1990; McLaughlin, 1989). In ethnic heritage, family, peer group, or, for a child, the
addition, these authors have looked beyond the linguistic classroom or school. This view of the self suggests that the
advantages of heritage language education and are pointing individual's feelings of self-worth involve an analysis not
to potential social benefits. Very often one of the stated or only of the personal or unique aspect of the self (i.e.,
implicit mechanisms by which heritage language education personal self-esteem), but also of the social or collective
is to improve academic success is through enhancement of aspects (i.e., collective self-esteem; Crocker & Luhtanen,
the child's self-esteem (see Appel, 1988; Cummins, 1989, 1990; Cross, 1987).
1990; Hernandez-Chavez, 1984). This view has gained con- Recognition of this distinction goes a long way in ex-
siderable acceptance among heritage language advocates. plaining the contradictory findings concerning self-esteem
It is widely accepted that children who think highly of and children from traditionally disadvantaged minority
themselves stand a much better chance of being successful groups. At a personal level, there are a number of effective
in school. Strong correlations between self-esteem and ac- mechanisms by which disadvantaged group members can
ademic success provide support for this belief (Covington, maintain high personal self-esteem (see Crocker & Major,
1989; Harter, 1986). The causal direction of this relation- 1989). However, faced with societal and educational struc-
ship is controversial. It has been proposed that rather than tures that contain both subtle and overt reminders of the low
higher self-esteem leading to increased achievement, it is status held by their group, minority children can suffer from
the success associated with high achievement that leads to a reduced collective self-esteem. A tradition of research that
more positive evaluation of the self (Rosenberg, Schooler, dates to the seminal doll studies of Clark and Clark (1939)
& Schoenbach, 1989). Although further investigation of the demonstrates a lower evaluation of the ethnic ingroup by
question of causal direction is needed, most educators be- young Black children. Others have replicated these results
lieve that self-esteem is an important determinant of aca- with Canadian Native (Corenblum & Annis, 1987; George
demic success. Evidence of the breadth of this acceptance is & Hoppe, 1979), Mexican American (Weiland & Coughlin,
found in the numerous programs that have been imple- 1979), and Chinese American children (Aboud & Skerry,
mented in an attempt to increase self-esteem with the ex- 1984). This lower evaluation of the ingroup is indicative of
pectation that this will translate into enhanced academic a negative collective self-esteem.
performance.

Heritage Language Education and Self-Esteem

Personal and Collective Self-Esteem
Heritage language education may have a positive effect
At present, considerable disagreement exists in the liter- on both personal and collective self-esteem. At the personal
ature on self-esteem and minority children (see Porter & level, minority language children will be spared the frustra-
Washington, 1979). Much of this controversy can be traced tion of not understanding much of what goes on in the
to a failure to recognize the distinction between two differ- (majority language) classroom (which can lead to poorer
ent dimensions or levels of self-esteem. Until recently, the performance and an associated lower self-evaluation). In
psychological literature has shared with mainstream North addition, heritage language education spares children from
American culture a conception of the self as an independent, the negative self-evaluation that is likely to result from
self-contained, autonomous entity (Markus & Kityama, social comparison with majority language speakers, who are
1991; Moghaddam, Taylor, & Wright, 1993; Sampson, likely to be more successful than themselves in majority
1988). For the most part, self-esteem has been represented language classrooms. Finally, when language is associated
as the evaluation of one's personal characteristic, of one's with cultural differences, the minority student may be dis-
independent self. Contrary to this view is the growing tanced from the activities and interpersonal interactions in
recognition that a child's (and an adult's, for that matter) the majority language classroom, and psychological isola-
 IDENTITY AND THE LANGUAGE OF THE CLASSROOM 243

tion could contribute to a negative self-image. These chil- alternatives. Inuit, White, and mixed-heritage children liv-
dren are less likely to face these problems in a heritage ing in an isolated subarctic community and attending school
language classroom. in Inuttitut, English, or French were tested at the beginning
In North America, considerable importance is placed on and the end of their kindergarten year.
skills and knowledge acquired in school. If English is the The present study expanded on previous strategies by
sole language of instruction, children may be led to believe including measures of both personal and collective self-
that the important knowledge and skills imparted at school esteem. Each child was shown a set of nine photographs,
must be learned in English. Thus, the English language is including four ingroup and four outgroup members as well
directly paired with the knowledge and skills necessary for as a photo of himself or herself. Personal self-esteem was
achievement and status. In contrast, if children who speak measured by the frequency with which the child selected his
the minority language are characterized as "deficient" or "in or her own photograph in response to positive attributes and
need of remediation," the child's heritage language is paired excluded it in response to negative attributes. Due to the
with lower status. The status of those who speak English is lack of previous research with Inuit children, no a priori
further enhanced in that most, if not all, of the high-status predictions were made about ethnic differences in personal
people at school (teachers, principals, etc.) speak English. self-esteem. However, it was hypothesized that when com-
As the minority language child becomes aware that mem- pared to second-language instruction, heritage language ed-
bership in the majority language group is an important ucation would have a positive impact on the children's
determinant of success, the relative value of his or her personal self-esteem.
ingroup is reduced. In all these cases, group level social Two measures of collective self-esteem were used. The
comparisons made in an English-only context can suggest to first involved comparing the children's evaluation of the
the minority child, either directly or subtly, that the majority ethnic ingroup with their evaluation of the ethnic outgroup.
group is inherently superior to his or her ingroup. When The frequency with which the children selected members of
these messages are combined with a general societal deval- their ethnic ingroup in response to positive attributes and
uation of the heritage language, the result may well be excluded them in response to negative attributes was used as
lowered collective self-esteem. a measure of their general evaluation of their ethnic ingroup.
The use of the heritage language as the medium of in- Similarly, the children's pattern of selecting and excluding
struction, on the other hand, is a clear affirmation of the members of the ethnic outgroup provided a measure of their
value and status of the heritage language and of those who general evaluation of the ethnic outgroup. A comparison of
speak it. When heritage language education involves coeth- these two measures provided an indication of the children's
nic educators, these models will affirm that ingroup mem- collective self-esteem. The second measure of collective
bers can hold high-status positions. The important skills self-esteem was the strength of the children's preference for
learned at school are not distinguished by language, and the ingroup versus outgroup members as potential friends.
children are not presented with a direct connection between
their language group and poorer performance. Several authors have claimed that a healthy collective
Despite the wide acceptance of the potential for heritage self-esteem should result in a mild preferential evaluation of
language education to have a positive impact on self-es- the ingroup. This should be accompanied by a mild prefer-
teem, very little research has tested this belief. Alexander ence for ingroup members as friends or friendship selection
and Baker (1992) argue that without such evidence it is without consideration of ethnicity (see Aboud, 1988). How-
irresponsible and even unethical to continue to promote ever, a large differential evaluation in favor of the ingroup
heritage language education as a mechanism for increasing or a near exclusive preference for ingroup members as
self-esteem. The present research provides a direct test of friends may reflect ethnocentric attitudes. A bias either in
the proposed relationship between language programs (ma- evaluation or friendship selection toward the outgroup, on
jority language vs. heritage language) and self-esteem. This the other hand, would be indicative of a devaluation of one's
analysis is also extended by including measures of both ethnic group and less than ideal collective self-esteem.
personal and collective self-esteem. Again, the lack of previous research with Inuit children
lead to no a priori predictions concerning ethnic differences
on the general level of collective self-esteem. However, it
Present Research Paradigm was hypothesized that, when compared to second language
instruction, heritage language instruction would result
The present procedures arise out of a history of research in higher collective self-esteem as measured by both the
dating back to the early Black and White doll studies of relative evaluations of the ethnic groups and friendship
Clark and Clark (1939; for reviews, see Aboud, 1988; preferences.
Brand, Ruiz, & Padilla, 1974; Williams & Morland, 1976).
To avoid problems found with previous studies of this type
(Aboud, 1988; Corenblum & Wilson, 1982; Newman, Liss,
& Sherman, 1983; Trent, 1964), the present research used
Method
coethnic testers; photographs rather than drawings or dolls This study was part of a longitudinal research project designed
as stimulus materials; and a multiple-alternative method to investigate the impact of heritage language education versus
rather than a forced choice between only a pair of stimulus second language education across a number of domains.
 244 STEPHEN C. WRIGHT AND DONALD M. TAYLOR

Community French (Francophones), with the remaining 2 being English speak-

ers (Anglophones). English was the first language for 30 of the
The community that served as the focus for this study is located mixed-heritage children, with 3 speaking French and 2 speaking
in the region of Northern Quebec, Canada, known to its inhabitants Inuttitut.
as Nunavik. This vast subarctic region contains 14 isolated com- Over the 4-year period, 36 Inuit children were enrolled in the
munities. The present study was conducted in the largest of these Inuttitut language program, 15 were enrolled in the French pro-
communities, whose population of 1,400 is made up of approxi- gram, and 13 were enrolled in the English program. Among the
mately 80% Inuit, 12% Francophones, and 8% Anglophones (Tay- White children, all of the Francophones (n = 11) were enrolled in
lor & Wright, 1989). the French program, one of the Anglophones enrolled in the
The Inuit of Nunavik remained extremely isolated from the English program, and the other enrolled in the Inuttitut program.
mainstream Canadian-American society until as late as the mid- Mixed-heritage children were almost equally divided between the
1950s, and the communities of Nunavik remain relatively isolated Inuttitut (n = 14), English (n = 11), and French (n = 11)
even today. They are accessible only by air, and many Inuit programs.
residents have never seen an urban center. Despite a growing trend
toward intermarriage, contact between the Inuit and White groups
of residents (particularly the Francophone and Inuit populations) Procedures and Materials
remains minimal. Most Inuit children will have had very little
contact with Whites prior to entering school. The self-esteem task was completed as part of a battery of tests
More than 90% of the Inuit from this region claim Inuttitut (the that included measures of language, arithmetic, and spatial ability.
heritage language of the Inuit) as their first language. Compared to The self-esteem task was always administered in the child's her-
virtually all other Native languages in North America, in Nunavik itage language by a native speaker of that language. Children were
Inuttitut remains a highly functional and vibrant language. In fact, taken from their school classes during regular instruction and
it has been described by several researchers as one of the few tested individually. Each child was tested twice, once during the
Native languages in North America that has a good chance of first week of the school year (fall) and again at the end of the
long-term survival (Foster, 1982; Priest, 1985). Despite these school year (spring).
optimistic claims, many have become concerned about the erosion The tester first took a photograph of the child. This photo was
of the language. Similarly, concerns have been raised about the added to a set of eight "head-and-shoulder" photos of children who
extent to which the growing White population exerts economic and were approximately the same age as the participant. Each set of
political control over the people and lands of Nunavik. photos contained 4 Inuit children (2 boys and 2 girls) and 4 White
children (2 boys and 2 girls) who were not known to the partici-
pants. The eight target photos were drawn from a larger set of
Participants and Programs photos that were pretested using four Inuit and four White adult
raters. Photographs of Inuit and White children were matched for
The participants included every child who entered kindergarten physical attractiveness, facial expression, and photograph quality.
over the 3-year period between 1989 and 1991 and approximately The child was asked to sort the photos 11 times on the basis of
half the children entering in 1992. Thus, data were collected from different characteristics. Testers used a standard request on each of
four cohorts. School board policy allows parents to register their these sorting trials: "Pick all the children who
children in one of three language programs in kindergarten: Inut- [are smart, nice, etc.] and put them here, and leave all the children
titut, English, or French. All instruction and most classroom ma- who are not here" (pointing to the surface in
terials are in the language of that program. The school board has front of the child). Before each request the tester shuffled the
made a considerable effort to provide books and other materials in photos and placed them randomly on the surface in front of the
Inuttitut for the Inuttitut program and to provide materials that child.
reflect Northern-Inuit culture in all three languages (Taylor, In the first three requests the tester asked the child to pick out
1990). However, White teachers in the English and French pro- the girls, the boys, and the Inuit. These requests served three
grams do make use of mainstream Canadian-American materials, functions. First, they ensured that the child understood the task.
most of which reflect mainstream White culture. Also, some Inuit Second, they were used to determine whether the children could
teachers occasionally use English materials (i.e., films, posters) in appropriately identify the children in the photos and whether they
the Inuttitut program. In Grade 3 the Inuttitut program is termi- were able to use gender and ethnicity categories. Finally, the third
nated, and children enroll in either English or French. question served as an initial measure of the child's ethnic self-
Turnover rates among White teachers are high in these isolated categorization.
northern schools, and many White teachers who do stay choose to In the next seven trials the child was asked to pick all the
move into higher grade levels. In addition, Inuit teachers some- children who are smart, nice, happy, and good at many things;
times choose to stay with their specific group of children for those who have lots of friends; those who like to go to school; and
kindergarten through Grade 2. Thus, over the 4-year period during those who the other children don't like. A score of 1 was given
which testing took place, there were three different English teach- each time the child included a photo in response to a positive
ers and a different Inuttitut and French teacher each year. The attributes and a score of - 1 was given if he or she included the
exact ethnic composition of each of the three language programs photo in response to the one negative attribute.
was different in each of the 4 years. However, in every case, The frequency with which the children selected themselves in
French classes contained a mixture of Inuit, White, and mixed- response to six positive attributes and failed to pick themselves in
heritage children, whereas the Inuttitut and English classes con- response to the single negative attribute provided a measure of
tained a mixture of Inuit and mixed-heritage children. their personal self-esteem. Thus, total scores could range from — 1
The final sample contained 64 Inuit, 13 White, and 36 mixed- to 6.
heritage (Inuit-White) children. Inuttitut was the first language of Separate scores were computed for the child's evaluation of
62 of the Inuit participants. The other 2 were native speakers of Inuit and White targets. These scores consisted of the frequency
English. Eleven of the White children were native speakers of with which the child selected the four Inuit targets (or the four
 IDENTITY AND THE LANGUAGE OF THE CLASSROOM 245

White targets) in response to six positive attributes and ignored girl—accounted for 83% of the errors. The possibility that
them in response to the negative attribute. Thus, each of these two there was some ambiguity about the gender of this particular
scores could range from - 4 (selecting all four Inuit targets only in child further supports the conclusions that the participants
response to the negative attribute) to 24 (selecting all four Inuit were able to comprehend the task and were aware of the
targets only in response to the six positive attributes). The child's gender distinction.
collective self-esteem along the dimension of ethnicity was as- Finally, two Inuit children were dropped from further
sessed by comparing his or her evaluation of ethnic ingroup and
outgroup targets. A healthy or strong collective self-esteem was
analyses because the accuracy of their answers on the ques-
demonstrated by a slightly higher score for the ingroup compared tions concerning gender (as well as the question on ethnic-
to the outgroup. A preferential evaluation of the outgroup would be ity) were less than chance. We interpreted this to mean that
indicative of a negative or low collective self-esteem. Finally, a they were unable to understand the nature of the task.
very high rating of the ingroup combined with a low evaluation of
the outgroup can be interpreted as demonstrating ethnocentric Ethnicity
attitudes.
In the final sorting trial the tester asked the child to pick the Self-Categorization. In the fall, 92% of the Inuit chil-
children that he or she would like to have as best friends. Separate dren identified their own photo as Inuit. This number in-
scores were computed for the child's interest in White and Inuit creased to 97% in the spring. In the fall, 93% of the White
targets as potential friends. These two scores were the total number
of Inuit and White targets selected. Thus, these two scores could
children identified their own photo as White, increasing to
range from 0 (no Inuit children selected) to 4 (all Inuit children 100% in the spring. These data indicate considerable accu-
selected). The child's relative preference for ingroup or outgroup racy in the self-categorization of both Inuit and White
members as friends provided a second indication of collective children, with the expected increase in accuracy over the
self-esteem. school year.
For mixed-heritage children the question of accuracy
becomes somewhat unclear. In a very real sense, these
Data Analysis children are "accurate" if they place themselves in either of
Separate analyses were performed on these three dependent these two ethnic categories, and the forced-choice format
measures—personal self-esteem, ingroup versus outgroup evalua- used in this question did not allow children to select both.
tion, and friendship preference. Unfortunately, only one White However, most mixed-heritage children willingly selected
child was enrolled in a second language program. The absence of an ethnic category for themselves, and the forced-choice
this cell precluded the performance of the complete factorial procedure provided interesting information about the child's
design. Therefore, for each dependent measure a general analysis strongest identification.
comparing the three ethnic groups was followed by separate anal- At the beginning of their kindergarten year (fall), 52% of
yses for each ethnic group. These separate analyses assessed the
generality of any overall findings and tested the impact of lan-
the mixed-heritage children identified themselves as Inuit.
guage of instruction on Inuit and mixed-heritage children. At the end of the year (spring) the number had dropped to
In all cases, simple main effects tests were used to investigate 39%. This difference failed to reach traditional levels of
interaction effects that result in the omnibus analysis. Where post statistical significance.
hoc pairwise comparisons of means were called for, Tukey hon- Categorization of others. A (3 X 2 X 2) Ethnicity of
estly significant difference (HSD) procedure was used (Winer, Participant (Inuit, White, Mixed Heritage) X Time of Test-
Brown, & Michels, 1991). ing (fall, spring) X Ethnicity of Target (Inuit, White) mixed
ANOVA was performed on the accuracy of the participants'
responses to the question concerning ethnicity of the target
Results children in the photos. This analysis yielded a significant
Preliminary Analysis main effect of time of test, F(l, 100) = 14.14, p < .001,
MSE = 7.0, indicating that the children were significantly
Gender more accurate about the ethnicity of targets at the end of
their kindergarten year (spring, 88%) than at the beginning
A (3 X 2 X 2) Ethnicity of Participant (Inuit, White, (fall, 77%).
mixed-heritage) X Time of Testing (fall, spring) X Ethnic- The main effect of ethnicity of target was also significant,
ity of Target (Inuit, White) mixed analysis of variance F(l, 100) = 6.68, p < .05, MSE = 5.1, but was subsumed
(ANOVA) was performed to investigate the accuracy of the under a significant Ethnicity of Participant X Ethnicity of
children's responses to the questions concerning the gender Target interaction, F(2,100) = 3.79, p < .05, MSE = 5.11.
of the children in the photos. This analysis yielded no Tests of simple main effect confirmed that White children
significant effects. Most of the children accurately catego- were significantly more accurate in their categorization of
rized the photos by gender at both testing occasions (86% members of the outgroup (89%) than they were at catego-
accuracy in the fall and 89% accuracy in the spring). This rizing members of their ingroup (68%), F(\, 12) = 4.62,
accuracy rate is consistent with previous research. In addi- p < .05. Inuit children also showed a similar, although not
tion, the vast majority of children appeared to understand statistically significant, tendency (outgroup, 88% and
the instructions and the nature of the task. ingroup, 83%). Mixed-heritage children showed little dif-
Additionally, an analysis of each of the target photos ference in the accuracy with which they identified the
demonstrated than one particular photograph—one Inuit ethnicity of the targets (Inuit, 81%; White, 80%).
 246 STEPHEN C. WRIGHT AND DONALD M. TAYLOR

Personal Self-Esteem test failed to reach traditional levels of significance. How-

ever, a strong effect size (TJ2 = .15)1 for this comparison
A personal self-esteem score was determined by the fre- appears to indicate a real increase in personal self-esteem
quency with which the child included himself or herself in for our sample of White children.
response to positive attributes and excluded himself or
herself in response to the negative attribute.
Inuit Children
Comparison of Three Ethnic Groups Inuit children enrolled in the French and English pro-
grams were combined into a single "second language pro-
A (3 X 2) Ethnicity of Participant X Time of Test mixed grams" group and were compared with those enrolled in the
ANOVA was performed. Both the main effect of ethnicity heritage language program (Inuttitut). A (2 X 2) Language
of participant, F(2, 101) = 11.45, p < .001, MSE = 1.95, of Instruction X Time of Test mixed ANOVA yielded a
and the main effect of time of test, F(l, 101) = 3.94, significant interaction, F(l, 57) = 3.89, p < .05, MSE =
p < .05, MSE = 1.75, were significant. 2.28, which is illustrated in the middle section of Table 1.
This analysis indicated no interaction effect between Although children in both language programs began the
these two variables (F < 1.0). Therefore, to investigate the year with almost identical scores, subsequent simple main
main effects of ethnic group, personal self-esteem scores effects tests confirmed that only the children in the heritage
were collapsed across the two testing times. Post hoc Tukey language program (Inuttitut) showed a significant increase
hsd tests indicated that this effect resulted from both White in personal self-esteem F{\, 31) = 6.85, p < .05, MSE =
(M = 5.46, SD = 0.87) and mixed-heritage (M = 5.34, SD 2.43.
= 0.81) children having significantly higher (p < .05)
personal self-esteem scores than Inuit children (M = 4.38,
SD = 1.09). Mixed-Heritage Children
The main effect of time of test indicated that, on average, Mixed-heritage children were also divided into two
the children showed an increase in self-evaluation over their groups: those receiving instruction in their heritage lan-
first year in school (fall, M = 4.56, SD =1.55; spring, M = guage and those receiving instruction in a second language.
5.04, SD = 1.30). A (2 X 2) Language of Instruction X Time of Test mixed
Separate analyses were performed for each of the three ANOVA was performed on the personal self-esteem scores.
ethnic groups. These analyses investigated the robustness of The Language of Instruction X Time of Testing interaction
this increase in personal self-esteem across the three groups approached traditional levels of significance, F(l, 34) =
and also allowed for the inclusion of the third variable of 2.64, p = .11, MSE = 0.82 (see the third section of Table
language of instruction for the Inuit and Mixed-Heritage 1). The similarity of this pattern to that of the Inuit children,
groups. combined with the relatively small sample size and a mod-
erate effect size (-n2 = .07), points to the potential impor-
White Children tance of this interaction. Mixed-heritage children in heritage
language instruction showed an increase in personal self-
As shown in the first row of Table 1, White children esteem from the fall to the spring. Conversely, those in
showed a considerable increase in personal self-esteem second language evidenced a small reduction in personal
from the fall to the spring. However, due to the very small self-esteem over this period.
sample size for this group and a possible ceiling effect in the
spring (M = 5.82 on a scale bound by 6.00), the statistical
Collective Self-Esteem: Ingroup vs.
Outgroup Evaluations
Table 1 The frequency with which ethnic ingroup targets were
Personal Self-Esteem Scores for White, Inuit, and Mixed- selected in response to positive attributes (and were ignored
Heritage Children by Language Program at the in response to the negative attribute) was compared to the
Beginning and the End of Their Kindergarten Year frequency with which outgroup targets were selected in
Time of testing response to positive attributes (and were ignored in response
to the negative attribute). This comparison provided the first
Fall Spring measure of the child's collective self-esteem along the di-
Child's ethnicity and
language program M SD M SD mension of ethnicity.
White
Heritage language (n = 12) 5.10 1.55 5.82 1.29 1
Inuit Cohen (1988) has established a widely accepted set of con-
Heritage language (re = 32) 4.10 1.69 5.10 1.22 ventions for interpreting effect size (see also Cohen, 1992). The TJ2
Second language (n = 27) 4.16 1.63 4.10 1.49 statistic represents the amount of variance accounted for in the
Mixed-Heritage dependent variable by the effect. Cohen proposes the following
Heritage language (n = 14) 4.91 1.11 5.37 1.05 conventions for interpreting this statistic: small TJ2 = .01; medium
Second language (n = 22) 5.31 0.87 5.05 0.75 T)2 = .06; and large TJ2 = .14.
 IDENTITY AND THE LANGUAGE OF THE CLASSROOM 247

Comparison of Three Ethnic Groups Table 3

Evaluation of Inuit and White Targets by Inuit Children
A (3 X 2 X 2) Ethnicity of Participant X Time of Test X in Heritage Language and Second Language Programs
Ethnicity of Target ANOVA was performed on the ingroup Ethnicity of target photo
and outgroup evaluation scores. A significant main effect of
ethnicity of target, F(l, 101) = 18.76, p < .001, MSE = Inuit White
22Ad, was qualified by a significant Ethnicity of Participant Language program M SD M SD
X Ethnicity of Target interaction, F(2, 101) = 7.23, p <
.01, MSE = 22.46. Table 2 shows that White and mixed- Heritage language (n == 32) 13.92 3.05 12.62 3.97
Second language (n = 27) 12.96a 3.96 14.40b 3.30
heritage children evaluated White targets more positively
Note. Analyses of simple main effects indicate that scores in the
than Inuit targets. Inuit children, on the other hand, showed same rows with different subscripts differ significantly (p < .05).
no difference in their evaluation of the two target groups. To
further explore these findings and to investigate the poten-
tial effects of language of instruction on collective self-
esteem, separate analyses were performed for each of the no interaction effect involving time of test (F < 1.0),
three ethnic groups. indicating that this pattern of bias was present both in the
White children. Because all but one of the White chil- fall and the spring.
dren received instruction in their heritage language, only the Mixed-heritage children. For the sample of mixed-her-
(2 X 2) Time of Testing X Ethnicity of Target repeated itage children, the main effect of ethnicity of target, F(l,
measures ANOVA was possible. A significant main effect 31) = 6.45, p < .05, MSE = 23.75, was the only significant
of ethnicity of target, F(l, 11) = 6.75, p < .05, MSE = effect in the ( 2 X 2 X 2 ) Time of Test X Ethnicity of
49.38, indicated that White children evaluated White targets Target X Language Program mixed ANOVA. This effect
more positively than Inuit targets (see the first line of Table indicates that mixed-heritage children were significantly
2). Although the two-way interaction only approached tra- more positive in their evaluations of White targets than Inuit
ditional levels of significance, F(l, 11) = 2.87, p = .11, targets (see the third line in Table 2).
MSE = 10.91, it was associated with a strong effect size Due to the mixed ethnicity of these children, it is difficult
(TJ2 = .21). Although White children evaluated White tar- to determine how this "pro-White bias" should be inter-
gets more positively than Inuit targets at both testings, this preted in terms of collective self-esteem. In a very real sense
White bias was greater in the spring (for White targets, M = these children belong equally to each of the two ethnic
15.25, SD = 5.15; for Inuit targets, M = 8.37, SD = 5.53) groups. However, one means by which to consider this issue
than in the fall (for White targets, M = 18.33, SD = 5.87; is to use the child's own self-identification. Recall that
for Inuit targets, M = 14.68, SD = 6.48). approximately 40% of these children identified themselves
as Inuit. If the pro-White bias simply represents a more
Inuit children. Inuit children were divided into two
positive view of the ingroup, we should predict that only
groups, those enrolled in second language programs (French
those mixed-heritage children who identified themselves as
or English) and those enrolled in the heritage language
White should show this bias. Those identifying themselves
program (Inuttitut). A (2 X 2 X 2) Time of Test X Ethnicity
as Inuit should not show this pattern.
of Target X Language Program mixed ANOVA yielded a
significant Language Program X Ethnicity of Target inter- A (2 X 2) Self-Categorization (Inuit, White) X Ethnicity
action, F(l, 57) = 7.00, p < .05, MSE = 15.41. Subsequent of Target mixed ANOVA yielded a main effect only for the
tests of simple main effects confirmed that this interaction ethnicity of target, F(l, 31) = 5.84,/? < .05, MSE = 23.51.
(see Table 3) results because Inuit children in the heritage No effect of self-categorization emerged (F > 1.0). The
language program showed a bias toward ingroup members, bias of mixed-heritage children who identified themselves
whereas Inuit children in second language programs exhib- as Inuit was virtually identical to the bias of those who
ited a significant bias toward the outgroup F(l, 26) = 4.48, identified themselves as White. Both groups evaluated
p < .05, MSE = 6.04. It should also be noted that there was White targets more positively than Inuit targets.

Collective Self-Esteem: Friendship Preferences

Table 2
Evaluation of Inuit and White Targets by Children From The final set of analyses involved children's selection of
Each of the Three Ethnic Groups Inuit versus White targets as friends. Friendship selection is
Ethnicity of target photo seen as a second indicator of the child's attitudes toward the
ingroup and the outgroup and, therefore, as another measure
Inuit White of collective self-esteem.
Child' s ethnicity M SD M SD
White (n = 13) 11.53a 5.14 16.79b 4.73 Comparison of Three Ethnic Groups
Inuit (n = 59) 13.49 3.50 13.42 3.75
Mixed-Heritage (n = 33) 13.92a 3.65 16.12b 3.85 A (3 X 2 X 2) Ethnicity of Participant X Time of Test X
Note. Analyses of simple main effects indicate that scores in the Ethnicity of Target mixed ANOVA yielded significant main
same rows with different subscripts differ significantly (p < .05). effects of ethnicity of participant, F(2, 99) = 5.17, p < .01,
 248 STEPHEN C. WRIGHT AND DONALD M. TAYLOR

MSE = 2.08, and ethnicity of target, F(\, 99) = 21.53, p < Discussion
.001, MSE = 1.64. These main effects were subsumed
under a significant Ethnicity of Participant X Ethnicity of Several findings emerged in the preliminary analysis that
Target interaction, F(2, 99) = 4.44, p < .05, MSE = 1.64. are deserving of comments. Other researchers have found
The means displayed in Table 4 indicate that, although that the age at which children can accurately use ethnic
children from all three ethnic groups preferred White targets categories varies across ethnic groups and across target
as friends, this preference was larger for White and mixed- groups (for a review, see Aboud, 1988). Relatively accurate
heritage children than for Inuit children. To further inves- distinctions have been found to develop in children as
tigate this interaction and to consider the potential effects of young as 4 to 5 years and as old as 8 years. It appears that
language of instruction on friendship preference, separate our sample falls at the young end of this age continuum. At
analyses were performed for each of the three ethnic groups. the beginning of kindergarten, most participants appeared to
White children. Because all but one of the White chil- understand the ethnic labels and could use them to catego-
dren received instruction in their heritage language, only the rize themselves and other ingroup and outgroup members.
(2 X 2) Time of Testing X Ethnicity of Target ANOVA was The general salience of ethnic categories appears to be a
possible. A significant main effect of ethnicity of target, key determinant of the pace of development and the strength
F{\, 11) = 7.05, p < .05, MSE = 2.84, indicated that White of children's awareness of ethnic categories (see Ramsey,
children generally chose more White targets than Inuit 1987). It is possible that the present social context makes
targets (see the first line in Table 4). ethnic categories particularly salient, even to very young
The friendship choices of White children in our sample children. Inuit and White children are highly distinguishable
demonstrate a clearly ethnocentric pattern. White children by physical appearance, language, cultural practices, and
chose nearly three times as many Whites as Inuit as poten- social behavior. Even this small community has "ethnic
tial friends. In addition, several findings indicate that this neighborhoods," and meaningful informal social interaction
pattern of ethnocentrism increased over their kindergarten is minimal. Thus, before entering school Inuit and White
year. In the fall 30% of the friends chosen by White children children are unlikely to have had significant interaction. At
were Inuit. In the spring only 10% of their chosen friends the same time, the community is small, and White and Inuit
were Inuit. Frequency data show that the percentage of preschoolers are likely to be aware of each other. These
White children choosing at least one Inuit friend dropped conditions may be optimal for children to learn and accu-
from 58% in the fall to 25% in the spring. rately use ethnic categories.
Inuit children. A (3 X 2 X 2) mixed ANOVA on the Also of interest is the tendency of White children, and to
friendship choices of Inuit children yielded no significant a lesser degree Inuit children, to be more accurate in cate-
main effects of or interaction for time of testing, ethnicity of gorizing members of the ethnic outgroup than members of
target, or language program. Inuit children in both heritage their ingroup. On the surface, this finding is unexpected.
and second language programs showed no particular pref- Children are likely to have had much more interaction with
erence for ingroup or outgroup members as friends. members of their own ethnic group and should be much
Mixed-heritage children. The ( 2 X 2 X 2 ) Time of Test more familiar with the characteristics and features of in-
X Ethnicity of Target X Language Program mixed group members.
ANOVA for the friendship choices of mixed-heritage chil- However, the "ingroup overexclusion effect" (Leyens &
dren yielded only a significant main effect of ethnicity of Yzerbyt, 1993) provides a possible explanation for the more
target, F{\, 31) = 11.21, p < .01, MSE = 1.24. Mixed- accurate categorization of the ethnic outgroup. In this case,
heritage children chose significantly more White targets the ingroup is protected by excluding individuals whose
than Inuit targets (see the third line in Table 4). The pref- group membership is ambiguous. We are unaware of any
erence for Whites as friends appears to occur equally among other demonstration of this phenomenon with young chil-
mixed-heritage children in both heritage and second lan- dren, and continued research is clearly necessary. However,
guage programs. that White children were most inclined toward this strategy
supports this interpretation. It is the White children who
showed the greatest level of bias toward the ingroup in other
areas of our data. In summary, the present findings for
Table 4 ethnic categorization support the conclusion that, under
Selection of White and Inuit Targets as Friends by some circumstances, children as young as 5 and 6 years of
Children From Each of the Three Ethnic Groups age may engage in categorization strategies that in effect act
Ethnicity of target photo to protect their ethnic ingroup.
Inuit White
Child's ethnicity M SD M SD Personal Self-Esteem
White (n = 13) 0.70a 0.84 2.00b 1.34
Inuit (n = 59) 1.86 0.93 2.04 1.06 It appears that in general Inuit children demonstrated
Mixed-Heritage (n = 33) 1.21a 0.75 1.90b 0.90 lower personal self-esteem than their White and mixed-
Note. Analyses of simple main effects indicate that scores in the heritage peers. Several explanations are possible for this
same rows with different subscripts differ significantly (p < .05). finding. The first is a cultural explanation that recognizes
 IDENTITY AND THE LANGUAGE OF THE CLASSROOM 249

the collectivist orientation of Inuit culture (Crago, 1993; Collective Self-Esteem

Williamson, 1987). Although positive self-regard is clearly
an important aspect of Inuit identity, individual recognition The patterns of response on the ingroup and outgroup
and public acknowledgment of personal achievement are evaluation measures (see Table 2) and on the friendship
not as central in Inuit culture as they are to White Canadian- preference measure (see Table 4) show considerable differ-
American culture. Thus, Inuit children's lower personal ences between the three ethnic groups.
self-evaluation scores may reflect a general cultural orien-
tation toward personal humility. White Children
Alternatively, the lower scores of the Inuit children may
reflect less conviction about their competence in the context The responses of White children on both measures dem-
of school (Marsh et al. 1991). Despite the presence of a few onstrated a clearly ethnocentric pattern. White children
Inuit teachers, the school atmosphere generally reflects evaluated Inuit targets significantly more negatively than
mainstream Canadian-American culture. Thus, for most of White targets and showed a nearly exclusive preference for
the Inuit children, the school environment is very different White friends. In addition, it appears that White children
from their home environment. Ogbu (1992) and others have became increasingly biased against Inuit targets over the
described a number of potential difficulties associated with course of their kindergarten year. By the spring testing,
a home-school cultural mismatch. In the present case, the White children were nearly twice as positive in their eval-
inconsistencies between the school and home environments uation of White targets as compared to Inuit targets, they
may have reduced Inuit children's certainty about their selected over three times as many White as Inuit friends,
abilities and competencies. and only 25% of them chose any Inuit targets as friends.
The general finding concerning the role of language of Of course, it is impossible to say whether the school and
instruction on personal self-esteem appears to be that kin- classroom experiences were the direct cause of this increas-
dergarten instruction in heritage language was associated ing bias against Inuit targets. Numerous other agents (e.g.,
with increases in personal self-esteem, whereas instruction parents, siblings, television, and the community) could also
in a second language was not. As Table 1 illustrates, White, have contributed. Nevertheless, the role of the school is
Inuit and mixed-heritage children in heritage language worth considering. Many of these White children would not
classrooms showed a considerable increase on our measure have had any real contact with Inuit children prior to kin-
of personal self-esteem. On the other hand, Inuit children dergarten. In every case, the class was taught by a White
educated in a second language did not experience this teacher in the White children's heritage language. Most, if
increase. Mixed-heritage children receiving instruction in a not all, of the Inuit children in these classes did not speak or
second language actually showed a slight decline in their understand the language of instruction. It is very possible
self-evaluation. that these White children simply recognized the lower status
In terms of personal self-esteem, minority students en- of the Inuit children in their classroom and generalized this
rolled in second language kindergarten programs appear to to a more global negative evaluation of the group.
"miss out" on the benefits experienced by their counterparts In summary, it is not clear whether the particular circum-
who receive heritage language instruction. On-going longi- stances found in these classrooms served only to confirm
tudinal research may be able to determine the impact of this negative stereotypes that the children brought with them
differential kindergarten experience on long-term involve- from home or elsewhere in the White community, or
ment in education and academic success. However, these whether the classroom situation contributed to the develop-
findings do provide initial support for claims concerning the ment of new negative attitudes about Inuit children. Never-
potential benefits of early heritage language education. theless, these findings support the conclusion that the school
The lack of a sample of White children educated in a was, at the very least, ineffective in preventing growing
second language prevents us from commenting on the im- negative attitudes among White children towards their Inuit
pact of second language education on majority language classmates, attitudes that do not augur well for future inter-
speakers. Evidence from other research indicates that the group relations between these two groups.
outcomes for this group may be very different. Lambert, More generally, these findings point to the need for
Genesee, and their colleagues (Genesee, 1983, 1987; Lam- schools and teachers to take an active role in attempting to
bert, Genesee, Holobow, & Chartrand, 1993) describe very improve the attitudes of majority children in multilingual-
few negative and numerous positive consequences of sec- multicultural educational settings. North American class-
ond language education for majority language children. rooms are becoming increasingly culturally and linguisti-
However, many of these studies involved English children cally diverse. The climate of the intergroup relations in
in French-immersion programs, which are specifically de- these classrooms can be an important determinant of suc-
signed for non-French speakers. Perhaps if English speak- cess for minority language students. Decades of research on
ing children were placed in a Spanish, French, or Inuttitut the contact theory (see Pettigrew, 1986, for a review) dem-
class with first-language speakers and a unilingual teacher, onstrates that simple contact, such as sharing a classroom,
their self-esteem might also be threatened. Clearly, more does not necessarily improve intergroup attitudes. The
research is necessary to determine the relative importance of present data illustrate a situation where linguistic diversity
language status in the classroom versus language status in in the classroom was associated with increasingly negative
the society outside the classroom. intergroup attitudes among majority students.
 250 STEPHEN C. WRIGHT AND DONALD M. TAYLOR

Inuit Children It is likely that this general bias resulted from the recog-
nition of the advantaged status of Whites in this arctic
Inuit children generally showed a very mild (nonsignifi- community. That 34 of the 36 mixed-heritage children in
cant) preference for outgroup members as friends and equal our sample were native speakers of English or French is an
evaluation of target White and Inuit children. Although indication that the superior status of White culture may be
most authors have described a slight preference for ingroup reflected in these children's homes. In addition, television
members as most indicative of a healthy positive collective may transmit messages about the relative position of Whites
self-esteem, the roughly egalitarian pattern shown by these over Inuit and other minority groups. The lack of a signif-
children could be interpreted as fairly healthy. icant interaction involving language of instruction indicates
However, this general trend was qualified by the effect of that language used in the classroom may not have ef-
language of instruction. Whereas heritage language instruc- fected this preference for Whites. However, it is possible
tion was associated with a modest ingroup bias—a healthy that the dominated position of White culture in the school
collective self-image—second language instruction was as- environment served to reinforce existing pro-White bi-
sociated with a bias toward the majority outgroup—a pat- ases. More research is need to untangle the multiple po-
tern indicative of a poorer collective self-image. The lack of tential influences.
a three-way interaction effect including the variable of time The experience of mixed-heritage children is of growing
of testing indicates that these biases where just as strong general importance in the North American contexts. To
when the children entered school in the fall as they were at date, there is a relative paucity of research on the experience
the end of the school year. Therefore, it does not appear that of this group. The present findings appear to support the
the experience of the second language classroom was the position that young children of mixed heritage are inclined
direct cause of the lower collective self-esteem. to identify with their majority rather than minority heritage.
Children were not randomly assigned to language of Even in a community that is almost 80% Inuit, children with
instruction. Parental selection of language program resulted one Inuit parent and one White parent were consistently
in a division of participants such that Inuit children enrolled more positive about Whites.
in the second language programs arrived at kindergarten
with a bias toward Whites. However, the kindergarten ex-
perience in a second language did nothing to reduce this Conclusion
outgroup preference and improve Inuit children's collective
self-esteem. On the other hand, those enrolled in the Inut- The present sample appears in many respects to be rather
titut program arrived at kindergarten with a mild ingroup unique. These children live in a small isolated northern
preference. The heritage language experience did nothing to community where the White "majority" population is small.
disrupt that positive collective self-image, or to increase it to The lifestyle and cultural habits of Inuit children may be
the point of ethnocentrism. very disparate from those of other language minority chil-
In most educational contexts, children are not randomly dren. This apparent uniqueness may raise questions about
assigned to language programs. The present findings sup- the generality of these findings to children in other multi-
port the possibility that minority language families who lingual school contexts. However, on closer examination the
enroll their children in second language instruction might experiences of the present sample may not be as unlike the
also be providing the child with other information about the experience of other minority groups in North America as it
higher status and importance of majority cultural and lan- first appears.
guage. If this is the case, it is important that teachers in these The very limited exposure of the Inuit children in our
classrooms be particularly diligent in their efforts to sample to mainstream White culture may appear somewhat
strengthen minority language children's perceptions of their unlike the experience of other language minority children,
cultural and linguistic ingroup. most of whom live in large towns and cities. However,
research shows that, for the most part, neighborhoods
throughout North America remain ethnically and racially
Mixed-Heritage Children segregated (Bickford & Massey, 1991; Farley, 1984, 1985).
This reality leads to a situation where preschoolers from
Generally, mixed-heritage children showed preferential many ethnic and linguistic minorities have very little con-
evaluation of Whites and a preference for White friends. tact with majority group members until entering school.
These children's mixed ethnicity obfuscates interpretation Whether the isolation results from hundreds of miles of
of this pro-White bias. For mixed-heritage children who arctic tundra or several city blocks and a railway track, the
identify themselves as White, this pattern may represent a world of many preschoolers includes relatively little direct
very positive collective identity, even ethnocentrism. contact with ethnic outgroup members.
However, the consistency of this bias among those who The present context may also seem unusual because the
identified themselves as Inuit points to a more general linguistic minority is in reality the numerical majority in the
bias toward the societally advantaged ethnic group. The community and the school. However, this is the usual oc-
trend for fewer mixed-heritage children to identify them- currence in virtually every Canadian and many American
selves as Inuit over the course of the year reinforces this Native reservation schools. It is also the growing reality for
interpretation. Spanish speaking children in communities in California,
 IDENTITY AND THE LANGUAGE OF THE CLASSROOM 251

New Mexico, and a number of other states. In all these guage acquisition: An Inuit example. TESOL Quarterly, 26,
cases, the majority of students are members of an ethnic or 487-505.
Crawford, J. (1989). Bilingual education: History, policies, theory
linguistic minority, whereas the school is dominated by the
and practice. Trenton, NJ: Crane Publishing.
English language, White teachers, and White mainstream Crocker, J., & Luhtanen, R. (1990). Collective self-esteem and
culture. intergroup bias. Journal of Personality and Social Psychology,
Clearly, the unique elements of the present sample and 58, 60-67.
educational context cannot be dismissed. The generality of Crocker, J., & Major, B. (1989). Social stigma and self-esteem:
these findings must be tested with other minority language The self-protective properties of stigma. Psychological Review,
groups in different educational settings. In addition, longi- 96, 608-630.
tudinal work will help to determine whether any of the Cross, W. E. (1987). The two-factor theory of Black identity:
present findings has an impact on long-term school reten- Implications for the study of identity development in minority
tion and academic success. However, the present findings children. In J. S. Phinney & M. J. Rotheram (Eds.), Children's
provide initial insights into the experiences of ethnolinguis- ethnic socialization: Pluralism and development (pp. 117-133).
Newbury Park, CA: Sage.
tic minority and majority students in terms of personal and
Cummins, J. (1989). Empowering minority students. Sacramento,
collective self-esteem. They also provide some support for
CA: California Association for Bilingualism Education.
claims that early heritage language education can have a
Cummins, J. (1990, June). Language development among aborig-
positive impact on the personal and collective self-esteem of inal children in northern communities. Paper presented at the
minority language students—a benefit that does not appear Circumpolar Education Conference, Umea, Sweden.
to be provided by second language instruction. Cummins, J., & Swain, M. (1986). Bilingualism in education:
Aspects of theory, research and practice. New York: Longman.
Farley, R. (1984). Blacks and whites: Narrowing the gap? Cam-
References bridge: Harvard University Press.
Farley, R. (1985). Three steps forward and two back? Ethnic and
Aboud, F. (1988). Children and prejudice. Cambridge, MA: Basil Racial Studies, 8, 4-28.
Blackwell. Foster, M. (1982). Indigenous languages: Present and future. Lan-
Aboud, F., & Skerry, S. (1984). The development of ethnic iden- guage and Society, 7, 7—14.
tity: A critical review. Journal of Cross-Cultural Psychology, Freeman, M. A. (1978). Life among the qallunaat. Edmonton,
15, 3-34. Canada: Hurtig.
Alexander, S., & Baker, K. (1992). Some ethical issues in applied Genesee, F. (1983). Bilingual education of majority-language chil-
social psychology: The case of bilingual education and self- dren: The Immersion experiments in review. Applied Psycholin-
esteem. Journal of Applied Social Psychology, 22, 1741-1757. guistics, 4, 1—46.
Appel, R. (1988). The language education of immigrant workers'
Genesee, F. (1987). Learning through two languages: Studies of
children in The Netherlands. In T. Skutnabb & J. Cummins
immersion and bilingual education. Cambridge, MA: Newbury
(Eds.), Minority education: From shame to struggle (pp. 57-78).
House.
Philadelphia: Multilingual Matters.
George, D. M., & Hoppe, R. A. (1979). Racial identification,
August D., & Garcia, E. E. (1988). Language minority education
preference and self-concept. Journal of Cross-Cultural Psychol-
in the United States: Research, policy and practice. Springfield,
ogy, 10, 85-100.
IL: Charles C Thomas.
Bickford, A., & Massey, D. S. (1991). Segregation in the second Harter, S. (1986). Processes underlying the construction, mainte-
ghetto: Racial and ethnic segregation in American public hous- nance, and enhancement of the self-concept of children. In J.
ing. Social Forces, 69, 1011-1036. Sulls & A. Greenwald (Eds.), Psychological perspectives on the
self (pp. 136-182). Hillsdale, NJ: Erlbaum.
Brand, E. S., Ruiz, R. A., & Padilla, A. A. (1974). Ethnic identi-
fication and preference: A review. Psychological Bulletin, 81, Hernandez-Chavez, E. (1984). The inadequacy of English immer-
860-890. sion education as an educational approach for language minority
Clark, K., & Clark, M. (1939). The development of consciousness students in the United States. In Office of Bilingual/Bicultural
of self and the emergence of racial identification in Negro Education (Eds.) Studies on immersion education: A collection
preschool children. Journal of Social Psychology, 10, 591-599. of United States education (pp. 144-183). Sacramento: Califor-
Cohen, J. (1988). Statistical power analysis for the behavioral nia State Department of Education.
sciences. Hillsdale, NJ: Erlbaum. Jordon, D. F. (1988). Rights and claims of indigenous people:
Cohen, J. (1992). A power primer. Psychological Bulletin, 112, Education and the reclaiming of identity. The case of the Cana-
155-159. dian Natives, the Sami and Australian Aborigines. In T. Skut-
Corenblum, B., & Annis, R. C. (1987). Racial identity and pref- nabb-Kangas & J. Cummins (Eds.), Minority education: From
erence in Native and White Canadian children. Canadian Jour- shame to struggle (pp. 189-222). Philadelphia: Multilingual
nal of Behavioural Science, 19, 254-265. Matters.
Corenblum, B., & Wilson, A. E. (1982). Ethnic preference and Lambert, W. E., Genesee, F., Holobow, N., & Chartrand, L.
identification among Canadian Indian and White children: Rep- (1993). Bilingual education for majority English-speaking chil-
lication and extension. Canadian Journal of Behavioural Sci- dren. European Journal of Psychology of Education, 8, 3-22.
ence, 14, 50-59. Leyens, J.-P., & Yzerbyt, V. (1993, October). A social judgeability
Covington, M. V. (1989). Self-esteem and failure at school: Anal- approach to stereotypes. Paper presented at conference of So-
ysis and policy implications. In A. M. Mecca, N. J. Smelser, & ciety of Experimental Social Psychology, Santa Barbara, CA.
J. Vasconcellos (Eds.), The social importance of self-esteem (pp. Markus, H. R., & Kityama, S. (1991). Culture and the self:
72—124). Berkeley: University of California Press. Implications for cognition, emotion, and motivation. Psycholog-
Crago, M. B. (1993). Communicative interaction and second lan- ical Review, 98, 224-253.
252 STEPHEN C. WRIGHT AND DONALD M. TAYLOR

Marsh, H. W., Craven, R. G., & Debus, R. (1991). Self-concepts psychologies of the individual and their role in personal and
of young children 5 to 8 years of age: Measurement and multi- societal functioning. American Psychologist, 43, 15-22.
dimensional structure. Journal of Educational Psychology 83 Taylor, D. M. (1990, June). Carving a new Inuit identity: The role
377-392. of language in the education of Inuit children in Arctic Quebec.
Marsh, H. W., & Shavelson, R. (1985). Self-concept: Its multifac- Paper presented at the Circumpolar Education Conference,
eted, hierarchical structure. Educational Psychologist, 20, 107- Umea, Sweden.
123. Taylor, D. M., Meynard, R., & Rheault, E. (1977). Threat to ethnic
McLaughlin, D. (1989). The sociolinguistics of Navajo literacy. identity and second-language learning. In H. Giles (Ed.), Lan-
Anthropology and Education Quarterly, 20, 275-290. guage ethnicity and intergroup relations (pp. 99-118). London:
Moghaddam, F. M., Taylor, D. M., & Wright, S. C. (1993). Social Academic Press.
psychology in cross-cultural perspective. New York: Freeman. Taylor, D. M., & Wright, S. C. (1989). Language attitudes in a
Newman, M. A., Liss, M. B., & Sherman, F. (1983). Ethnic multilingual northern community. Canadian Journal of Native
awareness in children: Not a unitary concept. Journal of Genetic Studies, 9, 85-119.
Psychology, 143, 103-112. Trent, R. (1964). The color of the investigator as a variable in
Ogbu, J. U. (1992). Understanding cultural diversity and learning. experimental research with Negro subjects. Journal of Social
Educational Researcher, 21(8), 5-14. Psychology, 40, 280-284.
Pelletier, W. (1970). Childhood in an Indian village. In S. Repo Tschantz, L. (1980). Native languages and government policy: An
(Ed.), This book is about schools (pp. 18-31). New York: historical examination. Native Language Research Series, Cen-
Vintage Books. ter for Research and Teaching of Canadian Native Languages,
Pettigrew, T. F. (1986). The contact hypothesis reconsidered. In University of Western Ontario, Ontario, Canada.
M. Hewstone & R. Brown (Eds.), Contact and conflict in inter- Weiland, A., & Coughlin, R. (1979). Self-identification and pref-
group encounters (p. 165-195). Oxford, England: Basil Black- erences: A comparison of White and Mexican-American first
well. and third graders. Journal of Cross-Cultural Psychology, 10,
Platero, D. (1975). Bilingual education in the Navajo Nation. In R. 356-365.
C. Troike & N. Modiano (Eds.), Proceedings of the first inter- Williams, E., & Morland, J. K. (1976). Race, color and the young
American conference on bilingual education (pp. 54-61). Ar- child. Chapel Hill: The University of North Carolina Press.
lington VA: Center for Applied Linguistics Williamson, K. J. (1987). Consequences of schooling: Cultural
Porter, J. R., & Washington, R. E. (1979). Black identity and discontinuity amongst the Inuit. Canadian Journal of Native
self-esteem: A review of studies of Black self-concept, 1968- Education, 14, 60-69.
1978. Annual Review of Sociology, 5, 53-74. Willig, A. C. (1985). A meta-analysis of selected studies on the
Priest, G. E. (1985). Aboriginal languages in Canada. Language effectiveness of bilingual education. Review of Educational Re-
and Society, 15, 13-19. search, 55, 269-317.
Ramsey, P. G. (1987). Teaching and learning in a diverse world: Winer, B. J., Brown, D. R., & Michels, K. M. (1991). Statistical
Multiculturalism education for young children. New York: principles in experimental design (3rd ed.). New York:
Teachers College. McGraw-Hill.
Rosenberg, M., Schooler, C , & Schoenbach, C. (1989). Self-
esteem and adolescence problems: Modeling reciprocal effects. Received February 23, 1994
American Sociological Review, 101, 1001-1008. Revision received December 8, 1994
Sampson, E. E. (1988). The debate on individualism: Indigenous Accepted December 15, 1994 •