+ All Categories
Home > Documents > A CROSS-CULTURAL STUDY OF NATIONAL STEREOTYPES HELD …

A CROSS-CULTURAL STUDY OF NATIONAL STEREOTYPES HELD …

Date post: 18-Dec-2021
Category:
Upload: others
View: 4 times
Download: 0 times
Share this document with a friend
13
Japanese Psychological Research 1962, Vol, 4, No.2, 65-78 A CROSS-CULTURAL STUDY OF NATIONAL STEREOTYPES HELD BY AMERICAN AND JAPANESE COLLEGE GRADUATE SUBJECTS YASUMASA TANAKA Universityof Illinois An important problem in scientific re- search is the extent to which the method- ology and conclusions of one discipline can be applied to data of another. This paper is an attempt to take a measurement instrument developed by experimental psychologists and apply it to data which has been essentially the concerns of socio- logists and social psychologists who are in- terested in the cross-cultural investigation of national stereotypes. Since Walter Lippmann (1922) invented the word stereotypes and defined it as a men- tal picture, there have been a number of studies of national stereotypes. Katz and Braly (1933) established one method for studying such stereotypes. They asked Princeton undergraduates to characterize a dozen ethnic groups, draw- ing from a list of 84 trait names. Thus, Jews were reported to be "shrewd", mercenary "and" industrious", while Negroes were regareded as" supersti- tious","happy-go-luck," etc. Buchanan and Cantril (1953) and Reigrotski and Anderson (1959) repeated this method in their cross-cultural studies. Some diffi- culties of this method have been pointed out however. Brown (1958), for instant, has contended that the adjectives used in such studies are commonly "evaluative" in nature and the property of applying them cannot be deteimined in any ab- solute sense. It would be extremely helpful, therefore, if some notion could be obtained of com- mon ingredients which makes up the meaningful judgments people make of each other. Furthermore, a double value may be derived from such an approach. Psychological instruments have mostly been developed and validated in the con- text of Western civilization. To what extent can conclusions based upon these instruments can be universally extended to other human races? This can only be investigated by studying peoples with a non-Western cultural and linguistic heri- tage. Those interested in studying national stereotypes and international communica- tions will also benefit if a technique de- veloped elsewhere proves to have cross- cultural validity and can be applied to problems of the organization and process of stereotyping. In the present sutdy, only limited numbers of problems of national stereo- types were examined. First, judgment of an object or a person is considered to in- volve what I will call "evaluative" and " denotative " elements . Evaluative judg- ment may most often reflect certain built- in criteria of preference and morals in the specific cultural context. A denotative judgment of ,the physical size, form or pat- tern of activity, of an object or a person , on 1This research was supported by the Cross -Cul- tured Project sponsored by the Society for Investiga- tion of Human Ecology and directed by Charles E. Osgood. The present writer is indebted to Pro- fessor Charles E. Osgood, Director of Institute of Communications Research, Univ. of Illinois, and T. Obonai of Tokyo Univ. of Education, for there helpful comments on a first draft of this paper, to Professors Edward E. `'are, and Murray S. Miron, of the Institute of Communications Research, for their helpful suggestions on data analysis, and to the University of Illinois Statistical Service Unit and Digital Computor Laboratory for the use of their facilities.
Transcript

Japanese Psychological Research1962, Vol, 4, No.2, 65-78

A CROSS-CULTURAL STUDY OF NATIONAL STEREOTYPES HELD BY AMERICAN AND JAPANESE COLLEGE

GRADUATE SUBJECTS

YASUMASA TANAKA

University of Illinois

An important problem in scientific re-search is the extent to which the method-ology and conclusions of one discipline can be applied to data of another. This paper is an attempt to take a measurement instrument developed by experimental psychologists and apply it to data which has been essentially the concerns of socio-logists and social psychologists who are in-terested in the cross-cultural investigation of national stereotypes.

Since Walter Lippmann (1922) invented the word stereotypes and defined it as a men-tal picture, there have been a number of studies of national stereotypes.

Katz and Braly (1933) established one method for studying such stereotypes. They asked Princeton undergraduates to characterize a dozen ethnic groups, draw-ing from a list of 84 trait names. Thus,

Jews were reported to be "shrewd", mercenary "and" industrious", while

Negroes were regareded as" supersti-tious","happy-go-luck," etc. Buchanan and Cantril (1953) and Reigrotski and Anderson (1959) repeated this method in

their cross-cultural studies. Some diffi-culties of this method have been pointed out however. Brown (1958), for instant, has contended that the adjectives used in such studies are commonly "evaluative" in nature and the property of applying them cannot be deteimined in any ab-solute sense.

It would be extremely helpful, therefore, if some notion could be obtained of com-mon ingredients which makes up the meaningful judgments people make of each other. Furthermore, a double value may be derived from such an approach. Psychological instruments have mostly been developed and validated in the con-text of Western civilization. To what extent can conclusions based upon these instruments can be universally extended to other human races? This can only be investigated by studying peoples with a non-Western cultural and linguistic heri-tage. Those interested in studying national stereotypes and international communica-tions will also benefit if a technique de-veloped elsewhere proves to have cross-cultural validity and can be applied to

problems of the organization and process of stereotyping.

In the present sutdy, only limited numbers of problems of national stereo-types were examined. First, judgment of an object or a person is considered to in-volve what I will call "evaluative" and " denotative " elements . Evaluative judg-ment may most often reflect certain built-in criteria of preference and morals in the specific cultural context. A denotative judgment of ,the physical size, form or pat-tern of activity, of an object or a person , on

1 This research was supported by the Cross-Cul-

tured Project sponsored by the Society for Investiga-tion of Human Ecology and directed by Charles E. Osgood. The present writer is indebted to Pro-fessor Charles E. Osgood, Director of Institute of Communications Research, Univ. of Illinois, and T. Obonai of Tokyo Univ. of Education, for there helpful comments on a first draft of this paper, to Professors Edward E. `'are, and Murray S. Miron, of the Institute of Communications Research, for their helpful suggestions on data analysis, and to the University of Illinois Statistical Service Unit and Digital Computor Laboratory for the use of their facilities.

66 Y.TANAKA

the other hand, may reflect a certain cogni-tive rule of relativity. Thus, for a tall

person, A, a smaller person, B, is labelled small. A person, C, smaller than A and B, however, regards both A and B as large, while he himself is labelled as small by both A and B. That many such evaluative and denotative judgments are based upon

previous learning leads to the second pro-blem.

As many studies have shown, stereotypes are not usually simple abstractions from direct personal experience ; rather, they have been"learned" verbally or non-verbally through a period of time. The

process of stereotyping most often is a process of "assign learning" (Osgood et al 1957). The meaning of assigns is literally "assigned"to them through association

with other signs rather than through direct association with the object or the person signified. Most often it is learned via mass media and other second-hand information. Such learning then may be culture-orient-ed and bound to cultural criteria. As a matter of fact, it has been revealed in several sociological studies that those in-dividuals who are in a position to know best about an ethnic group often do not subscribe to the culture-oriented stereo-types (Kramer 1951; Deutsch and Col-lins 1951 ;IED 1952).

Finally, there are in national stereo-types some important elements of national consciousness. National consciousness can be defined as the "attachment of secondary symbols of nationality to primary items of information" (Deutsch 1953). The sym-bols of nationality like"Japanese" or "American" are labels added to objects

or actions. Their lexical meaning is ob-vious. But their" connotative" or "of-fective" meaning is implicit and may vary according to the semantic environment in which they are presented. In this connec-tion, examples of a tendency of "extension of Ego"have been given by Buchanan and Cantril (1953). The authors reported that each sample population tended to

rate itself "best" relative to other nation-alities in terms of selecting a greater number of "positive" adjectives for the description of its own nationality.

Now, it is hoped that the present study will prove the validity and possible utility of a particular measuring device for in-vestigating national stereotypes and inter-national communications-the semantic dif-

ferential. It is also hoped that results of its use will correspond to those problems examined previously. First, its cross-cultural validity must be demonstrated. Furthermore, its usefulness as an analytic tool must be clearly shown. The instru-ment will be regarded as having cross-cultural validity if it reflects (1) cross-cultural generality, or (2) cross-cultural differences, known to exist as a result of

previous studies. It will also be considered as useful if it can be applied with ease and help formulate or test the hypotheses in question.

PROCEDURE

The Psychological Instrument: The measuring instrument under investigation is the semantic

differential developed at the University of Illinois by C. E. Osgood and associates (Osgood

et al 1957). The semantic differential is a technique for measuring the meaning of con-

cepts. Linguistic concepts like WOMEN are rated on a series of descriptive "scales" of

judgment. Each seven-point "scale" is de- fined by a pair of opposite adjectives such as

beautiful-ugly, industrious-lazy, thick-thine, or fast- slow. Subjects indicate the direction and the

intensity of the association between the concept and the adjectives defining each scale.

For the present study, the semantic differen- tial was considered an appropriate measuring

device for two reasons. First, it is possible to investigate the meaning of the concepts of national stereotypes by the same procedure

simultaneously, whereas they have been treated fragmentarily by previous investigators who

utilized the 'trait name' procedure (cf. Katz and Braly 1933). It is also possible to examine

the meaning of concepts more accurately in

Cross-Cultural Study of National Stereotypes 67

terms of the indicated direction and the in-tensity of the association between the concepts and the adjective pairs. Secondly, it provides a common tool with which it is possible to measure the meaning of concepts and compare it cross-culturally.

Furthermore, Osgood and associates have reported three main factors of meaning, which they regard as the primary dimensions that seem to organize semantic space (Osgood et al 1957). These three most salient factors, label-led Evaluation (good-bad), Potency (thick-thin), and Activity (fast-slow), have been found in sample populations with non-Western linguistic-cultural background (Kumata and Schramm 1956 ; Kumata 1958 ;Suci 1959 Miron 1961 ; Tanaka and Oyama 1961 Osgood 1961). Very similar, if not identical, semantic structures have been revealed in these studies.

Subjects : This experiment was run on two

groups of subjects, each representing a differ-ent language/culture base. The Japanese

group consisted of 26 males and 10 females with the average age of 29, and the American group of 20 males and 12 females with the average age of 28. They were selected from among the faculty and graduate students in natural and social sciences at the University of Illinois. The average length of stay for the Japanese in America was 27 months. No American subject had spent a period of time abroad, but they had been acquainted with at least one Japanese on the campus.

Material: The semantic differential was

given both in Japanese and in English. A total of 29 verbal concepts were used in this study. They are listed in Table 1 (a) in order of presentation. They included four adjectival concepts representing nationalities (JAPA-NESE, AMERICAN, RUSSIAN and CHI-NESE), five nouns of familiar objects (WOM-EN, SCIENCE, PRODUCTS, ARTS, and PRESS), and 20 adjective 'noun combinations

(hereafter named compounds). For the com-pounds, a Latin square design was used. Translation equivalence of each concept was tested by three Japanese bilinguals by means of back-translation.

A total of 12 scales were used in the semantic

differential. Eleven scales were selected on the basis of previous cross-cultural studies

(Kumata 1958; Miron 1961). Those previous results enabled the present writer to obtain the translation equivalence for the eleven scales used in this study and examine their rotated factor ladings for both cultures. As a result, it was shown that the scales, beautiful-ugly,

pleasant-unpleasant, wisefoolish, would cross-cul-turally represent the Evaluation factor ; thick-thin, heavy-light, powerful powerless, the Potency ; and active-passive, fast-slow, noisy-quiet, the Activity. The scale, industrious-lacc, was tentatively regarded as an Activity scale. Thus, ten out of the eleven scales appeared to have rather high loadings on one of the three most salient factors across cultures. To them was added another scale, democratic-undemocratic, which had not been used in the previous, cross-cultural studies. The 12 scales, randomized in their directions and orders, are listed in Table 1 (b) in order of presentation.

RESULTS

Cross-Cultural Generality in the Semantic Frame of Reference.

Our first interest lay in ascertaining the extent to which the subjects in different language, culture groups can be shown to be using the same factors in making their judgments.

In the present study, only the scale inter-correlation matrix was factor-analyzed in each culture. The factor analysis within culture was done according to Thurstone's centroid factor analysis routine with fixed communalities (Thurstone 1947). The factors thus derived were then rotated by the Varimax rotation scheme (Kaiser 1958), and related across groups by the Wrigley-Neuhaus Coefficient of Factorial Similarity (Wrigley-Neuhaus 1955). All correlational and factorial work was done on the University of Illinois digital compu-tor, (ILLIAC).

If a stable semantic frame of reference is operating, one can expect the same or highly similar factors to appear in both cultures. The first four factors in the un-

68 Y.TANAKA

TABLE. 1. Concepts and Scales

rotated centroid analysis accounted for 95% of the total variance for Japanese

and 93% for Americans. Table 2 gives the Varimax rotations for the scales for

Japanese and Americans. The first four factors were identifiable

as Potency, Activity, Pleasantness, and Industriousness, although somewhat differ-ent scales were involved. The Potency factor is most salient for both groups as it accounts for the highest percent of variance. The Potency factor is followed by Activity, Pleasantness, and Industriousness factors in order of per cent of variance for each

group. Both Pleasantness and Industrious-

ness factors can be identified with divisions of the Evaluation factor. The indices of factorial similarity reveals a remarkable degree of cross-cultural similarity along each dimension in the semantic frame of reference. The highest loadings scales and the indices of factorial similarity are given in Table 3.

From factor analysis results it was dis-covered that some scales were used differ-ently by Japanese and American. Power-

ful-powerless, and active-passive were used by Japanese like thick-thin in the Potency factor, while they were used by American like fast-slow in the Activity factor. The

Cross-Cultural Study of National Stereotypes 69

scale, democratic-undemocractic, was used quite differently between the two groups. For

Japanese it was associated with the fast-slow judgment, whereas for American with

pleasant-unpleasant.2 Although no fur-ther conclusion is possible in this stud,y such different uses of scales might be indi-cative of some inter-cultural difference in the organization of the semantic space. Inter-cultural Differences in the Meaning of Concepts. In accordance with the factor analysis

results, four composite scores, each repre-senting one factor dimension, were com-

puted for each concept. The composite score is defined as the means of sever-al scale means along the same factor di-mension and it serves to simplify inter-

pretation of data (Miron 1961). The composite scores were computed in this study on the basis of stable scales whose factor loadings were highest on the same factor for both cultural groups. They were based upon the means of thick-thin, large-small, and heavy-light scales for the Potency factor, fast-slow and noisy-quiet for Activity, pleasant-unpleasant and beautiful-ugly for Pleasantness, and industrious-lazy and wise foolish for Industriouness. The composite scores for 29 concepts along four factor dimensions are given in Table 4.

To simplify interpretations and limit the number of variables, Hotelling's T2 test results were obtained on the basis of four composite scores for each concept. Hotel-ling's T2 is the multivariate generalization of Student's t test (Hodges 1955). The t test is used to decide the statistical signifi-

cance of differences between the means of two groups of subjects on a single vari-

able. T2 is a statistical test of differences

between two groups on a set of variables. The use and the validity of T2 in a cross-cultural, anthropological study was fully discussed by Maclay and Ware (1961).

Three separate sets of T2 were obtained in this study in order to examine how dif-ferent the two cultural groups are on the basis of four composite scores in connec-tion with (1) the meaning of each concept,

(2) groups of several nationality-bound concepts, and (3) groups of several object-bound concepts. It was also possible to obtain t test results of the differences be-tween Japanese and Americans for each concept/composite score combination. All the statistical work was done on the IBM calculators and the University of Illinois digital computor (ILLIAC). t test of Differences for Concept/Composite Score Combinations.

Results of t tests of differences, pres-ented in Table 5, give a detailed picture of the cultural pair comparisons. Here 47 of the 116 comparisons show a significant difference beyond the .05 level. It is noted that the most of differences occur-red along the Potency and Activity dimen-sions, whereas along the two Evaluation dimensions there were only a total of 8 differences.

Dominant differences along the Potency and Activity seem to justify distinctions between denotative and evaluative judg-ments. In an individual, the denotative

judgments may involve a certain rule of relativity between the judge and the

judged. Such a rule can be shown in the following examples. Along the Po-tency dimension, those Japanese subjects who rated JAPANESE slightly impotent (3.79), rated AMERICAN almost fairly potent (4.97), while those Americans who rated AMERICAN slightly poetent (4.83), rated JAPANESE slightly impotent (3.32). It is possible for the analyst to compare the two groups on a single concept. That is,

Japanese rated WOMEN more impotentive (3.18) than did Americans (3.76). Simi-larly, Japanese rated RUSSIAN WOMEN

2 To those dealing with human problems of politic-

al thought and traditions, this may possibly con-stitute a point of interest. For it is known that both cultures have carride with them a quite distinct set of political thought and traditions. It has been only a decade and half since the Japanese people were "brain-washed" toward Democracy.

70 Y. TANAKA

TABLE 2

Varimax rotation of first four factors

TABLE 3

(a) Highest loadings scales on varimax rotated factors and per cent variance

(b) Cross-cultural i dices of factorial similarity

(J/, Japanese; Al, American; /PO, Potency; ffA, Activitym/PL, Pleasantness; /I, Industriousness)

Cross-Cultural Study of National Stereotypes 71

TABLE 4

Four composite scores for 29 concepts, by culture

•õ The four parenthesized scales represent the direction of original scales.

•õ•õ Composite scores were computed so that low values are toward left-hand scale term (e.g.,

lazy) and high toward right-hand term (e.g., industrious).

more potent (5.67) than did Americans

(5.11), while the latter rated JAPANESE WOMEN more impotent (2.76) than did the former (2.94) even if this difference did not prove statistically significant.

Similar trends were confirmed along the Activity dimension. Here, Japanese who

felt JAPANES WOMEN slightly inactive

(3.88), rated AMERICAN WOMEN fairly active (5.72), while Americans who felt AMERICAN WOMEN slightly active(4.91), rated JAPANESE WOMEN quite inactive (2.90). Again, the Japanese group rated AMERICAN more active (5.69) than did

72 •@ Y. TANAKA

TABLE 5

Results of t tests of differences between cultures for each concept/composite score combination, and F values derived from T2 tests of differences between

cultures for each concept

* Only those values significant beyond the .05 level are given. t values greater than 2.66 are

significant beyond the .01 level. ** Only those values significant beyond the .05 level are given. F values greater than 3.65 are

significant beyond the .01 level.

the American group (5.21), while the latter rated JAPANESE more inactive (3.49) than did the former (4.03).

It should be remembered that each of these judgments was made independently. Results given here were only based upon

comparisons of the position each concept occupies along the semantic dimensions. On this semantic map, relativity in deno-tative judgments was shown clearly . It appeared not to conflict with a culture-bound standard according to which dif-

Cross-Cultural Study of National Stereotypes 73

ferent peoples conceive and compare the size, form, or pattern of activity, of things and persons.

Along the Evaluation dimensions, many fewer differences were observed. Since results of t tests did not reveal statistically singificant differences between Japanese and Americans on most of con-cept/composite score combinations, we can only examine face-value differences rather than making a rigid statistical analysis. Due to limited space, only a few examples can be shown here.

First, along the Industriousness diman-sion, Japanese ranked JAPANESE them-selves highest (5.20), RUSSIAN second highest (4.88), AMERICAN third (4.75), and CHINESE lowest (4.50). Exaclty the same ordering was obtained from American data: that is, JAPANESE was reted highest (5.61), RUSSIAN second

(5.16), AMERICAN themselves third (4.95), and CHINESE lowest (4.89). Equally, along the Pleasantness dimension, the matched orderings were cross- cultural-ly noted. That is, Japanese rated JAPA-NESE themselves highest (5.11), AMERI-CAN second (5.06), CHINESE third (4.42), and RUSSIAN lowset (3.93). Americans ranked JAPANESE highest

(4.93), AMERICAN themselves second (4.74), CHINESE third (4.73), and RUS-SIAN lowest (3.68) . Note that both groups rated RUSSIAN as rather unpleasant.

Secondly, a sample of compounds was examined. WOMEN and SCIENCE were selected as the object componenets of the compounds. In this analysis, Japanese ranked RUSSIAN WOMEN highest (4.83) along Industriousness, JAPA-NESE WAMEN second (4.60), AMERI-CAN WOMEN third (4.33), and CHINE-SE WOMEN lowest (4.31), where Ameri-cans rated JAPANESE WOMEN highest

(5.11), RUSSIAN WOMEN second (4.97), CHINESE WOMEN third (4.73), and AMERICAN WOMEN lowest (4.53). Along Pleasantness, JAPANESE WAMEN was ranked highest (5.29) by Japanese,

AMERICAN WOMEN second (5.23), CHINESE WOMEN third (4.82), and RUSSIAN WOMEN lowest (4.70). Equally, Americans rated JAPANESE WOMEN highest (5.48), AMERICAN WOMEN second (5.26), CHINESE women third (4.91), and RUSSIAN WOMEN lowest (3.98). Thus, to both

Japanese and Americans, JAPANESE WOMEN and AMEIRCAN WOMEN were fairly pleasant, whereas, to them, both RUSSIAN WOMEN and CHI-NESE WOMEN were far less pleasant. What about SCINECE, then? Along Industriousness, AMERICAN SCIENCE was rated by Japanese as the first (5.65), followed by RUSSIAN SCIENCE (5.54),

JAPANESE SCIENCE (5.15), and CHINESE SCIENCE (4.76). By Ameri-cans, AMERICAN SCIENCE was also rated as the first (5.70), followed by RUS-SIAN SCIENCE (5.59), JAPANESE SCIENCE (4.75), and CHINESE SCI-ENCE (4.36). Now, along Pleasantness,

Japanese ranked AMERICAN SCIENCE highest (5.12), then, JAPANESE SCIENCE (4.65), RUSSIAN SCIENCE (4.47), and CHINESE SCIENCE (4.20), while Ameri-cans also rated AMERICAN SCIENCE highest (5.15), then, JAPANESE SCI-ENCE (4.71), RUSSIAN SCIENCE (4.46), and CHINESE SCIENCE (4.23).

Throughout the Evaluation dismensions, the two cultural groups gave a very similar pattern of ratings on most of 29 concepts. Since the judgments reflec-ted upon each concept/composite score conbination were independent, and were later compared cross-culturally along the same dimension, it certainly would be safe to conclude that there was a set of very similar criteria of the evaluative judgment across both Japanese and Americans.

Now, if the so-called "extension of Ego" implies a tendency of one nationality to rate itself best relative to other nationalities, then, it will also be observed in the analysis of the Evaluation dimensions of this

74 Y. TANAKA

study-Industriousness and Pleasantness . In order to see this tendency in our data, concepts, excluding five object concepts, were classified into one group of four nationality concepts and five groups of four object-bound concepts. Then, within each of these six groups, the composite score was compared with one another-

JAPANESE with AMERICAN, RUS-SIAN and CHINESE, or JAPANESE SCIENCE with AMERICAN SCIENCE, RUSSIAN SCIENCE and CHINCESE SCIENCE, etc. This comparison was made in each cultural group along both Industriousness and Pleasantness. And so, if Japanese rating nationalities rated JAPANESE best of all nationalities along Industriousness, or, if Americans rating a group of compounds rated AMERICAN SCIENCE best science of all nationalities along Pleasantness, then, it is evident that the " extension of Ego " certainly tended to be present in these cases. Summarizing the figures given in Table 4 in this way, we see the tendency in seven cases out of a total of twelve for Japanese and three for Americans. The seven cases for Japa-nese inclused JAPANESE, JAPANESE PRODUCTS, JAPANESE ARTS, JAPA-NESE PRESS along Industriousness, and JAPANESE, JPANESE WOMEN, JAPANESE ARTS along Pleasantness; and the three American cases AMERI-CAN SCIENCE, AMERICAN PRESS along Industriousness, and AMERICAN SCIENCE along Pleasantness. It is in this respect remarkable to see that both

Japanese and Americans agreed upon the ratings of JAPANESE, JAPANESE WOMEN, JAPANESE ARTS,and AMERICAN SCIENCE, as best along both Industriousness and Pleasantness. Thus, in most cases under investigation, the " extension of Ego " , even when it really existed, was not unilateral as it should be, for both Japanse and Ameridcans did agree upon the relative rankings of con-cepts. On the other hand, the fact that each cultural groups clearly discriminated

the ratings of object-bound concepts by nationalities, suggests the obvious presence of national consciousness. It is assumed that such discrimination is functioning in the process of conceptualization. Natio-nal consciousness as such, however, was not necessarliy an expression of patriotic tendencies. But it was rather a reflection of realistic discriminations among object-bound concepts. T2 tests of Differences for the Meaning of Concepts. Results of the first T2 test analysis are given in Table 5. The same table shows where the two groups differ in judging the meaning of each conccept along four separate semantic dimensions. The T2 tests summarize these t test results and give evidence as to whether the two "cultures " may differ in judging each con-cept on the basis of four separtae variables. In this analysis, significant differences were revealed between Japanese and Americans on 24 out of 29 concepts. The F values derived from T2 tests ranged from 2.56 (on JAPANESE WOMEN, CHI-NESE WOMEN, and AMERICAN SCIENCE) to 25.70 (on RUSSIAN ARTS). In summary, it was clear that, despite very similar eavluative judgments, that two culture groups assigned a signifi-cantly different meaning on the whole to most of the 29 concepts. The semantic differential proved to be useful in dis-solving the whole meaning of a concpet into several semantic dimensions so that one could examine where and to what extent meaning is rather similar or different between the different cultural groups.

Our second interest was in investigating wheter the two " cultures " differ when they judge not a single cocept, but a group of concepts. There were five groups of nationality-bound concepts and anothr five of object-bound concepts. Each

group of nationality-bound concepts con-tained one nationality concept (e.g., JAPANESE) and four compounds with the same nationality identification (i.e.,

Cross-Cultural Study of National Stereotypes 75

TABLE 6

Values derived from Ts trsts of differences between cultures for concepts groups

(a) Groups of nationality-bound concepts•õ (b) Groups of object-bound conceptstt

•õ Each group of nationality-bound concepts consists of one nationality concept and four com-

pounds with the same nationality identification. F values greater than 2.30 is significant beyond the

the .01 level.

•õ•õ Each group of object-bound concepts consists of one object concept and four compounds with

the same object identification. F values greater than 2.30 is significant beyond the .01 level.

JAPANESE WOMEN, JAPANESE SCIENCE, JAPANESE ARTS, and

JAPANESE PRESS).3 Each group of object-bound concepts, on the other hand, consisted of one object concept

(WOMEN) and four compounds with the same object identification (JAPANESE WOMEN, AMERICAN WOMEN, RUSSIAN WOMEN, and CHINESE WOMEN). 7-2 was obtained for each concept group on the basis of a total of twenty separate semantic variables (i.e., four composite scores for each of five con-cepts). Results are given in Table 6.

It was found that the two "cultures" were significantly different beyond the .01 level in all concept groups. A F value derived from T2 is assumed to indicate an extent to which Japanese and Americans were different when they judge the same group of concepts. It was revealed that such difference was smaller when the two groups

judged several concepts related to JAPA-NESE and AMERICAN, than when they judged those related to CHINESE and RUSSIAN. It was also smaller when they judged those concepts related to PRESS, PRODUCTS and SCIENCE,

rather than those related to WOMEN an ARTS.

DISCUSSION

We have considered several criteria that an instrument must meet before it is acceptable as a tool of analysis of national stereotypes. With respect to its ability to reflect cross-cultural and semantic gen-erality and divisions, the utility of the semantic differential is promising.

The results of factor analysis of scale relationships have demonstrated a remar-kable cross-cultural generality in the semantic frame of reference. Next, it was also shown that the meaning of a concept can be given by the isolated position it occupies in the multiple dimensional semantic space. The semantic difierenti-al has proved to be useful for dissolving the whole meaning of a concept into several semantic dimensions. This is useful not only because the number of variables (scales) can be reduced and interpretations of date are simplified, but also the analyst is enabled to make separate observations along each semantic dimen-sion. It has also been proved that the semantic differential can be used so that one can make both quialtative and quanti-tative analysis of meaning. In this study, it was discovered that meaning more often

3 Compounds with the component, PRODUCTS,

were excluded from this analysis because of com-

putational limitation on the ILLIAC.

76 Y. TANAKA

is different across the two cultural groups along the Potency and Activity dimensions than along the Evaluation.

Results thus obtained give us some notion of the validity of the semantic differential appliee in the study of national stereotypes. First, they confirmed the

presence of a rule of perceptive relativity with respect to the size, form, or pattern of activity, of things and persons. Secon-dly, there was also evidence that the sample groups shared some evaluative criteria in common. Thirdly, the so-called "extension of Ego " tendency was rather unconfirmed in this study. Fourthly, with respect to the over-all semantic difference between the two cultural groups, the findings of this study did not conflict with the findings of Kramer (1951), and Deutsch and Collins

(1951), who also found that individuals who knew an ethnic group well could

gain some immunity from culture-bound stereotypes about it. As was expected, the semantic difference between Americans and Japanese, acquainted with one another rather well, was amaller when they judged several concepts related to Japanese and American than when they judged several concepts related to Russian and Chinese. Similarly, the semantic difference be-tween he two groups for those concepts related to familiar modes of modern living, that is, science, products, and press, was smaller than those related to familiar objects of aesthetics, women and arts. In general, the sematic difference may in-dicate the extent of cultural lag between the different peoples. Such cultural lag was observed cross-culturally for RUS-SIAN and CHINESE and WOMEN and ARTS in the present study. Although further conclusion must await future in-vestigation, it can be assumed that dif-ference in meaning, or a cultural lag, reflect some differences in assign learning between the cultures involved.

On the whole, these analyses have dealt with psychological as well as linguistic problems of meaning. It is concluded

that results obtained have produced evi-dence against the Whorfian hypothesis of linguistic relativity-that our language determines our cognition (Whorf 1956). We consider the cross-cultuarl generality in the semantic frame of reference as such evidence. However, it has also been revealed clearly where and to what extent the meaning of concepts was different between Japanese and Americans who used the similar semantic frame of re-ference.

Due to the large number of statistical tests based upon a single sample, inter-

pretation of results presented here should be considered as hypothesis-formulating rather than hypothesis-testing. It must be remembered, for instance, that the test were administered in the American culture to those elite subjects with high intelligence and scholarstic standing. They were by no means representative of the average Japanese or American. With all this limitation, however, it is hoped that the semantic differential has proved to have its validity as well as utility as a general scientific instrument which might give additional dimension to cross-cultural studies of national stereotypes, and that the results of this study will serve for formulating a better hypothesis in the area.

REFERENCES BROWN, R. Words and Things. Glencoe : Free Press.

1958. Buchanan, W., & H. Cantril, How nations see each

other. Urbana : Univ. of Illinois Press. 1953. Deutsch, K. Nationalism and social communication.

Cambridge : Technology Press of M. I. T., 1953.

Deutsch, M., & Collins M. E., Interracial Housind: A psychological evaluation of social experiment. Min-neapolis : Univ. of Minnesota Press. 1951.

HODGES, J. L. Discriminatory analysis I. Survey of Discriminatory Analysis. Report No. 1, Contract No. AF-41 (128)-8, Univ. of Cali-fornia 1955.

Information and Education Division, United States War Department. Opinion about negro in-

Cross-Cultural Study of National Stereotypes 77

fantry platoons in white companies of seven divisions, In G. E. Swanson, T. M. Newcomb, & E. L. Hartley (Eds.), Readings in social psy-chology. New York : Holt. 1952, 502-506.

Kaiser, H. F. The varimax criterion for analytic rotation in factor analysis. Psychometrika. 1958. 23, 187-200.

KATZ, D., & BRALY K. W. Racial stereotypes of 100 college students. J. Abnorm. soc. Psychol., 1933, 48, 327-335.

KRAMER, B. M. Residential contract as a deter-minant of attitudes toward negroes. Un-

published Ph. D. dissertation, Harvard Univ. 1951.

KUMATA, H. A factor analytic investigation of the

generality of semantic structures across selected cultures. Unpublished Ph. D. dissertation, Urbana, Univ. of Illinois. 1958.

KUMATA, H., & SCHRAMM W. A pilot study of cross-cultural meaning. POQ, 1956, 229-238.

LIPPMANN, W. Public opinion. New York: Mac-millan. 1922.

MACILAY, H., & WARE E. E. Cross cultural use of the semantic differential. Behau. Sri., 1961, 185-90.

MIRON, M. S., A cross-linguistic investigation of phonetic symbolism. 3. Abno in. soc. Psychol.,

1961, 62, 623-630. C. E. OSGOOD, TANNENBAUM P. H. & Sucs G. J. The

measurement of meaning. Urbana : Univ. of Illinois Press. 1957.

OSGOOD, C. E., Studies on the generality of afective meaning system. Urbana : Univ. of Illinois, 1961.

REIGRGTSKI, F., & ANDERSON N. National stereo-types and foreign contacts. POQ, 1959-60,

515-528. Suct, G. J. An investigation of similarity between

the semantic spaces of five different cultures. J. Abnorm. soc. Psychol., 1960, 61, 25-30.

TANAKA, Y. & OYAMA T. A cross-culture and cross-onceptc study of the generality of Semantic Spaces, Unpublished resarch at the University of Illinois, 1961.

THURSTONE, L. L. Multiple Factor Analysis. Chicago : Univ. of Chicago Press, 1947.

WHORE, B. L. Language, thought, and reality ; Selected writings of Benjamin Lee Wharf. New York : Wiley, 1956.

WRIGLEY, G., & NEUHAUS J. O., The matching of two sets of factors. Contract Memorandum, Report A-32, Urbana, 1955.

(Received April 4, 1962)


Recommended