Some Potential Uses of the Cross-cultural Atlas of
Affective Meanings.*

Dr. Leon James (formerly Leon A. Jakobovits)
Center for Comparative Psycholinguistics of the University of Illinois

(c) 1967

1) Substantive characterization of the Coors-culturally common and the
cross-culturally unique.
2) Two examples of componential analysis.
3) An example of a hypothesis-testing application.
4) Summary


The Atlas of Affective Meanings is both a kind of a dictionary and a repository of information about cultural groups. For the comparative psycholinguist it may turn out to be what Murdock's Human Relations Area Files was for the anthropologist and social psychologist. Thus far in our work on what we refer to as "subjective culture" it has proved to be a useful tool for testing out certain culturological hypotheses. In this paper,, I would like to briefly outline some of the potential uses of the Atlas as a hypothesis-testing device and the nature of the information it contains about cultural groups.


A striking fact that emerges in the comparative study of cultures is the contrast between the cross-culturally variable and the cross-culturally constant. Yet this is no less and no more than what one should expect. For despite their varied nature (and frequent uniqueness), human societies everywhere share certain characteristics in common, a "world culture" that is imposed by the common sensory mold of the human species and the restrictions demanded by the requirements of cooperative interaction of social individuals.

What would a lexical inventory of world culture  look like? Table I shows a few such concepts selected from the Atlas. Ten cultures are represented in this

go to Table 1

comparison: AE for American English, DH for Delhi Hindi, FF for. Finnish, GK for *Paper prepared for delivery at the XI Interamerican Congress of Psychology, Greek, HC for Hong Kong Cantonese., JP for Japanese, LA for Lebanese Arabic, ND for Netherlands Dutch, SW for Swedish, and YS for Yugoslav Serbo-Croat. The columns represent the nature of the shared affect: E P A refers to a portion of the semantic space (or "octant") encompassed by the three affective dimensions of Evaluation, Potency,, and Activity and can conveniently be labeled as the "good-strong-active octant." Here we can find the concepts AIRPLANE, CHAMPION, and SPAM RABBIT is good-weak-active, SAVING is good-strong but neither active nor passive, while BEARD is potent but not particularly good or active. Notice a cluster of food objects in the good octant which is neutral with respect to Potency and Activity. Table 2 presents a more extended list involving 15 cultures - the previous 10

go to Table 2

and five additional ones: FR for French,, IT for Italian, MK for Mysore Kannada (Southern India), MS for Mexican Spanish, and TH for Thailand. The columns once again indicate the nature of the shared affect., and this time the differentiations are made along a single dimension only. There are in this list certain obvious and expected communalities which do not surprise: indeed, they demonstrate a face validity of the Atlas data whose absence would have been cause for worry. There is cross-cultural unanimity in the designation of COOPERATION, FRIENDSHIP, KINDNESS, PLEASURE, SUCCESS, TRUTH as good things. There is also perfect agreement as to the badness of ACCIDENT, DISEASE, POISON, THIEF, and the potency of ELEPHANT, MASCULINITY, POWER, and ROCK. Similarly, we find near perfect agreement on the weakness of BABY and RABBIT, the high activity of AIRPLANE and DANCING, and the passivity of OLD PEOPLE and PYRAMID.

On the other hand, certain cross-cultural communalities exhibited by this list would not have been easily predictable. Why should APRIL, MARCH, MAY, and AUGUST be good everywhere despite very different climatic environments? Although SUICIDE is intrinsically destructive of human life one would have expected that given its religious significance in some cultures (e.g. JP), some departure of its badness could have occurred. Not so. It is both instructive and, to this author at least, a source of profound satisfaction, to note the unanimous acceptance of the goodness of KNOWLEDGE, PEACE, and UNIVERSITY, the badness of CHEATING, GREED, and ANGER, the potency of JUSTICE, PHILOSOPHY, and SCHOOL. At the same time, it may be sobering to reflect upon the unanimous acceptance of the potency of ARMY, NATIONALISM, and REVOLUTION.

The cross-cultural similarities as well as the differences are reflected in the summary presentation of all the Atlas concepts in Table 3, which presents the

go to Table 3

distribution of concept allocations in the 27 octants of our semantic space for 10 cultures. The skewness of the distribution is a striking fact. The semantic world is not filled out equally in all regions. Good things outnumber bad things by five to one; strong things outnumber weak ones by 13 to 1; active things have a two to one edge over passive things. There is here a profound psycholinguistic truth whose lesson is at the moment not entirely clear. Why should the concept universe give such a decided edge for the good, the strong and the active over the bad, the weak, and the passive? The perception of these qualities is surely a subjectively imposed organization and one cannot find an easy answer in the objective reality that surrounds us.

If we now switch our attention to cross-cultural differences to be found in Table 3, they turn out to be no less intriguing than the similarities. Take for example the question of what is the most salient octant for the various cultures. We find that for AE and YS it is good and strong but neutral on activity; for GK and HC it is good, strong and active; for DH and LA it is good, strong, and passive; for FF it is good and active but neutral on potency . Note that for JP and ND there is no strongly salient octant, they apparently preferring a more differentiated concept universe. Note also the almost complete absence of concepts in the good, strong active octant for DH and IA. How is one to interpret these differences? There cannot be any doubt from these results that cultural membership is indeed an association with weighty consequences: it imposes upon the individual a world-view that is no less distinctively characteristic as it is far reaching and profound.


Up to now, the discussion has dealt with certain aspects of the Atlas data in terms of simple raw scores on the three main affective dimensions. It is possible to make use of the Atlas data in the form of comparisons at a more abstract level involving a componential analysis based on intuitive classifications of selected concepts within the Atlas. One such example is presented in Table 4. Thirteen

go to Table 4

concepts dealing with family kinship system ("kincepts") were isolated for this analysis. The seven components shown in the table were intuited by Professor Charles Osgood of the University of Illinois. Each kincept receives one of three values on-each of the components: a plus sign indicates the presence of a particular feature reflected by the component, e.g. BROTHER receives a "+" for Sex indicating the male feature, while SISTER receives a "-" indicating the absence of that feature (i.e. SISTER implies female). Similarly, AUNT is "-" on Sex, "+" on Generation (indicating above I, MYSELF which receives a "-"), "-" on Lineality (vs. "+" for FATHER), "-" on Nuclearity, Specificity (i.e. the identity of AUNT is potentially ambiguous), and Maritality (vs. a "+" for MOTHER or GRANDFATHER, which imply a state of being married). A zero sign indicates that the particular feature involved in a component is not relevant to the concept, e.g. BRIDE, Age ? Nuclearity? or I, MYSELF, Lineality? The reader can explore for himself the other relations implied in Table 4. The next step in this analysis is to isolate kincept pairs that contrast on one particular feature but are similar on all the other features - or at least, do not contrast on them. For example, the pair BROTHER/ RELATIVES (or BR/REL) contrast on the Nuclearity component (BR implies member of the nuclear family while REL does not) while they are either similar on the other components or do not contrast on them. Other relevant pairs contrasting on the Nuclearity component are BROTHER/UNCLE (or BR/UN), SISTER/AUNT SIS/AU), FATHER/ GRANDFATHER (FA/GF), and MOTHER/GRANDMOTHER (MO/GM). It should be noted that some of the pairs represent -sent imperfect contrasts in that they differ on one other component in addition to Nuclearity (e.g. BR/UN on Generation). Imperfect contrasts result in a potential confounding of the component that is responsible for affective differences between pairs of kincepts. The present analysis must therefore be considered as tentative only and is given here for illustrative purposes. Selection of concepts in greater depth for a particular category would insure the availability of perfect contrasts.

Table 5 presents a test of the affective correlates of the Nuclearity component involving comparisons across 15 cultures. For each of the five

go to Table 5

contrastive kincept pairs a test is made in terms of some affective score in the Atlas. For example, in terms of Activity the nuclear member of the pair (BR, SIS, FA, MO) is rated as more active in AE than the corresponding non-nuclear member of the pair (UN, AU, GF, CH, REL). This is indicated in the table by "+" signs. On the other hand, for LA the nuclear members of each pair are less active and this is indicated by a sign. At the extreme right of the table, the and "-" signs are summated in terms of a ratio score. As can be seen, the Nuclearity component is relevant to Evaluation, Activity, and Individual Polarity (or meaningfulness), as indicated by significantly disproportionate ratios of "+" and "-" signs. In other words, cross-culturally as a whole, the nuclear members of the family are perceived as more good, more active, and more meaningful (Individual Polarity) than the non-nuclear members. To this general characterization, we must note culture specific exceptions such as LA where the non-nuclear members are perceived as more active and AE where no consistent preference in Evaluation is exhibited, as well as other such lack of agreements in the table.

Table 6 presents a test of the Sex component involving the same 15 cultures.

go to Table 6

Here, five kincept pairs make up the relevant contrast taken from Table 4. The summary ratios on the extreme right indicate Potency as the relevant affective correlate of Sex. There is very little cross-cultural disagreement in assigning higher potency to the male pair of the family member.

Table 7 presents the results of a test of affective distance for the two components of Nuclearity and Sex for six cultures. The numbers in the table represent "D scores" which are geometric distance measures on three simultaneous dimensions (E, P, and A). The size of the number (which varies from zero on up) indicates how far in affective space the concepts are from each other. it can be seen that in every culture presented the affective distance between male-female members of the family is larger than the affective distance of the nuclear-non nuclear members. This indicates that sex differences are more salient than nuclearity differences in distinguishing affectively the members of the family.

go to Table 7

Note also that this contrast is much more pronounced in some cultures than others. Thus, MK does not differentiate Nuclearity as much as HC. On the other hand, HC differentiates Sex more than MK. These patterns ought not to be considered definitive until significance tests on these differences are carried out.

A second example of a componential analysis done on a different set of Atlas concepts is given in Tables 8 and 9. Six concepts were selected as members

go to Table 8

of the concept class of color. In this case physical and psycho-physiological considerations suggest three convenient components: Brightness, Saturation, and Hue. The terms WHITE ("+") and GREY ("-") represent an adequate contrast for Brightness. For Saturation RED and BLUE are contrasted with YELLOW and GREEN respectively, on the rationale that at equal physical purity (i.e. in terms of wave length) RED and BLUE appear psychologically more saturated than YELLOW and GREEN. The contrast for Hue is in terms of the longer wave length distribution (RED and YELLOW) contrasted with the shorter wave length distribution (BLUE and GREEN). Table 9 presents the test for affective relevance of the three color components

go to Table 9

for 15 cultures. The summary ratios at the extreme right indicate the following relationships. Brightness reflects a clear relevance for Evaluation and Activity such that brightness goes with high positive evaluation and high activity. Saturation is clearly relevant for Potency, viz. higher saturation goes with higher potency and less clearly with Activity, viz. higher saturation goes with higher activity. Finally, for Hue the B-G side of the spectrum is clearly more positively evaluated than the R-Y side. -It is interesting, and reassuring, to be able to report that recent data obtained with physical color patches varying It, known physical values of brightness, saturation, and hue corroborate this componential analysis based on color words. Color patches having higher brightness values while saturation and hue were kept constant, were rated significantly better in the four cultures tested. When brightness and hue were kept constant, more saturated color patches were rated significantly more potent and finally, when brightness and saturation were kept constant, BLUE was rated significantly better than RED or YELLOW. In addition to this corroboration the color patch h data were more powerful and showed additional contrasts not reflected by the color word ratings a finding which is hardly surprising.


This final section illustrates a different use of the Atlas, that of testing a specific hypothesis which in the present instance is based on Professor David McClelland's theory of need achievement in the economic sphere as expounded in his book on The Achieving Society. The aspect of Professor McClelland's theory that is under examination here is the notion that the economic achievement of a country is partially determined by the presence within a culture of a sufficient number of individuals with a high achievement motivation whose entrepreneuring spirit can be recruited to managerial and business positions where they can actively promote their country's economic development. The high achieving person is depicted as possessing certain characteristics which make him effective in promoting economic development. These characteristics include a perception of the self as active and potent and operating on an environment that is more passive and relatively weaker which gives him possibilities of manipulation and promotion of change. The high achieving person also has a dim view of tradition and the past, looks out towards a bright future, values knowledge and welcomes competition. This characterization of the high achieving person was generalized in the present Instance to the cultural level and it was argued that certain predictable differences in the affective meaning of concepts ought to occur between economically advantaged and disadvantaged cultures in our Atlas sample since the prevalent achievement syndrome within a culture ought to be reflected in that culture's predominant evaluations of concepts relevant to achievement. With this view in mind 22 concepts were selected from the Atlas which were thought on an intuitive basis to be relevant to the achievement motive. Predictions were then made with respect to the expected differences in Evaluation, Potency and Activity for the ratings of these concepts between the culturally advantaged and disadvantaged cultures in our Atlas sample. The concepts and predictions are shown in Table 10. The predicted difference in

go to table 10

each case is indicated by the numerals "1" and' "2" reflecting the rank order of means to be expected. Thus, I, MYSELF was predicted to be higher on Potency and Evaluation ("1") for the economically advantaged cultures; COMPETITION was predicted to be better, stronger, and more active; TRADITION was predicted to be less good, less potent, and less active and so on. In all 50 predictions were made. Table 11 presents the mean ratings of the 22 concepts on each of the three factors for six

go to table 11

economically advantaged cultures (AE, FR, IT,, JR, ND., SW) and six economically disadvantaged cultures (DH, HC, LA,, MK, MS, TH). In this table, there are 66 possible comparisons (22 concepts x 3 factors), 50 of which were included in the predictions. The results of the comparisons in relation to the predictions are indicated in Table 10. 'The predicted comparisons that were confirmed by the data are underlined. A solid line indicates confirmation by a difference that can be considered significant (i.e. a difference of half-a-scale unit or larger). A broken line indicates a difference in the predicted direction but which does not reach statistical significance. A test of the adequacy of the predictions is given by the following observations: (1) 32 of the 50 predictions were in the proper direction., and 11 of these were significant; (ii) none of the predictions made were ever significant in the opposite direction; (iii) of the IS predictions made with respect to the Activity factor, 16 were in the proper direction., and 9 of these were significant; (iv) of the 66 possible comparisons in Table 11, 10 were statistically significant; of theses, 9 were correctly predicted. They can be summarized as follows: higher activity on the part of the advantaged cultures for Is MYSELF., BODY, COMPETITION, and FUTURE; more passivity for NEUTRALITY., TRADITION, PAST., DEATH, and SUNDAY; more potency for COMPETITION; the one significant difference which was not predicted was higher evaluation for SUNDAY on the part of the advantaged cultures. It is clear then that the hypothesis can be considered adequately confirmed only in the case of the Activity factor.


The Atlas of Affective Meanings consists of affective ratings of 500 concepts given-by groups of individuals in some 15 language/culture communities around the world. This paper briefly discusses the nature of the data in the Atlas and presents a few illustrations of how that information can be used in culturological investigations in comparative psycholinguistics. The first section gives one possible characterization of the cross-culturally constant that may be viewed as substantive elements of a "world culture" and also examines some ways in which cultures impose a characteristically unique world-view upon the individual. The second section presents two examples of componential analysis involving the family kinship system and color. The third section illustrates a hypothesis-testing procedure of the Atlas data as applied to McClelland's theory of achievement motivation and economic development.


*Paper prepared for delivery at the XI Interamerican Congress of Psychology, Mexico, December 17-22, 1967. Published in proceedings and reprinted in: W.W. Lambert & R. Weisbrod (EDS.), Comparative Perspectives on Social Psychology. Boston: Little Brown, 1971. Pp. 164-174.

Applied Psycholinguistics in Social Psychology
Effects of Mere Exposure: A Comment
Rhetoric and Stylistics: Some Basic Issues in the Analysis of Discourse
Semantic Satiation and Cognitive Dynamics
Semantic Satiation as a Function of Initial Polarity and Scale Relevance
Some Potential Uses of the Cross-cultural Atlas of Affective Meanings

An Evaluation of People's Attitudes Toward Technostress and Techniques on How to Overcome it.
To Fear or not to Fear the Computer

More Leon James Articles here