Introduction

For close to a third of a century, students of Japan have commonly accepted the claim of Doi (1973; 1996) that Amae is distinctive to the production and reproduction of Japanese culture, and is what makes Japanese child rearing peculiarly different from that of Americans. The importance of the concept can be inferred from the fact that fifteen of seventeen chapters by imminent scholars in a recent compendium on Japanese childrearing (Shwalb & Shwalb 1996) cite the English translation of Doi's 1996 Japanese book, Amae no Koso.

Doi (1996: xv) defines amae as "indulgent dependency", rooted in the mother-child bond. The indulged (amaeta) child as a spoiled child is ethnocentrically Western from the Japanese-as unique view. Vogel (1996: 197) goes so far as to argue that ". . . I see amae (indulgence) as the universal basic instinct, more universal than Freud's two instincts, sex and aggression." According to Vogel (1966: 186), amae is experienced by the child as a "feeling of dependency or a desire to be loved", while the mother vicariously experiences satisfaction and fulfillment through overindulgence and overprotectiveness of her child's immaturity. The assumption is that subsequent Japanese social bonding—teacher-student, supervisor-subordinate, etc.— is patterned after the primary mother-child experience. This can be inferred from Vogel's observation that a large number of Japanese mothers blame the themselves for not being loving or giving enough when their children refuse to attend school. Essentially, Japanese mothers report feeling guilty if they are not all-giving to their children.

Doi asserts that European languages lack an equivalent word to amae. His argument is that the lack of an equivalent word implies lack of social recognition and need of feelings of dependency and the desire to be loved in the West. The closest Western equivalents might be the classical Greek concepts of eros, which assumes the child's immature need to be loved, versus agape, deriving from the mother's need to give unqualified love (Tillich, 1957).

In contradistinction, Hess and Azuma (1991) suggest that the American preoccupation with independence prevents us from noticing the extent to which the need for "indulgent dependence" expressed by amae positively influences educational aspirations through American parent-child and teacher-pupil relationships. Doi (1996) would agree; he asserts that the psychic feeling of feeling emotionally close to another human being is not uniquely Japanese—only the rich, semantic meaning of amae differentiates Japanese culture in his view.

This paper uses affect control theory (MacKinnon 1994) to understand the rich, semantic meaning of amae within the Japanese context, and to search for American English equivalents. First, we use special cross-cultural lexicons with exceptionally high reliability and validity to search for semantic cousins of concepts related to amae, independence, and dependence. The American and Japanese lexicons are publicly accessible through the affect control theory site with a Java web browser like Netscape 3.0 or Internet Explorer 3.0, or higher (Heise 1997). Then, we construct events that simulate the prototypical mother-child relationship in both cultures. We use an equation-based computer program with explained variance of .85 to .90 called JavaInteract (Heise 1997) to evaluate semantic differences and similarities in prototypical mother-child relationships.

Identity-Confirmation and Affect Control

Affect control theory (hereafter, ACT) postulates that humans try to engage in identity confirming events. A mother, in any culture, confirms her identity of mother through culturally appropriate behaviors and counter-identities. A Japanese mother—according to Doi's thesis- might optimally confirm herself as a mother through overindulging her child. An American mother, by the same token, would presumably confirm herself as a mother by engaging in acts express the individuality and independence of her child. ACT assumes that agreeable past experiences (e.g., the pleasant, identity-confirming feelings of having been overindulged as a child oneself) motivate humans to act in similar manner—as when a woman passes into the role of motherhood. In essence, cultural assumptions underlying the appropriateness or inappropriateness of any behavior derive from primal pleasant or unpleasant feelings attached through past experience.

When a human engages in culturally inappropriate behaviors, or with culturally inappropriate counter-identities, he or she experiences negative affect. Heise (1991) has empirically derived a measure he terms the Deflection Score, which measures the discomfort felt by a person who finds herself in an identity-disconfirming event. An event such as Mother Hugs Child is probably highly identity-confirming for both mother and child in any culture, leading to low deflections. Mother Scolds Child is likely to increase anxiety in both mother and child, but is typical enough that only mild increases in deflection should result. Mother Batters Child specifies a very high deflection score. The abnormality of such an event raises questions such as: What kind of mother would batter her child? What kind of child would cause a mother to child batter?

What is a normal or abnormal experience differs across cultures so that "dependence training" or "independence training" comes to carry different emotive meaning, cultural norms, role expectations for identities, and normative expectations, depending on early behavioral experiences of children. Over a half-century of psychometric work (Osgood, Miron & May 1975) demonstrates that these affectual differences are embodied in three culturally universal scales measuring the goodness, powerfulness, and liveliness of any concept.

Fundamental Cultural Sentiments

The Evaluation dimension denotes the goodness or badness of a concept. Collins and Kemper (1990) have shown that this dimension measures status cross-culturally. The Powerfulness dimension measures the potency or impotency of a concept and has also been demonstrated to have universal applicability by Collins and Kemper. The Activation dimension is defined by liveliness or inactiveness. Each scale has a range from -4 to +4, with zero denoting neutrality. The acronym of EPA is used in ACT to stand for the combination of these three scores.

The lexicons produced by ACT researchers provide quantitative measures of fundamental attitudes as these scores are highly resistant to change over time. For example, Heise (in press) and MacKinnon (1996) have shown virtually no quantitative EPA changes over two decades of data collection. Romney, Weller, and Batchelder (1986) suggest that these types of measures can provide highly reliable and valid measures of fundamental cultural knowledge with non-random sampling of as few as 25 informants.

It is possible to make valid comparisons cross-culturally of any concept semantically with these three scales. For example, American female informants rate Mother with fundamental EPA scores of 2.3, 1.9, and 0.0 in the United States. Compare these scores to Japanese female informants who rate Haha as 1.8, 1.5, and 1.0. The American fundamental identity is a significant half-point higher in status, 0.4 points higher in power, and 1 point lower in activation. (Roughly 0.4 point discrepancies denote statistically significant differences.)

Attributions and Identity-Disconfirming Events

Disconfirming events cause transient deviations away from fundamental cultural sentiments. ACT posits several solutions to identity-disconfirming events. One solution depends on culturally defined rules of attribution. Smith, Matsuno and Ike (unpublished) suggest that the Japanese and American processes of forming attributions differs in significant ways as shown in Figure 1. Americans and Japanese weigh the goodness or badness of an attribution and an identity equally in the identity-confirmation process, but Japanese weigh the potency and activation dimensions roughly twice that of Americans.

Table 1. The Semantics of Dependence (Izontekina) and Independence (Jiritsushita) for Americans and Japanese.

For characteristic attributions no problems are created for maintaining personal identities. For Japanese, the event Mother Hugs a Child is identity-confirming, causing Japanese to attribute such a mother with the characteristic emotion of satisfaction. However, the cultural differences in attributional rules in Figure 1 which leads to the image of a satisfied mother in Japan evokes a compassionate or affectionate mother in the United States.

Identity-disconfirming events—like a Mother Beats her Child — create greater attributional problems in Japan than the United States. Because deflection scores depend on squared discrepancies from expectation, and because attributes in Japan have double the weight of the United States, attribute-identity discrepancies exponential increase the hazard that Japanese will misattribute emotions (e.g., angry mother) or traits (e.g., hot-headed mother) compared to Americans.

Reidentification

As deflections rise to extremely high levels, ACT posits that a person's fundamental identity is unlikely to be maintained. A Mother Strangles a Child creates tremendous problems for maintenance of normal identities. The question "What kind of mother would strangle her child?" is unlikely to be satisfied by attributions of hot-headed or angry mother. The high deflections spoil the identity of mother in such a situation, leading the observer to relabel, or reidentify, the mother as a killer or psychopath.

The accumulated wisdom of social psychology, as operationalized by ACT suggests that humans prefer cognitive consistency. As MacKinnon (1994) explicates ACT, consistent emotions (satisfied mother), consistent counter-identities (mother-child), and consistent behaviors (hugs, feeds) lend credibility and confirmation to past events. By contrast, greater inconsistency leads to less credibility creating problems for the actors in, and observers of, events. ACT presumes that dispositional attributions like hot-headed mother are a more likely than reidentification through labels like killer. ACT assumes that it takes a particularly pernicious act like killing to completely spoil an identity.

The Logic of Analysis

The concept of amae is bound up with cultural expectations. Japanese culture stresses dependence while American culture underscores independence. First, we start out by comparing the affective meanings of dependence and independence across each culture semantically. Second, we search for behaviors that affectively confirm the dispositions of dependence and independence for Americans and Japanese. Third, we construct a series of identity-confirming and disconfirming events that simulate the embodiment, or lack, of amae based on the Japanese conceptualization of dependence for both cultures. The results indicate that there are greater similarities than dissimilarities in the roles of Mother and Child in both cultures than suggested by scholars since Doi first postulated the importance of amae to Japanese psychology.

The Affective Meanings of Dependence and Independence

The English concepts of independent and dependent are commonly translated into Japanese as jiritsushita and izontekina. Azuma (1991: 223) states that "independence prevails over other virtues" in American compared to Japanese culture. If Azuma is correct, we should expect Americans to give higher status to independence than Japanese do to jiritsushita, and Japanese higher value to izontekina than Americans do to independence. Table 1 indicates that Azuma may be incorrect. From column 2 it is clear that both Japanese men and women evaluate independence higher in status—and dependence lower in esteem—than American men or women. As Yamagishi (1996) shows in his theory of emancipation, Japanese social relationships allow for highly restrictive opportunities to leave groups. The Japanese concept of dependency implies high opportunity costs. The evaluation scores in column 2 are perfectly consistent with Yamagishi's theory, and inconsistent with the Asuma thesis.

Evaluation, potency, and activity scores are given in column 2 for both males and females. Column 3 displays the quantitative discrepancy between American men and women (A M/F), Japanese men and women (J M/F), American and Japanese men (M J/A), and American and Japanese women (F J/A). These deflections are computed by subtracting e, p, and a-scores of men and women, and Japanese and Americans. The differences are then squared and summed.

The smallest deflection (0.26) indicates strong semantic agreement exists for American men and women for the concept of independence. Similarly, Japanese men and women conceptualize izontekina (dependent) with little difference in basic meaning (0.37). There is slightly more discrepancy between American men and women in the meaning of being dependent with a D= 0.59. The greatest gender gap exists for the Japanese concept of jiritsushita. These gender differences warn us that the semantics of dependence-independence are influenced by sub cultures as well as cultures.

However, the cultural deflection scores in column 3 suggest that culture has much more influence than subculture, particularly for females. For males, the Japanese deflection for dependent-izontekina is 0.78, and for independent-jiritsushita it is 0.79. For females, the comparable deflections are 3.87 and 1.65. These numeric differences suggest that Japanese and American women must have very different experiences with dependence and independence. This is not surprising as amae is normally viewed as rooted in the mother's relationship to her child.

Column 4 uses the same deflection-scoring method to show which types of personal traits are viewed as clustering with particular concepts, by each sex and culture. All of the semantic equivalents in column 4 have deflection scores smaller than 0.4. For example, the American male concept of dependence is very close semantically to the Japanese male concepts of gullibility, unluckiness, and humiliation, and has the emotive meaning for Japanese females of drowsiness, being easily hurt, and intrusting. It is interesting to note that none of the four words in column 1 appear in column 4, underscoring the semantic gap that exists both between men

and women in the same culture, as well as the larger cultural gap separating Japanese from Americans. Amae may be very much immeshed in beliefs about dependence and independent, but this table alerts us to some of the cross-cutting effects of gender that we ought to consider.

The lack of overlap of traits by culture and gender suggests quite different traits are identity-confirming for males and females within Japanese culture, as well as when compared to American men and women. For example, an American male who displays traits that are a little nicer (compare 1.50 for independent to 1.83 for jiritsushita), a little less powerful, and considerably quieter than expected of an independent American male will likely be attributed as a clever American male. Similarly, a Japanese woman who displays traits in the range of an American woman's EPA profile for dependent, is likely to be perceived as drowsy or easily hurt by other Japanese women. Clearly, the cultural assumptions differ enough that what is denoted as the best translation connotes a very different set of cultural expectations—particularly when comparing Japanese and American women. Because amae is most strongly associated with cultural expectations for female behavior, this implies how amae-like behaviors and traits uniquely shape a Japanese mother's identity.

Behavioral Confirmation of Dependence and Independence

Just as we can use our measure of deflection, D, to search for clusters of traits, we can search for the behaviors that cluster semantically close to the traits of dependent, izontekina, independent, and jiritsushita. Because of the interaction of gender with culture in Table 1, Table 2 cross-classifies these clusters by sex of raters.

Table 2. Behaviors that Confirm Japanese and American Conceptions of Dependence and Independence by Sex of Raters.

As expected from the discussion of Table 1, Table 2 indicates the importance of affective meanings not only in terms of those evoked by language, but also in terms of identity confirmation and motivation in the course of social interactions. Schooler (1973) suggests that to understand the antecedents of adult psychological functioning, we need to examine differences in mothers' behaviors to their children..Table 2 clarifies the adult behavioral antecedents of independence and dependence by culture. Affect control theory predicts that the behaviors normatively expected of an izontekina Japanese woman would, in decreasing order, following, spurning, compromising with, serving, shunning, and coveting. A dependent American women, by contrast, ought to worship, submit to, beg, study, nudge, pamper, idolize, or sweet-talk.

Not to follow such advice can lead to a spoiled identity. For example, an American woman who indicates courting, bedding, warning, embracing, or sexually desiring is likely to be understood by other American as acting too much like an independent man—perhaps reidentified as butch, masculine, pushy, or aggressive. The most interesting part of Table 2 may be the jiritsushita row. Note that this is the only row in which men and women have considerable overlap in gender roles. That is, behaviors like cheering up and teaming with are perceived by the Japanese—and actions such as loving, saving, and curing by Americans—as genderless.

Identity-Confirmation of Amae in Mother-Child Relationship

Behavioral differences across cultures are understood by affect control theory through cultural norms, role expectations for identities, and normative relationships. The impression formation process is motivated by past experiences and the cultural assumptions underlying behavior. To examine the cultural expectations of amae, we use JavaInteract to advise us on the behaviors most semantically-associated with izontekina and amae for Japanese and Americans. The logic is to use JavaInteract to search for optimal behaviors associated with the EPA profiles for each of these traits within the confines of the classic mother-child relationship. The results are displayed in Table 3.

The top row of every cell identifies data for a female child; the bottom row a male child. Subtable A uses only Japanese EPA profiles and equations for identifying optimal behaviors and responses for various mother-child relationships. Although we have noted interesting gender differences in Tables 1 and 2, this table shows more similarities than differences. For mothers' optimal behaviors, the sex of the child makes no difference in either Japan or the United States. A mother's optimal response in Japan is to carry or hug her child of either sex. The Child's optimal response to being carried is to tickle or flatter the mother. The deflections are very low for both events, suggesting the normality of such sequences of events.

Table 3. Optimal Actor's Behaviors and Object Person's Responses for Mother and Child.

A. Using Japanese EPA profiles and equations:

B. Using Japanese EPA profiles for reidentifications; always with American equations:

Rows two and three of Part A are interesting because they show that a mother in either case is best off supervising a child who is amaeta (loved to death) or izontekina. In turn, an amaeta kodomo does have slightly different optimal responses to supervision, depending on sex. A mother ought to conduct an amaenbo. Although the original mother-child relationship has a low deflection, any of the child's reidentifications shown is slightly better, dropping from 2 or 3 to 1 point in all cases.

Subtable B, by contrast, indicates that a mother coaching a child, and a child playing with the mother, are even more likely than the optimal responses shown in Part A, Row 1 for Japanese. The most telling part of Part B is shown by the consistently higher deflections for rows 2,3, and 4 compared to the comparable rows in Part A. From an American's ethnocentric viewpoint, an amaeta kodomo is equivalent to a fool, an izontekina child to a bum, and an amaenbo to a squirt.

The asymmetry in the American mother-child relationship by comparison is fascinating. Japanese children whether reidentified or not reciprocate by playing a dependent role—begging, clinging to, following, serving, currying favor, tagging after, flattering, and buttering up—their mothers and generating equally low deflections in the process. But JavaInteract suggests that American children who take on those same characteristics respond differently by sex. Male children by challenging the mother, and grossly increasing the uneasiness of the relationship in the process; female children by becoming meek an submissive.

Discussion

Asuma (1996) claimed that amae is an ever-present positive feature in American as well as Japanese parent-child relationships, even where the American child is encouraged to become independent. The results of Table 3 suggest a more complex scenario in which young American boys are trained to be independent, but young American girls are encouraged to display dependence. Yet American mothers who attempt to confirm their identities through optimal behaviors with children who act outside of the normal child's role create little boys who rebel and little girls who are docile if they follow the Japanese rules of behaving.

An American mother best confirms her identity as a mother by coaching her child—an act that implies independence training. A Japanese mother is expected to carry or hug her child as connotated by the dependence inherent in amae. The Japanese version of JavaInteract mimics the overprotective and overindulgent attitudes of Japanese mothers. Still, this does not completely agree with Vogel (1996) who claims that Japanese mothers feel guilty if they are not all-giving, and who blame themselves if they do not appear all-loving. The Japanese mother who supervises or monitors her child is rewarded with uniquely identity-confirming responses like clinging and serving behaviors from the child not predicted for mothers and children in the United States.