Emotions and Culture 2

Psychological Bulletin Copyright 1991 by the American Psychological Association, Inc.
1991, Vol. 110, No. 3,426-450 0033-2909/91/J3.00
Culture and the Categorization of Emotions
James A. Russell
University of British Columbia
Vancouver, British Columbia, Canada
Some writers assume—and others deny—that all human beings distinguish emotions from non-
emotions and divide the emotions into happiness, anger, fear, and so on. A review of ethnographic
and cross-cultural studies on (a) emotion lexicons, (b) the emotions inferred from facial expressions,
and (c) dimensions implicit in comparative judgments of emotions indicated both similarities and
differences in how the emotions are categorized in different languages and cultures. Five hypothe-
ses are reviewed: (a) Basic categories of emotion are pancultural, subordinate categories culture
specific; (b) emotional focal points are pancultural. boundaries culture specific; (c) emotion catego-
ries evolved from a single primitive category of physiological arousal; (d) most emotion categories
are culture specific but can be defined by pancultural semantic primitives; and (e) an emotion
category is a script with both culture-specific and pancultural components.
Human beings divide the world into categories. We speakers morse, and an indefinable longing.. . . Under certain circum-
of English divide the colors into red, green, blue, and so on; the stances, however, it can have a very narrow meaning, a meaning as
definite, precise, and sharp as a well-honed cutting edge. I have
plants into trees, shrubs, grasses, and so on; and our kin into
never found an equivalent in other languages for this sense of the
aunts, uncles, cousins, and so on. People of other cultures word either, though I do not see how anyone can understand the
speaking other languages divide colors, plants, and kin into human soul without i t . . . . Litost is a state of torment caused by a
categories that are in some ways similar to and in some ways sudden insight into one's own miserable self.. . .Litost works like
different from what is done in English (Tyler, 1969). The topic a two-stroke motor. First comes a feeling of torment, then the
desire for revenge, (pp. 121-122)
of the present article is the categories into which people of dif-
ferent languages and cultures divide the emotions—not what I have been told of other emotion words in other languages
emotions exist or what outsiders observe, but what the insiders for which no word exists in English. An example from German
see among themselves. is the word Schadenfreude, which refers to pleasure derived
There are hints of unmistakable similarity in the categories from another's displeasure. Another is Angst: Walter Lowrie
of emotion even across great differences in language and cul- (1944) translated Kierkegaard's Der Begriff Angst under the title
ture. The following passage might have been taken from a mod- The Concept ofDread\»& said "the very title of this book reveals
ern textbook in psychology but is actually from the Li Chi, a a serious lack in our language: we have no word which ade-
Chinese encyclopedia compiled during the first century B.C. quately translates Angst" (p. ix). An example from Japanese is
from various documents of earlier periods: itoshii, which refers to longing for an absent loved one. Another
is ijirashii, which refers to a feeling associated with seeing some-
What are the feelings of men? They are joy, anger, sadness, fear, one praiseworthy overcoming an obstacle. An example from
love, disliking, and liking. These seven feelings belong to men
Bengali is obhiman, which refers to sorrow caused by the insen-
without their learning them. (Chai & Chai, 1885/1967, p. 379)
sitivity of a loved one. I was told by an Arab woman of her
There are also hints that different languages provide different delight on learning the English word frustration, because her
categories for the emotions. Here is a passage from Milan Kun- native language provided no word for that feeling. I read that
dera's (1979/1980) Book of Laughter and Forgetting: Homer's characters in the Iliad and the Odyssey had no concept
of guilt (Dodds, 1951).! also read of accidie, an emotion named
Litost is a Czech word with no exact translation into any other
in English but said to be extinct since the Middle Ages (Harre &
language. It designates a feeling as infinite as an open accordion, a
feeling that is the synthesis of many others: grief, sympathy, re- Finlay-Jones, 1986).
These examples raise the possibility that different languages
recognize different emotions. They carve up the domain of
This study was funded by a grant from the Social Sciences and Hu- emotion differently. Claims of this kind raise fundamental ques-
manities Research Council of Canada. tions for the psychology of emotion: Is it possible that the emo-
I thank Dare Baldwin, Peter Denny, Paul Ekman, Wally Friesen, tions are categorized differently in different languages? If so,
Paul Heelas, Carroll Izard, Batja Mesquita, Robert Levy, Catherine
how large and widespread are the differences? What emotions
Lutz, and Jane Wellenkamp for their comments on an earlier version of
might exist but go nameless in English? Might the concept ex-
this article. I thank Mita Banerjee, Lara Weick, Lisa Wong, and Janet
pressed by our word emotion be culture bound? What would
Wu for their bibliographic work.
Correspondence concerning this article should be addressed to such differences reveal about the nature of emotion categories
James A. Russell, Department of Psychology, University of British Co- or about where those categories come from in the first place?
lumbia, Vancouver, British Columbia, Canada V6T 1Z4. Electronic What would such differences reveal about the emotions them-
mail may be sent to userjaru at mlsg.ubc.ca. selves? What would be the implications for scientific theories of
426
CATEGORIES OF EMOTION 427
emotion stated in English? Would we be justified in using in guistic relativity, a hypothesis with a controversial empirical
other cultures our English words for emotions? On the other record (Brown, 1986; Kay & Kempton, 1984; Lucy & Shweder,
hand, is there good evidence of cultural differences; or are the 1979). An important study by Hoffman, Lau, and Johnson
examples cited above isolated curiosities, or perhaps even mis- (1986) underscores the possibility, however. Whereas most pre-
translated, romantic fantasies? Do lexical differences indicate vious studies of linguistic relativity had examined words in a
any real differences in how people think? What about evidence perceptual domain—color—Hoffman et al. turned to words in
familiar to all psychologists that recognition of facial expres- a social domain. The Chinese language provides labels for cer-
sions of emotion is universal, what does that evidence say about tain personality types (i.e, for certain clusters of personality
the role of culture in how emotions are categorized? characteristics) not labeled in English and vice versa. Hoffman
The relevant evidence—spread as it is across the literatures of et al. studied Chinese-English bilinguals to examine whether
psychology, linguistics, and anthropology—has not been these language differences influence other cognitive processes.
brought together, and these questions have not been given the Subjects were randomly assigned to have the experiment con-
attention they deserve. With this article, I attempt a step in that ducted in Chinese or in English. The result was an impressive
direction by examining evidence and hypotheses on how emo- display of linguistic relativity: Subjects' impression of and mem-
tions are categorized in different languages and cultures. I first ory for the same target's personality were found to vary with
examine the ethnographic record accumulated over the last cen- what language they used during the experiment.
tury on how emotions are labeled in different languages. I next I assume that categories are closely related to people's words
turn to cross-cultural studies on categorization of emotion by for the emotions. Neisser (1987, p. vii) argued that "language
facial expression and then to cross-cultural studies on concepts plays a key role in establishing categories, both developmentally
implicit in comparative judgments of emotion. Finally, I review and culturally" For example, the adult lexicon helps shape the
five hypotheses that attempt to account for the cross-cultural categories formed by the child (Mervis, 1987). Words like anger
evidence. designate not single events, but categories of events. Because
there are no Munsell chips for the emotions, one must examine
categorization of emotion indirectly, and much of the evidence
Scope, Perspective, and Rationale
necessarily involves words. Words express concepts by means of
My topic is words, categories, and concepts for emotions, which people categorize a part of their personal and social real-
rather than emotions themselves.11 draw no inferences about ity. There has been a recent interest in emotion categories specif-
emotions from people's words for the emotions, although such ically (Johnson-Laird & Oatley, 1986, 1989; Ortony, Clore, &
inferences have been drawn by writers on both sides of the Collins, 1988; Russell, 1991), and awareness of language differ-
nature-nurture question. Harre (1986) wrote, "Historians and ences may help in evaluating these hypotheses formulated on
anthropologists have established conclusively that there are his- the basis of the English lexicon. The present topic is thus part of
torically and culturally diverse emotion vocabularies. I claim the broad examination of the impact of culture and language on
that it follows that there are culturally diverse emotions" (p. 10). human cognition, an examination undertaken across the hu-
In contrast, Plutchik (1980) wrote, "The appearance in all lan- man sciences (Holland & Quinn, 1987). It is also an example of
guages of words like angry, afraid, and happy, suggests that a shift seen in most areas of psychology, a shift from the study of
these words represent universal experiences" (p. 102). a few convenient categories to the study of naturally occurring
Obviously we must get our lexical facts straight: Such is one ones.
purpose of this review. But the inferences one draws depend on Different traditions predict that categories into which people
one's theoretical assumptions as well as the facts. For example, divide the emotions would be found to be universal. One tradi-
one important theoretical idea in the psychology of emotion is tion stems from Darwin's (1872/1965) argument that the com-
that a person's emotional experience depends on how that per- munication of emotion, both its expression and its recognition,
son labels his or her own bodily state (Schachter & Singer, is part of our biological heritage. The theory that recognition is
1962). Imagine for a moment that this idea is correct. Imagine innate presupposes that the categories by which recognition
as well that words for the emotions vary from one culture to the proceeds are innate. A second tradition is the argument stem-
next, at least to some extent. It would follow that emotional ming from rationalist philosophy that most of our concepts are
experience, rather than being universal, would to that same innate (Fodor, 1981). A third is the idea, proposed by Boucher
extent vary with culture. More generally, those writers who em- (1979) and recently articulated by Johnson-Laird and Oatley
phasize the role of culture in shaping emotion (e.g, Averill, (1986, 1989), that words like fear and anger are undefinable
1980; Harre, 1986; Heelas & Lock, 1981; Hochschild, 1983; semantic primitives. Cultural differences in basic categories of
Lutz, 1980; Shott, 1979; Solomon, 1976) anticipate differences emotion would present difficulties for all these theories.
in the emotion lexicons of different cultures. Obviously, the topic of this article is also relevant to the psy-
I assume that emotion words are of interest because labeling chology of emotion. Consider the assessment of emotion.
of an emotion might play a role in subsequent cognitive pro- Words for emotion play a central role in ratings, judgment
cesses. For example, if emotion words vary with culture, then
persons from different cultures might encode, respond to, and
remember emotions in correspondingly different ways. This 1
Italics are used to emphasize the distinction between—on the one
specific possibility has not been subjected to empirical test and hand—words, categories, and concepts (e.g., emotion) and—on the
must today be viewed cautiously, because it is a version of lin- other—the events they are about (emotion).
428 JAMES A. RUSSELL
scales, and questionnaires. I take it for granted that psycholo- bull's (1961) ethnography of the BaMbuti Pygmies of the Congo
gists are interested in the emotions of all people, not just those is filled with descriptions of emotional life: "The Pygmy is not
who speak English. We must therefore understand the vocabu- the least self-conscious about showing his emotions; he likes to
laries of all people. laugh until tears come to his eyes and he is too weak to stand"
Perhaps the most far-reaching reason that the topic of this (p. 56), "with his chest puffed up with pride" (p. 61), "she was
article is of interest stems from the role that English words play very much in love with her husband" (pp. 132-133), "the girl is
in psychologists' theories of emotion. Anger, anxiety, sorrow, an object of suspicion, scorn, repulsion, and anger" (p. 186),
guilt, fear, happiness, and other English words for emotions "Saniwake was almost beside himself with grief" (p. 230). No-
have provided a ready-made means by which scientists classify where does Turnbull hesitate to attribute particular emotions to
emotions. The questions asked, the hypotheses formulated, and the BaMbuti, to use English words to describe those emotions,
the conclusions drawn in research on human emotion are built or to translate BaMbuti emotion words into English. Occasion-
with these words. Listen to what one linguist, Wierzbicka ally, and generally more recently, ethnographers have made this
(1986), had to say about the reason for studying words for the assumption explicit. For example, Shostak (1983) wrote, "My
emotions from different languages and cultures: hundreds of interviews with the IKung had shown me that
much of human emotional life was universal" (p. 7).
One of the most interesting and provocative ideas that have
Differences have also been observed concerning the fre-
been put forward in the relevant literature is the possibility of
identifying a set of fundamental human emotions, universal, dis- quency of, the causes of, the expression of, the importance of,
crete, and presumably innate; and that in fact a set of this kind has attitudes toward, beliefs about, and the regulation of emotion.
already been identified. According to Izard and Buechler (1980, p. In other words, it is assumed that what varies with culture are
168), the fundamental emotions are (1) interest, (2) joy, (3) sur-
events that surround the emotion. Emotion itself is universal.
prise, (4) sadness, (5) anger, (6) disgust, (7) contempt, (8) fear, (9)
shame/shyness, and (10) guilt. Although some writers do not share this point of view (Harre,
I experience a certain unease when reading claims of this kind. 1986; Lutz, 1982; Solomon, 1976), it is so widespread it can be
If lists such as the one above are supposed to enumerate universal termed the standard view. Two points need to be made regard-
human emotions, how is it that these emotions are all so neatly ing the standard view. First, the view cannot be stated in a
identified by means of English words? For example, Polish does
precise form because it is often unclear exactly what is assumed
not have a word corresponding exactly to the English word dis-
gust. What if the psychologists working on the "fundamental hu- to be the invariant universal emotion, once causes, conse-
man emotions" happened to be native speakers of Polish rather quences, expressions, and so on are removed. Second, the out-
than English? Would it still have occurred to them to include "dis- siders interpret a people's emotional life in this way does not
gust" on their list? And Australian Aboriginal language Gidjingali
necessarily imply that the insiders share this interpretation.
does not seem to distinguish lexically "fear" from "shame," sub-
suming feelings kindred to those identified by the English words
fear and shame under one lexical item (Hiatt, 1978, p. 185). If the
The Number of Words and Folk Theories of Emotion
researchers happened to be native speakers of Gidjingali rather
than English, would it still have occurred to them to claim that Languages differ in the number of words they provide to
fear and shame are both fundamental human emotions, discrete
categorize emotions. The ethnographic record does not give us
and clearly separated from each other? (p. 584)
enough information to make even an approximate count (which
Both Wierzbicka (1986) and the psychologists whom she ques- might, as we shall see, be impossible in any case) for each society
tions make the further assumption that the English word emo- to be reviewed here, but I can cite several attempts. At one
tion specifies the proper limits of the domain. That assumption extreme, Wallace and Carson (1973) found over 2,000 words for
too must be examined in light of culture and language. categories of emotion in the English language, although fewer
than one tenth of them are in most people's working vocabu-
lary Hoekstra (1986) found 1,501 words in Dutch for categories
The Ethnographic Record: Some Background Issues of emotion. Boucher (1979) found 750 words in Taiwanese Chi-
w nese and 230 words in Malay for categories of emotion. Lutz
Emotions Themselves
(1980) found 58 words in Ifalukian that were temporary inter-
Although my topic is words and concepts, the reader should nal states, although some of these would not be prototypical
understand something of ethnographers' view of the subject emotion words in English. At the other extreme, Howell (1981)
matter so conceptualized. Ethnographers have generally shown could find but 7 words in Chewong that could be translated as
only moderate interest in emotions (less in words or concepts categories of emotion.
for emotions). Mesquita and Frijda (in press) and Lutz and How many and what kind of categories for emotion are la-
White (1986) provided detailed reviews of the evidence on cul- beled in a language are related to a broader set of issues. No
tural variations in the emotions themselves. sharp line divides one's mental lexicon from one's mental ency-
Overwhelmingly, the ethnographies I have read on peoples of clopedia. That is, one's taxonomy for emotion cannot be com-
the world describe an emotional life easily understood in our pletely disentangled from one's cognitive model for the emo-
terms (Carstairs, 1967; Crapanzano, 1980; Eggan, 1970; Fortes, tions so taxonomized. By cognitive model, I mean one's struc-
1970; Hallowell, 1955; Hollan, 1988; Kracke, 1981; Lindholm, ture of beliefs concerning what brings each emotion about,
1982; Potter, 1988; Swartz, 1988; Tumbull, 1961,1972; Turner, what its mechanisms are, what to do about it, how to evaluate its
1967). Commonly, especially in the past, the ethnographer sim- occurrence, and so on. Levy's (198 3,1984) ethnographic studies
ply assumed the universality of emotion. For example, Turn- of the Tahitians and of the Newars of Nepal led him to conclude
that different societies possess different cognitive models of sia (Lutz, 1980,1983), the Chewong of Malaysia (Howell, 1981),
emotion. Levy described as hypercognized an emotion for nor the Samoans (Gerber, 1975, p. 187) have a word for emotion.
which a society possesses an elaborate cognitive structure. One Of course, in cultures lacking a word for emotion, the concept
index of hypercognition is a large number of lexical entries for might be implicit. For example, although the Tahitians have no
that emotion. That anger triri) is hypercognized in Tahitian word for emotion or feeling, Levy (1984) argued that emotion is
includes the fact that Tahitian has 46 separate terms for types of an implicit class in Tahitian thinking, noting that the various
anger, just as English has annoyance, rage, fury, irritation, and emotions have a set of features in common: They are thought to
the like. arise in the intestines, they involve the whole person rather than
Levy (1984) described as hypocognizedan emotion for which merely a part, they can lead to action, and they invoke the
a society possesses little knowledge. One index of hypocogni- relationship between the person undergoing the emotion to the
tion is having few or no lexical entries. That sadness is hypocog- external social or physical environment. What is not clear is
nized in Tahitian includes the fact that there is no concept of whether this grouping is made by the Tahitians or by Levy.2
sadness in Tahitian. People whom Levy would describe as sod Some languages contain words similar to emotion, although
would be described in Tahitian by means of more general terms emotions are grouped with words English speakers would not
such as pea pea, a generic word for feeling ill, troubled, or consider emotions. For example, the Samoans have no word for
fatigued. Levy (1973, p. 303) describes a man separated from emotion but do have a word, lagona. that groups together feel-
his wife and child. Interpreting the man's reaction as the emo- ings and sensations (Gerber, 1975). Gerber found that the clos-
tion of sadness, Levy saw the separation as its cause. Interpret- est she could come to denning emotion for the Samoans was to
ing his own reaction as pe'a pea, the man did not. talk about the loto, which is the bodily location of lagona, much
as the heart is the bodily location of emotion for speakers of
H English. \et, some of the most commonly used concepts for
The Ethnographic Record: Lexicons of Emotion feelings of the loto are not prototypical emotions in English—
for example, lotomaualuga and lotomaualalo, which are feelings
Concept of Emotion *"^
of arrogance and submissiveness, respectively Gerber tried to
By including the word emotion, the very title of this article delimit a set of words in Samoan that corresponded to the do-
presupposes that the domain of interest has already been delin- main of emotion. It is not clear how well she succeeded or if the
eated. The word emotion provides the English-speaking psy- set achieved corresponded to anything the Samoans recognize
chologist with a predefined field of inquiry The concept of as a class.
emotion is typically presupposed by already being embedded in The Ifaluk lack a word that translates as emotion (Lutz, 1982).
the questions asked: What is an emotion, what causes emotion, Instead, the Ifaluk distinguish nunuwan from tip, subsuming
and so on. One function of taking a cross-cultural perspective is both under the category niferash, which Lutz (1980) translated
to raise the question of whether other peoples speaking other as "our insides" (p. 97). The distinction between nunuwan and
languages draw the same boundary implied by the word emo- tip is between socially standard processes and individual, idio-
tion. Unfortunately, ethnographers have only recently raised the syncratic ones.
question whether emotion is a recognized domain in the cul- The Chewong implicitly group together feelings and thoughts
ture they described. by placing them both in the liver: The liver "is the seat of both of
There is some evidence that the concept emotion is wide- what we call 'thoughts' and 'feelings' and [the Chewong] do not
spread, if not universal. Brandt and Boucher (1986) assembled a make any conceptual distinction between the two" (Howell,
cross-cultural research team that included members of six non- 1981, p. 139). The Temiar of Malaysia also implicitly group
English-speaking societies. The languages studied were Indo- together feelings and thought by placing them both in the heart.
nesian, Japanese, Korean, Malay, Spanish, and Sinhalese. Col-
laborating researchers, who were fluent in English and their
native language, assured the authors that each language to be 2
In a personal communication in 1989, Levy replied: "\bu note that
studied contained a term conceptually equivalent to the English
it is not clear whether the grouping is made by the Tahitians or me for
word emotion. The criterion for conceptual equivalence was not the 'implicit' category of emotion. I made the grouping in the sense
stated and is unclear. For example, Japanese was one of the that it is my putting together of what I take to be, by analysis, a covert
languages said to have a word for emotion, but there is reason to but real aspect of their psychocultural organization—this is the way
question this assertion. Matsuyama, Hama. Kawamura, and one makes judgments about any 'covert' cultural form. Whether the
Mine (1978) presented an analysis of emotional words from the Tahitians would themselves recognize the category as being 'true' or
Japanese language. The word translated as emotion, jodo, cer- possible or interesting alheit un-named, I don't know—I didn't try to
tainly included states that would be considered typical emo- find out—although that would interest me now that you call my atten-
tion to it. But even if they didnt do any of those things it wouldn't
tions—angry, happy, sad, and ashamed—but jodo also in-
necessarily mean they didn't 'have' the 'implicit category' in some
cluded what might not—considerate, motivated, lucky, and cal-
sense. Actually of course the problem of what we might mean by ascrib-
culating.
ing to someone an unnamed implicit category is an interesting one.
There is also evidence that the concept of emotion is not Perhaps the most I should have stated is that among the phenomena
universal. Neither the Tahhians (Levy, 1973, p. 271), the Bimin- that I took to be feelings there are a group which shared several features
Kuskusmin of Papua New Guinea (Poole, 1985), the Gidjingali which did not characterize the others, and that this group was equiva-
aboriginesof Australia (Hiatt, 1978), the Ifalukiansof Microne- lent to what we (whoever we may be) take to he the 'social emotions.'"
The head is the location for language and expression (Roseman, ing about Luganda, which is spoken by the Buganda of Uganda,
1988): "Temiar distinctions differ from the traditional Western Africa:
dichotomy between thought and feeling, and turn rather on the
The difference between anger and sorrow is not stressed to the
distinction between inner experience [heart] and vocalized ex-
same extent as in English, and it is not uncommon to hear an
pression [head]" (pp. 11-12). interpreter using okusunguwala (to get angry) as a translation of to
The possibility that emotion is not a recognized domain in all get sad, nor would he really feel he had made a mistake even when
cultures is perhaps the most important issue about emotion "corrected." If one wishes to speak specifically about the grief of
mourning or of a friend's departure, then the verb okusaalirwa is
raised by the ethnographic record. At the same time, that possi-
used, but even then I have heard of men mourning at a funeral
bility creates difficulties that must not be ignored for the inter- being said to be basunguwadde (angry), (p. 3)
pretation of evidence to which I turn shortly. In a culture that
lacks the concept of emotion, it is difficult to know whether a Davitz (1969, chapter 6) asked Bugandan and American adoles-
given word in the language of that culture should be thought of cents to describe an incident in their lives involving anger. The
as referring to an emotion. To give an example, Riesman (1977, Bugandans were bilingual in English and Luganda. About a
p. 129) analyzed the central African Fulanfs concept of sem- third of the Bugandans mentioned that they cried during the
teende, which is commonly translated as shame or embarrass- incident, whereas none of the Americans had. This difference
ment. Closer analysis showed that semteende depends more on occurred whether the Bugandans gave their report in English or
the objective situation than on a person's inner feelings. Some- in Luganda.
one is in a state of semteende if the situation is appropriate to Some blurring between anger and sadness can also be seen
semteende, whatever the person may or may not feel. If so, sem- among the llongot, a head-hunting group living in the Philip-
teende may not refer to an emotion at all. Perhaps the compara- pines. Central to their culture is liget, to which I return, which is
ble situation would exist in English if guilt referred exclusively commonly translated as anger. Liget covers not only what anger
to legal guilt. Riesman didn't mention asking the Fulani covers, however, but a range of emotions including grief. In fact,
whether semteende is an emotion, and it would not have been according to the llongot, bereavement leads to such intense liget
possible if the Fulani lack the concept of emotion. that head-hunting is the natural form of release (R. I. Rosaldo.
1984).
Similarly, the Ifalukian word song covers what in English
Categories of Emotion
would be described as anger and sometimes sadness. Lutz
This section lists those claims in the ethnographic record (1980) translated song as justifiable anger, and the facial cues,
concerning differences between languages in labeled categories situations, and tendency to violence with song support that
of emotion. The focus is limited to the semantics rather than view. But song also indicates a state in which the person cries,
pragmatics of emotion words and on ordinary—as opposed to pouts, and inflicts harm on himself or herself, including sui-
artificial, technical, or scientific—words. Of necessity, not pref- cide.
erence, English is taken as the standard of comparison. Ex- Various claims touch on the English word shame. The En-
cluded are claims made by ethnographers without firsthand glish distinction between shame and fear is not made by the
experience with the particular language and cultural group Gidjingali aborigines of Australia (Hiatt, 1978). Both are cov-
studied, or what appeared to be mere speculation. Entries in ered by the same word, gurakadj. The English distinction be-
glossaries and dictionaries are also excluded unless supple- tween shame and embarrassment is not made by the Japanese
mented with adequate explanation. Likewise excluded are im- (Lebra, 1983, p. 194), by the Tahitians (Levy, 1973, chapter 10),
plicit claims; for example, with one exception indicated specifi- by the Ifalukians (Lutz, 1980, p. 209), by Indonesians (Keeler,
cally, I do not cite a language as lacking a specific word when the 1983, p. 153), or by the Newars of Nepal (Levy, 1983). Indeed,
only evidence would be that the ethnographer failed to mention Levy (1983) claimed that shame and embarrassment "seem to
that word. The claims can be divided into two classes, as de- be a lexically unified cluster in many or perhaps most parts of
scribed in the next two sections. the non-Western world" (p. 131). The llongot (M. Z. Rosaldo,
Some English emotion words have no equivalent in some other 1983, p. 141) use one word, belong, to cover shame, timidity,
language. For example, English distinguishes such words as embarrassment, awe, obedience, and respect. The Javanese use
terror, horror, dread, apprehension, and timidity presumably as isin to cover shame, guilt, shyness, and embarrassment (Geertz,
types or degrees of fear. In Gidjingali, an Australian aboriginal 1959, p. 233). And the Pintupi use one word, kunta, to cover
language, one word, gurakadj, suffices (Hiatt, 1978). Of course, shame, embarrassment, shyness, and respect (Myers, 1979, p.
with over 2,000 emotion-denoting words in English, it is not 361). Other ethnographers have simply pointed to a problem in
surprising that some are unique, and words for subdivisions translating shame into other languages. In an ethnography of
within basic emotions may have little significance. But even the Fulani of Central Africa, Riesman (1974/1977) studied the
some of the more important English words have no equivalent word semteende, previously translated shame, but concluded
in some languages. "this translation is not adequate" (p. 1347). Langness (1965)
Some languages do not distinguish clearly what English treats made a similar observation about the word usually translated as
as separate basic-level emotion categories. Leff 0973, p. 301) shame in the language of the Bena Bena of New Guinea. In an
pointed out that in some African languages the same word ethnography of the Balinese, C. Geertz (1966) studied lek, previ-
covers what we would distinguish as anger and sadness. Other ously translated as shame or guilt, but also concluded the trans-
evidence supports his claim. Orley (1970) observed the follow- lation was inappropriate. The studies cited so far took the En-
glish word shame as the reference point; Epstein (1984) exam- evidence for inclusion in this section, but there is one report
ined various Melanesian words translated as shame and found that merits special attention. Howell (1981,1984) studied the
that they differ from one another. Chewong, a small hunter-gatherer group in the rain forest of
Several similar claims are noted by single ethnographers. Sa- Malaysia. Howell (1981) made "special efforts to record as large
moans use one word for hate and disgust and one word, alofa, a vocabulary as possible" (p. 133) in the realm of emotions and
for love, sympathy, pity, and liking (Gerber, 1975, p. 3). Briggs other inner states. Table 1 sets out all the words Howell man-
(1970, p. 326) detailed how the Utku, an Inuit band in Canada, aged to obtain that pertain to inner states. Of these, I classed
do not distinguish between feelings of kindness and gratitude seven as among the emotions. Think of all the emotional states
and instead have one word, hatuq, to designate both. coded in English, subtract these seven, and you have an extraor-
There are also claims of important English words missing dinary number of emotions not coded in Chewong.
altogether in another language. The two that have been noticed Some languages have words without an equivalent in English.
the most frequently were ones needed for psychiatric classifica- German distinguishes Schadenfreude from other types of plea-
tion.3 Marsella (1981) reviewed various studies and concluded sure. In English, disgust refers to the feelings associated with
that there is no word for depression among many non-Western decaying matter as well as the feelings associated with moral
cultural groups. Collaboration comes from observations of the indignation. Ifalukian distinguish these cases, using niyabut for
\bruba of Nigeria (Leighton et al, 1963), from various North the former, song for the latter (Lutz, 1980, pp. 183-184). The
American Indian languages (Termansen & Ryan, 1970), from Tahitians distinguish fear caused by a ghost from other sorts of
Malay (Resner & Hartog, 1970), from Chinese (Chan, 1990; fear; they also have separate words for emotional states in which
Tseng & Hsu, 1969), from Eskimo (Leff, 1973), from Fulani in an inner feeling differs from the outer expression (Levy, 1973,
Africa (Riesman, 1977, p. 156), from the Kaluli of Papua New pp. 96-98). Indonesian, which does not distinguish shame
Guinea (Schieffelin, 1985), and from the Xhosa of Southern from embarrassment, does distinguish shame/embarrassment
Africa (Cheetham & Cheetham, 1976). brought on by one's own deeds, malu, from shame/embarrass-
Apparently no word equals anxiety among the Eskimos and ment brought on by someone elsels deeds, dipermalukan
Yorubas, either (Leff, 1973, p. 304). Cheng (1977, p. 15.1) found (Keeler, 1983, p. 153). Similarly, English distinguishes annoy-
no exact translation in Chinese for anxiety, although two words ance, fury, rage, and irritation, presumably as types or degrees of
(perhaps better translated as tension and worry) come very close. anger. Malay distinguishes types of anger, marah, but the types
Johnson, Johnson, and Baksh (1986) found no word for worry do not correspond to the English distinctions (Boucher, 1979,
among the Machiguenga of Peru. p. 171).
Levy (1973) observed that the Tahitians have "no word which Briggs's (1970) study of the Utku yielded no exact equivalent
signifies anything like a sense of guilt" (p. 342). A word for guilt for love because naklik (which is love for those who need protec-
is missing from the Sinhala language of Sri Lanka (Obeyese- tion, such as babies, puppies, or the sick) is distinguished from
kere, 1981, p. 79), from the llongot language of the Philippines niviuq (which is love for those who are charming or admired).
(M. Z. Rosaldo, 1983, pp. 139-140), from the Pintupi language Ifaluk does not have a term that corresponds to surprise be-
of aboriginal Australians (Morice, 1978, p. 93), and from the cause pleasant surprise, kcr, is distinguished from unpleasant
Samoan language (Gerber, 1975). Guilt is subsumed under me- surprise, rus (Lutz, 1980). Ifaluk does not have a general word
tagu (fear or anxiety) for the Ifalukians (Lutz, 1980, p. 223). for fear but distinguishes fear of future events, metagu, from
Indeed, Gerber (1975, p. 3) wrote that there is a "notorious the panic, surprise, and fear that is due to confrontation with a
absence of a term equivalent to guilt in many Asian and Pacific present event, rus (Lutz, 1980, p. 188). Similarly, Utku contains
languages." no exact equivalent of fear because fear of physical injury, iqhi,
The Quichua of Ecuador lack a word for remorse (Tousig- is distinguished from fear of social injury, Him. Moreover, Uira
nant, 1984). The Ifaluk lack a word for surprise, which has been also includes what in English might be termed respect. Morice
considered a basic discrete emotion by English-speaking emo- (1978) found no Pintupi word for fear but lists 15 words refer-
tion theorists (P. Ekman, 1972, Izard, 1977). The Nyinba of ring to kinds of fear or its concomitants. The divisions in Pin-
Nepal lack a word for love (Levine, 1988). No word covers both tupi do not correspond to those in English: Ngulu is fear of
parental and sexual love. Parental love and similar feelings to- another seeking revenge; nginyiwarrarringu is a sudden fear that
ward the weak and dependent are assimilated to concepts that causes the person to stand up to see what caused it; wurrkulinu
are close to compassion. Sexual love is assimilated to concepts is worry over land or relatives. In contemplating these particu-
that are close to greed or desire.
Because sadness has been listed as a basic emotion, let me
3
repeat Levy's (1973) observation, mentioned earlier, that the Also, numerous descriptions of pathological states, labeled by the
Tahitians lack a word for sadness. The Tahitians use more ge- culture, have no equivalent English term (Gobeil, 1973; D. Johnson &
Johnson, 1965; Kiev, 1968; Langness, 1965; Rubel, 1964; Westermeyer,
neric words for sadness, treating sadness, fatigue, longing, loneli-
1972). For example, amok is a term found mainly in the Malay archi-
ness, depression and the like not as emotion but as something
pelago. It refers to an intense state characterized by delusions and vio-
closer to physical illness (Levy, 1973). lent assaults directed against friend and foe alike, followed by amnesia
Unless an observer is specifically looking for the indigenous and deep sleep (Carr & Tan, 1976). We are not inclined to include such
equivalent of a particular English word (as when needed for states as amok among the emotions, but perhaps this is simply because
psychiatric diagnosis, for example), its absence may go unno- amok is not an emotion term in English. For good overviews of this
ticed. Absence of a word from a glossary was not adequate area, see Kirmayer (1989) and Simons and Hughes (1985).
Table 1 feeling state associated with respect, is something peculiarly

Chewong Lexicon for Inner States Javanese" (p. 233). H. Geertz gave several other Javanese words
that appeared to lack exact English equivalents. A commonly
Chewong word English translation
used word, bingung, refers to being upset, confused, and lack-
Emotion words ing a sense of direction. Kaget refers to being startled by some-
thing that happens outside oneself, so that one becomes bin-
Chan Angry
gung. Iklas refers to a state of pleasant, or at least indifferent,
Hanrodn Proud
Hentugn Fearful, frightened frustration.
Lidya Ashamed, shy According to Wikan (1989), the Balinese word tekajut, ini-
Meseg Jealous tially translated as fright, appears on closer scrutiny to be
Punmen Like unique. Unlike fright, tekajut is sharply distinguished from
Osayagn (sayang)" Fond of
fear. Tekajut is the unpleasant and upsetting response to the
Other states unexpected. Unlike startle, however, tekajut need not require
suddenness. Tekajut is considered beyond choice or personal
Abud Hot (body)
Full (stomach) responsibility, whereas the Balinese equivalents of fear, anger,
Bohi
Gadd Thirsty and jealousy involve considerable responsibility.
Greno Sexually aroused Gerber (1975) analyzed 44 commonly used Samoan words
Haratn Know, understand that come as close as possible to being categories of emotion.
Hengkong Hungry
Gerber's translations did not always even sound like emotions,
Imeh Want
Kenjed Mean, stingy even though her Samoan informants assured her that these
Ion Want very much were feelings in the loto: lotomalie (agreeableness), matamau
Mund Miss, remember (generosity), bnosa'i (patience), faalotolotolua (indecision). Per-
Opriya Lie haps the clearest example is lotomaualalo, which refers to a
Pesedd Pain
pleasant feeling—an absence of malice, anger, or resentment in
Sedeig Cool (body)
Tokad Cold (body) situations of potential conflict in which these feelings might be
Aga(agak)4 Guess expected to arise.
Btani (berani)1 Brave Fajans (1983) described awumbuk as a "peculiarly Baining
Duga (duga)* Esoteric knowledge
sentiment" (p. 177); the Baining are a people of Papua New
Malas (mains)' Lazy
Moda (mudah)* Generous Guinea. Awumbuk is a sadness, lassitude, tiredness, and bore-
Olupa (lupa)1 Forget dom caused by the departure of visitors, friends, or relatives.
Rayitn (rajin)* Energetic The IKung have a word, kua, for the combination of awe,
respect, and fear (Shostak, 1983) associated with the formally
Note. From "Rules Not Words" (pp. 133-143) by S. Howell, 1981, in P.
and ceremonially recognized milestones in one's life, but it can
Heelas & A. Lock, Indigenous Psychologies: The Anthropology of the
Self, San Diego, CA: Academic Press. Copyright 1981 by Academic also occur in response to danger, such as encountering a lion or
Press. Adapted by permission. walking alone at night.
* Malay (or derivative of Malay) words. Levy (1973, p. 307) examined the Tahitian word ri'ari'a, com-
monly translated as fear, but concluded that it could not be an
exact translation. He found that although riaria covers many
lar distinctions, one must not assume that the Ifaluk think of instances labeled as fear in English, it does not include fear
metagu and rus as types of fear or of ker and rus as types of about possible future events, which is central to the English
surprise, or assume that the Utku think ofnaklik and niviu as notion of fear. Tahitian includes a separate term, mata'u, for fear
types of love or iqhi and Him as types of fear, or assume that the of the future. On the other hand, riaria does include the feeling
Pintupi think of nginyiwarmrringu and wurrkulinu as types of of being mildly repulsed, as, for example, in response to food.
fear, just because that is how we speakers of English would In her study of the Ifaluk, a people of Micronesia, Lutz (1985)
classify these various states. studied the word nguch, which captures what in English must
In the cases above, English equivalents might be formed as be said metaphorically as "sick and tired" or "fed up." Nguch
disjunctions of the foreign words. In other cases, this is not so. also includes feelings of boredom and lethargy that are due, for
Doi (1973) concentrated years of study on a single Japanese example, to extreme heat, weariness, or illness.
word, amae, for which no word exists in English. Amae is a The most thorough analysis of a single emotion word may be
pleasant feeling of dependence on someone: the feeling Catho- M. Z. Rosaldo's (1980) study of the Ilongot liget, which is com-
lics have toward Mary, the mother of Jesus, or an infant has monly translated anger: "I began to see in a term that I had
sucking the sweet milk of its mother. Doi (pp. 20-21) remarked understood initially to mean no more than 'anger' a set of prin-
that the closest he has come to hearing this idea expressed in ciples and connections with elaborate ramifications for Ilongot
Western thought was the psychoanalytic notion of passive ob- social life" (M. Z. Rosaldo, 1980, p. 45). Like anger, liget can be
ject love, with an emphasis on passive. caused by insult or injury. But liget can also be aroused by a
In her study of the Javanese, H. Geertz (1959) wrote: "Wedi communal, all-night song fest; pride of accomplishment; or the
and isin, although complex, are close enough to American ideas death of a loved one. Liget can be manifested in irritability or
to be translated as 'fear' and 'shame' or 'guilt,' but sungkan, a violence, but it also can be manifested in the sweat of hard
work. Liget is shown when a man hunts with courage and con- which emotion is described in English suited that society and
centration or when a woman prepares a good meal. Liget is a assumed that native words could be accurately translated into
highly valued force, vital to social and personal life. English.
Some of the examples so far suggest that emotions named in
other languages can be described in an English phrase. Lutz Translation Equivalence
(1980) provided the clearest counterexample, however, in her
analysis of a single emotion word, fago, which is in daily use None of the HRAF reports just reviewed attempted to estab-
among the Ifaluk. Her observations on how fago is used in daily lish the actual equivalence of words in different languages. Re-
life, interviews about experiences of fago, and requests for ex- ports cited here earlier sometimes pointed out that what was
plicit definitions of fago leave little doubt that ft is difficult to once taken as a translation equivalent turned out, on closer
translate into English. Fago is felt when someone dies, is needy, inspection, not to be so (Davitz, 1969; Levy, 1973; M. Z. Ro-
is ill, or goes on a voyage, but fago is also felt when in the saldo, 1980; Wikan, 1989). This work raises a fundamental ques-
presence of someone admirable or when given a gift. Fago is tion: Even where translation equivalents initially appear to ex-
used in some situations in which English speakers would use ist, are they really equivalent?
love, empathy, pity, sadness, and compassion—but not in all Five studies have taken up the question of translation equiva-
such situations. lence for emotion words. Tanaka-Matsumi and Marsella (1976)
first used the standard procedures of translation, back transla-
tion, and consultation of bilingual dictionaries to obtain the
Prevalence best Japanese translation for the English word depression. The
How widespread are differences in the lexicon of the emo- result was yuutsu. They then compared the free associations
tions? Ideally, I would tell you how many languages categorize given by English speakers to the word depression with those
emotions differently than does English and, for each language, given by Japanese speakers to yuutsu. The associations were
how many differences. Aside from the problem of counting quite different. Tanaka-Matsumi and Marsella (1977) then
emotion words, it is difficult even to estimate the number of asked subjects to rate depression and yuutsu on semantic differ-
equivalences, because ethnographers may tend to report differential scales. Separate analyses for the terms yielded different
ences more than similarities. Similarities may be less notice- factor structures—a damning piece of evidence if, as Osgood
able. Or because of the doctrine of the psychic unity of human- (1969) argued, the semantic differential measures affective
kind, especially when emotion is the topic, ethnographers may meaning. Independent studies by Imada (1989) and Imada,
sometimes take similarities for granted and not bother with Araki, and Kujime (1991) came to a similar conclusion regard-
them. Thus when mentioned at all, similarities tend to be men- ing yuutsu/depression as well as two other common transla-
tioned in passing, often to contrast with differences. Before tions: kyofu/fearand. fuan/anxiety. These six words were emp iri-
describing in great detail differences in certain emotion words, cally denned in terms of Davitz's (1969) empirically obtained
Lutz (1985) mentioned that "relatively adequate American En- list of 556 statements. The overlap of defining statements was
glish glosses can be found for many Ifalukian emotion words" 48.4% for yuutsu/depression, 42.0% for kyofu/fear, and only
(p. 43). Although they could find no Machinguengan word for 9.8% for fuan/anxiety. Furthermore, semantic differential rat-
worry, A. Johnson et al. (1986) did find what they considered ings showed significant differences between Juan and anxiety.
exact translations for fear, happiness, anger, and sadness. Many Finally, Chan (1990) obtained free associations for you-yu, the
ethnographers have assumed that all words for categories of closest translation of depression in Chinese and a relative of the
emotion in the language of the culture being studied could be Japanese yuutsu. Although no comparison group speaking an-
translated into English. All this suggests that there is a great other language was included, Chan concluded that you-yu and
number of equivalences or at least similarities. depression appeared to be quite different.
To obtain preliminary information on the question of preva- Of course, free associations, semantic differential ratings,
lence required some means of sampling the entire ethnographic and Davitz^ (1969) list may not be definitive assessments of
record. I therefore turned to the Human Relations Area Files meaning. No more definitive tests have been offered, and the
(HRAF) at the University of British Columbia. The HRAF lead of these pioneering studies has not been followed. Never-
contain ethnographic reports on 324 different societies, with theless, the cautious researcher will no longer assumes that
the reports annotated for content. The HRAF were searched for emotion words in different languages can be translated one-to-
every report that fit the following criteria: (a) the report con- one. If so, a methodological mainstay of cross-cultural research,
cerned a non-Western society and (b) at least four pages of mate- translation and back translation by bilinguals, may be problem-
rial in the report had been annotated as concerned with emo- atic. The back-translation criterion of success is insufficient be-
tion. cause it can only achieve the best translation, which might not
This search yielded 114 records, which were read for infor- be an exact equivalent.4
mation about the way in which that society spoke about the
emotions. The idea was to record the number of ethnographic 4
Levy (1973, p. 303) observed that some Tahitians bilingual in Tahi-
reports that claimed some difference in that society's emotion tian and French incorrectly used the French word trisle(sad) as synony-
lexicon and then to count the number of such differences. The mous with fatigued and the like. The Tahitian language does not in-
result is easy to describe: Not a single example was obtained. clude a word for sad but includes sad feelings within a broader category
That is, in every case, the ethnographer assumed that the way in with fatigue, loneliness, and lacking enthusiasm. Davitz (1969) ob-
Discussion in English. Gerber (1975) presented the Samoan word alofa as

meaning love, sympathy pity, and liking. But it is also conceiv-
I have just discussed claims that English words often as-
able that alofa and fago are similar or equivalent in meaning.
sumed to denote natural basic categories of emotion—words
That is, fago might be rendered approximately as love, sympa-
such as love, anger, sadness, fear, and of course emotion itself—
thy pity, and liking; or, alternatively, alofa might have no precise
have no equivalents in some other languages. And other lan-
equivalent in English. Another example might be the IKung
guages provide commonly used emotion words with no equiva-
concept kua (Shostak, 1983), presented as without a simple
lent in English. Although such differences do not appear to be
translation. Kua may be like the words found in various societ-
prevalent, even the possibility is important. Furthermore,
ies that combine shame, respect, awe, and embarrassment.
words now taken to be equivalent might not be, and this possi-
In short, the precise claim in the ethnographic record may
bility further opens the door, perhaps to widespread differ-
depend on the style of reporting used by the ethnographer. The
ences. The ethnographic record reviewed here generally im-
emotion lexicon of the culture being studied is often described
plied that something more than lexical differences were at
in relation to the English lexicon for emotion. But English may
stake, that differences in concepts were revealed. Yet, before
not provide an adequate means by which to convey the meaning
drawing any conclusions, I must ask how strong is the evidence
of words in all other languages. Put another way, we lack a
for the claim of lexical differences.
precise criterion on what constitutes equivalence and what con-
One difficulty is that the evidence available focuses on those
stitutes nonequivalence. Wierzbicka (1986) has most clearly ar-
categories coded into single words. Languages can express
ticulated this problem and calls for the creation of a metalan-
many more ideas than are coded in single words. Even if a
guage in terms of which any emotion word in any language can
language lacks a word for, say, guilt, there remains the possibil-
be denned. A corollary to the lack of a criterion of equivalence
ity that guilt may be expressed in a phrase, or metaphorically, or
is that no mechanism exists to resolve disputes. Wierzbicka
even nonverbally. My topic was the lexicon; other aspects of
claimed that the Polish word tesknic has no exact equivalent in
language also convey emotional meaning (Besnier, 1990), al-
English. Kolenda (1987) claimed that ft does: longing.
though the concepts of emotion implicit in other aspects of
One could also question the interjudge reliability of ethno-
language have not been articulated. Any discussion of this evi-
graphic reports. The claims reviewed here vary in how much
dence must also point out that ft concerns current practices in a
information was given in support. Unsupported assertions
culture, not capacities. There is no reason to believe that one's
were indicated as such, but for others, anecdotes, quotations,
language or culture limits one's ability to comprehend or learn
and responses to questions in varying amounts were offered.
other practices.
The claims also varied in how many writers made that observa-
Another difficulty is that all claims rely on the ethnographic
tion. For example, differences surrounding shame were widely
method. I presented claims only in cases in which I believed the
reported, but for most claims, the modal number of ethnogra-
support was good, in which the ethnographer had sufficient
phers was one. That this is a problem can be seen by looking at
familiarity with the language and people, and in which the
instances in which different ethnographers examined the same
claim was supported by anecdotes, quotations, responses to
or similar cultures. Hiatt (1978) claimed that the Gidjingali
questions or incidents, and the like. There is no standard means
aborigines of Australia use a single word for fear and shame,
of supporting such a claim; therefore, it would be cumbersome
lacking any terms for differences within this broad class. Hiatt
to include in this article supporting information in each case.
also speculated that other Australian aboriginal languages did
At the same time, a word of caution is in order. Until these
the same (I ignored speculations of this sort in the review
observations are confirmed through other methods, it remains
above). In contrast, Morice (1978,1979) described the Pintupi
difficult to say how much credence they deserve. I say this for a
aborigines of Australia as lacking a generic word for fear but
number of reasons.
distinguishing 15 kinds of fear. Morice (1979) also claimed that
There is today no adequate means by which an ethnographer
Pintupi distinguishes various kinds of anger. But Myers (1988)
can state the meaning of an emotion word found in another
wrote: "On the basis of my own ethnographic observation, I am
language. For example, Hiatt (1978) said that the Gidjingali
not as confident as Morice is about the differentiation in types
word gurakadj includes what is included in the two English
of anger" (p. 608). In fact, Meyers refers to Morice's definitions
words shame and fear. But Hiatt may not have been asserting
as "misleadingly overspecific" (p. 607).
that the boundaries of gurakadj coincide exactly with a bound-
For another example, recall Lutz's (1980) ethnography of the
ary around shame and fear. Hiatt attempted to show that gura-
Ifalukians, which provided a number of the claims reported
kadj is used for fear as well as for shame, but he did not attempt
above. Burrows and Spiro (1953/1957) had written an earlier
to show that gurakadj is used in all and only those cases. Lutz
ethnography of the Ifalukians. Although Burrows and Spiro did
presented the Ifalukian word fago as one without an equivalent
not focus on words or on emotions, they did not entirely neglect
the topic, either. They described the Ifalukians as undergoing
various emotions and unhesitatingly used English words to de-
served a similar phenomenon when bilingual Bugandan subjects re-
scribe the Ifalukians' emotions and to translate Ifalukian emo-
ported crying during anger; Luganda, the Bugandan language, does
not sharply differentiate anger and sadness. Bilinguals may therefore tion words. To be specific, Lutz (1980) emphasized that Ifalu-
be influenced by their first language in assigning meaning to words in kian emotion words are defined not by introspection of internal
their second. If so, this influence would constitute another bias in the states, but by reference to external circumstances. She also said
translation/back-translation procedure. that the English word sadness maps onto two distinct Ifalukian
words, song (justified anger) and fago (love, compassion, sad- culture to test the prediction that the new subjects would assign
ness). Contrast these claims with a sentence from Burrows and each photograph to the same emotion category (i.e, its transla-
Spiro, (1953/1957): "The Ifalukian word for'happiness' iseratr tion equivalent) as had the English speakers.
tipei, which literally means 'my belly is good'; and the word for The first matter to note is that the forced-choice method used
'sadness' is e ngau tipei, which literally means 'my belly is bad' " in these studies means that they cannot show precise equiva-
(P. 317). lence of the emotion concepts in the different cultures. This is so
To summarize, I am not saying that some other method is even if 100% of the choices agreed with prediction. To illustrate,
superior to or should replace ethnography; I believe that no imagine you are a subject in one of the studies. You are shown a
single method can ever be relied on completely. The ethnogra- photograph of a young woman with a bright smile. \bu are
phers whose work was reviewed here often had other purposes asked to describe how she feels by selecting one word from the
in mind. I am proposing that for present purposes ethnography following list: sad, angry, disgusted, afraid, surprised, happy.
be complemented by as many other methods as researchers can Most likely, you'd select happy. But now suppose that happy had
devise. The claims from the ethnographic method can be ac- been replaced on the list with elated. Given the alternatives,
cepted only tentatively, until verified by other methods. youa have little choice but to select elated. If happy were succes-
sively replaced with serene, satisfied, excited, grateful, and
triumphant, you'd again probably select each of these words in
Categorization of Emotion Expressed in the Face
turn. Indeed, substitute for happy any clearly positive word (or
The idea of differences in how cultures categorize the emo- perhaps any subcategory of happy) from contented to ecstatic,
tions could be viewed with skepticism by psychologists who and the conclusion remains the same. If so, the judgment task
have read that recognition of facial expressions of emotion is used in these studies is insensitive to the precise meaning of the
pancultural (Izard, 1980). It is therefore important to under- terms involved. Cross-cultural evidence gathered with this
stand the implications of research on facial expression for the judgment task could show at best that people from different
topic of this article. cultures give similar interpretations to facial expressions.
In contrast to an ethnographic study, a cross-cultural study Problems with forced-choice method are not limited to sub-
attempts to compare directly different cultures, typically by categories of happiness, or to shades of meaning within a cate-
including more than one society within the same study. In a gory. The results obtained also depend on the exact composi-
series of cross-cultural studies, people from different societies tion of the prespecified list. For example, if the word most
have been asked to assign smiles, frowns, scowls, sneers, and applicable to the facial expression is omitted, a consensus can
other facial expressions to emotion categories. The purpose of nevertheless appear on the next most applicable word. In one
these studies was to investigate the nature and origin of facial study, the same facial expression was rated as angry by one
expressions and their association with emotion. I do not ad- group, sad by another group, and upset by a third, depending
dress those issues here. Indeed, for this article, I assume that on which alternatives were available on the list (Russell, 1989).
under the conditions of the experiment the same emotional Because cross-cultural studies almost always begin with a list of
state is manifested in the same facial changes in all human emotion categories specified by English words and then obtain
beings. With this assumption, I can then ask how peoples of translations for them, we have no guarantee that the list of
different cultures categorize the emotional states so mani- choices presented to subjects in another culture contains the
fested. word that those subjects would find most applicable.
The results of these studies are well-known. Much the same This methodological concern is reinforced by the findings
meaning is found in facial expressions by people of different from two studies (Boucher & Carlson, 1980; Izard, 1971) in
cultures and languages. Seven, plus or minus two, different fa- which subjects were allowed to choose their own label for each
cial expressions can be matched to different categories of emo- facial expression. Subjects generated a much greater number of
tion. Initially, this evidence would appear to say that at least that labels than allowed in the forced-choice method. Furthermore,
same number of categories for the classification of emotion is even with a generous criterion for agreement whereby any label
universal. I believe that a closer scrutiny of these studies sup- from a set was counted as agreement, agreement was less than
ports a slightly different conclusion: There is great similarity, that obtained with the forced-choice method. Free labeling is
but not always identity, across cultures in the way in which thus the more interesting task, for present purposes. Unfortu-
emotions communicable by facial expression are categorized. nately, the few data gathered with this method were analyzed by
There is also, surprisingly, evidence for differences as well. scoring subjects' responses as correct or incorrect against a
standard defined by English language categories. The free classi-
fication of facial expressions has yet to be exploited for what it
Literate Societies
reveals about how peoples of different cultures categorize emo-
In most of the studies of literate societies, a similar procedure tion.
was followed. Still photographs of facial expressions thought to I now return to the studies using the forced-choice method.
represent the basic emotions were obtained. The association Even though these studies may exaggerate the amount of agree-
between expression and emotion was first established with ment, they did not, in fact, yield 100% agreement with predic-
norms from English speakers, who were asked to select for each tion. It is also known that cultures vary in a systematic and
expression one emotion label from a short, prespecified list. reliable way in the amount of agreement with prediction (Ma-
The photographs were then shown to subjects from another tsumoto, 1989). Thus, the amount of disagreement may yield
clues as to whether the emotion categories involved might vary happiness, surprise, and sadness (Japanese range = .77 to .98)
with language. To get a rough idea of the influence of linguistic but not on anger, disgust, and fear (Japanese range=.31 to .70).
similarity on the results, I examined the results separately for Moreover, an analysis of variance on quantitative ratings
Indo-European and non-Indo-European languages. Table 2 showed a significant Emotion X Language/Culture interaction.
summarizes the results from the available studies (i.e., those Of course, the reliability of concept-specific differences must
that followed the procedure described above and that included be tested further, and there are various other possible explana-
the relevant comparison groups).5 Because different studies tions for the lower agreement that must be examined. In the
used slightly different methods, stimuli, and response formats, studies listed in Table 2, all the faces judged were Caucasian,
the second column gives the results from the normative En- and Vinacke (1949) and Vinacke and Fbng (1955) reported evi-
glish-speaking group or groups as a standard of comparison. dence that Caucasians and Orientals do slightly better when
In the third column are results for groups speaking another judging facial expressions of their own race (see also Wolfgang
Indo-European language. The proportion of these subjects who & Cohen, 1988; but see Boucher & Carlson, 1980; E Ekman &
agreed with prediction was higher in two cases and lower in Heider, 1988; and Matsumoto & Ekman, 1989, for contrary
eight than that obtained with the normative English-speaking results). Another possibility concerns familiarity with Western
sample. The median difference was 2.2 percentage points. So culture. Among Amharic-speaking students, those living in
whether subjects speak English or another Indo-European lan- closer contact with Westerners showed higher degrees of agree-
guage influences the results minimally. ment with prediction (Ducci, Arcuri, W/Georgis, & Sineshaw,
In the fourth column are results for groups speaking a non- 1982). Neither of these alternatives would explain why the prob-
Indo-European language. These subjects produced a lower pro- lems arose with certain emotion categories and not others. The
portion of agreement than the English comparison group nine available evidence thus provides no definitive answer but en-
out of nine times (p < .05 by a two-tailed sign test). The median courages further research on the idea that some emotion catego-
difference was 10.4 percentage points with a range of 3.8 to ries in non-Indo-European languages differ enough from their
33.7. So. although those speaking a non-Indo-European lan- assumed translation equivalent in English to influence the cate-
guage agreed with prediction to a high degree, there was a cul- gorization of facial expressions.
tural/language difference not accounted for.
Moreover, the lower proportion of agreement seen with non- Preliterate Societies
Indo-European languages may be an average between nearly
equal agreement on some categories of emotion and lower I now turn to studies of proliferate societies. P. Ekman, Soren-
agreement on others. Consider Japanese, the only non-Indo- son, and Friesen (1969) used the method described in the pre-
European language included in more than one study. In Izard's vious section to obtain data on two preliterate groups with little
(1971) study, the Japanese sample gave results similar to the contact with Western culture, the Fore of New Guinea and the
American normative sample for all emotion categories save dis- Sadong of Borneo. Once again, communication by facial ex-
gust, anger, and shame. Izard (1971, p. 258) also mentioned pression across cultural boundaries far exceeded chance. But
another unpublished study of seven facial expressions, four of the results were much weaker than those of studies of literate
which were categorized as contempt and three as disgust by cultures. For example, the number of facial expressions for
American students. Japanese students disagreed, reversing the which 70% or more of the observers agreed with one another
classification for three of the seven. In the P. Ekman (1972) was 23 out of 24 for English speakers, 6 out of 24 for the Fore,
study, the only two failures to predict the modal response oc- and 6 out of 23 for the Sadong.
curred when Japanese labeled as (what was translated) disgust A later expedition to two preliterate societies, the Fore again
what the normative group said was another emotion. In the P. and the Dani of Papua New Guinea, obtained results appar-
Ekman et al. (1987) study, the Japanese sample yielded similar ently more consistent with prediction, but with a change in
results for all emotion categories save fear, anger, and disgust. method: The subject picked one of three (or for children, one of
In short, one or more of the Japanese words translated as fear, two) faces for the protagonist of a story told by the experimenter
anger, disgust, shame, or contempt may differ enough from (P. Ekman, 1972; P. Ekman & Friesen, 1971). This method was
their English counterparts to influence the results of recogni- designed for the original authors' purposes but is ambiguous for
tion studies. One possibility consistent with these results would present purposes. Some stories omitted the emotion word alto-
be that the Japanese categories are broader, admitting of a gether. The surprise story was, "He is just now looking at some-
greater range of referents, than their English counterparts. thing new and unexpected" (Ekman & Friesen, 1971, p. 126).
Serendipitously, a test of this hypothesis is available. Matsu- Results with this story may show subjects' knowledge of facial
moto and Ekman (1989) asked Japanese and American college responses to novel events but cannot show how subjects classify
students to judge the emotion expressed in 48 photographs of
facial expressions. Half of the photographs were of Japanese, !
Other studies, not included in Table 2, also show the similarity
half of Americans; all had been preselected by P. Ekman and
across cultures; see CUceloglu (1970), Dickey and Knower (1941),
Friesen's (1976) criterion as prototypical expressions of basic
Ducci, Arcuri, W/Georgis, and Sineshaw (1982), P. Ekman and Heider
emotions. The proportion correct obtained for each emotion is (1988), Graham, Ricci Bitti, and Argyle (1975), Mandal, Saha, and
shown in Table 3 (along with the results from the P. Ekman et Palchoudhury (1986), McAndrew (1986), Shimoda, Argyle, and Ricci
al, 1987, study). The new results replicated P. Ekman et al. Bitti (1978), Triandis and Lambert (1958), Winkelmayer, Exline,
(1987): Japanese and American performance was similar on Gottheil, and Paredes (1978), and Wolfgang and Cohen (1988).
Table 2
Percentage of Judgments Agreeing With Prediction
Other
English* Indo-European Non-Indo-European
Study Language % Language % Language %
Izard(1971) American, 80.7 Swedish 83.4 Japanese 65.4

British French 82.2
German 80.6
Swissb 79.6
Greek 75.1
P.EkmanfWZf American 85.8 Spanish 83.4 Japanese 78.4
Portuguese 83.3
Boucher & Carlson (1980) American 76.6 Malaysian 64.4
Kilbride & Yarczower (1980) American 94.1 Chi-Nyanja 60.4
P. Ekmanetal. (1987) Scottish, 87.6 Italian 85.5 Estonian 83.2
American Greek 82.0 Cantonese 82.7
German 80.2 Turkish 80.3
Japanese 77.2
Minangabau 74.7
• For English, when more than one English-speaking group was included in the sample, an average across
groups listed is given. " Language was not specified but presumably was French, German, Italian, or
Rhae-Romance. c Means were first calculated for each emotion category. Then a mean of those means
was calculated. For the Spanish language entry, a mean was taken for results from Chile and Argentina.
emotions. Other stories gave both an event and a name for the Another interesting observation was that both Fore and Dani
emotion. The happy story was, "His friends have come, and he languages lack an exact equivalent of the English word surprise
is happy" (Ekman & Friesen, 1971, p. 126). Results with this but distinguish pleasant surprise from unpleasant surprise (Ek-
story cannot show whether subjects know the facial response man, personal communication, 1980). Moreover, both groups
used in greeting friends or the facial expression of happiness. "confused" surprise with fear when asked to pick a face for the
Sorenson (1976) described additional data from the expedi- surprise story. Boucher and Carlson (1980) later reported a simi-
tion to the Fore. One hundred Fore were asked to label the lar finding for the Malay: They lack an exact equivalent for
Ekman pictures in their own language. Many of the responses surprise and confused surprise with fear expressions.
fit the predicted pattern, but others did not. Sorenson wrote,
"The most striking result was that the Fore saw anger more
Conclusion
often than did Westernized subjects" (p. 142). In fact, among
the Fore with least contact with Westerners, anger (or more Smiles, frowns, and other facial expressions are given similar
precisely, the Fore word translated as anger) was the modal re- meaning in all cultures studied (as are vocal expressions, Ma-
sponse for expressions of (what are labeled in English) anger, tsumoto & Kishimoto, 1983; van Bezooijen, Otto, & Heenan,
contempt, and sadness. Thus, the Fore word assumed to be 1983). There is a core of emotional communication that has to
equivalent to anger might have expressed a broader concept. do with being human rather than with being a member of a
(Recall the claims in the ethnographic record of concepts sub- particular culture. Beyond saying this, the available evidence
suming anger and sadness) admits a range of interpretations. Writers in psychology tend to
Table 3
Percentage of Judgments Agreeing With Prediction
Matsumoto and Ekman

Concept P. Ekmanetal. (1987) (1989)
English Japanese English* Japanese English Japanese
Happiness Shiawase 96 90 98 98
Surprise Odoroki 90 94 94 88
Sadness Kanashimi 89 87 95 77
Anger Ikari 82 67 87 70
Disgust lyake 82 60 78 68
Fear Osore 85 65 71 31
• Average between Scotland and United States.

emphasize the great similarity seen; I have emphasized what Variously named, the three are (a) evaluation, valence, pleasure,
room remains for the influence of culture or language on catego- or positivity, (b) activity, arousal, or activation, and (c) potency,
rization. power, control, or dominance. I use names for the three dimen-
Labeling a facial expression is of course not the same as con- sions that capture their affective interpretation: pleasure,
ceptualizing emotion. Less than a dozen emotions are asso- arousal, and dominance, respectively. This evidence is impor-
ciated with recognizable facial expressions. Perhaps these are tant for two reasons. First, it suggests (not proves) the hypothe-
the emotions most likely to be classified similarly across cul- sis that the three dimensions are somehow part of the meaning
tures. Even so, and even with an overly generous forced-choice of emotion words. On one hypothesis, the features denning
method, less than perfect agreement in labeling facial expres- excitement would be pleasure, high arousal, and dominance;
sions has been found in cultures speaking languages not of the features defining calm would be pleasure, low arousal, and
Indo-European origin. Results with preliterate societies, which dominance; the features defining nervousness would be dis-
speak languages perhaps least similar to English, showed even pleasure, high arousal, submissiveness; and the features defin-
more latitude for the influence of language or culture on catego- ing anger would be displeasure, arousal, and dominance (Rus-
rization. The Fore, Dani, and Malay languages also lack a word sell & Mehrabian, 1974,1977). (On this hypothesis, the Japa-
for surprise, which has been claimed to be one of the basic nese word amae would be defined as pleasure, low arousal, and
emotions communicated by facial expression. submissiveness.)
It is time to go beyond the either-or question of whether The second reason is that these same three dimensions ap-
communication of emotion by facial expression is universal pear to be pancultural. Osgood, May, and Miron (1975) showed
versus culture specific. I see no need for further studies that seek that the same three dimensions emerge whatever language is
to establish above-chance communication. It is time to seek the studied. Osgood (1969) interpreted this evidence to mean that
precise nature of the universal core of communication and the human beings universally respond to the objects and events in
precise role played by culture. Future research could then focus their world in terms of three basic bipolar affective dimensions.
on specific hypotheses. Studies would be useful that do not For example, pleasure and dominance have appeared as pan-
presuppose the universality of such concepts as surprise, anger, cultural dimensions in the conceptualization of interpersonal
and the rest. Also useful would be studies that select societies behavior (White, 1980), a domain closely linked to emotion
on something other than an ad hoc basis. Some hypotheses (Kiesler, Horner, Larus, & Chapman, 1987). Pleasure and
could be inspired by ethnographers' reports of culture-specific arousal have appeared as pancultural dimensions of aesthetic
emotion categories. How, for example, would the Gidjingali, judgments, another related domain (Berlyne, 1975; Berryne,
who lexically do not distinguish fear and shame, categorize Robbins, & Thompson, 1974).
those facial expressions identified by English speakers as proto- One worry, of course, is that these three dimensions are some-
typical of fear and of shame? how dependent on method: words rated with the semantic dif-
Important research could be inspired by the general litera- ferential technique. Fortunately, it is now known that the same
ture on cross-cultural psychology For example, collectivist cul- dimensions emerge with other methods, the most important of
tures have been found to differ from individualist cultures in which is multidimensional scaling. Multidimensional scaling
various basic psychological processes (Kagitcibasi & Berry, relies on simple judgments of similarity provided by subjects
1989), including the accuracy with which they identify particu- and thus imposes no concepts on them whatsoever. Multidi-
lar emotions (Matsumoto, 1989). Jeffery and Patterson (1987) mensional scaling of English language emotion terms generally
showed that accuracy in identifying emotions from facial ex- yields the pleasure, arousal, and dominance dimensions (Bush,
pression is related to field dependency, which, in turn, is known 1973; Daly, Lancee, & Polivy, 1983; Neufeld, 1975,1976; Rus-
to vary with culture (Kagitcibasi & Berry, 1989). Hofstede sell, 1978,1980; Storm & Storm, 1987), although not always all
(1980) has identified other broad dimensions of cultural varia- three in the same study. Nor is this finding somehow limited to
tion, and Matsumoto (1989) has begun the study of how these emotion terms: Multidimensional scaling of still photographs
dimensions relate to the categorization of emotion. of facial expressions of emotions yields the same three dimen-
sions, although again not always all three in the same study
(Abelson & Sermat, 1962; Fillenbaum & Rapoport, 1971; Rus-
Dimensions of Judgment
sell & Bullock, 1985,1986). In other words, asked to judge the
A separate research enterprise has sought the basic dimen- similarities and differences among emotions, whether ex-
sions of emotion. The evidence gathered has often been inter- pressed in words or faces, subjects made their judgments in
preted as concerning the structure of subjective experience. terms of degree of pleasure, arousal, and dominance expressed.
Here I examine that evidence from another point of view, Moreover, at least the pleasure and arousal dimensions have
namely what it says about concepts of emotion. The dimensions been demonstrated to be the same dimensions as those ob-
of emotion obtained are taken here as concepts implicit in hu- tained in semantic differential studies (Russell, 1978).
man understanding of emotion. All the studies cited in the preceding paragraph were con-
The semantic differential technique has played a key role in ducted with English-speaking subjects. Studies in other lan-
this search. When fear, anger, and the like are rated on seman- guages have produced the same three dimensions, although
tic differential scales (which are pairs of opposites such as good- again not always all three in the same study When Block's
bad), three bipolar dimensions appear to underlie the ratings (1957) study was repeated in Norwegian, the same two dimen-
(e.g., Averill, 1975; Block, 1957; Russell & Mehrabian, 1975). sions, pleasure and arousal, were found. Herrmann and Ray-
beck (1981) studied samples of subjects in Spain, Vietnam, Two further studies found pleasure and dominance. Gehm
Hong Kong, Haiti, and Greece as well as in the United States. and Scherer (1988) multidimensionally scaled 235 German
Multidimensional scaling of similarity judgments for IS emo- words related to feelings. Lutz (1982) multidimensionally
tion terms yielded the same two dimensions, although only the scaled 31 words commonly used on Ifaluk to describe emo-
first, pleasure-displeasure, was easy to interpret. When Rus- tions. Lutz (1982) specifically noted the absence of anything
sell's (1980) study was replicated in Chinese, Croation, Esto- like the dimension of arousal seen in studies of English emotion
nian, Greek, Oujarati, Japanese, and Polish, the same two di- words.
mensions, pleasure and arousal, were found (Russell, 1983; Finally, Corraliza (1987) reported a factor analysis of Spanish
Russell, Lewicka, & Niit, 1989). Because the words scaled were emotion-related terms and found all three factors: Pleasure,
not a representative sample of emotion-related words in each Arousal, and Dominance.
language, the salience of these two dimensions remains to be That some studies found arousal whereas other studies found
established. Nevertheless, this evidence does show that at least dominance seems to suggest that languages differ in the dimen-
these two dimensions exist and are the means by which similar- sions of emotion beyond pleasure. Such also is the suggestion of
ity is judged within some set of emotion-related words in every Getter's (1975) and Lutz's (1980) ethnographic work in the
language studied to date. South Pacific. More research would be required before decid-
Nor are emotion-related words needed to obtain this result. ing that arousal or dominance is lacking from any language,
Two multidimensional-scaling studies of the feelings conveyed however. No single study has found a difference associated with
by facial expressions both yielded the same pleasure and different languages. Those studies that suggest differences in-
arousal dimensions in different languages: Osaka (1986) stud- cluded but one language and differed from the others in the
ied Japanese- and English-speaking subjects; Russell et al. sample of words scaled, data-analytic procedure, and theoreti-
(1989) studied Greek-, Chinese-, and English-speaking subjects. cal stance of the investigator. In all studies in which these fac-
The panhuman nature of dimensions of facial expression can tors were held constant and only language varied (Block, 1957;
also be seen in replications of Schlosberg's (1952) work in Greek Herrmann & Raybeck, 1981; Osaka, 1986; Russell, 1983; Rus-
by Triandis and Lambert (1958) and in Dutch by Frijda (1953). sell et al, 1989), the same dimensions emerged. More telling,
Other multivariate studies of emotion words have included Abele-Brehm and Brehm (1986) and Gehm and Scherer (1988)
but one language, but a consistent pattern is seen. All the stud- both conducted their studies in German, but pleasure and
ies have corroborated the pancultural nature of the pleasure arousal emerged in the former, and pleasure and dominance
dimension. In some studies, pleasure was the only dimension emerged in the latter. A similar point is made by studies con-
found. Fillenbaum and Rapoport (1971) multidimensionally ducted in English.
scaled 15 Hebrew words denoting emotions, but only a single Why arousal emerged in some studies and dominance
dimension was interpretable: pleasure. A multidimensional emerged in others is unknown. My guess would be that the
scaling of 35 Japanese words yielded pleasure as a first dimen- sample of words scaled is the major factor. Perhaps domi-
sion, but no further dimension analogous to anything found in nance-submissiveness emerges as the second dimension when
studies of English (Yoshida, Kinase, Kurokawa, & Yashiro, the sample of words emphasizes interpersonal contexts. And
1970). Cheng (1977) interviewed Chinese residents of Hong perhaps arousal-sleepiness emerges as the second dimension
Kong, one group in English and another in Chinese. Subjects when the sample of words emphasizes noninterpersonal con-
were asked to describe various emotional experiences, to label texts. Recall that dominance was the second dimension in
the emotion, and to rate it on various verbal scales. These ver- White's (1980) work, which was specifically focused on inter-
bal scales were specifically aimed at seven dimensions derived personal behavior. Dominance (strength vs. weakness) was the
from studies carried out in English. The results were as follows: second dimension in Lutz's (1982) multidimensional scaling of
Ifaluk emotion words; recall that the domain of emotion in
The pleasant-unpleasant dimension. . . was the only dimension Ifaluk emphasizes interpersonal emotions and de-emphasizes
of the entire seven dimensions that the author believed as defi-
intrapersonal states. Dominance was the second dimension in
nitely present in the subjects' construction of emotion. None of
the subjects of the entire sample has difficulty in understanding it Gehm and Scherer's (1988) study. The set of terms they scaled
and in indicating that their emotions were pleasant or unpleasant, included such interpersonal conditions as (I give English trans-
(p. 192) lations) affectionate, compassionate, disdainful, respectful, and
vengeful. In contrast, arousal may emerge as the second dimen-
Two further studies found pleasure and arousal. G. Ekman sion in noninterpersonal contexts. Recall that arousal (activity)
(1955) used multidimensional scaling with 23 Swedish emo- was the second pancultural dimension in Berryne's studies of
tion-denoting words. These data have been analyzed in various aesthetic judgments. Arousal was the second dimension in stud-
ways, often with slightly different results (G. Ekman, 1955; Fil- ies of emotional facial expressions (Abelson & Sermat, 1962;
lenbaum & Rapoport, 1971; Lundberg & Devine, 1975; Micko, Russell & Bullock, 1985,1986; Russell et al, 1989). Arousal was
1970; Shepard, 1962; Stone, 1971; Stone & Coles, 1970). Never- the second dimension in Russell's (1980, 1983) multidimen-
theless, one bipolar dimension closely related to pleasure sional scaling of emotion-related words, which included intra-
emerged in most of the analyses, and a clear arousal dimension personal terms: tired, aroused, excited, tense. And arousal was
emerged in at least one (Micko, 1970). Abele-Brehm & Brehm the second dimension in Abele-Brehm and Brehm's (1986) fac-
(1986) reported a factor analysis of German emotion-related tor analysis, which included such items as (I give English trans-
terms that found pleasure and arousal. lations) active, full of energy, relaxed, and quiet.
Of course, it takes more than two or three dimensions to in the language of emotion. Boucher's (1979) hypothesis recon-
define words for the various emotions. When the pleasure di- ciles language facts with the theory of basic emotions.
mension is held constant, further dimensions can be obtained Boucher (1979) defines the domain of emotion as a vertical
through multidimensional scaling of emotion-related words class-inclusion hierarchy, as can be done for the domain of ob-
(Russell, 1978). Using different methods, investigators have of- jects. Fruit, orange, and navel orange illustrate such an arrange-
fered evidence for a score of dimensions beyond pleasure-dis- ment, in which each succeeding category is included within,
pleasure. These dimensions seem to be aspects of the cognitive and hence less abstract than, the former category. English words
appraisal of the circumstances leading up to the emotion (I. J. for emotions appear to fit this pattern: emotion, love, romantic
Roseman, 1984; Scherer, 1984; Smith & Ellsworth, 1985) or love. In Boucher's version, emotion is the superordinate, most
aspects of the behaviors and motives pursuant to the emotion inclusive, category. At the next lower (basic) level, emotion is
(Frijda, 1987). Although Frijda (1987) carried out his work in divided into the 7± 2 basic emotions, that is, anger, fear, and
Dutch, Scherer (1984) in German, and Smith and Ellsworth the like. Each of these, in turn, is subdivided, forming a subordi-
(198S) and I. J. Roseman (1984) in English, no cross-language nate, least inclusive level. Most emotion words therefore label
comparisons have been reported. subordinate categories of emotion: as when annoyance, rage,
In short, studies on what is similar in how emotions are un- and fury are subtypes of anger. Boucher's hypothesis is that the
derstood across cultures point principally to bipolar dimen- domain of emotion takes this hierarchical form in all lan-
sions: pleasure-displeasure surely, arousal-sleepiness and dom- guages, that the top two levels are universal, and that the lowest
inance-submissiveness probably. No one has yet shown any in- level is culture specific. More recently, other writers have pro-
fluence of language or culture on the nature of the dimensions posed similar ideas (Agnoli, Kirson, Wu, & Shaver, 1989; John-
found in these studies. Researchers are just beginning to go son-Laird & Oatley, 1986, 1989; Shaver, Schwartz, Kirson, &
beyond these few dimensions. Including different languages O'Connor, 1987). Johnson-Laird and Oatley (1986,1989) pro-
and cultures in their research should provide much important posed five universal semantic primitives—happiness, fear,
information, including evidence on some of the differences hy- anger, disgust, and sadness—in terms of which all other emo-
pothesized in the ethnographic record. tion words can be defined.
Boucher's (1979) hypothesis is reinforced by research on cate-
gorization in other domains. Rosch (1977) observed that one
Hypotheses on the Nature of Emotion Categories level in category hierarchies is basic, in the sense that various
signs point to its being psychologically salient. Although the
Words that people use to categorize the emotions do not ap-
same object could be called an object, a fruit, an orange, or a
pear to be equivalent in all languages, although there is good
navel orange, most people feel that one name, orange, seems to
reason to believe they are often similar. Dimensions that people
say what that object really is. The basic-level names are learned
use to conceptualize the emotions may be universal, but only a
first by children and emerge first in historical development of a
few have been studied. How can there be both similarities and
language. Berlin's (1978) work on folk taxonomies for plants and
differences in words, and how can categories and dimensions
animals shows further that the basic level, which Berlin termed
be integrated? The available evidence is sparse, and what there
the generic level, is most likely to be similar in different cul-
is does not force us to a single conclusion. Any number of hy-
tures. Categories subordinate to the basic level are formed for
potheses could be proposed to account for this evidence. I first
cultural utility (Brown, 1986, p. 479) and hence show cultural
present the available hypotheses briefly but then focus my at-
variability.
tention on one additional possibility, that of a script, not ad-
Boucher's (1979) hypothesis also accounts nicely for some of
vanced in this context before.
the observations found in the ethnographic record. It easily
accounts for the Gidjingali having one word for fear, but no
Boucher subtypes, or for Malay distinguishing among types of anger in a
way not done in English.
Boucher (1979) is among those writers in the field of emotion On the other hand, some claims in the ethnographic record
(Buck, 1988; P. Ekman, 1972; Izard, 1977; Leventhal, 1974; are not accounted for. Those languages lacking a specific super-
Plutchik, 1980; Tomkins, 1984) drawn to the idea that there are ordinate equivalent to emotion is one problem. Another is that
a few discrete, universal—basic—emotions. Indeed writers anger, fear, and other basic-level categories may not be univer-
throughout history have compiled lists of the basic emotions sal; see Table 4. Boucher's hypothesis might be reconciled with
(Gardiner, Metcalf, & Beebe-Center, 1937). Today, 7± 2 basic this evidence, but it is puzzling why a language would fail to
emotions are most often listed: anger, fear, happiness, sadness, provide a single word for an important, salient, discrete, and
and disgust, with surprise, contempt, interest, or shame some- possibly innate category of experience—if such exists.
times added. These theoretically represent universal categories Boucher (1979) used cluster analysis to represent the Malay
of subjective emotional experience (Izard, 1977) and universal, lexicon for emotion and found results supportive of his hypoth-
possibly innate, categories used in the recognition of emotional esis, but other cluster analyses are less encouraging. Boucher
signals (Buck, 1988, based on Darwin, 1872/1965). assembled a cross-cultural research team representing eight so-
At first glance, this idea does not square with our English cieties, two of which speak English. Cluster analysis of similar-
lexicon with more than 7 ± 2 words for categories of emotion. ity judgments within each language did not yield a universal
Moreover, the ethnographic record indicates cultural variation category structure. For example, in four societies, a separate
Table 4
Possible Problems for the Universality of Hypothesized Basic-Level Emotion Categories
Basic-level term Problem
Happiness Missing in Chewong (Howell, 1981)

Surprise Missing in Fore, Dani (P. Ekman, personal communication, 1980), Malay
(Boucher &. Carlson, 1980), and Ifaluk (Lutz, 1982)
Anger Overlaps with sadness or grief in Luganda (Davitz, 1969; Leff, 1973; Orley,
1970), Illongot (R. I. Rosaldo, 1984), and Ifaluk (Lutz, 1980)
Fear Missing in Ifaluk (Lutz, 1980), Utku (Briggs, 1970), Pintupi (Morice, 1978);
not distinguished from shame in Gidjingali (Hiatt, 1978)
Sadness Missing in Tahitian (Levy, 1973) and Chewong (Howell, 1981); not sharply
distinguished from anger in Luganda (Davitz, 1969; Leff, 1973; Orley,
1970)
Disgust Missing in Polish (Wierzbicka, 1986), Ifalukian (Lutz, 1980), and Chewong
(Howell, 1981); not distinguished from hate in Samoan (Gerber, 1975)
Note- See text for fuller explanation. For example, fear is said here to be missing in Utku (Briggs, 1970);
fear per se is missing, but Utku do have separate words for fear of physical injury and fear of social injury.
basic-level cluster representing depression emerged, but in the derstanding emotion. There is no guarantee that human beings
other four societies, no depression cluster emerged (Brandt & have got the matter right. Those who believe in basic emotions
Boucher, 1986). Other researchers have obtained from 2 to 18 thus have a choice whether to assume a separate and further
basic clusters (Agnoli et al, 1989; Dietze, 1963; Fillenbaum & hypothesis that the English words like anger and fear denote
Rapoport, 1971; Gerber, 1975; Hoekstra, 1986; Lutz, 1980; Niit those basic emotions, and a still further choice whether to as-
& Valsiner, 1977; Shaver et al., 1987; Storm & Storm, 1987; sume that the concepts expressed by these words are pancul-
Stringer, 1967). The clusters do not necessarily correspond to tural.
what Boucher would list as basic emotions. For example, dis-
gust and anger were in the same cluster, and irritation was in a
Levy
different cluster, in Fillenbaum and Rapoport's (1971) reanaly-
sis of G. Ekman's (1955) data on Swedish words for emotions. Levy's (1984) account resembles and complements Boucher's
On the other hand, hierarchical cluster analysis does not pro- (1979). He too presupposed some number of universal basic
vide a good test of Boucher's hypothesis. Clusters from a cluster emotions, although approaching them from a psychodynamic
analysis need not correspond to cognitive categories in the perspective. Levy also relied on the idea of a vertical hierarchy
minds of the subjects. In none of these studies has the re- and the idea that languages differ in which and what number of
searcher ruled out the null hypothesis that two data sets are subtypes are recognized. Already mentioned was Levy's (1984)
samples from the same population. view that a culture develops a cognitive model for each emotion.
Although Boucher (1979) does not take this tack, writers in Where the model is elaborate, many subtypes of the emotion
this tradition often use another way of accounting for words are recognized. Where the model is poorly developed, few or no
other than the core 7± 2: Some terms might label blends of the subtypes may be recognized. The culture need not even recog-
basic emotions. For example, Plutchik (1980) suggested that nize a basic emotion as an emotion, as when the Tahitians have
pride is the blend of joy and anger. The notion of blend might no concept of sadness.
be inconsistent with Boucher's hierarchy, because in a true hier- To account for cultural differences, Levy (1984) proposed an
archy, categories at the same level are mutually exclusive: Any- account that is based on the notion of a focal point. Reminis-
thing that is joy is not anger. cent of work such as Berlin and Kay's (1969) on a universal
So, there remain several problems with Boucher's (1979) ac- pattern within cultural differences in color vocabulary, Levy
count. It is therefore more a promise than a worked out theory. (1973, pp. 16-17) proposed that there may be cross-cultural
One would have to see a specific analysis of different lexicons unrversals in the focal points, or best examples, of particular
before knowing how well this sort of theory can be made to fit emotions. Where languages differ is in the category bound-
the evidence. There is also no consensus on which terms in aries. The Ilongot word liget illustrates Levy's (1984) account.
English label basic emotions, which blends, and which sub- Even though liget seems to include a broader range of states
types. The lack of consensus suggests that the concepts of basic than does anger, they might have the same focal point, perhaps
category, blend, and subtype are not sufficiently specified in the a prototypical furious reaction. Where one language subdivides
domain of emotion to allow a resolution of this issue. In addi- a category can be seen easily from Levy's (1984) spatial meta-
tion, although not contradicted by it, Boucher's hypothesis fails phor as adding a boundary. Levy's account would have an easier
to account for the evidence of universal bipolar dimensions. time than Boucher's with evidence summarized in Table 4.
Finally, in evaluating Boucher's (1979) hypothesis, one must Levy's (1984) account is phrased metaphorically in terms of
recognize that the existence of basic emotions does not entail points and boundaries. In the same metaphor, we would have to
nor is entailed by the existence of universal categories for un- ask about nature of the space in which the points are placed and
the boundaries drawn. Or, put another way, along what dimen- and [c] when one thinks that one cannot cause it not to happen.
sions do we define best, intermediate, and borderline examples (P. 592)
of an emotion category? One possibility would be to combine
Her style of definition therefore provides a universal (etic) frame-
Levy's account with Osgood's (1966) idea of a panculturaJ three-
work in which (emic) words from different languages can be
dimensional space denned by the factors of the semantic differ-
compared.
ential. According to Osgood (1966) different discrete emotions
Some psychologists have misunderstood Wierzbicka (1986)
correspond to different regions of the space.
as offering a classical definition in which Conditions a through
c are each necessary (see Russell, 1991). Let me therefore em-
Leff phasize the crucial word as in the quotation above. If Condi-
tions a through c were necessary conditions, there would be
Leff(1973,1977,1981) provided an account that emphasizes obvious counterexamples in which one of the alleged necessary
historical changes in the meaning of emotion words. At an conditions is missing. Wierzbicka's conditions are not meant to
earlier historical stage, one word denoted the state of unpleas- specify the necessary circumstances of fear, but the prototypi-
ant somatic arousal. Slowly the meaning of that word came to cal circumstances.
include, and then to focus on, the psychological experience Acceptance of Wierzbicka's (1986) definition requires accep-
accompanying the somatic state. Such experience was initially tance of her theory as to what constitute universal semantic
undifferentiated, and hence the single word was broad in primitives. In addition, the psychologist is inclined to ask about
meaning. Later this root word split into a number of variants to the psychological reality of the proposed definitions. To return
differentiate distinct psychological experiences. As a result, to fear, are Conditions a, b, and c what someone knows who
"we find that emotions we consider as distinct, namely, anger, knows the meaning of fear? The answer would appear to be no
fear (anxiety) and sadness, at one stage in the development of for neonates, who have been said to recognize fear (and there-
English were all represented by words which derived from the fore have the concept of fear), presumably before they can rea-
same hypothetical Indo-Germanic root Angh" (p. 300). As a sonably be said to possess Wierzbicka's universal semantic com-
consequence, languages differ today by including more or fewer ponents constituting fear.
distinctions among emotional states. And indeed, at least those
examples cited above from ethnographer's reports that were
described as just such combinations of or distinctions within The Hypothesis of Scripts
English categories fit Leff's notion of differentiation. As with
A script is a knowledge structure for a type of event whereby
Boucher and Levy, a hierarchy is envisioned, although without
the event is thought of as a sequence of subevents. Although we
the suggestion of a basic level of 7± 2 basic emotions.
often speak of an emotion as a thing, a sequence of subevents is
Leff's is less than a fully detailed account. For example, con-
a more apt description. For example, Table 5 gives a script for
sider the original, undifferentiated emotion category Leff sug-
anger.
gested that the root word referred to any state of unpleasant
The sequence given might rarely occur in just that way, but
somatic arousal. But it is unclear on this account what to say
for each emotion concept, we know some such prototype se-
about pleasant (joy) or even neutral (surprise) emotions. Further-
quence. Fear, we know, is typically caused by a danger and
more, some emotions (sadness) are felt to involve low arousal
typically leads to some sort of running away. For some con-
(Russell & Mehrabian, 1977). Finally, Leff did not specify how
cepts, the story is simple. In happiness, you desire something,
differentiation between emotions takes place.
get it, feel pleasure, smile, and, perhaps feel kind toward others.
Leff's focus on an undifferentiated state of somatic arousal
For other concepts, the story is more complicated. Jealousy
suggests joining Leff's account with Schachter and Singer's
might include anger, but jealousy implies a surrounding situa-
(1962) theory of emotion. On the revised account, the initial
tion, a social relationship involving three people, specific mo-
state is arousal, which can be pleasant or unpleasant. Further
tives, behaviors, and consequences. These implications must be
differentiation between emotions is based on the social situa-
understood to know what the word jealousy means.
tion in which the emotion takes place. Leff's account, or this
In short, according to the script hypothesis, categories of
modified version of it, thus allows considerable room for cul-
emotion are defined by features. The features describe not hid-
ture-specific emotion concepts.
den essences but knowable subevents: the causes, beliefs, feel-
ings, physiological changes, desires, overt actions, and vocal
Wierzbicka and facial expressions. These features are ordered in a causal
sequence, in much the same way that actions are ordered in a
Wierzbicka (1986) argued that an emotion word in one lan- playwright's script. To know the meaning of a term like happi-
guage often has no exact equivalent in another. But each emo- ness, fear, or jealousy is to know a script for that emotion. In
tion word can be defined in terms of universal semantic primi- other words, the script hypothesis is that the meaning of each
tives, hypothesized to be concepts such as want, think, good, and such word, the concept it expresses, is a script. No emotional
bad. To illustrate, here is Wierzbicka's definition of fear: essence is added to the features, just as biologists found no vital
essence needed to spark chemical constituents to life.
X was afraid = X felt as one does
[a] when one thinks that something bad can happen to one Unfortunately, the word script means different things to dif-
\b] when one wants to do something to cause it not to happen ferent writers. For example, I am treating emotion scripts some-
Tables
An Anger Script
Step Subevent
1 The person is offended. The offense is intentional and harmful. The person is innocent. An
injustice has been done.
2 The person glares and scowls at the offender.
3 The person feels internal tension and agitation, as if heat and pressure were rapidly mounting
inside. He feels his heart pounding and his muscles tightening.
4 The person desires retribution.
5 The person loses control and strikes out, harming the offender.
Note. This anger script is partly based on Lakoff's (1987) analysis of anger.
what differently than did Abelson (1981) when he first pro- than to be mutually exclusive. Actual events tend to be catego-
posed them. Abelson thought of emotion scripts as involved rizable into more than one category; the same case can be
both in understanding (a knowledge schema) and in behavior (a anger, fear, disgust, and sadness. Some overlap each other al-
response program); I am restricting my treatment to the former. most completely, others to a high degree, others to a minimal
Abelson thought of the emotion as one of the features within an degree, and some not at all.
emotion script: Someone interferes, you become angry, you hit. The script hypothesis offers a simple and straightforward ac-
In contrast, I am using the notion of script to define the emo- count of cross-cultural similarities and differences. Those lan-
tion concept. guages with fewer emotion categories would have more general
Other authors have proposed thinking of emotion concepts scripts: Each script would have fewer features and cover a
as something like a script (Abelson, 1981; de Sousa, 1980; Fehr broader range of phenomena. Languages with many emotion
& Russell, 1984; Lakoff, 1987; Sabini & Silver, 1982; Shaver et categories have more specific scripts: Each script would have
al, 1987; Tomkins, 1979). The current hypothesis presented more features and cover a narrower range of phenomena. More-
here illustrates this line of thinking; I would not like to be taken over, some features are culture specific, and others are pancul-
as attempting to differentiate within this family of ideas. Just tural. Or, better, culture specific and pancultuml define two
how abstract or concrete are the features that constitute the ends of a continuum. Some features may be limited to few cul-
script remains an empirical question. Some people may under- tures; others found in al] or almost all.
stand emotion in terms of more abstract scripts, others in terms Consider first the similarities: Those bipolar dimensions of
of more concrete exemplars (Kahneman & Miller, 1986). More- affective meaning found universally are choice candidates for
over, different people, even within the same culture, might pos- pancultural features of emotion scripts. These dimensions de-
sess slightly different scripts for the same emotion. There may scribe universal aspects of feeling, or mood change (Russell &
be more agreement for fear and anger than for disdain and Snodgrass, 1987). Other universal or near universal features are
melancholy. also likely to be found, describing universal aspects of anteced-
The present script hypothesis is closely tied to the principle ents (Boucher & Brandt, 1981), motives, desires, or action ten-
that the meaning of any concept is related to the network of dencies (Frijda, 1987), facial or vocal expressions (R Ekman,
concepts within which it is embedded. Meaning wholism, or the 1972), and physiological changes (P Ekman, Levenson, & Frie-
theory-laden quality of meaning, is somewhat of an orthodoxy sen, 1983). Such features would quite likely be noticed and in-
in philosophy (Searle, 1983) and has been emphasized in psy- corporated into the mental representation of the emotion, the
chology by Medin (1989). script. Whether any such feature is actually universal is natu-
The script hypothesis is also tied to Roach's (e.g, 1973,1975, rally an empirical question. Even if universal, a particular fea-
1977) prototype theory of natural language categories of ob- ture may be emphasized, deemphasized, or ignored. For exam-
jects. A script is to an event what a prototype is to an object. ple, even though arousal feelings appear to be a good candidate
Elsewhere my colleagues and I have pointed to aspects of the for a universal dimension, the Ifaluk pay them little attention
emotion domain clarified by Rosch's theory (Bullock & Russell, (Lutz, 1982).
1984,1985; Fehr & Russell, 1984,1991; Fehr, Russell, & Ward, Moreover, these features form natural patterns. That is, not
1982; Russell, 1991; Russell & Bullock, 1986). Rather than prop- all combinations of the features are equally likely to occur. As
erly defined, happiness, sadness, fear, anger, and other natural Rosch (1973,1975,1977) emphasized, features in the world are
language categories of emotion are fuzzy: (a) Borders between correlated: Creatures with beaks and wings are likely to have
categories are vague, rather than clear-cut. Although some ac- feathers. In the emotion world, frustration of a goal is likely to
tual events are clear cases of, for example, anger and other ac- lead to feelings of displeasure and arousal, which often leads to
tual events are clearly not anger, some events straddle the fence some effort to change the situation; this sequence is more likely
and are difficult to decide one way or the other, (b) Membership accompanied by a frown than a smile and by physiological
within a category is a matter of degree rather than all or none. arousal. On the other hand, getting what you want often leads
Actual cases of anger vary in how well they exemplify the con- to pleasure, which often leads to no action or merely maintain-
cept, (c) Different categories tend to overlap one another rather ing the current situation, to smiles, and lowered physiological
arousal. Such patterns might not be absolutely universal, but main appear to vary, as do divisions within the domain. Thus,
the more widespread the patterns the more likely they will re- neither the word emotion nor words for even alleged basic emo-
sult in cross-cultural similarity of human concepts of emotions. tions, such as anger and sadness are universal. Different lines of
Now consider what might be culture specific about emotion evidence converge on this conclusion: Intensive ethnographic
concepts. One possibility seems to concern what has been studies of specific emotion words, the large number of reports
called the cause of the emotion. Causal antecedents appear to by ethnographers of noticeable differences in emotion words,
be involved in emotion terms in English: Fear implies that a the large variation across languages in the number of emotion
danger has appeared, whereas anxiety implies that the cause is words, the experimental evidence of differences in what were
vague or unknown. Guilt implies that you yourself are the previously taken to be translation equivalents, and the small
causal agent of a bad outcome, whereas anger implies that an- difference between Indo-European and non-Indo-European
other has caused some harm. Weiner (1982) has provided evi- languages in categorization of facial expressions.
dence and a conceptual analysis of the role of causal thinking in According to some accounts of emotion, this conclusion im-
English language emotion concepts. Note that this is not to say plies that emotions themselves, the events referred to by the
that a particular cause is a necessary feature, just a prototypical word emotion, are culture specific. On this point, I would like to
feature. avoid misunderstanding. This article did not address the ques-
The same is true in other languages. Litost is caused by a tion of what about emotion itself is panhuman and what is
sudden insight into one's own miserable self. Schadenfreude is culture bound. It addressed the question of what in the categori-
caused by another's displeasure. Ijirashii is caused by seeing zation of the emotions is universal and what is culture bound.
someone praiseworthy overcoming an obstacle. Culture can From a logical point of view, answers to these two separate
emphasize one cause or another. People can react emotionally questions may coincide, but they need not. Peoples of different
to different things in different cultures. Different causes can cultures could impose their own categorization on a universal
thus be incorporated into the meaning of emotion-descriptive emotion reality Conversely, people could impose universal cate-
terms. Anthropologists have described how people of certain gories on a culture-bound reality.
cultures attribute emotions to such things as the soul leaving the Differences in emotion categories do, nevertheless, support
body, a curse from an enemy, bewitchment, demonic posses- one inference regarding the emotions themselves. Some writers
sion, disfavor of the gods, and visitations of ghosts. For exam- assume that emotions have to be classified as we do in English
ple, the Tahitian word mehameha refers to fear caused by a —in terms of anger, fear, anxiety, depression, and so on. If
ghost (Levy, 1973). In short, I propose that causal antecedent English language categories regarding emotion are not univer-
can be an aspect of the meaning of an emotion term and that sal, then we have no guarantee that emotion, anger, fear, and so
this aspect of meaning can be part of what varies with culture. on are labels for universal, biologically fixed categories of na-
Another sort of culture-specific aspect of emotion concerns ture. Rather, they are hypotheses formulated by our linguistic
what has traditionally been called consequences. Litost leads to ancestors.
a desire for revenge. The Pintupi distinguish various forms of I also tentatively conclude that there is great similarity in
distress by their consequences: Watjilpa is worry that leads to emotion categories across different cultures and languages. Dif-
physical illness (Morice, 1978, p. 92). More commonly, the con- ferent lines of evidence converge on this conclusion: the wide-
sequences are behavioral, such as emotionally expressive ges- spread assumption in the ethnographic record that emotion
tures. For example, Levy (1973, p. 96-98) found 26 terms in and emotion categories are everywhere the same (as evidenced,
Tahitian in which the inner feeling differs from what is out- for example, in the HRAF), the acknowledgment by many eth-
wardly displayed. Similarly, the Samoan word bbna refers to nographers who found differences in emotion lexicons that
anger that is not expressed (Gerber, 1975). According to P. Ek- many words mean the same or something similar, the apparent
man (1972, 1980), different cultures establish different norms universality of dimensions in judgments of emotion, and the
about the control of emotional expressions. These display rules high similarity across cultures in perception of facial expres-
might dictate that at a funeral, for example, grief should be sions of emotion.
inhibited, displayed, or exaggerated. Peoples of different cul- As banal as the conclusion of great similarity may sound, I
tures thus expect different behavioral consequences of specific mean it to imply that I found no evidence for claims of a radical
emotions. Again, I propose that these expectations are incorpo- kind. For example, in various publications, Solomon (e.g., 1976)
rated into the meaning of terms and that this aspect of meaning claimed that whole groups of emotions go nameless and do not
can vary with culture. exist within particular cultures. Calhoun and Solomon (1984)
wrote the following about the Utku:
Conclusion It is not just that they do not express anger, they do not feel angry,
either. Indeed, they do not even have a woid for anger in their
Studies of different lexicons, of judgments of emotion from vocabulary (the closest word to it, significantly, means "childish"),
(p. 34)
facial expressions, and of the dimensions implicit in compara-
tive judgments of emotion cannot tell us directly how people 1 believe this statement is based on a misinterpretation of
categorize emotions. Nevertheless, considering these three Briggs's (1970) ethnography of the Utku, in which there are
sources of evidence, I tentatively conclude that people of differ- both incidents of and words for anger.
ent cultures and speaking different languages categorize the Another example would be Lutz's (eg, 1982) claim about the
emotions somewhat differently. The boundaries around the do- referent of emotion words:
Internal feeling states have commonly been assumed to be the derives from its role in a scientific theory (Cronbach & Meehl,
primary referents of emotion words in Western thought, both so- 1955), the meaning of a folk concept of emotion derives from its
cial-scientific and l a y . . . . Examination of the use of emotion
role in the folk theory of emotion. A further step would be to
words among several Oceanic peoples. . . reveals an alternative
view of emotion. In these societies, emotion words are seen as
examine the consequences of categorizing one way rather than
statements about the relationship between a person and an event another and of believing in one such folk theory as opposed to
(particularly those involving another person), rather than as state- another. These consequences remain largely unknown, but ex-
ments about introspection on one's internal states. (Lutz, 1982, citing work has begun to explore this question (C. Geertz, 1966;
P. 113) Heelas & Lock, 1981; Lutz, 1988; M. Z. Rosaldo, 1980,1983).
I take this assertion to mean, for example, that the Ifalukian The most glaring difficulty with the evidence reviewed here
word song, commonly translated as anger, refers not to the is that different conclusions are associated with different meth-
angry person's internal state, but to something external. There ods of gathering evidence. We need further evidence of all
is first the question of whether Lutz's claim is consistent with kinds, but we especially need new methods. Conclusions drawn
her own ethnographic evidence. Lutz (1980) had earlier indi- from current methods need to be subjected to empirical tests
cated that song refers to niferash, which she translated as "our that are based on other methods. Several exciting examples of
insides." Second, there is the conceptual issue of how a word in cross-fertilization therefore bear emphasis. Lutz (1982) in-
any language that does not refer to an internal state could be cluded the psychometric techniques of multidimensional scal-
said to be an emotion word. If song were a member of a class of ing and cluster analysis in her ethnography of the Ifaluk. P.
words that, like marriage or kinship, referred to a relationship, Ekman and HeiderO 988) used psychometric methods in a cul-
then the reason for calling song an emotion word is unclear. ture chosen on ethnographic grounds. Scherer and his asso-
Lutz may be conflating sense and reference. The sense of ciates have begun a large-scale collaborative project comparing
song can involve anger-inducing events and external relation- various cultures on common instruments (Scherer, 1988;
ships, although the reference of song is still an internal state. Scherer, Wallbott, & Summerfield, 1986). Perhaps further re-
Consider the sentence, "My grandmother lives in Los Angeles." search and the development of new methods is best stimulated
The word grandmother here refers to a particular person, my by the contrast of specific hypotheses. For this reason, the ac-
maternal grandmother, Besse. But the sense of grandmother counts developed by Boucher (1979), Leff(1977), Levy (1984),
involves other people, relationships, and events. Thus, the deter- and Wierzbicka (1986) and the script hypothesis may come to
mination that Besse is a grandmother is made not by inspection play an important role in guiding future research. My conclu-
of Besse, but by the existence of other people (my mother and sion of both similarities and differences across cultures in the
me) and certain events (her daughter's having given birth, etc.). categories of emotion is only as good as the available evidence.
The word grandmother does not refer to me or to my mother or The conclusions drawn here are thus hypotheses, which must
to anyone's birth. It refers to Besse. So, the proper thing to say be tested empirically and which may serve to stimulate nonbe-
appears to be that song refers to an internal state created when lievers into marshalling evidence or arguments against them.
certain external circumstances occur.
In this regard, Ifaluk emotion words are more like than un-
like English emotion words. Evidence of the kind reviewed in References
this article highlights the role of situational antecedents in de-
nning emotion concepts. This conclusion was explicit or im- Abele-Brehm, A., & Brehm, W (1986). Zur Konzeptualisierung und
plicit in all five hypotheses regarding the nature of emotion Messung von Befindlichkeit [On the conceptualization and measure-
concepts. Even Boucher (1979), who believes in basic emotions, ment of mood], Diagnostica, 32, 209-228.
was led to a similar analysis for those emotion concepts beyond Abelson, R. P. (1981). Psychological status of the script concept. Ameri-
names for the 7± 2 basic emotions. Research on the psychologi- can Psychologist, 36, 715-729.
cal representation of emotion provides strong complementary Abelson, R. P., & Sermat, V (1962). Multidimensional scaling of facial
expressions. Journal of Experimental Psychology, 63, 546-554.
evidence that emotion concepts in English are differentiated by
Abu-Lughod, L. (1986). Veiled sentiments: Honor and poetry in a
their situational antecedents (Conway & Bekerian, 1987; Dore
Bedouin society. Berkeley: University of California Press.
& Kirouac, 1986; Harrison, 1986). Thus, jealousy refers to an Agnoli, F, Kirson, D., Wu, S, & Shaver, P. R. (1989). Hierarchical analy-
internal state created when certain circumstances occur. If so, sis of the emotion lexicon in English, Italian and Chinese. Paper deliv-
English and Ifalukian follow a remarkably similar way of defin- ered at a meeting of the International Society for Research on Emo-
ing an emotion. tion, Paris.
More generally, emotion concepts are embedded in a system Averill, J. R. (1975). A semantic atlas of emotional concepts. JSAS:
of beliefs about psychological and social processes. This system Catalog of Selected Documents in Psychology, 5, 330. (Ms. No. 421).
has been called a cognitive model, folk theory of mind, ethnopsy- Averill, J. R. (1980). A constructivist view of emotion. In R. Plutchik &
H. Kellerman (Eds.), Theories of emotion (Vol. 1, pp. 305-340). San
chology, or indigenous psychology (Heelas & Lock, 1981). Simi-
Diego, CA: Academic Press.
larities and differences in emotion concepts may be but the tip
Berlin, B. (1978). Ethnobiological classification. In E. Rosch & B. B.
of the iceberg, where the iceberg would be similarities and dif-
Lloyd (Eds), Cognition and categorization (pp. 9-26). New York:
ferences in the folk theory of mind, of self, of society, of nature, Wiley.
and so forth (Holland & Quinn, 1987). Moreover, understand- Berlin, B, & Kay, P. (1969). Basic color terms: Their universality and
ing the folk model may be necessary to understanding con- evolution. Berkeley: University of California Press.
cepts. In the same way that the meaning of a scientific construct Berlyne, D. E. (1975). Extension to Indian subjects of a study of explor-
atory and verbal responses to visual patterns. Journal ofCross-Cul- Conway, M. A, & Bekerian, D. A. (1987). Situational knowledge and
tural Psychology, 6, 316-330. emotions. Cognition and Emotion, 1,145-191.
Berlyne, D. E., Robbins, M. C, & Thompson, R. (1974). A cross-cul- Corraliza, J. A. (1987). La experienciadelambiente[The experience of
tural study of exploratory and verbal responses to visual patterns the environment]. Madrid, Spain: Tecnos.
varying in complexity. In D. E. Berlyne (Ed.), Studies in the new Crapanzano, V (1980). luhami: Portrait of a Moroccan. Chicago: Uni-
experimental aesthetics: Steps toward an objective psychology of aes- versity of Chicago Press.
thetic appreciation (pp. 259-278). Washington, DC: Hemisphere. Cronbach, L. J., & Mcehl, P. E. (195 5). Construct validity in psychologi-
Besnier, N. (1990). Language and affect. Annual Review of Anthropol- cal tests. Psychological Bulletin, 52,281-302.
ogy, 19. H9-451. Ciiceloglu, D. M. (1970). Perception of facial expressions in three cul-
Birdwhistell, R, L. (1970). Kinesics and context. Philadelphia: Univer- tures. Ergonomics, 13, 93-100.
sity of Pennsylvania Press. Daly, E. M., Lancee, J. W, & Polivy, J. (1983). A conical model for the
Block, J. (1957). Studies in the phenomenology of emotions. Journal of taxonomy of emotional experience. Journal of Personality and Social
Abnormal and Social Psychology, 54, 358-363. Psychology, 45. 701-711.
Boucher, J. D. (1979). Culture and emotion. In A. J. Marsella, R. G. Darwin, C. (1965). The expression of the emotions in man and animals.
Tharp, & T. V Ciborowski (EdsJ, Perspectives on cross-cultural psy- Chicago: University of Chicago Press. (Original work published
chology^. 159-178). San Diego, CA: Academic Press. 1872)
Boucher, J. D., & Brandt, M. E. (1981). Judgment of emotion: Ameri- Davitz, J. R. (1969). The language of emotion. San Diego, CA: Aca-
can and Malay antecedents. Journal of Cross-Cultural Psychology, 12, demic Press.
272-283. deSousa, R. (1980). The rationality of emotions. In A. O. Rorty(Ed.),
Boucher, J. D, & Carlson, G. E. (1980). Recognition of facial expression Explaining emotion (pp. 127-151). Berkeley: University of Califor-
in three cultures. Journal of Cross-Cultural Psychology, 11,263-280. nia Press.
Brandt, M. E., & Boucher, J. D. (1986). Concepts of depression in emo- Dickey, E. C, & Knower, F. H. (1941). A note on some ethnological
tion lexicons of eight cultures. International Journal of Intercultural differences in recognition of simulated expressions of emotions.
Relations, 10. 321-346. American Journal of Sociology, 47.190-193.
Dietze, A. G. (1963). Types of emotions or dimensions of emotion? A
Briggs, J. L. (1970). Never in anger: Portrait of an Eskimo family. Cam-
comparison of typal analysis with factor analysis. Journal of Psychol-
bridge, MA: Harvard University Press.
ogy, 56,143-159.
Brown, R. (1986). Linguistic relativity. In S. H. Hulse & B. F. Green, Jr.
Dodds, E. R. (1951). The Greeks and the irrational. Berkeley: University
(Eds.), One hundred years of psychological research in America: G.
of California Press.
Stanley Hall and the Johns Hopkins tradition (pp. 241-276). Balti-
Doi, T. (1973). The anatomy of dependence. Tokyo: Kodansha.
more: Johns Hopkins University Press.
Dare E Y, & Kirouac, G. (1986). Reliability of accuracy and intensity
Buck, R. (1988). Human motivation and emotion. New York: Wiley.
judgments of eliciting situations of emotions. Canadian Journal of
Bullock, M., & Russell, J. A. (1984). Preschool children's interpretation
Behavioral Science. IS, 92-103.
of facial expressions of emotion. International Journal of Behavioral
Ducci, L., Arcuri, L., W/Georgis, T, & Sineshaw, T. (1982). Emotion
Development, 7,193-214.
recognition in Ethiopia. Journal of Cross-Cultural Psychology, 13,
Bullock, M., & Russell, J. A. (1985). Further evidence on preschoolers'
340-351.
interpretations of facial expressions of emotion. International Jour-
Eggan, D. (1970). Instruction and affect in Hopi cultural community.
nal of Behavioral Development, 8,15-38.
In J. Middleton (Ed.), From child to adult (pp. 109-133). Garden City,
Burrows, E. G., & Spiro, M. E. (1957). An atoll culture: Ethnography of
NJ: The Natural History Press.
Ifaluk in the Central Carolines. New Haven, CT: Human Relations
Ekman, G. (1955). Dimensions of emotion. Ada Psychologica, 11,
Area Files. (Original work published 1953).
279-288.
Bush, L. E, II. (1973). Individual differences in multidimensional Ekman, P. (1972). Universal and cultural differences in facial expres-
scaling of adjectives denoting feelings. Journal of Personality and sions of emotions. In J. K.. Cole (Ed.), Nebraska symposium on moti-
Social Psychology, 25, 50-57. vation, 1971 (pp. 207-283). Lincoln: University of Nebraska Press.
Calhoun, C, & Solomon, R. C. (1984). What is an emotion? Classic Ekman, P. (1980), Thefaceofman: Expressions of universal emotions in
readings in philosophical psychology. New Tcbrk: Oxford University a New Guinea village. New York: Garland STPM Press.
Press. Ekman, P, & Friesen, W Y(1971). Constants across cultures in the face
Carr, J. E., & Tan, E. K. (1976). In search of true amok: Amok as viewed and emotion. Journal of Personality and Social Psychology, 17,124-
within the Maly. American Journal of Psychiatry, 133,1295-1299. 129.
Carstairs, G. M. (1967). The twice-born: A study of a community of Ekman, P, & Friesen, W Y (1976). Pictures of facial affect. Palo Afto,
high-caste Hindus. Bloomington: Indiana University Press. CA: Consulting Psychologists Press.
Chai, C., & Chai, W (Eds.). (1967). Li Chi: Book of rites (Vol. 1, J. Legge, Ekman, P, Friesen, W Y, O'Sullivan, M., Chan, A., Diacoyanni-Tarlat-
Trans.). New Hyde Park, England: University Books. (Original work zis, I, Heider, K., Krause, R., Le Compte, W A., Pitcairn, T, Ricci-
published, 1885) Bitti, P. E., Scherer, K., Tomita, M, ATzavaras, A. (1987). Universals
Chan, D. W (1990). The meaning of depression: Chinese word associa- and cultural differences in the judgments of facial expressions of
tions. Psychologia, 33,191-196. emotion. Journal of Personality and Social Psychology, 53,712-717.
Cheetham, W S., & Cheetham, R. J. (1976). Concepts of mental illness Ekman, P., & Heider, K. G. (1988). The universality of a contempt
amongst the Xhosa people in South Africa. Australian and New expression: A replication. Motivation and emotion, 12, 303-308.
Zealand Journal of Psychiatry, 10, 39-45. Ekman, P., Levenson, R. W, & Friesen, W Y(1983). Autonomic ner-
Cheng, T. (1977).^ phenomenological study of emotional experience: A vous system activity distinguishes among emotions. Science, 221,
search for cultural differences and similarities in the construction of 1208-1210.
emotion by a Hong Kong Chinese sample. Unpublished master's the- Ekman, P., Sorenson, E. R., & Friesen, W Y (1969). Pan-cultural ele-
sis, University of Hong Kong. ments in the facial displays of emotions. Science, 164, 86-88.
Epstein, A. L. (1984). The experience of shame in Melanesia. London: of person cognition. An English-Chinese comparison. Journal of
Royal Anthropological Institute. personality and Social Psychology, 51,1097-1105.
Fajans, J. (1983). Shame, social action, and the person among the Bain- Hofstede, G. (1980). Culture's consequences. Beverly Hills, CA: Sage.
ing. Ethos, 11,166-178. Hollan, D. (1988). Staying "cool" in Toraja. Ethos, 16, 53-72.
Fehr, B., & Russell, J. A. (1984). Concept of emotion viewed from a Holland, D., & Quinn, N. (1987). Cultural models in language and
prototype perspective. Journal of Experimental Psychology: General, thought. Cambridge, England: Cambridge University Press.
113, 464-486. Howell, S. (1981). Rules not words. In P. Heelas & A. Lock (Eds),
Fehr, B., & Russell, J. A. (1991). The concept of love viewed from a Indigenous psychologies: The anthropology of the self {pp. 133-143).
prototype perspective. Journal of Personality and Social Psychology, San Diego, CA: Academic Press.
60, 425-438. Howell, S. (1984). Society and cosmos: Chevfong of peninsular Malay-
Fehr, B, Russell, J. A., & Ward, L. M. (1982). Prototypicality of emo- sia. London: Oxford University Press.
tions: A reaction time study. Bulletin of the Psychonomic Society, 20, Imada, H. (1989). Cross-language comparisons of emotional terms
253-254. with special reference to the concept of anxiety Japanese Psychologi-
Fillenbaum, S., & Rapoport, A. (1971). Structures in the subjective lexical Research, 31,10-19.
con. San Diego, CA: Academic Press. Imada, H., Araki, M., &Kujime, Y (1991). Comparisons of 'concepts of
Fodor, J. A. (1981). Representations. Cambridge, MA: MIT Press. anxiety, fear, and depression in English and Japanese languages.
Fortes, M. (1970). Social and psychological aspects of education in Manuscript in preparation, Kwansei Gakuin University, Depart-
Taleland. In J. Middleton (Ed.), From child to adult (pp. 14-74). Gar- ment of Psychology, Hyogo, Japan.
den City, NJ: The Natural History Press. Izard, C. E. (1971). The face of emotion. New York: Appleton-Century-
Frijda, N. H. (1953). The understanding of facial expression of emo- Crofts.
tion. Acta Psychologies 9, 294-362. Izard, C. E. (1977). Human emotions. New York: Plenum Press.
Frijda, N. H. (1987). Emotion, cognitive structure, and action ten- Izard, C. E. (1980). Cross-cultural perspectives on emotion and emo-
dency. Cognition and Emotion, 1,115-143. tion communication. In H. C. Triandis & W. Lonner (Eds), Hand-
Gardiner, H. M., Metcalf, R. C, & Beebe-Center, J. G. (1937). Feeling book of cross-cultural psychology (pp. 185-221). Boston: Allyn &
and emotion: A history of theories. New "Ybrk: American Book Co. Bacon.
Geertz, C. (1966). Person, lime, and conduct in Bali: An essay in cultural Izard, C. E., & Buechler, S. (1980). Aspects of consciousness and per-
analysis. New Haven, CT: Yale University Southeast Asia Studies. sonality in terms of differential emotions theory. In R. Plutchik & H.
Geertz, H. (1959). The vocabulary of emotion: A study of Javanese Kellerman (Eds.), Emotion: Theory, research, and experience (Vol. 1,
socialization processes. Psychiatry, 22, 225-237. pp. 105-187). San Diego, CA: Academic Press.
Gehm, T. L, & Scherer, K. R. (1988). Factors determining the dimen- Jeffery, G. H., & Patterson, D. L. (1987). Field dependence and sensitiv-
sions of subjunctive emotional space. In K. R. Scherer (Ed^, Facets of ity to facial affect cues. Semiotica, 64, 335-341.
emotion: Recent research (pp. 99-113). Hillsdale, NJ: Erlbaum. Johnson, A., Johnson, Q, & Baksh, M. (1986). The colours of emotions
Gerber, E. (1975). The cultural patterning of emotions in Samoa. Un- in Machiguenga. American Anthropologist, 88, 674-681.
published doctoral dissertation, University of California, San Diego. Johnson, D, & Johnson, C. A. (1965). Totally discouraged: A depres-
Gobeil, O. (1973). El susto: A descriptive analysis. International Journal sive syndrome of the Dakota Sioux. Transcultural Psychiatric Re-
of Social Psychiatry, 19, 38-44. search Review, 2,141-143.
Graham, J. A., Ricci Bitti, P, & Argyle, M. (1975). A cross-cultural Johnson-Laird, P. N., & Oatley, K. (1986). The meaning of emotions: A
study of the communication of emotion by facial & gestural cues. cognitive theory and a semantic analysis. Unpublished manuscript,
Journal of Human Movement Studies, 1,68-77. Cambridge University, Applied Psychology Unit, Cambridge, En-
Hallowell, A. I. (1955). Culture and experience. Philadelphia: Univer- gland.
sity of Pennsylvania Press. Johnson-Laird, P. N, & Oatley, K. (1989). The language of emotions:
Harre, R. (1986). An outlineof the social constructionist viewpoint. In An analysis of a semantic field. Cognition and Emotion, J, 81-123.
R. Harre (Ed.), The social construction of emotions (pp. 2-14). Ox- Kagitcibasi, C, & Berry, J. W (1989). Cross-cultural psychology:
ford, England: Basil Blackwell. Current research and trends. Annual Review of Psychology, 40,493-
Harre, R., & Finlay-Jones, R. (1986). Emotion talk across time. In R. 531.
Harr£ (Ed.), The social construction of emotions (pp. 220-233). Ox- Kahneman, D., & Miller, D. T. (1986). Norm theory: Comparing reality
ford, England: Basil Blackwell. to its alternatives. Psychological Review, 93,136-153.
Harrison, R. H. (1986). The grouping of affect terms according to the Kay, P, & Kempton, W (1984). What is the Sapir-Whorf hypothesis?
situations that elicit them: A test of a cognitive theory of emotion. American Anthropologist, 86, 65-79.
Journal of Research in Personality, 20, 252-266. Keeler, W (1983). Shame and stage fright in Java. Ethos, 11,152-165.
Heelas, P., & Lock, A. (Eds.). (1981). Indigenous psychologies: The an- Kiesler, D. J., Horner, M. S., Larus, J. P, & Chapman, R. C. (1987).
thropology of the self. San Diego, CA: Academic Press. Measurement of evoked emotion in interpersonal transactions. Unpub-
Herrmann, D. J., & Raybeck, D. (1981). Similarities and differences in lished manuscript, Department of Psychology, Virginia Common-
meaning in six cultures. Journal of Cross-Cultural Psychology, 12, wealth University.
194-206. Kiev, A. (1968). Curanderismo: Mexican-American folk psychiatry.
Hiatt, L. R. (1978). Classification of the emotions. InL. R. Hiatt(Ed.), New York: Free Press.
Australian aboriginal concepts (pp. 182-187). Princeton, NJ: Hu- Kilbride, J. E, & Yarczower, M. (1980). Recognition and imitation of
manities Press. facial expressions: A cross-cultural comparison between Zambia
Hochschild, A. R. (1983). The managed heart. Berkeley: University of and the United States. Journal of Cross-Cultural Psychology, 11,281-
California Press. 296.
Hoekstra, H. A. (1986). Cognition and affect in the appraisal of events. Kirmayer, L. J. (1989). Cultural variations in the response to psychiat-
Groningen. The Netherlands: Rijksuniversteit te Groningen. ric disorders and emotional distress. Social Science and Medicine,
Hoffman, C, Lau, I, & Johnson, D. R. (1986). The linguistic relativity 29, 327-339.
Kolenda, K. (1987). On human emotions. American Anthropologist, tures. In H. Triandis & J. Draguns (Eds), Handbook ofcross-cultural
89, 946-947. psychology (pp. 237-289). Boston: Allyn & Bacon.
Kracke, W H. (1981). Kagwahiv Mourning I: Dreams of a bereaved Matsumoto, D. (1989). Cultural influences on the perception of emo-
father. Ethos, 9. 258-275. tion. Journal oj'Cross-Cultural Psychology, 20, 92-105.
Kundera, M. (1980). The book of laughter and forgetting (M. H. Heim, Matsumoto, D, & Ekman, P. (1989). American-Japanese cultural dif-
Trans.) New York: Knopf. (Original work published 1979). ferences in intensity ratings of facial expressions of emotion. Motiva-
Lakoff, G. (1987). Women, fire and dangerous things: What categories tion and Emotion, 13,143-157.
reveal about the mind. Chicago: University of Chicago Press. Matsumoto, D., & Kishimoto, H. (1983). Developmental characteris-
Langness, L. L. (1965). Hysterical psychosis in the New Guinea High- tics in judgments of emotion from nonverbal vocal cues. Interna-
lands: A Bena Bena example. Psychiatry, 28, 258-277. tionalJoumalofInterailluralRelations, 7, 415-424.
Lebra, T. S. (1983). Shame and guilt: A psychocultural view of the Matsuyama, Y, Kama, H., Kawamura, Y, & Mine, H. (1978). An analy-
Japanese self. Ethos, II, 192-231. sis of emotional words. The Japanese Journal of Psychology, 49,229-
Leff, J. (1973). Culture and the differentiation ofemotional states. Brit- 232.
ish Journal of Psychiatry, 123, 299-306. McAndrew, F. T. (1986). A cross-cultural study of recognition thresh-
Leff, J. (1977). The cross-cultural study of emotions. Culture, Medicine, olds for facial expressions of emotion. Journal ofCross-Cultural Psy-
and Psychiatry, J, 317-350. chology, / 7, 211-224.
Medin, D. L. (1989). Concepts and conceptual structure. American
Leff, J. (1981). Psychiatry around the globe: A transcultural view. New
Psychologist, 44,1469-1481.
York: Marcel Dekker.
Mervis, C. B. (1987). Child-basic object categories and early lexical
Leighton, A. H., Lambo, T. A., Hughes, C. C, Leighton, D. C, Murphy,
development. In U. Neisser (Ed.), Concepts and conceptual develop-
J.M.,&Mack\in,nB.(l963).PsychiatricdisorderamongtheYoruba.
ment: Ecological and intellectual factors in categorization (pp. 201-
Ithaca, NY: Cornell University Press.
233). Cambridge, England: Cambridge University Press.
Leventhal, H. (1974). Emotions: A basic problem forsocial psychology.
Mesquita, B, & Frijda, N. (in press). Cultural variations in emotions.
In C. Nemeth (Ed.), Social psychology (pp. 1-51). Chicago: Rand
Psychological Bulletin.
McNally.
Micko, H. C. (1970). A "halo"-model for multidimensional ratio scal-
Levine, N. E. (1988). The dynamics of polyandry: Kinship, domesticity,
ing. Psychometrika, 35,199-227.
and population on the Tibetan border. Chicago: University of Chi-
Morice, R. (1978). Psychiatric diagnosis in a transcultural setting: The
cago Press.
importance of lexical categories. British Journal of Psychiatry, 132,
Levy, R. I. (1973). Tahilians. Chicago: University of Chicago Press.
87-95.
Levy, R. I. (1983). Introduction: Self and emotion. Ethos, 11.128-134.
Morice, R. (1979). Personality disorder in transcultural perspective.
Levy, R. I. (1984). The emotions in comparative perspective. In K. R.
Australian and New Zealand Journal of Psychiatry, 13, 293-300.
Scherer & P. Ekman (Eds.), Approaches to emotion (pp. 397-412).
Myers, F. R. (1979). Emotions and the self: Atheoryofpersonhoodand
Hillsdale, NJ: Erlbaum.
political order among Pintupi aborigines. Ethos, 7, 343-370.
Lindholm, C. (1982). Generosity and jealousy: The Swat Pukhtun of
Myers, F. R. (1988). The logic and meaning of anger among Pintupi
Northern Pakistan. New York: Columbia University Press.
aborigines. Man, 23, 589-610.
Lowrie, W (1944). Translator's preface. In S. Kierkegaard & W Lowrie
Neisser, U. (1987). Concepts and conceptual development: Ecological
(Trans.), The concept of dread. London: Oxford University Press.
and intellectual factors in categorization. Cambridge, England: Cam-
Lucy, J. A., & Shweder, R. A. (1979). Whorf and his critics: Linguistic
bridge University Press.
and nonlinguistic influences on color memory. American Anthro-
Neufeld, R. W J. (1975). A multidimensional scaling analysis of schizo-
pologist, 81, 581-615.
phrenics' and normals' perceptions of verbal similarity Journal of
Lundberg, U, & Devine, B. (1975). Negative similarities. Educational Abnormal Psychology, 84, 498-507.
and Psychological Measurement, 35, 797-807. Neufeld, R. W J. (1976). Simultaneous processing of multiple stimulus
Lutz, C. (1980). Emotion words and emotional development on Ifaluk dimensions among paranoid and nonparanoid schizophrenics. Mul-
Atoll. Unpublished doctoral dissertation, Harvard University. tivariate Behavioral Research, 11, 425-441.
Lutz, C. (1982). The domain of emotion words in Ifaluk. American Niit, T, & Valsiner, J. (1977). Recognition of facial expressions: An
Ethnologist, 9,113-128. experimental investigation of Ekman's model. Ada el Commenta-
Lutz, C. (1983). Parental goals, ethnopsychology, and the development tionesUniversitatis Tarvensis, 429, 85-107.
ofemotional meaning. Ethos, 11, 246-262. Obeyesekere, G. (1981). Medusa^ hair: An essay on personal symbols
Lutz, C. (1985). Ethnopsychology compared to what? Explaining be- and religious experience. Chicago: University of Chicago Press.
haviour and consciousness among the Ifaluk. In G. M. White & J. Orley, J. H. (1970). Culture and mental illness. Nairobi, Kenya: East
Kirkpatrick (Eds), Person, self, and experience: Exploring Pacific Africa.
ethnopsychobgies (pp. 35-79). Berkeley: University of California Ortony, A., Clore, G. L., & Collins, A. (1988). The cognitive structure of
Press. emotions. Cambridge, England: Cambridge University Press.
Lutz, C. (1988). Ethnographic perspectives on the emotion lexicon. In Osaka, N. (1986). Cross-cultural differences in the perception of facial
V Hamilton, G. H. Bower, & N. H. Frijda (Eds.), Cognitive perspec- expressions of ambiguous Noh faces. Bulletin of the Psychonomic
tives on emotion and motivation (pp. 399-419). Norwell, MA: Kluwer Society, 24, 427-430.
Academic. Osgood, C.E. (1966). Dimensionality of the semantic space for commu-
Lutz, C, & White, G. M. (1986). The anthropology of emotions. Annual nication via facial expressions. Scandinavian Journal of Psychology,
Review of Anthropology, 15, 405-436. 7,1-30.
Mandal, M. K., Saha, G. B., & Palchoudhury, S. (1986). A cross-cultural Osgood, C. E. (1969). On the whys and wherefores of E, P, and A.
study on facial affect. Journal of Psychological Researches, 30,140- Journal of Personality and Social Psychology, 12,194-199.
143. Osgood, C. E., May, W H., & Miron, M. S. (1975). Cross-cultural univer-
Marsella, A. J. (1981). Depressive experience and disorder across cul- sals of affective meaning. Urbana: University of Illinois Press.
Plutchik, R. (1980). Emotion: A psychoevotutionary synthesis. New ety in terms of emotional response factors. Journal of Consulting and
York: Harper & Row. Clinical Psychology, 42, 79-83.
Poole, F. J. P. (1985). Coming into social being: Cultural images of Russell, J. A., & Mehrabian, A. (1975). The mediating role of emotions
infants in Bimin-Kuskusmin folk psychology. In G. M. White & J. in alcohol use. Journal of Studies on Alcohol, 36,1508-1536.
Kirkpatrick (Eds.), Person, self, and experience: Exploring Pacific Russell, J. A., & Mehrabian, A. (1977). Evidence for a three-factor
ethnopsychologies (pp. 183-242). Berkeley: University of California theory of emotions. Journal of Research in Personality, 11, 273-294.
Press. Russell, J. A., & Snodgrass, J. (1987). Emotion and the environment. In
Potter, S. H. (1988). The cultural construction of emotion in rural Chi- D. Stokols & I. Altman (Eds.), Handbook ofenvimnmental psychology
nese social life. Ethos. 16,181-208. (pp. 245-280). New York: Wiley.
Resner, G., & Hartog, J. (1970). Concepts and terminology of mental Sabini, J., & Silver, M. (1982). Moralities of everyday life. Oxford, En-
disorder among Malays. Journal of Cross-Cultural Psychology, I, gland: Oxford University Press.
369-381. Schachter, S., & Singer, J. E. (1962). Cognitive, social, and physiological
Riesman, P. (1977). freedom in Fulani social life: An introspective eth- determinants of emotional state. PsycA0fogKM/.Revievii <SS*. 379-399.
nography (M. Fuller, Trans.). Chicago: University of Chicago Press. Scherer, K. R. (1984). On the nature and function of emotion: A compo-
(Original work published 1974). nent process approach. In K. R. Scherer & P. Ekman (Eds.), Ap-
Rosaldo, M. Z. (1980). Knowledge and passion: Ilongot notions of self proaches to emotion (pp. 293-317). Hillsdale, NJ: Erlbaum.
and social life. Cambridge, Cambridge University Press. Scherer, K. R. (1988). Facets of emotion. Hillsdale, NJ: Erlbaum.
Rosaldo, M. Z. (1983). The shame of headhunters and the autonomy of Scherer, K. R., Wallbott, H. G., & Summerfield, A. B. (Eds.). (1986).
self. Ethos, 11,135-151. Experiencing emotion: A cross-cultural study. Cambridge, England:
Cambridge University Press.
Rosaldo, R. I. (1984). Grief and a headhunter's rage: On the cultural
Schieffelin, E. L. (1985). The cultural analysis of depressive affect: An
force of emotions. In S. Plattner & E. M. Bruner (EdsJ, Text, play,
example from New Guinea. In K. Kleinman & B. Good (Eds.), Cul-
and story: The construction and reconstruction of self and society (pp.
ture and depression: Studies in anthropology and psychiatry of affect
178-195). Washington, DC American Ethnological Society.
and disorder (pp. 101 -13 3). Berkeley: University of California Press.
Rosch, E. (1973). On the internal structure of perceptual and semantic
Schlosberg, H. (1952). The description of facial expressions in terms of
categories. In T. E. Moore (Ed J, Cognitive development and acquisi-
two dimensions. Journal of Experimental Psychology, 44, 229-237.
tion of language (pp. 111-144). San Diego, CA: Academic Press.
Searle, J. R. (1983). Intentionality: An essayin the philosophy of mind.
Rosch, E. (1975). Cognitive representations of semantic categories.
Cambridge, England: Cambridge University Press.
Journal of Experimental Psychology: General, 104,192-233.
Shaver, P, Schwartz, J., Kirson, D, & O'Connor, C. (1987). Emotion
Rosch, E. (1977). Human categorization. In N. Warren (Ed.), Studies in
knowledge: Further exploration of a prototype approach. Journal of
cross-cultural psychology (Vol. 1, pp. 1-49). San Diego, CA: Aca-
Personality and Social Psychology, 52,1061-1086.
demic Press.
Shepard, R. N. (1962). The analysis of proximities: Multidimensional
Roseman, I. J. (1984). Cognitive determinants of emotion: A structural scaling with an unknown distance function. Psychometrika, 27,
theory. In P. Shaver (Ed.), Review of personality and social psychology 125-139,219-246.
(Vol. 5, pp. 11-36). Beverly Hills, CA: Sage. Shimoda, K., Argyle, M, & Ricci Bitti, P. (1978). The intercultural
Roseman, M. (1988). Head, heart, odor and shadow: The structure of the recognition of emotional expressions by three national racial
self and the emotional world. Unpublished manuscript, Tufts Univer- groups: English, Italian, and Japanese. European Journal of Social
sity, Department of Anthropology, Medford, MA. Psychology, 8.169-179.
Rubel, A. (1964). The epidemiology of a folk illness: Suslo in Hispanic Shostak, M. (1983). Nisa: The life and words of a IKung woman. New
America. Ethnology, 3, 268-283. York: Vintage.
Russell, J. A. (1978). Evidence of convergent validity on the dimensions Shott, S. (1979). Emotion and social life: A symbolic interactionist
of affect. Journal of Personality and Social Psychology, 36, 1152- analysis. American Journal of Sociology, 84,1317-1334.
1168. Simons, R. C, & Hughes, C. C. (Eds). (1985). The culture-bound syn-
Russell, J. A. (1980). A circumplex model of affect. Journal of Personal- dromes: Folkillnesses of psychiatricand anthropological interest. Bos-
ity and Social Psychology, 39,1161-1178. ton: Reidel.
Russell, J. A. (1983). Pancultural aspects of human conceptual organi- Smith, C. A., & Ellsworth, P. C. (1985). Patterns of cognitive appraisal
zation of emotions. Journal of Personality and Social Psychology, 45, in emotion. Journal of Personality and Social Psychology, 48, 813-
1281-1288. 838.
Russell, J. A, (1989). Measures of emotion. In R. Plutchik &H. Keller- Solomon, R. C. (1976). The passions. Garden City, NJ: Anchor Press.
man (Eds.), Emotion: Theory, research, and experience (Vol. 4, pp. Sorenson, E. R. (1976). The edge of the forest. Washington, DC: Smith-
83-111) San Diego, CA: Academic Press. sonian Institution Press.
Stone, L. A. (1971). Congruent multidimensional scaling results ob-
Russell, J. A. (1991). In defense of a prototype approach to emotion
tained using the halo-model and the Stone-Coles method model.
concepts. Journal of Personality and Social Psychology, 60, 37-47.
Perceptual and Motor Skills, 33, 524-526.
Russell, J. A, & Bullock, M. (1985). Multidimensional scaling of emo-
Stone, L. A, & Coles, G. J. (1970). Correlational similarity: The basis
tional facial expressions: Similarities from preschoolers to adults.
for a new revised method of similarity analysis. Studio Psychologica
Journal of Personality and Social Psychology, 48,1290-1298.
(Bratislava), 12, 258-265.
Russell, J. A., & Bullock, M. (1986). On the meaning preschoolers Storm, C, & Storm, T. (1987). A taxonomic study of the vocabulary of
attribute to facial expressions of emotion. Developmental Psychol- emotions. Journal of Personality and Social Psychology, 53,805-816.
ogy, 22. 97-102. Stringer, P. (1967). Cluster analysis of non-verbal judgments of facial
Russell, J. A., Lewicka, M., & Niit, T. (1989). A cross-cultural study of a expressions. British Journal of Mathematical and Statistical Psychol-
circumplex model of affect. Journal of Personality and Social Psychol- ogy, 20,71-79.
ogy, 57, 848-856. Swartz, M. J. (1988). Shame, culture, and status among the Swahili of
Russell, J. A., & Mehrabian, A. (1974). Distinguishing anger and anxi- Mombasa. Ethos, 16, 21-51.
Tanaka-Matsumi, J, & Marsella, A. J. (1976). Cross-cultural variations Vinacke, W E. (1949). The judgment of facial expressions by three
in the phenomenological experience of depression: I. Word associa- national-racial groups in Hawaii: I. Caucasian faces. Journal of Per-
tion studies. Journal of Cross-Cultural Psychology, 7. 379-396. sonality, 17, 407-429.
Tanaka-Matsumi, J., & Marsella, A. J. (1977). Ethnoculturat variations Vinacke, W E., & Bong, R. W (1955). The judgment of facial expres-
in the subjective experience of depression: Semantic differential. Un- sions by three national-racial groups in Hawaii: II. Oriental faces.
published manuscript, University ofHawaii, Department ofPsychol- Journal of Social Psychology, 41,184-195.
ogy, Honolulu. Wallace, A. E C, & Carson, M. T. (1973). Sharing and diversity in
Termansen, R E.. & Ryan, J. (1970). Health and disease in a British emotion terminology. Ethos, 1,1-29.
Columbian Indian community. Canadian Psychiatric Association Weiner, B. (1982). The emotional consequences of causal attributions.
Journal, 15,121-127. In M. S. Clark & S. T. Fiske (Eds.), Affect and cognition (pp. 185-209).
Tomkins, S. S. (1979). Script theory: Differential magnification of af- Hillsdale, NJ: Erlbaum.
fects. In H. E. Howe& R. A. Dienstbier(Eds), Nebraska symposium Westermeyer, J. (1972). A comparison of amok and other homicides in
on motivation, J97S (pp. 201-236). Lincoln: University of Nebraska Laos. American Journal of Psychiatry, 6, 703-709.
Press. White, G. M. (1980). Conceptual universals in interpersonal language.
Tomkins, S. S. (1984). Affect theory. In K. R. Scherer & P. Ekman American Anthropologist, 82, 759-781.
(Eds.), Approaches to emotion. Hillsdale, NJ: Erlbaum. Wierzbicka, A. (1986). Human emotions: Universal or culture-spe-
Tousignant, M. (1984). Pena in the Ecuadorian Sierra: A psy- cific? American Anthropologist, 88, 584-594.
choanthropological analysis of sadness. Culture, Medicine and Psy- Wikan, U (1989). Illness from fright or soul loss: A North Balinese
chiatry, 8, 381-398. culture-bound syndrome? Culture, Medicine and Psychiatry, 13, 25-
Triandis, H. C, & Lambert, W W (1958). A restatement and test of 50.
Schlosberg's theory of emotion with two kinds of subjects from Winkelmayer, R., Exline, R. Y, Gottheil, E, & Paredes, A. (1978). The
Greece. Journal of Abnormal and Social Psychology, 56, 321-328. relative accuracy of US., British, and Mexican raters in judging the
Tseng, W-S., & Hsu, J. (1969). Chinese culture, personality formation emotional displays of schizophrenic and normal US. women. Jour-
and mental illness. International Journal of Social Psychiatry, 16, nal of Clinical Psychology, 34,600-608.
5-14. Wolfgang, A., & Cohen, M. (1988). Sensitivity of Canadians, Latin
Turnbull, C. M. (1961). The forest people. New York: Simon & Schuster. Americans, Ethiopians, and Israelis to interracial facial expressions
Turnbull, C. M. (1972). The mountain people. New York: Simon & of emotion. International Journal of Intercultural Relations, 12,139-
Schuster. 151.
Turner, V (1967). The forest of symbols: Aspects ofNdembu ritual. Ith- Yoshida, M., Kinase, R., Kurokawa, J, & Yashiro, S. (1970). Multidi-
aca, NY: Cornell University Press. mensional scaling of emotion. Japanese Psychological Research, 12,
Tyler, S. A. (1969). Cognitive anthropology. New York: Holt, Rinehart & 45-61.
Winston.
van Bezooijen, R., Otto, S. A., & Heenan, T. A. (1983). Recognition of
vocal expressions of emotion: A three-nation study to identify univer- Received September 15,1989
sal characteristics. Journal of Cross-Cultural Psychology, 14, 387- Revision received April 2,1991
406. Accepted April 17,1991 •

Emotions and Culture 2

Uploaded by

Copyright:

Available Formats

Emotions and Culture 2

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Emotions and Culture 2

Uploaded by

Copyright:

Available Formats

Psychological Bulletin Copyright 1991 by the American Psychological Association, Inc.

1991, Vol. 110, No. 3,426-450 0033-2909/91/J3.00

Culture and the Categorization of Emotions

Table 1 feeling state associated with respect, is something peculiarly

Discussion in English. Gerber (1975) presented the Samoan word alofa as

Study Language % Language % Language %

Izard(1971) American, 80.7 Swedish 83.4 Japanese 65.4

Matsumoto and Ekman

English Japanese English* Japanese English Japanese

• Average between Scotland and United States.

Basic-level term Problem

Happiness Missing in Chewong (Howell, 1981)

You might also like