A Taxonomy of Emotions: How Do We Begin

47. A Taxonomy of Emotions: How Do We Begin[1]?

Thomas J. Scheff

Abstract: This paper proposes the need for a taxonomy (system of classification) of emotions, and a way we might go about developing one. We need a web of emotion names clearly defined as concepts, rather than using vernacular words. The first question is: why? It would seem that the meaning of vernacular emotion names, even in research, is extraordinarily ambiguous.  The second question: aren’t existing taxonomies adequate? It is clear that there is no consensus, or even much similarity, between the various proposals. The third question: if we wish to develop a new taxonomy, how should we go about it? Most of the paper is devoted to this last topic. The problem is to devise a way of generating emotion concepts that are closely related to the actual particulars that the concepts are to represent. This question can be treated as a part/whole problem, with the concepts being wholes, the empirical particulars the parts. Studies by Goffman are considered that conceptualize embarrassment/shame, on the one hand, and attunement (mutual awareness) on the other.  His forte was analyzing many concrete examples, the parts, before using them to help define a concept, the whole. In conclusion, several implications of the method proposed here are offered.

1. Why Do We Need A Taxonomy?

Although there are by now a large number of studies of emotions, almost all of them fail to define the emotion they are investigating. Existing studies, by and large, use vernacular words for emotion names. One reason that we need concepts is that in the English language, particularly, these vernacular words are ambiguous.

For brevity, only a few of the many possible examples will be offered: pride, love, grief/distress, and shame/embarrassment. In English, unlike most other languages, the word “pride” can mean either genuine/justified pride, on the one hand, or arrogance/hubris (“Pride goeth before the fall”), on the other. Oddly, in English, if the word pride is used without an adjective, as it usually is, it connotes hubris rather than justified pride.

The word love in English, far more than in any other language, is so broad and inclusive as to be wildly ambiguous, as Solomon (1981; 1982) has shown[2]. There are 24 meanings attributed in unabridged dictionaries, so that diverse feelings are all included: infatuation, lust, and heartbreak, for example. The broadness of usage allows the inclusion of even highly pathological states (Women Who Love Too Much).

The meaning of the word shame in English, on the other hand, is so narrow as to exclude most of the kind of feelings that other languages define as members of the shame family. In English, shame refers to an intensely painful emotion of disgrace. But other languages include many more feelings, such as ordinary embarrassment, modesty, and even the anticipation of shame (modesty or “sense of shame.”)

There is also a second reason that emotion concepts are needed. Emotion researchers use a wide variety of emotion names: there are many different names used for what seems to be the same emotion, each seemingly connoting a subtle or sometimes a flagrantly different meaning.

One example would be the emotion that is usually called grief in the clinical literature, that follows from loss of an attachment, or anticipation of that loss. There very large literatures on attachment and on child development that use the term distress instead. Distress is much broader than grief since it connotes physical as well as emotional pain, and implies consciousness more than grief.

Silvan Tomkins (1962) seems to have started the use of the word distress rather than grief. In the first three volumes of Affect/Imagery/Consciousness (1962; 1963; 1965; 1992) the word distress is used frequently, with the word grief occurring only on one page (V.2, p. 6). However, in Volume 4 (1992), there is an abrupt change:  distress disappears, its place apparently taken by grief.

In the first three volumes it is fairly clear what Tomkins means by distress, because he connects distress to loss and crying. In IV, he makes this connection using only the word grief. What happened? As far as I know, there has been no published response to this dramatic change in nomenclature.

The original studies of facial expression of emotion followed Tompkins first usage: neither Ekman and his colleagues nor Izard refer to grief. However, later works, such as Harre and Parrott, refer only to grief, never to distress. Plutchik (2003) also refers only to grief. Others use the word sadness, rather than distress or grief. Volkan (1988, 1997, 2004), one of the leading theorists of conflict, uses an entirely different nomenclature. What I would call unresolved grief, a standard diagnostic category in psychiatry, plays a central role in his work. Yet instead of referring to it, he uses only the phrase “the failure to mourn.”

I have found only one explicit discussion of the relationship between distress and grief, in Izard (1977). What he proposed, that distress is the primary affect of which grief is only one ingredient, seems to me the exact opposite of the majority understanding: grief is the primary affect. However, in a recent publication (2004), it is clear that Izard has, like Tomkins, switched terminology. He doesn’t switch from distress to grief, as Tomkins did in his 1992 volume, but from distress to sadness. As with Tomkins, there is no explanation of the change.

There are many other emotions words run amok in studies and discussions of emotion. The word anxiety, for example, has come to mean almost any diffuse emotion. The ambiguities of the word fear, and the cover words that are used in its place, is a smaller but still highly significant domain. If, as it seems reasonable to believe, that fear should refer only to the instinctive reaction to physical danger, what could it mean in  “I fear rejection?” In the English language, at least, many, many words are used to avoid saying the word shame. Perhaps the use of the word fear in the sentence above is just one of these instances.

There have been several surveys of the occurrence of specific emotions in populations, but the results are ambiguous. For example, when a subject is asked about his or her anger events, a detailed definition of anger that is inclusive of the cognate emotion words the subjects might use is not offered by the researcher. Without such a definition, however, it is not clear whether the distribution that results refers to emotions or to the meaning of emotion words. I have noticed for example, that my students often use the term “pissed off” rather than angry, and surprisingly, that some of them do not connect this feeling with anger. These latter students, if asked about their anger events, might reply that they don’t have them.

Aaron Lazare told me about a similar experience he had in an actual anger study he did. Apparently he was meeting with a large group of elderly Jewish women in NYC to investigate their experiences of emotion. When he came to anger, however, they all denied its occurrence. He tried many cognates (irritated, annoyed, etc) , but there was silence until he tried “aggravation.” Everyone responded enthusiastically with raised hands and murmurs of recognition. One woman cried out: “Oy gewalt! Have we got aggravation!”

In studies in English in which emotions go undefined, there are likely to be different understandings by researchers, the subject, and readers. We need concepts of emotions so that these different groups will understand each other within and between the three groups.

2. What about existing taxonomies? As far as can be seen in publications, experts disagree on almost everything about classifying emotions. Several studies have pointed out the lack of agreement on which emotions are basic. Ortony et al (1988, p.27) show no agreement on this issue among twelve investigators, some leading experts in the field. Even the number of such emotions, much less the specific emotions, is in contention; the fewest proposed is two, the most, eleven. There is not a single emotion word that shows up on all 12 lists. Plutchick (2003) also shows complete disagreement (see the table of 16 theorists, p. 73.)

This disagreement involves emotion words in only one language, English. The comparison of emotion words in different languages opens up a second chaos. Anthropological and linguistic studies suggest that just as the experts disagree on the number and names of the basic emotions, so do languages. Cultural differences in emotion words can only be mentioned in passing here; it is so complex as to require a paper in itself.

There are also several monocultural and crosscultural studies that seek to classify vernacular emotion words in terms of emotion concepts. These studies may be premature because we lack a taxonomy of emotion concepts that is consensual among researchers, and between them and the subjects. A classification of vernacular emotion words, a second taxonomy, will be needed, but only after we have organized the first one.


There are by now many studies that compare either emotion words or still photos of facial expressions in different languages/cultures, finding mostly similarities. The problem with these studies is that in order to use a quantitative format, they have focused entirely on the words or still photos themselves, omitting nonverbal and contextual elements. But the meaning of emotion words, particularly, is largely dependent on these extra-verbal components. As already indicated, the phrase “I love you” can mean everything or nothing, depending on how it is said, and in what context. Leaving out either nonverbal components (as in Shaver    ) or contextual ones (as in the many still photo studies of facial expressions) may invalidate the findings.

In their study of limbic communication, Lewis et al (2000) make a similar observation with respect to formulaic approaches to psychotherapy. Applying their idea to the present topic, comparisons of languages that ignore nonverbal and/or contextual components of emotion words will find them “like Reader’s Digests condensed books --- where, by purging the particular, the stories are strangely identical (p. 184).”

3. How should one begin to develop emotion concepts, which as Lewis et al suggest, means getting down the particulars? Most of the attempts to generate taxonomies referred to above not only disagree with each other, but also seem arbitrary. In the “theoretical” studies, there is no attempt to link the classifications to particulars, such as data or even examples. They are theoretical in the sense that they seem to be generated out of thin air.

The few taxonomic studies that are based on data also seem arbitrary, but in a different sense, as discussed above.  Surveys of the meaning and/or occurrence within a single language group, and those which are cross-cultural, are acontextual, and fail to sufficiently explore the meaning that subjects attribute to the emotion names. Similarly, the many studies that employ still photos have the same limitations.

It may be that there is a step preliminary to systematic theoretical and empirical investigation. This step would involve the clearing of tropes, since vernacular emotion words are all tropes (taken-for-granted metaphors)[3].

The idea of trope clearing has been proposed by the philosopher Quine (1979):

The neatly worked inner stretches of science are an open space in the tropical jungle, created by clearing tropes [metaphors] away” (1979, p. 160. Quoted by Manning 1992, p. 147).

That is to say, it often happens that ­before scientific procedures are applicable, an obstructive metaphor has to be overthrown. A trope is a ruling metaphor that is taken for granted. The history of the physical sciences is full of examples of the clearing away of obstructive tropes. Progress in the astronomy of planetary motion was delayed for over a century because of the trope that the earth was the center of the universe.

Astronomers, like everyone else, took for granted that the planets circled around the earth. In the l6th century, Brahe, a great scientist and mathematician, had made an accurate charting of the transit of Venus[4]. But he could not plot the shape of the orbit because he assumed it was around the earth. Kepler, who inherited the data after Brahe’s death, was equally puzzled for many years.

The idea of logocentric universe was so ingrained that Kepler hit upon the solution only inadvertently. In his frustration, he devised a geometric model of the planetary orbits based on solid figures representing polyhedrons. The model was ridiculous except for one feature; Kepler had inadvertently placed the sun, rather than the earth, at the center.

Similarly, Einstein began work on relativity as a joke that challenged the trope of the absoluteness of time and space. Although he had a doctorate in physics, Einstein knew little mathematics. He had to get help to put his anti-trope into mathematical form. He was a trope clearer before a scientist.

Parts and Wholes

If we don’t seem to be making much progress with emotion concepts either by theoretical taxonomies or by systematic empirical studies, what is left? How can we develop emotion concepts that are closely related to the reality they are supposed to represent? One way to approach this problem is to treat it as a part/whole problem: how to relate abstract concepts, the wholes, to the particulars of actual experience, the parts (Scheff 1997).

This problem came up in an unusual way in an interview with the novelist Muriel Spark concerning her novel The Bachelors. The novel describes the lives of bachelors of varying ages and stations in life in London in remarkable detail.  Ms. Spark, a middle-aged unmarried woman at the time, was asked how she could possibly know so much about such men. Her answer was “A lifetime of combing lint.” By lint, Spark seems to be referring to the particulars of the lives of many people that she had learned out about over her own lifetime. This idea might be as useful in the early stages of science as it seems to be in the writing of novels. As William Blake put it, “... Art and Science cannot exist but in minutely organized particulars.”

In one chapter of Richard Lazarus’s last book (1997), he suggests a new approach, at least for him, to the study of emotion. His initial discussion, at least, implies that with respect to emotions, some lint-combing might be in order. At the beginning of Chapter 8, he proposes that one might derive a classification of emotions by close study of narratives. He gives one example, a paragraph describing what seems to be an actual marital quarrel. From this one narrative, he derives four types of anger: inhibited, righteous, and sullen anger, and hostility.

However, he provides only one narrative. In the rest of the chapter, he goes on to derive still another theoretical taxonomy for all the major emotions, seemingly forgetting his own suggestion about the use of narratives. He proposes many abstract concepts of emotions, but only one particular, the narrative about the marital quarrel.

One last problem. Although narratives are much better than thin air, they are still quite abstract, being verbal descriptions. To get to the real core of emotions, we will need verbatim recordings of discourse. Such records, whether audio only, or video, make available for study the verbal and non-verbal indicators of emotion.

The idea that concepts and theories need to be closely linked to concrete particulars is the central theme of what is called “grounded theory” as proposed by Glaser and Strauss (1967). A drawback of their work is that it isn’t sufficiently explicit about specific steps in this direction. They imply that ethnographic work is usually a prerequisite to theory formulation, but without specifying much about the actual methods of getting from parts to wholes.

In particular, for the development of a concept, how many parts as compared to how many wholes? It doesn’t seem likely that there should be many more wholes than parts, as in Lazarus’s chapter 8. More likely, there should be many more parts than wholes, as is the case in the research to be described here: how the concepts of embarrassment/shame and attunement are generated in Goffman’s work. Using his approach as an example, it may be possible to be explicit about the steps needed in order to ground concepts and theories in concrete particulars.

Goffman’s Approach

In his Presentation of Self in Everyday Life (1959), Goffman seems to be developing, in a round about way, two concepts: embarrassment/shame, on the one hand, and attunement[5] (intersubjectivity), on the other. In this particular book, he provides many concrete examples of both ideas, but no definitions of either concept. However, in later work, he is clear that he is seeking to define both concepts. In the chapter on embarrassment in his 1967 book, he actually provides a conceptual definition, and at least a preliminary operational definition. In still another work, in 1969, and several other places, he seems to be attempting to define attunement (intersubjective states that Goffman calls “mutual awareness.”)

Although he casually uses metaphors (such as “mystic union”) many times to refer to speakers who are talking to each other, he also offers a fairly elaborate and complex definition of “being in a state of talk.”  Since his definition requires an entire page of text, I will not repeat it all here. Suffice it to say that it contains phrases that imply mutual awareness, and awareness of that awareness: “…An understanding will prevail [among the speakers] as to how long and how frequently each speaker is to hold the floor…” (1967, 35; a similar formulation occurs earlier, on p. 34). The definition that comes closest to explicitly defining attunement also comes in this line:

 “…A single focus of thought and attention, and a single flow of talk, tends to be maintained and to be legitimated as officially representative of the encounter (Goffman 1967, 34, emphasis added).”

 The phrase “a single focus of thought and attention” implies not only that the speakers are attending to the same thing, but they are having the same thought about it. Its significance becomes more apparent if it is compared to a similar phrase, “joint attention” used by the psychologist Bruner (1983), when he is explaining how an infant learns to become attuned with its caretaker. The mother, he says, is only trying to teach a new word.  She places an object (such as a doll) in her own and the baby’s line of gaze, shakes it to make sure of the baby’s attention, saying “See the pretty DOLLY.” In this situation, the baby is likely to learn not only the meaning of a word, but also, since both parties are looking at the same object, how to have, jointly with the mother “a single focus of thought and attention”, to use Goffman’s phrase.

A conceptual definition of intersubjectivity is as far as Goffman goes in attempting to explicate this idea; he didn’t provide objective indicators. Perhaps Goffman was uncomfortable about the implications of flatly stating and following up an idea that is anathema in individualistic modern societies, that we are all “members one of another.” Although church members recite this idea every Sunday, most would be loath to take its meaning literally, as Goffman did.

But with the other interior strand of Goffman’s work, embarrassment, he was not content to give only a conceptual definition, but also followed up, offering elements of an operational definition:

An individual may recognize extreme embarrassment in others and even in himself by the objective signs of emotional disturbance: blushing, fumbling, stuttering, an unusually low- or high-pitched voice, quavering speech or breaking of the voice, sweating, blanching, blinking, tremor of the hand, hesitating or vacillating movement, absentmindedness, and malapropisms. As Mark Baldwin remarked about shyness, there may be "a lowering of the eyes, bowing of the head, putting of hands behind the back, nervous fingering of the clothing or twisting of the fingers together, and stammering, with some incoherence of idea as expressed in speech." There are also symptoms of a subjective kind: constriction of the diaphragm, a feeling of wobbliness, consciousness of strained and unnatural gestures, a dazed sensation, dryness of the mouth, and tenseness of the muscles. In cases of mild discomfiture, these visible and invisible flusterings occur but in less perceptible form (Goffman 1967, emphasis added).

This definition links an interior emotion with surface observables.  With uncanny instinct, in the last sentence he hints at the need for further elaboration of the operational definition: “these visible and invisible flusterings [that accompany embarrassment], but in less perceptible form.” This clause seems to point toward the development of more elaborate coding systems for the vernacular verbal and gestural indicators of shame and embarrassment, such as the one by Retzinger (1991; 1995).

A weakness in Goffman’s method is that in some cases he resorts to hypothetical situations. Although they help the argument along, hypotheticals have many serious drawbacks in science. The most glaring one is that they always lack the ambient details, the minute, seemingly irrelevant particulars that often provide the key.  Elias’s work on shame, whatever else its deficiencies (as discussed below), at least used only verbatim material as data.

Yet Goffman’s attempt at defining embarrassment is extraordinary in the context of the then contemporary social science. The few social science theorists who emphasized emotions seldom defined them, even conceptually[6]. An example would be Elias’s masterwork, The Civilizing Process (1939). His proposition that the threshold for shame is advanced in the civilizing process is the central thread of the entire work. In a later work of Elias’s, The Germans (1996), shame is again frequently evoked, though not explicitly as in the earlier study.

Elias offered no definition of shame in either book, seeming to assume that the reader would understand the concept of shame in the same way that he did. The absence of any definition of shame and a systematic way of identifying it is particularly glaring in The Civilizing Process.  This study entailed an extensive and detailed analysis of verbatim excerpts from advice and etiquette manuals in five languages over six centuries. The analysis of the excerpts is completely intuitive, and in most cases, highly inferential. That is, the word shame is sometimes used in the excerpts that he selected, but much more frequently it is not.

Elias relied on intuitive and unexplicated interpretations of what Retzinger (1995) would call cue words, in context. Even if his interpretations were fairly accurate, which they might be, he still gave little direction to future research on the subject. Unlike Elias and most other analysts of emotion, Goffman took at least the initial step toward overcoming this problem. Like Muriel Spark, he was a lint comber[7]


This paper has proposed that a new taxonomy of emotion names is needed, and a way that we might go about generating such a scheme. Perhaps the first step would be the forming of a group dedicated to creating a new taxonomy, since it would be too much for one or even two persons to handle. The second step, following Goffman’s example, would be for each for each of the parties to amass and analyze a large number of verbatim examples that might involve the particular emotion that each party is to investigate. As a result of their separate investigations, each party might contribute a trial emotion concept to the group. Working as a group, each concept could be located with respect to every other concept, resulting in a new taxonomy. The next step would be to use the new emotion concepts to generate a taxonomy of vernacular emotion words, so that we could understand how emotions are conveyed in ordinary language.


Bruner, Jerome. 1983. Child’s Talk. New York: Norton.

Elias, Norbert. 1978, 1982, 1983. The Civilizing Process: V. 1-3. New York: Pantheon.

Glaser, Barney, and Anselm Strauss. 1967. The Discovery of Grounded Theory. Chicago: Aldine

Goffman, Erving. 1959. Presentation of Self in Everyday Life. New York: Anchor.

_______________l967. Interaction Ritual. New York: Anchor.

  _____________ 1969. Strategic Interaction. Philadelphia: U. of Penn. Press.

Harre, R. and G. Parrott. 1996.  The emotions: social, cultural and biological dimensions. London: Sage

Izard, Carroll. 1977. Human Emotions. New York: Plenum

Izard, Carroll, K. A. King, and C. J. Trentacosta. 2004. Lawful Processes in Naturally Occurring Emotions. Social Science Information 43 (4): 599-608.

Koestler, Arthur. 1959. The Sleepwalkers. New York: Grosset and Dunlap.

Lazarus, Richard. 1997. Stress and Emotion. London: Free Association Press.

Lewis, Thomas, Fari Amini, and Richard Lannon. 2000.  A General Theory of Love. New York: Random House.

Oatley, Keith. 2005. “Words and Emotions: Shakespeare, Chekhov, and the Structuring of Relationships.” Journal of General Psychology. 

Ortony, Andrew, Gerald Clore, and Allan Collins. 1988. The Cognitive Structure of Emotions. New York: Cambridge University Press.

Plutchick, Robert. 2003. Emotions and Life. Washington, D.C.:  American Psychological Association.

Quine, William. 1979. A Postscript on Metaphor. In S. Sacks (Editor), On Metaphor. Chicago: University of Chicago Press.

Retzinger, Suzanne. 1991. Violent Emotions. Newbury Park: Sage.

        1995. Identifying Shame and Anger in Discourse. American Behavioral Scientist 38: 541-559)

Shaver, Philip,  Wu, S., & Schwartz, J. C. (1992). Cross-cultural similarities and differences in emotion and its representation: A prototype approach. In M. S. Clark (Eds.), Emotion (pp. 175-212). Newbury Park: Sage.

Solomon, Robert. 1981 Love : emotion, myth, and metaphor. Garden City, N.Y. : Anchor Press/Doubleday.

   1992. About Love: Re-inventing Romance for our Times. Lanham, Md.: Littlefield Adams.

Stern, Daniel. 1977. The First Relationship. Cambridge: Harvard University Press.

  Tomkins, Silvan. 1962; 1963; 1965; 1992. Affect/Imagery/Consciousness. New York: Springer.

  Volkan, V. D. 2004. Blind Trust: Large Groups and Their Leaders in Times of Crisis and Terror. Charlottesville, Virginia: Pitchstone Publishing.


_________1997. Bloodlines: from ethnic pride to ethnic terrorism. New York: Farrar, Straus and Giroux.


___________1988.  The need to have enemies and allies: from clinical practice to international relationships. Northvale, N.J.: J. Aronson, Inc.

May 23  05  4559       atax

[1] The basic idea of this paper was first presented in Bengt Starrin’s seminar on emotions at the U. of Karlstadt. I am indebted to him for his encouragement in general, and in this paper in particular, his calling to my attention the chapter by Richard Lazarus to be discussed below.

[2] However, after noting the many narrow meanings of love, Solomon goes on to propose an interesting and important cognitive/emotional definition, but one so narrow that it leaves out a physical component of love, attachment.

[3] For a discussion of emotion words as metaphors, see Oatley  2005.

[4] My discussion of the discovery of the orbit of Venus follows Koestler  1959.

[5] Attunement is Stern’s (1977) term for intersubjective accord, (in Goffman’s language, “mutual awareness.”) This is an awkard idea in Western cultures, because they visualize the individual as an uncombinable unit. Attunement implies that momentarily at least, two or more persons may become united in thoughts and feelings.

[6] However, in her extensive treatment of shame, Lynd (1968) also took a step toward explicit definition. In a rare miss, Goffman did not reference Lynd’s work. 

[7] I realized in writing this paper that it was Goffman himself who first told me the story about Muriel Spark. For some reason long departed from my memory, I was at his house in Berkeley. As I was admiring his very large library, I asked him if he got the examples he used in his writing from books. Rather than answer, he told me the story about Muriel Spark.