C. George Boeree
Shippensburg University

        Structural analysis is essentially the search for meanings that have been "embedded" in products or artifacts  --  e.g. verbal expressions, literature, social structures, cultural rituals, artwork, technologies, techniques  --  by others.

        In phenomenology "proper," we are describing what's going on in someone's experiencing  --  our's or someone else's.  At most, we ask for protocols, which we normally check with follow-up interviews.  But these things are not always possible:

        1.  The person or people may not be around, so all you have is their records (literature, autobiographies, diaries, art, music, crafts...).

        2.  Protocols, interviews, or dialog assume a degree of shared understanding to begin with, a certain "research sophistication" one cannot always assume.

        3.  Much of noeses is unconscious.  Phenomenologists agree with Freud that the unconscious can be made conscious; but Freud also pointed out that much that's unconscious "wants" to stay that way.

        Each of these problems is regularly faced by students in several fields:  Problem one is faced in art and literary criticism, cultural archeology, and psychohistory; problem two is faced in cultural anthropology and sociology, and in cross-cultural psychology; problem three is faced in linguistics, cognitive development, and psychoanalysis.

        The most common method for dealing with these situations is structural analysis:  You look at a set of phenomena and attempt to dis-cover the meaning structures on the "far side" of the phenomena, i.e. the meaning structures of others.

Here's how the structuralist Roland Barthes explains it (in Lane, 1970):

        Taken like this, we are really just describing science.   Theory-building is the construction of a simulacrum or model.  Structuralists are simply focusing on meaning structures, rather than physical ones.

        You might, for example, create a computer model of human problem-solving, which imitates the "visible" aspects of that ability.  We know how our computer works and suggest that the human mind works like our model.  We do not suggest that the human mind is "nothing but" a wet computer -- this is the fallacy of reductionism, or the fallacy of "nothing-but-ism," as I prefer to call it.  This is mistaking the map for the territory.

        A good way to get comfortable with structural analysis is to play certain games  --  for example,  Mastermind.  One player chooses four marbles from a set with six colors, duplication of color permitted, and arranges them in a particular order.  The other player attempts to figure out the order and colors from clues given by the first player in response to his previous guess:  A white peg says that you have guessed one correct color in its correct location; a black peg says that you have guessed a correct color, but the position is wrong.  Four white pegs, therefore, would indicate that you have guessed all four marbles correctly.  The game can be nicely played by substituting A, B, C, D, E, and F for the colored marbles, and * and +  for the white and black pegs, respectively.

        There are two things I would like you to get out of this and the next couple of exercises:  First, I want you to develop habits of patient, even pain-staking, analysis.  Most of us are so used to doing things as quickly and (supposedly) efficiently, that we have lost these habits.  Structural methods -- and qualitative methods generally -- require patience.  Think of them as something akin to working with a teaspoon and a paint brush at an archeological dig!  Second, I want you to develop a feel for patterns.  With practice, patterns (structures, essences...) will begin to jump out at you, like finally seeing the hidden image in a Magic Eye picture.  You may find yourself amazed at how perceptual these supposedly cognitive things can be!

        (The old board game Clue is probably the best known game of this type.  Another game is called Nature, and works like this:  One player is "nature" and lists on a piece of paper one, two, three or more rules for discarding cards  --  e.g. if the last card discarded was black, the next must be red  --  making sure that the rules will permit the discarding of all cards at some point.  The other players are dealt equal numbers of cards and are required to guess the rules by discarding and having that discard accepted or rejected.  First player to get rid of all his or her cards wins and becomes the next "nature.")


        Let's get right into a structural analysis.  Get into your "small group."  First, describe the structure of tic tac toe, including the discriminations a player must be able to make and the rules of play.  Then describe the structure of the strategy of tic tac toe, i.e. what you must do to win!

        Please note that there are many things in life that have the same basic structure as games.  Social rituals, for example, have things you must recognize and rules to follow.  Yet they too permit one to �play� within their basic structure.  Some of us are good at the game, some of us are competitive, some are cooperative, some play badly, and some don�t want to play at all!  Some social scientists believe that all human interaction can be understood as games, tactics, and strategies.

        Alternative exercises:  Do an analysis of the rules or the strategy of any game or sport.  Or do an analysis of the rules of grammar for some language, or the rules of phonetics or spelling.


        The two areas making most use of structural analysis are anthropology and linguistics.  In anthropology, it is most clearly found in a part of ethnography (the description of cultures) called  ethnoscience,  which looks to discover how a different culture organizes its understanding of some domain -- botany, zoology, musicology, and so on.  But an older area than ethnoscience is the analysis of kinship systems by means of a technique called  componential analysis.

        Let's warm up by looking at English kinship:  The basic terms are mother (M), father (F), sister (S), brother (B), son (s), daughter (d), wife, husband, aunt, uncle, and cousin.  Qualifying words, affixes, and phrases include "grand-," "great," "first," "second," etc., and "once-removed," "twice-removed," etc.  We won't be concerned here with "-in-law," "step-," "half," and so on.  The first step is, in fact, to list as many terms as possible.

        The next step is to have an informant (in this case, yourself) define these terms for you, occasionally probing for detail and variations:
        Mother is a female parent.
        Father is a male parent.
        An uncle is a brother of your father or mother, or the husband of an aunt.
        A great grandfather is the father of the father or mother of your father or mother.

        ...and so on.  Another approach is to collect examples of each specific term: A cousin includes the children and other descendants of uncles and aunts, the children and other descendants of great uncles and aunts, etc.

        The next step is to organize the terms on the basis of similarities of or patterns in their definitions:

        Mother and father "belong together."
        Mother and daughter should be close together -- they are "opposites."
        Cousins are "the most different" from the more "nuclear" terms.

        ...and so on.  The point is to discover the basic components of meaning in the kinship system.  For example:

        Mother and father are both parents, female and male respectively.
        Uncles and aunts, like brothers and sisters, are siblings (or the husbands or wives of siblings).
        Parents, grandparents, great grandparents, etc. are ancestors at various distances.

        When you have your components, you go back and redefine your kinship terms in terms of these components:
        Mother is a female parent.
        An uncle is a male sibling of a parent, or the spouse of a female sibling of a parent.
        A cousin can be specified in terms of first, second, and so on, by counting how many generations back from oneself or from the cousin, whichever number is larger, you have to go to find the siblings that tie you together; and a cousin can be specified in terms of once-removed, twice-removed, and so on, by subtracting the shortest of those two distances from the largest.  (Check it out, or check out Geoffrey Leech's discussion of English kinship in his book  Semantics,  1974.)

        But English kinship is too easy for you.  This project involves the analysis of the kinship system of the Seneca Iroquois Indians, following one of the earliest formal componential analyses by Floyd Lounsbury (1964).  On the next page is a partial list of kinship terms, with very approximate English translations, and with a list of specific relatives that one would call by each kinship term.  Note that "MFSd," for example, means "mother's father's sister's daughter."

        To be precise, the task is this:  If I were to give you the details of the relationship between you and 'alter' (that is, a relative of yours, such as MMBdsd -- your mother's mother's brother's daughter's son's daughter), what rules would you need to know to determine the proper term of relationship, so that you'd never need a chart again!

        Here are the steps to take in a structural analysis:

        1.  List the specific instances.  In this case, I'm giving you the list.  "In the field," you would have had to collect them by asking many people many questions.

        2.  Organize the specific instances.  Lounsbury has made it easier for you by arranging and dividing the terms into revealing patterns.

        3.  Deduce the rules from the organization.  The better you've done the previous step, the easier this one will be.

        4.  Go back and carefully test your rules with some "free variation."

        Divide into small groups and give it your best shot.



1. ha'nih  'my father' -- F, FB, FMSs, FFBs, FMBs, FFSs, FFFBss, etc.

2.  no'yêh  'my mother'  -- M, MS, MMSd, MFBd, MMBd, MFSd, MMMSdd, etc.

3.  hakhno'sêh  'my uncle' -- MB, MMSs, MFBs, MMBs, MFSs, MMMSds, etc.

4.  ake:hak  'my aunt' -- FS, FMSd, FFBd, FMBd, FFSd, FFFBsd, etc.

5.  hahtsi'  'my elder brother' -- B, MSs, FBs, MMSds, FFBss, MFBds, FMSss, MMBds, etc. (when older than ego)

6.  he'kê:' 'my younger brother' -- (same, when younger than ego)

7.  ahtsi' 'my elder sister' -- S, MSd, FBd, MMSdd, FFBsd, MFBdd, FMSsd, MMBdd, etc. (when older than ego)

8.  khe'kê:'  'my younger sister -- (same, when younger than ego)

9.  akyä:'se:'  'my cousin' -- MBs, FSs, MMSss, FFBds, MFBss, FMSds, MMBss, etc.  also:  MBd, FSd, MMSsd, FFBdd, MFBsd, FMSdd, MMBsd, etc.

10.  he:awak  'my son' -- (a) s, Bs, MSss, FBss, MBss, FSss, MMSdss, etc. for male ego

-- (b) s, Ss, MSds, FBds, MBds, FSds, MMSdds, etc. for female ego

11.  khe:awak  'my daughter' -- (a) d, Bd, MSsd, FBsd, MBsd, FSsd, MMSdsd, etc. for male ego

-- (b) d, Sd, MSdd, FBdd, MBdd, FSdd, MMSddd, etc. for female ego

12.  heyê:wô:tê'  'my nephew' -- Ss, MSds, FBds, MBds, FSds, MMSdds, etc. for male ego

13.  hehsô'neh  'my nephew' -- Bs, MSss, FBss, MBss, FSss, MMSdss, etc. for female ego

14.  kheyê:wô:tê'  'my niece' -- Sd, MSdd, FBdd, MBdd, FSdd, MMSddd, etc. for male ego

15.  khehsô'neh  'my niece' -- Bd, MSsd, FBsd, MBsd, FSsd, MMSdsd, etc. for female ego


        Some students get rather bored at this point and ask the famous question "Why are we doing this?"

        First of all, kinship is the most basic system of social relationships there is.  Some would argue that all others are based on it (e.g.  Eric Berne, 1966).

        Secondly, we tend to see our own system as "the" system, and others as somehow "wierd."  I don't want you to be so egocentric!  And we have to break our tendencies to see the mind of the "primitive" as somehow childish or simple.  Any mind that can handle Australian Aboriginal kinship or the marriage rules of the Yanomamo Indians of Venezuela is as capable as any in our society.

        But the main reason is that I want you to learn to "think," that is, to perceive subtle patterns in complex information.  With practice at kinships analysis (etc.), you develop a "sense for structures."  You could go into an alien society and discover their kinship system and their language and their social, political, and economic systems.

        Are you interested in medicine?  You can go to a famous diagnostic physician -- one with "great intuition" -- and look at his diagnoses,  listen to him talking to himself, record his specific observations -- and, with structural analysis, write a program that reveals his intuitions to medical students, or might even provide other doctors with an instant "second opinion."

        Are you interested in industrial and organizational psychology?  You can look at a company's organization (both formal and informal), at its culture, communications systems, specific "unconscious" skills of its employees, at training programs, and go on to suggest changes in these and other aspects of the company's functioning.

        Are you interested in clinical psychology?  You can look at the communications within a family and reveal them to the family.  You can look at the pragmatics of interaction between client and therapist from a theory-free point of view. And so on.


        Our next project is an elicitation interview, that is, a conversation with someone with the purpose of getting data about their image of reality which we might then submit to structural analysis.  Until now, we have started with our lists ready made.  We have been cheating!

        For example, we might want to develop an understanding of someone's "ethnozoology" -- i.e. how they categorize animals.  Here's an imaginary dialog:

        R (researcher):  Tell me, would you, what is that (pointing to an animal)?
        I (informant):  That is a cat.
        R:  What kind of animal is a cat?
        I:  It is a meat-eater.
        R:  What other kinds of meat-eaters are there?
        I:  Oh, there's dogs, wolves, bears, rats, a few others.
        R:  Do some of these meat-eaters belong together?
        I:  Yes:  Dogs and wolves belong together.  So do rats and similar animals.  Bears are different.
        R:  Is there something you call both dogs and wolves?
        I:  No.
        R:  How about rats and the similar animals -- is there something you call that group of animals?
        I:  Yes, we call them vermin.
        R:  I see.  What kind of animals are meat-eaters?
        I:  I don't know what you mean:  They are just animals.
        R:  Are there other kinds of animals?
        I:  Sure.  There are vegetable-eaters.
        R:  What kinds of vegetable-eaters are there?
        I:  Well, there's the little ones and the hoofed ones.
        R:  Can you give me some examples of the little ones?
        I:  Yes:  There are rabbits, squirrels, chipmunks, woodchucks, and...that's all I can think of.
        R:  Do any of these belong together?
        I:  Yes, the squirrels and the chipmunks do -- they are tree-dwellers.
        R:  And the hoofed animals -- what are some examples of these?
        I:  Deer, Cattle, Horses, Moose, Elk....
        R:  Any of these belong together?
        I:  Yes, the deer, moose, and elk belong together.
        R:  Is there a name for them?
        I:  They are the antlered animals.
        R:  Tell me, what kind of animal is that (pointing to a bird)?
        I:  Ah, that's not an animal.  That's a spirit.

        There are actually only a few questions the researcher is asking:
        1.  What is that?
        2.  "Moving up" (What kind of x is that?)
        3.  "Moving down"  (What kinds of x are there?)
        4.  Clustering (Do any of these belong together, and what do you call these clusters?)

        There are other questions:
        1.  At the very top of a hierarchy, you might have to ask:  What else is there besides what we mentioned?  Our researcher here assumed that we would get to birds under animals, but the informant considered birds a whole different form of reality.
        2.  At the very bottom of the hierarchy, we might need to check for completeness by asking for full lists -- for example, of those other "vermin" the informant mentioned.

        And, if we are truly interested in how our informant construed his reality, we need to ask questions like:  What's the difference between x and y? (e.g. between animals and spirits, between vermin and other meat eaters, between deer and elk).  And how do you tell?

        Remember:  Your "anchors" in all of this are (1) specific words or phrases and (2) phenomenal referents -- i.e. specific experiences that can be shared between researcher and informant (this animal here, this particular feature, this structural characteristic, etc. -- things more easily shown than described).  If you are using more than one informant (as is smart, if you are trying to get to an  ethno-zoology, and not an idiosyncratic one), it is again these two -- terms and referents -- that form your anchors, this time in terms of common points-of-reference.

        In the example, we were eliciting a taxonomy.  You can do the exact same thing to elicit a rule system.  The precise questions change, that's all:  What's that?  What happens next?  What happens if...?  How do you know? and so on.

        The project for today is to find a partner and elicit each other's understandings of the varieties of modern popular music.  This is what ethnoscientists call an "ethnomusicology."





Return to top

Go to part three

Return to table of contents

Return to homepage

Questions or comments

Copyright 1998 by C. George Boeree