Cultural Frame Analysis

The Kinship Map: Image, Idea, and Organization

We are concerned here with finding and analyzing ideas of a certain kind. These are the ideas  people use to form their social organizations.   Anthropologists since the beginning of the discipline have recognized that people do this. They form organizations using ideas of what those organizations are. But they have focused much more on the results of using the ideas than the properties of the ideas  that make the results possible.   In fact,  these ideas have a specific type of organization.  They form systems.  The systems have formal properties. And the results are the direct consequence of these formal properties.  Many other idea systems exist with different formal properties. They produce different results, but cannot produce these results.

It may seem strange to say so, but one of the properties is of these ideas is that they have names. This has caused considerable confusion among a line of anthropologists who have argued that a name only has meaning by referring to a physical thing and the meaning is nothing other than the physical thing that it refers to.  The names in the formal idea-systems that are used to produce social organizations have a different function.  They are not names for material things; they are abstract tokens.  Their use is to provide anchors or attachments for definitions. They are, therefore very much like the concepts of mathematical systems.  For example, the word “triangle” is a name for a figure in geometry.   The figure is a concept, not a physical thing.  The concept is a definition, which is made up of other concepts with other names.  We learn how to make one concept clear by learning how to make all of the related conceptions clear, to understand the whole as a system.  The way we learn such systems recognizes their formal structure.  We learn them in a classroom as part of a deductive system. We learn it has a forma structure: premises, corrolaries, branches, and rules of deduction. When do so, we also learn that physical things may approximate the concept but can never match it exactly. That does not affect the truth of the concept, or the need to adhere precisely to definitions that make it up. The word “father” in the English kinship system is the name for an idea in the same way. It is part of an idea system just as the idea of a triangle  is.   Only the subject matter differs. The figures of geometry are what we use to objectify physical space.  People have been using it this way for about two thousand years.  The concepts of kinship relations are what we use to objectify kinship space in exactly the same way. As far as we can tell, people (our species, Homo sapiens) have been using them this way for over thirty thousand years.

In a functioning society,  very few people have to master the concepts that define their physical spaces. But practically everybody  above the age of about 11 has to master the concepts that define their social spaces.   In that sense, these ideas systems must be present in almost universal consensus in every community.

When we say that an idea-system is present in practically universal consensus, we do not mean that it is fully formed in the imagination of every person in that community, such that if we ask people to say what it is individually, they will all say exactly the same thing. Nor do we mean that a high proportion of those in the community “believe” it in some sense.  Matters of opinion may be interesting, but we are after something much more important.   We are after systems of interrelated ideas that people  in a community use in forming their organizations, and that are present in  the widest possible consensus.   People impose them on themselves and each other with moral force.  They are what one must act upon, or assume, in social action.   The male child of a father is not a son as a matter of opinion, he is a son. An executive officer in a military unit is not subordinate to the commanding officer as a matter of opinion, he is subordinate to the commanding officer. It is enforceable, and it is enforced.   Like “father” and “commanding officer,” they include the names of kinds of organizations and the names of positions within such organizations.

Anthropologists have greatly underestimated the extent to which these  systems are formal, coherent, and complete, just in the way that the idea-systems of geometry and algebra are, and the way theories in chemistry and biology are. A major reason why they have underestimated this is that they have not paid attention to the problem of eliciting such systems of ideas in a clean and systematic manner. The present approach to kinship is an application of a social theory focused on recognizing the logical properties of these ideas, and a general method for eliciting them. The theory is given elsewhere (Leaf 2009, Leaf and Read 2012).  So is a published description of the method (Leaf 2006, 2009; Leaf and Read 2012).    We focus here on demonstrating the method rather than on describing it.

The Kinship Map

Lewis Henry Morgan’s Systems of Consanguinity and Affinity in the Human Family (1877) established the importance of kinship terms for investigating kinship organizations. It also set an assumption for conducting such investigations that continues to mislead. This was once summarized by David Schneider as “a kin term is a term for a kinsman.”  On this assumption, anthropologists since Morgan have tried to figure out what kin terms mean by figuring out what kinship is as something apart from the terms themselves, that the terms are names for. This has focused on reproductive relationships, biology,  and genealogy.   In the late 1960s studies of kinship on this basis, reinforced with arguments and claims drawn from philosophical Positivism, reached an apogee in componential analysis and the “structuralism” associated with Claude Lévi-Strauss. Then, between 1987 and about 2003, it collapsed.  When this line of analysis began, it conveniently failed consider its starting assumptions carefully. In the end, this failure caught up with it.

Saying a kin term is a term for a kinsman is not absolutely wrong, but it is powerfully wrongheaded.  It is wrongheaded  in precisely the same way that it is wrongheaded to say that geometry is a bunch of terms for shapes, and then trying to set out independently to find what the shapes were.  In fact, we learn geometry as a deductive system, grounded in  a specific set of  theorems and corollaries and rules to use them in making deductions.  What anthropologists have investigated under the misleading label of “kin terms” or “kinship terminology” is precisely the same, although they have not been able to see it because of this misleading initial assumption. The only differences that it produces different kinds of shapes in different kinds of space. Instead of producing geometrical shapes in geometrical space, it produces kinship positions and relations in kinship space.

Distinctive terminologies

All generative ideas systems  are associated with distinctive vocabularies.  This marks them off from less well-organized ideas of ordinary life as well as from each other. The terms name types of entities and relations among them. They may also name rules of computation or association for extending these relations.  So if we were doing and anthropological description of the United States we could identify “government terminologies,” “economic terminologies,” religious terminologies, military terminologies, and so on. But we do not do so, because this is our society and we know better.  It never occurs to us to ask why their processes are not named uniformly. It is easy to see the answer: it would cause confusion. It would make it difficult to know which kind of organizations being spoken of.   It also never occurs to us to think that it is the names that are important, rather than the ideas that are brought together in defining them.

The same applies to the organizations we find in the kind of small-scale, “traditional,” societies that anthropologists have historically dealt with. And it also applies to the types of organizations that are common to both:  kinship, community government, adjudication in a general sense, and many kinds of occupational organizations.

 Once we have identified a type of organization, with the distinctive vocabulary associated with it, it is usually not difficult to find some terms that are given much more emphasis in conversations than others.  When this happens it is not simply the linguistic unit in the sense of the sound or phrase that is being emphasized but rather the idea that it represents, and when we find several such often emphasized ideas we can usually find that they are interconnected.   This is where we start to search for the generative premises the tie all the other  apparently related concepts together.

Generative Premises

Since these idea systems are used to generate public  understandings through public communication, their generative premises must also be public. In fact, they are enacted in ceremonies and other public performances where the key terms are usually both explained and demonstrated.  And they appear in many forms in ordinary communication.

In speech, the premises take the form of readily envisioned images and metaphors, but the most important ones are normally quite abstract, or minimal.  This is because they function more as conceptual hooks that one can hang additional ideas on than detailed and complete pictures in their own right.  For government in the United States the idea of “three branches” is a familiar example.   Everyone agrees on the phrase, but what it means is subject to constant discussion.  In the American economy, the imagery of a market as a place where people buy and sell things or money, is equally common, widely used, and equally open to interpretation once you start to think about it.  This combination of being firmly and widely held while being minimally agreed upon, this minimality, is an important characteristic of such ideas.

For kinship, the primary generative system is the kinship map, as already noted.  This defines all of a person’s possible relatives, arranged around a central self or “I.”  In some systems, the there is one such map for males and another for females.  In others, gender of the user does not matter. Figures 1 through 5 are kinship maps for English, Punjabi, Tamil Male Speaker, Tamil Female Speaker, and Hausa. All are from field elicitations using the method demonstrated here.  The method involves building the diagram with your informants, drawing it out in such a way that they can see and understand what you are doing.  This usually involves a lot of erasing and redrawing.  So they have to be drawn more neatly for publication.  Figures 1 through 4 are the final neat drawings.  Figure 5, for Hausa, is a scan of a page in the original field note-book.  It is a cleaned up version made after the initial interview to check with other Hausa speakers.


Figure 1 – English Kinship Map


Figure 2: Punjabi Kinship Map
Figure 2: Punjabi Kinship Map


Figure 3: Tamil Kinship Map Male Speaker
Figure 3: Tamil Kinship Map Male Speaker


Figure 4: Tamil Kinship Map Female Speaker
Figure 4: Tamil Kinship Map Female Speaker


Figure 5: Hausa Kinship Map


Kinship Map differences

These contain much too much information to be digested at a first look, but a few key features should be noticed.  First, they all have different shapes. English is like half of a pine tree, Punjabi is like a butterfly, Tamil is a butterfly whose antennae and feet come together at the top and bottom.  Hausa is a pole.

The shapes come from the number of nodes and the ways they are interconnected.  Since the characteristics of the nodes (male, female, or either) and the ways they are connected are their definitions, each in relation to the others, this means that the different shapes mean that each map defines their relations differently.  And this means in turn that each map also represents a different idea of kinship in general.  These differences can be very far-reaching. For example, if you look at the rules for English, you see that it has provisions for extending one’s relations upward and downward, which means into the past and into the future, indefinitely.  There is no end. Any ancestor is a relation; any descendant is a relation. But for Punjabi, the top and bottom positions are endpoints. The father (bap) of the top ancestors on the mother’s side (parnana/parnani) is not a relative; the father (bap) of the top ancestors on the father’s side  (nakardada/nakardadi) is not a relative. So in English the idea of a relative is very closely related to the idea of descent, or genealogy–which is a cultural bias Western European anthropologists have unconsciously tried to impose on everyone else.  For Punjabi, a relative is someone you have a relationship with; interaction is what counts, not ancestry.

Kinship Map Similarities

But along with these differences are fundamentally important similarities. First of all, they all have a central “self.”  A semi-thoughtful critic might say this is only an artifact of the elicitation method.  We looked for a self and started the elicitation from there. So we have imposed it.  But in fact we have not.  We looked for a self and the concept is clear enough that we could have been told there wasn’t one. But we were told there was one.  And when we began the elicitation from there we got answers and those giving the answers did so with great confidence and nearly perfect agreement. If we were not asking about something already in their cultural knowledge, this could not have happened.  There is a difference between a starting assumption that is arbitrary and one based on fact. This is based on fact, and acting on it provides a further opportunity to test it.

Around each self, there are positions that are immediately connected in three directions: up, down, and to the sides.  But the numbers vary.  These are the “direct” relations.  The positions on opposite sides of the self-position are reciprocals. That is, the reciprocal of father in English, one link up, is son–one link down.  The reciprocal of “son” in English, one link down, is “father”–one link up.  The reciprocal of “brother” in English–one link to the side–is “sister” –one link to the other side, and vice versa.  This also introduces the idea that some relations are self-reciprocal; the reciprocal of brother can also be brother, also one link to the side. Such nuances can be stated as rules.

Since every direct relation is also a self, they, too, have the same configuration of direct relations.  So we can ask what their direct relations are to the original self. What is father of father to you, what is mother of father to you, and so on around.  This is the method for eliciting the rest of the kinship map. The direct kin around self are the eliciting frame. Since this is culturally defined, this is a cultural frame analysis.

We keep applying the cultural frame analysis to all new positions until we reach the boundaries of the system. The boundaries are of two kinds only.  Either we encounter a rule for going on indefinitely, which we can record as a rule.  Or we ask about relation of a relation and we are told that they are not related to the original self.

Boundaries on the kinship map as a conceptual system is not the same thing as a limit on the number of people who can be kin. It is not unusual in the logic of kinship maps that any human being can ultimately be a relative, but the system of concepts will nevertheless have definite boundaries.  English speakers may say everyone is related if we go back to Adam.  Yet there is also no relation beyond cousin.  There are logical and formal reasons why generative systems must be bounded.  The simplest is that if they are not, there can be no way to restrict the logic of one system to a definite domain. Another is that a premise cannot be precise if the domain of conceptual objects it pertains to is not precise.

Kinship Maps in Social Thought

Once we see what kinship maps are, we can reflect on how they appear in consciousness, or what we mean when we say they are indigenous ideas. In most cases, the complete map has no direct indigenous counterpart.  We can find diagrams of the English kinship map in lawyer’s offices and reference books, but many societies lack lawyers and writing and have no similar system of pictorial representation.  Even when they do have writing, they often do not have a literature that includes kinship maps.  And even if they do have such a literature, as American’s do, most of the people in the community are not aware of it.  Nevertheless, as the kinship map emerges in the course of the elicitation, those providing the information have no trouble recognizing what it is, reading it, and suggesting corrections.  So it is in their thinking somehow.  But how?

While people rarely have in their imaginations a single image corresponding to their kinship map as a drawing, they do have it consciously in mind in two other forms. The first is as linguistic patterns of stems, affixes, and prefixes that usually rhyme in such ways as to suggest the visual pattern.  The second is that they have numerous metaphors and imagistic ways of speaking that suggest the overall shape or the shape of the generative core.

The linguistic imagery duplicating the drawn shape is easy to see (or hear) in figures 1 through 5. In English the line directly from one’s ancestors to oneself and one’s descendants is marked by the “father” -“mother” stem in ascending generations and “son” -“daughter” in descending generations;  lines branching off are marked by the “uncle” and “aunt” stems; and generational distance is marked by the prefix [none], “grand,” “great-grand,” and so on.   In Tamil and Punjabi the two wings have different stems and  the stems in each side rhyme with each other;  male and female terms are marked by male and female endings; and generations are again marked by distinctive and reciprocal prefixes.  In Hausa stems rhyme on each generation and contrast across generations; there are male and female endings; and distinct-sounding prefixes indicate the important distinction between direct and indirect relations.   So when a user of any of these kinship maps sees them drawn and written for the first time they recognize that the visual pattern which is new to them corresponds to the sound pattern that is familiar.

 This is supplemented by many other kinds of imagery that represent parts or aspects of the kinship map.   Again focusing on American English  as the most familiar example, one set of images is hierarchical.  People commonly speak as though they descend from a single ancestor, although in fact the number ancestors a person has doubles with each generation that they would count back.  That is, the spoken imagery corresponds to the kinship map imagery, and not what people should imagine if they were really thinking in biological terms.   Similarly, people readily speak of “lines” and “branches off” of these lines.   There is the idea of an “immediate family” with whom one occupies a household and a more extended family, such as you might invite to a wedding or a family reunion. And of course the idea of “reunion” suggests that at one time one unity that has become dispersed.  Americans also speak of relatives as being “close” or “distant,” and of “generations.”   People also speak about families as though they have “sides,”  as in “on his mother’s side,”  but if you press them on it the imagery is not like Punjabi or Tamil but rather  still the English hierarchy of the mother or father in question– or husband or wife.

We should also probably include the “nuclear family” as an image of this type: two parents and their children. It may have begun as a sociological generalization but it is now widely used in stories, films, public media, and the like. It is often taken as what a family stereotypically “should be.”  The interesting thing about this is that very few families in a statistical sense really are of this type. It seems to be normative, but it is not statistically normal.  So where does the idea come from?   What makes it plausible? The obvious answer is that it corresponds to the generating core, the configuration of direct kin. Familiar and natural as such images may seem to Americans, none of them are universal and most of them would seem quite odd to users of different idea systems.

The first step in starting any elicitation of a kinship map is to identify its generative core. Of course you should ask for direct kin, once you can figure out how to do this in the indigenous language.  And in order to figure this out, it is very good practice to have listened for metaphors of this kind.  Then, as the elicitation proceeds, you should be able to see where they fit in.

Kinship Maps and “Norms”

Anthropologists and  sociologists  often speak of kinship positions as involving “norms” for behavior, rules that this or that relative should follow. But they have had very little success in specifying what these norms are.  It is easy to point to a specific role for a specific relation in one ceremony or context, but equally easy to find a seemingly contradictory “norm” in some other ceremony or context that the same person might also participate in.   It is also difficult to know what a norm is.  Is the common American stereotype of a mother-in-law a norm?  Is it a norm for wives to be overbearing, or should they be supportive?  Is divorce a norm?  What is involved in being “fatherly” or “avuncular?” When  we see what kinship maps are, we can see how this this lack of specification comes about.  Detailed prescriptions of behaviors or powers is not needed.  The positions are  well-defined in entirely formal terms. The most definite definition of each position is nothing other than the other positions that it is attached to.   Otherwise, it is undefined.  The result is that it can establish a relationship between two people with absolute clarity but still leave the behavioral content of that relationship up to them or others to fill in as might be needed. And when we consider that kinship relations are usually defined as permanent  or at least long-term, it is easy to see how such  A combination of extreme stability with extreme flexibility would be  extremely useful.

The Videos

Three videos are included. Each simulates the process of eliciting one of the kinship maps whose final version is given above: English, Punjabi, and Hausa.  They are not videos of the interview itself.  None were made and it would not have been be a good idea to try.  It would be distracting.  But if they had been made all they would show would be Leaf  talking with different groups of people.  This would probably not be very instructive. They are videos representing the drawings he made in the process. They show what you see as an analyst building up your diagram in the course and elicitation. It shows how the elicitation starts with the direct kin and a clear understanding of what the elements are in representing them. It shows how this initial core is used as the frame for constructing the rest. And it shows how you know when you have reached the edge or boundary and what to do about it.

 In the field elicitation, the diagram was drawn with a pencil and eraser on paper in the indigenous language in full view of those who are answering the questions. It was frequently redrawn as the elicitation proceeded.   The basic principle was to have one symbol representing a position for one name.  This is not possible in all cases  in a single diagram on a flat piece of paper. Sometime the logic of the definitions would be better represented in three dimensions, with overlays.  Sometime it is thought of as a cylinder, sometime as a central sphere or column with outward rings, and so on.  But it should always be clear that each name corresponds to a single idea,  meaning that it has a single coherent definition.

The term “kinship map” actually came out of this process in the first elicitation, of Punjabi, in 1964.   Leaf met with the largest family in the village.  They were well regarded and well-respected–so not likely to hold bizarre or strange views of something so basic.   Members of the family made up five different households.   They cooperated with each other economically.   A few of those present also spoke English, so the process did not rely only on Leaf’s knowledge of Punjabi. When Leaf began, several in the group asked what he was trying to do.  The only Punjabi word he could think of that conveyed the idea was naksha, meaning map. So he said that he wanted to ask if they could give him a map of Punjabi kin relations, Panjabi risten da naksha.   They agreed to do so.  Punjabi was used for all the positions and relations.  Everything was written in the Gurumukhi  script, which most of those present could read.   No English was included, no English translations were asked for or given.   In the end, everybody agreed that this was indeed a risten da naksha, and there has been no reason since then to call it anything else.

The videos here also use only symbols that were agreed on to represent indigenous ideas and indigenous terms, but the script is Roman.    As near as memory permits, Leaf describes the questions he asked, the answers he was given, and how this information was incorporated in the diagram as it was constructed.   The format is mp4.  Since this is done on an iPad using Adobe Sketch and an Apple pencil, the space is much more restricted than in the field situation, and the stylus is not as sharp or easy to control as a lead pencil.   There are a few places where Leaf  loses focus and does not write down what he is saying, although this is later corrected. Try to watch for them.  Such lapses and corrections are almost unavoidable.  You cannot restart from the beginning every time you make a mistake.  The original interviews took substantially longer than the videos.  Despite what you may be told in methods classes,  however, anthropological interviews that take one to four hours to complete are not problematic as long as the people whose time you are asking for understand and agree with the importance of what you are asking for and consider it something they want you to know.  An anthropologist in the field is a guest and a student, not a boss, an authority, or a master.