A cloud of claims and counter-claims, a smoke-screen of charges
and counter-charges, has descended over anthropological inquiry
in the past two decades. And the discipline has spawned a proliferation
of theoretical models, a morass of new methodology, until one's
vision of the goals of anthropological inquiry and the major issues
involved in our study of socio-cultural phenomena can become quite
obscured. Suddenly the spoils of the game go not to the ethnographer
who after years of toil under difficult circumstances provides a
description of a socio-cultural system in greater or lesser detail.
It is the age of the theoretician, and one wonders at times what
he is really theorizing about. Consequently, I thought it might
be useful to discuss here what I conceive to be some of the major
epistemological issues in the fields of social structuralism, opportunism,
cognitive structuralism, and synthetic structuralism. I am not going
to attempt an analysis of all the major issues, but instead, what
I shall consider are those issues that separate these fields. Even
so, to attempt such a coverage in a limited paper such as this would
either be conceived of as an act of anthropological arrogance or
naivety. As for the sins which I will commit here, I would like
to excuse them by claiming epistemological naivety. For I am not
going to try to present here a reconstructed logic of these methods
of inquiry as a proper epistemologist would do. I am not interested
in reconstructing a normative model, but, instead, I am interested
in the cognitive style (Kaplan 1964:3-33) of the social structuralist,
the opportunist, the cognitive structuralist and the synthetic structuralist,
or what Kaplan calls their Alogic-in-use@.
In addition to briefly analysing the issues that separate these
approaches to ethnographic reality, I shall also be concerned with
the nature of the models built and the explanations offered for
the data produced by the models. I will of necessity have to be
brief here, but before I go on I should like to make explicit my
own interest in these issues. I conceive of myself as a British
social structuralist inasmuch as I am primarily concerned with social
action. But I believe that there are certain deficiencies in the
theoretical model of social structuralism, and I am attempting to
deal with these deficiencies in my own research, as I shall enlarge
on later in this essay. Therefore, the attacks on the question-set
of social structuralism by other approaches to ethnographic reality
and their epistemological claims have been of concern to me in my
attempts at developing a more appropriate theoretical model. My
approach to this problem has thus been to make an ethnographic study
of the various theorists, and as social anthropologists will do,
I found that what people said ought to be done, what they said they
did do, and what they actually did, frequently were at variance
with one another. But more about this later.
In discussing structural theories and methodologies certain difficulties
arise because of the polysemic nature of the term 'structuralism'.
Structuralism in Piaget's (1970) and Boudon's (1971) terminology
certainly encompasses all the various methods of inquiry that we
will be discussing here. Therefore, it is necessary to make certain
distinctions so that our argument will not be confused over the
meanings of the term 'structuralism'. First of all, I am not going
to deal with the metatheory of structuralism, for Piaget and Boudon
have produced excellent analyses of this, and there are in addition
a number of other works available. Piaget's version of structuralism
might be termed 'operational structuralism', as he has himself indicated
(1970:8), and I will use this terminology if necessary to distinguish
the distinctive features of his approach. Piaget has also referred
to Lévi-Strauss's version as 'analytical structuralism' and
states that it is based on a deductive approach. But in my opinion
this fails to distinguish the major thrust of this theoretical position,
which is to answer the question as to what are the innate structures
and processes of the mind through which we can provide an explanation
for the surface structure of observable cultural phenomena. Therefore,
because so much of the work in this field deals with symbols and
concepts, I at first believed that it would be more appropriate
to refer to this version of structuralism as 'symbolic structuralism'
to distinguish it from other mentalistic approaches, such as 'cognitive
But some of the participants in the SUNY symposium at which I gave
an earlier version of this paper took the position that they also
dealt with symbols, but not in terms of French structuralism. I
then considered the term 'ideational structuralism', since one of
the features of this form of structuralism has been to focus on
archetypes in a Platonic sense. Yet this also did not seem to identify
the major features of this approach in contrast to other forms of
structuralism, and seemed to imply as well conscious mental activity,
while Lévi-Strauss has argued that his structures are unconscious.
Consequently, I have now decided, for lack of a better term, to
refer to the form of structuralism represented in the work of Lévi-Strauss
as 'synthetic structuralism' in that it shares two features with
the concept of 'synthesis': (1) the reasoning is frequently deductive
with an emphasis on the dialectic combination of thesis and antithesis
into a higher stage of synthesis; and (2) attempts are made to synthesize
various structures into one common mother structure.
Finally, as the term 'structuralism' is also sometimes used in
the frame of reference of British social anthropology (for example,
van Velson 1967; Mitchell 1969), I shall distinguish this form from
others by using the commonly accepted term 'social structuralism'.
Approaches to Ethnographic Reality
In the various approaches to ethnographic reality it seems to me
there are three basic issues in controversy: (1) the issue as to
whether social behaviour or mental behaviour constitutes the basic
datum of research; (2) the controversy over whether ethnographic
data are to be interpreted primarily in terms of their meaning to
a particular cultural system, that is, cultural-specific or system-specific
meaning, or whether they should be interpreted in terms of a theory
of behaviour held by the investigator, that is, in terms of theory-meaning;
(3) the controversy over whether the individual or the collectivity
is the locus of ethnographic reality. This last issue is connected
to the controversy over mental behaviour versus social behaviour,
but it is not isomorphic with it, as I shall attempt to demonstrate.
Also this distinction shares many features with the controversy
over methodological individualism versus methodological collectivism,
but again it is not completely isomorphic with it.
These issues may be plotted in a matrix as follows, with the various
fields of inquiry in anthropology entered in the resulting boxes
along the dimensions that most closely represent their approach
to a resolution of these issues (Figure 1).
Figure 1: Issues in Anthropological Theory
| Theory Meaning
|| Social Behavior
Unfortunately, in using such matrix displays, one has to force the
evidence into boxes where it may not in fact entirely belong. Any
inquiry is an ongoing, developing growth, so that there may be appearing
on the peripheries developments that differ from the central tendencies
of the inquiry. It would be impossible to cover all these peripheral
growths within the limits of this essay, and so I shall confine
myself to defining the ideal types, realizing fully the difficulties
in using these. Thus many may disagree with my conclusions. But
the evidence for my conclusions will be drawn primarily from what
the practitioners of these approaches say they are doing in published
materials, not what we may believe them to be doing, and certainly
not what they say they are doing in private conversations. And at
times I will contrast this with what they in fact do do.
However, this method tends to over-emphasize boundary features
rather than the continuum along which the various approaches fall.
Therefore, I have attempted to display these distinctions in a three
dimensional model along the axes of which the various approaches
can be plotted. This will not only allow a greater latitude for
describing central tendencies, but it also permits you to shift
the fields of inquiry along any of the dimensions to represent more
precisely your view of where they lie in terms of these basic issues
Figure 2: Issues in Anthropological Inquiry
This distinction between focusing on boundary features rather than
on continua, in phrasing relations in digital rather than in analogue
terms, is an important one in our analysis of the various approaches
to anthropological inquiry and will appear again later on in our
discussion of Lévi-Strauss’s method of analysis and
his use of linguistic models.
Various terms have been used to identify this field of inquiry.
The Oxford anthropologists (Ardener 1971; Needham 1971; Rivière
1971), as many others, tend to prefer the term ‘functionalism’.
The term ‘structural-functionalism’ has also been used,
as well as ‘structuralism’ (van Velson 1967; Mitchell
1969). I prefer to refer to this field of inquiry as social structuralism
to distinguish it from the various forms of mentalistic structuralism,
and this also permits any reference to functionalism to be deleted,
which I believe is useful since functionalism is not a necessary
feature of social structuralism. While functionalism was originally
part of the social structural approach, it is entirely possible
and desirable to provide a structural description of a social system
without reference to functional explanations (see Appell 1965).
Furthermore, by referring to this field as functionalism we identify
it by one variant of its method of explanation, although we do not
refer to other fields of inquiry by the types of explanation they
In indicating that social structuralism deals with social behaviour,
we do not mean to imply that social structuralists have not dealt
with cognitive aspects of social systems such as religion and belief
systems, and made important contributions. But, more frequently
than not, social structuralism has considered belief systems and
religions in final reference to the problem of social control or
how they articulate to social structure (see, for example, Firth
1967 and Nadel 1954).
But this does not dispel the difficulties in identifying social
structuralism with behaviour. We have here the problem of what the
natives say versus what they do. For example, social anthropologists
say they deal with behaviour. Mair (1965:9) refers to “the
structural order by means of which the behaviour of people is interpreted...”
But in fact, most realize that the social structuralists are dealing
with social action; that is, not simply with behaviour but also
with the aim, intention, and goals of behaviour, and these are part
of the cognitive organization of the actor (see Nadel 1953:30).
Thus while the cognitive style of social structuralism is ‘behavioural’,
the strength of social structuralism is that it incorporates both
social and cognitive behaviour. But this is also its weakness, for
social structuralism has not explicitly come to grips with cognitive
models and therefore they have not systematically dealt with the
problem of eliciting indigenous distinctions, particularly with
regard to social units. This is why I have entered this approach
along the theory-meaning dimension.
For example, Firth (1963) defines the kindred, cognatic stock,
and other concepts so that it is not clear whether the social units
he is concerned with are the entification of the anthropologist
or the members of the society being described. Radcliffe-Brown’s
approach to corporate descent groups is similarly confused (1950:41).
And Gulliver (1971:9) writes, contra Appell (1967), that “It
seems to be more efficacious to ...[include affines in the kindred]
in analysis, even where a people themselves for some purposes culturally
distinguish between the two kinds of kin.” (For other examples
see Appell 1973, 1974a, 1974b, n.d.).
Thus, social structuralism has never dealt systematically with
system-specific meaning, which eventually led to the position articulated
by Goodenough that in focusing on behaviour in the analysis of kin
groupings and in ignoring the cultural principles organizing this
behaviour, anthropologists failed to identify non-unilineal descent
groupings and instead forced some of the empirical evidence for
these into a unilineal mould (Goodenough 1955, 1961). Thus he concludes
(1964a:11-12) that an observer can perceive this kind of statistical
patterning, that is, the phenomenal order of observed events, in
a community without any knowledge whatever of the ideas, beliefs,
values, and principles of action of the community’s members,
that is, their ideational order. This position, while undoubtedly
based on some reality, does overstate the situation, since many
philosophers of science would argue that it is impossible to classify
human behaviour without some inference as to its intention, even
if it means, though, an imperfect knowledge of the actors’
Cognitive structuralism and methodological individualism
The controversy over methodological individualism and methodological
collectivism or socialism has had a long history (see, for example,
O’Neill 1973), and consequently I cannot discuss here the
various skirmishes in this debate. What I am essentially trying
to discriminate by this contrast might perhaps be better phrased
under the concept of social locus of ethnographic reality. Thus
I am asking here whether the fields of inquiry which we are considering
concern themselves with the individual as the real locus for ethnographic
investigation or whether the inquiry focuses on the collectivity,
the social grouping, the institution, or the social system.
Goodenough, for example, has taken the position that social anthropology
has focused on observed events, which are the property of the community
as a material system of the people, their surroundings, and their
behaviour. These are, furthermore, artefacts of the ideational order,
which is not a property of the community but of its members, existing
in their minds (Goodenouqh 1964a:11-12).
Thus, in cognitive structuralism the individual is conceived of
as being the locus of ethnographic reality. This stems from the
view that culture is a code and therefore amenable to analysis by
the same procedures used in structural linguistics with a Bloomfieldian
pedigree (see Goodenough 1951, 1957; Frake 1964). However, this
linguistic model in the beginning led the cognitive structuralists
astray. For, as Aberle has pointed out (1960:13), in structural
linguistics the assumption is made that “the idiolect is,
loosely speaking, isomorphic with the language”.
As a result, componential analysis as a tool of the cognitive
structuralists was claimed to represent the ‘real’ cognitive
world of the members of a society, until Burling (1964a) pointed
out that there were a number of logical possibilities for ordering
componential data. Thus, for the analyst to claim the cognitive
saliency of his particular method required that he introduce additional
methods to establish that his results were in fact psychologically
However, as Wallace (1965) and Goodenough (1965) consequently
pointed out, the fact that more than one model of a semantic system
can be constructed raised profound implications for a cultural theory
that presumed that a society’s culture was shared by its members.
Aberle, as I have noted, had earlier detailed the implications of
this with regard to the use of linguistic models in culture and
personality theory. And he distinguished between sharing and participating,
such as in a communication system and in a social system (1960:14-15).
The degree of sharing of cognitive maps is an important empirical
question, originally raised by Wallace (1961, 1962), but one which
is difficult to deal with because our analytical concepts have generally
included the assumption of some degree of sharing. That is, with
the exception of Wallace’s concept of mazeways. Nevertheless,
it is interesting to note that social structuralism did not fall
into this trap as social structuralists early on realized that a
social system was heterogeneous and that its members did not have
identical knowledge of its cultural domains.
But to return to the issue of behaviour - in my previous critique
of cognitive structuralism (1973) I was misled by Goodenough’s
attack on social structuralism, by his statement that behaviour
was only an artefact of the ideational order, and by the attack
on statistical summaries of behaviour by cognitive anthropologists
(see Appell 1969 for a discussion of this issue). This supposed
view that the locus of ethnographic reality lay in the minds of
informants also apparently led Burling astray (1964b). He pointed
out the difficulties of dealing with mental phenomena, of getting
into the cognitive organization of informants, and spoke for the
primacy of behaviour as the focus of ethnographic investigation.
But the important point, overlooked by many, is that even in early
programmatic statements the cognitive structuralists did not ignore
behaviour. Thus, Frake (1962:85) wrote:
Consequently, a strategy of ethnographic description that gives
a central place to the cognitive processes of the actors involved...will
give us productive descriptions of cultural behavior, descriptions
which, like the linguists’ grammar, succinctly state what
one must know in order to generate culturally acceptable acts
and utterances appropriate to a given socio-ecological context
(Goodenough 1957) [emphasis added].2
However, the use of behaviour, that is, acts, decisions, and so
forth, as a method of establishing the validity of statements of
cultural rules began to grow in importance as cognitive structuralism
moved on from an exclusive concern over folk classifications to
decision models. Kay discussed this aspect of cognitive structuralist
There are at least two ways of checking empirically that one’s
construction of the native actors’ decision making process
is correct. First, one can assess native actors’ reactions
to actions on the part of their fellows. The cognitive model predicts
which actions will be judged appropriate under which circumstances...
However, this is not the only kind of test one can make of a cognitive
model. The stronger test requires that one predict not only the
natives’ judgments of appropriateness after the behavioral
facts but that one predict actual behaviors before the fact. To
do so one obviously has to have information outside of the cognitive
model, namely, the inputs to that model. For example, suppose a
cultural rule states in part “reside post-maritally with the
husband’s matri-sib if that sib (a) is localized (ie, is in
possession of territory) and (b) owns uncultivated land.”
If we wish to predict where a given individual X will reside after
marriage, we will have to know, in addition to the fragment of a
cultural rule just given, the sex of X, which matri-sib X (or X’s
husband) belongs to, whether that matri-sib is localized, if so
where, and whether or not that matri-sib controls unused land, among
To predict distributions of actual residence patterns on an aggregate
scale, we have to furnish as input to the postulated cognitive
model the joint distributions of matri-sib membership, matri-sib
localization, wealth in land and so on, for the entire population.
This approach has been successfully employed to predict several
sorts of statistical distributions of economic and demographic
data reflecting the outcomes of individual applications of cultural
Thus, in sum, Kay says:
cognitive models alone do not predict overt behavior. But when
the cognitive model is supplied with information it specifies
as necessary for reaching a decision, it can predict overt behaviour
What we have here seems to be a partial return to the study of
social action and social structures, albeit an occluded image of
such. It first of all is a reflection of social structuralism in
that behaviour is generated by cultural rules. This is reminiscent
of the concern of the social structuralists over the nature of constraints
on human behaviour.3 But on the other hand it has focused on what
I have called the opportunity structure, as, for example, decision
models. But as I have pointed out elsewhere (1974b) and will discuss
here, many theoreticians of opportunism take the position that social
forms and the social order are generated by the multiple decisions
and transactions of individuals operating in their own self-interest,
not by cognitive constraints.
Unresolved problems of cognitive structuralism
Cognitive structuralism focuses on the statics of labeling behaviour
or on the statics of cultural rules and their relationships to other
behaviours. Thus, it has not yet dealt with process, something for
which it should not be criticized since this claim was also made
against social structuralism and relates to its particular stage
of development. There seems to be a history in structuralist analyses
whereby the statics must first be isolated and defined before attention
can be devoted to process.
Furthermore, the assumption that labelling behaviour somehow reflects
social behaviour is an interesting assumption but it needs to be
contextualized. In what circumstances is this so and when is it
not? For instance, Moerman (1966) points out that the relationship
between two cognitive categories, “things that have price”
and “kinsmen” has remained the same among the Lue, although
there have been major social and cultural changes in the society.
This has resulted in different inventories of behaviours now found
under these two terms as well as a shift in the function of these
two terms in justificatory discourse for action. Diebold, in what
is without a doubt an important article on kinship terminology (1966),
uses empirical field data to test various hypotheses as to the relationship
of labelling behaviour to kin behaviour. He concludes that behaviours
may first change, producing then a change in labels. Thus, lineal
family structures among the Mareño use collateral merging
terminology while the terminology used in stem family structures
differentiates between lineal and collateral kinsmen. But the important
evidence is that while all lineal families use merging terminology,
a few stem families do as well. This thus indicates that in the
shifts to new behaviours, that is, the stem family, labelling practices
have lagged behind.
In sum, I would argue, with Bateson (1972, 1979), that a cybernetic
model of cognition and behaviour would be more productive. While
cultural rules and social sanctions do direct behaviour, nevertheless
the actors in the social system also constantly scan the behaviour
of others to determine whether their application of the cultural
rules, whether their activation of the opportunity structure, is
as successful as that of others. Thus, behaviour also generates
modification of cognitive structures in the process of being produced
by them. For example, at some point the accumulation of new types
of decisions in the opportunity structure may become recognized
as different from previous behaviours. The next step may involve
their legitimization by an act of a jural institution in the social
system. This legitimization of behaviour in the opportunity system
then transforms it out of the opportunity structure into the social
structure, that is, into the structure of cultural rules (Appell
Growing opposition to the ‘structural-functional’
paradigm in social anthropology resulted in a developing interest
in synthetic structuralism on one hand and opportunism on the other.
I have collected under this rubric ‘opportunism’ all
those fields of inquiry that are concerned with how the individual
can manipulate, control, and create his own social world. Thus,
opportunism focuses on the question of how the individual attempts
to resolve conflicting principles of organization and cope with
discrepant values (see Garbett 1970:219); how social forms are generated
by transactions and multiple decisions (but cf. Appell 1974b for
a critique of this and a different approach to the problems raised
by this); and how norms and values are manipulated by the individual
in furthering his own self-interest. Such approaches are found under
labels such as ‘exchange theory’, ‘transactionalism’,
‘situational analysis’, ‘network theory’,
and so forth. And they attempt to “incorporate conflict as
a normal’ rather than ‘abnormal’ part of social
process” (van Velson 1967:129).
In terms of the matrix of issues that we are constructing here,
opportunism is actor-centred, rather than institution-centred, it
deals with social behaviour rather than cognitive behaviour; but
it has not explicitly dealt with the problem of how the analyst
isolates system-specific behaviour (see, for example, Barth 1966;
van Velson 1967; Mitchell 1969). Thus, although one would assume
that the goal of opportunism must indeed be that of describing system-specific
behaviour, I have entered it along the dimension of theory meaning.
To illustrate my point, while Barth claims for his version of
transactionalism that values, which form a crucial part of his theoretical
model, are not the analyst’s construct but are views held
by the actors themselves (1966:23), he does not provide the observational
procedures by which they may be isolated. Nor does he ever systematically
deal with the problem of isolating culture-specific entities. Thus,
in his theory of the generation of social form, Barth argues that
transactions and/or the results of multiple decisions create the
social forms. But such social forms must be system-specific and
isomorphic with the naturally occurring social entities. Yet it
is far from clear that they are in fact so. Consequently, the question
arises as to whether the actual act of generation lies in the mind
of the anthropologist rather than being located in the target society.
For example, I would argue that the ‘unilineal descent groups’
generated by Barth from multiple decision makings are social forms
whose precise system-specific status is questionable since this
is not delineated by his method. For his method does not specifically
detail whether these are jural entities holding rights as a unit;
whether they are jural aggregates whose members hold rights in severalty;
or whether they are jural collectivities. Such social isolates are
permitted in the jural system of which they are a part to have a
representative sue on their behalf even though the rights are not
held by the group but by the individual members (see Appell 1974b).
Thus I find it interesting that there has been no extended discussion
in opportunism as to how one gets at system-specific meaning, although
one would assume from their concerns that the opportunists would
in fact be particularly interested in system-specific meaning. This
is certainly the central problem of decision models, for decisions
can only be made within the terms of system-specific valuations.
Therefore it should be an integral aspect of the theory. This is
why I have temporarily entered decision models along the system-specific
dimension, though in parentheses.
One might conclude from this that there is something in the nature
of social action, as opposed to mental behaviour, which leads the
empiricist to the conclusion that what he sees the actor also sees.
Finally, those who work with the approach of transactionalism
claim that social structure does not determine behaviour but that
it emerges from the ongoing transactions and exchanges in the society.
Likewise those dealing with negotiated social orders (see Strauss
et al. 1964) argue that social structure rather than determining
behaviour is in fact being continually renegotiated in the ongoing
social processes of individuals attempting to achieve their own
interests. On the other hand, those who deal with network theory
or situational analysis claim that social structure alone does not
fully delineate the social order but needs to be supplemented (see
van Velson 1967); or that while it complements network analysis
it nevertheless orders ethnographic reality at too high a level
of abstraction. Thus while it is useful for some things, it does
distort actual social processes (Mitchell 1969:49).
I would like to reiterate that those who claim that social structure
is emergent are in opposition to the cognitive structuralists who
claim that cultural rules generate behaviour. This crucial epistemological
distinction, which puts cognitive structuralists in the same field
as social structuralists in opposition to opportunists, is not isolated
in the matrix and box displays.
Synthetic structuralism, like cognitive structuralism, bases its
approach on the view that culture is a code. While the cognitive
structuralists argue that behaviour is a product of the code, the
cultural rules, the synthetic structuralists take the position that
“all manifestations of social activity, whether it be the
clothes that are worn, the books that are written or the systems
of kinship and marriage that are practised in any society, constitute
languages...” (Lane 1970:13-14). 4
However, the synthetic structuralists, using a Chomskian analogy,
argue that they are concerned with symbolism at the level of deep
structure, at the unconscious level, while the cognitive structuralists
tend to focus on the surface manifestations. System-specific meaning
therefore is at the surface level of cultural phenomena, while the
synthetic structuralists are concerned with identifying the unconscious
structures that produce the surface phenomena and which are lodged
in the innate organization of the mind.
Thus, the synthetic structuralists are dealing with mental behaviour
but only in terms of a theory of the investigator. They do not state
that they are explicitly attempting to delineate culture-specific
meaning, the rational meaning of behaviour to the members of society.
Nor have they attempted to develop methods for testing their reconstructions
against newly generated materials, that is, against behaviour, as
the cognitive structuralists have done in order to test for error.
The test of a synthetic structural analysis can only lie in its
internal logic, consistency, and fit with received data. But more
about this later.
The point of my argument here is that synthetic structuralism would
thus lie, in the terms that I have delineated here, in the theory-meaning
Finally, synthetic structuralism does not explicitly deal with
the problem of the social locus of its phenomena. One might argue
that, since the synthetic structuralists are dealing with mental
behaviour the locus of ethnographic reality must lie in the minds
of individuals, yet they have shown little concern with the nature
of individual’s articulation with his socio-cultural system.
They instead use a myth or a ritual, or a segment of other cultural
behaviour, as a collective representation of a society.
Question-sets and goals of anthropological inquiry
I use the term ‘question-set’ to refer to the direction
a field of inquiry is taking. The term ‘paradigm’ was
never supposed to be used at this level of specificity, and it has,
furthermore, been semantically eviscerated by its widespread and
In summary, the basic goal of social structuralism is to answer
the question of how a social system operates. The basic goal of
cognitive structuralism is to answer the question-set of what must
someone know, and how does he achieve this knowledge, to be able
to interpret correctly verbal and non-verbal messages and thereby
act appropriately or anticipate appropriate behaviour. The goal
of synthetic structuralism is to answer the question of what are
the universal mental forms and processes, and, according to Leach,
to ask the question: “How does man perceive himself to be
in relation to the world and to society?” (1970a:195).The
basic goal of opportunism is to answer the question of how does
the individual actor manipulate, control, and create his social
universe (Garbett 1970).
Features of the Theoretical Models
At this point it is important to make a distinction between the
use of theoretical models and descriptive integration to order empirical
data. In writing an ethnography, the anthropologist has the problem
of how he is going to present his data. If he does not have a precise
theoretical model to order his data (and in fact collect his data),
he can select a theme, or a salient value, or a form of social action
that is repeated in different cultural domains, and use this as
a basis on which to present his data. Such an approach is not replicable;
it is instead highly personal. And I have called this method ‘descriptive
Recently another approach has arisen that appears in scientific
guise, but which in fact I believe to be a variant of descriptive
integration. I refer here to the presentation of ethnographic materials
in terms of the contrast between culture and social structure. Culture
here is referred to as an ideational system that is integrated in
logico-meaningful terms, while structure is phrased as behaviour
that has a causal-functional method of integration. But since this
approach posits no basic units, since it does not distinguish ritual
behaviour from culture (Wyllie 1968), since ideational phenomena
can be viewed in causal terms while behaviour has both logical aspects
and meaningful aspects in fact — it cannot be otherwise or
it would not be human behaviour — I find that this is not
so much a theory of socio-cultural behaviour as a technique of organizing
ethnographic materials for presentation. I should also add that
I find that transactionalism also has certain aspects that suggest
that it is only a technique for descriptive integration rather than
a theory of action, but I shall return to this later.
In any event, any theory that lays claim to scientific responsibility
must establish its basic units, must provide observational procedures
to move from its theoretical constructs to observables, and must
attempt some explanation of the relationship that exists between
its basic units. Furthermore, for theories to be productive, as
I have pointed out elsewhere (1973), such as those of structural
linguistics and social structuralism, they should view the ethnographic
phenomena that they are investigating in terms of a system. And
finally, because the phenomenal world and man’s relation to
it is segmented, defined, and organized differently within each
cultural system, there is in addition the problem of determining
that one’s basic units are phrased in terms of abstract, analytical
concepts, which carry no cultural loading from the observer’s
own socio-cultural system or from the system where they were first
isolated. As such they will not distort the system-specific reality
of the target society and will permit the delineation of the system-specific
boundaries between the cultural domains.
Features of the theoretical model of social structuralism
Radcliffe-Brown’s use of the organism analogy for the social
system indicates that social structuralists deal with phenomena
in terms of a system, although this is not always made explicit
(see Appell 1973). The boundaries to the system are implicitly there
from this same analogy, but again this problem is not often explicitly
dealt with. I have also pointed out elsewhere (1973) that the social
structuralists frequently fail to delineate their concepts so that
they would be devoid of cultural contamination from the observer’s
own society or that society where they were first used, and therefore
the social structuralists have not adequately dealt with the problem
of developing abstract analytical concepts.
The basic unit of social structuralism is the person as an occupant
of various roles (Nadel 1953:91-8; Radcliffe-Brown 1952:193-4).
The interesting aspect of this basic unit is that operations have
been developed in social structuralism whereby the basic unit may
be expanded into groups, networks, quasi-groups, and institutions
at other levels of analysis. Thus, for example, Nadel (1953:107-44),
in his expansion of the person to an institution comprising any
number of persons, includes purposes or aims, sanctions, rules,
and action patterns as standardized modes of co-activity. The expansion
rules for a grouping include the principles whereby individuals
are recruited; context of institutionalized activities in which
groups operate and are visible; factors making for cohesion and
endurance of the grouping, the internal order of social relations;
and the external order or relations with other social entities (Nadel
One of the major controversies in social anthropology today is
whether this expansion is culturally determined or under the control
of the actor himself, as I have mentioned before.
I have also pointed out elsewhere (1973) that social structuralism
has generally failed to provide observational procedures whereby
their theoretical constructs can be systematically linked to observables,
although with regard to persons, and some aspects of networks and
groupings, some standardization of these procedures has been attempted.
The method of articulation of persons to one another is via the
concept of social relations, more specifically normative social
relations. With respect to the explanations advanced to account
for these articulations, social structuralism has frequently been
accused of focusing primarily on functional explanations. However,
this is only partially so, for one of the characteristics of social
structuralism is that explanation is pansophic. In some instances
associational or statistical explanations are advanced, which are
of course on a different level from the functional; in other instances
teleological explanations are used. Other explanations involve causal
relationships, and in still others the focus is primarily on meaning
to the actors, although, as I have noted, the social structuralists
have not systematically developed theoretical constructs or observational
procedures to isolate the naturally occurring units of a social
Features of the theoretical model of network analysis
Whether network is by definition a system or not I find difficult
to ascertain. Certainly it includes basic units, persons (in some
instances actors) and their interrelationships, which suggests at
a very minimum a non-adaptive type of system, that is, a system
without feedback. Network analysts have not yet made any definitive
statements along these lines.
The boundary problem of networks has produced a certain conflict
of views. Barnes discusses the various aspects of this problem,
and concludes that boundaries can be drawn by the observer for the
convenience of his analysis (1969:69). Mitchell also concludes that
boundaries are fixed in relation to the social situation being recognized
(1969:40).5 Thus, it should be noted here that emphasis has not
been put on identifying the naturally occurring boundaries or break
points in a network or between networks.6
There has been little discussion of the concepts used in network
analysis in terms of whether they are abstract analytical ones.
In part they are, particularly in those analyses in which the basic
unit used is the person. In these the network theorist normally
focuses on the gross morphological features of the network. For
other theorists who focus on the instrumental and transactional
aspects of networks (for example, Garbett 1970:220-21), the basic
unit tends to be more specifically the actor operating in a frame
of individualized action.
The nature of the necessary observational procedures is still
being discussed and there is some difficulty in finding techniques
by which the observer can in fact plot an ongoing network. The method
of articulation of the basic unit, the person, is in terms of social
relations and, as Mitchell writes:
The interest in these studies focuses not on the attributes of
the people in the network but rather on the characteristics of
the linkages in their relationship to one another, as a means
of explaining the behaviour of the people involved in them (1969:4).
The explanation of social relationships and behaviour in network
analysis may follow the full range of explanation found in social
structural analysis. First, explanation is advanced in terms of
system specific meaning:
the most important interactional aspect of the links in a person’s
network is that which concerns the meanings which the persons
in the network attribute to their relationships... (Mitchell 1969:20).
Functional explanations are also attempted, and causal explanations
are always in the background:
A high frequency of contact, however, does not necessarily imply
high intensity in social relationships (Mitchell 1969:29).
And associational explanations are frequent, as, for instance,
when it is hypothesized that a greater variation of norms is found
in a loose knit network (Mitchell 1969:37). But primarily it is
teleological explanations that are advanced:
At times these recognized relationships may be utilized for a
specific purpose - to achieve some object, to acquire or pass
on some information, to influence some other person in a desired
direction (Mitchell 1969:26).
Features of the theoretical approach of social situationalists
In the approach to ethnographic reality of the social situationalists,
there is little emphasis on treating the data in terms of a system.
This is related to their method of dealing with the boundaries of
the social situation, which also illustrates their failure to come
to grips systematically with the problem of system-specific meaning.
The situation as a unit of analysis, therefore, is defined by
the observer...This is different from the treatment of situations
by, for example, the symbolic interactionists... where it is the
actor’s definition of the situation which is significant
Thus there has been no emphasis on devising abstract, analytical
concepts. On the contrary, there seems to be a positive search for
the empirically salient. In other words, the facts are delineated
in terms of the folk concepts of the social situationalists themselves.
This is also illustrated by their approach to their basic units.
The basic unit of the social situationalist is the actor, not the
person. Van Velson writes:
it [situational analysis] requires a greater emphasis in fieldwork
on the recording of the actions of individuals as individuals,
as personalities, and not just as occupants of particular statuses
However, one wonders whether it is possible to observe and record
behaviour as distinct from roles, for observation here may be similar
to linguistic inquiry in which it is indeed very difficult to provide
a completely phonetic recording of a language without having immediately
begun phonemicization of it.
The nature of the observational procedures for the collection
of data under the approach of social situational analysis is not
entirely clear or spelled out. However, van Velson states that the
focus of social situational analysis is to learn how individuals
cope with conflicting norms and how they deal with social conflict.
As a result,
in collecting and presenting data on the actual behaviour of
individuals reference must always be made to the norms which govern
or are said to govern that behaviour. Thus one will be able to
assess whether deviation from certain norms is general or exceptional,
why such deviation occurs, and how it is justified... [Furthermore,
as] the norms of society do not constitute a consistent and coherent
whole ...[this] allows for their manipulation by members of a
society in furthering their own aims, without necessarily impairing
its apparently enduring structure of social relationships. Situational
analysis therefore lays stress on the study of norms in conflict
(van Velson 1967:145-6).
The method of articulation of the basic units is apparently social
relationships but specifically actual social relationships, and
the explanations of these are made primarily in terms of purpose,
meaning to the actors, and choice.
Features of the theoretical model of transactionalism
The transactionalists do not to my knowledge define their theoretical
models explicitly in terms of a system. That it is a system they
are dealing with, sometimes a non-adaptive system, sometimes a system
with feedback (Barth 1966), is clear, but its features are not specifically
delineated. The problem of boundaries also does not feature in their
discussions of their theoretical models. And while their analytical
constructs might appear to be abstract, that is, devoid of cultural
debris and therefore applicable to all socio-cultural systems, this
aspect in the construction of productive theoretical models is not
raised in their discussions.
The basic unit of the analysis appears to be the actor or person
(Paine 1973). However, in Barth’s statement one at times begins
to feel that it is the transactional act itself which forms the
basic unit. In any event, in his discussion of the structure of
a fishing vessel crew the basic unit is a person constrained by
statuses. Then in his discussion of the processes of integration
in culture, it is the evaluating actor. But few observational procedures
are detailed to enable the observer to move from his theoretical
concepts to the observables.
The basic units are linked by transactions, sometimes of material
goods and services (Barth 1966), in other instances by messages
(Paine 1973). And it is these transactions that generate role behaviour
(Barth 1966:9), or social forms; or, in the instance of the evaluating
actor, the transactions generate consistency between values.
The observed phenomena are explained in terms of choice and purpose,
that is, maximizing value, and sometimes in terms of a cybernetic
model of the transactional processes.
In conclusion, I find Barth’s argument that social forms
are generated by choice and transactions unconvincing, and not only
because the forms are not isolated in system-specific terms. His
approach is a simple behaviouralist’s one rather than being
based on social action, which would include the participants’
cognitive organization of events as well. Thus, in dealing with
a social system which has for the observer no beginning, only changes,
how can we factor out the prime mover, behaviour or the cognitive
organization? They are instead intertwined. Thus, if there exists
in the cognitive organization of the participants a hamlet as the
place of residence, can we really say the social form of the hamlet
was produced by the choices of multiple individuals or that the
social form of the hamlet channelled the choices of individuals
from a disordered array to an ordered array? There is no question
that in certain instances of change there may be a shift in choices
in some sectors of the opportunity structure so that they pile up
in a different manner than previously realized. But this does not
create a new social form until it is scanned by the participants,
brought into their cognitive organization at one level, and then
finally legitimized by an act of an accepted social forum as a new
social isolate (see Appell 1974b).
Thus all that Barth in his theory of social forms offers is a
new type of descriptive integration. For example, his analysis of
the role behaviour of the netboss cannot be founded on the set of
transactions he maintains that it is, for one could predict the
joking role of the netboss from experimental evidence on the structure
of small groups that have the same distribution of power but which
operate in completely different cultural and ecological settings.
The formal relationships remain unaffected by the substance of the
transactions or the nature of the tasks.
Features of the theoretical model of cognitive structuralism
Cognitive structuralism can be broken up into various subfields,
and I have suggested three such subfields which might be productive
to consider here. These are all closely articulated, however, through
their similar emphasis on the lexeme as the basic unit.
Kinship semantics. It is assumed, but nevertheless unspecified
or unexamined in detail, that the field of their inquiry forms a
system. Boundaries become an important issue in kinship semantics,
as well as in folk classification, for there is concern to ensure
that system-specific boundaries be delineated and not imposed by
the outside observer’s own presuppositions (Conklin 1964:467).
The concepts are generally framed in terms of abstract, analytical
features on the basis of a linguistic analogy, but again this aspect
does not receive the attention that it might (see Appell 1973),
even though the concern is to discover system-specific meaning.
In any event, the basic unit of the lexeme is a postulated universal.
But as we shall see in dealing with the subfield of event analysis,
the lexeme has a limited field. Here it is restricted to a terminologically
distinguished array of objects (in contrast with events), referred
to as a ‘segregate’ (Frake 1962; Conklin 1967).
Observational procedures have been detailed not only on how to
embody this concept with the substantive materials from any cultural
system but also on how to determine system-specific boundaries.
The cognitive structuralists have been extremely conscious of the
problem of observational procedures which are culture-free and have
been very explicit in detailing them. However, the nature of unlabelled
categories has presented certain difficulties, since these may in
fact be only behaviourally defined (Dentan 1970), and this problem
I don’t think has yet been satisfactorily resolved (cf. however,
Berlin et al. 1973).
Furthermore, one of the paradoxes in this field, and one of the
major problems for its methodology, arises from the analysis of
the ethnographic materials gathered by others not sharing the goals
or concerns of the cognitive structuralists. Cognitive structuralism
arose partially in response to the problem of collecting field materials
uncontaminated by the observer’s own culture and presuppositions.
Various techniques were devised to insure that the materials did
represent the cognitive reality of the target community, and then
this material was subjected to analysis by various methods, including
componential and extensional analysis. But the paradox arises when
these same techniques, which are claimed to reach some level of
cognitive reality of the members of the target community, are used
on materials gathered by methods that have been charged as being
contaminated. Thus, there is every reason to suspect such bodies
of ethnographic materials for distorting system-specific reality,
particularly along the boundaries of the domains being studied,
for here is where contamination and distortion is most likely to
The method of articulation of basic units was originally in terms
of contrast and inclusion, but the ordering of these digital relationships
has provided considerable ground for discussion (Colby 1966; Werner
and Fenton 1970). In addition, Bright and Bright (1965) and Dentan
(1970) have also challenged this digital view of native categories
and suggested that many folk classifications are in terms of more
or less, therefore analogue in nature, rather than in terms of contrast
or opposition, the presence or absence of a feature.
Thus Dentan (1970), in dealing with taxonomies among the Semai
of Malaya, finds that labelled categories sometimes fade away at
their boundaries so that instead of being defined by the presence
or absence of a particular characteristic they are defined by the
degree to which a characteristic is present. Dentan concludes that
there are two major processes of learning, discrimination and stimulus
generalization; and he suggests that componential analysis is an
appropriate tool for dealing with those areas in which the learning
has been by discrimination. For domains involving polysemic categories
for which there appears to be an ideal type, Dentan suggests that
the process by which they are learned may be based on stimulus generalization.
Therefore, in the analysis of such categories an extensionist model
might be more appropriate.
The explanation of the nature of articulation of the basic units
in kinship semantics, as well as in folk classifications, has been
first in terms of the meanings these have to the participants. But
at another level of explanation, Tyler (1969) has claimed that explanation
in cognitive structuralism is in logico-mathematical terms while
explanation in social anthropology is in causal-functional terms.
As I have pointed out earlier, this view raises serious questions,
for social behaviour can also be viewed in terms of its meaning
to the participants, in terms of its mathematical relations (Kaplan
1965), and in terms of its logical relations to other aspects of
culture viewed in either system-specific terms or theory-meaning
terms. Furthermore, one can see that certain fields dealing with
cognitive behaviour, such as that of cognitive dissonance, deal
with causal explanations; and the functions that beliefs perform
have always been a central concern of all the social sciences, including
Folk classifications. I believe that I have covered most of the
features of this subfield under my discussion of kinship semantics,
and I will not go into it further here.
Percipient events. I use the term ‘percipient events’
to distinguish analyses of events in terms of the perceived or cognitive
aspect of the event from analyses that focus on behaviour or the
social action of the event.
The question as to whether the ethnographic materials in percipient
events are perceived in terms of a system can be answered in the
same manner as with the previous two subfields of cognitive structuralism:
it is assumed to be so but it is not specified in these terms. Boundaries
are to be explicitly determined in terms of system-specific criteria,
although in Frake’s study of law (1969) it is presumably in
terms of boundaries specified by Conklin for the lexical set:
a lexical set consists of all semantically contrastive lexemes
which in a given, culturally relevant context share exclusively
at least one defining feature...(1967:124).
In Agar the boundaries are presumably identified by breaks in ‘prerequisite
links’, that is,
Two events may be cognitively related in that the outcome of
one provides a prerequisite for the next (1973:20).
As with ethnosemantics, abstract analytical concepts are used,
but their formal nature is not explicitly discussed.
The basic unit of percipient events raises some problems. Agar
first states that the basic unit is the lexeme (1973:12), but he
then states that he uses the concept of an ‘event’ as
a primitive term (1973:15). Thus, the basic unit does not consist
of all possible lexemes but only those dealing with events, which
Agar states will expectedly be verbs encoding events or, more specifically,
action-process verbs that encode an actor engaging in some act which
is also a process acting on some ‘patient’ (1973:15).
Frake includes as constituent units of events the following: setting,
provisions, paraphernalia, roles of participants, and routines (1964:472).
In a later article (1969) the constituent units are: focus, purpose,
roles, and integrity, that is, the extent to which an activity is
construed as an integral unit as opposed to being part of other
activities. Agar suggests that it is too early yet to isolate such
constituent units (1973:14-16).
Frake states that the lexemes of events are articulated by contrast
and inclusion (1969:149). Agar (1973) suggests that event concepts
are linked by a logical sequence of necessary prerequisites, which
are not necessarily temporally ordered.
These articulations are explained in meaningful terms as well as
logico-mathematical. But lurking in the background is an ordering
and thereby an explanation in terms of paradigmatic and syntagmatic
(To be concluded next part.)
This is a revised version of a paper delivered at the SUNY Symposium,
Conversations in the disciplines: the mutual relevance of structural
anthropology and cognitive anthropology, Geneseo, New York, 13 April
1974. I want to thank the organizers of the Symposium, Russell A.
Judkins and Gerald M. Erchak, for a well-organized and most productive
Symposium. I also want to thank Barbara Henry and Eric G. Schwimmer
for their useful comments. And I want to express my appreciation
to Paul Garvin for his good criticisms and particularly for his
very useful suggestions on how my discussion of linguistic models
might be more profitably revised. I am solely responsible, however,
for any errors of omission and commission. The final editing of
this paper and its submission for publication was done while I was
a Visiting Fellow in the Department of Anthropology, Research School
of Pacific Studies, The Australian National University in 1980.
This visit was supported by NSF Grant No.BNS-7915343.
1 Ardener expresses (personal communication) his dissatisfaction
with referring to structural-functionalism as ‘social structuralism’
and structuralism as ‘symbolic structuralism’ with some
reason, preferring the terminology of ‘structural-functionalism’
2 “A society’s culture consists of whatever it is
one has to know or believe to operate in a manner acceptable to
its members and to do so in any role they accept for any of themselves”
3 “...a growing awareness that ‘culture’ is best
seen as a set of control mechanisms - plans, recipes, rules, instructions
- which are the principal basis for the specificity of behavior
and an essential condition for governing it...” (Durbin 1973:470).
4 “Therefore totemic ideas appear to provide a code enabling
man to express isomorphic properties between nature and culture.
Obviously, there exists here some kind of similarity with linguistics,
since language is also a code which, through oppositions between
differences, permits us to convey meanings and since in the case
of language as well as in that of ‘totemism’, the complete
series of empirical media provided in one case by vocal articulation,
and in the other by the entire wealth of the biological world, cannot
be called upon, but rather (and this is true in both cases) only
a few elements which each language or each culture selects in order
that they can be organized in strongly and unequivocally contrastinq
pairs” (Lévi-Strauss 1963c:2).
5 “How far the links of a network need be traced depends
entirely upon the field-worker’s judgement of what links are
significant in explaining the behaviour of the people with whom
he is concerned” (Mitchell 1969:13).
6 This lack of system-specific perspective is reflected in Mitchell’s
view of the analyst’s approach to the data: “It could
be argued that from an actor’s point of view no relationship
has only one content. This is true. The observer, however, for his
purpose may be justified in treating a set of actions as being so
dominated by one set of identifiable norms that for all intents
and purposes the relationship is single-stranded...The perceptions
of the strands in relationship, however, depend upon the analytic
purpose of the observer” (Mitchell 1969: 22-3).
7 Yet Garbett later on writes (1970:217): “From this perspective,
a situation is viewed as occurring within a field setting whose
circumstance expands and contracts according to the changing interests
and values of the actors in the situation. The field is defined
in terms of the interest and involvements of the participants in
the process being studied and its contents include ‘the values,
meaning, resources and relationships employed by the participants
in that process’ (Swartz 1968:9).” In any event, the
locus of the boundaries of the social situation have not become
a major concern of the social situationalists.
EPISTEMOLOGICAL ISSUES IN ANTHROPOLOGICAL INQUIRY
Social Structuralism, Cognitive Structuralism, Synthetic Structuralism
[This is the concluding part of an essay published in two parts.
In the first part (Vol. 3 No. 2) the author introduced the problem
occasioned by a discipline which has “spawned a proliferation
of theoretical models, a morass of new methodology” such as
to obscure one’s vision of the goals of anthropological inquiry.
He suggests a solution to this problem through the establishment
of a comparative framework in which several extant models and methodologies
may be compared, one with another, according to their (borrowing
from Kaplan 1965) “logic-in-use”. Part 2 opens with
his concluding remarks on comparative features of the theoretical
models established. These remarks are then followed by the author’s
Conclusion, his propositions regarding discrepancies between what
various theorists and methodologists say they do and in fact do,
and his proposals for directions out of the “morass”.]
Features of the theoretical model of synthetic structuralism8
The discussion of the features of theoretical models up to this
point may have been a bit tedious, but it is in dealing with synthetic
structuralism that I find this technique to be the most productive.
First, the synthetic structuralists explicitly state that they
are considering ethnographic materials in terms of a system, one
modelled after the linguistic system (Lévi-Strauss 1963a:33).
But it should be noted that the characteristics of the linguistic
system and its applicability to sociocultural phenomena, as in cognitive
structuralism, has not been sufficiently explored (see Appell 1973).
And here is where much of the criticism against synthetic structuralism
Where the boundaries to the system are to be drawn also raises
certain questions. Leach writes (1971:23): “In contrast [to
functional interpretations], in a structuralist interpretation,
the analyst considers each myth, or ritual sequence as a whole.”9
This would seem a reasonable locus of boundaries, yet Lévi-Strauss
frequently carries his mythic analysis across cultural and temporal
boundaries (see his analysis of the Oedipus myth in Lévi-Strauss
1963a:206-31). Thus, the boundary problem seems to be unresolved
in synthetic structuralism, or perhaps I should say rather fluid
and reactive to the analyst’s interests rather than to naturally
The problem of devising abstract analytical concepts seems to
be little considered by the synthetic structuralists, although this
can only be satisfactorily dealt with in the discussion of the basic
The basic unit of synthetic analysis defies the epistemological
sleuth. Barnes argues with respect to kinship studies that “whereas
many modern descriptions of a language begin with a list of its
phonemes, Lévi-Strauss never enumerates exhaustively the
fundamental elements used in structural models. He merely describes
duality, alternation, opposition, and symmetry as ‘basic and
immediate data of mental and social reality’...” (1971:116).
However, Barnes also writes that “models of kinship systems...are
built up out of a single type of ‘elementary structure’
consisting of a woman, her brother, her husband, and their son.
This constitutes the unit or atom of kinship...” (1971:116-7).
In other writings Lévi-Strauss refers to his basic units
as mythemes or gustemes, depending on the context of the analysis.
Let us look for a moment at mythemes. Lévi-Strauss argues
that myth falls in the domain of language, and like the rest of
language is made up of constituent units. These units belong to
a higher and more complex order than do the constituent units of
language, phonemes, morphemes, and so forth. They are to be found
at the level of sentences (Lévi-Strauss 1963a:210-11).
These gross constituent units consist of a relation. And it may
in fact be that the ‘relation’ is the basic unit of
synthetic structuralism. Thus, Leach writes: “the elements
of symbolism are not things in themselves but ‘relations’
organized in pairs and sets... The crucial point is that the ‘element
of structure’ is not a unit thing but a relation X “
(1973:48-9). And there have been various attempts to phrase this
in abstract, analytical terms as illustrated in the use of signs
such as those for minus (-) and equals (=) (see Lévi-Strauss
Whether the basic unit is an X-eme or a relation X, the point
is that if culture in all its aspects can be viewed as communication
there must be fundamental units for its analysis at each level,
as in linguistic analysis. Gustemes and mythemes in this view are
contradictory to the premises of the argument since they are not
reducible to the same basic unit. To argue that clothing, food,
and other domains of culture form different languages (Leach 1970b:46)
misses the point, for it confuses content with structure. And the
synthetic structuralists are concerned here with the structure of
the code carrying the message. The contradictions in the approach
of synthetic anthropology thus stem from this failure to develop
and apply systematically abstract analytical concepts that have
relevance for the analysis of all cultural systems. One gets the
impression that, while insight grows, there is no cumulative development
of theory and method, for, without explicit basic units that are
formulated in abstract analytical terms so that they apply to all
social systems and all aspects of social systems, and without detailed
observational procedures by which the basic units can be related
to empirical materials, many find it difficult to replicate the
work of Lévi-Strauss.
With respect to the observational procedures that Lévi-Strauss
does provide, they are only partially mapped out for isolating the
X-emes and the relations. Lévi-Strauss writes, for example:
The method we adopt, in this case as in others, consists in the
(1) define the phenomenon under study as a relation between
two or more terms, real or supposed;
(2) construct a table of possible permutations between these
(3) take this table as the general object of analysis which,
at this level only, can yield necessary connexions, the empirical
phenomenon considered at the beginning being only one possible
combination among others, the complete system of which must
be reconstructed beforehand (1963b:16).
The basic technique involves the search for opposition and the
mediation of this opposition at a higher level. Thus, Lévi-Strauss
writes that mythical thought always progresses from the awareness
of oppositions towards their resolutions:
...two opposite terms with no intermediary always tend to be
replaced by two equivalent terms which admit of a third one as
a mediator; then one of the polar terms and the mediator become
replaced by a new triad, and so on (1963a:224).
The position 1 take here that the observational procedures are
only partially mapped out is based on the fact that we have no means
to select without personal bias what X-emes or relations are the
salient functioning discriminations at whatever level, the cognitive
or the unconscious, for the society under study. But I shall discuss
this further shortly.
In the analysis of kinship systems the method of articulation
of social units is by means of communications of women, messages,
and goods. At a higher remove this puts units into opposition, and
thus at a higher level the method of articulation of units in his
structural models is by opposition and transformation. However,
Leach has argued that some relations are best viewed in terms of
a continuum rather than in binary opposition (1964, 1970a, 1971),
but I shall return to this shortly.
Finally the explanation of the articulation of units is posited
in teleological terms in kinship analysis at one level, but at another
level in terms of logico-mathematical relations (see Leach 1971:23),
particularly paradigmatic and syntagmatic structures (Leach 1973:45).
But surely one can see gross causal explanations creeping in as
well as associational ones.
Unresolved Issues in Synthetic Structuralism
I shall not attempt to touch on all the issues that critics have
raised in their analysis of synthetic structuralism; instead I shall
only deal with those that arise from the contrast of the theory
and methodology of cognitive structuralism, social structuralism,
and opportunism with synthetic structuralism.
Locus of ethnographic reality and linguistic models
As I have previously noted, the synthetic structuralists have
yet to deal with the problem of the locus of ethnographic reality.
They state that they deal with mentalistic phenomena, which would
lead one to assume that the locus of ethnographic reality lies in
the mind. And to a certain extent this is true, but not entirely
so. For the basic data used in the analysis of synthetic structuralism
is the product of collective minds, that is, sociocultural phenomena.
And so they deal not with individual mental processes but shared
mental processes, or, perhaps better, shared mental processes inferred
from social, cultural, and linguistic behavioural evidence. This
position is justified, as well as explicated, by their reference
to linguistic models, which implies that the models constructed
by the synthetic structuralists are shared by all members of the
referent society. Thus, using a linguistic model for their research,
but of a Jakobson rather than a Bloomfield variety, they nevertheless
have become trapped in the same errors as were the cognitive structuralists.
They do not investigate the social distribution of their data. Or,
to put it in another way, they do not test their assumption that
all members of the society share the same structure of codification,
the same sets of meanings, the same knowledge. Yet we know that
the cognitive organizations of the individual members of any society,
and from this presumably the organizations of their unconscious,
are not isomorphic but instead only overlap. This lack of isomorphy
is one of the major findings of cognitive structuralism. However,
the degree to which cognitive structures do not overlap is an unresolved
problem of cognitive inquiry. Thus, the synthetic structuralists,
as the cognitive structuralists, have yet to come to terms with
this problem and detail the social distribution of their structural
models (see Appell 1973).
One additional but tentative conclusion might be drawn at this
point. If the synthetic structuralists are dealing with sociocultural
phenomena, which by definition are shared, perhaps Lévi-Strauss’s
goal of delineating universal mental structures is unrealistic.
What, instead, he might be doing in some of his analyses is detailing
the social processes by which individual members of a society reconcile
their conflicting views of the sociocultural field in which they
must act. We would thus be dealing with the products of innumerable
efforts at strategic interaction and manoeuvring rather than with
mental processes directly. The locus of ethnographic reality, if
this is true, would lie not in the minds of individuals but in the
processes that occur between individuals as they manoeuvre and the
products of this manoeuvering. Thus, the mental processes of the
participants may be in fact structured on entirely different grounds
from those proposed by the synthetic structuralists.
The premise of cultural relativity
To the question of where the locus of ethnographic reality is,
the synthetic structuralist would answer, I suspect, by saying that
he is first of all not concerned with system-specific meaning. He
is concerned instead with the unconscious, universal structures
of the human mind, and these are phrased in terms of the universal
relations between the gross constituent units.
The social structuralist or cognitive structuralist might reply
by asking: Why therefore bother with ethnographic evidence? Why
not use the evidence occurring in one’s own society? As soon
as one considers ethnographic materials one must deal with sociocultural
systems, and there are certain accepted methods of handling these.
That is, there is an assumption that each system is unique, to a
greater or lesser degree, and one of the tasks of the ethnographer
is to determine this uniqueness. Certainly, since the synthetic
structuralist uses a linguistic model, he is also forced to this
conclusion, as linguistic analysis is similarly based on the assumption
that each system is unique and the substantive, meaningful categories
of one system cannot be transferred ungeneralized for the analysis
of another. Thus, while the linguist is also searching for potential
relations, this is always done in terms of a specific system and
in terms of a context, a specified environment in that system. Furthermore,
the linguist’s relations are not phrased in terms of meaning,
while the synthetic structuralist’s relations are. They both
assume and have sufficient evidence to conclude that meaning must
indeed be system specific, that is - and I leave here my postulated
discussion between synthetic and other structuralists - unless there
are certain universal symbols. And the arguments presented here
are not to deny the validity of searching for such universals. But
the point is that the assumptions and methods of one question-set
must not be confused with another.
I would like to present now two brief examples to illustrate how
the synthetic structuralists apparently violate the premises of
the question sets of social and cognitive structuralism without
adequate explanation. Maranda defines myth in the following terms:
Myths display the structured, predominantly culture-specific,
and shared, semantic systems which enable the members of a culture
area to understand each other and to cope with the unknown (1972:12-13).
If the term ‘culture-specific’ refers to the “members
of a culture area” my mind boggles. Take Kroeber’s Southwestern
Culture Area (1947). This is composed of Navajo, Apache, and the
various Pueblos. On what grounds are we to establish that this range
of cultural systems share semantic domains, and then how are we
to explain this similarity of meaning systems? Either an explanation
is in order or we must assume that synthetic structuralism is a
retrograde development in anthropological inquiry.
But as the proposition now stands, the cognitive structuralist might
reply that it is pure nonsense, and for reasons already given. The
opportunist would argue that a myth is also one of the cultural
items that are constantly being brought into the arena of negotiation
and manipulation, so that its ‘meaning’ can only be
perceived in terms of the system of transactions and negotiations
that are going on in a specific social situation. And the social
structuralist would try to ascertain whether there are any social
relations extant between the members of the culture area sharing
the myth, if in fact they do share it, and then attempt to ascertain
how the myth modulates these social relations.
In sum, the assumption of uniqueness made by the other structuralists
may have to be modified by the research done by the synthetic structuralists,
but this must be demonstrated by reference to evidence and not by
Leach’s analysis of the Garo and Kachin social systems provides
another illustration of the problem. The Kachin of north Burma are
patrilineal, and the Garo of Assam are matrilineal and are located
about 100 miles to the west of the Kachin. Leach writes:
But a structuralist way of looking at things shows that these two
marriage rules are versions of the same principle...and modern fieldwork
has shown that the two cultural systems are in fact remarkably similar
right across the board. The contrast patrilineal descent/matrilineal
descent being the only major difference between them. A structuralist
therefore regards the two systems as transformations of a single
This position immediately raises more questions, and crucial questions,
than it in fact answers. Where is the locus of this ‘single
structure’? Is it in the mind of the analyst? Is it an aspect
of his metatheory? Or does it lie in the minds of the members of
these two societies? Thus, is it a feature of the ethnographic model
of Garo and Kachin societies?
The concept of transformation also implies that there is some kind
of continuity between the mother structure and its daughters. But
what is it? Does the mother structure lie back in history at some
point when Garo and Kachin were the same society? Or is it an unconscious
and unfulfilled opposition in the minds of the Garo and Kachin today?
Both these questions imply that there must be certain mechanisms
that determine change from one structure to another. And is this
possible without first going through a cognatic phase?
Or are the Garo and Kachin articulated together in some way even
though geographically separated so that the operation of one system
is seen as a means of establishing social identity in opposition
to the other system?
Although it seems hardly credible, Lévi-Strauss turns out
at one level to be a naive empiricist.11 I refer here to his attitude
towards ethnographic evidence. He assumes that the ethnographic
facts as presented represent the ethnographic reality of the societies
under study — except where the facts do not fit his models.
But he overlooks the point that a fact is only a fact in terms of
some ideological system. Thus, he ignores not only the implications
of his own theory but his arguments at other levels. Thus, since
Lévi-Strauss does not challenge the veracity of the ‘facts’,
except as noted, even though the fully cognizant cognitive structuralists
would argue that they have come to him screened through the bias
of the original observer, he does not appear to appreciate the problem
of the validity of the texts he works with and the possibility of
Leach, however, argues that synthetic structuralism is “intuition
free”. And he concludes that
the distinctive characteristic of structuralist procedure is
that the analyst tries to avoid making symbolic substitutions
which are not already quite overtly specified in the evidence
In the ethnographic example he gives, he states that there is
a contrast in the two versions of the myth between the overemphasis
of kinship bonds and the underemphasis of kinship bonds. But we
know that contrast can only occur within a specific environment,
a specific context. And the synthetic structuralists have not produced
any observational procedures by which the semantic domain in which
the contrast occurs, in which the opposition is found, can be isolated
and described. Therefore, we are always a bit uncomfortable with
the structuralist’s claims, since one begins to suspect that
the fields of contrast are not culture-specific but observer-specific.
Thus Bott writes:
so many equivalences can be established that one can arrive at
a considerable range of possible interpretations and the choice
of one rather than another depends on ‘feel’ as well
as logic (1972:280).
In a sense, if Wilden is right that Western culture is overdigitalized
(1972), one might suspect that the native texts which have gone
through the collection and processing by the Western mind might
have already distorted the cultural-specific organization, might
have already digitalized the native’s analogic relations,
so that by the time Lévi-Strauss studies them they are in
fact the myth of myths.
Let me discuss one final example. Lévi-Strauss depends
on the opposition between nature and culture as a basis for much
of his work. But as we have noted, he does not provide procedures
by which these culture-specific domains may be isolated for each
society. Therefore, one wonders whether he might not in fact be
imposing his own view of what constitutes nature and culture on
the indigenous sociocultural systems. For example, one wonders how
he would have dealt with the observations made by Hallowell among
the Ojibwa. Hallowell writes, with regard to Ojibwa ontology and
Since their cognitive orientation is culturally constituted and
thus given a psychological ‘set’, we cannot assume
that objects, like the sun, are perceived as natural objects in
our sense (1960:29).
The sun is, instead, a ‘person’ in the other-than-human
class. Or take stones. They are linguistically included in the same
linguistic category of animate along with persons. Hallowell writes:
Since stones are grammatically animate, I once asked an old man:
Are all the stones we see about us here alive? He reflected a
long while and then replied, ‘No! But some are. (1960:24).
Before we conclude this section, let us look at the problem of
observer bias from a different angle, which raises the question
of cultural relativity again but from another perspective. It has
been the experience of anthropological inquiry that its concepts
are usually contaminated by the presuppositions from the culture
of its practitioners (see Appell n.d.). As a result, anthropological
inquiry has continually tested them against a variety of ethnographic
evidence in order to refine them and make them more universally
valid. For example, take the problem of definition of marriage (see
Leach 1961; Rivière 1971). Yet such attempts at falsification,
at correcting for error, do not seem to be part of the synthetic
structuralist endeavour. This charge is, of course, not new but
I have attempted to put it into a more anthropological perspective.
Whether the fault lies primarily in the phrasing of the theoretical
model of synthetic structuralism or in the practice of synthetic
structuralists themselves I will not attempt to delineate here.
In any event, one of the fundamental contrasts between cognitive
and synthetic structuralism is that cognitive structuralism questions
the reliability of the ethnographic evidence gathered by methods
other than those stipulated by the cognitive structuralists.12 They
cogently argue that the possibility of cultural contamination of
ethnographic data from the investigator’s own culture has
been demonstrated in comparative studies, and they thus take the
position that it must be avoided at all costs. The synthetic structuralists
on the other hand seldom question the validity of the ethnographic
evidence that they use, nor have they developed field methods, as
have the cognitive structuralists, to prevent any cultural contamination.
In sum, the point I want to make here is that had the symbolic
structuralists provided a more precise definition of their basic
units, had they developed these in terms of abstract analytical
concepts, had they detailed the observational procedures by which
one could move from the theoretical concept to the observables and
isolate them in terms of system-specific discriminations, then the
cognitive structuralists and the social structuralists might be
more convinced by their arguments.
These same problems arise in dealing with the claim that structuralism
deals with unconscious structures.
The nature of unconscious structures
Lyons (n.d.) argues that Lévi-Strauss uses the concept
of the unconscious in two senses: (1) in the linguistic sense meaning
out of awareness; and (2) in the Freudian sense of something hidden.
Rossi, however, in his extensive review of the concept of the unconscious,
argues that Lévi-Strauss shares with Freud the conviction
that a genuine meaning lies behind an apparent one but concludes
that the only contribution that Freud has made to Lévi-Strauss’s
view of the unconscious is the postulate that what is unconscious
is more important than what is conscious. Thus, Rossi argues, the
linguistic concept of the unconscious as a structuring activity
and the related emphasis on form over content distinguishes the
structuralist approach from the psychoanalytic (Rossi 1973:29-30).
Yet there is a paradox in Lévi-Strauss’s use of the
linguistic analogy. While he maintains that he is dealing with unconscious
phenomena in his structural analyses in a method similar to that
used by the linguist to detail the unconscious infrastructure of
language, the linguist does not deal with meaning at this level
as the structuralist does. For the linguist, meaning at the level
of the unconscious is not the product of the analysis but is only
used as a means of discovery. Thus, it can be argued that if any
degree of cultural relativity is assumed, synthetic structuralism
must only be concerned with abstract, formal relationships and not
relationships that involve content and meaning, for if the analysis
does impute meaning, it destroys the validity of the linguistic
analogy. Furthermore, the emphasis on meaning and symbolism, the
stress on myth mediating contradictions on the surface level, leads
one to conclude that Lévi-Strauss does indeed lean more on
Freudian processes, depends more on Freudian concepts, than he or
others give him credit for.
Thus, at one level, if the structuralists are going to deal with
unconscious meanings they are going to have to develop a theory
of the unconscious, which would include the process through which
meaning originates and is then repressed, the nature of defence
mechanisms which prevent the meanings from rising to the conscious,
how the repressed is returned, as well as procedures for interpreting
the meaning of these and their emotive saliency.
On the other hand, if the synthetic structuralists choose the
linguistic route to deal with meaning, they are then going to have
to develop a theory of contrastive environments, as I have noted
previously. This would include how boundaries of contrastive sets
are identified in the texts or in sociocultural phenomena, how one
determines for any specific sociocultural system which element contrasts
with what other element or elements, and how one determines the
level of contrast operating. For it is not only likely but highly
probable, from our experience with the cultural contamination of
data, that the analyst may choose as contrastive elements items
which in the social system under study are in fact not in contrast.
For example, while the actual system is represented in Diagram 1,
the analyst without proper controls and testing for error in his
analysis might conclude that the structure of systems is as in Diagram
2 because of the distortion of his own cultural presuppositions.
Furthermore, in any specific analysis the unconscious nature of
the results can only be postulated, not established. For the symbolic
relationships uncovered by the analyst may appear explicitly and
consciously in other cultural statements not covered by the analyst;
or the relationships may appear to native speakers, if capable of
being tested, as quite straightforward. But they are not tested,
and one also runs up against the same problem as in Freudian analysis
when the conclusions reached by the analyst are denied by informants.
Thus, as it now stands, the concept of the unconscious covertly
provides a nice excuse for observer bias and a justification for
the failure to apply rigorous methods. For without a theory of the
return of unconscious materials by which they can be identified
(using a Freudian analogy, if indeed it is appropriate), or without
a method to test one’s analysis with native speakers (using
a linguistic analogy), it is impossible, due to the way the concept
of the unconscious is presently framed, to test for error and verify
Now the points I raise here cannot be dismissed, as some have
done with similar criticism, as graffiti of the empirical mind.
A contribution to knowledge is not a contribution until it is shared,
and until we have the methods by which it can be shared, doubt troubles
the mind. Piaget in his discussion of structuralism raises a related
logico-mathematical deduction of a set of laws is not sufficient
for their explanation, at least not so long as deduction remains
formal; explanation requires, besides, that something be supposed
to underlie phenomena and that these hypothetical objects really
act upon one another. Now the striking fact is that frequently the
action of such inferred entities resembles our own operations, and
it is precisely to the extent that there is such a correspondence
between inner and outer that we feel we “understand”.
Thus, understanding or explaining is not just a matter of applying
our operations to the real and finding that “it can be done”.
Such “application” does not break through to causes;
it keeps us within the realm of laws. Causal explanation requires
that the operations that “fit” the real “belong”
to it, that reality itself be constituted of operators...Then and
only then does it make sense to speak of “causal structures”,
for what this means is the objective system of operators in their
effective interaction (1970:39-40).
I would of course not restrict Piaget’s conclusions just
to causal explanations.
Culture as Communication and the Problem of Linguistic
Over the past several decades anthropologists have been envious
of the theoretical models of the linguists and have attempted to
adapt them for the analysis of their own cultural and behavioural
data. However, there has not been any substantive effort to analyse
the nature of linguistic behaviour, or the models built for describing
this, in order to assess the fit of such to the sociocultural domain.
As a result, although it may at first seem paradoxical, two markedly
divergent fields, both cognitive and synthetic structuralism, have
arisen claiming theoretical paternity from linguistics. While one
is derived from Bloomfieldian and the other from Jakobsonian linguistics,
these both nevertheless share a sufficient number of common features,
such as the belief in the relative uniqueness of each language,
to raise questions as to the validity of the linguistic analogy.
Thus, when one hears claims for linguistic modelling one begins
to get the impression that we are dealing with justificatory and
not scientific discourse. Consequently, I would like to consider
here briefly some of the criticisms of the applicability of linguistic
models for the analysis of sociocultural domains.
Aberle makes the important point that language is characterized
by the fact it selects a small number of actualizations from a large
number of possibilities. And while selection is also necessary in
culture, he points out that it is not simply a reduction in random
behaviour as is linguistics. On the contrary it has other important
adaptive functions in addition to making face-to-face relationships
I have also attempted to point out (Appell 1973:19-20) that the
linguistic model differs from cultural behaviour in that the linguistic
model includes the assumption of automatic behaviour at certain
speech levels, that is, behaviour carried on without conscious knowledge
on the part of the subject, and there is in addition in linguistic
behaviour the lack of organized community-wide sanctions. Thus,
there are many aspects of social organization that do not occur
with sufficient frequency to form a corpus of social behaviour large
enough for the type of analysis that the linguist uses in speech
behaviour. And finally, as opposed to linguistic behaviour, members
of a society frequently have very clear-cut ‘theories’
as to how their society works and use these ‘theories’
to guide behaviour. These are outlined most frequently and explicitly
at times of jural disputes and therefore the features of the system-specific
isolates of a target society can often be determined in such an
Keesing points out usefully that the organization of knowledge
in nonlinguistic domains may not be similar to the organization
of grammatical knowledge and furthermore that cultural rules are
basically concerned with specifying appropriate messages and
social context; and ‘linguistic rules’ are basically
concerned with the conversion of messages into verbal form (1972:315)
Bateson argues that “coding devices characteristic of verbal
communication differ profoundly from those of kinesics and paralanguage”
Furthermore, he writes that
the kinesics of men have become richer and more complex, and
the paralanguage has blossomed side by side with the evolution
of verbal language. Both kinesics and paralanguage have been elaborated
into complex forms of art, music, ballet, poetry, and the like,
and, even in everyday life, the intricacies of human kinesics
communication, facial expression, and vocal intonation far exceed
anything that any other animal is known to produce (1972:418).
He continues, in discussing the inappropriateness of the linguistic
model to the analysis of non-verbal behaviour,
our iconic communication serves functions totally different from
those of language and, indeed, performs functions which verbal
language is unsuited to perform (1972:418).
He argues that the discourse of non-verbal communication is precisely
concerned with matters of relationships such as love, hate, respect,
fear, dependency, and he concludes:
If this general view of the matter be correct, it must follow
that to translate kinesics or paralinguistic messages into words
is likely to introduce gross falsification due not merely to the
human propensity for trying to falsify statements about ‘feelings’
and relationships and to the distortions which arise whenever
the products of one system of coding are dissected onto the premises
of another, but especially to the fact that all such translation
must give to the more or less unconscious and involuntary iconic
message the appearance of conscious intent (1972:419).
He thus distinguishes the “arbitrary” and digital coding
characteristics of verbal behaviour from the iconic and analogic
coding of depiction (1972:133). This position thus casts doubt on
the argument that social behaviour may be viewed as a code amenable
to the same type of analysis that verbal behaviour is.
Wilden also criticizes the structuralists for their mistake of
treating a context-free system of oppositions between the acoustic
characteristics of “bits” of information (distinctive
features) as if it were isomorphic with myth, which is a system
with context. And he states that this error arises as the result
of their confusion of communication with language (1972:8). Lévi-Strauss
thus mistakenly uses a digital system of analysis, a linguistic
model, for describing continuous or analogue relations and as a
result introduces bias and distortions.
Leach (1964, 1970a, 1971) has also pointed out that synthetic
structuralist arguments are often unsatisfactory precisely because
they tend to reduce all discriminations to the binary ones and leave
no room for discriminations of the quantative “more/less”
type. Thus he writes: “binary analysis is a possible rather
than a necessary procedure” (1970a:197). And this brings us
back again to the problem faced by the cognitive structuralists
of separately establishing whether or not the logic of the analysis
replicates the logic in use of the members of the referent society.
The fact that communication is frequently confused with language,
Keesing’s argument (1972) that the unproductivity of cognitive
structuralism is accounted for by its dependency on a passé
model of linguistic analysis, and Burling’s argument (1969)
that ethnographic materials can be ordered and analysed like linguistic
materials, all point to one fundamental conclusion that to my knowledge
has not yet been explicitly discussed. Those using a linguistic
model tend to act as if there were some sort of ultimate reality
involved rather than a relative reality defined by our own interests
and constrained by the culturally programmed panhuman wiring of
the brain. For example, if we base our analysis and justify it on
the basis of a transformational linguistic model, as suggested by
Keesing, can we be sure that it will not be superseded by another
model, which is again claimed to represent the truth?
Thus, I believe that our position should be to view all models,
structural linguistic models, transformational models, teleological
models, causal models, etc., only as useful tools rather than reflecting
specific ontological reality. This is particularly true, it seems
to me, since it has been the experience of anthropological inquiry
that most if not all models that man creates to process information
always carry a load of cultural detritus from the domain for which
they were first established, as for example with linguistic models.
As a result, and to improve the tools of our analysis, I believe
that in transferring any model from one domain to another our position
should be to determine first the formal characteristics of it, removing
its cultural detritus, and then to use it wherever it may seem appropriate
for the analysis of our empirical data. In this view, the formal
aspects of the structural linguistic model are never passe but,
as one method of ordering data, they retain a potential for the
analysis of empirical materials where relevant. The fact that linguistic
materials can be more fully illuminated by a newer model does not
destroy the usefulness of the structural linguistic model in its
formal mode, unless we are incapable of generalizing it by removing
its contamination by specific linguistic materials.
It is this position that I believe enlightens Burling’s
claim that ethnographic analysis is similar to linguistic analysis
(1969). Sociocultural materials are not equivalent to linguistic
materials, I would argue. They may appear to be so only because
the models used in the analysis of each of their domains share certain
formal characteristics. In other words, while I differ with Burling
on many points, these two procedures may nevertheless be similar
in some respects, and these are not the results of the applicability
of a linguistic model. The similarities arise at those points where
the formal characteristics of each model are dominant over the substantive
materials it is ordering. Thus, the claim of productivity for a
specific model should not get confused with the nature of its contents
from the domain where it was first developed, as this promotes illegitimate
passions and questionable claims of similarities with the content
of other domains (see Shankman 1969 for a refutation of Levi-Strauss’s
linguistic model when applied to the domain of cooking).
In sum, I take the position that all methods of analysis should
be formalized and considered to be relative, or nominal, rather
than representing specific ontological reality. And finally they
should be then linked, if possible, with the operations used by
the members of the referent society we are attempting to analyse,
as Piaget has argued (1970).
If claims to knowledge were in fact contributions to knowledge,
synthetic structuralism would far out-produce the various other
approaches to anthropological knowledge, closely followed of course
by cognitive structuralism. So we must not be misled by claims to
knowledge into assuming that they do generate knowledge; nor must
we be misled into assuming that just because a particular question-set
has greater cognitive interest at a particular time within the discipline
of anthropology that it also generates greater contributions to
Furthermore, in attempting to assay these various claims to knowledge,
it seems to me that we as anthropologists must not dwell too deeply
and devote too much time to epistemological detective work. We may
just end up chasing our own tails, for there are many epistemological
questions that are just not resolvable at the present time, and
we have more important work to do!
The strength of anthropological inquiry has always been in the dialectical
relation between ethnographic investigation and theoretical formulation.
And what I have implied here is that this relation has become unbalanced.
The major contribution to Western knowledge of anthropological inquiry,
in my opinion, has been not its theoretical formulations but its
vast body of ethnographic data. But where are the students today
who are willing to devote two and three years of their lives to
the ethnography of a remote people? Where have all our Malinowskis
gone? The present time is indeed critical, for all over the world,
more rapidly than previously envisioned or experienced, the indigenous
societies are disappearing. In Borneo alone since World War II we
have averaged only 1.5 social-anthropological expeditions per year.
And we list urgent anthropological project after urgent anthropological
project and have not yet had one response! At this crucial time
I find it paradoxical that we are theorizing more and more in our
armchairs and doing less and less fieldwork of the quality and extent
that has been in the past the diacritical mark of the anthropologist.
Thus, it seems to me that we must somehow shift the centre of
concern in anthropological inquiry back to the field. This does
not mean that our conceptual tools are perfect. Indeed they are
not. But we cannot wait any longer to perfect them. In fact I would
argue that we can only perfect them by the close interplay between
new ethnographical data and theory, whereby better and more refined
theory is developed.
In other words, I am saying let us get on with the task of writing
ethnographies rather than critiques of critiques, with the thought
in mind that it is not the best theory that makes the best ethnographer.
Instead it takes a lot of time, a lot of practice, a lot of patience,
and a certain amount of mildew and dirt to develop a good ethnographer
and the good ethnographies. And perhaps by turning back to the field
we can also bring back to anthropology the joy and exuberance that
has been lost. We no longer sit up to all hours in enthusiastic
discussions of new discoveries and new experiences, but have settled
for a darkened image of the larger society and have become harried
cultural accountants, ridden with suburban anxieties.
8 Ardener is developing a more comprehensive theory of ‘structuralism’
in which the processing of social and ecological features into ‘p-structures’
(paradigmatic) and ‘s-structures’ (syntagmatic) is involved.
However, I have been unable to consult his most recent paper in
time for this conference.
9 However, Leach later on wrote “in this field the essential
innovation in Lévi-Strauss’s approach is the recognition
that mythological stories always exist as sets rather than isolates.
The individual members of the set constitute permutations of the
same theme” (1973:51).
10 Thus, in his interpretation of the Oedipus myth, Lévi-Strauss
brings in information for his interpretation from the mythology
of the American Indians: “This immediately helps us to understand
the meaning of the fourth column. In mythology it is a universal
characteristic of men born from the Earth that at the moment they
emerge from the depth they either cannot walk or they walk clumsily.
This is the case of the chthonian beings in the mythology of the
11 At still another level he is a confused idealist. If mind structures
reality, then arguments on the basis of the theoretical results
of other sciences, such as genetics, to prove the ontological basis
of binary oppositionism is pure nonsense. For if mind does structure
reality, then the study of the results of other scientific endeavours
is really only mind perceiving itself operating in another realm.
And thus the advocate commits the prime fallacy of disappearing
up his anus along with his argument.
12 This is so with respect to some cognitive structuralists, but
others, ignoring the implications of their own theory and the logic
of their own arguments, have made formal analyses of data gathered
by ethnographers who did not use the methods of cognitive structuralism
(see Appell 1973 for a discussion of this).
Aberle, David F. 1960. The influence of linguistics on early culture
and personality theory. In Gertrude E. Dole and Robert L. Carneiro
(eds), Essays in the science of culture. New York: Thomas Y. Crowell.
Agar, Michael. 1973. Ripping and running: a formal ethnography
of urban heroin addicts. New York: Seminar Press.
Appell, G.N. 1965. The nature of social groupings among the Rungus
Dusun of Sabah, Malaysia. Ph.D. thesis, The Australian National
1967. Observational procedures for identifying kindreds: social
isolates among the Rungus of Borneo. Southwestern Journal of Anthropology
1969. Social anthropological census for cognatic societies and
its application to the Rungus of northern Borneo. Bijdragen tot
de Taal-, Land- en Volkenkunde 125:80-93.
1973. The distinction between ethnography and ethnology and other
issues in cognitive structuralism. Bijdragen tot de Taal-, Land-
en Yolkenkunde 129:1-56.
1974a. An overview of research in Sarawak. Studies in Third World
Societies No. 2.
1974b. The analysis of property systems: the creation and devolution
of property interests among the Rungus of Borneo. Paper delivered
at the Association of Social Anthropologists, Keele, 27-30 March,
n.d. Observational procedures for social structure: the comparative
analysis of three cognatic societies of Borneo.
Ardener, Edwin. 1971. The new anthropology and its critics. Man
Barnes, J.A. 1969. Networks and political processes. In J. Clyde
Mitchell (ed.), Social networks in urban situations. Manchester:
Manchester University Press.
1971. Three styles in the study of kinship. London: Tavistock.
Barth, Frederic. 1966. Models of social organization (RAI Occasional
Paper 23). London: Royal Anthropological Institute.
Bateson, Gregory. 1972. Steps to an ecology of mind. San Francisco:
1979. Mind and nature: a necessary unity. New York: Dutton.
Berlin, Brent, Breedlove, Dennis E., and Raven, Peter H. 1973.
General principles of classification and nomenclature in folk biology.
American Anthropologist 75:214-42.
Bott, Elizabeth. 1972. A rejoinder to Edmund Leach. In J.S. LaFontaine
(ed.), The interpretation of ritual. London: Tavistock.
Boudon, Raymond. 1971. The uses of structuralism (trans. Michalania
Vaugn). London: Heinemann.
Bright, Jane O. and Bright, William. 1965. Semantic structures
in northwestern California and the Sapir-Whorf hypothesis. In E.A.
Hammel (ed.), Formal semantic analysis. American Anthropologist
(Special Publication) 67.
Burling, Robbins. 1964a. Cognition and componential analysis: God’s
truth or hocus-pocus? American Anthropologist 66:20-8.
1964b. Burling’s rejoinder. American Anthropologist 66:120-2.
1969. Linguistics and ethnographic description. American Anthropologist
Colby, B.N. 1966. Ethnographic semantics: a preliminary survey.
Current Anthropology 7:3-32.
Conklin, Harold. 1964. Ethnogenealogical method. In Ward H. Goodenough
(ed.), Explorations in cultural anthropology. New York: McGraw-Hill.
1967. Lexicographical treatment of folk taxonomies. In F.W. Householder
and Sol Saporta (eds), Problems in lexiography. Bloomington: Indiana
Dentan, Robert K. 1970. Labels and rituals in Semai classification.
Diebold, A. Richard, Jr. 1966. The reflection of coresidence in
Mareño kinship terminology. Ethnology 5:37-79.
1973. Cognitive anthropology. In J. J. Honigmann (ed.), Handbook
of social and cultural anthropology. Chicago: Rand McNally.
Firth, Raymond. 1963. Bilateral descent groups: an operational
viewpoint. In I. Schapera (ed.), Studies in kinship and marriage.
London: Royal Anthropological Institute.
1967. Tikopia ritual and belief. London: George Allen and Unwin.
Frake, Charles O. 1962. The ethnographic study of cognitive systems.
In Thomas Gladwin and William C. Sturtevant (eds), Anthropology
and human behavior. Washington: Anthropological Society of Washington.
1964. A structural description of Subanun “religious behavior”.
In Ward H. Goodenough (ed.), Explorations in cultural anthropology.
New York: McGraw-Hill.
1969. Struck by speech: the Yakan concept of litigation. In Laura
Nader (ed.), Law and culture in society. Chicago: Aldine.
Garbett, G. Kingsley. 1970. The analysis of social situations.
Goodenough, Ward H. 1951. Property, kin and community on Truk (Yale
University Publications in Anthropology 46). New Haven, Conn.: Yale
1955. A problem in Malayo-Polynesian social organization. American
1957. Cultural anthropology and linguistics. In Paul L. Garvin
(ed.), Report on the seventh annual round table meetings on linguistics
and language study (Monograph Series in Language and Linguistics
9). Washington: Georgetown University Press.
1961. Review of G.P. Murdock (ed.), Social structure in Southeast
Asia. American Anthropologist 63:1341-7.
1964a. Introduction. In Ward H. Goodenough (ed.), Explorations
in cultural anthropology. New York: McGraw-Hill.
1964b. Componential analysis of Konkama Lapp kinship terminology.
In Ward H. Goodenough (ed.), Explorations in cultural anthropology.
New York: McGraw-Hill.
1965. Yankee kinship terminology: a problem in componential analysis.
In E.A. Hammel (ed.), Formal semantic analysis. American Anthropologist
(Special Publication) 67.
Gulliver, P.H. 1971. Neighbours and networks. Berkeley: University
of California Press.
Hallowell, A. Irving. 1960. Ojibwa ontology, behavior, and world
view. In Stanley Diamond (ed.), Culture and history. New York: Columbia
Kaplan, Abraham. 1964. The conduct of inquiry. San Francisco: Chandler.
1965. Noncausal explanation. In Daniel Lerner (ed.), Cause and
effect. New York: Free Press.
Kay, Paul. 1970. Some theoretical implications of ethnographic
semantics. In A. Fischer (ed.), Current directions in anthropology
(Bulletin of the American Anthropological Association 3 (3):Pt.2).
Keesing, Roger M. 1972. Paradigms lost: the new ethnography and
the new linguistics. Southwestern Journal of Anthropology 28:299-332.
Kroeber, A.L. 1974. Cultural and natural areas of native north
America. Berkeley: University of California Press.
Lane, Michael. 1970. Introduction. In Michael Lane (ed.), Structuralism:
a reader. London: Jonathan Cape.
Leach, Edmund. 1961. Rethinking anthropology (LSE Monographs on
Social Anthropology 22). London: Athlone.
1964. Anthropological aspects of language: animal categories and
verbal abuse. In Eric H. Lenneberg (ed.), New directions in the
study of language. Cambridge, Mass.: MIT Press.
1970a. Telstar and the Aborigines or La pensee sauvage (translated
and reprinted). In Dorothy Emmet and Alasdair MacIntyre (eds), Sociological
theory and philosophical analysis. London: Macmillan.
1970b. Lévi-Strauss. London: Fontana/Collins.
1971. Kimil: a category of andamanese thought. In Pierre Maranda
and Elli Kongas Maranda (eds), Structural analysis of oral tradition.
Philadelphia: University of Pennsylvania Press.
1972. The structure of symbolism. In J.S. LaFontaine (ed.), The
interpretation of ritual. London: Tavistock.
1973. Structuralism in social anthropology. In David Robey (ed.),
Structuralism: an introduction. Oxford: Clarendon.
Lévi-Strauss, Claude. 1963a. Structural anthropology. New
York: Basic Books.
1963b. Totemism. Boston: Beacon Press.
1963c. The bear and the barker. Journal of the Royal Anthropological
Lyons, Harriet D. n.d. Moses Maimonides: bad anthropologist or
good native? Jewish laws of ritual purity and the problem of conscious
model. Paper delivered at SUNY Conference on Structure and Cognition,
Geneseo, 13 April 1974.
Mair, Lucy. 1965. An introduction to social anthropology. Oxford:
Maranda, Pierre. 1972a. Introduction. In P. Maranda (ed.), Mythology.
Baltimore: Penguin Books.
Mitchell, J. Clyde. 1969. The concept and use of social networks.
In J. Clyde Mitchell (ed.), Social networks in urban situations.
Manchester: Manchester University Press (for Institute of African
Studies, University of Zambia).
Moerman, Michael. 1966. Kinship and commerce in a Thai-Lue village.
Nadel, S.F. 1953. The foundations of social anthropology. Glencoe:
1954. Nupe religion. Glencoe: Free Press.
Needham, Rodney. 1971. Introduction. In Rodney Needham (ed.), Rethinking
kinship and marriage (ASA Monoqraph 11). London: Tavistock
O’Neill, John. 1973. Modes of individualism and collectivism.
Paine, Robert. 1973. Transactions as communicative events. Paper
delivered at the meeting of the Association of Social Anthropologists,
Piaget, Jean. 1970. Structuralism. New York: Basic Books.
Radcliffe-Brown, A.R. 1950. Introduction. In A.R. Radcliffe-Brown
and Daryll Forde (eds), African systems of kinship and marriage.
London: Oxford University Press.
1952. Structure and function in primitive society. Glencoe: Free
Rivière, P.G. 1971. Marriage: a reassessment. In Rodney
Needham (ed.), Rethinking kinship and marriage (ASA Monograph 11).
Rossi, Ino. 1973. The unconscious and the anthropology of Claude
Le’vi-Strauss. American Anthropologist 75:20-48.
Shankman, Paul. 1969. Le roti et le bouille: Lévi-Strauss’s
theory of cannibalism. American Anthropologist 71:54-69.
Strauss, Anselm et al. 1964. Psychiatric ideologies and institutions.
New York: Free Press.
Swartz, M.J. 1968. Introduction. In M.J. Swartz (ed.), Local-level
Politics. Chicaqo: Aldine.
Tyler, Stephen A. 1969. Introduction. In S.A. Tyler (ed.), Cognitive
anthropology: readings. New York: Holt, Rinehart and Winston.
Van Velson, J. 1967. The extended-case method in situational analysis.
In A.L. Epstein (ed.), The craft of social anthropology. London:
Wallace, Anthony F. 1961. Culture and personality. New York: Random
1962. Culture and cognition. Science 135:351-7.
1965. The problem of the psychological validity of componential
analysis. In E.A. Hammel (ed.), Formal semantic analysis. American
Anthropologist (Special Publication) 67.
Werner, Oswald and Fenton, Joann. 1970. Method and theory in ethnoscience
or ethnoepistemology. In R. Naroll and R. Cohen (eds), A handbook
of method in cultural anthropology. New York: Natural History Press.
Wilden, Anthony. 1972. System and structure: essays in communication
and exchange. London: Tavistock.
Wyllie, Robert W. 1968. Ritual and social change: a Ghanaian example.
American Anthropologist 70:21-33.