LeviStrauss1951

Language and the Analysis of Social Laws

Claude Lévi-Strauss 
American Anthropologist, New Series, Vol. 53, No. 2. (Apr. -Jun., 1951), pp. 155-163. 
Stable URL: JSTOR Link 

Can there be a cybernetic anthropology?

Wiener doesn't think so

IN A recent work, whose importance from the point of view of the future 
of the social sciences can hardly be overestimated, Wiener poses, and re- 
solves in the negative, the question of a possible extension to the social sciences 
of the mathematical methods of prediction which have made possible the 
construction of the great modern electronic machines. He justifies his position 
by two arguments.[1]

The problem of reflexivity (as I call it)

In the first place, he maintains that the nature of the social sciences is 
such that it is inevitable that their very development must have repercussions 
on the object of their investigation. The coupling of the observer with the 
observed phenomenon is well known to contemporary scientific thought, and, 
in a sense, it illustrates a universal situation. But it is negligible in fields which 
are ripe for the most advanced mathematical investigation; as, for example, 
in astrophysics, where the object has such vast dimensions that the influence 
of the observer need not be taken into account, or in atomic physics, where the 
object is so small that we are only interested in average mass effects in which 
the effect of bias on the part of the observer plays no role. In the field of the 
social sciences, on the contrary, the object of study is necessarily affected by 
the intervention of the observer, and the resulting modifications are on the 
same scale as the phenomena that are studied.

The problem of the short run

In the second place, Wiener observes that the phenomena subjected to 
sociological or anthropological inquiry are defined within our own sphere of 
interests; they concern questions of the life, education, career, and death of 
individuals. Therefore the statistical runs available for the study of a given 
phenomenon are always far too short to lay the foundation of a valid induction. 
Mathematical analysis in the field of social sciences, he concludes, can bring 
results which should be of as little interest to the social scientist as those of 
the statistical study of a gas would be to an individual about the size of a 
molecule.

But Wiener focuses on one kind of data

These objections seem difficult to refute when they are examined in terms 
of the investigations toward which their author has directed them, the data of 
research monographs and of applied anthropology. In such cases, we are deal-
ing with a study of individual behavior, directed by an observer who is him- 
self an individual; or with a study of a culture, a national character, or a pat- 
tern, by an observer who cannot dissociate himself completely from his culture, 
or from the culture out of which his working hypotheses and his methods of 
observation, which are themselves cultural patterns, are derived. 

We should use language as our model

There is, however, at least one area of the social sciences where Wiener's 
objections do not seem to be applicable, where the conditions which he sets 
as a requirement for a valid mathematical study seem to find themselves 
rigorously met. This is the field of language, when studied in the light of struc- 
tural linguistics, with particular reference to phonemics. 

Language not affected by reflexivity

• Language's rules are unconscious and unaffected by awareness of them

Language is a social phenomenon; and, of all social phenomena, it is the 
one which manifests to the greatest degree two fundamental characteristics 
which make it susceptible of scientific study. In the first place, much of lin- 
guistic behavior lies on the level of unconscious thought. When we speak, we 
are not conscious of the syntactic and morphological laws of our language. 
Moreover, we are not ordinarily conscious of the phonemes that we employ 
to convey different meanings; and we are rarely, if ever, conscious of the phono- 
logical oppositions which reduce each phoneme to a bundle of differential 
features. This absence of consciousness, moreover, still holds when we do be- 
come aware of the grammar or the phonemics of our language. For, while 
this awareness is but the privilege of the scholar, language, as a matter of fact, 
lives and develops only as a collective construct; and even the scholar's lin- 
guistic knowledge always remains dissociated from his experience as a speaking 
agent, for his mode of speech is not affected by his ability to interpret his lan- 
guage on a higher level. We may say, then, that as concerns language, we need 
not fear the influence of the observer on the observed phenomenon, because 
the observer cannot modify the phenomenon merely by becoming conscious 
of it. 

Language is ancient

Furthermore, as regards Wiener's second point, we know that language 
appeared very early in human history. Therefore, even if we can study it 
scientifically only when written documents are available, writing itself goes 
back a considerable distance, and furnishes long enough runs to make language 
a valid subject for mathematical analysis. For example, the series we have at 
our disposal in studying Indo-European, Semitic or Sino-Thibetan languages is 
about four or five thousand years old. And, where a comparable temporal 
dimension is lacking, the multiplicity of coexistent forms furnishes, for several 
other linguistic families, a spatial dimension that is no less valuable. 

We thus find in language a social phenomenon which manifests both in- 
dependence of the object and long statistical runs; which would seem to indi- 
cate that language is a phenomenon fully qualified to satisfy the demands of 
mathematicians for the type of analysis Wiener suggests. 

Language lends itself to computation

• Note L-S's conflation of cybernetics with computation
• L-S shows his interest in database technology here (as elsewhere)
• Basic idea: structure = elements + rules of combination

It is, in fact, difficult to see why certain linguistic problems could not be 
solved by modern calculating machines. With knowledge of the phonological 
structure of a language and the laws which govern the grouping of consonants 
and vowels, a student could easily use a machine to compute all the combina- 
tions of phonemes constituting the words of n syllables existing in the vocabu- 
lary, or even the number of combinations compatible with the structure of 
the language under consideration, such as previously defined. With a machine 
into which would be "fed" the equations regulating the types of structures 
with which phonemics usually deals, the repertory of sound which human 
speech organs can emit, and the minimal differential values, determined by 
psycho-physiological methods, which distinguish between the phonemes closest 
to one another, one would doubtless be able to obtain a computation of the 
totality of phonological structures for n oppositions (n being as high as one 
wished). One could thus construct a sort of periodic table of linguistic struc- 
tures that would be comparable to the table of elements which Mendeleieff 
introduced into modern chemistry. It would then only remain for us to check 
the place of known languages in this table, to identify the positions and the 
relationships of the languages whose first-hand study is still too imperfect to 
give us a proper theoretical knowledge of them, and to discover the place of 
languages that have disappeared, are unknown, yet to come,, or simply possible.

An example from Jakobson

• Metastructure
• "Law of the Group" (See L-S 1955)

To add a last example: Jakobson has recently suggested that a language 
may possess several coexisting phonological structures, each of which may 
intervene in a different kind of grammatical operation.[2] Since there must ob- 
viously be a relationship between the different structural modalities of the same 
language, we arrive at the concept of a "metastructure" which would be some- 
thing like the law of the group (loi du groupe) consisting of its modal structures. 
If all of these modalities could be analyzed by our machine, established mathe- 
matical methods would permit it to construct the "metastructure" of the 
language, which would in certain complex cases be so intricate as to make it 
difficult, if not impossible, to achieve on the basis of purely empirical investi- 
gation.

The Model of Language

• Language = phonemes + rules of combination
• "Universal laws which regulate the unconscious activities of the mind"
• Criteria for structure:
  1. is unconscious
  2. has elements
  3. has rules
  4. does communication
• Grammar is a good synonym

The problem under discussion here can, then, be defined as follows. Among 
all social phenomena, language alone has thus far been studied in a manner 
which permits it to serve as the object of truly scientific analysis, allowing us 
to understand its formative process and to predict its mode of change. This 
results from modern researches into the problems of phonemics, which have 
reached beyond the superficial conscious and historical expression of linguistic 
phenomena to attain fundamental and objective realities consisting of systems 
of relations which are the products of unconscious thought processes. The 
question which now arises is this: is it possible to effect a similar reduction 
in the analysis of other forms of social phenomena? If so, would this analysis 
lead to the same result? And if the answer to this last question is in the affirma-
tive, can we conclude that all forms of social life are substantially of the same 
nature-that is, do they consist of systems of behavior that represent the pro- 
jection, on the level of conscious and socialized thought, of universal laws which 
regulate the unconscious activities of the mind? Obviously, no attempt can be 
made here to do more than to sketch this problem by indicating certain points 
of reference and projecting the principal lines along which its orientation might 
be effective. 

The example of Kroeber on fashion

• About Kroeber's essay:
  • Women’s dress fashion over 300 years
  • Skirt length changed in a periodic cycle.
  • No direct cause (political or economic cycles) -- cannot be explained by outside factors; evolves according to its own internal laws.
  • This independent cultural realm = the "superorganic"

Some of the researches of Kroeber appear to be of the greatest importance 
in suggesting approaches to our problem, particularly his work on changes in 
the styles of women's dress[3] Fashion actually is, in the highest degree, a phe- 
nomenon which depends on the unconscious activity of the mind. We rarely 
take note of why a particular style pleases us, or falls into disuse. Kroeber has 
demonstrated that this seemingly arbitrary evolution follows definite laws. 
These laws cannot be reached by purely empirical observation, or by intuitive 
consideration of phenomena, but result from measuring some basic relation- 
ships between the various elements of costume. The relationship thus obtained 
can be expressed in terms of mathematical functions, whose values, calculated 
at a given moment, make prediction possible.

Teissier on zoology

Kroeber has thus shown how even such a highly arbitrary aspect of social 
behavior is susceptible of scientific study. His method may be usefully com- 
pared not only with that of structural linguistics, but also with that of the 
natural sciences. There is a remarkable analogy between these researches and 
those of a contemporary biologist, G. Teissier, on the growth of the organs of 
certain crustaceans.[4] Teissier has shown that, in order to formulate the laws 
of this growth, it has been necessary to consider the relative dimensions of 
the component parts of the claws, and not the exterior forms of these organs. 
There, relationships allow us to derive constants-termed parameters-out 
of which it is possible to derive the laws which govern the development of these 
organisms. The object of a scientific zoology, in these terms, is thus not ulti- 
mately concerned with the forms of animals and their organs as they are usual- 
ly perceived, but is to establish certain abstract and measurable relationships, 
which constitute the basic nature of the phenomena under study. 

Levi-Strauss on kinship

• Describes results from The Elementary Structures of Kinship.
• Elements = partners, groups, households
• Rules = incest prohibition, marriage rules, inheritance, locality

An analogous method has been followed in studying certain features of 
social organization, particularly marriage rules and kinship systems.[5] It has 
been shown that the complete set of marriage regulations operating in human 
societies, and usually classified under different headings such as incest prohibi- 
tions, preferential forms of marriage, and the like, can be interpreted as being 
so many different ways of insuring the circulation of women within the social 
group, or, of substituting the mechanism of a sociologically determined affinity 
for that of a biologically determined consanguinity. Proceeding from this 
hypothesis, it would only be neccesary to make a mathematical study of every 
possible type of exchange between npartners to enable one almost automatical- 
ly to arrive at every type of marriage rule actually operating in living societies 
and, eventually, to discover others which are merely possible; one would also 
understand their function and the relationships between each type and the 
others. 

Reciprocity as special case of exchange

This approach was fully validated by the demonstration, reached by pure 
deduction, that the mechanisms of reciprocity known to classical anthropology 
-- namely, those based on dual organization and exchange-marriage between 
two partners or whose number is a multiple of two-are but a special instance 
of a wider kind of reciprocity between any number of partners. This fact has 
tended to remain unnoticed, because the partners in those matings, instead 
of giving and receiving from one another, do not give to those from whom they 
receive, and do not receive from those to whom they give.,They give to and 
receive from different partners to whom they are bound by a relationship that 
operates only in one direction. 

This type of organization, no less important than the moiety system, has 
thus far been observed and described only imperfectly and incidentally. Start- 
ing with the results of mathematical study, data had to be compiled; thus, the 
real extension of the system was shown and its first theoretical analysis ffered.[6] 
At the same time, it became possible to explain the more general features of 
marriage rules such as preferential marriage between bilateral cross-cousins 
or with only one kind of cross-cousin, on the father's side (patrilateral), or on 
that of the mother (matrilateral). Thus, for example, though such customs had 
been unintelligible to anthropologists,[7] they were perfectly clear when regarded 
as illustrating different modalities of the laws of exchange. In turn, these were 
reduced to a still more basic relationship between the rules of residence and 
the rules of descent.[8]

The result of viewing kinship and marriage as communication

• Clothing and women are signals

Now, these results have only been achieved by treating marriage regula- 
tions and kinship systems as a kind of language, a set of processes permitting 
the establishment, between individuals and groups, of a certain type of com- 
munication. That the mediating factor, in this case, should be the women of 
the group, who are circulated between clans, lineages, or families, in place of 
the words of the group, which are circulated between individuals, does not at all 
change the fact that the essential ~pect of the phenomenon is identical in 
both cases. 

The Origin of Language

Kinship ay help shed light on the origin of language

• Kinship is more static and has more data
• If it is a communication system, and if all communication systems share a common code, then kinship theory can shed light on language
• "The original impulse which compelled men to exchange words must be sought for in that split-representation which pertains to the symbolic function."
  • Split-representation = meaning + value?

We may now ask whether, in extending the concept of communication so as 
to make it include exogamy and the rules flowing from the prohibition of in- 
cest, we may not, reciprocally, achieve insight into a problem that is still very 
obscure, that of the origin of language. For marriage regulations, in relation 
to language, represent a complex much more rough and archaic than the latter. 
It is generally recognized that words are signs: but poets are practically the 
only ones who know that words have also been values. As against this, women 
are held by the social group to be values of the most essential kind, though we 
have difficulty in understanding how these values become integrated in systems 
endowed with a significant function. This ambiguity is clearly manifested in 
the reactions of persons who, on the basis of the analysis of social structures 
referred to,[9] have laid against it the charge of "anti-feminism," because women 
are referred to as objects.[10] Of course, it may be disturbing to some to have 
women conceived as mere parts of a meaningful system. However, one should 
keep in mind that the processes by which phonemes and words have lost -- even 
though in an illusory manner -- their character of value, to become reduced 
to pure signs, will never lead to the same results in matters concerning women. 
For words do not speak, while women do; as producers of signs, they can never 
be reduced to the status of symbols or tokens. But it is for this very reason 
that the position of women, as actually found in this system of communication 
between men that is made up of marriage regulations and kinship nomenclature, 
may afford us a workable image of the type of relationships that could have 
existed at a very early period in the development of language, between human 
beings and their words. As in the case of women, the original impulse which 
compelled men to exchange words must be sought for in that split-representa- 
tion which pertains to the symbolic function. For, since certain terms are 
simultaneously perceived as having a value both for the speaker and the 
listener, the only way to resolve this contradiction is in the exchange of comple- 
mentary values, to which all social existence reduces itself. 

Hypothesis: are different aspects of social life generated by the system code?

These speculations may be judged utopian. Yet, granting that the assump- 
tions made here are legitimate, a very important consequence follows that is 
susceptible of immediate verification. That is, the question may be raised 
whether the different aspects of social life (including even art and religion) 
can not only be studied by the methods, and with the help of concepts similar 
to those employed in linguistics, but also whether they do not constitute 
phenomena whose inmost nature is the same as that of language. That is, in 
the words of Voegelin, we may ask whether there are not only "operational" 
but also "substantial comparabilities" between language and culture.[11]

To verify, go deep ...

How can this hypothesis be verified? It will be necessary to develop the 
analysis of the different features of social life, either for a given society or for 
a complex of societies, so that a deep enough level can be reached to make it 
possible to cross from one to the other; or to express the specific structure of 
each in terms of a sort of general language, valid for each system separately 
and for all of them taken together. It would thus be possible to ascertain if 
one had reached their inner nature, and to determine if this pertained to the 
same kind of reality. In order to develop this point, an experiment can be at- 
tempted. It will consist, on the part of the anthropologist, in translating the 
basic features of the kinship systems from different parts of the world in terms 
general enough to be meaningful to the linguist, and thus be equally applicable 
by the latter to the description of the languages from the same regions. Both 
could thus ascertain whether or not different types of communication systems 
in the same societies -- that is, kinship and language -- are or are not caused by 
identical unconscious structures. Should this be the case, we would be assured 
of having reached a truly fundamental formulation. 

The data

If then, a substantial identity were assumed to exist between language 
structure and kinship systems, one should find, in the following regions of the 
world, languages whose structures would be of a type comparable to kinship 
systems in the following terms: 

Indo-Eurpopean

• Simple terms, many combiniations

1. Indo-European: As concerns the kinship systems, we find that the mar- 
riage regulations of our contemporary civilization are entirely based on the 
principle that, a few negative prescriptions being granted, the density and 
fluidity of the population will achieve by itself the same results which other 
societies have sought in more complicated sets of rules; i.e. social cohesion 
obtained by marriage in degrees far removed or even impossible to trace. 
This statistical solution has its origin in a typical feature of most ancient Indo- 
European systems. These belong, in the author's terminology, to a simple 
formula of generalized reciprocity (formule simple de 1'6change g6n6ralis6).[12]
However, instead of prevailing between lineages, this formula operates be- 
tween more complex units of the brastsvo type, which actually are clusters 
of lineages, each of which enjoys a certain freedom within the rigid framework 
of general reciprocity in effect at the level of the cluster. Therefore, it can be 
said that a characteristic feature of Indo-European kinship structure lies in 
the fact that a problem set in simple terms always admits of many solutions. 
Should the linguistic structure be homologous with the kinship structure 
it would thus be possible to express the basic feature of Indo-European lan- 
guages as follows: The languages have simple structures, utilizing numerous 
elements. The opposition between the simplicity of the structure and the multi- 
plicity of elements is expressed in the fact that several elements compete to 
occupy the same positions in the structure.

Sino-Thibetan

• Simple elements, complex structures

2. Sino-Thibetan kinship systems exhibit quite a different type of complex- 
ity. They belong to or derive directly from the simplest form of general reci- 
procity, namely mother's brother's daughter marriage, so that, as has been 
shown,[13] while this type of marriage insures social cohesion in the simplest 
way, at the same time it permits this to be indefinitely extended so as to in- 
clude any number of participants. 

Translated into more general terms applicable to language that would 
correspond to the following linguistic pattern, we may say that the structure 
is complex, while the elements are few, a feature that may be related to the 
tonal structure of these languages.

African

• Intermediate between 1 and 2

3. The typical feature of African kinship systems is the extension of the 
bride-wealth system, coupled with a rather frequent prohibition on marriage 
with the wife's brother's wife. The joint result is a system of general reciproc- 
ity already more complex than the one with the mother's brother's daughter, 
while the types of unions resulting from the circulation of the marriage-price 
approaches, to some extent, the statistical mechanism operating in our own 
society. 

Therefore one could say that African languages have several modalities 
corresponding in general to a position intermediate between 1) and 2).

Oceanic

• Simple structure, few elements

4. The widely recognized features of Oceanic kinship systems seem to lead 
to the following formulation of the basic characteristics of the linguistic pat- 
tern: simple structure and few elements.

American

• Many elements, simple structures

5. The originality of American kinship systems lies with the so-called Crow- 
Omaha type which should be carefully distinguished from other types showing 
-the same disregard for generation levels.[14] The important point with the Crow- 
Omaha type is not that two kinds of cross-cousins are classified in different 
generation levels, but rather that they are classified with consanguineous kin 
instead of with affinal kin as it occurs, for instance, in the Miwok system. But 
systems of the Miwok type belong equally to the Old and the New World; 
while when considering the differential systems just referred to as Crow-
Omaha, one must admit that, apart from a few exceptions, these are only typi- 
cal for the New World. It can be shown that this quite exceptional feature of 
the Crow-Omaha system results from the simultaneous application of the two 
simple formulas of reciprocity, both special and general (tchange restreint and 
echange generalise),[15] which elsewhere in the world were generally considered 
to be incompatible. It thus became possible to achieve marriage within remote 
degrees by using simultaneously two simple formulas, each of which independ- 
ently applied could only have led to different kinds of cross-cousin marriages. 
The linguistic pattern corresponding to that situation would be that cer- 
tain of the American languages offer a relatively high number of elements, 
which succeed in becoming organized into relatively simple structures by com- 
pelling these to assume an asymmetrical form.

Concluding remarks

Up to linguists to interpret the test

It must be kept in mind that in the above highly tentative experiment, the 
anthropologist proceeds from what is known to what is unknown to him: 
namely from kinship structures to linguistic structures. Whether or not the 
differential characteristics thus outlined have a meaning in so far as the re- 
spective languages are concerned, remains for the linguist to decide. The author, 
being a social anthropologist, and not a linguist, can only try to explain briefly 
to which specific features of kinship systems he is referring in this attempt 
toward a generalized formulation. Since the general lines of his interpretation 
have been fully developed elsewhere,[16] short sketches were deemed sufficient 
for the purpose of this paper. 

If valid, then we are much closer to the goal

If the general characteristics of the kinship systems of given geographical 
areas, which we have tried to bring into juxtaposition with equally general 
characteristics of the linguistic structures of those areas, are recognized by 
linguists as an approach to equivalences of their own observations, then it 
will be apparent, in terms of our preceding discussion, that we are much closer 
to the understanding of the fundamental characteristics of social life than we 
have been accustomed to think. 

Toward a comparative structural analysis of institutions

The road will then be open for a comparative structural analysis of customs, 
institutions, and accepted patterns of behavior. We will be in a position 
to understand basic similarities between forms of social life, such as language, 
art, law, religion, that, on the surface, seem to differ greatly. At the same time, 
we will have the hope of overcoming the opposition between the collective 
nature of culture and its manifestations in the individual, since the so-called 
"collective consciousness" would, in the final analysis, be no more than the 
expression, on the plane of individual thought and behavior, of certain time and 
space modalities of these universal laws which make up the unconscious 
activity of the mind.

ENDNOTES

[1] Wiener, N., 1948, p. 189-191. 
[2] Jakobson, R.,1948. 
[3] Kroeber, A. L. and Richardson, J., 1940. 
[4] Teissier, G., 1936. 
[5] Levi-Strauss,C., 1949,passim. 
[6] Ibid., pp. 278-380. 
[7] Ibid., pp. 558-566. 
[8] Ibid., pp. 547-550. 
[9] Ibid., p. 616. 
[10] Ibid., p. 45 sq. 
[11] "Language and Culture: substantial and operational comparabilities" was the title given 
by C. F. Voegelin to the symposium held at the 29th International Congress of Arnericanists, 
New York, 5-12 September, 1949, where these reflections were first offered. 
[12] LBvi-Strauss, C., 1949, pp. 583-591. 
[13] Ibid., 1949, pp. 291-380. 
[14] From this point of view, G. P. Murdock's suggestion that the Crow-Omaha type be merged 
with the Miwok type (1949, pp. 224,340) should be challenged. 
[15] Levi-Strauss, C., 1949, pp. 228-233.
[16] Ibid.

BIBLIOGRAPHY

JAKOBSON, R., 1948, The phonemic and grammatical aspect of language in their interrelations. 
Actcs du 6O Congrds Internetional des linguistes, Paris. 


KBOEBER, A. L., and J. RICHARDSON, 1940, Three centuries of women's dress fashions. Anthro- 
pologicd Records, Berkeley. 


LEVI-STRAUSS, C., 1949, La Structures Elementaires de la Parente, Paris. 


MURDOCH, G. P., 1949, Social Structure, New York. 


TEISSIER, G., 1936, La description mathkmatique des faits biologiques, Revue de M6taphysique 
et de Morale, Paris, Jan. 


WIENER, N., 1948, Cybernetics, or Control and Communication in the Animal and the Machine, 
Paris, Cambridge, New York.