Vous êtes sur la page 1sur 7

54.

The Dravidian Kinship Terminology as an Expression of Marriage


Author(s): L. Dumont
Reviewed work(s):
Source: Man, Vol. 53 (Mar., 1953), pp. 34-39
Published by: Royal Anthropological Institute of Great Britain and Ireland
Stable URL: http://www.jstor.org/stable/2794868 .
Accessed: 30/11/2012 00:39
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at .
http://www.jstor.org/page/info/about/policies/terms.jsp

.
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of
content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms
of scholarship. For more information about JSTOR, please contact support@jstor.org.

Royal Anthropological Institute of Great Britain and Ireland is collaborating with JSTOR to digitize, preserve
and extend access to Man.

http://www.jstor.org

This content downloaded by the authorized user from 192.168.82.203 on Fri, 30 Nov 2012 00:39:46 AM
All use subject to JSTOR Terms and Conditions

Nos. 53. 54

Man

parchment, as has been suggested by Fran,oise Henry.25


Mr. C. H. Roberts hasexpressedthe following opinion: 'If
we may assumethe existence of pre-Christian"books" in
Ireland, we should expect them to be in the roll form.
Equallyone would expect the codex form to be introduced
by the earliestChristianmissionaries.'On the other hand it
is strange that Theodor Birt does not refer at all to this
unusualway of holding a roll.26The posturemay, however,
have a symbolical meaning as yet unknown to us.
The contrast between the lifelike sculptureof the seated
ecclesiastic and the other three stereotyped figures is
striking. I should like to suggest as a hypothesis that the
latter are copies of seventh-centurysculptures.Similar sets
of figures may have existed in other more accessible
churches,where they suffereddestructionwhilst the White
Islandfigures survived owing to their isolation. If this was
the case, a local sculptor may have been commissioned to
add furtherportraitsof persons closely associatedwith this
church. This would (i) explain the single head (Plate Cd,
left) and the rejected stone mentioned above; and (ii)
bridge the time gap between the seventh-century original
statuesand the younger text of The TripartiteLife.
The question now ariseswhether the three seventh-century sculptures, or copies of them, may not even have
inspired some of the above quoted passages in The TripartiteLife. We cannot, of course, offer any definite answer,
but may at least seem justified in adducing those texts as
pertinent evidence. The compilers of The TripartiteLife
may, indeed, have remembered that the figures of White
Island, or possibly similar copies, or the originals themselves were representationsof St. Patrick, King Loiguire
and King Enna.27
I Eric

MARCH, I953
3

The PreliminarySurvey of the AncientMonumentsof Northern

Ireland (Belfast, I940), p. I48.


4 A.J. Butler, TheAncientCopticChurches
ofEgypt(Oxford, I884),
Vol. II, fig. 20, p. IIO.
5 IrishMonasticism
(Dublin, I93I), p. i85.
6 Ed. tr. Wh. Stokes (London, i887), pp. 75, 567.
7 The use of pairsof animalsin substitutionritualis alsomentioned
by ProfessorS. H. Hooke, 'The Theory and Practiceof Substitution', Vetusrestamentum,
Vol. II, No. I, pp. 6-8.
8 The MichaelmasSheep Story,' Tripartite
Life, op. cit.,pp. ss7f.;
James F. Kenney, The Sourcesof the EarlyHistoryof Ireland(New
York, I929), p. 350.
9 'The Iconographyof St. Patrick,'Down and ConnorHistorical
Society'sjournal,Vol. VII (Belfast,I936).
IO Op. cit., pp. i83, 326.
I" Ibid.,pp. 35, i64, 22I.
I2J. A. MacCulloch,'Tonsure,'E.R.E., Vol. XII, p. 386a.
I3 'The Form of the CelticTonsure,'P.S.A.S., Vol. XXX, p. 326.
I4

Op. cit., Vol. II, p. 229.

H. F. McClintock, Old Irish and HighlandDress, 2nd ed.


(Dundalk,I950), Plate 8; Sexton, op. cit., figs. 48, 52, 55.
i6 Op. cit., p. 79.
I7 TheBanquet
of Dun-nan-Gedh(Dublin, I842), note on pp. 38f.
I8 The Life of St. Gall (London, I927), p, 73, note I.
'9 Op. cit., pp. II4, I20f, 476-9.
I5

Ibid., p. 249.
2nd ed. (Leipzig, I884), note on p. 42, quoted from Morgenblatt, i853, No. 34, p. 4I5. See alsoJ.R.S.A.I., Vol. XXX, p. 237,
20

21

T. J. Westropp, 'The Clog an Oir, or Bell Shrineof Scattery.'The


Secretaryof the Royal IrishAcademykindly informedme by letter
that 'the Minutes of the R.I.A. for iS March, igig, contain three
pagesrelatingto the history of the bell. ...'
22Adolf Mahr, ChristianArt in Ireland(Dublin, I932), plate i8,
No. 5.
23 Op. cit., p. 383.
24 Op. cit., p. 358.

IrishArt (London I940), p. I0o.


Die Buchrollein derKunst(Leipzig,I907).
27 We should regard the squatting female figure (Plate Cd,
centre) as a Sheela-na-Gig,and hope that the originaland probably
paganmeaningof this whole group of femalestatueswill some day
be fully explained.I am greatlyindebtedfor advice in the preparation of thisarticleto Fatherkomuald Bauerreiss,O.S.B., of Munich,
and to ProfessorI. LI. Foster, Sir John L. Myres and Mr. C. H.
Roberts of Oxford.
25

26

Notes
H. L. Sexton, Irish Figure Sculptures(Portland, Maine,

1946), p. 299.
2 Ibid.,p. 298; the Revd. D. O'DriscollinJ.R.S.A.I., Vol. LXXII,
pp. ii6ff.; R. A. S. Macalister,The Archa?ology
of Ireland,2nd ed.
(London, 1949), p. 358.

THE DRAVIDIAN
KINSHIP TERMINOLOGY
EXPRESSION
OF MARRIAGE*

AS AN

by
L. DUMONT
Instituteof SocialAnthropology,
Universityof Oxford

54

This paper'springsfrom two sources.Field acquaintance with Dravidian kinship terminology


made me feel very strongly its systematic,logical character;
I could not help thinking that it centred in marriage, and
that it should be possible to expressthose two featuresin a
simple formula. But, in trying to do so, a considerableresistance from currentanthropologicalideas was experienced.
Therefore a few generaland criticalremarkssuggest themselves.
* A paperreadbeforetheIVInternational
and
ofAnthropological
Congress
EthnologicalSciences,Vienna,September,I952, and herepublishedby
ofthe OrganizingCommittee.With 4 text figures
permission

PRELIMINARY

Its main features are well known: classificationaccording to generations, distinction of sex, distinction of two
kinds of relatives inside certain generations, distinction of
age.
Since Morgan, who based his second or 'punaluan'
family on the Dravidian and the Seneca-Iroquoissystems,
this type of terminology, known as Seneca or DakotaIroquoistype, and one of the most widely spread,has challenged anthropologists. Rivers, studying the Dravidian
system, saw that its main feature was the distinction of

34

This content downloaded by the authorized user from 192.168.82.203 on Fri, 30 Nov 2012 00:39:46 AM
All use subject to JSTOR Terms and Conditions

MARCH,

Man

1953

No. 54

parallelandcrosscousins,andrightlyconnectedsomeof its
-I
I
featureswith cross-cousinmarriage,but, to accountfor it
F
M's Br
asa whole,he turnedtowardsa hypotheticalpreviousstage
of dualorganization.Lesssatisfactorydescriptions,when and it is very likely that the principle of the opposition lies
foundin modernliterature,
witnessto thedifficultyscholars in that relationship.Possibly our preconceived ideas resist
encounterin becomingfamiliarwith this importantand such a view, but should they not give way if the facts im-

relatively simple terminology. As late as 1947 we find


maintainedthe denomination of 'bifurcate merging' type
introduced previously with the explanation: ' . . . bifur-

cate, becausepaternal and maternal kin are distinguished,


mergingas far as thereis a partialmerging with the parents,
a definitionobviously inaccurateand misleading,as the distinction is not between paternaland maternalsides, which
are, on the contrary, treatedexactly according to the same
principle,as alreadymade clear by Rivers. Even when the
'principleof the solidarity of the sibling group' is emphasized, we return to the same confusion, since the paternal
auntis assimilatedwith the father, the maternaluncle with
the mother.2
All this would requirean explanation,and some of what
I believe to be the factors producing these misconceptions
will be found below. But perhapsit may be said in general
that the terminology was not consideredfor a moment in
itself but in terms of other aspectsof kinship, in fact related
to but differentfrom it; at the same time it was still felt as
irrationaland one hastened to explain without accurately
describing.This is so true that when Kirchhoff,on the contrary, only wants to describe it, he comes close to the
explanation.He states,in his type D, that there is 'a common word for fatherand father'sbrother,but anotherword
for mother's brother' (etc., in two columns).3 Let us proceed from this point to some furtherobservations.Here, in
the father'sgeneration, there are two kinds, and two kinds
only, of male relatives.They are two classes,and we should
not, because the father and the mother's brother respectively fall into these two classes,by stressingthem in fact
substitutethe idea of a dyadic relationshipfor that of a class,
as we do if we suppose, for example, 'mother's brother' to
be the basic meaning, and the others to be extensions.4
Moreover, the 'mother's brother' is also the 'father-inlaw,' and the common assumptionthat the affinalmeaning
is here secondary, the cognatic meaning being primary, is
based upon nothing but the common notion that one's
kinshipposition necessarilyprecedesone's marriage,an idea
quite out ofplace here, as only the analysisofthe system can
reveal the real meaning of the category. All these arbitrary
assumptionsarise from our own way of thinking, unconsciouslysuperimposed upon the native way of thinking.
We must,therefore,refuseto indulge them and keep before
us the question: what is the principle of the opposition between those two classesof relatives exemplified by what
we call fatherand mother'sbrother?Provided that we consider this opposition as standingin its own right and do not
assumethat the principle of the opposition lies in the relation with the Ego, and provided that we view it against
the background of the whole system (see note 7 below),
we can find some approach to the answer. Briefly, in
this case the relationship between father and mother's
brother is:

pose it? This relationshipwe shall call an alliancerelationship, as the relationshiparisingbetween two male (or two
female) persons and their siblings of the same sex, when
a 'sister' (a 'brother') of one is marriedto the other:

or, more generally:


and

O[ =]o.
It expresses the fact that if marriage creates a relation
between two personsof differentsexes, it connectsalso their
groups. As an equivalent formula I shall speak also of two
men (or women) having an alliancerelationshipas male (or
female) affines.
There is another way of expressingthe same fact, which,
although not altogetherwrong, is I think less accurate,and
the criticism of which will throw some more light on the
anthropologist'sunconsciousresistanceto the classificatory
idea. It is possible to extend the distinctionbetween-parallel
and cross cousinsand to speakof paralleland crossrelatives,
the principle of the distinctionbeing that 'there is, or there
is not, a change of sex when passingfrom the direct line to
the collateral line.' I followed this doctrine in a monographicstudy of kinship in a Tamil-speakingcommunity.5
But the whole passage, although tending to a synthetic
view, is, I am afraid,obscure.Moreover, the formula is not
satisfactoryfor two reasons: (i) in spite of the fact that the
natives do, when tracingrelationships,passfrom one line to
another, these are not among their basic categoriesand are
not in the least expressedin their theory: (ii) the system has
much to do with marriage, and this should appear more
clearly,if possible,in its formula. In fact, it is the anthropologist alone who is responsiblefor the introduction of this
unsatisfactoryconcept of a 'change of sex'; he does so
becausehe wants to trace through a relative of the opposite sex a relationshipwhich the native conceives-when he
thinks classificatorily-in a different manner. For instance
we introduce the mother as a link between Ego and his
mother'sbrother, where in fact the latteris just opposed to
the father. Two errors converge here: (i) the 'extension'
tendency confusesa classwith the actualmother's brother,
(ii) the introduction of the latter's compounded, western,
descriptivename bringsin the mother, who is only relevant
at this level as the link by which the relation between
father and mother's brother comes into existence. If, however, we agree to consider the terms for the two sexes
separately (as is normal in a system where the terms for
females are distinct, and not mere feminine forms of the
terms for males), and in a classificatoryperspective, the
difficultyvanishes.

35

This content downloaded by the authorized user from 192.168.82.203 on Fri, 30 Nov 2012 00:39:46 AM
All use subject to JSTOR Terms and Conditions

Man

No. 54

After this lengthy but necessarydiscussion,we can now


define the problem.
LIMITS AND NATURE OF THE ANALYSIS

Since Morgan, it has been recognized that the terminological systems used by most of the communities speaking
one of the four written Dravidian languages (round about
70 million people) are very much alike. What does this
amount to, when each language uses differentterms, when
againin each language the actuallist of terms differsslightly
from one group to another, and when, moreover, only a
few such lists are recordedfrom among the vast number of
those which exist? Is it possible to abstractanything like a
common terminological system? It is, thanks to the systematic characterof a remarkablyconstant structure.And it
will not be denied that the attempt will be logical rather
than statistical. Not all groups conform to the perfect
schema outlined below-for instance, some Tamil Brahmins alter the system considerably by the introduction
of a number of individualizingterms, or Nayar at the present day do not distinguishbetween cousins (accordingto
Mlle Biardeau)-but on the whole most lists can be said to
centre in a common scheme, from which they differ
slightly and individually. Both the Tamil lists and the published Kanareseexamples illustrateit almost perfectly.6
The limits of the analysiswill be drawn close to the vital
nucleus of the system: I shall consider only the common
classificatoryfeatureswithin a range of live generations.
One important point is that the nature of the task compels us to considerthe distinctivenessof the terms denoting
the classes, quite irrespective of their concrete linguistic
form. This is fortunate, because it allows the analysis to
develop at the basic level of the structure of the system,
whereas such analysesusually become mixed up with linguistic considerations as well as with considerations of
attitudesor institutionswhich belong to a differentlevel of
analysisand which are excluded here by the very diversity
of the background. The need to stress the cross-cousin
marriagewill appear the more striking as our analysisdevelops.
A brief explanation is needed of the expression used
above: 'the distinctiveness of the terms denoting the
classes.'The distinctivenessof the terms is the main matter,
as they are used to distinguish (i.e. to oppose) classes.But
conversely, linguistic differences which are not used to
oppose classesare irrelevanthere, and it is for this reason
that I add the words 'denoting the classes.'For instance,
different words applied to exactly the same relatives are
irrelevant, or again secondary differences within a class
(obtainedby affixation, etc.) are irrelevantin so far as they
do not alterits unity (becausefor instancethe classword or
root is kept in all). Again, linguistic resemblances may
exist between terms of different classes, in so far as the
classes are not in direct opposition. All such facts are of
interest, and may even be found to be common to all our
terminologies; but they do not form part of the basicstructure. (Considerationsof space preclude these points being
developed and exemplified here as they should be.) Our
situation is similar to that of the phoneticist: just as he
36

MARCH, I953

retains among phonetic particularitiesonly those which


differentiatemeanings, we here retain from linguistic particularitiesonly those which differentiaterelatives,and even
(for the time being) the fundamental classes of relatives
only.
The system as just defined classifiesall relatives of five
generationsfrom grandfatherto grandsoninto i6 dassesby
using i6 distinctive (setsof) terms. The generationsare as a
rule absolutely distinguished; there is no assimilation of
relatives belonging to different generations. Additionally,
Ego's generation is split into two by distinguishing relatives older and younger than Ego: this distinction of age
will be treatedas analogousto the distinctionof generations.
(The distinctionof age in other generations,e.g. the father's,
is marked,not by distinctterms, but by prefixed adjectives;
hence it is not relevant here, as stated in our previous
point). Some of the terms have a masculineand a feminine
form, some have only one form, either masculine or
feminine, and this is the rule wherever the central, critical
distinction which follows is fully maintained. In each
generation (or age) group, the relativesof the same sex are
distinguishedinto two classes.In the chart (fig. i), every
class is designated by a letter, from A to P, and they are
distributedsymmetrically to stressthe opposition.
Generation

son
grandson

E >Ego
1<Ego

A (+fem.A')

grandfather
father

K
N( +fem-N')

IO[-=k+N]*( +fem.O')

P(+fem. P')

* For instancein Tamil,wherek probablymeans'marriage.'The


linguisticconnexionbetweenN and0 is stressedhereas an
exception.
FIG. I
B the 'fathers,'C the 'mothers,'
A are the 'grandfathers,'
E the 'mother's
D the 'father'ssisters'and 'mothers-in-law,'
brothers'and 'fathers-in-law,'F the 'brothers'older than
Ego,J the younger,I andM, 'male crosscousins,'olderand
younger;G, K, 'sisters,'and H, L, 'femalecrosscousins'
respectivelyolder and younger;N the 'sons' (fem. for the
'daughters');0 the 'sons-in-laws'(fem.for the 'daughtersin-law').

Although, for the reader's convenience, I give the


ordinary equivalents, we shall not rely upon them in the
least, but on the contrary try to deduce the meaning of
each class from its situation in the whole.
Some qualificationsare necessary,as regardsthe value of
the chart. ClassD has a tendency to split among the Tamil
groups that I studied,but the cleavageis never the same, and

This content downloaded by the authorized user from 192.168.82.203 on Fri, 30 Nov 2012 00:39:46 AM
All use subject to JSTOR Terms and Conditions

Man

MARCH, I953

thetwo termson whichit is basedarelargelyinterchangeable,so thatalreadyin Tamilit is not possibleto takeit asa
generalfeature.In the region HILM I had to choose
betweentwo variants,the othervariantnot applyingdistinctionof age to this group.Both will be found equally
consistent.For N and 0 this is the Tamilsituation,while
elsewherethe centraldistinctionandthe distinctionof sex
aremorein evidence.
We now proceedto discover,or ratherconfirm,the
natureof the principleof the centralopposition,and thus
definethe fundamentalmeaningof eachclass(as distinct
fromitslinguisticmeaning;seeabove),andto tryto understandthe way in whichthe differentdistinctionsarecombined,andthe rangeof theirapplication.
FATHER

No. 54

two abstractoppositionsoperatingcrosswise: (i) communityand differencein generation;(ii) communityand


differencein kin, i.e. kin andalliance.The latter,in which
the categoryof allianceis broughtto light by opposition
to thekincategory,is of paramount
importance.Compared
with Morgan'sMalayansystem,wherethe two categories
are not distinguished,it emphasizesthe importanceof
alliance,i.e. of marriageas a relationbetween groups.
Moreoverboth ideasare given together,andspringfrom
one another:no kin withoutalliance,no alliancewithout
kin.

S GENERATION

We have seen alreadythat the alliancerelationshipdefinesthe mother'sbrotherby referenceto the father.But


thefatherhimselfis definedby referenceto theEgo. Letus
considernow the natureof the latterrelationand both
together.In doing so we shouldnot forgetthat,although
we have taken the particular,genealogicalfather as
example,we aredealingin factwith the 'fathers'asa class.
In the relation,or as I preferto say, in the opposition,betweenEgo andEgo'sfather,therearetwo elements,one of
whichis commonto them both, while the otherdifferentiatesthem; the elementwhichis commonto both terms
of the oppositionI call the 'basis'of the opposition,the
differentiating
elementI callthe 'principle'of the opposition.Theprincipleis clear:it is the distinctionbetweentwo
But whatis the basis,whatis it that
successivegenerations.
is commonto Ego andEgo'sfather?Obviously,theanswer
liesin the context:whatthey havein commonis opposed
to what makestheir relation(morepreciselythe father's
relation)with the mother'sbrother,i.e. to the alliance

. ..... ...

Father
*A

,,

I's brother

distinction
of Kin(alliance)
distinction
of generation

*"f

1*

SERVICE DESSIN DU MUSEE DE L'HOMME

FIG. 2

A few more remarksmay be added. (i) We understand


why there are no special terms (at the present level) for
affmes; the basic meaning of the terms for the 'cross' category is affinal-my mother's brother is essentially my
father's affme. (ii) We have in fact taken the two oppositions as a way leadingfrom Ego to the father and from the
father to the mother's brother; are we then perhaps not
entitled to speak of a structuresensustricto?But here lies
(flg.
2).
Fatherand Ego are relatedby a link which excludes the characteristicof a kinship terminology as compared
alliance,andwhichI proposeto call 'kin link.' One quali- with other kinship groupings, that it is a constellation
ficationregardingsex mustbe added:whereasthe 'fathers' revolving around the Ego. The only differencefrom cusand the 'mother'sbrothers'are respectivelymale sibling tomary views on the subjectlies in the way we have taken,
groups,the sex of Ego is irrelevant(the termsfor father, not the way through the mother, as suggested by our own
of thesex of Ego). Thetwo vocabulary, but, I believe, the native way, as imposed by
etc.,beingthesameirrespective
generations
opposedto oneanotherin thekin groupareone the terminology. (iii) What is here called kin has, of
generationof male siblings,and the generationof their course, nothing to do with actual groups, being only an
children,bothmaleandfemale.In otherwords,the distinc- abstractionarisingfrom the oppositions; this again centres
tion of sex, if it is the preliminaryconditionof the distinc- in Ego, and is only a part of what the terminology suggests
tion of kin, is unrelatedto the distinctionof generation; as such, because we had to abstract it on the male side;
turning to female relatives, we shall find its feminine
thisshouldbe remembered.
If we now considertogetherthetwo oppositionsbetween counterpart. The whole could be called 'terminological
Ego, his fatherandhis mother'sbrother,we see thatEgo kin' to avoid confusion, and opposed to 'terminological
andthe fatheraresimilarin kin anddifferentin generation, affines'. This is only a framework which is used and
while fatherandmother'sbrotheraresimilarin generation shaped by each group according to its particularinstituand differentin kin (i.e. are allied).Each of the two ele- tions.
In the same generation,we can deal exactly as above with
ments (generationandkin) servesunderits negative(difformasprincipleof one opposition,andunder the opposition between the 'mother' and the 'father's
ferentiating)
sister,'and connect it with the opposition between Ego and
its positive(uniting)formasthe basisof the other.
The two concreteoppositionsnot only haveone termin Ego's mother. We shall leave out the intermediary link,
is builtupon this time the father, as a mere agent bringing about (and
common(thefather),but theircolncatenation
37

This content downloaded by the authorized user from 192.168.82.203 on Fri, 30 Nov 2012 00:39:46 AM
All use subject to JSTOR Terms and Conditions

Man

No. 54

hencecontainedin) the alliancerelationshipbetweenthe


two women.Thekin grouparisingherewi be formedof
a generationof femalesiblings,the mothers(opposedto
tlhcirfemaleaffines),andof the generationof theirchildren
of both sexes.This kin categoryis not differentfrom the
precedingone; it is the same,opposedto allianceas above,
thloughwe takeanotherview of it in accordancewith the
distinctionof sexesin the system.In orderto insistupon
theclassificatory
character,we give here (fig. 3) a generalized schema;a similarone could,of course,be drawnfor
males.

How canwe in our turnreproducein othergenerations


what we said in the father'sgeneration?If the alliance
may be supposedto be similar,the generation
relationship
will be different.
relationship
marriage(or
generation,cross-cousin
Inthe grandfather's
an equivaleht)leadsone to supposean affmallink between
and this is the very reasonwhy
Ego's two grandfathers,
they cannotbe distinguished,and why thereis normally
only one term for both of them, for both are kin in one
way, andaffmesin another:motheras well asfatheris kin
to the Ego, 4ndso aretheirfathers,who haveat the same
so thatwe may considerone
time an alliancerelationship,
of themA askin, andthe otherB asaffme,or, equally,B as
kin and A as affine: the two categoriesmerge in that
generationandthe distinctionof kin doesnot applyto it.
The same may be said about grandsons:allianceworks
asa principleof oppositionfor (two or) threegenerations
only, whereasall relativesmerge in the fifth and the

OQ Li

of genHeratio
distinction

) oOr(

(=
ALLIANCE
REPRESENTS
FIG.

=)

first.7
There is no theoretical difficulty in Ego's son's generation, but rather a practical one: in Tamil at least, the
alliance opposition weakens (the basis is emphasized by
the use of the same word, with the addition of a prefix on
one side, ratheras with 'son' and 'son-in-law'), and at the
same time the sex opposition disappears('daughter' is the
feminine of 'son'). This is consistent, but I can offer no
structuralexplanation, although there is probably a common background.
In Ego's generation (males), something interestinghappens if we try to apply the same procedureas in the father's
generation: on one side the alliance opposition is present,
the male affinesbeing sister'shusbandand wife's brotheras
well as sons of the father's(male) affinesand of the mother's
(female) affmes. On the other side, the generation opposition vanishes, as Ego and his brothersmight be considered
indifferently, but a new principle is invoked in order to
replace as it were the waning principle, i.e. relative age is
distinguished, and the generation is split into two halves
under Ego's older brother and Ego's younger brother. The
two distinctions(generationand age), one of which relieves
the other, have a common backgroundof age connotation
and are closely connected.8

Having ended the part of the demonstration which is


most likely to arouse controversy,and before extending it
to the other generations,we may pausehere and get a first
glimpse of the whole. There will be no difficulty, as one
can imagine, in showing that Ego's 'cross cousins' are
essentially Ego's affines, just as the 'mother's brother'
proved to be essentiallythe father'saffme. This means that
the alliance which we considered horizontally in one
generation acquiresa new, a vertical dimension, and runs
through generations.
ALLIANCE AS AN

sameasto saythata certainmarriageregulationis observed.


Theoretically,to maintainthe relation,one marriagein
eachgenerationis enough,but the moremarriagesof that
will be. The
type occur,the firmerthe alliancerelationship
mostimmediateandcomplete,the totalformulafor thatis
In fact,whatwe
marriage'of anydescription.
'cross-cousin
is nothingbut
marriage
to
call
cross-cousin
areaccustomed
the alliancerelationthe perfectformulafor perpetuating
ship from one generationto the next and so makingthe
alliancean enduringinstitution-a very particularand
queernamefor a factof a very generalandlogicalcharaccustomaryand
ter. Indeed,it is only the anthropologist's
peculiarvocabulary,expressingalliancein terms of kin,
which concealsthissimpletruthinsteadof revealingit.
OTHER GENERATIONS

KID

[j

MARCH, I953

ENDURING

INSTITUTION:

CROSS-COUSIN

MAIRIAGE

It is not another alliance,but one and the same relationship transmittedfrom one generationto the next, inherited;
what we have consideredup to now as an alliancerelationship was ouly a horizontal section of it. And could it be
opposed to kin if it did not transcendgenerations?It is this
alliance as an enduring institution that is embedded in the
terminology, that provides it with its fundamental and
characteristicopposition.
But to say that an alliancerelationshipis inherited is the
38

This content downloaded by the authorized user from 192.168.82.203 on Fri, 30 Nov 2012 00:39:46 AM
All use subject to JSTOR Terms and Conditions

Man

MARCHr,1953

Now we can proceed with the elder brother as with the


father:he is opposed to Ego, as older, and he is opposed to
the 'cross cousin, older than Ego' as a sibling to an affine.
The same for the younger brother, opposed to a 'younger
crosscousin' (fig. 4a), but we here crossthe generationaxis
of the structure,and the age order between Ego and his kin
is inverted.
As previously stated, our chart gives for the affineshere
only one of the two variants actually found. The other
variantpresentsno distinctionof age among the affmesand
has only one term for males equivalent to HL. For this we
can account very simply: in that variant, Ego's generation
is taken as a whole, the male affine is opposed directly to
Ego, and the age distinction, although introduced among
brothers,does not replace structurallythe generation distinction, and therefore, is not extended to the affines
(fig. 4b).

No. 54

The third distinction (which alone is in no way biological) is the most important; the system embodies a
sociological theory of marriage taken in the form of an
institution following the generations, and supposes-as
well as favours-the rule of marrying a cross cousin as a
means of maintainingit. Hence also the fact, well preserved
in Indian groups, that the two categoriesof kin and affines
comprehend all relatives without any third category. This
may be understoodwithout resortingto dual organization;
the opposition between kin and aflmes constitutesa whole
-the affme of my affme is my brother-marriage is in a
sense the whole of society, which it unites, and at the
same time separatesin two from the point of view of one
Ego.9
No wonder, then, if India makes it the paramountceremony, and perhapsit is also an explanationfor the stability
and vitality of the Dravidian terminology which has
puzzled many anthropologistssince Morgan.
Notes
wish to expressmy thanksto ProfessorEvans-Pritchardfor his
discussionof this paper,andto D. Pocock, for his help in its preparation. The structuralapproach, although different, is largely influencedby Cl. Levi-Strauss,'L'Analysestructuraleen linguistiqueet
en anthropologie,' Word,New York, Vol. I, No. 2, August, I945.
For a structuralapproachto attitudes,see E. E. Evans-Pritchard,
'The Study of Kinship,'MAN, I929, I48, and I932, 7.
2 Lewis H. Morgan, AncientSociety,London, I877, pp. 424-52;
W. H. R. Rivers, Kinshipand SocialOrganization,London, I9I4,
pp. 47-9, 73; see also 'The Marriageof Cross Cousins in India,'
J.R.As.Soc., I907, pp. 61I-40.
'Bifurcate-merging': R. H. Lowie, 'A Note on Relationship
Terminologies,'Amer.Anthrop.,I928, pp. 265f.; cf. G. P. Murdock,
'Bifurcate-Merging,a Test of five Theories,'Amer.Anthrop.,I947,
pp. 56-68.
'Solidarity of the Sibling Group': A. R. Radcliffe-Brown, in
AfricanSystemsof KinshipandMarriage,Introduction,p. 25.
3 P. Kirchhoff, 'Verwandtschaftsbezeichnungen
u. Verwandtenheirat,'Zeits.fur Ethnol.,Vol. LXIV (I932), pp. 4I-72; cf Lowie,
Social Organization(I948), London, I950, p. 63.
4 For a strong protest againstthis kind of 'extension' see A. M.
Hocart, 'Kinship Systems,' Anthropos, Vol. XXXII (I937),
PP. 545-5I (reprintedin TheLife-givingMyth,London (Methuen),
II

AHt

EG
O

.I

...........................
EGO

.I

(b)

(a)
FIG. 4

Moreover, it is in this part that the actual terminologies


differmost from our chart. Severalfactorsare at work, one
of which is. of a classificatorynature. It is a tendency to
stressthe relative sex of the person compared to the Ego,
as is quite naturalwhere prospective mates are found. This
tendency combinesin various ways with the elder-younger
distinction, and the matter is still more complicated by
other factors, so that it requiresa special treatment.
In the preceding paragraphwe have already anticipated
the classification of female relatives, which should be
extended from the mother's to the other generations.This
is not necessary,as the structureis symmetrical (with the
exceptionjust mentioned).

CONCLUSION

I have shown, I hope, that the Dravidiankinshipterminology, and with it other terminologies of the same type, can
be considered in its broad features as springing from the
combination in precise configurationsof four principlesof
opposition: distinction of generation (qualified as an
orderedscale), distinctionof sex, distinctionof kin identical
with alliance relationship,and distinction of age.

I952,

p. I73).

5 Formulafrom Cl. Levi-Strauss,Les Structures


Elementairesde la
Paris (P.U.F.), I949, p. i65. I hope that my emphasison
Parente',
marriagewill be found in keeping with the generalinspirationof
that work. L. Dumont, 'Kinshipand Allianceamong the Pramalai
Kallar,'EasternAnthrop.Lucknow, Vol. IV, No. i., Sept.-Nov.
(I950-I),
pp. I-26 (but with many misprints); see pp. 5-I2 as a
first attemptin the presentdirection.
6 Most completeareMorgan'slists (Systems,.pp.
5i8f.) for Tamil,
Telugu andKanarese.Referencesto recentmonographsin Dumont,
loc. cit. (not restrictedto written languages,but see for Kanarese
Srinivas,Marriage.. ., for MalayalamAiyappan,Nayadis... and
Iravas... .). Listsof 'common' terms, unspecifiedand unlocalized,
arefound in grammars,etc. I have takenhereinto accountlistsfrom
severalgroups in Tamil, to be published.
7 This featureis fundamental,and our analysisrestslargely upon
it. The whole structureis differentwhen grandsonand grandfather
are identified, as in Kariera(with two terms for each).
8 The close connexion between age and generationin the structure may constitute the basis of an important exception to the
generationprinciple, ratherof a diachronicnature, as stressedby
Mrs. I. Karvein a study to be published(oralcommunication).
9 This does not happenalways,but only when certainconditions
are present.

39

This content downloaded by the authorized user from 192.168.82.203 on Fri, 30 Nov 2012 00:39:46 AM
All use subject to JSTOR Terms and Conditions

Vous aimerez peut-être aussi