You are on page 1of 9

Classificatory Systems of Relationship.

Author(s): A. L. Kroeber
Source: The Journal of the Royal Anthropological Institute of Great Britain and Ireland, Vol. 39
(Jan. - Jun., 1909), pp. 77-84
Published by: Royal Anthropological Institute of Great Britain and Ireland
Stable URL: http://www.jstor.org/stable/2843284
Accessed: 15-02-2016 14:08 UTC

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.

Wiley and Royal Anthropological Institute of Great Britain and Ireland are collaborating with JSTOR to digitize, preserve
and extend access to The Journal of the Royal Anthropological Institute of Great Britain and Ireland.

http://www.jstor.org

This content downloaded from 200.130.19.157 on Mon, 15 Feb 2016 14:08:52 UTC
All use subject to JSTOR Terms and Conditions
77

CLASSIFICATORY SYSTEMS OF RELATIONSHIP.

BY A. L, KROEBER.

THE distinction, betweenclassificatory and descriptivesystemsof relationshiphas


been widelyaccepted,and has foundits way into hanidbooksand generalliterature.
According to the prevalent belief the systems of certain nations or languages
grouptogetherdistinctrelationshipsand call themby one name,and are therefore
classifying. Other systems of consanguinityare said to indicate secondary
differences of relationshipby descriptiveepithets added to their primaryterms,
and to be thereforedescriptive.
Nothingcan be morefallaciousthan thiscommonview. A moment'sreflection
is sufficientto show-that everylanguagegroupstogetherunder single designations
manydistinctdegreesand kinds of relationship. Our word brotherincludesboth
the oldefand the youngerbrotherand the brotherof a man and of a woman. It
thereforeembraces or classfies four relationships. The English word cousin
denotes both men and--womencousins; cousins on the father'sor on the mother's
side; cousins desc6endedfromthe parent's brotheror the parent'ssister cousins
respectivelyolder or youngerthan one's self, or whose parents are respectively
older. or youngerthan the speaker's parents; and cousins of men or womern.
Thlirty-two different relationshipsare thereforedenoted by this one English word.
If the termis not strictlylimitedto the significance of firstcousin,the number of
distinct ideas that it is capable of expressingis many times thirty-two.Since
thenit is not onlyprimitivepeople that classifyor fail to distinguishrelationships,
the suispicionis justifiedthat the currentdistinctionbetween the two classes or
systemsof indicatingrelationshipis subjective,and has its origin in the point of
view ofinvestigators, who,on approachingTextoforeignlanguages,have been impressed
with theirfailureto discriminatecertainrelationships
Texto betweenwhichthe languages
of civilizedEurope distinguish,and who,in the enthusiasmof formulatinggeneral
theoriesfrom such facts,have forgottenthat theirown languagesare filledwith
entirelyanalogousgroupingsor classifications which custom has made so familiar
and naturalthat theyare not feltas such.
The total numberof different relationshipswhich can be distinguishedis very
large,and reaches at,least manyhundred. No languagepossessesdifferent terms
forall of these or even forany considerableproportionof them. In one sense it is
obviousthat a languagemust be moreclassificatory as the numberof its termsof
relationshipis smaller. The number of theoreticallypossible relationships

This content downloaded from 200.130.19.157 on Mon, 15 Feb 2016 14:08:52 UTC
All use subject to JSTOR Terms and Conditions
78 A. L. KROEBER.-Classiflcatory SystemsofRelationship.

remainingconstant,there must be more ideas grouped under one termin pro-


portionas the ilumberof termsis less: Followiiig the accepted understandingof
what constitutesclassificatoryconsanguinity,English, with its twentyterms ol
relationship,mustbe not less but more classificatorythan the languages of all
primitivepeople who happen to possess twenty-five, thirty,or moreterins.
It is clear that if the phrase classificatoryconsanguinityis to have any
meaningit must be soughtin some morediscriminating way. The singlefactthat
another people group together various relationships which our language
distinguishesdoes not mnaketheirsystemiclassificatory.If thereis a generaland
fundamentaldifferencebetween the systems of relationshipof civilized and
uncivilizedpeople,its basis must be looked forin somethingmoreexact than the
roughand ready expressionsof subjectivepoint of view that have been customary.
It is apparent that what we should tryto deal withis not the hundredsor
thousands of slightly varyingrelationshipsthat are expressedor can be expressed
by the various languages of man,but the principlesor categoriesof relationship
whichunderliethese. Eight such categoriesare discernible.
1. The differencebetweenpersonsof the same and of separategenerattions.-The
distinctionsbetween father and grandfather,between uncle and cousin, and
betweena personand his father,involvethe recognitionof this category.
2. The difference betweenlineal and collateralrelationship.-Whenthe father
and the father'sbrotherare distinguished, this categoryis operative. When only
one termis employedforbrotherand cousin,it is inoperative.
3. Ditf'erenceof age withinonegeneration.-Thefrequentdistinctionbetween
the olderand the youngerbrotheris an instance. In Englishthis categoryis not
operative.
4. The sex of the relative.-This distinctionis carriedout so consistentlyby
English,the one exception being the foreignwordcousin,that the discrimination
is likely to appear self-evident. By many people, however,many relationships
are not distinguishedfor sex. Grandfatherand grandmother, brother-in-law
and
sister-in-law, and
father-in-lawand mother-in-law, even such close relationshipsas
son and daughter,are expressedrespectivelyby singlewords.
5. The sex of thbespeaker.-Unrepresentedin English and most European
languages,this category is well known to be of importance in many other
languages. The father,mother,brother,sister,and more distant relatives may
receiveone designationfroma man and anotherfromhis sister.
6. The sex of theperson throughwhom relationshipexists.-Englishdoes not
expressthis category. In consequence we frequentlyfindit necessaryto explain
whetheran -uncleis a father'sor a mother'sbrother,and whethera grandmother is
paternalor maternal.
7. The distinctionof bloodrelativesfromconnections bymarriage.-While this
distinctionis commonlyexpressedby most languages,there are occasionallapses;
-justas in familiarEnglish speech the father-in-law is oftenspokenof as father.
Not strictlywithin the domain of relationship,but analogous to the occasional

This content downloaded from 200.130.19.157 on Mon, 15 Feb 2016 14:08:52 UTC
All use subject to JSTOR Terms and Conditions
SystemsofRelationtship.
A. L. KROEBER.-Classificatory 79

failureto express this category,is the frequentignoringon the part of primitive


peopleofthedifference betweenactual relativesand fictitious
elan or tribalrelatives.
8. The conditionof life of thepersonthroughwhomrelationship exists.-The
relationshipmay be either of blood or by marriage; the person servingas the
bond of relationshipmay be alive or dead, marriedor no longermarried. Many
North American Indians refrainfrom using such terms as father-in-lawand
mother-in-lawafter the wife's death or separatioll. Some go so faras to possess
termsrestricted to such severed relationship. It is natural that the uncle's
relation to his orphaned nephew should tend to be somewhatdifferent fromhis
relation to the same boy while his natural protector,his father,was livinig.
Distinct terms are thereforesometimesfoundforrelativesof the uncle and aunt
groupafterthe death of a parent.
The subjoined table indicates the representation of the eightcategories,aild
the degreeto whichtheyfind expression,respectivelyin English and in several of
the Indian languagesof North America.

N.A. Indian. CaliforniaIndian.

0~~~~~~~ iI

~ ~ ~ ~~~~~ ce
mo Ca 'A
9 Ca~

No. of terms 2'... 2 20 31 19 18 28 24 27 28 24 28 34 35

Generation ... ... 21 20 31 11 13 23 24 21 27 24 22 30 26

Blood or marriage ... ... 21 19 131 17 18 26 24 27 28 24 28 32 34

Lineal or collateral 21 10 20 5 11 25 24 21 28 18 26 34 28

Sex of relative ... 20 18 29 17 2 12 16 21 20 20 17 18 22


relative
Sex of connecting 0 6 6 2 0 20 13 13 14 10 14 19 21
Sex of speaker ... ... ... 0 3 18 4 0 15 3 3 10 2 12 10 14

Age in generation ... 0 3 7 2 2 2 3 4 4 4 4 12 8

Conditionof relative
connecting 0 0 0 0 8 1 0 0 0 0 2 0 1

It appears that English gives expression to only four categories. With the
exception,however,of the one and foreignword cousin, every term in English
involves the recognitionof each of these four categories. All the Indian
languages express kfromsix to eight categories. Almost all of them recognize
1 All termsare omitted,suchas great grandfather,
great-uncle,and second-cousin,, which
are not generallyused in ordinaryspeechand existprincipallyas a reserveavailableforspecific
discrimination on occasion.
2 Termsdenoting relativesby marriageundergoa vocalicchangeto indicatethe death of
the connecting relative.

This content downloaded from 200.130.19.157 on Mon, 15 Feb 2016 14:08:52 UTC
All use subject to JSTOR Terms and Conditions
80 SystemsofRelationship.
A. L. KROEBER.-Classiftcatory

seven. But in all the Indian languages the majority of the categories
occurringare expressed in only part of the terms of relationshipfound in the
language. There are even Indian laniguages,such as Pawnee and Mohave, in
whichnot a single one of the seven or eight categoriesfindsexpressionin every
term. While in English the degree of recog,nition which is accorded the repre-
sentedcategoriesis indicableby a percentageof 100 in all cases but one,when it
is 95, in Pawnee correspondingpercentagesranigevariouslyfromabout 10 to 90,
and in Mohave from5 to 95. All the otherIndiail languages,as comparedwith
English,closelyapproachthe conditionof Pawnee and Mohave.
It is clear that this differenceis real and fundamental. English is siinple,
consistent,and, so faras it goes,complete. The Indian systemsof relationshipall
start froma more elaborate basis, but carry out their scheme less comipletely.
This is inevitable fromthe fact that the total number of terms of relationship
emnployed by themis approximatelythe same as in English. The additionof only
one categrory to those found in English normallydoubles the numberof terms
requiredto give full expressionto the system; and the presence of three addi-
tional categoriesmultiplies the possible total by about eight. As the numrber of
termsoccurringin any of the Indian languages under considerationis not much
morethan halfgreaterthan in English,and sometimesis not greaterat all, it is
clear that at least some of theircategoriesmustfindonly verypartial expression.
In short,as faras the expressionof possible categoriesis concerned,Eniglishis
less complete than any of the Indian languages; but as regardsthe giving of
expressionto the categorieswhich it recognizes,English is more coinplete. In
potentiality, the English scheineis poorerand simpler; but fromits own point of
view it is bothmorecompleteand moreconsistent. As English may evidentlybe
taken as representativeof European languages,it is in this point that the real
difference is to be foundbetweenthe systemsthat have been called classificatory
and thosethat have been called descriptive.
The so-called descriptivesystemsexpress a small number of categoriesof
relationshipcompletely;the wrongly-named systemsexpressa larger
classificatory
number of categorieswith less regularity. Judged frornits own point of view,
English is the less classificatory;looked at fromthe Inidianpointof view it is the
more classificatory, inasmuch as in every one of its terms it fails to recognize
certaindistinctionsoftenmade in other languages; reggarded froma general and
comiiparative pointof view,neithersystemis moreor less classificatory.
In short,the prevalentidea of the classificatory systemnbreaks down entirely
underanalysis. And in so far as there is a fundamentaldifferencebetween the
languagesof European and of less civilized peoples in the method of denoting
relationship, the difference can be determinedonly on the basis of the categories
described and can be best expressedin termsof the categories.1

A tendencytowardreciprocalexpressionis sometimesof importanceand mayinfluence


the degreeto whichcategoriesare given expression. Reciprocaltermsare such that all the
personsincludedin the relationshipexpressedby one termcall by one name all the persons

This content downloaded from 200.130.19.157 on Mon, 15 Feb 2016 14:08:52 UTC
All use subject to JSTOR Terms and Conditions
A. L. KRoEBER.-Classificatory
SystemsofRelationship. 81

The categoriesservealso to indicatethe leading characteristicsof systemsof


the same general order. It is obvious, for instance, that the most important-
differencebetweenDakota and Arapahois the strong tendencyof the forinerto
recognizethe sex of the speaker. Chinook is notable forlayingmorestress on
the sex of the speaker and of the connectingrelation than on the sex of the
relative.' General differences such as naturally occur between the languages of
one regionand of anothercan also be expressed in termns of the categories. All
the Californiiasystems,for instance,lay much more stress upon the sex of the
connectingrelativethan do any of the Plains languages examined. The. Plains
systemsare conspicuousfor their weak developinentof the distinictionbetween
lineal and collateralrelationship,this findingexpressionin two-thirdsof all cases
in Dakota, half in Arapaho,one-fourth in Pawnee. In seven Californialanguages
the corresponding values lie betweenthree-fourths and complete expression. The
methodcan be applied successfullyeven in the case of smaller and contiguous
geographicalareas. Of the seven Californialanguages Luisefnoand Mohave are
spokenin southernCalifornia. Their systemsshow a unityas compared with the
systemsof the five languages from northernanld cenltralCalifornia. Both the
southern California languages have a greater number of terms; both are
strongerin the expressionof the categoriesof the sex of the connectingrelative
and of age withinthe same generation; and both are weaker in the categoryof
sex of the relative,than the others. Aoain, Chinlookand Skokomish,both of the
NorthPacificCoast, are alike in indicatingthe colndition of the connectinarelative
and in failing, on account of the possession of grammatical sex gender, to
distin-guishthe sex of relativesthemselvesin mnany termsof relationship. There is
a verydeep-goingdifference between them,however,in the fact that Skokomish

who apply this term to them. In the most extremeformof reciprocity the two groupsof
relativesuse the same term. The paternalgrandparents call theirsons' children,whetherboys
or girls, by the sametermwhich these children,both boys and girls,apply to theirfathers'
parents. Nevertheless, thereciprocalrelationis just as clear, thoughless strikinglyexpressed,
wheneachofthegroupsusesa differentterm fortheother. OurEnglishwordsfatherand child,or
brotherand sister,are not reciprocal,for the termchild is employedalso by the mother,and
brotheris used by the brotheras well as by the sister. In factthe onlyreciprocaltermin
English is cousin. The tendencytowardreciprocalexpressionis developedin many Indian
languages. It is particularlystrongin California. In some languages this tendency has
broughtit about thatdifferent categoriesare involvedin the termsappliedto a pair ofmutual
relationships.The termfather'ssisterindicatesthe sex of the relativebut not ofthe speaker.
The exactreciprocalof father'ssisteris woman'sbrother'schild. This term,however,does not
recognizethe sex of the relativeindicated,but does implythe sex of the speaker. The two
reciprocaltermstherefore each involvea categorywhichthe otherdoes liot express. If the
same categorieswere representedin the two terms,brother'sdaughterwould correspondto
father'ssisterand exact reciprocity would be impossible. When, therefore, the termsfound
are father'ssister and woman's brother'schild, it is clear that the tendencytoward the
establishment of exactlyreciprocaltermshas been strongerthan the feelingfavoringthe
consistentuse or neglect of certaincategories;in otherwords,the extentto whichcertain
categoriesare expressedhas been determined by the vigorof the reciprocaltendency.
1 No doubt, as has been pointedout, owing to the fact that the sex of the relativeis
indicableby purelygrammaticalm=eans in thisgnd cortlinothierI nguages.
VOL. XXXIX, G

This content downloaded from 200.130.19.157 on Mon, 15 Feb 2016 14:08:52 UTC
All use subject to JSTOR Terms and Conditions
82 SystemsofRelationship.
A. L. KROEBER.-Classiftcatory

is as freeas English fromrecognizingthe sex of the speaker and of connecting


relatives, while Chinook generally expresses both categories. In short, the
categoriespresenta means of comparingsystemsof terms of relationshipalong
the basic lines of their structureand of expressing their similarities and
differences withoutreferenceto individualtermsor details.
The reasonwhy the vague and unsatisfactory systemof
idea of a classificatory
consanguinity has foundstuchwide acceptanceis not to be sought in. any primary
interest in designationsof relationshipas such, but in the fact that terms of
relationshiphave usually been regarded principally as material from which
conclusionsas to the organizationof societyand conditionsof marriagecould be
inferred. If it had been more clearly recognizedthat terms of relationshipare
determinedprimarilyby linguisticfactors,and are only occasionally,and then
indirectly,affectedby social circumstances, it would probablylong ago have been
generallyrealized that the difference between descriptiveand classificatory systems
is subjectiveand superficial. Nothingis moreprecariousthan the commoni method
of deducing'the recent existence of social or marital institutions from a
ser ento designation
Isso parece
of relationship. Even when the social conditionagrees perfectlywith
uma
problemtica expressions of relationship,it is unsafeto concludewithoutcorroborative evidence
dos estudos
sobre
that these expressions are a direct reflectionor resultof the condition.
parentesco
ou
In the Dakota language,accordingto Riggs,thereis only one wordfor grand-
terminologia fatherand father-in-law.Followingthe mode of reasoning sometimesemployed,
particulares
it mightbe deduced fromthis-that these two relationshipswere once identical.
Worked out to its implications,the absurd conclusionwould be that miarriage with
the motherwas once customaryamongthe Sioux.
In the same language the words for womnan'smale cousin and for woman's
brother-in-law have the same radical,differing onlyin a suffix. Similar reasoning
would induce in this case that miarriageof cousins was or had been the rule
among the Sioux, a social conditionutterlyopposed to the basic principles of
almostall Indian society.
The use of such identicalor simnilar termsfordistinctrelationshipsis due to a
considerablesimilaritybetween the relationships. A woman's male cousin and
her brother-in-law are alike in sex, are both of oppositesex from the speaker,are
of the same generationas herself,and are both collateral,so that they are similar
underfourcategories. In view of the comparativepaucityof terms as compared
with possible relationships, it is entirelynaturalthat the same word,or the same
stem, should at times be used to denote two relationshipshaving asw;muchin
comimon as these two.
No one would assume that the colloquial habit in modernEnglish of speaking
of the brother-in-law as brotherimplies anythingas to formof marriage,for
logicallythe use of the term could only be an indicationof sistermarriage. lt is
easily conceivablethat in the futuredevelopmnent of English the morecumbersome
of thesetwo termsmightcome into completedisuse in daily life and the shorter
take its place, withoutthe-leastchangein social or maritalconditions,

This content downloaded from 200.130.19.157 on Mon, 15 Feb 2016 14:08:52 UTC
All use subject to JSTOR Terms and Conditions
A. L. KROEBER.-Classiftcatory
Systemsof Relationship. 83

The causes which determinethe formation, choice,and similaritiess


of termsof
relationshipare primarilylinguistic. Whenever it is desired to regard terms of
relationshipas due to sociologicalcauisesand as indicativeof social conditions,the
burdenof proofmustbe entirelywith the propounderof such views.
Even the circumstancesthat the father'sbrotheris frequentlycalled fatheris
not necessarilydue to or connectedwith the custom of the Levirate; nor can
group marriagebe inferredfromthe circumstancethat there is frequentlyno
othertermformother'ssister than mother. A woman and her sister are more
alike than a woman and her brother,but the difference is conceptual,in other
wordslinguistic,as well as sociological. It is true that a woman'ssistercan take
her place in innumerablefunctionsand relationsin whicha brothercannot; and
yet a womanand her sister,being of the same sex, agree in one more categoryof
relationshipthan the same womanand lherbrother,and are thereforemoresimilar
in relationshipand morenaturallydenoted by the same term. Thereare so many
cases where the expression of relationshipcannot have been determinedby
sociological factorsand must be purely psychological,as in the instancesjust
discussed,thatit is fairto requirethat the preference be givento the psychological
cause, or that this be admittedas of at least equal probability,even in cases where
eitherexplanationis theoreticallypossibleand supportingevidenceis absent.
On the whole it is inherentlyveryunlikelyin any particularcase that the
use of identical terms for similar relationshipscan ever-be coninected with such
special customsas the Levirateor groupmarriage. It is a muchmoreconservative
view to hold that such formsof linguisticexpressionand such conditionsare both
the outcomeof the unalterablefact that certainrelationshipsare more similarto
one anotherthan others. On the one hand this facthas led to certainsociological
instiPutions;on the other hand,to psychologicalrecognitions and theirexpression
in la.aguage. To connect the institutionsand the terms causally can rarely be
anything but hazardous. It has been an unfortunatecharacteristicof the
anthropolog,y of recent years to seek in a greatmeasurespecificcauses forspecific
events,connectionbetweenwhich can 'be establishedonly throughevidencethat
is subjectively selected. On wider knowledge and freedomfrommotive it is
becomingincreasingly apparentthatcausal explanationsof detachedanthropological
phenomenacan be but rarelyfound in other detachedphenomena,and that it is
even difficultto specifythe mostgeneraltendenciesthat actuate the formstaken
by culture,as the immediatecauses of particularphenomena.
The followingconclusionsmay be drawn:-
1. The generally accepted distinctionbetween descriptive and classi-
ficatorysystemsof termsof relationishipcannotbe supported.
Crtica ao Morgan,
talvez 2. Systemsof terms of relationshipcan be properlycompared through
an examination of the categories of relationship which they
involve and of the degree to which they give expressionto these
Degree, ou seja, at que
ponto um termo pode
corresponder relao
social por ele designada categories.
3. The fundamentaldifference betweensystemsof termsof relationship
G 2

This content downloaded from 200.130.19.157 on Mon, 15 Feb 2016 14:08:52 UTC
All use subject to JSTOR Terms and Conditions
84 ofRelationship.
A. L. KROEBER.-Classiftcatory Systems

of Europeans and of AmericaniIndians is that the formerexpress


a smaller numberof categoriesof relationshipthan the latterand
expressthemmorecompletely.
4. Terms of relationshipreflectpsychology,not sociology. They are
determinedprimarilyby languageand can be utilizedforsociological
inferencesonlywithextremecaution.

This content downloaded from 200.130.19.157 on Mon, 15 Feb 2016 14:08:52 UTC
All use subject to JSTOR Terms and Conditions

You might also like