Sunteți pe pagina 1din 31

On Ethnic Unit Classification [and Comments and Reply]

Author(s): Raoul Naroll, Ronald M. Berndt, Frank D. Bessac, Eliot D. Chapple, Gertrude E.
Dole, Harold E. Driver, Paul Ducey, Melvin Ember, Helmuth Fuchs, Hans Hoffmann,
Mervyn Jaspan, David Landy, Edmund Leach, Otto Von Mering, Simon D. Messing, Frank
W. Moore, Ramkrishna Mukherjee, George P. Murdock, Artur Hehl Neiva, Gideon, Andree
F. Sjoberg, Andre F. Sjoberg, Leigh M. Triandis, C. F. Voegelin, Linvill Watson, John W.
M. Whiting...
Source: Current Anthropology, Vol. 5, No. 4 (Oct., 1964), pp. 283-312
Published by: The University of Chicago Press on behalf of Wenner-Gren Foundation for
Anthropological Research
Stable URL: http://www.jstor.org/stable/2740256
Accessed: 04-08-2017 13:39 UTC

REFERENCES
Linked references are available on JSTOR for this article:
http://www.jstor.org/stable/2740256?seq=1&cid=pdf-reference#references_tab_contents
You may need to log in to JSTOR to access the linked references.

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.

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at
http://about.jstor.org/terms

Wenner-Gren Foundation for Anthropological Research, The University of Chicago


Press are collaborating with JSTOR to digitize, preserve and extend access to Current
Anthropology

This content downloaded from 62.204.192.85 on Fri, 04 Aug 2017 13:39:37 UTC
All use subject to http://about.jstor.org/terms
On Ethnic Unit Classification1
by Raoul Naroll

THIS PAPER is a discussion of the problem of defining of prime importance. As Ralph Gerard, a biologist
the "tribe" or "society," conceived of as the basic who has worked seriously with cultural problems, asked
culture-bearing unit-that group of people whose in a seminar I attended: "Exactly where is the skin of a
shared, learned way of life constitutes a whole culture?"
"culture" rather than a mere "subculture" on one hand
or a culture area or culture cluster on the other. I UNIT DEFINITION IN THE LITERATURE
propose a general concept of such a unit, whose appli- Considerable attention to the problem of unit d
cation to varying situations results in four types of has been given by English-speaking anthropologists.
culture-bearing units. Brief surveys of opinions among them have been
My interest in this problem first arose in connection published by Schapera (1953), Berndt (1959), and
with sampling design for cross-cultural surveys. In suchLeVine (1961). In reviewing this body of thought, it is
research, two important problems present themselves. important to distinguish opinions about the nature of
First, probability sampling theory requires that each the basic culture-bearing unit from opinions about the
sampling unit in the universe sampled have a mathe- proper unit to be used in comparative studies. The term
matically definable chance of being selected. Obviously,
"tribe" has often been used to mean a non-literate
the first step in meeting this requirement must be to culture-bearing unit (Berndt 1959:81); but this usage is
define the units being considered. For cross-cultural by no means universal. For example, Evans-Pritchard
surveys this task heretofore had never been rigorously (1940:278-79) uses c"tribe"' to designate a political
accomplished. Secondly, the validity of results of cross- unit within the Nuer people; but it is the Nuer as a
cultural surveys has been questioned in discussions with whole of whom he regularly speaks in his descriptions
anthropologists on the ground that the units were not of Nuer culture.
comparable. One man asked me at dinner table, with Reichard (1938:413f.) obviously has the basic
a note of disdain in his voice, "How can you compare culture-bearing unit in mind in her discussion of the
a little tribe like the Yahgan with a complex people primitive tribe as a unit. "By a tribe," she says, "we
like the English?" (See further discussion in Whiting usually mean an economically independent group of
1954:527-28). people speaking the same language and uniting to
Obviously, however, interest in this problem is not defend themselves against outsiders." But she notes that
confined to those concerned with comparative method some 'etribes," like the Toda, are not economically
but is shared by all students of culture, since the nature independent, while other "tribes," like the Dobuans,
of the culture-bearing unit is itself a scientific question act only rarely as a unit. Further, she adds, a tribe is
a closed society, with laws and morals applying only
RAOUL NAROLL is Associate Professor of Anthropology at to its members. Radcliffe-Brown (1 940a:xii-xiii)
Northwestern University. In the summer, he serves at Los speaks of Australian aborigines as "divided into some
Angeles with the Institute for Cross-Cultural Studies. Born in hundreds of separate tribes, each with its own language,
1920 in Toronto, Canada, he was educated at the University
of California, Los Angeles (A.B., 1950; Ph.D., 1953). He has
done field work in Germany and in the Austrian Tyrol, 1 Work on this paper began in 1954 at the Center for
Currently he is engaged in cross-cultural research on two Advanced Study in the Behavioral Sciences, where the writer spent
major problems, culture stress and deterrence. He has been much time in discussion of the problem in a work group with
working out methodological problems of cross-cultural studies Allan Holmberg, and where James Bosch served as a research
for the last ten years. This paper is directed at one of the assistant, gathering relevant data. Sol Tax repeatedly urged the
main problems encountered in intersocietal comparisons. It was importance of the problem upon the writer and repeatedly called
begun in 1954, and has been discussed with many anthro- upon him to prepare a manuscri-pt. But although a draft embodying
pologists since then. substantially the present scheme was written in 1956, not until
The present article, submitted to CURRENT ANTHROPOLOGY 3 XII recently has the writer had time to put the material in final form.
62, was sent for CA* treatment to 64 scholars of whom the For this opportunity he is indebted to Grant M-382 1, National
following responded with written comments: Ronald M. Berndt, Institutes of Mental Health, U.S. Public Health Service; Grant
Frank B. Bessac, Eliot D. Chapple, Gertrude E. Dole, Harold NSF G-13141, National Science Foundation; and a contract under
E. Driver, Paul R. Ducey, Melvin Ember, Helmuth Fuchs, Hans Project Michelson, U.S. Naval Ordnance Test Station, China Lake,
Hoffmann, Mervyn Jaspan, David Landy, Edmund Leach, Otto California-all three subventions being in support of aspects of the
von Mering, Simon D. Messing, Frank W. Moore, Ramkrishna larger War, Stress and Culture project of which this unit definition
Mukherjee, George P. Murdock, Artur Hehl Neiva, Gideon classification forms a part. Colleagues who read the manuscript
and Andree Sjoberg, Leigh M. Triandis, Carl F. Voegelin, and offered valuable comments include Paul J. Bohannan, Donald
Linvill Watson, and John W. M. Whiting. The comments T. Campbell, Robert A. LeVine, Frada Naroll, Herbert Lewis
written for publication are printed in full after the author's and-especially helpful-Francis L. K. Hsu. I alone am responsible
text and are followed by a reply from the author. for errors and shortcomings.

Vol. 5 * No. 4 * October 1964 283

This content downloaded from 62.204.192.85 on Fri, 04 Aug 2017 13:39:37 UTC
All use subject to http://about.jstor.org/terms
organization, customs, and beliefs." To Fortes (1940: Nadel (1951:187) on the other hand recommends the
239) the term "tribe" likewise denotes a basic culture- "political group," that is
bearing unit, a "well-defined political or cultural entity
the aggregate of human beings who coordinate their
differentiated from like units." Fortes calls attention effort for the employment of force against others and for
to the difficulty of using such a concept in discussing the elimination of force between them, and who usually
the Tallensi or their neighbors: count as their principal estate the possession and utilization
of territory.
no "tribe" of this region can be circumscribed by a precise
boundary-territorial, linguistic, cultural or political. Each Schapera (1956:8-10) likewise favors the political
merges with its neighbors in all these respects. In the unit, however small, as the basic unit of comparison.
transition zones between two "tribes" well communities
For the universe of study, within which comparisons
equally linked by residential contiguity and by structural
ties to both.
are to be made, he, like Radcliffe-Brown and Leach
(1954) favors the region. As Schapera (1953:359-60)
The concept of society as that of a basic culture- explicitly pointed out, the results of these regional
bearing unit is clearly expressed by Fortes and Evans- comparisons ultimately were themselves to be com-
Pritchard (1940:23): pared; but this presumably would be the work of
The social structure of a people stretches beyond their future generations.
political systems, so defined, for there are always social Driver's regional study (1956:15), comprising the
relations of one kind or another between peoples of entire North American continent,
different autonomous political groups. Clans, age-sets, ritual
divided extensive territorial groups into the minimum
associations, relations of affinity and of trade, and social
number of units necessary to display all the variation
relations of other kinds unite people of different political
present in the data on the six topics under consideration ...
units. Common language or closely related languages,
similar customs and beliefs, and so on, also unite them. Uniformity in trait inventory of two neighboring groups
Hence a strong feeling of community may exist between who spoke different languages was not regarded as sufficient
groups which do not acknowledge a single ruler or unite reason to lump them.
for specific political purposes. Community of language and Thus we have at least six criteria proposed for
culture, as we have indicated, does not necessarily give rise
defining whole societies, or other units of comparison:
to political unity, any more than linguistic and cultural
dissimilarity prevents political unity. (1) Distribution of particular traits being studied.
(2) Territorial contiguity.
These views may be compared with some others.
(3) Political organization.
Murdock (1953:478-79) proposes (4) Language.
to define a culture as including all local cultural variants (5) Ecological adjustment.
exhibited by communities within a particular geographical (6) Local community structure.
area which speak mutually intelligible languages and have
Before discussing the theory of ethnic unit definition
essentially similar forms of economic adjustment.
or proposing my taxonomic scheme, let me review the
Ember (1963) defines his sampling unit as "a con- theoretical and practical difficulties connected with
tinuously distributed population whose members speak these six criteria from various points of view.
a common language or lingua franca which is different Distribution of particular traits being studied.-
from the dominant languages of any neighboring Where the societies being studied include all those found
societies." within a single geographical area, this is operationally
One common thread runs through the foregoing feasible and methodologically clean. For Driver's North
comments. Reichard, Radcliffe-Brown, Fortes, Evans- American study, it seems to have been successful. But
Pritchard, Murdock, and Ember all conceive of the for comparative studies in which scattered samples are
basic culture-bearing unit as a unit in which a common to be studied, the method is prohibitively expensive
language is spoken. Had the Tallensi differed sharply because it requires the study of all the neighbors of each
from their neighbors in speech, instead of blending society in the sample, and their neighbors, and their
gradually along a linguistic continuum, Fortes would neighbors, until boundaries have been reached for all
not have complained of the difficulty of defining the traits being studied-or, if a linguistic criterion is
"tribal" boundaries in the Tallensi country. On the also used, as by Driver, until a sharp language bound-
other hand, none of these writers thinks that speech ary is reached.
community alone can be relied on to define the basic Territorial contiguity. -This criterion seems impor-
culture-bearing unit, and they do not agree on what tant where neighboring societies have generally similar
else needs to be considered. cultures and speak inter-intelligible dialects but are
Another point of disagreement among English- separated by uninhabited geographical gaps, whether
speaking writers lies in the nature of the unit to be water gaps or wasteland gaps. Such gaps seem to be the
examined in comparative studies. Murdock and Ember crucial boundaries between many Polynesian and
wish to compare whole cultures. Radcliffe-Brown Eskimo groups. On the other hand, water gaps seem
(1940b:4-5) proposes we take "any convenient less important to peoples who share membership in a
locality of a suitable size" and study "the structural single political unit. Also, short gaps of water or waste-
system as it appears in and from that region, i.e., the land obviously are unimportant; the gaps between the
network of relations connecting the inhabitants home islands of Japan or between the Samoan islands
amongst themselves and with the peoples of other of Savaii, Upolu, and Tutuila obviously are inconse-
regions." Whiting (1954:526) likewise recommends the quential. Operationally, large gaps are usually easy to
local comxmunity as the basic unit of comparative study. find out about.

284 CURRENT ANTHROPOLOGY

This content downloaded from 62.204.192.85 on Fri, 04 Aug 2017 13:39:37 UTC
All use subject to http://about.jstor.org/terms
Political organization.-States-territorial units em- Naroll: ON ETHNIC UNIT CLASSIFICATION

bracing a considerable number of local communities


into a political unit wielding exclusive war-making Yagua to other Peban dialects (Peba, Yameo, Tucuman)
authority-are generally treated by both the people is not at all clear (Steward and Metraux 1948:713,
involved and the ethnologists as distinct cultures. 728-29; Rivet 1911:176-77, 182). Eastern and
Murdock himself so treats neighboring states not dif- Western Timbira probably are mutually unintelligible
fering importantly in ecological adjustment and speak- (what is our definition of intelligibility?) but clearly
are on a linguistic continuum (Snethlage 1930:187-
ing intertelligible (mutually intelligible) dialects, e.g.,
Zulu and Swazi. On the other hand, where no such 199; Nimuendaju 1915; 1946:6). According to Powder-
territorial units exist, and where each local community maker (1933:15, 31), the Lesu dialect was understood
is politically independent, retaining full control over only in four other nearby villages; but there seems
war-making, not only the people involved but also reason to suspect that Lesu is on a linguistic continuum
ethnologists commonly ignore the criterion of political running the length of not only New Ireland but also
organization in defining tribes. Thus to give only a fewNew Britain (Capell 1962:89-94). Hupa seems to be
examples, the following societies aboriginally containedmutually intelligible with neighboring Athapaskan
many independent political organizations (made up, dialects to the south (Leldin, Hleltin) and west (Whil-
in some, of the single nuclear family): Copper Eskimo,kut) (Goddard 1903:7; 1911:92-93). The Flathead
Hopi, Southern Paiute, Samoans, Ainu, Chuckchee, dialect seems to be on a Salish linguistic continuum
Nama Hottentots, Naron Bushmen, Yahgans, Onas. (Swadesh 1950:159) and evidently is readily inter-
Data on political organization membership are usually telligible with Kalispel and Pend d'Oreille (Teit 1930:
easy to find and interpret. 303). Ulithians speak a Central Caroline dialect; but
Language.-Where people are separated either by these dialects are not mutually intelligible and it is an
substantial water or wasteland gaps, or by state open question how intertelligible Ulithi (Mogmog) is
boundaries as above described, it is common for peoplewith the dialect of nearby Sorol or Fais (Damm 1938:
speaking intertelligible dialects to deny membership230-31,
in 297ff., 372ff.; Kramer 1937:390ff.; Cheyne
a common "tribe" and for ethnologists to treat them1852:195;
as Fritz 1911:6-7).
belonging to distinct cultural units. Furthermore, it notNama does not appear to be mutually intelligible
seldom happens that people speaking intertelligible dia-
with Korana, Cape Hottentot or Gricqua but does seem
lects are commonly classified as members of distinct to be readily intelligible with Bergdama, although there
cultural units if they differ markedly and obviouslyarein some differences (Schapera 1930:45, 224; Hoernle
their ecological paterns and occupy distinct and fairly
1925:3-4; Stow 1910:265; Godee-Molsbergen 1916:
large contiguous territories. On the other hand, it is215-233; Barrow 1801:382-83, 389; Beach 1938:
rather uncommon to class a group of people as a single8, 181; Wandres 1927). Lepcha would seem to be non-
"tribe" who speak neither a generally intelligible dia-
intertelligible with its nearest relatives among the non-
lect nor a lingua franca (understood as a second pronominalized Himalayan dialects of the Tibeto-
language by leading men in every local community). Himalayan group of the Tibeto-Burmese family, but
This uncommon classification is probably applied most the evidence on this score is not from informants' state-
frequently to simple foraging peoples like the Anda- ments nor from linguistic experiments but only from
manese and the Yahgans. examining the available word lists (Hodgson 1857;
However, the operational difficulties of applying theKunow 1906; Campbell 1868).
linguistic criterion are formidable, and neither Murdock But even where the linguistic data is clear, the classi-
nor Ember claims to have systematically examined thefication problem often remains formidable. What do
linguistic data of the societies in their sample to see we do about one-way intelligibility, where speakers of
whether in fact they meet the linguistic criteria pro- Dialect A understand speakers of Dialect B but not
posed. If they had done so, they could hardly have vice versa? Operationally, linguistic classification is
avoided discussing the operational problems involved extremely technical and difficult at best; with the
in working with linguistic continuums or in defining materials available in existing ethnographic literature
mutual intelligibility. I have systematically surveyed it is often sheer guess-work. If, as seems to be the case,
the linguistic evidence for the thirty tribes of my first Eastern and Western Timbira are mutually unintelli-
pilot study (Naroll 1956). I, find serious linguistic gible dialects but are connected by over a dozen geo-
classification problems in eleven of the thirty: Yahgan,graphically intermediate dialects, each of which is
Yagua, Ona, Fiji, Lesu, Ramkokamekra Timbira, readily intelligible to its neighbor, where does Eastern
Hupa, Flathead, Ulithi, Nama, and Lepcha. Timbira stop and Western Timbira start? Where is the
Although Yahgan dialects are said to be mutually skin of the culture here? Paul Bohannan in conversation
intelligible, they are also said contradictorily to differhas spoken of linguistic continuums in Africa whose
as much as Scotch and Cockney (Lothrop 1928:120). poles are radically different languages and yet which
Nothing but formal linguistic experiments of the nowhere present a sharp linguistic break. Romance
Voegelin-Biggs type would persuade me that Ayreshirelanguage peasant dialects seem to present a linguistic
rural Scottish and Cockney are at all mutually intelli- continuum including dozens of mutually unintelligible
gible. Eastern Ona (Haush) and Western Ona differ variants and including literary Portuguese, Spanish,
similarly (Bridges 1948:92). It is easy to see that some French and Italian within their range. The German
nineteenth-century Fijian dialects were mutually un- linguistic continuum is the classic example; how do
intelligible but it is not so easy to divide them up into we treat a village, all of whose members claim to
mutually intelligible groupings (Williams 1858:216; speak mutually intelligible variants of German but
Capell 1941:464; Fison 1886). The relationship of who in fact do not? What is our operational definition

Vol. 5 N No. 4 * October 1964 285

This content downloaded from 62.204.192.85 on Fri, 04 Aug 2017 13:39:37 UTC
All use subject to http://about.jstor.org/terms
of intelligibility and how do we apply it to the of the Supreme Court in Washington, D.C., and vice
available word lists? Do we take an informant's word versa. Two of us (Naroll 1956; Ember 1963) are
for it? Most Viennese claim to understand Tyrolean interested in the size of the largest settlement in a
readily; but this is a myth; they usually cannot. society as a variable for study; such a variable by its
What do we do about bilinguality? trilinguality? very definition requires a society comprising several
Many Luisefio today seem equally at home in Luiseino, local communities. Folk-urban interrelationships
Spanish and English; many Hopi Tewa speak fluent cannot be understood without considering both the city
Hopi and English as well as Tewa. A Western and its hinterland; a Tyrolean village like Feichten in
Papago man may speak fluent English, but his wife not Kaunertal cannot be understood without looking care-
a word of it. Is he a member of American society be- fully at the small town of Prutz, the larger town of
cause he speaks the lingua franca and his wife not be- Landeck, and the cities of Innsbruck and Vienna.
cause she doesn't? And what about the children? In Finally, no one claims that the local community
some Tyrolean villages, children of six or seven neither usually embodies the full society; proponents of that
understand nor speak literary German, but only their concept are talking about the design of comparative
local dialect. Their parents are bilingual and the studies, not about the nature of the culture-bearing unit.
children are expected to become- bilingual.
Ecological adjustment.-I know of no case involving A NEW UNIT: THE "CULTUNIT"
societies who possess aboriginal states where this
criterion has been considered important by the people I now propose as an "ideal type" a concept of the
themselves or by ethnologists. But it is a criterion culture-bearing unit-the cultunit-which yields four
applied to a number of fairly well known "tribes" taxonomic classes. This taxonomy is proposed primarily
which lack aboriginal states, a criterion distinguishing as a conceptual norm to aid anthropologists in their
"tribes" who occupy contiguous adjacent territories thinking,
and rather than an actual unit of study either in
who speak intertelligible dialects. There are the desert- field work or in comparative studies. Field workers
foraging Papago and the riverine-farming Pima; and will often prefer to define their ethnic unit in terms of
similarly the desert-foraging Yavapai and the riverine- local conditions. In concomitant variation research
farming Havasupai; the cattle-herding Nama and the designs (e.g., Redfield 1941), the ethnic unit definition
foraging Bergdama; the reindeer Chukchi and the obviously must be suited to the study in question.
maritime Chukchi; the reindeer Koryak and the Comparativists in cross-cultural surveys often will
maritime Koryak-to give the best known examples. prefer to use the units in Murdock's Outline of World
Theoretically, this criterion is bothersome because it Cultures (1958), in his World Ethnographic Sample
is ignored by everyone in considering the inhabitants (1957), or the units of the Human Relations Area Files.
of large, complex societies where division of labor is However, I appeal to anthropologists in the field to
commonplace and no one follows a "typical" ecologicalidentify the cultunit or cultunits involved in their
adjustment. Operationally, it is formidable to apply; study, by collecting the relevant linguistic, political,
where societies get part of their subsistence from and geographical data. Thus in studying the village of
farming and part from foraging, it often becomes aFeichten in Kaunertal, (Naroll and Naroll, 1962) the
problem to determine the predominant ecological cultural unit of reference for that field study is first of
pattern. Both Murdock (1957)-in his World Ethno- all the commune of Kaunertal and second, the cultural
graphic Sample-and Hobhouse, Wheeler and Gins- district of Oberinntal; but both these units form a small
berg (1930) have classified peoples widely by ecological part of the cultunit which as far as we can tell seems to
criteria, but they have not presented evidence that their consist of the peasants of Tyrol west of the town of
classifications are reliable, or trustworthy (which is Hall-in-Tirol.
by
no means the same thing; see Naroll [1962] on data
DEFINITION OF THE CULTUNIT
quality control). However, if this criterion were
applied only to stateless societies, and even then onlyCultunit.-People who are domestic speakers of a
when its crucial importance is explicitly claimed by common distinct language and who belong either to the
the ethnographer, this would meet the need presented same state or the same contact group.
by the examples given and solve the operational Distinct Language.-All those dialects mutually in-
problem. telligible to speakers of a stated dialect.
Local community structure.-Restricting the unit of State.-A territorially ramified territorial team
study to certain local communities is clearly a good idea whose leaders assert and wield the exclusive right to
in certain kinds of comparative study but clearly declare and conduct warfare.
impossible in certain other kinds. In a cross-cultural Territorially Ramified.-Made up of a mumber of
survey which seeks a random or stratified random component territorial teams. For example, the United
sample (Ember 1963; Naroll 1961b), such a unit defi- States is made up of fifty territorially defined states.
nition is operationally worthless. It is hard enough to Territorial Team.-A group of people whose mem-
get something resembling a complete list of societies, butbership is defined in terms of occupancy of a common
impossible to get anything even vaguely like a complete territory and who have an official with the special
inventory of local communities. Furthermore, to this function of announcing group decisions-a function
operational difficulty must be added a theoretical one. exercised at least once a year.
Where the variable being studied is a political variable, Common. Territory. A geographically contiguous
often no single local community presents the trait as territory within which are found not only the dwellings
such. No one can understand the recent events in of the people of the team but also the lands of their
Oxford, Mississippi, unless he is aware of the behavior usual subsistence activities ashore.

286 CURRENT ANTHROPOLOGY

This content downloaded from 62.204.192.85 on Fri, 04 Aug 2017 13:39:37 UTC
All use subject to http://about.jstor.org/terms
Contiguous.-Accessible without crossing the land Naroll: ON ETHNIC UNIT CLASSIFICATION
territory of others. (N.B. Thus water gaps are ignored,
however great.) (Where word lists exceed two hundred entries, use only
Warfare.-Public lethal group combat between ter- those on the Swadesh glotto-chronology list [Hymes
ritorial teams. (N.B. Thus blood feuds between non- 1960]). If 80 per cent or more of the words examined
territorially defined kin groups are not considered war- constitute "recognizable cognates" consider the two dia-
fare.) lects mutually intelligible; otherwise, not. (2) Where
Contact Group.-People who belong to no state but no such comparative word lists are available follow
who speak a common distinct language and who are all statements by informants or ethnographers about intel-
interconnected by successive contact links. ligibility. (Such statements can probably be trusted to
Contact Link.-Two nuclear families constitute a agree with the "recognizable cognate" word-list method
contact link if their dwellings are not more than two except when the dialects concerned are in the "partially
hundred airline kilometers apart at some time in their intelligible" class.) Use availability or unavailability
annual cycle. of such word lists as a data quality control factor
Domestic Speakers.-People who predominately use (Naroll 1962:14-18).
a given dialect for speech within the nuclear family,
CLASSIFICATION OF THE CULTUNIT
that is, among husband and wife and their minor
children. It seems useful to distinguish four types of cultunit:
Hopi type.-People who belong to no state but who
DISCUSSION OF THE CULTUNIT CONCEPT
speak a common distinct language and who are all
This concept of a cultunit as a basic culture-bearing interconnected by successive contact links. Examples:
group makes use of three of the six criteria discussed Hopi, Bella Coola, Naron, Hupa, Nuer.
earlier. The criteria of language and territorial con- Flathead type.-People who belong to a state all of
tiguity are invariably used. The criterion of political whose members speak a common distinct language.
organization is also used wherever there is a sufficient- Examples: Flathead, Swazi, Tikopia.
ly authoritative political structure transcending the Aztec type.-People who belong to a state in which
local community. mutually unintelligible dialects occur and who are
Deciding dialect affiliation according to the dialect domestic speakers of a dialect intelligible to speakers of
predominately used domestically resolves most ques- the lingua franca of the state, that is, the dialect in
tions of bilinguality or trilinguality readily. Occasion- which the state officials usually transact their business.
ally, informants claim to belong to bilingual nuclear Examples: Aztecs, Incas, Zulus.
families which use two unintelligible languages with Tarascan type.-People who belong to a state in
Ccequal" frequency. Mathematically, this claim seems which mutually unintelligible dialects occur and who
implausible; such statements might mean only that the are domestic speakers of a dialect not intelligible to
informant does not know which language is more fre- speakers of the lingua franca of the state. Examples:
quently used in his household, or they might mean that Tarascans, Aymara, Zulu-ruled Thonga.
a lower prestige language is used more frequently. These four cultunit types emerge when we consider
No precise definition of mutual intelligibility is stated two different kinds of boundries; a linguistic boundary
above. The problem of dialect intelligibility has been and a communication-link (state or contact group)
studied in the field by several linguists whose interest in boundary. A cultunit boundary is formed by either of
this problem was stimulated by Carl Voegelin (Voegelin these boundaries. If a state has more than one distinct
and Harris 1951; Biggs 1957; Pierce 1954). These language spoken within it, then the political unit, as
studies seem to suggest that where two dialects have established by the political boundaries, is subdivided
approximately 90 per cent of their vocabulary cognate, into cultunits by the linguistic boundaries within the
little trouble is ordinarily experienced when speakers state. That cultunit formed by the domestic speakers
of one try to communicate with speakers of the other; of the lingua franca of the state is classified as an Aztec
apparently the substantial element of redundancy in type cultunit; the other cultunits within that state are
language is able to accommodate itself to this much classified as Tarascan type cultunits. Thus by definition
informational "noise." On the other hand these studies we cannot have an Aztec type cultunit except in asso-
suggest that where two dialects have as little as 70 per ciation with Tarascan type cultunits. (Theoretically,
cent of their vocabulary cognate, speakers of one we could have a state made up entirely of Tarascan
commonly have great difficulty communicating effec- type cultunits, provided the lingua franca of the state
tively with speakers of the other. Two dialects with is not spoken domestically by any one. This may indeed
more than 70 per cent but less than 90 per cent of prove to be the case in Switzerland, where the pre-
cognates might perhaps be termed "partially intelli- dominant lingua franca may well be literary German;
gible." literary German seems to be spoken little if at all in the
While the definitive solution of this problem must be home by the Allemanic Swiss, who prefer one or another
left to linguists, as a practical working method for of the several mutually unintelligible varieties of
comparative ethnologists today, I propose the follow- Schweizerdeutsch. The question of identifying the
ing: (1) Where comparative word lists including at actual lingua franca of supposedly multilingual states
least twenty words are available in a consistent phone- like Canada, Belgium, South Africa, and Switzerland
tic notation, consider as "recognizable cognates" only who theoretically have more than one is a matter for
those words which are reported to share at least one research; presumably one language in each such state
meaning in common and which also have two-thirds of is most commonly used to transact business among
their phonemes in common and in the same order. officials.)

Vol. 5 N No. 4 - October 1964 287

This content downloaded from 62.204.192.85 on Fri, 04 Aug 2017 13:39:37 UTC
All use subject to http://about.jstor.org/terms
Where a state is linguistically homogeneous, the ity as a whole than patrilineal descent, matrilineal
state boundary is followed and we get a Flathead descent, or double descent. Murdock was as much in-
type unit. Where there is no state, linguistic boundaries terested in double descent, a rare phenomenon found
only are followed unless contact gaps occur which only among a few exotic peoples, as he was in bilateral
break up speech communities into cultunits within and patrilineal descent, one or the other of which is
which speech communication presumably flows freely found among the vast majority of mankind. Conse-
but between which speech communication presumably quently, the number of people who model their be-
is rare and unimportant; and the speech communi- havior upon a given culture pattern is irrelevant per
cation groups thus defined form Hopi type cultunits. se; if as occasionally happens a particular culture
It is important to remember that some cultunits pattern varies in its social implications with the numbe
differ very little in culture from one another while of people involved, then but only then is the popu-
others differ very much. Frequently, neighboring lation of the culture-bearing group-the cultunit-
cultunits may differ not at all with respect to certain relevant. In such a case, social population is ex hypo-
traits being studied. But methods of avoiding the thesi a factor influencing the situation being studied
fallacy of counting neighboring and closely related and needs to be considered in some sort of multiple
cultunits repeatedly as independent trials of a func- variable analysis along with other factors being con-
tional hypothesis being tested are presented elsewhere sidered. If an investigator suspects that social popu-
(Naroll 1961a; Naroll and D'Andrade 1963). lation may be relevant to the problem he is studying,
In evaluating the cultunit concept, anthropologists he should collect population data and test its relevance
should bear in mind that it is offered as an ideal type, directly.
a standard of reference, and primarily to solve sampling In his latest statement about comparative statistical
problems in cross-cultural surveys. Field description studies, Leach (1963:174) points out correctly that
units often will not correspond to the cultunit. Thus social facts may be statistical phenomena like demo-
the cultunit is an analytical device parallel to that of graphic or economic information, or they may be non-
the great cultural tradition as used by Hsu (1963: statistical jural rules; for the statement of jural rules,
chap. II). The cultunit sometimes may be a genuine Leach correctly reminds us, statistics are wholly irre-
social system but often will be only theoretical-an levant. From this he infers that statistical procedures
analytical abstraction. Considered as an analytical are likewise irrelevant in testing theories about jural
abstraction, the cultunit concept is valuable even rules. Not so. Within each cultunit we search for the
though as Leach (1954) points out, a given human being predominant jural rule. But in a statistical study of a
may function simultaneously as a member of two number of cultunits, we may count the frequencies with
distinct societies, based on two distinct cultunits. which certain sorts of jural rules occur and may com-
Indeed, the distinction between society and culture pare these frequencies with the implications of a theory
becomes critical when we consider Leach's informant about jural rules. Such a comparison is the purpose of
who functioned both as a Shan and a Kachin. A culture a cross-cultural survey. A cross-cultural survey uses
is a pattern, a set of plans, a blueprint for living. Every facts about particular cultunits to test theories about
culture includes as an element a social system, that is, culture in general.
a plan for social interaction. A society is an actual Since the cultunit concept is offered as a tool for
system of social interaction. Among human beings, comparative statistical studies, it is desirable to have
such a system usually resembles rather strikingly the that concept as nearly uniform as possible. I have thus
plans in the minds of its members; but this resemblance reduced the number of proposed cultunit types to four,
seldom if ever attains complete identity. Social practice which seems to me the feasible-minimum. Were we to
in most if not all societies departs somewhat from adopt instead the policy of using the socio-cultural
social theory. The distinction between culture and unit recognized by the natives being studied, or by
society becomes evident the moment we look at social their ethnographers, a number of inconveniences would
insects. These have societies but lack cultures. The result. Some peoples have no consciousness of them-
social organization plan, which in a human society is selves as a culture-bearing unit, no name for them-
taught anew to each generation of children, is among selves as a group, no native definition of their own
social insects communicated genetically. The cultunit ethnic unit. Others define themselves by traits peculiar
is offered as a unit of comparative statistical analysis to themselves, as logracing among the Timbira. Still
of sets of plans-of social and cultural patterns as they others recognize a nest of unit affiliations whose
exist in the minds of culture bearers. Study of actual relative importance is difficult to assess. Thus the
social systems is of course relevant to the study of the peasants of the hamlet of Unterhaiuser near Landeck
plans which are supposed to govern them, but the are in their own minds first of all parishioners of
unit of comparison is the plan, not the society. Feichten; second, citizens of the commune of Kauner-
That is why it is usually irrelevant how many tal; third, Oberinntalers; fourth, Tyroleans; fifth,
people follow the plan-whether only a few hundred Austrian citizens; sixth, Germans; seventh, Europeans;
or many million. The idea of a comparative statistical and eight, Christians of the Roman Catholic persua-
study, a cross-cultural survey, is not to assess the general sion. They pay little attention to those dialect dif-
trends among the human race as a whole by counting ferences which make their speech unintelligible to
heads, as Leach suggests (Leach 1960:137). Instead, it eastern Tyrolean peasant, and most are probably un-
is to compare culture patterns in order to see if there aware that these differences exist at all. Their state
are general tendencies governing their construction. affiliation with Austria seems less important to them
In Sociazl Structure, Murdock had no interest in learning than their character as Tyroleans.
if bilateral descent was more characteristic of human- Ethnographers follow no consistent pattern of unit

288 CURRENT ANTHROPOLOGY

This content downloaded from 62.204.192.85 on Fri, 04 Aug 2017 13:39:37 UTC
All use subject to http://about.jstor.org/terms
definition. The perils of following them blindly are Naroll: ON ETHNIC UNIT CLASSIFICATION
best illustrated by the difficulties of Hobhouse,
Wheeler, and Ginsberg (1930): for example, these men "mouth to mouth"-i.e., from mind to mind-are
counted the Indians of Lower California collectively obviously key elements in defining the culture-bearing
as one unit, and the Waicuri and Pericui of Lower group. A speech community tends to form such a com-
California as two other units; the Hopi as one unit munication net, even though its boundaries may be
and the Moqui (another name for Hopi) as another blurry; translation problems constantly remind us that
unit; the Azande as one unit and Niam-niam (another even closely related languages often lack equivalent
name for Azande) as another unit. In short, in ethnic terms for key cultural concepts. Thus literary German
unit classification as in other concept definition prob- has no equivalent for the English word "fair" as in
lems, the rule by Goodenough (1956:37) applies: the "fair play"; nor has English any equivalent for the
concepts which are suitable for describing behavior German Heimat, which is neither "home" nor "home-
patterns of particular cultures are not necessarily the land," but a little of both. A political state, with its
same as those suitable for comparing these cultures with specialist leaders, forms a communication net of another
others. sort, one in which key concepts can be coined and
Having said all this, it is worth remarking that while argued or ordered into use by the expert leaders of
the cultunit concept often diverges from the concepts the state. In the absence of state organization, a gap
used by natives to identify their own groups or by of two hundred kilometers between the nearest settle-
ethnologists to describe native cultures, yet these diver- ments of technologically primitive people must ordi-
gences are rarely extreme or arbitrary. Usually the narily represent a clear break in the speech commu-
divergences are moderate and within the range of nity's communication net. Samoans and Tongans spoke
variation of divergences among varying native and varieties of Polynesian sufficiently similar so that the
varying ethnographic concepts. The cultunit, I submit, crucial communication question was less often "Can
generally resembles fairly closely at least some you understand what they are saying?" than "Do you
operating social unit within the society wherever these ever talk to them?"
units have sharp boundaries. Its most arbitrary element
is its delineation of boundaries along linguistic con- THE CULTUNIT IN TIME
tinuums from a bibliographic vantage point (see below). The problem also arises of defining cultunits in time
The only ethnic unit definition criterion not con- rather than space. This problem has often been neg-
sidered in the foregoing discussion is that of religion. lected by comparativists in cross-cultural surveys, with
In most civilized parts of Eurasia, as Francis Hsu results which are sometimes disastrous (Pilling 1962).
reminded me in conversation, religion has long been In practice, societies which have kept no written
an important ethnic unit identification criterion. Wher-records might fall into any of three chronological
ever Christianity, Islam, or Buddhism has spread, re- types:
ligious cult membership has involved major cultural Palaeoethnography.-Societies described before
differences. And we are all familiar with the cultural 1492. Herodotus and Tacitus are the best known
significance of such religious groups as Jews, Parsees, European palaeoethnographers; but much palaeo-
Sikhs, and Mormons. In the twentieth century, secular ethnography has been written by Arab and Chinese
ideologies which embody many traditional Judeo- observers.
Christian moral attitudes often play a quasi-religiousAboriginal period.-Cultures described after 1492,
role: cultural differences between Communist and but anti-while still politically independent.
Communist Koreans, Chinese, or Vietnamese seem Colonial period.-Cultures described after 1492,
parallel to seventeenth-century differences between after European conquest. E.g., present-day Navaho
Catholic and Protestant Germans, French, or Ro- culture.
mansch. However, among the vast majority of native Societies which keep written records (including of
cultures, unconverted to religious or secular universal course societies taught to do so by Europeans) present
ideologies stemming from India, the Middle East, or an entirely different problem of periodization. The
Europe, ideological organization membership is almost only general attempts at periodization of history have
never an ethnic definition criterion. been made in connection with theories of rise and fall
Finally, the cultunit concept is not entirely without of civilization, like those by Spengler (1926), Toynbee
theoretical interest. A culture is a plan of behavior. (1947), and Quigley (1961)-of these, Quigley's is by
Although, as Schapera (1956:9) has pointed out, therefar the most nearly acceptable scheme for me. Of
are many Naron bands, there seems to be only one course, conventional historians construct and reach
standard Naron plan for band organization which all fairly broad agreement on particular periodization
the Naron band members seem to have in mind and schemes for particular historical traditions, although
which all Naron bands more or less model themselves they commonly haggle over precise boundary lines;
upon. Culture patterns however are not fixed and very often, of course, histories of monarchies are
immutable; on the contrary they are constantly divided dynastically; and art styles are, of course, also
changing, sometimes very slowly, sometimes very commonly used as period markers.
quickly. These changes presumably occur when The development of a satisfactory general theory of
opinion leaders deliberately or unconsciously change historical periodization is an urgent task, but one
the rules or the texts in repeating them and when fur- which I cannot attempt here. For sampling purposes,
thermore these changes are accepted by their associates,at least, it seems satisfactory simply to use arbitrary
whether deliberately or tacitly. The communication time units, like centuries. (Incidentally, for scientific
nets through which these changes are passed from purposes, it would be a vast improvement on time no-

-Vol. 5 * No. 4 * October 1964 289

This content downloaded from 62.204.192.85 on Fri, 04 Aug 2017 13:39:37 UTC
All use subject to http://about.jstor.org/terms
tation if the Christian era were dropped. Why not do with the unit, or some portion of it; regional
so by commencing time reckoning from the year ethnographies which deal collectively with groups of
10,000 B.C.? Thus the year 1 B.C. becomes the year sampling units (hence presumably with groups of
10,000 absolute; the year 1 A.D. becomes the year cultunits) are not to be considered. However, it would
10,001 absolute; the year 376 A.D. becomes the year seem wise to admit as a monograph on a particular
10,376 absolute; the year 1962 A.D. becomes the year cultunit a book which, while purporting to deal with
11,962 absolute; while the year 776 B.C. becomes the the entire sampling unit, actually is known to draw
year 09,225 absolute. Such a system would encompass its data chiefly from the cultunit in question. For
dates of all written records. Archaeologists might preferexample, Gusinde (1937), while purporting to deal
to begin an archaeological reckoning 110,000 B.C. or with the Yahgans as a whole, actually draws his
1,110,000 B.C., leaving all absolute historical dates un- data almost entirely from the Ushaia Yahgan; Gusinde
changed in their last five digits while keeping all dates (1931) while purporting to deal with the Onas as a
positive). whole, actually draws his data chiefly from the Ona
proper rather than the Haush Ona; and Radcliffe-
THE CULTUNIT IN CROSS-CULTURAL SURVEYS
Brown (1922:viii, 13, 24-25) while purporting to
Now some comments about the use of the cultunit deal with the Andaman Islanders as a whole, actually
concept in cross-cultural surveys. Ideally, random draws his data chiefly from the North Andaman
sampling should be made from a list of cultunits whose group, whose dialects seem to have been mutually in-
linguistic boundaries or characteristics are described telligible.
and to which a bibliography of basic ethnographies is Some of these units will prove to embrace more than
attached. Such a list should ideally include all cultunits one cultunit as I have defined it above. Some of them
about which a minimally acceptable ethnographic will prove to constitute a cultunit as so defined. Some of
literature exists. I have taken a random sample of all them will prove to comprise only a part of a cultunit.
societies in Murdock's Outline of World Cultures Some of them will prove to be on linguistic continuums
(1958) within world-wide diffusion strips or arcs such that cultunit definition on the terms proposed here
(Naroll 1961a:24-25). Ninety per cent of those which is not possible until a particular "stated dialect" is first
had calendared topically organized monographs on specified. And some of them will prove impossible to
cultunits or on portions of cultunits which containedclassify because the necessary linguistic data is una-
both data on aboriginal periods and certain specified vailable (in some case, forever unavailable because no
data on suicide or homicide were included in Murdock's informants survive). Cultunit classification should
World Ethnographic Sample (Murdock 1957). Thus a proceed with the dialect spoken by the people described
universe which consists of Murdock's World Ethnog- in the monograph which qualified the sampling unit
raphic Sample plus the record-keeping nations repre- bibliographically. Rarely will more than one such
sented in the United Nations Demographic Yearbook monograph involving mutually unintelligible dialects
includes 40 per cent of the modern cultures which are turn up for a single OWC rubric.
extensively described in calendared ethnographic Since cross-cultural surveys usually seek correlations,
literature. (In this sample, no literature was considered the ethnic unit definition problems involved in them
unless it was listed in one of the standard ethnographic boil down to two. First, does the actual sampling unit
bibliographies or was in the Human Relations Area definition used bias the sample and thus affect the
Files). correlations found in the study? Second, does variation
At present, a low-budget cross-cultural survey then in the type of unit studied affect the correlations
would seem to be well-advised to use the World found in the study? This second question is the classic
Ethnographic Sample as its sampling universe; a mod- problem of comparability of units-if the variation in
erate-budget survey would seem to be well-advisedtype to of unit does not affect the correlations the units
sample from the Outline of World Cultures, at least are obviously comparable.
for those areas in which good ethnographic biblio- To answer the first question above, control the
graphies exist (Negro Africa, Oceania, and America sample by classifying it into two groups: Group 1-
north of Guatemala); while a high-budget survey those sampling units which clearly include all of one
might use the Outline of World Cultures as its but no more than one bibliographically acceptable
sampling universe for the whole world. (Certain cate-cultunit; Group 2-the remainder of the sample, i.e.,
gories of OWC units are of course to be ignored if the sampling units which depart from the model or
used as an ethnographic universe for sampling pur- whose boundaries are not clear. Then correlate this
poses.) dichotomy with the traits being studied to see if the
Whether the World Ethnographic Sample or the two groups differ importantly and significantly. For
Outline of World Cultures is used as a sampling uni- example, in a study investigating the relationship
verse, the comparativist will find that many of the between residence rules and descent rules, the sampling
units randomly chosen must be rejected because they
unit control consists in seeing whether Sampling
fail to meet bibliographic requirements. (These requi-
Group 1 and Sampling Group 2 differ in their pro-
rements should be specified objectively and precisely,
portions of matrilineal, patrilineal, bilateral or double
of course.) So sampling must continue in a pre-descent and in their proportions of matrilocal, patri-
determined manner until an acceptable number local,
haveneolocal or bilocal residence. If they do not so
been identified. differ, no harm has been done by the departures from
The bibliographic unit of examination must bethe
the
model in Group 2. If they do so differ, then Group
unit named in the sampling universe list. Only those
2 should be examined to see whether it contains a higher
ethnographies can be examined which purport to deal
proportion of "hits" (cases confirming the hypothesis

290
CURRENT ANTHROPOLOGY

This content downloaded from 62.204.192.85 on Fri, 04 Aug 2017 13:39:37 UTC
All use subject to http://about.jstor.org/terms
being tested) than does Group 1. If not, then there is Naroll: ON ETHNIC UNIT CLASSIFICATION
still no harm done; but if so, then sampling errors have
biased the results of the study and the investigator of these anthropologists thought that language alone
must allow for these errors by differential weighting was an adequate unit definition criterion.
or by eliminating the excessive proportion of "hits" A new concept of culture-bearing unit, the cultunit,
from Group 2. To answer the second question above, is proposed for use in cross-cultural surveys. The con-
proceed similarly, seeing whether Hopi type units, cept is offered primarily as an ideal type, a base of
Flathead type units, Tarascan type units, Aztec type reference, rather than as an actual working category.
units differ from one another with respect to the traitsA cultunit is defined as a group of territorially con-
being studied, and if so, whether they differ likewise intiguous people who not only are domestic speakers of
proportion of "hits." mutually intelligible dialects but also belong to the
same state or contact group.
SUMMARY Cultunits are classified into four types: the Hopi
type (stateless), the Flathead type (linguistically
This discussion of the problem of defining the "tribe" homogeneous state), the Aztec type (ruling group of a
or "society," that is, the basic culture-bearing unit, linguistically diverse state), and the Tarascan type
begins by reviewing the opinion of a number of (subject group of a linguistically diverse state).
English-speaking anthropologists. Six criteria proposed The problem of classifying units in time as well as
by them are considered: (1) distribution of particular space is considered. Ethnographic data might well be
traits being studied, (2) territorial contiguity, (3) po- classified chronologically as palaeoethnography (be-
litical organization, (4) language, (5) ecological adjust-fore 1492), aboriginal (after 1492 but while still in-
ment, and (6) local community structure. The criterion dependent) and colonial (after conquest by record-
of language is considered essential by all but there are keeping societies).
serious practical difficulties in applying it: (1) defining Finally, a statistical method is proposed to control
and measuring intelligibility of dialects, (2) marking biases due to non-comparability of units or to discrep-
boundaries along linguistic continuums, and (3) ancies between the cultunit concept and the actual
classifying multilingual speakers. Furthermore, none sampling units.

Abstract who are domestic speakers of mutually intelligible dia-


lects and who also belong to the same state or contig-
This paper discusses the general concept of the basic uous contact group. Four types of cultunits can be dis-
culture-bearing unit and proposes a new definition- tinguished: theHopi type, a contiguous linguistic group
the cultunit. This proposal is a response to the need for who belong to no state; the Flathead type, a state whose
units of cross-cultural surveys to be comparable and to members all speak a single language; the Aztec type,
be rigorously defined if these studies are to be validated domestic speakers of the lingua franca of a linguisti-
statistically. Such anthropologists as Schapera, Berndt, cally diverse state; and the Tarascan type, domestic
Whiting, Evans-Pritchard, Reichard, Radcliffe-Brown, speakers of a language other than the lingua franca of
Fortes, Ember, Murdock, Nadel, Leach and Driver a linguistically diverse state. Three periods of time of
have given much thought to the problem of defining any of these four types can also be usefully distin-
basic culture-bearing units. From their writings, we guished: palaeoethnographic periods, aboriginal periods,
can see at least six possible criteria for defining societal and colonial periods. Although the cultunit has some
units: (1) language (which nearly everyone thinks is theoretical justification, it is offered as an arbitrary
important), (2) political organization, (3) territorial definition whose justification is its convenience in
contiguity, (4) distribution of particuar traits being cross-cultural surveys. Through its use, sampling biases
studied, (5) ecological adjustment, and (6) local com- from inconsistent sampling units can be statistically
munity structure. The cultunit concept uses the first controlled, and the importance of diversity of societal
three of these: a cultunit is defined as a group of people type can be statistically assessed.

discussion, one may reasonably expect sidering acceptance or nonacceptance


Comments the descriptions to be modified or of "new" terms and definitions.
elaborated according to the demands In proposing a new term, "cultunit"
By RONALD M. BERNDT* which are made on them, for example, (which could easily be mistaken for
Perth, Western Australia 31 I 64, as new evidence comes to hand. meaning a social unit made up of
Methodologically, then, they can be members of a religious, magical, or
Definitions are, in one sense, shorthand useful; but one of the tests of their secular unit), Naroll takes as his point
descriptions of frequently used con- usefulness is their relevance to specific of departure the concept of "tribe"
cepts or terms, more or less agreed- empirical situations-how far they do or "society." Both these are, he says,
upon formulae enabling discussion to indeed "fit" what they are intended unsatisfactory for comparative pur-
continue without constant explanatory to represent in the real situation. We poses, although he is able to draw
diversions. But in the course of such should keep this in mind when con- upon two (and in part, a third) of the

Vol. 5 * No. 4 * October 1964 291

This content downloaded from 62.204.192.85 on Fri, 04 Aug 2017 13:39:37 UTC
All use subject to http://about.jstor.org/terms
criteria emphasized in discussion of of any one interact with those ol research workers in the New Guinea
them, in constructing the definition ofseveral others to form a small social Highlands to use this term loosely for
his new term. Naroll rightly sees a world of their own, which overlaps such disparate units as language unit,
need for rigorous definition of "the with other small social worlds. phratry, district, and so on, and its
basic culture-bearing unit" and "the 2. This position is comparable, up use in respect of Aboriginal Australia
proper unit to be used in comparative to a point, as regards many other is so general that I doubt whether a
studies." He claims that "it is im- Australian Aboriginal "tribes." For substitute would meet with any but
portant to distinguish opinions about" instance, in northeastern Arnhem very limited acceptance. Also, I have
these two kinds of unit, but combines Land, each of two exogamous not been able to include a number of
them in putting forward his "new patrilineal moieties includes a number features which would have to be taken
concept of culture-bearing unit," of named, territorially-based, (exo- into account even in "comparing"
designed "for use in cross-cultural gamous) dialect units (mada) and these 3. Nevertheless, I am not
surveys." I found Naroll's discussion "clans" (mala) in a variety of com- convinced that the proposed "ideal
in the main body of his paper a useful binations; each person belongs by type cultunit does, in fact, improve on,
sorting-out of views and detail on descent to a mada-mala pair. The or point up more specifically, what is
this general topic, and the section extent of vocabulary held in common usually designated by such terms as
on "Language" particularly so. As a between any two or more mada has "tribe" and "society," or "cultural
clearing of the ground, so to speak, not been assessed, but it seems fair to bloc" and "an area of common
or (to mix the metaphor) a sorting- say (without going into detail here) culture." Nor do I follow Naroll in
out of the plants already growing that they are not mutually intelligible:his discussion of the concept of
there, the article certainly fills a gap.communication is possible because society as it appears within his cult-
I do not, however, feel the same each adult has learnt some vocabulary unit.
enthusiasm for the new concept of of at least one other mada. With The assertion that if we were "to
cultunit. As presented, it reads to me patrilocal residence, the domestic adopt instead the policy of using the
like a first draft of such a concept- language is predominantly the mada socio-cultural unit recognized by the
showing distinct possibilities, not least of the husband-father, but, a woman natives being studied, or by their
as regards the linguistic distinctions it uses her own mada in speaking to ethnographers, a number of incon-
incorporates, but requiring a great her young children. Overall, and veniences would result" is not open to
deal more work even if "proposed allowing for these differences, the question; what is at issue is (a)
primarily as a conceptual norm to aid culture is very similar: the "vocabu- whether such inconveniences would
anthropologists in their thinking." lary of culture" parallels the dis- worth surmounting, and (b) whether
"Ideal type" or not, its definition tinction between dialects, not that this "policy" would entail more of
leaves much to be desired. For between distinct and separate lan- them than does the new design he
example, the requirement that a guages. Members of adjacent and presents.
"territorial team" should "have an affiliated mada and mala come
official with the special function of The rationale is indicated in the
together for (for example) ritual and
announcing group decisions" is a warning that the cultunit concept is
ceremonial performances, but members
subsidiary part of the definition ofofa the whole cultural bloc never act "an analytical device," "a standard of
"state." Cultunits of "state" type reference," which need have no exact
corporately, or meet as a group.
make up 3 of the 4 types which counterparts in reality. He affirms
3. Or, take a further example from that the cultunit "resembles fairly
Naroll distinguishes, the number
the eastern Highlands of New Guinea. closely at least some operating social
being so limited in order "to have
that concept as nearly uniform as What Naroll would call non-inter- unwit within the society wherever
telligible languages serve as broad these units have sharp boundaries."
possible." This means, of course, that
the "state"-less type must be used to membership-labels for a number of But he is emphatic that in his view
"big names" (native districts), each "the unit of comparison is the plan,
cover a diversity of sub-types, in-
containing two or more villages or not the society." Study of actuality
ferring a far higher degree of
hamlets sited at various "small names" "is of course relevant," but only in a
uniformity than is warranted, and
within them. District boundaries are subsidiary way: "The cultunit is
raising queries about the "rigor" of
territorially defined, and the district offered as a unit of comparative
the definitions involved.
is, broadly speaking, the political unit. statistical analysis of sets of plans-of
To glance at 3 presumably Hopi- Again, apart from linguistic dif- social and cultural patterns as they
type examples: ferences, throughout the whole area exist in the minds of culture bearers."
1. In 1959, my own dissatisfaction there is a more or less common cul- This is all very well if it is seen as
with the use of the term "tribe" in ture. Members of contiguous districts only one kind of comparative ap-
the Western Desert of Aboriginal combine or oppose one another across proach: and certainly the difficulties
Australia led me to look again at linguistic boundaries for fighting, entailed in comparing plans pale
that term, as Naroll has done (Berndt intermarriage, and ritual purposes, and beside the difficulties of comparing
1959:81-107). I suggested that the these shifting spheres of interaction, actual behaviour and systems-in-
units usually called "tribes" in this or potential interaction, are the widest operation. But Elizabeth Colson's
area are in fact dialectal units, ter- social units relevant to these people. warning on this score, published a
ritorially anchored although relatively Clan (village or hamlet) and, except decade ago, is very apposite here
mobile, with flexible territorial and in the south, lineage exogamy, and (Colson 1954, e.g., 44-45, 58). Re-
linguistic boundaries, and sharing an ideal of patrilocal residence, moved from the realm of empirical
what may virtually be called a define the dominant domestic lan- reality, plan-comparison can be-
common culture, with both local guage: but here too a woman speaks come little more than a philo-
authority and links of communication to her young children in her own sophic exercise-or a parlour game,
based very largely on ritual and myth. language-which may or may not be with appropriate (that is, trivial)
Kinship aside, exogamy hinges on the same as her husband's. prizes, if, in fact, "the idea of a
local and totemic, not on linguistic, I would not use the term "tribe" in comparative statistical study, a cross-
affiliations. But the dialectal unit is any of these briefly noted examples, cultural survey, is... to compare cul-
not the widest social unit: members although it is becoming common for ture patterns in order to see if there

292 CURRENT ANTHROPOLOGY

This content downloaded from 62.204.192.85 on Fri, 04 Aug 2017 13:39:37 UTC
All use subject to http://about.jstor.org/terms
are general tendencies governing their Naroll: ON ETHNIC UNIT CLASSIFICATION
construction"
The definition of "cultunits in time" proximate actual ethnic units much of tive units, the omission of the ethnic
is even more arbitrary than that of the time. name as a criterion leads to over
"cultunits in space;" for example, in The concept of the ethnic unit particularization.
conjunction with the cultunit-type implies that some people tend to My reason for adding the ethnic
labels of Hopi, and so on, one might interact more with each other than name as a variable in defining the
suspect an America-centred perspective with others. This statement can be cultunit is not only to bring it into
from the date-line of 1492, that made in spite of the existence of closer proximity with ethnic unity,
distinguishes the aboriginal period marginal individuals and of ethnic but also because I am interested in
from the colonial period. (Wasn't continua. People who interact with making statements about the relation-
this the time when "Columbus sailed each other over long periods of time ship between interethnic or inter-
the ocean blue," so blazing the trail, tend to distinguish themselves from cultunit relations and the internal
so to speak, for the cultunit, or series others by the use of a common name social orders of the cultunits. The
of cultunits, now known as the United as well as by tending to speak the typology of the cultunit does not
States?) I mention this last point, also, same language and to form a common now indicate ethnic consciousness with
in relation to unit comparability; the government. Even when such groups the degree of assurance necessary if
"nest of unit affiliations" referred to are separated from each other and we are going to study problems
as an example of self-definition among exist under different governments, dealing with interethnic relations and
the inhabitants of Unterhauser would they continue for some time to use with the general area of diffusion
seem to have much in common, in this a common name as a symbol of ethnic dynamics and integration.
respect, with the differing cultunits oneness. Contrariwise, groups which Naroll suggests that it is the nature
delineated for Feichten in Kaunertal happen to live together under the of some problems that reference to the
in Oberinntal. In other words, the same rule, and even at times speak cultunit will not be particularly help-
new term which is proposed will, I the same language, may have different ful towards their solution. It might
am sure, involve more or less the samenames for themselves, for instance, be said that my argument for adding
difficulties (in its definition and in its the Chinese Moslems and the Chinese. the ethnic name to the criteria used for
use) as those currently in common use. Because the use of an ethnic name determining cultunits should be dis-
At least in its present form, it does not implies a great deal about how missed in these terms; the cultunit
seem to me to be any more useful for members of a group feel about them- cannot be used to serve all masters.
comparative purposes than those we selves, I suggest that the use of ethnic However, interethnic relations include
already have. names be added to those criteria which a large area of social organization and
define the cultunit. A case in point are so important in influencing culture
By FRANK D. BESSAC* are the Tumet Mongols of Inner that any general ethnic type scheme,
Mongolia. The Tumets of age 50 to 80 if it is to be acceptable for arranging
Appleton, Wisconsin, U.S.A. 14 II 64 material cross-culturally, should give
spoke little Mongolian as children,
nor did their parents speak Mongolian. recognition to this broad range of
The cultunit is not a culture area, i.e., problems (cf. van Nieuwenhuijze
a contiguous area of cultural similarity, The Tumet Banner is controlled by 1963).
nor does it arrange the material so the Chinese, and, because it is south
of the Yellow River, the people who Admittedly, it is not always easy
that it can be studied in terms of a
inhabit it are separated from other to delimit the use of an ethnic name
limited number of related problems,
Mongols. Therefore, according to correctly. However, my experience
like the culture type schemata of
Naroll's system of classification, they has been that this is no more difficult
Steward (1955). On the contrary, than the accurate use of the criteria
culture content is kept at a minimum were not Mongols but Chinese. How-
ever, these people called themselves now proposed to determine cultunits.
in determining the cultunit so that
Mongols, and, believing that lan- The use of an additional criterion
the nature of the construct will not
guage was a means of maintaining creates additional types. The classifica-
by itself give an improper bias in tion I propose is as follows:
cross-cultural sampling. Because of this ethnic unity, they had their children
aspect and due to the nature of the learn Mongolian. Thus, the group Hopi type.-People who belong to
no state but share a common ethnic
criteria that define the cultunits, they became Mongols! If the criterion of the
ethnic name had been included in name, speak a common distinct lan-
appear to be an approximation to
ethnic units-which are sometimes establishing cultunits, these people guage, and are all interconnected by
would have been continuously clas- successive contact links.
called tribes, societies, nations-in
such a form that the ways of life of sified as Mongols, a classification Yuki type.-People who belong to
these peoples can be compared cross- which better approximates the actual no state, have no ethnic name for
ethnic condition. themselves, but share a common
culturally. Naroll is to be congra-
tulated because he goes a long way in The cases of the Tumet Mongols distinct language and are inter-
presenting a good solution to a diffi- and Chinese Moslems illustrate in- connected by successive contact links.
cult problem. stances where the criteria of language, Mongol type.-People who form at
But what is an ethnic unit? I would territorial contiguity, and political least one state, share a common
define it as a group of people who organization put people of different ethnic name, and speak a common
often share a common, self-applied ethnic units into one cultunit to no distinct language.
appellaeion, have a sense of common advantage, as far as I can see. In Kazak type.-People who live in
identity, and share a belief in these instances, the approach results in one or more states in which mutually
a common heritage. Although this over generalization. In South China, unintelligible dialects occur and who
definition might be acceptable, it is where there are many different lan- are domestic speakers of a dialect not
difficult to make it operational. It guages, but where the speakers of intelligible to speakers of the lingua
appears to me that the cultunit, even some of these do not identify them- franca of the state, but who have a
though it does not attempt to define selves primarily by linguis,tic area but common ethnic name.
ethnic groups, is defined by criteria first of all as Chinese and secondly Aztec type.-People who share a
which, in fact, will make it ap- in terms of geographic and administra- common ethnic name and belong to

Vol. 5 * No. 4 * October 1964 293

This content downloaded from 62.204.192.85 on Fri, 04 Aug 2017 13:39:37 UTC
All use subject to http://about.jstor.org/terms
a state in which mutually un- Naroll's comment that "the cultunit in language implies that different
intelligible dialects occur, and who generally resembles fairly closely at cultural units are to be distinguishedby
are domestic speakers of a dialect least some operating social unit within it. But it seems to me that the
intelligible to speakers of the lingua society wherever these units have problem of identifying comparable
franca of the state. sharp boundaries" seems to indicate a ethnic units, each of which bears a
Mohave type.-People who share a more useful direction of inquiry; that whole culture, does not entail distin-
common ethnic name and belong to a is, a precise consideration of the factors guishing different cultures, or cul-
state all of whose members speak a which make it possible to isolate social ture types. Whether a culture differs
common language. systems, systems of social interaction, from, is similar to, or is interrelated
Theoretically, it is possible for a within cultures. Quantitative factors with other cultures is irrelevant to the
state to exist whose people all speak applied to the frequency of relation- problem of identifying the culture-
a common distinct language but which ships promise more effective cut-off bearing unit. For this purpose, the
contains no group possessing a points for cross-cultural survey language spoken by one unit need
common ethnic name. The factors purposes than the criteria Naroll uses. not be distinct from that spoken by
which might give rise to such a Their limitations-that sufficient data another.
condition could exist in the transition are not always available to isolate the For the purpose of identifying units
to or from feudalism. systems significantly-is hardly reason which bear whole cultures, the most
for not using them in future work. useful criterion is political autonomy.
By ELIOT D. CHAPPLE* (This is, in fact, Naroll's principal
Orangeburg, New York, U.S.A. 25 II 64 By GERTRUDE E. DOLE* criterion for classifying units bearing
New York City 13 II 64 cultures on the state level.) If separate
In his concern with developing
local groups participate in the same
sampling designs for cross-cultural I applaud Naroll's attack on the
political organization, we can be
surveys, Naroll has attempted to problem of defining an ethnic unit to
certain that they bear a common
define a homogeneous unit, which he be used in comparative studies of
culture. But if two separate local
labels a "cultunit," his shorthand for cultures.
groups are autonomous, they do not
a culture-bearing unit. His review of If I interpret his paper correctly,
have the same political organization.
the various attempts by others to face Naroll is interested in defining a
The political aspects of their cultures
up to the same problem leads him to unit of people in such a way that each
set up 6 criteria for definition; these may differ from one another signi-
unit will represent a whole and
involve very different orders of ab- ficantly. Because of their separate and
separate culture. For this purpose, the
straction. He regards 3 of the criteria distinct political systems, these two
unit must be defined in terms of a
as common to all preceding definitions: sovereign groups bear not one culture
criterion or multiple criteria which
(1) common language (which he but two cultures.
are functionally related to cultural
defines both phonetically and seman- unity. Moreover, the political group is the
tically with a minimum rule of There is no inherent relation between easiest unit for ethnologists to identify
"recognizable cognates"), (2) territorial cultural unity and the two criteria in the literature, since, by definition,
contiguity (contiguity being in- used by Naroll, common distinct lan- a political system serves to distinguish
troduced, instead of propinquity, to guage and territorial unity. For this an autonomous group from all other
deal with land or water barriers, as in reason, it is impossible to classify all groups. Other aspects of the culture,
Polynesia), and (3) political organiza- societies according to a typology which on the other hand, frequently manifest
tion. Here he is torn between formal is based on linguistic and territorial overlap and, in fact, connect societies
structure and the occurrence of re- unity with the expectation that the with their neighbors in a virtually
cognized leadership hierarchies (which units identified will in each case bear infinite chain through trade, inter-
can be defined operationally as sets a common culture. marriage, ceremonies, and verbal com-
[see Chapple and Coon 1942]), since Virtual linguistic unity mav occur for munication.
all of such structures are dynamic in a considerable time among groups who An instance of autonomous neigh-
the individuals included within them. have separated and are at permanent boring groups which are connected
However, important as is the enmity with one another, as among with one another by complex inter-
problem, Naroll's solution is less than some of the Northern Cayapo groups relationships is the Upper Xingu'
satisfactory. Granted, that he is of eastern Brazil. Some of these groups region of Brazil. Here the local groups
careful to say he is dealing with would constitute a single contact group intermarry, trade, and conduct
"ideal types"-though whether out- in Naroll's terms, but, because of their ceremonies with one another. Some of
right adherence to Platonism is suf- mutual hostility and various cultural these groups speak interintelligible
ficient excuse seems arguable. He who differences, they can hardly be con- dialects and would therefore consti-
wishes to compare cultures as entities sidered as forming an ethnic unit with tute an ethnic unit on the basis of
has the responsibility of coming up a common culture. Moreover, some linguistic similarity and territorial
with unequivocal operations which groups such as the Nyul-Nyul and unity. But others speak entirely dif-
do not involve a penumbra of un- Bard of northwestern Australia speak ferent languages and would thus form
certainty (and not in Heisenberg's an interintelligible dialect and con- separate ethnic units on this basis.
sense either). stitute a contact group but, never- Yet, in spite of linguistic differences,
Further, we must realize that in theless, have different kinship systems all of these groups resemble one
statistical theory we have to concern as well as other cultural differences. another equally in most other aspects
ourselves with homogeneity. Naroll's Hence, these groups would not of culture. However, all have some
statement that each sampling unit has together constitute a single cultural cultural differences, such as economic
a mathematically definable chance of unit. specialties, which suggest that they
being selected is true, but not suf- As Naroll has pointed out, the should all be considered as separate
ficient; he must be sure that his difficulties of dividing peoples into culture-bearing units. The difficulty of
sampling constructs do not introduce units with common distinct languages determining a consistent classification
heterogeneity which he plasters over are formidable. It would facilitate of these groups is easily solved by
with operationally non-comparable classification if this step could be separating them along the lines of
criteria. avoided. The criterion of distinctness political autonomy.

294 CURRENT ANTHROPOLOGY

This content downloaded from 62.204.192.85 on Fri, 04 Aug 2017 13:39:37 UTC
All use subject to http://about.jstor.org/terms
By HAROLD E. DRIVER* Naroll: ON ETHNIC UNIT CLASSIFICATION

Bloomington, Indiana, U.S.A. 7 II 64


minimum number of units necessary to list of 200 are cognates is certainly not
display all the variation within the generally true. The languages cited by
My views on ethnic unit classification
data, referred principally to these Lees (1953) average about 80% of
will be made clearer if I first describe
chains and meshes of language dis- cognates shared after 1,000 years of
what I have done with North
tribution in the Arctic, Sub-Arctic, separation, and it is doubtful if any of
American data. In 1948, I began to
and Great Basin areas. I had no the pairs of languages separated by
collect data covering the entire North
problem in other areas where the lan- this interval of time are mutually
American continent with a new
guage units matched those of the intelligible.
technique. Instead of using cards or
ethnographic literature quite closely. Naroll's classification-of ethnic units
paper slips to excerpt information
In short, I used the language unilts as into Hopi, Flathead, Aztec, and Ta-
from sources, I entered the data direct-
ethnic units over most of the continent, rascan types introduces some order
ly on maps in parallel hatching of
but subdivided these in the 3 areas into the present chaos. I mildly object
various colors so that multiple subject
mentioned above. At this date I would to his using the names of localized
units could be superimposed on the
territory of a single ethnic unit. None
add that differences in biological type, peoples for ethnic units of world-wide
as in Africa, are an additional basis on scope. I would prefer descriptive catch
of the Indian "tribal" maps available
at the time had enough names of tribes,
which to separate ethnic units. phrases such as stateless-language type
In world-wide cross-cultural studies for Hopi type, state-language type for
villages, and bands on them to take
care of most of the names of ethnic by Naroll and the Yale-Harvard- Flatnead type, state-lingua-franca type
units which kept cropping up in the Pittsburgh schools, the ethnic unit for Aztec type, and state-multiple-
ethnographic literature, and none of becomes more critical because the langzages type for Tarascan type.
the maps could be purchased in out- samples include a smaller percentage The notion shared by many world-
line form by the 100. Therefore, I of the total number of cases than my wide cross-cultural researchers, that
solicited help from 5 ethnologists and North American Indian work and they can eliminate diffusion and
drafted a new map (Driver et al. also because geographical continuity genetic heritage factors by choosing
1953), and made outline copies of it is broken. The world-wide cross-cul- their samples so that each ethnic unit
available in any quantity needed. This ture researcher has no opportunity to falls in a different culture area and a
map employed 3 kinds of lines: a solid draw historical or evolutionary in- different language family, is wishful
line for genetic language family ferences from relatively complete thinking. Almost all culture areas have
boundaries; a broken line for geographical distributions, because he been determined impressionistically;
boundaries between mutually unin- often deliberately selects his ethnic rarely has an explicit quantitative
telligible languages belonging to the units to avoid geographical continuity method been employed. Furthermore,
same genetic language family; and a- and language family co-membership. culture areas are of multiple
dotted line to separate mutually in- Having thus eliminated the most sizes or orders: Wissler (1917) and
telligible dialects of the same language obvious aids to interpretation of his Kroeber (1923) both cut up the
or geographically distinct ethnic units data, he must rely more heavily on the Americas into 15 or more culture area
with no dialectic differences. The lan- mere statistical results to obtain the for descriptive purposes and inter-
guage families were the ones pre- causal and evolutionary sequences he pretations of localized phenomena, yet
viously listed and described by Voe- is trying to establish. In my work, the they treated the two American con-
gelin and Voegelin (1941), and Hoijer statistics have been ancillary to the tinents as a single culture area, center-
(1946). In the jargon of glotto- conventional distribution methods of ing in Middle America, for their more
chronology, they are genetic families comparative ethnology, as Kroeber speculative theories of origin and
with time depths up to about 5,000 and I said (Driver and Kroeber 1932: dispersal of widespread culture
years. Naroll seems to have missed the255). Other limitations of world-wide elements and complexes. These men
significance of these 3 kinds of lines. cross-cultural method have already also slit their 15 or more areas into
The above key map was used in been discussed rather fully by Kobben smaller subdivisions. Although culture
two later publications in which (1961) and need not be gone into areas are supposed to package geo-
considerable numbers of page-size further here. graphically and historically connected
maps displayed the geographical Naroll's discussion of ethnic unit traits together, an alarming number
distributions over the North American classification is definitely the fullest of items spill over the boundaries of
continent of a total of about 500 and best so far, but his proposed any and every culture area scheme, as
subject units (Driver and Massey 1957; solution of the problem seems to me the many maps in Driver and Massey
Driver 1961). A casual glance at these less happy than his assemblage of the (1957) show beyond all doubt. There-
maps reveals the fact that the shades of issues involved. For instance, the term fore, the selection of only one ethnic
gray employed are, in almost all in- "cultunit" seems awkward. It doesn't unit from each culture area and each
stances, coterminous with the linguistictell one whether it refers to an ethnic language family can lessen the number
boundaries. However, in the Arctic, unit (village, band, tribe, state) or to of historical connections in a corpus
Sub-Arctic, and Great Basin areas, the a subject unit (culture element, trait, of data, but it cannot eliminate them
large numbers of dotted lines indicate complex, assemblage, theme, pattern); and tell where they occur and where
a number of chains or meshes of and a novice might even think it has they do not.
mutual intelligibility (Swadesh 1959). something to do with a religious cult. In spite of such criticisms of world-
This means that neighboring forms of Territorial contiguity can scarcely be wide cross-cultural studies, I believe
speech so separated are mutually in- demanded of all ethnic units, because that the majority of the generalizations
telligible, but that forms of speech migratory peoples are sometimes split they assert, often from small samples,
separated by considerable distances into two or more territories, and will be confirmed later by2 larger
within such chains or meshes may not modern nations, such as Pakistan, samples and more detailed analyses.
be mutually intelligible. In my study sometimes occupy two separated Naroll's practical suggestions in the
referred to by Naroll (Driver 1956: regions. The notion that two languages last section of his paper are good ones
15), my remarks about dividing ex- are mutually intelligible when 80% and will improve cross-cultural
tensive territorial groups into the of the words in the glottochronology research. To these I would add only a

Vol. 5 * No. 4 * October 1964 295

This content downloaded from 62.204.192.85 on Fri, 04 Aug 2017 13:39:37 UTC
All use subject to http://about.jstor.org/terms
plea for more regional comparative defining the unit, the "type" should specified set of criteria).
studies in which all, or almost all, encompass all levels of integration. Let us consider the Nootka and
available ethnic units are utilized. Where would Naroll place the Swiss, Kwakiutl. They are located next to
Austrians, or Americans, either as each other and have many cultural
societies or in terms of component similarities, presumably because of ex-
By PAUL DUCEY* ethnic units? tensive borrowing and because of
Cairo, Egypt, U.A.R. 25 I 64
Finally, I am in complete disagree- parallel adaptations to similar en-
ment with his idea of "the cultunit vironments. They would still be
in time." There are several major separate and independent cases if we
No one would deny that the problem
questions here. Why do we need a chose to define a sample unit as a
which Naroll has set for himself is
distinction between "palaeoethnogra- continuously distributed population
an important one. Clarification of such
phy" and the "aboriginal period?" whose language or lingua franca is
basic concepts is long overdue. Un-
Were the "palaeoethnographers" less different from the dominant languages
fortunately, however, his discussion of
observant than Columbus and the of neighboring populations (cf. Ember
the problem confuses me.
explorers who followed him? Why 1963) and if they both had been
There are two implicit assumptions
advocate dropping the Christian era selected randomly from some list of
which are particularly disturbing. One
in favor of a pre-Christian time scale societies.
is the correlation of "tribe" and
which assumes the Christian era and The supposedly thorny problem of
"society." Needless to say, tribes may
arbitrarily sets the beginning of the the comparability of units cannot be
be societies, but societies are not
time scale at 10,000 B.c.? Such a resolved by any statistical maneuvers,
necessarily tribes. The other is his
system would, of course, encompass such as the method of inspecting
equating the peasants of Tyrol or the
"dates of all written records," but so proportions of "hits" that Naroll
people of the United States with the
would any date beyond 4,000 B.C., or suggests. Nothing will convince his
cultunit, as defined and, especially, as
perhaps, 3,500 B.C. Archaeologists have disdainful dinner companion that the
presented in terms of 4 types of cul-
considered the problem of time scales, Yahgan and the English are "com-
tunits.
but only two alternatives have parable," no matter how many tables
Turning to Naroll's explicit position,
attracted attention. One may use years or numbers are presented, and -no
a number of points are disturbing. To
before present or a B.C./A.D. date. It attempt should be made to do so. In
begin with, there is the proposed term
really does not matter. Scholars can cross-cultural research, we are not
"cultunit." Why should we continue
interpret either in terms of the tilme interested in global comparisons of
to coin new terms when established
scale understood in their respective cultures, as was the old "comparative
vocabulary would serve our purposes?
ethnic unit orientation. method." Rather we are concerned
Ethnic unit, it seems to me, would be
with examining relationships between
equally appropriate and avoids
specific variables-e.g., degree of
possible religious associations. Second-
By MELVIN EMBER* economic and degree of political
ly, there is the use of the terms
development-in a sample of societies.
"tribe," "society," and "state." After Yellow Springs, Ohio, U.S.A. 17 II 64
The definition of a sample of societies.
presenting and analyzing a sampling
The definition of a sample unit
of definitions of tribe and briefly com- Since Naroll's paper is a thoughtful
depends solely upon the purpose of the
menting on the concept of society, contribution to the methodology of
study. To put the issue simply, if our
Naroll proceeds to obscure the concept cross-cultural research, and since it
intent is to examine the relationship
of the "state" by equating it with may therefore be consulted by those
between volume and weight in "fruit,"
"tribe." Certainly, if the "state" is to who are interested in conducting such
it does not matter one whit that we
be meaningful, there must be a quali- research, a few errors which appear
are dealing with "apples" and "pears"
tative distinction between tribe and in it should be corrected at once.
or even "watermelons."
state. Moreover, there must be At the beginning of the paper, Naroll
distinctions between tribe, nation, con- observes that cross-cultural studies
federation, state, and empire. heretofore have never rigorously
By HELMUTH FUCHS*
A third problem is presented by the defined the units being considered. At
emphasis on "war-making authority" the risk of appearing immodest, I Caracas, Venezuela 17 II 64
in both the analysis of "tribe" and must point out that the study of mine
"cultunit." Unless one uses a very to which he refers is an exception The criteria Naroll furnishes for the
artificial definition of "war-making (Ember 1963:234-236). definition and delimitation of the
authority," few ethnic units are Naroll says it is fallacious to count minimum culture unit (as opposed to
characterized by having leaders who neighboring and closely related cul- ethnic unit) for classificatory purposes
can "assert and wield the exclusive tunits as independent cases. This is not are, if they are necessary at all, only
right to declare and conduct warfare" necessarily so. The independence of some of those already considered in a
or "lethal group combat." As Hsu any two given cases has nothing to docentury of anthropological literature.
reminded Naroll, religion is an im- with how geographically close they There is a considerable number of
portant criterion. I would go further are to each other or how similar they publications whose titles indicate,
and say it far outweighs the ability are culturally. It is strictly a statistical implicitly or explicitly, that -their
or right to wage war. Undoubtedly, matter. The independence of cases is authors were aware of the dangers of
one of the major trends in the cultural guaranteed by one condition, namely, false generalization and abstraction
development of man has been the that every sample case has had the in this respect.
increasing restriction of the right of same probability of being included in It is difficult in a few words to
ethnic units to wage war. the sample (which means that the summarize the manifold factors in the
The fourth thing that bothers me is selections have been made according to problem of establishing units in
the distinction of the four types of some kind of random procedure). anthropology, and especially in cul-
cultunits. Subsuming the Aztec, Inca, Naturally, every sample case must tural anthropology. However, some
and Zulu under one heading cannot be also be a separate case (which means principles should be stated and some
justified. Moreover, in so far as cross- that every sample case must be defined questions should be asked. For
cultural surveys are the basis for as different according to some example, is an ethnic unit (if it exist

296 CURRENT ANTHROPOLOGY

This content downloaded from 62.204.192.85 on Fri, 04 Aug 2017 13:39:37 UTC
All use subject to http://about.jstor.org/terms
per se as a culture bearing entity) a Naroll: ON ETHNIC UNIT CLASSIFICATION
simple or complex phenomenon?
Further, what is its grade of integra- 1949), we might consider for our pur- state a partition rule (which he does),
tion, and is it mono- or polydi- pose the numerical value of culture- but he must also indicate why this rule
mensional? Second, is any category- bearers (F) of a culture and the range results in "comparable" sets (which
defined by a member of the culture (R) of a culture within a given he does not do). This is a critical
or by an outsider-of a supposed cul- sequence of cultures. The two extremes, defect.
ture entity or a part of one (a) (1) all the inhabitants of the world Many of the potential solutions to
congruent with regard to quality having the same culture, and (2) each his problem are meaningless in
(specification, function, structure), inhabitant of the world having his anthropological investigations. For
quantity (numerical value of in- own culture, are either utopic or else example, Naroll points out that the
cidents), genesis and development, possible only in the case of single number of people in each "tribe" (i.e.,
chronological position and duration, survivors. The bulk of the data may the number of elements in each set) is
and spatial extension, with any other behave in accordance with Zipf's law irrelevant in a comparative study of
category of the same or another of equilibrium:
their culture. Hence a definition of
culture entity or a part of one? And "comparable" sets that involves their
(b) does such a defined category log F = p - log R
cardinality is useless. Further, the
coincide with a likewise self- or other- "tribes" of the world are made up of
defined ethnic unit or a part of one? in good approximation of a hyperbola.
distinct individuals, i.e., they are
Third, is this problem restricted to It is also likely that, as in the case of
disjoint sets (with rare exceptions such
ethnic units and cultural units, or do language, the formation of a "new"
as the person cited by Naroll who
similar problems arise in other anthro- culture is easier, the smaller the group
functioned both as a Shan and a
pological fields, with regard to (minimum value of F). This scheme
Kachin). Hence it is futile to define
archaeological units, linguistic units, effectively illustrates the relative in-
"comparability" in terms of equality
sociological units, educational units, adequacy of classificatory systems, of sets.
religious units, biological units, and so insofar as it does not matter whether
Since these formal properties of sets
on? Fourth, can the result of such an or not the unit (R) is strictly definable
are not very relevant in an anthro-
over-all cross-study, undertaken a or not: the scheme is built upon the
pological context, we move on to
priori or a posteriori, be classified at relation between two more or less
consider the formal properties of the
all? And if so, is classification a arbitrarily chosen units. We have
partitioning rule. Such a rule permits
sufficient statistical method to serve ignored the question of whether cul-
one to decide whether any given pair
anthropology as a tool and not a toy? tural data really do behave in ac-
of people belong in the same set or not.
cordance with Zipf's law, but it is
Instead of searching for a general For example, consider the following
highly probable that they do. If not,
"skin" of culture or for specific skins rule (a): If x has the same sex as y,
then we may continue packaging
of culture categories, and defining they go into the same set; if not, they
cultural data into classificatory
units for classificatory purposes, we go into different sets. This rule will
systems, and then, certainly, cultunits
must first understand the usefulness of partition a universe of people into two
will be of great interest to comparative
classificatory systems and statistics in sets: male, female. A second rule (b)
anthropology, but they will be subject
comparative anthropology. Possibly involving marital status would parti-
to revision of a statistical, rather than
"The Use of Computers in Anthro- tion the universe into three sets: un-
an anthropological, order.
pology" will throw new light on this married people; married people;
problem. widows and widowers. A conjunction
Meanwhile, the question arises By HANS HOFFMANN* of these two rules (a and b) would
whether the present emphasis on partition the universe into six sets:
sampling cultural data for a kind of Clinton, New York, U.S.A. 17 II 64 bachelors, husbands, widowers,
IBM evaluation will be adequate to spinsters, wives, widows. Formally, a
arrive at the results comparative an- A paper that offers a new funda- conjunction may be interpreted as a
thropology expects. At present, we mental unit to a science immediately single rule.
package material into classificatory raises two questions: What is the ra- At this point we (somewhat ar-
systems which later prove too large or tionale for this particular unit? How bitrarily) define "comparable sets" to
too small. Anthropologists react to is it superior to existing formulations? be those resulting from a single parti-
these defects by changing the Naroll does not discuss the second tion rule. Since all six of the above
package-by constructing "new" questions. He documents the absence sets were defined by a single rule,
systems of classification. This fact of concensus among anthropologists we consider them to be comparable to
probably demonstrates the relative about the basic unit in their investiga-each other. We may compare this rule
unfitness of such classificatory systems
tions. He does not, however, explicitly with another that includes a dis-
for generalization in anthropology. relate his "cultunit" to the problems junction of partitioning criteria. An
Language and political structure are raised by this discussion. Nor does he example would be: x is a member of
the main criteria in Naroll's "Classi- discuss its advantages over such units the same set as y if x and y practice
fication of the Cultunit"-in spite of as those of the Human Relations the same type of economy and live in
the fact that nine criteria (which were Area Files. Thus the potential impact settlements that are either patrilineal
by no means sufficient) were presented of his cultunit remains obscure. or socially stratified. This will parti-
in the foregoing "Definition of the Naroll delimits the first question by tion the universe into many sets. The

His problem is to define the word


Cultunit." Zipf's law of balance should stating that the cultunit (or "tribe") question is, are these sets "comparable"
be remembered. This law shows a is offered primarily to solve sampling in the sense defined earlier; i.e., are
mathematical behavior regarding the problems in cross-cultural research. they the result of a single partition
self-structuring of language. I would His objective is to partition the rule?
not be surprised if a similar law of ethnographic universe into "tribes" To answer this question we have to
balance (equilibrium) could not be (sets that are comparable to each other.look at the form of this rule, which is
shown to apply to the self-structuring (a and [b or c]). It can easily be
of culture. Following Zipf (1935; 1946; "comparable." Not only must Naroll proved that the statement (a and [b or

Vol. 5 - No. 4 * October 1964 297

This content downloaded from 62.204.192.85 on Fri, 04 Aug 2017 13:39:37 UTC
All use subject to http://about.jstor.org/terms
c] implies the statement ([a and b] which corresponds to Naroll's, ter- By DAVID LANDY*
or [a and c]). In other words, some ritorial contiguity in the sense- of "a
Pittsburgh, Pennsylvania, U.S.A. 21 11 64
of the sets resulted from applying the geographically contiguous territory"
rule (a and b), i.e., same economy is usual but not universal. Thus the
The problem to which Naroll ad-
and patrilineal descent, while the otherI1/2 million Javanese in Bantam, West
dresses his paper is one that seems
sets resulted from the rule (a and c); Java, are divided by about 300 kilo-
limited to the social sciences, since zoo-
i.e., same economy and degree of social metres of Sundanese country (thus ex-
logists studying forms of life other
stratification. Hence, this partitioningceeding Naroll's limit of 200 kilo-
than human seem generally content,
of the universe resulted from the ap- metres) from the main body of Ja-
for comparative purposes, to use such
plication of two different rules and, vanese with whom they nonetheless
concepts as "aggregation," "colony,"
therefore, does not consist of com- share the remaining 4 criteria in my
"crowd," "population," and even
parable sets. (There are further list). In Southwest Sumatra, the high-
"society," each of which, in com-
problems here; i.e., are these sets land Lembak are separated from the
parative terms, becomes almost
disjoint? but that is another story.) Lowland Lembak by the Redjang and
coterminous with species (e.g., Allee
We may conclude that a partition rule the Barisan Mountains but still con-
1958). In zoology generally, the limits
containing a disjunction will not, in stitute a single ethnic unit. Similarly,
or boundaries of species remain un-
general, yield comparable sets. the Redjang Abeus of the Trans-
solved and vexing, and, as in the case
We are now in a position to return Barisan plains are separated by the
of defining society and culture in
to Naroll's partition rule: x is in the Malays of Rawas from the main body
anthropology, the problem seems to
same set as y if x and y are speakers of Redjang in the Barison highlands,
increase in complexity with the
of the same language and belong either but the two still constitute one ethnic
growth of knowledge. Kroeber and
to the same state or the same contact unit. In my view, absence or partial
Kluckhohn's (1963) brilliant tour de
group. invalidity of any one of the 5 criteria
force performed a useful service in
To abstract the structure of a does not invalidate the other 4.
examining hundreds of definitions of
sentence from the morass of its actual With regard to the time factor, I culture, but in the end they seemed
and implied meanings requires a clear suggest the addition of a fourth no nearer a universally operational
head and high courage. Hence it is period, that of post-colonial or in- definition than their predecessors.
with some diffidence that I interpret dependent, unless Naroll were to insist It is just possible that these kinds of
Naroll's rule as a disjunction. On the that the new independent states of taxonomic dilemmas are in their
other hand, I suspect that a modest Africa and Asia stand in much the nature insoluble and the concepts
amount of tinkering may be sufficient same relation to their constituent involved must remain fluid and
to convert it to a conjunction. ethnic units as did the previous unbound. Admittedly, if one wishes
colonial governments. to compare "culture-bearing units"
I agree with Naroll that the term statistically (or any elements of them),
By MERVYN JASPAN*
"ethnic unit" is unsatisfactory but do one has to define the unit. The animal
Canberra, A.C.T., Australia 13 II 64 not consider his neologism "cultunit" biologists may at least attach limits to
any improvement. In laying stress on their groupings in terms of fairly
Naroll's paper summarises the many culture rather than social structure specific gross anatomical structures,
imprecise and dissimilar uses that are and territorial contiguity, I find this except, perhaps in the instance of
made of the terms "society," "tribe," innovation no better than "ethnic microscopic life. But the anthropolo-
"culture," and "people" to denote an unit." On the contrary, it introduces gist is dealing with populations which
ethnic unit. After discussing several new facets of imprecision and possible must be defined not in terms of
definitions of ethnic units, he discerns ambiguity. Naroll has himself in- physical forms but cultural forms. It is
at least 6 (no more than 6 are sub- dicated this in remarking that "field almost as if the zoologist were called
sequently considered) criteria used by description units will often not upon to classify his groupings in terms
anthropologists in their definitions. To correspond to the cultunit." It is by of functions-what the animals do-
my mind the first of these ("distribu- no means certain that cultures have instead of forms or structures.
tion of particular traits being studied") clearly discernible boundaries, and Assuming that the problem is "signi-
is meaningless in such a definition un- least of all in a spatial sense. Until ficant," there is no question of the
less we are told what the comparison ethnographers can agree on the un- highly stimulating manner in which
is for. The sixth criterion ("local mistakable boundaries or "skins" of Naroll has approached it. However,
community structure") might well be cultures, attempted digital counts and he has not, in my opinion, added
linked with the third ("political or- classifications are unlikely to serve materially to its solution. His concept
ganization") in a single criterion of any useful purpose and may, on the of the "cultunit" is another linguistic
social structure. The fifth ("ecological contrary, prove misleading. Never- monstrosity that does not clarify the
adjustment") is really an extraneous theless, Naroll insists correctly on a notion of society or ethnic unit; it
variable as irrelevant to the definition tightening up of the definition of appears at first glance to have been
as marital adjustment. In my defini- ethnic unit. emitted by a computer that had been
tion (Jaspan 1957), I suggested the The errors of earlier ethnographers fed a question having to do with
following criteria: who mistook for ethnic units such certain socio-religious entities. Naroll
moieties as Budi-Tjaniago and Koto- claims that his "new unit" is "propos-
1) Territorial contiguity or proximity
Piliang among the Minangkabau, and ed primarily as a conceptual norm
2) Language
social strata such as the Pubian and to aid anthropologists in their
3) Social structure Pepadon among the Lampong, have thinking, rather than an actual unit
4) Important elements of common folklore made the systematic classification of of study either in field work or in
and history
Indonesian ethnic units an unenviable comparative studies." Now all con-
5) Consciousness of a common identity, task (Jaspan 1959). Much avoidable cepts, particularly concepts of units or
however minimal or incipient (This is
confusion has arisen from the use of entities, are ideal types, but their
a cumulative product of the foregoing
such vague combinations as "peoples, usefulness would be limited indeed if
criteria.)
tribal groups, and tribes" (Kennedy the scientist did not attempt to use
With regard to the first of these, 1945). them as more than simple guides; after

298 CURRENT ANTHROPOLOGY

This content downloaded from 62.204.192.85 on Fri, 04 Aug 2017 13:39:37 UTC
All use subject to http://about.jstor.org/terms
all, Naroll assumes that his concept Naroll: ON ETHNIC UNIT CLASSIFICATION
does have a taxonomic function, and
in fact he addresses an "appeal to Incidentally, where, I wonder, will is the major economic activity; in
anthropologists in the field to identify "the Chinese" be allocated in the four- 1940, most Kachin communities were
the cultunit or cultunits involved in fold cultunit schema (page 2E)? China small hamlets of three or four houses,
their study by collecting the relevant is a single state, but there are millions but a small minority were compact
linguistic, political, and geographical of persons of "Chinese culture" who villages of over 100 houses; most
data." Nevertheless, subsequently he live outside that state. Persons of houses were occupied by "an entire
says, "Field description units often will "Chinese culture" speak many mutual- lineal family," but some houses
not correspond to the cultunit." if ly unintelligible languages, but contained 50 or more people, and in
they do not, then one must question educated Chinese can write one the days when (f) was true, the ma-
even the normative function of such another intelligible letters even when jority of the population lived in such
a concept. Attempts to use the concept they cannot converse in mutually in- "long houses;" while most couples
will be even more confounded by telligible speech. But this, we shall be reside patrilocally, the political struc-
Naroll's assertion that "the cultunit told, is a very special case, and cul- ture is entirely dependent on the fact
sometimes may be a genuine social tunits are only intended as "ideal that a significant proportion of house-
types." Alas, I fear that if anyone holds reside uxorilocally; statistically
system but often will be only
tries to use this new concept to sort out speaking, marriage with a matrilateral
theoretical-an analytical abstrac-
practical problems, nearly every in- cross-cousin is rare, though marriage
tion." No matter at what level of ab-
stance will turn out to be a very with a kinswoman of the category to
straction a concept is formulated, and
special case! which a matrilateral cross-cousin
no matter to what degree the data in
My disagreement with Naroll is belongs is obligatory. It is not that
specific cases deviate from the con-
quite basic. Murdock's World Murdock's tabulators misread their
ceptual norm, the user of the concept
Ethnographic Sample and all work sources; it is simply that the ethnog-
should be able to identify the data as
that derives from it rest on the raphic facts of the case will not fit
members of the class or classes sub-
fundamental assumption that the units tidily into tabulated categories. I
sumed by the concept. Various
of discourse, call them "tribes," "cul- believe that this is true of all human
organs and organ systems, for example, tures," "cultunits," or whatever, are societies.
are never identical in any two human "species objects" which can be I have written in the first person,
bodies (Williams 1958), but, no matter described taxonomically by a list of but I would suppose that most British
what the degree of variance, the characteristics. Just as a species of social anthropologists would share my
physiologist or anatomist should have beetle may be described. This is a views on this particular issue.
no difficulty in recognizing a kidney, proposition which I simply do not
stomach, or spleen. accept. If I write a monograph about
In another commentary on Naroll's the Kachin (as I have done) and By OTTO VON MERING*
paper, Murdock has indicated why his Murdock chooses (as he has done) to Pittsburgh, Pennsylvania, U.S.A. 20 II 64
classification now in progress of have that book treated as if it were a
cultural types (Ethnographic Atlas taxonomic description of a particular Naroll wishes to help his fellow
1963) is preferable for providing an cultural unit (Murdock 1957:680), he surveyors of the ways of man by
appropriate universe from which to is acting within his rights, but from presenting them with the new con-
draw samples for cross-cultural studies. my point of view he is producing cept, "cultunit." Ail those who wish
I am in agreement with the notion of tabulated nonsense. Naroll tells us that to accept his offering are invited to
"cultural types" and in general the the object of a comparative statistical choose whether they will regard it as
criteria used by Murdock for orderingstudy, a cross cultural survey, is "to an "ideal type," or a "conceptual
them. Perhaps we should distinguish compare culture patterns in order to norm," or a "unit of comparative
in the future between cross-cultural see if there are general tendencies statistical analysis of a set of plans...
and cross-societal studies. For the governing their construction." This ob- as they exist in the minds of culture
former, we should be interested not sojective presupposes that "culture bearers," or a "standard of reference ...
much in the "culture-bearing unit," patterns" are discrete entities composedto aid... in their thinking."
as in representatives of known cultural of further discrete entities which are The infancy of any science is said to
types. Cultural types would seem to capable of being tabulated by code be characterized by a search for
have some finite universe, whereas letters or numbers. I do not accept taxonomy, and this applies to anthro-
culture-bearing units will vary in either of these suppositions, and it pology. However, the pressing issue is
number depending upon the selection seems to me that tabulations which that, unless you have a sense of
of classificatory criteria and the are made on this basis are bound to be problem, your taxonomy will do-
purpose for which they are postulated. grossly misleading no matter how the minate your method, rather than help
units are defined. Let us take the you locate the real data.
Kachin as a case in point. Murdock In the course of Naroll's paper, the
By EDMUND LEACH*
(1957:680) tells us that (a) fishing is reader is treated to yet another
Cambridge, England 15 I 64 "absent, insignificant, or sporadic;" (b) taxonomy of culture. In this case, it is
Kachins live in compact villages or a fourfold one which even includes
The first part of Professor Naroll's towns; (c) there is "normal occupancy a Flathead cultunit. For good measure,
paper states the problem very clearly, of a single dwelling by an entire lineal this is coupled with an "appeal to
but I entirely fail to understand how (small extended) family;" (d) residence anthropologists in the field to identify
the cultunit device will solve anything. is "normally" patrilocal, with no the cultunit ... by collecting relevant
I have previously argued that any "patterned alternatives;" (e) matri- linguistic, political, and geographical
system of cross-cultural comparison lateral cross-cousin marriage is "pre- data." Later in the text, however, the
which can be so used as to make "the ferred;" (f) the system includes reader comes upon the demurrer that
Tikopia" and "the Chinese" units of hereditary aristocrats and hereditary the cultunit is not really meant to be
comparable type is self-evidently slaves. Are these statements true or "an actual unit of study, either in
absurd, and I do not see that the false? I suppose it all depends upon fieldwork or in comparative studies."
absurdity is reduced by calling the what you mean by "normality." I "The operational difficulties of
units "cultunits" instead of "tribes." know of Kachin villages where fishing applying the linguistic criterion...

Vol. 5 - No. 4 * October 1964 299

This content downloaded from 62.204.192.85 on Fri, 04 Aug 2017 13:39:37 UTC
All use subject to http://about.jstor.org/terms
[to] whole societies or other units of diverse cultural solutions to a selected are tested in chemical laboratories.
comparison" are discussed in extenso. problem has always been the prologue Now that anthropology has reached
Despite his caveats, Naroll appears to new and original field investigation. the second stage, applied anthropology
willing to settle for his linguistic At this stage of the development of can supply the laboratory for this
boundary, "a group of people who are anthropology as a science, it is doubt- third stage, by testing factors of
domestic speakers of mutually in- ful whether its practitioners would be change held constant in one region for
telligible dialects," as the construct the richer for conducting the proper a limited period, against other factors
which is capable of forming a cultural study of man on the basis of or change which are permitted to
boundary. the cultunit or a similar taxonomic operate in an equivalent region.
At another point, we are told that contrivance. Naroll has, however, Here the "cultunit" concept could
since "the cultunit concept is offered performed the service of directing the be helpful, if qualified demographical-
as a tool for comparative statistical attention of his colleagues to a most ly; for instance:
studies" of sets of plans, it has been perplexing theoretical and methodolo- Hamletcultunit.-limited toprimary
made "as nearly uniform as possible." gical issue that deserves all the interest, group up to 100 persons.
Next, we are assured that it "generally talent, and experience that anthro- Village or camp cultunit.-100 to
resembles fairly closely at least some pologists can bring to bear on it. 2,500 persons.
operating social unit within the Town area cultunit.-2,500 to
society," and is therefore capable of 10,000 persons who are stable residents
By SIMON D. MESSING* and incorporated.
solving sampling problems. Neither
these assertions, nor the claim that "it Tampa, Florida, U.S.A. 30 I 64 City-state or provincial cultunit.
is irrelevant how many people follow linguistic or territorial unit whose
a plan-whether a few hundred or There is indeed need for definition ofmembership exceeds that of the cul-
many million," will cause problems of statistically usable units of culture- tunit.

complexity and organization to dis- bearing "wholes." But Naroll's pro-


appear. posed "cultunit" lacks the demographic By FRANK W. MOORE*
I am puzzled by Naroll's divergent dimension. How can it be irrelevant
assertions and pleas in his approach whether "a few hundred or many New Haven, Connecticut, U.S.A. 8 II 64
to the difficult problem of finding a millions follow 'the plan'"? The cul-
meaningful and useful unit of culture. tunit would then be limited to the Naroll has presented a proposed solu-
His proposals are both exceedingly humanities, to poetry and philosophy, tion to one of the most pressing and
elaborate and disconcerting. and not usable in the construction of difficult problems facing not only
It is possible, that, without realizing statistical sampling procedures. Yet cross-cultural studies but the entire
it, Naroll is actually advocating the quantitative surpluses over subsistence field of cultural anthropology. What-
simple biological or atomistic family, needs have long been shown to be theever the merits or failures of his
or some other man-made social institu- basis for qualitative cultural develop- scheme, merely to call attention to this
tion, as the ideal or basic sampling ment (specialists, multiple institutions, problem is a worthwhile effort.
unit in cross-cultural study. To be etc.). In recent years, populatioln ex- As Naroll points out, ethnographers
sure, this has the value of simplicity. plosion has placed considerable strain have been remiss in not making a
Though it may be as troublesome as on certain exotic institutions, and cul- serious effort to place their units of
the cultunit to apply faithfully in tural particularism may be affected study into an overall context.
practice, it would at least remove the by the size of the group. Scientists in other fields who failed as
obligation of classifying the Swiss as Suppose one is faced with the consistently as ethnographers to
"a state made up entirely of Tarascan problem of designing accurate and identify satisfactorily their units of
type cultunits." precise sampling of attitudes relating study would scarcely be allowed to
It is reassuring that most anthro- to health and sickness in an ethnically publish results. The problem, of course,
pologists still use as their unit of and ecologically diverse country of 20 is somewhat beyond the scope of the
definition and discourse "the socio- million people. Suppose further that individual ethnographer. The entire
cultural unit recognized by the natives time and various logistics limit the field has been at fault for not sup-
being studied." Restated, they study research to four pairs of "study" and porting the effort to develop and use
people who do the calling, rather than "control" communities, plus pretest. an effective and flexible nomencla-
people who have been called some- The practical problem is then to ture.
thing by others. Although Naroll avers choose among the major cultunits in Naroll's survey of previous work ont
that "a number of inconveniences order to give meaningful representa- the subject indicates that language
would result from this," he appears tion to those groups and regions that mutual intelligibility is the most
quite willing to live with them in his are most significant in both numbers widely accepted single criterion for
study of the Austrian peasants of Un- and cultural considerations. isolating cultural units. Despite
terhauser. Indeed, he does not even Anthropological enterprise has Naroll's recognition of some of the
mention the possibility that the largely shifted from description of problems in applying the language
villagers might constitute a Tarascan "unique," small, isolated groups, to criteria, he proceeds to use it as the
type cultunit. analytical, problem-oriented research basic trait in distinguishing his four
Clearly, for a great many problems useful for cross-cultural comparison. types of culture units. Aside from
and ways of living that the anthro- Perhaps anthropologists can learn the operational objections Naroll
pologist studies, neither cross-cultural from the sequence of events in medical 1nentions, there is the much more
surveys nor the cultunit will be the research. When a new disease is en- serious objection of the lack of identity
answer. The comparability of the countered, physicians gather case between language and culture areas.
chosen units of comparison, and the histories and write wordy, descriptive Territorial contiguity is also a
validity of the results will always reports. After a considerable amount somewhat tricky criterion to apply.
depend on the particular problem of of data has been collected, operational What about groups such as the Miao,
living that has been selected, and on models of the development of the who are dispersed over a wide area
the fact the units have some intrinsic disease and its relevant dimensions are with numerous intervening culture
meaning or relevance for the people designed. In this second stage, des- units, but who retain a very large
who make up those units. cription is replaced by less wordy measure of cultural homogeneity?
A survey of stored information about analysis. In the third stage, the models I am also disturbed by the omission

300 CURRENT ANTHROP O LO GY

This content downloaded from 62.204.192.85 on Fri, 04 Aug 2017 13:39:37 UTC
All use subject to http://about.jstor.org/terms
of a hierarchical ordering of the Naroll: ON ETHNIC UNIT CLASSIFICATION
culture units for comparative purposes.
It seems to me that a system is 1957:viii). Without such precision in presenting various criteria (which
needed which will permit ready terminology, one may easily be led Naroll attempts to define, and which
identification of the precise culture astray or may lead others that way may be elaborated in the light of
units or portions of culture units (as is not uncommon in other fields of further research), and they are dis-
employed through a uniform nomen- social research; e.g., in referring to criminated among themselves in terms
clature, perhaps adapted from the one "village" as an entity in understanding of a panel of "cultural determinants,"
proposed by Murdock (1953). the rural society of India [Mukherjee the panel remaining unlimited in
To apply Naroll's scheme without 1961]). coverage, theoretically.
this preliminary first step of classifica-Naroll stresses the need to identify
tion by levels would be analogous to a the "skin of a culture," and he
By GEORGE P. MURDOCK*
botanist studying and comparing attempts a typology of the cultunit.
plants with red-colored flowers The analogy of "skin" from biology Pittsburgh, Pennsylvania, U.S.A. 21 I 64
regardless of the content of his sample may not be particularly relevant in
in terms of families or isolated sub- the present context. For "cultures" are Naroll has addressed himself to an
varieties. not discrete entities; on the contrary, important issue, and much of what
Until a widely accepted and utilized however they may be looked at and he says is eminently sensible and
system of hierarchical ordering is whatever may be their specific pertinent. Nevertheless-and despite
available, it is doubtful that Naroll's determinants, their fields overlap. So, my complete accord with his ultimate
scheme can be entirely effective. if the probable density of a set of objectives-I find myself unable to
cultural determinants defines Ci and accept any of his concrete recom-
distinguishes it from Cj, a typology mendations.
By RAMKRISHNA MUKHERJEE*
of the cultunit, at this present hypo- The concept of "cultunit" is valid
Calcut#a, India 12 II 64 thetical stage of equating cultunit to and important, but the term itself is
Si in reference to Ci, may cut off, atan ambiguous barbarism. What it
Naroll wishes to initiate "a discussion its very beginnings, the rigorous connotes, if anything, is a component
of the problem of defining the 'tribe' development of ethnic unit classifica- congregation of a religious cult. Are
or 'society,' conceived of as the basic tion which Naroll has so seasonably there any serious objections to the
culture-bearing unit." That is, his initiated. For typological thinking of common practice of designating a
ultimate objective is to identify a set this nature may do more harm than culture-bearing unit as "a society"?
of people, Si, which represents a good, as has been pointed out by I emphatically reject the fourhold
distinct "cultural" situation = a set of Dobzhansky (1962), Garn (1962), and classification into Hopi, Flathead,
"culture" Ci. Following Naroll, (1) Si others. Aztec, and Tarascan' types. To inject
may be denoted as a "tribe," or any Naroll has raised a vital point about cultural content, such as the presence
one of its components Sij may be so cross-cultural surveys, when he asks or absence of a "state" or a "lingua
denoted; and (2) Ci may be composed how "sampling biases from in- franca," into the over-all definition
of Cil, C2, .-. ., Cii,... Ci. as "sub- consistent sampling units can be of a culture-bearing unit seems to me
cultures," while a "culture area" or statistically controlled, and the im- quite indefensible. Cross-cultural
"culture cluster" = Ci may be formed portance of diversity of societal type studies have demonstrated strong
of a conglomeration of Ci, Cj, Ck, and can be statistically assessed." He correlations between the occurrence
so on. asserts that this problem can be solved of states and other such cultural
But the method Naroll has adopted through the use of his concept of phenomena as complexity of social
does not distinguish Ci from Cj, and cultunit. But how is that possible, until stratification and degree of urbaniza-
on that basis differentiate Si cor- and unless it is established that the tion. If Naroll were to be taken
respondingly from Sj. Instead, Naroll cultunit presents the nature of varia- literally, however, any such study
examines six criteria which are tion specified for distinct sets of C and would be invalid because of the non-
assumed to define mutually distinct and thus does not lead to false dis- comparability of the units. For cross-
categories having concomitant rela- crimination in this respect? cultural comparisons to be possible,
tions with distinct sets of C. And, Naroll's enunciation of the problem any culture must be comparable with
purely on the basis of such an is important, and his formulation of any other, and this means that there
examination, he has formulated his cultunit may prove useful; but his can be only a single kind of "cultunit"
concept of the "cultunit" by taking deductions, interpretations, and in- -not four or any other number of
into account three of the six criteria ferences are a little premature. Cul- noncomparable types.
(one of only ancillary importance), tunit is proposed as a concomitant It seems to me that the only
after spelling out precisely what he variable in the perspective of the meaningful definition of a cultunit is
would like the attributes of each given relation between Si and Ci. social unit which carries a single total
criterion to be. Therefore, before elaborating this culture. Such a unit cannot be smaller
It follows from the above that, concept by means of its taxonomy than a single community, but it can

of such culture-bearing units, each


while cultunit would divide the field or its proclaimed powerto discriminate include several or many communities,
into mutually distinct groups of people in cross-cultural surveys, its validity provided the cultural differences be-
and thus could refer unequivocally to will have to be ascertained in the tween them are insufficient to yield
Si (and not either to Si or Sij), it willabove context. As has been pointed out appreciably different results in cross-
remain, all the same, an arbitrary and in the course of a discussion of racial cultural comparisons. Where a number
hypothetical concept in reference to classification (Garn 1962; Hiernaux
the relation between Si and Ci, until 1962), and as has been demonstrated with its "little tradition," share
cultunit is demonstrated from empirical in other fields of social research elements of a common "great tradi-
evidence to equate to Si. Hence, it may (Mukherjee and Bandyopadhyay 1964), tion" in consequence of economic
be more justified to label the concept, ethnic unit classification may require Interaction or political or religious in
for the purpose of C1, as "cultvariable," the application of statistical techniques tegration, such shared elements mu
meaning thereby "a varying quantity like Mahalanobis's D2, distance be- :f course, be counted as part of each
where the nature of the variation is tween clusters or groups: the groups rcultunit."
unspecified" (Kendall and Buckland are formed by precise attributes re- Naroll's purpose is "primarily to

Vol. 5 - No. 4 * October 1964 301

This content downloaded from 62.204.192.85 on Fri, 04 Aug 2017 13:39:37 UTC
All use subject to http://about.jstor.org/terms
solve sampling problems in cross- Nagel 1934:223-44; or more generally how Soviet archaeologists use the
cultural surveys." To him, the chief in Benjamin 1937:121-22, 134-37), term plemya, "tribe"), not only because
problem is the definition of the thus incurring the criticism formulated we cannot uproot them from the lan-
sampling unit, which to me seems by Kluckhohn (1960:135). Further- guage, but also because, somehow, they
easily soluble. What is in my opinion more, I would prefer to call the are useful for their connotations,
delimitation of the sampling universe, cultunit a "logical construct," and not those subtler shades of meaning they
the crucial problem, namely, the an "ideal type" which has a definite possess beyond their actual meaning
delimitation of the sampling universe, meaning in the philosophy of science or denotation. For these reasons, I
seems less important to Naroll. He (see Watkins 1953). It is only fair to doubt that such a precisely clarified
recommends that either my "World say, however, that "ideal type," with concept as the cultunit will have the
Ethnographic Sample" or my "Out- the connotation of opposable polarity, same usefulness as, say, Cooper's
line of World Cultures" be used as the has been widely though loosely used, typology of South American cultures
sampling universe in cross-cultural and is preferred by eminent sociologists (Cooper 1944) which, although much
studies. I must reject this suggestion, (e.g., Parsons 1958:320, 321). But less precise than the cultunit, provided
however flattering, for I am much too good usage prefers "construct," even a splendid basis for Steward's in-
acutely conscious of the serious in sociology (McKinney 1957:224-28). valuable Handbook, as he himself un-
defectiveness of both my efforts from
This is not purely a matter of equivocally states (Steward 1946:4).
the point of view of establishing an semantics, since it has epistemological However, as a tool for compara-
adequate stratification of the cultural implications (Nagel 1946:319, passim, tive statistical studies, the cultunit may
universe for sampling purposes. and references to Russell; also Benja- have a function, provided it is not
In my opinion, as set forth elsewhere min 1937:184-85). used only for extremes of minute
(1963), the appropriate sampling The second point is more pragmatic. empirical studies which employ refined
universe for cross-cultural research is It deals with the usefulness of Naroll's statistical methods but are unable to
not a universe of cultures or cultunits proposed typology. Typological con- reach a deeper understanding of
but a universe of "known cultural stracts must be (a) useful (Murdock sociocultural phenomena. There are
types" which still remains to be 1960:183; McKinney 1957:225) and certain qualitative aspects of socio-
laboriously worked out. When it is (b) able to explain (McKinney 1957: culture life which cannot be analyzed
worked out, however, a random 225-27). on a purely quantitative basis, how-
sample can be drawn in which every The cultunit, as carefully defined byever perfected, through simplifying
known cultural type in the world will Naroll, is certainly an improvement hypotheses or isolates introduced to
have an equal chance of being re- over the existing scandalous im- permit mathematical treatment. The
presented. The particular cultures or precision of the meanings of "tribe." very loss of their complexity precludes
cultunits which will represent the typesEven current textbooks give con- understanding them in full, and we
thus selected can then be drawn, eitherflicting conceptions. While Keesing will only be able to see the trees for
randomly or on the basis of the (1958:290), Beals and Hoijers (1959: the forest-if at all. Besides, qualita-
adequacy of the ethnographic evidence, 502-3), and Honigmann (1959 :325) tive judgments can also be systematic
using some of the devices which Naroll stress its political aspect, Hoebel and rigorous (cf. Kluckhohn 1960:138-
has elsewhere proposed for solving (1958:661) does not think it is 39).
"Galton's problem" and preventing "necessarily organized politically." My last point relates to culture
plural representation of historically However, by abstracting the three history. I liked the section on "the
contaminated cultunits. criteria of territorial contiguity, lan- cultunit in time," including the
guage, and political organization (as proposal on time notation, although
By ARTUR HEHL NEIVA* was his privilege), Naroll may have I believe that it will have only a
left only a skeleton in place of a vanishingly small chance of being
Rio de Janeiro, Brazil 31 I 64 living body, thus reducing the use- accepted. I also enjoyed the periodiza-
fulness of the concept. This loss tion of preliterate societies, although
Naroll's paper is important, first, would have to be balanced against Naroll's colonial period is somewhat
because after description, classification the undoubted gains in precision and inapplicable in countries outside the
is basic to scientific procedure; second, mathematics, is intrinsically vague U.S.A., where the process of contact
because clear-cut concepts, even at the(Russell 1948:162-63, 274-83), and and acculturation with aboriginals
risk of increasing jargon, are that all of us are used to this some- followed different cources (well
essential for any scientific analysis to times extraordinary vagueness in summarized in Ribeiro 1957:44-48)
be worthwhile; third, because Naroll anthropology, even for basic concepts and extended far beyond the date of
is not afraid to tackle some of the such as "culture" (cf. Kroeber and the country's political independence,
most awkward operational difficulties Kluckhohn 1952). When the situation which ended historically its colonial
encountered in fieldwork, such as becames unmanageable, anthropologists period. This is the case in many
appropriate sampling; and last, be- will suggest dropping a word al- Latin American countries, e.g., Brazil,
cause he may thus prepare for a together, as the UNESCO statements where such tribes as the Xavantes were
greater precision in handling cross- of 1950 and 1951 proposed for "race" only pacified after 1949, although this
cultural surveys, always vitally needed (cf. Montagu 1960:692-709; Barnicot process had begun much earlier (cf.
when new insights into ethnological 1963:117). Or they will consider it Alencastre 1864: passim, especially
problems are required. "unfortunate as a scientific term," as 320, 328-37).
How far has Naroll succeeded? Radcliffe-Brown said of "culture" But how does Naroll propose, for
Although his well thought out scheme (White 1954:462), or they will throw instance, to classify within his
is a step in the right direction, I want up their hands in despair and try to typology the Greek-speaking "tribes"
to raise three objections. avoid it, as Steward and Faron (1959: between the 11th and 8th centuries
The first concerns the philosophy of 17, 21) did for "tribe," declaring, after B.C. (8,900-9,200 new style)? Since
science. I have the impression that a rather good analysis, "the term tribe, the suggested taxonomic classes are
Naroll ignored the formal rules of thus having no clear meaning, will be non-exhaustive (see my remarks in the
fundamentum divisionis required in a generally avoided in this volume." But first point above), these preliterate
classificatory system (as stated in we continue to use these words every- Greeks, all speaking a common distinct
detail in Stebbing 1961:30, 246-49, where, sometimes compounding the language, cannot be subsumed under
433-39, 461-63; or in Cohen and confusion (cf. Krader 1959 :168, on either the Aztec or the Tarascan types,

302 CURRENT ANTHROP O LOG Y

This content downloaded from 62.204.192.85 on Fri, 04 Aug 2017 13:39:37 UTC
All use subject to http://about.jstor.org/terms
but neither can they be included in Naroll: ON ETHNIC UNIT CLASSIFICATION
Hopi type-because they were not
stateless-nor in the Flathead type- sense, logically or empirically. Naroll Also, Naroll's statement: "Societies
for the belonged to several states, apparently thinks that the mere which keep written records (including
each governed by some petty basileus spatial or physical proximity of two of course societies taught to do so by
(cf. Starr 1962 :chaps. 1-4, especially groups leads to "symbolic communica- Europeans)" invites the inference that
61-63, 80-81, 111, 114-15, 120-35). tion" or at least to some diffusion of Europeans have been the only people
This instance suffices to show the in- objects or ideas. But any student of to have taught writing to preliterate
adequacy of Naroll's typology, which divergent cultures knows that a groups.
would encounter the same difficulties "kilometer" varies with a society's 4. The final issue concerns the
with other protohistoric or proto- cultural orientation and technological author's efforts to operationalize his
literate peoples of his palaeoethnog- development; nor can we ignore linguistic concepts. First, in the matter
raphic period. Even in his aboriginal "environment" as a limiting condition. of mutual intelligibility, Naroll's
period this would happen, e.g., with The impact of technology is particular- analysis rests solely upon the spoken
the Tupi-speaking peoples of Brazil as ly dramatic; even in the United States, language. He ignores, as do so many
described by the 16th century what constituted 200 kilometers two other American anthropologists, the
chroniclers and epistolarians (e.g., centuries ago is a very different thing written language. In China, peoples
Nobrega 1955; Anchieta 1933; Cartas for the actors involved than 200 speaking languages that are not
Avulsas 1931; Staden 1930), and kilometers today. Or, conversely, two mutually intelligible have for millen-
probably also with the Xerente and preliterate communities lying only a nia sustained extensive communication
Xavante dichotomy (Nimuendaju' few kilometers apart but separated by links via the written language. The
1942 :1-7). a high mountain range may have little Chinese case casts doubts upon
In short, while congratulating or no contact, whereas communities Naroll's very definition of cultunit-
Naroll on his endeavor, I think that separated by thousands of miles of the subject of his paper. For he defines
his suggested typology should be ocean may be closely linked by this as "people who are domestic
reformulated, first to adapt it better modern technological developments. speakers of a common distinct lan-
to the more exacting requirements of 2. Naroll asserts that what he is guage and who belong either to the
logic and philosophy of science, and presenting is an "ideal type." It is not same state or the same contact group.--
second, in consequence, to widen it so at all certain that he comprehends And recall that his definition of a
it can be applied to peoples and tribes what is meant by an "ideal type," at "distinct language" rests upon mutual
not presently encompassed by it, there- least as formulated by Weber, who intelligibility. Under these conditions,
by increasing its practical usefulness saw his "ideal type" as an exaggera- China does not qualify as a cultunit,
as a tool not only for cultural anthro- tion of reality, as a construct that the culture-bearing entity, which will
pology but also for other social does not appear in concrete form come as a surprise to scholars.
sciences. (Weber 1949). Often Naroll seems to Moreover, while Naroll admits the
be using a "constructed type" in presence of difficulties in determining
Becker's sense rather than an "ideal what is mutual intelligibility among
By GIDEON and
type" in the Weberian sense (Becker languages or "dialects" and also what
ANDREE F. SJOBERG*
1950). are cognates, he nevertheless proposes
Austin, Texas, U.S.A. 17 II 64 3. In Letter to Associates No. 20, an easy way out: ". . .consider as
the Editor of CA argued for the use 'recognizable cognates' only those
'ie problem Naroll has set for him- of "nonpartisan" language in the words which are reported to share at
self is exceedingly important for journal and, by implication at least, least one meaning in common and
scholars engaged in cross-cultural advised against "ethnocentrism" in which also have two-thirds of their
research. Although this is admittedly the language of contributors to the phonemes in common and in the same
a difficult area, we have serious journal, viz: cc... its English must order." The basis for his choice of the
reservations about the author's schema. transcend the biases of the nations figure "two-thirds" is not at all clear.
1. Throughout Naroll confuses dif- to which it is native." But isn't it just Apparently, for Naroll certain
ferent levels of analysis. On the one as important to avoid ethnocentrism numbers possess an almost "magical"
hand, he attempts to establish certain in conceptual categories? Naroll's quality.
analytical distinctions on a rather proposed classification is blatantly
abstract plane. On the other, he seeks Western and, more than this,
"Americanist." For he seeks to ca- By LEIGH M. TRIANDIS*
to provide "operational definitions" of
certain concepts. But he does not keep tegorize in American Indian terms- Urbana, Illinois, U.S.A. 31 XII 63
these levels distinct. Hopi, Aztec, Flathead, Tarascan-
Consider one illustration of the phenomena that are world-wide and in Naroll does not include a list of
author's confused reasoning. Naroll many cases far earlier in time than societies classified by his four types
defines a "contact link" as follows: Naroll's denominations indicate. cultunit, nor does the paper discuss
"Two nuclear families constitute a More disturbing still is the ethno- procedures for classifying groups that
contact link if their dwellings are not centrism that is built into his chronolo- fit some of the criteria of a cultunit
more than 200 airline kilometers apart gical types of cultunit: the palaeo- but not all of them. Surely such groups
at some time in their annual cycle." ethnographic, the aboriginal, and the will exist, and I am sure that anthro-
Though he seems unaware of the fact,colonial. This leads Naroll to mix pologists will not want to exclude
this is on a very different level of absolute time with relative time. The them from consideration. The feasi-
abstraction than his definitions of colonial type, which for him includes bility of the classification remains to be
"cultunit," "state," "distinct lan- those cultures described after 1492, is seen. I am a psychologist, and I have
guage," or even "contact group." In an especially troublesome one. Here, always found variables to be more
other words, Naroll does not define logically, Naroll would- have to useful than categories in the psy-
"contact link" in general, analytical exclude the colonial expansion of, chological
say, description of behavior, but
terms; instead he offers a crude the Chinese, Indian, or Arab empires- this may not be the case when one
operational definition of this concept. which spread widely and enfolded deals with groups.
Moreover, Naroll's operational defi- numerous preliterate groups, long My other comment is that the unit of
nition of "contact link" makes little before the discovery of the Americas. study depends upon the problem to be

Vol. 5 * No. 4 * October 1964 303

This content downloaded from 62.204.192.85 on Fri, 04 Aug 2017 13:39:37 UTC
All use subject to http://about.jstor.org/terms
investigated. Naroll's classification distinguish between situations of operationalization of more rigorous
will be very useful for many types of egalitarian bilingualism, and bil- controls delimiting valid units for
problems, but no doubt he does not ingualism in which one language is the statistical cross-cultural study as such,
intend it as the only sampling method. prestige language and the other not. it is also to be commended for in-
(3). "Distinct Language.-All those dicating provocatively the fuller con-
dialects mutually intelligible to text of theoretical issues within which
By C. F. VOEGELIN* speakers of a stated dialect." Stated his particular concern is imbedded.
Bloomington, Indiana, U.S.A. 6 II 641 this way, it would be almost impossible For some purposes it is certainly a
to discuss distinct languages in ab- question of no small moment, how
When the successive terms of the original Australia, or in colonial people categorize their own behavioral
definition of the cultunit are indexed Polynesia today, or in other areas system, for instance, whether, or not
by the addition of a parenthetic where so-called "chain dialects" or they see it as emphatically unique in
number after each term, it is possible "dialect clusters" may be either neigh- plan, or only relatively so, within
to show that what is pivotal to the boring or distant (e.g., Polynesian certain or uncertain boundaries, which
cultish-seeming cultunit is language. Outliers). It would facilitate matters may or may not be conceptually
"Cultunit.-People (1) who are the to discuss "chain dialects" or "dialect emphasized and charged with value.
domestic speakers (2) of a common clusters" in terms of language barriers, To what extent, and in what pattern-
distinct language (3) and who belong and to inquire, whenever a barrier is ed ways, does their shared plan for
either to the same state (4) or the same encountered, whether the language existence take into account divergent
contact group (5)." beyond the barrier can be learned in a and contrasting plans, both con-
(1) In Naroll's context, "people" matter of days or weeks-or much ceptually and in terms of differential
contrast with "social insects," whose longer, as in the case of really different evaluations
societal behavior is genetically given languages. The term "ethnocentrism" has long
rather than learned, as culture is. But (4) "State. A territorially ramified been used by sociologists for invidious
in other contexts, for animals in territorial team whose leaders assert awareness of either cultural difference
general and especially for primates, it and wield exclusive the right to or social distinctness, both between
has been shown that some social declare and conduct warfare." subgroups of one system and between
behavior is learned, and some sub- Three types are distinguished, more fully distinct systems. Ethno-
human animals, especially when according to whether the lingua franca centrism is, of course, a highly complex
domesticated, do learn from man. The of the state is (a) the language of the phenomenon, which among other
cut-off point seems to be where further domestic speakers, or (b) not in- things may be more or less explicit,
learning would require knowledge of telligible with the language of the more or less intense, and may vary
human language. People seem to have domestic speakers; or (c) the simple considerably in the specificity or
a monopoly here; there may be a type in which no lingua franca is comprehensiveness with which shib-
genetic limitation (that precludes the needed because one language is shared boleth functions are attached to cul-
chimpanzee's learning human lan- by all. tural features.
guage) as much as a genetic map that (5) "Contact Group.-People who It has been very interestingly
matures as behavior without learning belong to no state but who speak a brought out by Hans Wolff, for
or teaching. common distinct language and who are adjoining Ijaw communities in the
(2) "Domestic Speakers.-People who all interconnected by successive contact Niger delta, how the interintelligibility
predominantly use a given dialect for links." Some Polynesian Outliers of dialects may become more problem-
speech within the nuclear family, that provide examples of a contact group atical where on one side people wish
is, among husband and wife and their without contact-who are not inter- to understand and be understood, but
minor children." In Indonesia today connected by successive contact links on the other side a superiorly aloof
there may be many families that and yet retain understanding of the attitude prevails, so that one group
virtually lack "domestic speakers"- language of the place from which claims interintelligibility and the
and so also in Israel and other new their forebears emigrated. other denies it (Wolff 1959). Pre-
nations which have designated one As defined, culture units or cultunits sumably there is more here than
language to be a national language. can be identified only with knowledge fudging or hypocrisy; rather it is a case
Though the national language is not of dialect distance and language of cultural distinctness -being in fact
the native language of either parent, it barriers. Pathetically little knowledge more real as an effective barrier, or
may be the native or first language of this kind is in the books or in less so, as affected by the culturally
of the children. Thus, a friend from research programs. Sometimes a little determined attiitudes of heightened or
Java spoke Bahasa Indonesian to both knowledge (e.g., reliance on "re- muted ethnocentrism.
his parents, who had at least a passive cognizable cognaties" of 80% or more Though presumably in the vast
knowledge of this national language, between two dialects to classify them majority of ethnographic materials
which they were slowly learning as as mutually intelligible) is a dangerous such difficulties may not disturb the
their children grew up. The father thing. It would be safer to ask speakers applicability of Naroll's methods,
spoke Javanese as his native lan- of each of two dialects whether they those methods or comparable ones
guage, and spoke Javanese at home can understand the other. But even if should not obscure the considerable
in the presence of his Javanese rela- this ask-the-informant method is residual non-comparability of social
tives; the mother spoke Sundanese as generally reliable, it does not advance (and related cultural) boundaries,
her native language, and was learning our theoretical understanding of dialect differences in the value-weighted
Javanese as well as Bahasa Indonesian distance or language barriers. nature of which are nicely suggested
while her children were growing up. in his mention of the "nesting" of self-
For the children, Bahasa Indonesian is identification frameworks for his
By LINVILL WATSON*
their mother-tongue, but not their Tyroleans.
mother's mother-tongue. But one can Easton, Pennsylvania, U.S.A. 18 II 64 As anthropology comes more and
extrapolate to the next generation, more to be concerned with interpreting
in which the parents will be bilingual, While Naroll's closely-argued pre- materials on the organizational level
as will the children, so that "domestic Sentation iS clear and cogent regarding of Naroll's linked Tarascan-and-Aztec
speakers" will have two native lanl- the overriding problem to which it is types (cf. the formula "national in
guages, and it will be important to explicitly addressed, namely the form, socialist in content" for

304 CURRENT ANTHROP OL OG Y

This content downloaded from 62.204.192.85 on Fri, 04 Aug 2017 13:39:37 UTC
All use subject to http://about.jstor.org/terms
relating nationalities within the Soviet Naroll: ON ETHNIC UNIT CLASSIFICATION
Union, and roughly analogous com-
plexities and ambiguities of "pluralism" generally corresponds to Naroll's vehtory of local communities." This
elsewhere in the modern world), it Tarascan type. A tribe, as I would like stricture does not take account that
will be still more important to consider to define it, is at present always sub- the unit most intensively studied by
in a sophisticated way how socio- ject to a sovereign nation, operational- standard ethnographic techniques is, in
cultural units are compounded and ly defined as being a member of the fact, a community rather than a larger
how boundaries change their mean- United Nations or as having applied unit. The anthropologist in the field
ings. to the United Nations for member- typically pitches his tent in a hamlet
ship. Thus, it does not have complete or village or wanders with a band.
autonomy and must obey at least some Although he will ordinarily get some
By JOHN W. M. WHITING*
of the laws of the nation within whose material from other communities of
Cambridge, Massachusetts, U.S.A. 19 II 64 borders it resides. Minimally a tribe the tribe by visiting some of them
does not have the right to declare war. during the course of his investigation,
The great increase in the frequency However, to be a tribe, some degree the detailed workings of the social
of world sample comparative studies of sovereignty is required. I would life is documented from observations
during the last 15 years makes Profes- follow Murdock's modification of on the community in which he lives.
sor Naroll's stimulating paper most Swanson's definition (Murdock 1962). Here, after having taken a census,
timely. I do not, however, agree that Sovereignty is characterized by made a map, and taken geneologies of
there is any one "ideal type" of unit, "original and definitive jurisdiction all the families, he participates in the
but rather feel that its definition over some sphere of social life in intimate details of daily life in a
should depend upon the problem being which the organization (tribe) has the manner vividly described by Ma-
investigated. Thus, a comparative in- legitimate right to make decisions linowski in the introduction to
vestigation of a problem in human having a significant effect on its Argonauts of the Western Pacific
biology or genetics should consider a members; e.g., distribution of food, (1922). The method is also described
set of breeding populations or allocation of productive resources, in Whiting & Whiting (1960). Such
endogamous groups as the appropriate punishment of delicts, assignment or techniques are clearly not suited for
universe; while for problems in lin- conscription of labor, levying of taxes, obtaining a description which will
guistics, speech communities comprise initiation of war and peace." No appropriately represent the culture of
the appropriate universe. For problems tribe subject to a nation has complete nations or even of large tribes. Here
of environmental adaptation, the in- sovereignty, but a tribe must have one must either depend upon materials
habitants of ecological regions-e.g., some degree of sovereignty. It has, as collected by censuses and surveys
desert peoples, tropical rainforest suggested above, relinquished the right carried out by the local government,
peoples-might provide the best to initiate war and peace, and it may the collation of results from com-
sample. have relinquished other rights as well. munity studies, opinion polls con-
Even if the problem is cultural, I How much sovereignty can be relin- ducted by others, or one must engage
do not feel that a single definition quished before the unit should be in an immense data gathering enter-
such as that offered for the cultunit isconsidered an ethnic minority rather prise that is seldom if ever undertaken
appropriate for all problems. As will than a tribe is an important considera- for the purpose of obtaining ethnog-
be discussed below, nations or states tion that must, again, depend upon the raphic information Furthermore, the
might be more appropriate for the problem being considered. The Hopi of type of material gained from national
study of problems involving political New Oraibi are generally classed as a censuses and opinion polls is of quite
variables, whereas local communities tribe, whereas the Chinese Americans, a different order from that gained
may be the more appropriate unit for even when they are living in a ghetto from participant observation in a
the study of kinship or child-rearing. and speak little English, are more community.
Naroll's types were not, as I under- often referred to as an ethnic minority. When studies of a larger scope than
stand them, designed to be used for For certain problems, a unit with the local community are undertaken
different purposes. They seem to be even less sovereignty may be required. by anthropologists, their usual techni-
based on empirical rather than theo- In an intensive study of family life, que is to sample communities from a
retical considerations. the local community is, as I have tribe or language group rather than
For problems which involve relating argued elsewhere (Whiting 1954), the attempting to sample individuals.
one aspect of culture with another- appropriate sampling unit. Such a unit Traditional ethnographic techniques
such as relating child-rearing practices may be a nomadic band, a neighbor- can thus be employed. If a sufficient
with projective systems-the unit must hood of scattered homesteads, a hamlet, sample of communities is obtained,
both be organized and have the or a barrio or ward of a town or it may be said to represent the larger
members in communication with one city. Some degree of sovereignty group. The cooperative research being
another. This suggests that the group should also be a required attribute of carried on by Romney, Vogt, and
have a common territory, speak a the community. Thus, the members of others in Chiapas, Mexico, is an ex-
mutually intelligible language, and such a group feel that there are certain cellent example of this method.
have some degree of sovereignty. This matters that are their concern and no As a consequence of the use of the
last requirement follows from the else's and that decisions with regard to field methods described above, the
assumption that a social unit must be such matters binding on the members great majority of ethnographies that
organized. The extreme diffusionist of the group. purport to describe the culture of a
"shreds and patches" view of culture While Naroll admits that the local tribe, and are so listed, might also
would not require sovereignty as a community "is clearly a good idea in be said to describe the culture of a
criterion. I would object on these certain kinds of comparative study," community within that tribe. The

-ni etelpmoc a ekil yleugav gnihtyna


grounds to including the stateless he claims that "in a cross-cultural nature and name of the community
Hopi type along with the sovereign survey, which seeks a random or or communities intensively studied are
Flathead and Aztec types, and the stratified random sample, it is opera- generallv described in the ethnogra-
dependent Tarascan type tionally worthless. It is hard eniough phy. Thus, a search of the ethno-
The most commonly used unit in to get something resembling a complete graphic literature on tribal units
recent cross-cultural studies is what is list of societies, but impossible to get would provide a list of communities.
commonly called a tribe. This unit If one and only one community were

Vol. 5 * No. 4 * October 1964 305

This content downloaded from 62.204.192.85 on Fri, 04 Aug 2017 13:39:37 UTC
All use subject to http://about.jstor.org/terms
taken as a unit from each listed tribe Each of these groups includes the one In sum, there is no single unit that
or nation, they would provide a following it, and thus each unit is at is appropriate for all purposes;
reasonable list from which to draw a a different level of abstraction. Since sovereignty should be an important
sample, and would, I believe, meet this procedure is followed throughout consideration in unit definitions when
Naroll's objection to the community the Outline, it is clearly inappropriate problems of cultural integration are
as a unit. for any sampling list. The outline being considered; the local community
This leads me to one further com- could, however, form a basis for is a viable unit especially suited to
ment. Naroll suggests that for "a compiling a number of lists each of comparative research utilizing existing
moderate budget survey" an appro- which would be appropriate for ethnographic material; and both
priate sampling unit might be the different problems. As we have sug- degree of sovereignty and level of ab-
Outline of World Cultures. If this gested, a list of nations, of tribes, and straction should be taken into
listing were taken literally, it might possibly of language groups could account in listing the sampling
yield a quite curious sample. The first easily be drawn. The task of listing universe. Finally, I hope that Naroll's
"culture" listed is the world. Then communities is more difficult and paper results in the compilation of
we have Asia, Korea, North Korea, would require a search of the sampling lists that are more suitable
and Quelpart (Specific data on... ethnographic literature, but it is for problems in cross cultural research
Quelpart Island and its inhabitants). clearly possible. than any now existing.

validation. They constitute a type of


stratification. They purport to group of other important traits; neither are
Reply
together similar tribes. If in fact they political organizations, endogamy
do, sampling is improved. groups, or largest meaningful social
By RAOUL NAROLL If they do
not, no harm is done. But as a formal unit. Going beyond these North
Many Murdock's
taxonomy, of these comments
culture type point
is up key American trait distribution studies,
issues.
still in Dole, Fuchs,
the proposal stage. Jaspan,
He has no Leach, looking at what is known of the
Moore, the
yet published Mukherjee,
taxonomic and
data Whiting
on a dis- histories of such tribal groups as the
singlecuss
type.theHis fundamental nature of the
proposal (Murdock Iroquois, the West Greenland Eskimos,
1963: culture-bearing
249-53) names 41unit. types, Berndt,
listing Chapple, the Nguni-speaking peoples, and the
Ducey,
for each typeFuchs,
by name Moore, Mukherjee,
the tribe or and Nahua-speaking peoples, I think it
tribesMurdock
included.propose alternate
He proposes general
(p. 251) safe to say that culture-bearing units
approaches
to publish in the to its definition.
future "tabulationsChapple, are also not congruent with respect to
Ember, Hoffman,
of ethnographic data" asLeach, and Watson
an "attempt chronological position and duration or
consider
to indicate the the difficulty
evidence on which of validation.
our pattern of genesis and development.
Hoffman,
classification Neiva,
of known and the
culture types Sjobergs
is Culture-bearing units are not species-
based."raise
I hopequestions
that byaboutthis he the logic of the objects; in Leach's phrase, a cultunit
means
cultunit
something more definition
than mereand the resultant
tabulations specifies a particular culture-bearing
taxonomy.
of general data of theBerndt,
sort he Bessach,
has beenDucey, group and not a category of such
Jaspan,
publishing Mering,
in his and Messing
Ethnographic Atlas.argue in groups as a zoological species design-
Such favor of using
tabulations, other particular
however useful as criteria ates a category of animals. If neat
than mine. Driver and
reference compilations, are inadequate Leach question congruencies between groups of traits
the validity
as evidence in support of ofthe cross-cultural
a taxonomy. were invariable, the problem of unit
surveyhis
To validate method
taxonomyas a whole;
Murdock Ember, definition would be easy to solve, and
needsMurdock,
to state for andeach
Whiting
genustake (e.g.,issue on there would be no need for such a
Southparticular
African Bantu) elementswhatoftrait(s)
it. There
it are discussion as this. Any particular
has inalso many which
common objections to matters of
differentiate world-wide definition of the culture-
it fromincidental detail, especially
the remainder of mankind, the choice bearing unit is sure to be arbitrary,
of particular
and again words
for each species as names
(e.g., Ngoni for to follow some boundaries and hence
type) concepts
to state what ("cultunit,"
traits it"Hopi"
has in type to ignore others. Consequently, there
common etc.)which
and the selection it
differentiate offrom
arbitrary cannot by any particular trait or trait
cutting points
the remainder of each(1492;
genus.the This200sortkilometer group which will be universally suit-
rule). able. But remember Goodenough's
of taxonomy will be most valuable not
only as a sampling list butt also as a rule: what we do as ethnographers is,
Fundamental Nature of the and must be kept, independent of what
general contribution
Culture to knowledge.
Bearing Unit
But please note that none of this we do as comparative ethnologists.
solvesMost
the of thedefinition
unit questions about the funda-
problem in To vary the unit definition to suit
mental nature of the culture-bearing
cross-cultural surveys. It simply sub- the problem being studied is a sensible
unit can be answered by a careful
stitutes a classified list of "societies" practice wherever it is feasible. But the
study of the North-American data, asfeasibility is not merely a matter of
for an unclassified one. It does not tell
compiled and mapped in 2 impressive devising theoretically suitable cate-
us in any rigorous way what a society
studies: (1) the voluminous California gories; it depends upon the ability of
is. Nor does it tell us what a "culture
type"trait element definition
is. Murdock's distribution
of a studies, the investigator, with the resources
whose theoretical results are sum- at his disposal, to apply those
culture type (p. 249) does not meet
marized
Chapple's in Kroeber
operational (1936); of
criterion and (2) categories in practice.
the continent-wide surveys of Driver
validity. It is vague and equivocal.
and Massey (1957) and Driver (1961).
No rigorous definition of a culture
General Approaches
These studies show that the
type is required in order to construct
culture- I have some questions about the
bearing unit is a complex phenomenon general approaches suggested by
a valid taxonomy; but such a defini-
rather than a simple one, that units are several commentators. Murdock's cul-
tion is required in order to make
not congruent with regard to quality, ture type approach is attractive,
rigorous inferences about a universe
quantity, or spatial extension. They especially as a sampling device.
from a sample taken from a list of
show that language groupings are Viewed as a sampling device, his cul-
alleged types. If the particular culture
often not congruent with distributions ture type groupings require no formal
CURRENT ANTHROPOLOGY
306

This content downloaded from 62.204.192.85 on Fri, 04 Aug 2017 13:39:37 UTC
All use subject to http://about.jstor.org/terms
being grouped is defined as a cultunit, Naroll: ON ETHNIC UNIT CLASSIFICATION
this unit definition problem is solved.
Leach has questioned whether any are often very large, I think. Would of the cultunit definition data on
single culture-much less a culture not the entire Northwest Coast con- 58 society War, Stress, and Culture
type-can be characterized by par- stitute the widest relevant social unit probability sample.
ticular traits found universally through- for a study of aboriginal Northwest My dissatisfaction with Ember's
out the culture but nowhere outside Coast tribes? Were not all these tribes definition, as with the earlier general
it. My cultunit concept does not linked in important trading, slaving, definitions of Murdock, arises from his
require any homogeneity of culture and military networks? The relation- failure to offer any evidence that he
with respect to any traits other than ship between traditional enemies is a tried to apply it in practice. Ember's
those used in the definition itself. social relationship of great importance, definition involves distinguishing one
Murdock's culture type classification and no unit which excludes such a language or lingua franca from
does require homogeneity with respect relationship could be properly called another. Yet he does not tell us what
to the as yet unidentified characteristic the widest relevant social unit. he means by a "different language."
traits of each type. As an operational Jaspan suggests we use 5 criteria, That is why I do not consider his
policy, Murdock might be well taking any 4 as a unit definition. I definition a rigorous one. His claim
advised to stick to criteria whose am not sure this would produce of rigor for it is evidence that he did
distribution is readily identifiable by mutually exclusive categories; and it not check the linguistic data on his
unskilled ethnographers because it is would certainly add to the labor of sample to see if his societies did, in
readily seen by run-of-the-mine in- applying the unit definition concept. fact, differ from their neighbors in lan-
formants. Cousin terminology or Otherwise, theoretically, it too seems guage. In short, he seems unaware of
residence rules, for example, would a sound approach. the problem of validation, of showing
not be operationally suitable because Mukherjee calls for us to identify a that the definition proposed actually
informants are often unaware of culture trait or group of traits whose applies to the units which are claimed
variations in such patterns wi thin a mapped distributions would yield to conform to it.
society. Since fieldworkers rarely have sampling units. My approach has been Chapple's criterion of validation is
validated their descriptions of whole rather to identify communication splendid. Indeed we must "come up
cultures by systematic sampling units, assuming that, since culture is with unequivocal operations which do
methods, most ethnographies are constantly changing, these uniits will not involve a penumbra of uncertain-
indeed, as Whiting points out, des- tend to be culturally distinct. Recent ty." The rival approaches discussed
criptions of a culture as it is seen work on Galton's problem (Naroll here need, first of all, to be validated
by informants of a single community. 1961a; Naroll and D'Andrade 1963; by showing in practice their ap-
I agree with those like Moore and Naroll 1964) offers protection against plicability to the data at hand "with
Ducey, who follow Murdock (1953: failures of this assumption in correla- unequivocal operations which do not
478) in seeking to distinguish a hier- tional inferences from cross-cultural involve a penumbra of uncertainty."
archy of culture bearing units. surveys. However, I would like to see I have designed the cultunit with
Although the task is beyond the one I someone try the statistical approach precisely this criterion in mind; but
have set myself here, ultimate solution Mukherjee suggests on panels of data my colleagues must, of course, with-
of the problem must certainly taken, say, from the California trait hold their judgment on this crucial
encompass such a hierarchy. element distribution studies. point until I publish the data on a
Perhaps the final solution will in- Zipf's "Law" of Equilibrium seems sample. They are likely to regard the
volve a cultunit concept along the worth looking into. But I have 2 validation problem of other unit
lines favored by Bessac, Jaspan, or me, reservations. First, the kind of data concepts with equal reserve, until
used in combination with the culture freely available to Zipf for his studies their proponents publish the relevant
type concept favored by Murdock, is not always available to cross-cul- data on a sample. The proof of the
Whiting, and Landy. By setting up tural comparativists in the library; pudding is in the eating.
units like the cultunits with clear hence in practice Fuchs is talking more
boundaries, one can then define the to fieldworkers than to comparativists.Logic
characteristic traits of the species in We do not often have collections of The new criterion of comparability
terms of predominance rather than of texts from successive communities proposed by Hoffmann and the old
universality, thus producing a species along a linguistic chain, for example. criteria of classification referred to by
differenit in character from that used Second, Zipf's demonstration did not Neiva both seem unimportant to me.
in zoology-a difference reflecting the establish what he claimed-it only On the other hand, the objection to
difference between cultural and biolog- showed that the traits he studied confusing levels of abstraction raised
ical evolution. followed a certain stochastic distribu- by the Sjobergs is well taken. We
Chapple's suggestion that we seek to tion model, in which the probability of should indeed distinguish theoretical
measure frequency of relationships a further increment is related to the concepts from operational indicators,
sounds promising in theory but dif- magnitude of the unit to be in- as I failed to do.
ficult in practice. He does not say cremented (see Simon 1955). Hoffmann's criterion of com-
whether we must wait for new field- parability requires that our categorie
work to be done all around the world The Validation Problem form conjunctive rather than dis-
or whether we can use this approach Triandis puts her finger on this junctive sets. His objection would be
on the data in existing ethnographic problem when she calls attention to satisfied if I redefined a cultunit as a
literature. He does not offer any my failure to include a list of tribes group of people who speak a common
concrete suggestions on how these classified by my system. This omission distinct language and who belong to
measurements are to be made either by was no oversight but rather a the same communication group. I
fieldworkers or by comparativists in reluctance to publish at this time my would ithen define a communication
the library. data on the 30 tribes of my 1956 group as a group of people who belong
Berndt's suggestion that we seek the sample because I still was not satisfied to the same state or to the same
widest relevant social unit seems to with it after extensive investigations, contact group. Thus, by a semantic
me sound advice for fieldworkers but some of whose results were reported in trick, I would move the disjunction
a poor basis for a unit of comparison my discussion of linguistic problems. I out of the definition of the set itself
in cross-cultural surveys. Such units do plan to publish a detailed discussion into the definition of one of the set's

Vol. 5 * No. 4 * October 1964 307

This content downloaded from 62.204.192.85 on Fri, 04 Aug 2017 13:39:37 UTC
All use subject to http://about.jstor.org/terms
criteria. The trivial nature of this Homogeneous Language.-A set of now cite the corroboration of this
change illustrates the unimportance of dialects the speakers of any one of rule in the study of the reconstruction
the logical goal achieved. If I were which can understand all the others, of printed English by Chapanis (1954).
to make this change, I would achieve but none of which is intelligible as Accordingly I suggest that we also call
comparability by Hoffmann's rule, as such to speakers of any dialect of any 2 words "similar cognates"-regardless
he predicted I could, but people who other set. of their linguistic histories-if 80% of
were concerned before about com- Language Chain.-A set of dialects their phonemes are reportedly iden-
paring Yaghans and English would be which are arrangeable in sequences tical and in identical order, and if
concerned still. such that, while each dialect in the they share one meaning in common.
Similarly, it is quite true that my sequence is intertelligible with the Glottochronological data is irrelevant
classification ignored the medieval preceding and the following dialect, here, for it considers all cognates, how-
logician's canon of fundamentum and all such intertelligible neighbors ever dissimilar phonetically. Thus
divisionis, according to which all are included, some dialects in the set English- "through" and German
species of a particular genus are to be are not intertelligible with others in "durch" are obvious cognates, yet
distinguished by the same "basis of the set. phonetically entirely unlike. The 80%
division" or fundamental criterion. (N.B. What I have here called a rule would rule out even English
This rule is not heeded by biological "language chain," following Voegelin, "hand" and German Hand, but would
taxonomists, and Neiva offers no is nothing else but what I earlier accept English "tobacco," German
evidence or argument in favor of it, called a "linguistic continuum," "tabak."
but only the authority of Clyde following Bloomfield. Notice that In cross-cultural surveys, it would
Kluckhohn. I could reply that Cohen "Homogeneous Language" and "Lan- be necessary in practice to concern
and Nagel (1934:242) advise against guage Chain" are mutually exclusive oneself with chain link classification
it. Like Hoffmann's rule, it is funda- and collectively exhaustive categories. only where it appeared that neigh-
mentally trivial, because a classifica- Everv dialect must belong to one or boring links were each represented by
tion which fails to conform to it can be the other of these; none can belong bodies of ethnographic literature
easily rearranged into a more cumber- to both.) meeting the bibliographic criteria of
some, less elegant one which does. All Chain Link.-One of a set of the sample in question and not
I would need to do is add 2 additional divisions of a language chain made in separated by state boundaries. For such
non-working categories to my system such a way as to produce the smallest chain links, the comparativist can
and proceed by dichotomies. Thus I possible number of divisions each reasonably expect to find or to be able
could divide cultunits into those with- containing only intertelligible dialects to construct enough parallel word lists
out states (Hopi type) and those and, wherever consistent with the to permit him to make a chain link
within states; then I could subdivide foregoing requirement, dividing neigh- classification.
those within states into those within boring dialects with maximum dif- I defend the 200 km. rule as a
linguistically homogeneous states (Flat- ferences. (In other words, one seeks practical operational criterion for
head type) and those within lin- boundaries which separate compara- identifying contact links. In the
guistically heterogeneous states; final- tively dissimilar dialects wherever absence of other data, I would assume
ly, I could subdivide those within lin- alternate division plans present them- that families within such a distance
guistically heterogeneous states into selves. This proposal assumes that lan- were contact links, those further apart,
those politically dominant (Aztec guages chain or dialect continuums not. Since the contact link comes into
type) and those not politically do not procede by exactly equal alter- use only in Hopi type cultunits-
dominant (Tarascan type). What has nations in dialect from community to those independent of any state,
been gained by all this logical ap- community, but rather that the whether native or colonial-it is un-
paratus? amount of the difference varies con- likely that this rule will often be mis-
However, I would agree with Cohen siderably. I understand that the dialect leading because of the effects of
and Nagel that, where a manifold maps of Germanic and Romance modern communication technology. It
empirical classification turns out on peasant dialects do reveal such migh,t help a little, however, to distin-
deeper examination and further variations.) guish water gaps, flatland gaps, and
research to have a fundamentum di- ContactLink.-Two nuclear families mountain gaps, providing different
visionis which its authors did not constitute a contact link if every year rules for each. I have not done so
perceive, such a finding suggests that one of the members of each speaks because all the instances I know of
the field of stucdy in question is directly to one of the members of the in which this rule needs to be used
scientifically maturing. The periodic other. are those involving native travel in
table of chemical elements is a fine Ethnographic Time Period.-That pre-European boats across the open
example. Thus it would be impressive period of time in which the cultunit sea.
indeed if, by fieldwork, it were type remains constant. The periodization rule now proposed
demonstrated that alike in a Hopi Having modified the foregoing con- calls for us to distinguish for example,
type, a Flathead type, a Tarascan type cepts to eliminate purely operational pre-colonial, colonial and post-colonial
and an Aztec type cultunit, the cul- criteria, let me now discuss their use periods, if the people belong to a
tunit boundary is, in fact, the major in practice. To divide a language Hopi, Flathead, or Aztec type cultunit
boundary of interaction frequency. chain into chain links, one needs before and after colonial conquest.
While I do not consider either of detailed dialect data. With the data It also calls for us to distinguish
the foregoing changes in my system now in libraries, it can be done changes in cultunit type in the forma-
useful, I do think changes are needed thoroughly for rather few language tion of a native state, for example,
to remove its confusion of theoretical chains. Many dialect groups like when the Zulu state organized a
and operational concepts, as the Timbira and Central Caroline can be Flathead type unit from the undif-
Sjobergs wisely say. Let me now so classified if one uses a frequency ferentiated 18th-century Nguni, who
redefine 3 of the basic concepts: definition of intelligibility, rather were then all members of a Hopi
distinct language, contact link and than an informant's statement. In type cultunit. (The concept of
time period. support of my proposal for the 80% palaeoethnography belongs to still
Distinct Language.-In homogeneous rule of "similar cognates," a rule another level of abstraction; it is a
languages, the languages itself. In derived from the results of the Biggs classification of ethnographic writings,
languages chains, the chain link. study of Yuman languages, let me not of peoples being described, and it

308 CURRENT ANTHROPOLOGY

This content downloaded from 62.204.192.85 on Fri, 04 Aug 2017 13:39:37 UTC
All use subject to http://about.jstor.org/terms
has its place in a sampling protocol Naroll: ON ETHNIC UNIT CLASSIFICATION
rather than in a unit definiition
taxonomy.) produces additional taxonomic cate- suggest that, in any cross-cultural
Particular Definitional Criteria gories, as Bessac made clear. survey, the comparativist note the
Ducey proposes that we also classify cultunit boundaries associated with
However important marriage ex-
according to level of complexity. A each unit in his sample. If he can use
change patterns may be in defining
step in this direction would be to a sampling list made up of cultunits,
ethnic units in Australia and New
distinguish between Hopi type societies so much ithe better. Thus the com-
Guinea, they are not, I think, general-
with politically organized settlement parativist can, by a distinct calcula-
ly so outside these regions. Variations
(bands, villages) and those without. tion, see whether, in the unit of
in such patterns ought to be considered
Naroll (1956), Freeman (1957), and references of the ethnographers he
as a variable in the research design
Carneiro and Tobias (1963) offer studies, variations from the cultunit
whenever its relevance to the problem
indices or scales of level of coni- standard affect his correlations or not.
at hand is suspected.
plexity. As will be shown in a forth- In other words, I propose that we
Native name is, however, a common
coming paper by Naroll and Tatje, strive to order all data by cultunits
and widespread, if bv no means
these indices and scales are highly for purposes of cross-cultural survey
universal ethnic unit definition cri-
correlated with each other, though comparisons. (This is not to deny the
terion. Bessac's proposal thus has
they use different measuring devices; value of other sorts of comparisons,
strong attraction, especially since the
hence, it seems clear that they are all which may well make use of other
reference group of a native name is so
tapping the same underlying factor- sorts of units.) To comparativists, then,
often very easily identified. Bessac's
precisely the factor which Ducey I suggest that you arrange your data
proposal is to subclassify my Hopi
wishes to consider. Any of them can by cultunits as nearly as you can and
type into those who have a name for
be used to classify or measure this take note of discrepancies which you
themselves (Bessac's Hopi type) and
factor. It is not necessary to classify cannot avoid.
those who do not (Yuki itype); and
by level of complexity in order to My proposal to fieldworkers is
to subclassify my Flathead type into
control for it. One can use any of different. I suggest that they determi
groups of linguistically homogeneous
these measures in a multiple factor their unit of study, of data collection,
states who have a common name for
analysis or partial correlation to see with primary regard to local cultural
theselves, presumably the same as their
conditions. Fieldworkers in parts of
own name for their language (Mongol whether relationships between other
traits being studied merely reflect Australia or New Guinea, for example,
type), and those in which language
covariation in level of complexity. might well need to pay greatest heed
and state boundaries are coterminous
Use of gross population as a to marriage exchange grouping. How-
(Mohave type). Finally, as to lin-
taxonomic criterion is easy enough. ever, I urge them to ascertain and
guistically heterogeneous states, Bessac
But since it is equally easy to use report the cultunit boundaries involved
retains my Aztec type, but he sub-
cultunit population as a control in their study, (i.e., the linguistic
stitutes for my Tarascan type his
factor, and since gross population is boundaries and the state or contact
Kazak type in which peoples speaking
probably important chiefly as it affects group boundaries) so that their reports
a distinct language and having a name
societal complexity through the in- may be accurately classified by coni-
for themselves are lumped together
parativists.
regardless of state boundaries. Bessac's tervening variable of settlement size,
I see no particular point in classifying As an ethnographic classification,
approach is promising, but it needs
by it. intended to serve for data on all the
some more work. At present, it is not
people, not merely on elites, the cul-
an exhaustive classification. What about In my cultunit taxonomy, I classified
only by those traits which were tunit is defined in terms of spoken lan-
2 linguistically heterogeneous states
necessary to provide an operational guage, not written language. Con-
having mutually intelligible lingua
unit definition. These traits were sequently, the Chinese written lan-
francas? The English-speaking peoples
guage, which used to be as widely
know themselves collectively as Anglo-selected for their relationship to geo-
understood in Korea and Viet Nam
Saxons, but they operate a number of graphical variations in culture, not for
their general importance. They are all as in China (and quite widely under-
linguistically heterogeneous states.
3 aspects or elements of communication stood in Japan as well), plays no part
Jaspan's suggestion that we consider
in the classification of the Chinese
also "important elements of common systems. If human beings formed
small, discrete isolated communication people. The Chinese Peoples Republic
folklore and history" poses a problem.
systems, those would be the units I of today is a single state, as far as I
It is clear that "ex-cultunit" groups
know. Within its mainland boundaries,
often continue to constitute ethnic units would prefer to sample. Since they do
not, I offer the cultunit as an ap- control of conduct of warfare is
of importance in the eyes of culture
proximation, and I claim for it the claimed and effectively exercised by
bearers and ethnographers alike. The
merit of being rigorously operational the authorities in Peking. I would
Jews today are less a religious sect
in the library most of the time. Other classify its inhabitants by their
than an ex-state; that is, they are
domestic speech into one Aztec type
aware of their descent from the formerimportant variations in culture may
properly be the basis for other sorts cultunit (Mandarin speakers) and a con-
citizens of the ancient Jewish state.
of taxonomies. Mine is a taxonomy of siderable number of Tarascan type
The Basque communities today are
units qua units. cultunits, among which would be
better defined as communities of those
found speakers of such Chinese
aware of their descent from Basque Application of the Cultunit dialects as Cantonese, Hakka, Fu-
speakers than as those who speak in Practice kienese, etc. I would pay no attention
Basque still. Such important "ex-
In proposing the cultunit as an "ideal at all to the fact that those inhabitants
cultunits" always have names for
themselves. Operationally, Jaspan doestype"-or "logical construct," if you who are literate employ a single
prefer that term-I did not make clear common ideographic script which they
not suggest how he dis,tinguishes im-
enough how I mean it to be used. pronounce differently when reading
portant elements of common folklore
Several commentators failed to grasp aloud.
and history from unimportant ones.
my meaning. I offer it to compara- Similarly, I would divide the
Particular Taxonomic Criteria tivists in cross-cultural surveys to be peoples of the United Kingdom into
Obviously, if either Bessac's or used whenever enough data is available one Aztec type and a number of Ta-
J aspan's proposal is accepted, it to permit its use. In other words, I rascan types. The Aztec type would

Vol. 5 * No. 4 * October 1964 309

This content downloaded from 62.204.192.85 on Fri, 04 Aug 2017 13:39:37 UTC
All use subject to http://about.jstor.org/terms
consist of those inhabitants of the number of basic problems of method, States is commonly called behavioral
U.K. who are domestic speakers of calls for their solution, and predicts an science. In this country, a large
some English dialect intertelligible important role for studies so improved. proportion of social or cultural an-
with that used to conduct business by Leach takes quite another position. thropologists consider themselves be-
Her Maies,ty's government. I would He considers the cross-cultural survey havioral scientists. Thus they associate
distinguish as Tarascan types, domestic method utter nonsense and a complete themselves with most American so-
speakers of those surviving dialects of waste of time, and he says he speaks ciologists and social psychologists as
English which are not intertelligible for most British social anthropologists. well as with the behavioral group of
with the Queen's English (including He gives 2 reasons- for this view here; American political scientists and with
one or more types of Cockney and one elsewhere (Leach 1963; Leach 1964) he such economists as Kenneth Boulding.
or more types of Lowland Scots, and gives still others. What he perceives In Britain, behavioral science is well
perhaps many others). I would also as difficulties of categorization seem to represented at such red-brick universi-
distinguish as additional Tarascan me rather difficulties of quality con- ties as London and Manchester, but
types the speakers of one or more trol (Naroll 1962). I am not so ready not I think at the ancient, prestigeful
varieties of Welsh and one or more as Leach to pronounce impossible a centers of Oxford and Cambridge.
varieties of Gaelic. Similarly in correct Ethnographic Atlas entry on There the study of human behavior is
France, I would distinguish the the Kachins. Either Leach wrote them overwhelmingly humanistic, over-
domestic speaker of literary Parisian up misleadingly, or the comparativists whelmingly descriptive, analytical,
on the one hand and of non-inter- did not read him carefully. Similarly, and intuitive.
telligible provincial dialects (patois) it is notorious that the term "clan" Needless to say, humanistic studies
on the other; and further distinguish has been used inconsistently by anthro-are well represented on all American
as necessary among Breton and Basque pologists (Leach 1964). If the com- campuses and are fully respected by
speakers. In what is left of the over- pilers of the Ethnographic Atlas did behavioral scientists. Description, ana-
seas British and French empires (not not take this problem into considera- lysis, and intuition are essential, we all
counting militarily and diplomatically tion, they fell short of the ideal agree. But at Oxford or Cambridge,
independent associates), I would con- standards of care. By classifying not only in the Department of Social
sider as part of the metropolitan Aztec reports according to date of compila- Anthropology, but on the campus as a
type cultunit domestic speakers of tion and nationality of author, one can whole, social studies are almost entire-
standard English or literary Parisian, expect errors of this sort to be detected ly seen as exclusively humanistic. The
respectively. and sort out domestic if they are frequent and if they pro- possibility of a natural science of
speakers of other dialects (whether duce biases raither than random errors. human behavior, in which regularities
English or French pidgins or not) as so Thus one can protect oneself from in- are identified through rigorous studies,
many Tarascan type cultunits. ferring correlations which actually is emphatically denied. Thus Leach is
reflect pairs of biases of this sort, talking at cross-purposes with those of
The Validity of the Cross-Cultural rather than behavior of the people us who are behavioral scientists. He
Survey Method as a Whole studied. wants only to describe structure and to
However, this sort of protection is classify; we want also to study the
I agree with Driver in calling for full of no value to Leach. He takes no interrelationships between variables.
studies of entire regions, of the sort interest in correlations. By "theory" Nevertheless, critiques from keen,
he has done so well, along with cross- he and many of his British colleagues knowledgable observers like Leach are
cultural surveys. Driver is calling our do not mean the analysis of relation- helpful to us; they keep us on our toes
attention to the need for a good "mix" ships (hopefully causal relationships) and remind us of questions we must
in our professional research effort as a among variables. They simply mean answer. Our best answers will come in
whole. We must not neglect ithe categorization. This attitude is espe- the form of truly rigorous cross-cul-
regional study, or the concomitant cially prevalent at Oxford and Cam- tural surveys which demonstrate the
variation study of the Redfield-Eggan bridge. It is not confined there to existence of correlations and show that
type, or the analytical case study of social anthropologists but is a general these cannot be plausibly explained
the single group. I also agree with characteristic of social studies. The away as artifacts of unit definition in-
Driver in his endorsement of Kobben's cross-cultural survey is an element in consistency, of sampling bias, of data
critique of the cross-cultural survey a broader interdisciplinary approach reporting or coding error, or of mere
method. Kobben lists a considerable to social studies which in the United diffusion.

BEACH, DOUGLAS M. 1938. The phonetics ton S. Coon, pp. 84-116. New York-
References Cited of the Hottentot language. Cambridge, Holt.
England: Hefner. CAMPBELL, A. 1868. On the tribes around
ALENCASTRE, J. M. PEREIRA DE. 1864. Darjeeling. Transactions of the Ethno-
BEALS, RALPH L. and HARRY HOIJER.
Annaes da Provincia de Goyaz. Revista logical Society of London 7:144-59.
1959. An introduction to anthropology.
Trimensal do Instituto Historico, Geogra- CAPELL, ARTHUR. 1941. A new Fijian dic-
New York: Macmillan. [AHN*]
phico e Ethnographico do Brasil, Toma tionary. Sydney: Australasian Medical
BECKER, HOWARD. 1950. Through values
27, Parte Segunda (Vol. 29), Quarto Pub. Co.
to social interpretation. Durham, N.C.:
Trimestre, pp. 229-349. [AHN*] --. 1962. Revised edition. A linguis-
Duke Unive-sity Press. [G&AS*]
ALLEE, W. C. 1958. Revised edition. The tic survey of the southwcstern Pacific,
BENJAMIN, A. CORNELIUS. 1937. An intro-
social life of animals. Boston: Beacon Noumea: South Pacific Commission.
duction to the philosophy of science. New
Press. [DL*] CARNEIRO, ROBERT L. and STEPHEN F.
York: Macmillan. [AHN*]
ANCHIETA, JOSEPH DE, S. J. 1933. Cartas, TOBIAS. 1963. The application of scale
informac6es, fragmentos histcricos e ser- BERNDT, RONALD M. 1959. The concept
analysis to the study of cultural evolution.
m5es do... (1554-1594). Publicag6es of "the tribe" in the western desert of
Transactions of the New York Academy
da Academia Brasileira. II. Hist6ria, Car- Australia. Oceania 30:81-107.
of Sciences, Series II, 26:196-207.
tas Jesulticas 3. Rio de Janeiro. [AHN*] BIcGs, BRUCE. 1957. Testing intelligibility Cartas Avulsas 1550-1568. 1931. Publica-
BARNICOT, NIGEL A. 1963. Racial taxon- among Yuman languages. International g6es da Academia Brasileira. II. Hist6ria,
omy. Current Anthropology 4:117-18. journal of American Linguistics 23: Cartas Jesui'ticas 2. Rio de Janeiro.
[AHN*] No. 2, pp. 57-62. [AHN*]
BARROW, JOHN. 1801-4. An account of BRIDGES, THOMAS. 1948. "The Canoe In- CHAPANIS, A. 1954. The reconstruction of
travels into the interior of South Africa. dians of Tierra del Fuego," in A reader abbreviated printed messages. Journal of
2 vols. London: T. Cadell & W. Davies. in general anthropology. Edited by Carle- Experimental Psychology 48-.496-510.

310 CURRENT ANTHROPOLOGY

This content downloaded from 62.204.192.85 on Fri, 04 Aug 2017 13:39:37 UTC
All use subject to http://about.jstor.org/terms
CHAPPLE, ELIOT D. and CARLETON S. COON. Naroll: ON ETHNIC UNIT CLASSIFICATION
1942. Principles of anthropology. New
York: Holt. [EDC*] Bielicki. Current Anthropology 3:27-28. method in social anthropology," in Read-
CHEYNE, ANDREW. 1852. A description of [RM*] ings in cross-cultural methodology. Edited
islands ... London: Potter. GODDARD, PLINY E. 1903-4. Life and cul- by Frank W. Moore, pp. 175-92. Human
COHEN, MORRIS R. and ERNEST NAGEL. ture of the Hupa. Hupa Texts. Univer- Relations Area Files, New Haven,, Conn.
1934. An introduction to logic and scien- sity of California Publications in Ameri- [HED*]
tific method. New York: Harcourt, Brace. can Archeology and Ethnology 1:1-88, KRADER, LAWRENCE. 1959. "Recent trends
COLSON, E. 1954. "The intensive study of 89-368. in Soviet anthropology," in Biennial
small sample communities," in Method --. 1911. Athapascan. Handbook of review of anthropology 1959. Edited by
and perspective in anthropology. Edited American Indian languages. Bureau of Bernard J. Siegel, pp. 155-84. Stanford,
by R. Spencer, pp. 43-58. Minneapolis: American Ethnology, Bull. No. 40, Part Calif.: Stanford University Press.
University of Minnesota Press. [RMB*] I: 85-158. Washington, D.C.: U.S. [AHN*]
COOPER, JOHN M. 1944. Areal and tem- Govt. Printing Office. KRAMER, AUGuSTIN. 1937. Zentralkaro-
poral aspects of aboriginal South Ameri- GODPiE-MOLSBERGEN, EVERHARDUS C., ed. linen. "I. Halbband: Lamotrek - Oleai
can culture. Annual Report of the Smith- 1916. Reizen in Zuid-Afrika in de Hol- -Feis," in Ergebnisse der Sfidsee-Expedi-
sonian Institution for the year 1943: landse Tijd, Vol. XI. 's-Gravenhage: tion 1908-1910. Edited by G. Thilenius,
pp. 429-61. [AHN*] Publication of the Linschoten Society. Vol. 10. Hamburg: Friederichsen.
DAMM, HANS. 1938. Zentralkarolinen. "II. GOODENOUGH, WARD H. 1956. Residence KROEBER, A. L. 1923. Anthropology. New
Halbband: Ifaluk - Aurepik - Faraulip rules. Southwestern Journal of Anthro- York: Harcourt, Brace and Co. [HED*]
- Sorol - Mogemog," in Ergebnisse der pology 12:37. 1936. Culture area distributions III:
Sfidsee-Expedition. Edited by G. Thile- GUSINDE, MARTIN. 1931. Die Selk'nam. Area and climax. University of California
nius, Vol. 10. Hamburg: Friederichsen. Die Feuerland Indianer, Vol. 1. Modling Publications in American Archeology and
DOBZHANSKY, THEODOSIUS. 1962. Comment bei Wien: Anthropos. Ethnology 37:111-12.
on "Issues in the study of race: Two 1937. Die Yamana. Die Feuerland KROEBER, A. L. and CLYDE KLUCKHOHN.
views from Poland," by A. Wiercinski Indianer, Vol. 2. Mbdling bei Wien: 1952. Culture: A critical review of con-
and T. Bielicki. Current Anthropology 3:
Anthropos. cepts and definitions. Papers of the Pea-
26-27. [RM*] HIERNAUX, J. 1962. Comment on "Issues body Museum of American Archaeology
DRIVER, HAROLD E. 1956. An integration in the study of race: Two views from and Ethnology, Vol. 47, No. 1. Cam-
of functional, evolutionary and historical Poland," by A. Wiercinski and T. Bielicki. bridge, Mass. [AHN*]
theory by means of correlations. Indiana Current Anthropology 3:29-30. [RM*] 1963. Culture: A critical
University Publications in Anthropology HOBHOUSE, L. T., G. C. WHEELER and M. concepts and definition. New York: Ran-
and Linguistics, Memoir 12. GINSBERG. 1930. The material culture dom House. [DL*]
1961. Indians of North America. and social institutions of the simpler KUNOW, STEN. 1906. "General Introduc-
Chicago: University of Chicago Press. peoples. London: Chapman and Hall. tion, Specimens of the ... Himalayan
DRIVER, HAROLD E., JOHN M. COOPER HODGSON, BRIAN H. 1857. Comparative Dialects .. ." (Part I, "Tibeto-Burman
PAUL KIRCHHOFF, DOROTHY LIBBY, WM. vocabulary of the language of the broken Family." Edited by George A. Grierson),
C. MASSEY, and LESLIE SPIER. 1953. tribes of Nepal. Journal of the Royal in Linguistic Survey of India, Vol. III.
Indian tribes of North America. Indiana Asiatic Society of Bengal 26:317-49. Calcutta.
University Publications in Anthropology HOEBEL, E. ADAMSON. 1958. 2d edition. LEACH, EDMUND. 1954. Political systems of
and Linguistics, Memoir 9. [HED*] Man in the primitive world. New York: highland Burma. London: Bell.
DRIVER, HAROLD E. and A. L. KROEBER. McGraw Hill. [AHN*]
1960. Review of: Organization of
1932. Quantitative expression of cultural HOERNLE, AGNES W. 1925. The social
work: A comparative analysis of produc-
relationships. University of California organization of the Nama Hottentots of
tion among non-industrial peoples, by
Publications in American Archaeology and southwest Africa. American Anthropol-
Stanley H. Udy, Jr. (New Haven: HRAF
Ethnology 31:211-56. [HED*] ogist 27:1-24.
Press 1959). American Sociological Re-
DRIVER, HAROLD E. and WILLIAM C. HOIJER, HARDY. 1946. Introduction to view: 25:136-38.
MASSEY. 1957. Comparative studies of Linguistic structures of native America.
North American Indians. Transactions of Edited by H. Hoijer, pp. 9-29. Viking --. 1963. Comment on: "Forms and
the American Philosophical Society, 47.
problems of validation in social anthro-
Fund Publications in Anthropology, No.
pology," by William J. McEwen. Current
Philadelphia. 6. [HED*]
Anthropology 4:174.
EMBER, MELVIN. 1963. The relationship HONIGMANN, JOHN J. 1959. The world of
between economic and political develop- man. New York: Harper. [AHN*] - 1964. Comment on scheffler's note
ment in non-industrialized societies. Eth- Hsu, FRANCIS L. K. 1963. Clan, caste and on the Mangaian Kopu. American Anthro-
nology 2:228-48.
pologist 66:427-29.
club. Princeton: Van Nostrand.
EVANS-PRITCHARD, E. E. 1940. "The Nuer HYMEs, DELL. 1960. Lexicostatistics so far. LEEs, ROBERT B. 1953. The basis of glotto-
of the Southern Sudan," pp. 272-96 in Current Anthropology 1:3-44. chronology. Language 29:113-27.
African political systems. Edited by Meyer JASPAN, M. A. 1957. Bangsa dan Suku- [HED*]
Fortes and E. E. Evans-Pritchard. London: bangsa di Indonesia. Jogjakarta: Kemen- LEVINE, ROBERT A. 1961. Anthropology
Oxford. terian Penerangan. [MAJ*] and the study of conflict: an introduc-
FISON, LORIMER. 1886. Specimens of Fiji - . 1959. A provisional list of Indo- tion. Journal of Conflict Resolution 5:
dialects. Edited by A. S. Gatschet. Inter- 3-15.
nesian ethnic groups. Sosiografi Indo-
nationale Zeitschrift fair Allgemeine nesia 1:75-90. [MAJ*] LOTHRQP, SAMUEL K. 1928. The Indians
Sprachwissenschaft 2:193-208. KEESING, FELIX M. 1958. Cultural anthro- of Tierra del Fuego. Contributions,
FORTES, MEYER. 1940. "The political sys- pology, the science of custom. New York: Museum of American Indians, Heye
tem of the Tallensi of the northern Rinehart. [AHN*] Foundation, Vol. 10.
territories of the Gold Coast," pp. 239- KENDALL, M. G. and W. R. BUCKLAND. McKINNEy, JoHN C. 1957. "Methodology,
71 in African Political Systems. Edited 1957. A dictionary of statistical terms. procedures and techniques in sociology,"
by Meyer Fortes and E. E. Evans- London: Oliver and Boyd. [RM*] in Modern sociological theory in con-
Pritchard. London: Oxford. KENNEDY, R. 1945. Biography of Indo- tinuity and change. Edited by Howard
FORTES, MEYER and E. E. EVANS-PRITCHARD, nesian peoples and cultures. New Haven: Becker and Alvin Boskoff, po. 186-235.
eds. 1940. African political systems. Lon- Human Relations Area Files. [MAJ*] [AHN*]
don: Oxford. KLUCKHOHN, CLYDE. 1960. "The use of MALINOWSKI, B. 1922. Argonauts of the
FREEMAN, LINTON C. 1957. An empirical typology in anthropological theory," in western Pacific. London: George Rout-
test of folk urbanism. Ann Arbor: Uni- Men and cultures, selected papers of the ledge and Sons. [JW*]
versity Microfilms, No. 23502. Vth International Congress of Anthro- MONTAGU, M. F. ASHLEY. 1960. 3d edition.
FRITZ, GEORG. 1911. Die Zentralkaroli- pological and Ethnological Sciences. Edited An introduction to physical anthropology.
nische Sprache. Berlin: Reimer. by Anthony F. C. Wallace, pp. 134-40. Sprinigfield, Ill.: C. C. Thomas.
GARN, STANLEY M. 1962. Comment on Philadelphia. [AHN*] [AHN*]
"Issues in the study of race: Two views KOBBEN, ANDRE J. 1961. "New ways of MUKHERJEE, RAMKRISHNA. 1961. A note on
from Poland," by A. Wiercinski and T. presenting an old idea: The statistical village as unit or variable for studies of

Vol. 5 *No. 4 *October 1964 311

This content downloaded from 62.204.192.85 on Fri, 04 Aug 2017 13:39:37 UTC
All use subject to http://about.jstor.org/terms
rural society. Eastern Anthropologist 14: PIERCE, JOE E. 1954. Crow vs. Hidatsa in SWADESH, MoRRIS R. 1950. Salish internal
3-29. rRM*] dialect distance and glottochronology. relationships. International Journal of
MUKHERJEE, RAMKRISHNA and SuRAj BAN- International Journal of American Lin- American Linguistics 16:157-67.
DYOPADHYAY. 1964. "Social research and guistics 20:134-6. 1959. The mesh principle in com-
Mahalanobis's D2," in Contributions to PILLING, ARNOLD R. 1962. Statistics, sor- parative linguistics. Anthropological Lin-
statistics (in press). Calcutta: Statistical cery and justice. American Anthropologist guistics 1:7-14. [HED*]
Publishing Society, and Oxford: Pergamon 64:1057-59. TEIT, JAMES A. 1930. The Salishan tribes
Press. [RM*] POWDERMAKER, HORTENSE. 1933. Life in of the Western Plateaus. Annual Reports
MURDOCK, GEORGE P. 1953. "The process- Lesu. New York: Norton. of the Bureau of American Ethnology
ing of anthropological materials," in QUIGLEY, CARROLL. 1961. The evolution XLV: 37-197. Washington, D.C.: U.S.
Anthropology today. Edited by A. L. of civilizations. New York: Macmillan. Govt. Printing Office.
Kroeber, pp. 476-87. Chicago: Univer- RADCLIFFE-BROWN, A. R. 1922. The An- TOYNBEE, ARNOLD J. 1947. Abridged
sity of Chicago Press. daman Islanders. Cambridge, Eng.: Uni- edition. A study- of history. New York:
--. 1957. World ethnographic sample. versity Press. Oxford.
American Anthropologist 59:664-87. --. 1940a. "Preface," pp. xi-xxiii in VAN NIEUWENHUIJZE, C. A. 0. 1963.
--. 1958. 2nd edition, revised. Outline African political systems. Edited by Cross-cultural studies. The Hague: Mou-
of world cultures. New Haven: Human MEYER FORTES and E. E. EVANS- ton and Co. [FDB*]
Relations Area Files. PRITCHARD, London: Oxford. VOEGELIN, C. F. and ZELLIG S. HARRIS.
--. 1960. "Typology in the area of social --. 1940b. On social structure. Journal 1951. Methods for determining intelligi-
organization," in Men and cultures, of the Royal Anthropological Institute bility among dialects of natural languages.
selected papers of the Vth International 70:1-12. American Philosophical Society Proceed-
Congress of Anthropological and Ethno- REDFIELD, ROBERT. 1941. Folk culture of ings 95:322-29.
logical Sciences. Edited by Anthony F. C. VOEGELIN, C. F. and E. W. VOEGELIN.
Yucatan. University of Chicago Press.
Wallace, pp. 183-88. Philadelphia. 1941. Map of North American Indian
[AHN*] REICHARD, GLADYS A. 1938. "Social life,"
in General anthropology. Edited by Franz languages. American Ethnological Society
1962. Ethnographic atlas. Ethnology Publications, No. 20. [HED*]
Boas, pp. 409-86. New York: Heath.
1:269. [JW*] WANDRES, C. 1927. Tiernamen in der
RIBEIRO, DARCY. 1957. Linguas e culturas
--. 1963. Ethnographic atlas. Ethnology Nama- und Bergdama Sprache etymo-
ind'genas do Brasil. Rio de Janeiro:
2:249-53. logisch erlautert. Festschrift Karl Mein-
Centro Brasileiro de Pesquisas Educa-
NADEL, SIEGFRIED F. 1951. The founda- hof. Sprachwissenschaftliche und andere
tions of social anthropology. London, cionais. [AHN*]
RIVET, PAUL. 1911. La famille linguistique
Studien: 125-233. Hamburg: Friederich-
Cohen and West. sen, de Gruyter & Co.
NAGEL, ERNEST. 1946. "Russell's philo- Peba. Journal de la Societe des Ameri-
WATKINS, J. W. N. 1953. "Ideal types and
sophy of science," in The philosophy of canistes de Paris, n.s., 8:173-206.
historical explanation," in Readings in
Bertrand Russell. Edited by Paul Arthur RUSSELL, BERTRAND. 1948. Human knowl-
the philosophy of science. Edited by
Schilpp, pp. 317-49. Evanston, Ill.: The edge, its scope and limits. London:
Herber Feigl and May Brodbeck, pp.
Library of Living Philosophers. [AHN*] George Allen and Unwin. [AHN*]
723-43. New York: Appleton-Century-
NAROLL, RAOUL. 1956. A preliminary in- SCHAPERA, I. 1930. The Khoisan peoples
Crofts. [AHN*]
dex of social development. American of South Africa. London, Routledge.
WEBER, M. 1949. The methodology of the
Anthropologist 58:687-715. --. 1953. Some comments on compara- social sciences. Translated and edited by
--. 1961a. Two solutions to Galton's tive method in social anthropology. Ameri-
E. A. Shils and H. A. Finch. Glencoe,
problem. Philosophy of Science 28:15-39. can Anthropologist 55:353-62. Ill.: Free Press. [G&AS*]
--. 1961b. MS. Two stratified random --. 1956. Government and politics in WHITE, LESLIE A. 1954. Review of The
samples for a cross-cultural survey. tribal societies. London: C. Watts. nature of culture, by A. L. Kroeber, and
--. 1962. Data quality control. New SIMON, HERBERT. 1955. On a class of skew Culture: A critical review of concepts and
York: The Free Press of Glencoe. distribution functions. Biometrika 42:425. definitions, by A. L. Kroeber and Clyde
--. 1964. A fifth solution to Galton's SNETHLAGE, HEINRICH. 1930. Unter nord- Kluckhohn. American Anthropologist 56:
problem. American Anthropologist 66: ostbrasilianischen Indianern. Zeitschrift 461-68. [AHN*]
No. 4, August.
fur Ethnologie 62:111-205. WHITING, JOHN W. M. 1954. "The cross-
NAROLL, RAOUL and FRADA NAROLL. 1962.
SPENGLER, OSWALD. 1926. The decline of cultural method," in Handbook of social
Social development of a Tyrolean vil-
the West. New York: Knopf. 2 vols. psychology. Edited by Gardner Lindzey,
lage. Anthropological Quarterly 35:103-
(Translated from German by Charles Vol. 1:523-31. Cambridge, Mass.: Ad-
20.
Francis Atkinson.) dison-Wesley.
NAROLL, RAOUL and RoY G. D'ANDRADE.
1963. Two further solutions to Galton's STADEN, HANS. 1930. Viagem ao Brasil. WHITING, J. and B. WHITING. 1960. "Con-
problem. American Anthropologist, Oct. Versao do texto de Marpurgo, de 1557, tributions of anthropology to the methods
1963, in press. por Alberto L6fgren. Publicasoes da of studying child rearing," in Handbook
NIMuENDAJu, KURT. 1915. Vokabular der Academia Brasileira. II. Hist6ria. Rio de on research methods in child development.
Timbiras von Muranhao und Para. Zeit- Janeiro. - [AHN*] Edited by P. H. Mussen, New York: John
schrift f&r Ethnologie 47:302-5. STARR, CHESTER G. 1962. The origins of Wiley. [JW*I
--. 1939. The Apinaye. Catholic Univer- Greek civilization, 1100-650 B.C. Lon- WILLIAMs, ROGER J. 1958. Chemical
sity of America Anthropology Series, don: Jonathan Cape. [AHN*] anthropology-an open door. American
No. 8. (Translated from German by STEBBING, L. SUSAN. 1961. 7th edition. A Scientist 46:1-24. [DL*]
Robert H. Lowie.) modern introduction to logic. London: WILLIAMS, THOMAS. 1858. Fiji and the
--. 1942. The gerente. Translated by Methuen. [AHN*] Fijians, Vol. 1. Edited by George R.
Robert H. Lowie. Publications of the STEWARD, Rowe.
JULIAN London:
H. A. Heylin. "Introduc
1946.
Frederick Webb Hodge Anniversary Pub- in Handbook of South American Indians. WISSLER, CLARK. 1917. The American
lication Fund, 4. Los AngeJes: The South- Edited by Julian H. Steward. Smithsonian Indian. New York: D.C. McMurtrie.
western Museum. [AHN*] Institution, Bureau of American Ethnology [HED*]
--. 1946. The eastern Timbira. Univer- Bulletin 143, 1:1-9. Washington, D.C. WOLFF, HANS. 1959. Intelligibility and
sity of California Publication in Ameri- [AHN*] inter-ethnic attitudes. Anthropological Lin-
can Archeology and Ethnology, Vol. 41. 1955. Theory of culture change. guistics 1:34-41. [LW*]
(Translated from German by Robert H. Urbana, Ill.: University of Illinois Press. ZIPF, G. 1935. The psycho-biology of
Lowie.) [FDB*] language. Boston: Mifflin. [HF*I
N6BREGA, P. MANUEL DA. 1955. Cartas do STEWARD, JULIAN H. and AFRED ME- ---. 1946. "The psychology of language,"
Brasil e mais escritos do... (opera TRAUX. 1948. The Peban tribes. Hand- in Encyclopaedia of psychology. Edited by
omnia). Coimbra: Universidad do Coim- book of South American Indians, III: P. H. Harriman, New York: Philosophical
bra. [AJN*] 728-36. Library. [HF*]
PARSONS, TALCOTT. 1958. Revised edition. STOW, GEORGE W. 1910. The native races --.1949. Human behaviour and the
Essays in sociological theory. Glencoe, Ill.: of South Africa. London: Swan Sonnen- principle of least eff ort. Cambridge, Mass.:
The Free Press. [AHN*] schein. Addison-Wesley. [HF*]

CURRENT ANTHROP OL OG Y

This content downloaded from 62.204.192.85 on Fri, 04 Aug 2017 13:39:37 UTC
All use subject to http://about.jstor.org/terms

S-ar putea să vă placă și