Sie sind auf Seite 1von 31

Linguistic Relations across Bering Strait

Author(s): Morris Swadesh


Source: American Anthropologist, New Series, Vol. 64, No. 6 (Dec., 1962), pp. 1262-1291
Published by: Wiley on behalf of the American Anthropological Association
Stable URL: http://www.jstor.org/stable/667850 .
Accessed: 10/01/2014 11:08

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

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

Wiley and American Anthropological Association are collaborating with JSTOR to digitize, preserve and
extend access to American Anthropologist.

http://www.jstor.org

This content downloaded from 200.54.94.206 on Fri, 10 Jan 2014 11:08:42 AM


All use subject to JSTOR Terms and Conditions
LinguisticRelationsAcrossBeringStrait
MORRIS SWADESH
UniversidadNacionalAutonomade MGxico

THE archeological and other evidence indicating that America has re-
ceived various waves of imigration in the last 25,000 years suggests that
there may be linguistic relationships also. If there had been only one ancient
movement many millennia back, the American languages would by now have
so far diverged from any kin still in the Old World as to defy any ordinary
efforts to discover similarities deriving from their common origin. Only an
arduous long-term effort to reconstruct very old protolanguages in both
hemispheres might eventually permit a successful comparison between them.
However, since there were in fact later migrations, the wonder is that there
is as yet no generally accepted instance of related languages in both hemi-
spheres, with the exception of the obvious unity of Siberian and American
Eskimo. The difficulty may lie in the remoteness in time of the latest migra-
tions, or in the limited intensity of comparative studies up to now. Possibly
both circumstances are involved. At any rate, there can be no doubt that the
closest intercontinental affinities are fairly distant. None can be compared,
say, to those which interconnect the Germanic languages, a group whose
separation goes back 2,000 years. In fact, as estimated by the index of basic
vocabularly retention, the time depth of the nearest possible relationships is
evidently greater than 4,000 years.
CIRCUMSTANCESOF RECENT MIGRATIONS
The fact that there are no linguistic relatives with a closer time-depth than
4,000 years in Asia and America does not imply that no migrations have taken
place in all that time. It is perfectly possible that a group of people having
arrived speaking a new language eventually was absorbed into an already
established linguistic community. Or the pioneers could have been joined by
more and more of their co-linguals, and, after the new language was well
established in the New World, those who still remained behind could have
been absorbed into some other expanding community. There are many paral-
lels for such variant possible developments in recorded history. Familiar
examples of languages which did not perdure in new territory are those of
Dutch in New York, Norman French in England, or Arabic in Spain. The
opposite case, of persistence only in the new territory, may be shown by
Dominican Carib in Central America, Magyar in Europe, Indic languages in
India.
Although preservation in both areas is a fairly common development, it
is evident that this did not occur between Asia and America during recent
millennia, and this fact is of considerable importance to prehistory. It suggests
1262

This content downloaded from 200.54.94.206 on Fri, 10 Jan 2014 11:08:42 AM


All use subject to JSTOR Terms and Conditions
[SWADESH] Linguistic Relations Across Bering Strait 1263
that by about 3000 B. C. there may have been a fairly complete settlement of
the areas around the entrance to North America, with populations capable of
annihilating or assimilating new bands. At the same time it may imply that
the pressure of expanding populations in Asia was pushing peripheral tribes
into the north coast areas so strongly that residual groups there were either
forced to follow their advance guard into America or be absorbed on the far
side. Some of the details of these events will certainly some day be recon-
structed by archeological findings studied in the light of linguistic and other
evidence.
These considerations show the importance of determining as accurately as
possible the interrelations of American languages and Asian languages among
themselves and their intercontinental connections. The conditions of the
problem require that relationships be established in considerable depth. That
is, if the nearest intercontinental affinities go back 5,000 years, it will be
necessary to solve whatever technical problems may be involved in such remote
ties. Is this too tall an order? I believe not, and in the present paper I will
attempt a preliminary demonstration of one of the key relationships, that
between Eskimo-Aleutian in American and Chukotan in Asia.

THEORIES OF INTER-HEMISPHERICRELATIONSHIP
To better orient our analysis, it is well to examine briefly a few of the more
serious theories of linguistic affinities between the Old and New World.
Rasmus Rask, one of the greatest and most able of the pioneer comparative
linguists, found evidence relating Eskimo and Aleut to Uraltaian, but his
effort did not become generally known until the present century. Uhlenbeck
renewed the question in 1905. Sauvageot, in response to his suggestion, pub-
lished data comparing Uraltaian with Eskimo and attempting to establish
phonetic laws (1924). Rask's comparisons (see Thalbitzer 1921) and some of
Sauvageot's appear to be fairly convincing, but many of the latter's seem
forced, so that his phonological equations remain doubtful. The matter re-
mained dormant for a number of years, but was reexamined a few years ago
by Bergsland (1959) on the basis of his very successful work of systematizing
the comparative study of Eskimo and Aleut. This scholar presents interesting
evidence of structural and vocabulary similarities between Uralic and Eskimo-
Aleutian which tend to support the theory of their relationship, but he him-
self does not consider it proved. Thus, he states (p. 14) that "In lexical matters
the Eskimo-Uralic hypothesis will probably remain a non-liquet..." and that
"The number-both absolute and relative-of seemingly possible morpho-
logical comparisons is fairly large . . . but the element of chance cannot be
eliminated...." It is possible that this inconclusive finding is partly due to
the author's skepticism from the outset, since he begins by saying, "Like most
problems of remote linguistic relationship the subject of this paper belongs to
the realm of speculation rather than to that of scientific linguistics." Never-
theless, he analyzes the question in a fashion which will unquestionably be
helpful for future study.

This content downloaded from 200.54.94.206 on Fri, 10 Jan 2014 11:08:42 AM


All use subject to JSTOR Terms and Conditions
1264 American Anthropologist [64, 1962
In1935, C. C. Uhlenbeck published a set of fairly impressive comparisons
between Eskimo and Indo-European. Jensen (1936) criticized some details
of these comparisons, without either accepting or rejecting kinship. Ham-
merichin 1951 and Thalbitzer in 1952 added to Uhlenbeck's list. Since some
linguists (e.g., Collinder 1934) are convinced of agenetic relationship between
Indo-European and Uraltaian, Rask's theory and Uhlenbeck's need not be
regarded as mutually contradictory.
Thalbitzer (1952) called attention to similarities between Eskimo and
Chukchi, but his explanation was that Chukchi must have had an Eskimo
substratum. Recently Shimkin (1960) listed similar words between Chukchi
and Aleut, mentioning common origin as a possibility. Ankeria (1951) has
compared Chukotan with Uralic.
In 1925, Sapir announced the discovery of a link between Nadennean in
Americaand Sinotibetan in Asia. In correspondence with Laufer and in some
of his classes, he cited portions of the evidence, which covered vocabulary,
structural features, and specific structural elements. And at his death he left
twolarge notebooks with hundreds of comparisons. His reason fornot publish-
ingany of this was that he felt he should first complete his work on Nadennean,
a task which was cut short by his death. In 1952, Shafer, knowing of Sapir's
workbut without having consulted it, published an independently assembled
set of comparisons, along with a reconstruction of the phonology. After that
(also 1952) the present author published a few items from notes on Sapir's
class lectures, along with some additional comparisons of his own.
1925 also saw the publication of Rivet's comparison of Malayo-Polynesian
and Hokan, with a long and impressive list of similar radical and affix ele-
ments. However, these points of contact were culled from a fair number of
American languages and the great many which make up Malayo-Polynesian,
a circumstance which led some linguists to feel that the similar forms were
not necessarily more numerous than might result from the chance factor.
In the following year, Rivet came out with his even bolder theory of con-
tact between Australia and South America, with evidence offered for the
specific relationship of Chonan with the Australian languages as a whole. The
case is open to the same criticism as the previous one. However, in another
publication (1961), the present author finds the agreements to be far in excess
of the factor of chance, even when one takes material from only one language
in each of the two groups. On the other hand, the same holds for Australian
languages compared with Quechua, Wintun, or various other languages,
which can hardly be held to have come in over the Antarctic. By glottochrono-
logic estimate, the time depths are far greater than the 5,000 years supposed
by Rivet. Moreover, Chon is no isolate in America, for it shows relationship,
much closer than that to Australia, with many other New World languages.
With this demonstration that there is no special connection between Chon
and Australia, there remains the problem of the more generalized Australian-
American relationship. Possible solutions include rejecting the concrete evi-
dence of greater-than-chance resemblance, attributing it to hypothetical

This content downloaded from 200.54.94.206 on Fri, 10 Jan 2014 11:08:42 AM


All use subject to JSTOR Terms and Conditions
SWADESH] Linguistic Relations Across Bering Strait 1265
common psychological factors, or taking
it to be a persisting reflection of a
very ancient common origin of the languages of both
hemispheres. To the
present author the last solution appears to be the most
reasonable of the
three. The first, denying the evidence, would be
unscientific; the second,
possible psychological causes, is
not suggested by the actual instances of
agreement. The third might have appeared an
equally vague speculation five
years ago, but researches of the past several years tend
to give it some sem-
blance of probability.
Recent researches of various scholars have been
ofsome of the large linguistic unities demonstrating the reality
suggested in the past and have even added
new broad relations. The present
author, while trying to clarify some of the
geneticproblems of the American continent, kept
finding more and more con-
nectionsuntil it appeared that all
thelanguages of the NewWorld must have
hada common origin. This naturally raised the
question of intercontinental
ties,and indeed exploratory tests with samplings of
material seemed to bear
out the possibility. The next requirement is to attempt deeper studies to
determinewhether the first appearances will be
confirmed or found wanting,
andthat has been the author's principal task the
last three years. To untangle
andrejoin the thousands of bits of evidence in a
major problem of compara-
tivelinguistics is of course no simple
job, and the present one will have to
continuefor a long time, but it is perhaps well to make
some of it known with-
outundue delay. This Ido in thepresent
paper.
PERSPECTIVE
AND FOCUS
Theprincipal difference between my recent
comparative research and
thatwhich Ihad attempted previously has been
the effort to see the separate
languagesand the groups in an ever broader
perspective. In thecourse of this
examination,it was found that large sets of linguistic stocks took on
ofvast intergrading networks. That the aspect
is, complexes of great time depth were
seento have, on an enlarged scale,
something of the characteristics of the
meshof local dialects making up a single
language. Within one speech com-
munity,the shading off of speech forms tends to be
ofdistance, in that neighboring dialects are fairly
a simple function
more similar to each other than
thoseseparated by
intervening types, but the continuity may be broken at
some points as a result of population movements or the
spread of some dialects
atthe expense of others. Although the breaks in the
erallymore marked as one takes ever broader intergradation aregen-
perspectives, there still remains
some correlation between distance and linguistic
A affinity.
methodological implication of this circumstance is that, in cases of
break between linguistic entities, one has the wide
possibility of working step by
stepthrough series
a of off-grading congeners. Thus, if it is hard
connection between language A and language to trace the
N,one may try comparing A
withB, then B with C, andso on. Instead of attempting to compare Green-
landicEskimo with Finn, as
with Sauvageot did, it may be easier to connect Eskimo
Chukchi, then Chukchi with
Uralic, and finally Eskimo with Uralic.

This content downloaded from 200.54.94.206 on Fri, 10 Jan 2014 11:08:42 AM


All use subject to JSTOR Terms and Conditions
1266 A merican Anthropologist [64, 1962
For another thing, phonologic and semantic anomalies, that defy solution
in one part of the network, may be due to local changes in one area that per-
haps did not take place somewhere else. This means that one may perhaps
find an operating linguistic pattern somewhere in America that explains a
whole class of apparent irregularities in some Eurasian linguistic group, or
vice versa. Once one has the clue, it is likely he will then be able to recognize
many bits of evidence which once appeared disconnected, so that the solution
can actually be confirmed within the affected part of the broad linguistic net.
This sort of thing happens not only with reference to phonetic variations but
also in matters of meaning. What seems to be a single somewhat ramified set
of derivatives from a single root when seen within a smaller linguistic family,
may turn out to be the confusion of two or more originally distinct elements
when seen in the perspective of the broader network. Such additional light
on comparative problems is welcome not only because it corrects premature
inaccuracies but because a better analysis is needed in order effectively to
cope with the problems of distant linguistic relationship.
In the present paper we focus on the relationship of Eskimo-Aleutian of
America with Chukotan of Asia, but in our research these have been studied
as part of a relationship network which also includes Wakashan in America,
and Uraltaian, Ainu, Gilyak, and Sumerian in Eurasia, to speak only of some
of the less distant of the congeners. Farther removed in the sequence of rela-
tionship but very helpful because of the great time depth of the already
reconstructed material, is Indo-European. This whole complex of relation-
ships is in accord with the previous theories of Rask and Uhlenbeck, connecting
Uraltaian and Indo-European with Eskaleutian. The involvement of Chuk-
chian, already suggested by Thalbitzer and Shimkin, is in accord with its
geographical position in between Eskaleutian and Uraltaian. Sapir's theory of
Nadennean affinities with Sinotibetan will not be taken up at this time, but
our preliminary tests give them a place in the same network. The general
perspective is given in a previously published work (Swadesh 1960b).
THE DIMENSIONS OF VARIABILITY
There is a special problem in all comparative linguistics which is generally
recognized and successfully met in problems of relatively close comparison,
but which unfortunately tends to be avoided in remote comparison. It is the
multiplicity of variation. In general terms, there are three dimensions of
difference in comparative philology. One that is never forgotten, because it is
a foundation stone of the science, is that of historic change. Because of the
thoroughly proven principle (or "law") of phonological regularity, the com-
paratist knows that he can and must discover equations of phonologic agree-
ment between any two once-identical languages, for example in word-initial,
English t=Latin d (e.g., English two, Latin duo). However, this is not the
only kind of difference.
Another dimension of variance is spatial, consisting in the fact that two
local variants of the same language may have different treatments of the

This content downloaded from 200.54.94.206 on Fri, 10 Jan 2014 11:08:42 AM


All use subject to JSTOR Terms and Conditions
Linguistic Relations Across Bering Strait 1267
SWADESH]
same original phoneme, thus,
Latin k before front vowels is represented by an
s-sound in most local forms of Spanish, but by a
O-sound (like th in Eng.
think) in Castille and neighboring provinces. There is no problem as long as
the divergent products are confined to separate
areas, but sometimes there
is a mixture of dialects, that is, the copying of individual
forms from one area
in the speech of another.
A third form of change is functional
as in English sang as against sing ("symbolic" in Sapir's terminology),
(vocalic alternation), or believe as against
belief(consonantal The special feature of these cases is that the
alternation).
change of vowel (only in certain verbs in modern English,
although some
thousandsof years back the variation was not so
limited) express a difference
in tense; the consonant change, which happens to be
historically secondary,
dependingon the presence or absence of following vowel, marks verb versus
noun. Here, also, the old existence of a phonemic
alternation is no problem
forthe comparatist as long as the process remains active.
But, if it goes out
ofuse with sometimes one or another of the forms and
occasionally both re-
tained,with or without a distinction of meaning, the comparatist
may fail to
inferfrom the evidence he has that such was the case. he
If then attempts to
establishphonologic equations between two
languages, both of which show
effectsof the alternation, he may be a loss as to why a phoneme in one
at
languageseems to correspond to various things in the other. For
example, if
alinguist were to attempt the comparison of English with Spanish without
knowledgeof older stages of both languages when vowel
change was more
usual,he would not understand
whythe rounded vowel of English foot should
correspondwith a front unrounded sound in Spanish pie.
Recognizing the
equationof English f to Spanish p, he might well conclude that
these forms
agreeonly with respect to the first consonant, and that
they must therefore
beonly accidentally similar. He would thus lose hold of
one of the important
bitsof evidence which can help establish the common
origin of English and
Spanish,and to reconstruct one of the old morphological
processes which
characterized the common language which gave rise to these.
Fortunately,science is not a passive product of past methodological limita-
tions.The mere fact that a problem continues to be
studied is likely to lead
the discovery of the once hidden realities. This is
to evidently what is coming
tolight with some of the interhemispheric and other
comparisons where a
certain pessimism had developed with respect their possible solution. The
difficulties to
encountered by those who sought to relate Uraltaian with
Eskaleu-
tiannow appear to be due to unsuspected phenomena of old dialect
and mixture
of functional alternation. If the search is now
renewed with the specific
intentof discovering and reconstructing these lost
features, the present diffi-
cultiesmay well yield to scientific study. A partial solution to the
problem of
ancient phoneme alternation has perhaps already been achieved for
Indo-
European(Swadesh and Altaic (Swadesh In the
1960a) 1962). present in-
stance,even though we do make the proof of
not relationship rest on old
alternants,the theory has proved enormously helpful in
marshalling the evi-

This content downloaded from 200.54.94.206 on Fri, 10 Jan 2014 11:08:42 AM


All use subject to JSTOR Terms and Conditions
1268 American Anthropologist [64, 1962
dence, in that a large number of convincing cognates were discovered with its
help which would surely have been missed otherwise.
ESKALEUTIANAND CHUKOTAN
Eskaleutian falls into two well-marked divisions, Eskimo and Aleut,
whose relationship was indicated long ago by Rask (Thalbitzer 1921), but
which has only recently been provided with a clearly defined phonology
(Marsh and Swadesh 1951, Bergsland 1951, 1953, 1956, 1959). By glotto-
chronologic estimate the amount of basic vocabulary divergence is something
like three times as great as that between German and English, but the diffi-
culties of finding the cognate elements have proved considerably greater than
might be suggested by this analogy. The number of phonetic changes is per-
haps not very large, but some have been fairly drastic.
Fortunately, some perspective is gained on the Eskimo side by the existence
of a number of dialects falling into at least two distinct languages, Alaska-
Siberia, called Yuit or Yuk or Yupik, and Alaska-Canada-Greenland, called
Inuit, Inuk, or Inupik. Aleut also has dialect variations, study of which has
helped clarify Eskaleutian phonology.
In Chukotian, too, there are two main divisions, Kamchadal and Chukchi-
Koryak (Skoryik 1958). By glottochronologic estimate, the divergence is only
about two-thirds that of Eskaleutian, but the phonologic complication may be
as great. Kamchadal, now reduced to one form and much mixed with Russian
and Koryak, had considerable dialect variation. Three principal dialects have
been noted in Koryak, called Koryak proper, Kerek, and Aleutor. Apparently
Chukchi itself does not have any pronounced variations.
The order of relationship between Eskaleutian and Chukotan (glotto-
chronologically estimated as around 45 minimum centuries) is approximately
like that between Germanic and Romance, considered in terms of the present-
day languages. From the outset structural similarities were evident, which on
closer examination proved to be even greater than they had originally ap-
peared. Lexical agreements were also noteworthy from the beginning. Some
of them at first appeared to involve enormous phonologic difficulties, but
many of these were eventually solved in terms of precise rules of positional
variation. Moreover, these same rules uncovered a great many additional
cognates not noticed before. Some of the comparisons were found to contain
alternate forms, requiring the development of a theory of paradigmatic inter-
change, which in turn proved helpful in uncovering additional cognates. The
task of tracking down common elements is hardly more than well begun, yet
there are already over 200 strictly parallel sets (like English seven, French
sept) and many additional oblique sets (like English foot, French pied). In the
present paper, because of its introductory character, we cite only enough
cognates to demonstraate each of the protophonemes and the structural
parallelism of the stocks being compared.
It has been possible in a relatively short time to make a reasonably thorough
search of available Kamchadal, Chukchi, Aleut, and Eskimo dictionaries only

This content downloaded from 200.54.94.206 on Fri, 10 Jan 2014 11:08:42 AM


All use subject to JSTOR Terms and Conditions
Linguistic Relations Across Bering Strait 1269
SWADESH]
by the help of electronic equipment and the cooperation of the
Centro de
C,lculo Electr6nico of the Universidad Nacional Aut6noma de Mexico. We
owe thanks principally to
Knut Bergsland, Georgiy Menovshchikov, and
Dean S. Worth for help in obtaining materials.
Bergsland in addition gave
data and suggestions for some of the
Eskimo-Aleut comparisons. Thomas A.
Sebeok and Dell Hymes have given very helpful
suggestions about limiting
the focus of the paper and arranging the evidence in
convenient form.
GRAMMATICAL FEATURES

In keeping with what is normally to be expected in languages related at


a time-depth of about
5,000 years, Chukotan and Eskaleutian show many
grammaticalsimilarities, some differences,
andvarious instances in which one
stockshows mere vestiges of processes still active in the
other.
Withreference to morphological processes, both stocks
make principal use
ofsuffixation. Chukotan also uses a limited number of
prefixes, while Aleutian
hasnone at all and Eskimo only one, the demonstrative
otherfree demonstratives. Chukotan employs stem ta(s)- combined with
composition, aprocedure
absent in Eskaleutian, which compensates by
agreater use of suffixation.
Chukotanmakes common use of reduplication, with the
special office of
markingthe absolute singular of certain nouns; in Eskaleutian this
isnot used. process
Vowelalternation in Chukotan is partially amere matter of
vowelharmony, but also serves to differentiate noun cases and verbal voices,
particularlyin Kamchadal; in Eskaleutian there is no functional vowel
inter-
change. Ingeneral, the Asian stock has wider variety of formal
a
whilethe American one has specialized on the use of mechanisms,
suffixes, but the latter
appearsto have vestiges of each of the missing processes.
Hundredsof examples
can be cited of reduplicated noun absolutes in
Chukotan,like yilyil tongue to
itin Eskaleutian, except that someinflectional base yil-, andnothing exactly like
nouns have astem form quite parallel to
theChukotan absolute. Thus, Inuk
siksik, sikyik ground squirrel (from
*zikizik),Eskimo li9niq, nipniq goose, Yuktixtiq humpback salmon, Aleut
cugcugcodfish; possibly Eskimo tuktu, tuntu deer (perhapsfrom *tuktuk or
*tu0tun). Inat least two cases, there is comparative evidence to support the
presumedreduplication. Thus, Aleut susu-
forcognacy with Chukchi wicowc pus meets the phonological rules
eruption, since each form would bethe
normal phonological development of an original
dalzinazin-teen *wisowiso. Similarly, Kamcha-
(in counting from 11 to 19)and Aleut sisa- hundredappear
to
be two different plans of reduplicating the stem
the first being derived from *sinasina and the second *sini light, reason, calculates,
from *sinsan. Although
only a few instances have been identified so
far, they lend some likelihood to
the former existence of reduplication in an earlier form of
Eskaleutian, which
could have coincided with pre-Chukotan.
The
large number of suffixes in Eskimo, some with
isevidentlya secondary development. This is veryconcrete meanings,
suggested by the fact that it is
not even matched in Aleut, belonging to the same
stock, and also by the fact

This content downloaded from 200.54.94.206 on Fri, 10 Jan 2014 11:08:42 AM


All use subject to JSTOR Terms and Conditions
1270 American Anthropologist [64, 1962
that some of the suffixes are found to agree with stems in Eskimo itself or in
theother languages. For example, Inuk -palaq bad,miserablemay be compared
with Inuk palaq- is quiet, Chukchi palqat- is old; -piluk bad, wicked with
Inuk piluk- is poor; -tuq- swallows with Inuk tuqluaq throat; -kan-niq nearly,
with Inuk katigiaq- goes to meet; -kuluk little with Kamchadal kul-ic youngest;
-suk- is about to, -suaq- tries to, with Aleut su- takes; -anik- stops, with Inuk
ani- goes out. In similar fashion, some of the affixes of Chukchi are related with
stems, as has been recognized by Bogaraz (1937). In sum, then, some of the
features of affixation in which Chukchi and Eskimo seem to be farthest apart
may be due to recent developments. That area of suffixation in which they
agree presumably reflects the earlier state of affairs. In other respects they
have drifted apart, with Chukchi adding elements at the beginning and Eskimo
at the end of the word. In these matters, Kamchadal and Aleut probably give
the best representations of archaic structure.
In both stocks, verbal inflection is expressed principally by complex suffixes,
including tense-mood and person, added more or less mechanically to the stem
(which may be simple or complex). However, an alternate treatment, by
means of auxiliary particles which receive the endings, is found both in
Chukotan and in the Aleut branch of the American stock. One of the auxiliary
stems of other languages occurs in suffix position in Eskimo, where it surely
represents just one more example of the suffixing trend.
A detail in which Chukotan and Eskaleutian seem to diverge sharply is
in the placement of the pronominal object. In Eskimo it is part of the inflec-
tional ending; in Aleut it is an independent particle; in Chukotan it varies
between suffix, prefix, and independent. Aleut may give the best represen-
tation of the old system, since an independent form might have become
agglutinated at the beginning or end of the word, depending on the prevailing
word order in each dialect.
Meaningful consonant alternation is found only in Chukotan, where c is
the diminutive variant of 1 in all the languages and also of n in Kamchadal.
In the vowels, there are interchanges which principally obey rules of vowel
harmony but which are in some details meaningful; they have been reported
(Skoryik 1958) to be regular in Chukchi, almost regular in Koryak, sporadic
in Kamchadal, vestigial in Aleutor, and entirely absent in Kerek. Thus, Kerek
agrees with Eskaleutian. The nonoccurrence of the interchange is related to
the absence of the vowels e and o, which are obviously essential to maintaining
the interchange in the form found in Chukchi, involving the pairs: u/o, i/e,
e/a. Probably more significant for determining the old system is the scheme
used in Kamchadal, which includes the sets u/o/a and i/e/a. While true
functioning phonemic alternation today occurs only in Chukotan, vestiges of
the process are found in all the languages. The evidence can best be discussed
after we have treated the historical phonology.
In the matter of inflectional categories, Chukotan and Eskaleutian show
many more agreements than differences. The noun is always inflected for
number and case, the verb for tense-mood and pronominal subject. The

This content downloaded from 200.54.94.206 on Fri, 10 Jan 2014 11:08:42 AM


All use subject to JSTOR Terms and Conditions
Linguistic Relations Across Bering Strait
SWADESH] 1271
numbers are singular, plural, and dual in both branches of
Eskaleutian and
in Koryak on the other side; the remainder of
Chukotan also shows traces of
having formerly possessed all three categories, if
we may judge by the fact
that the form characteristic for the dual in Eskaleutian
is found as a plural
in certain uses. Cases found in both branches are
absolute, locative, ablative,
dative, comitative, and instrumental. The
handling of possession in the two
stocks is different, since Chukotan has a possessive
case, while Eskaleutian
puts the possessor in superordinate and attaches a
a pronominal element to
the possessed noun. Moods common to both
groups are indicative, optative,
subjunctive, and imperative; certain
participial moods and the interrogative,
foundin Eskimo, are peculiar to it.
To complete the structural comparison of the
betaken of instances of coinciding two stocks, account must
morphological elements. These are presented
in later section, following the treatment of
a phonology.

PHONOLOGY
Oneof the chief requisites for reasonable
a demonstration of common ori-
gin,serving directly as proof and at
thesame time providing atest of common
originfor specific morphemes, is consistent
a phonological theory. This re-
quirementcan be met for
Chukotan andEskaleutian. The main facts are
summarizedin the table below, in which the
contemporary language families
areeach represented by a specific dialect.
Thephonemes of the reconstructed
protolanguage are shown in the top
sectionof the table, in rows and columns,
according to positions and types of
articulation.The development of each phoneme is then to
be found in the
correspondingpoint of the section belonging to each language. The
usedto indicate loss of a phoneme. Arrows are used sign #is
where convenient to show
thedirection of change in the cases of fused items.
The diagonal indicates
alternatechanges according to phonetic environment. The
items are not simply
phonemesin the specific sense, but rather super
phonemes covering any
mechanicalvariation in each language based on phonetic
environment. Thus,
forexample, k in Greenlandic corresponds to k
before vowels, g before a
stop, rj before a nasal. Where a master phoneme could be
confused with
another,the more changeable of two is represented by small
the
N in Chukchi is that n which is replaced a cap letter.
Thus, by y before a consonant cluster.
The details of these mechanical alternations
inthe descriptive works on each can (forthe most part) be found
language; for present purposes, it may be
sufficient to give a simple summary of variations, without
to ex-
plain the conditions that give each form. is to be found attempting
It after the table.
Itwill be noted that in fewa particulars our phonemic analysis differs
from that of other authors. We do
not insist that ours is necessarily the best
treatment, but we believe makes
it all the necessary distinctions. Itshould be
for other scholars to recognize the
easy forms and thereby to realize what new
interpretations are being made, with the possible exception of
which we believe Kamchadal,
has never had an adequate phonemic interpretation. This

This content downloaded from 200.54.94.206 on Fri, 10 Jan 2014 11:08:42 AM


All use subject to JSTOR Terms and Conditions
1272 American Anthropologist [64, 1962
RECONSTRUCTION
TENTATIVE AND
OEPROTOPHONEMES
REFLEXES
PRINCIPAL

Proto P ((?) c 1 q
language p t c s k q x
b d z g p u i i
m w n r 1 X y ni 0 o a e

West p c 1 ?
Kamchadal p t c z k k(?) G
w - - z Y g u i i
m #/w n r 1 c #/y c e o a e

Chukchi [ r g v
p t c < k q #
w r R Y/W g u i i'
m w n r 1 c y N 1' a

Aleut [ I I #
h/hm t c s k q #/p
$ t/d c/s k/g #/p u/u a/A i/I
h/m #/m t/n ->t/1 -- #/y t/n 13 u a i

Greenland 1 1 i #
Inuk p t -- s k q #/p
p/w t/s s/s k/g #/p u/I I i/I
m #/w n-on/l #- #/y n/N 13 u a i

SUMMARY OF PRINCIPAL REGULAR INTERCHANGES


West Kamchadal: Chukchi:
p, 15/w/# y/# u/o/# p/m r/n Y/#
t/c Y= y/w/g/c/# i/e/# t/n R=r/t Y=y/g/#
k, l~/g w/# i/e/# k/r 1/n w/#
c, c/z w= w/g/# g/? q/? N=n/y w=w/g/#
G=g/#
Aleut: Greenlandic:
t/h w/# u=u/# p/w/m w/p/m/# N=n/y
k/g y/# I=-i/# t/s/n n/t s/t/y/#
q/P d/n A--a/# k/g/13 g/k/1/# s/t
q/P P/q/# I= i/a/#

language evidently has considerable positional and some free variation; the
sound or sounds written g, x, and h may be one phoneme, and similarly f/w,
z/s/z/s, palatalized and nonpalatalized dentals (perhaps an intermediate
type, identifid by the early Slavic-speaking phoneticians at times with one
and at times with another of their contrasting types). It is not clear whether
k and q are one phoneme, variously written for lack of adequate criteria, or
two of them frequently confused. For practical purposes it is easiest to assume
one entity.

This content downloaded from 200.54.94.206 on Fri, 10 Jan 2014 11:08:42 AM


All use subject to JSTOR Terms and Conditions
SWADESH] Linguistic Relations Across Bering Strait 1273
Several points regarding phonemic symbols may
require mention; y is the
palatal semiconsonant; like English ch;
c, X fi are palatalized 1 n respectively;
i is a central or back unrounded vowel; q p x represent velar sounds, respec-
tively the voiceless stop, the voiced stop or weak
fricative, the strong voiceless
fricative.
The variant developments indicated
by slant lines, in the case of the con-
sonants, generally depends on initial versus noninitial position.
However, for
*gto Y/W,it also depends on the
vowel: in stem initial, Y is regular; in medial,
it is w instead if the basic form has u in contact with the consonant (even
though the vowel itself is lost). Other variations, connected with
in medial or final, or due to contact occurrence
phonemes, are implicit in the rules of
regularinterchanges. The phoneme
*tj apparently did not occur initially in
thecommon period.
Forthe vowels, the main conditions of
varying development are related
to thesyllable of occurrence and the nature of the
vowels. The latter fall into
a weak and a strong set, consisting respectively of the high and the low
categories.In Eskimo, original *i and *uare reduced to I in the second
ofthe word; perhaps also in the first syllable if the syllable
second vowel is strong. In
Aleut,*u ii are lost in the first syllable if the
vowelof the second syllable is
strong,but are preserved as u a i subject to positional loss) if the latter
isweak; they u (not
give A I in the second syllable in all instances. In Chukotan,
anyvowel of the second syllable is lost or reduced to unless it
i falls in the
pre-paradigmaticposition; that is, the reduction takes place in the
compound-
ingform and in some derivatives treated in similar fashion.
haveundergone a reduction of *u and *i Kamchadal may
to in the first syllable also; since i
i
issubject to loss with fewer restrictions than in
clustersare both extensive and numerous in Chukchi-Koryak, consonant
Anotable feature of the Kamchadal.
old consonant system is thevery limited number
of
instances of *p
glottalized (reminiscentof the rarity of *b in Indo-European)
andthe apparent absence of *f.
ROOT
FORM
AND THE CHANCE FACTOR
Various facts lead to the
conclusion that the normal root form in proto
Chukoto-Eskaleutianwas consonant-vowel-consonant-vowel, which may be
symbolicallysummarized as CVCV.
Vowel-initial forms in the derivative lan-
guages,except for some formations with prefixed
vowelon the old level, are due
tothe loss of one or more of the old consonants. Longer forms are old com-
pounds and derivatives. In many cases the added
morphemes can be identified
as
affixes still existent
in one or more of the present-day languages. The basic
formsof morphological elements,
including demonstratives, personal pro-
nouns,inflectional signs, etc., was
CV. Occasionally, new roots were
formed by the combination of a demonstrative with suffix.CVCV This general con-
ceptionof the shape of the morphemes is useful a
a guide in the search for cog-
nate
forms in the two stocks and also provides basis
a for estimating the
chancefactor.

This content downloaded from 200.54.94.206 on Fri, 10 Jan 2014 11:08:42 AM


All use subject to JSTOR Terms and Conditions
1274 American Anthropologist [64, 1962
We have reconstructed 24 consonants capable of occurring in initial posi-
tion, 25 medial, and 6 vowels. If all the possibilities could be clearly distin-
guished on both sides of the comparison and if all consonants and all vowels
were equally frequent, we would calculate the chance factor for any particular
combination as 1 in 24X6X25X6, or 1 in 21,600. However, in the real situa-
tion, the second vowel is often coalesced or contracted with a following ele-
ment, and even the first vowel may be obscured in Aleut and Kamchadal.
There are various other instances of phonemes which fall together in one or the
other stock or which cannot be defined with certainty because we lack forms
from some of the languages. Unequal frequency (for example the high fre-
quency of k) has an affect similar to cutting down the total number of pho-
nemes. To allow with a generous margin for all these factors, we may figure the
chance factor on the basis of 10X3X10, making it 1 in 300. The corpus of
comparison consists of less than 2,000 distinct root forms from each stock. We
should therefore expect as the product of pure chance a total of seven cognate
sets with like meanings. If we allow for a reasonable amount of semantic varia-
tions, our chance factor is increased, possibly by about three, thus calling for
about 21 apparent cognates to equal the estimated chance factor. To safely
clear the hazard, we could take three times this figure.
In addition to the over-all problem of chance, it is desirable to present evi-
dence for all the phonemes. We propose to give several examples of each con-
sonantal equation in initial position, omitting only *niand *X,whose reality can
be shown without question by the diminutive interchanges in Chukotan but
cannot be separated from *n and *1 in Eskaleutian initial position. We give
about 8 examples of each of the high frequency phonemes, 2 or 3 for the most
difficult low-frequency sounds, and 5 in other instances; the examples simul-
taneously illustrate the noninitial phonemes. The total number of roots shown
comes to over 120. There are separate examples of many oblique correspond-
ences and of structural elements.

INITIAL-PHONEMEEQUATIONS
The languages are shown by abbreviations, made up of one letter for
Eskimo, Inuk, Yuk, Aleut, and two letters for Kamchadal, Kerek, Koryak,
Chukchi; E. (Eskimo) means both I. and Y. Dialect or free variants are
ordinarily listed without special indications. Normalized phonemic writing is
used in all instances to the extent that the source material permits it. Morpho-
phonemic symbols (consisting of small caps where two entities have to be dis-
tinguished) are used, when the data is available, in the parts of the cited ele-
ments where interchange is possible. Longer forms are broken with hyphens
when the separate elements can be identified. At the head of each set of forms
are given the reconstructed phoneme and the phonologic equation.
(*p-). Ka-Ch. p = A. h = E. p.
Ka-Ch. pamya- stockings; A. hamga- sleeve. From *pami-ga. Original
meaning of *pami perhaps extending member, found also in Ch. pamyayp-

This content downloaded from 200.54.94.206 on Fri, 10 Jan 2014 11:08:42 AM


All use subject to JSTOR Terms and Conditions
SWADESH] Linguistic Relations Across Bering Strait 1275
goes to sleep; E. pamiu-, pamyu- tail, drum
limb pawlu handle, pawlia- shovel
handle, A. hamyu-ga- hare (tail-having).
Ch. paa- ceases; A.
hapa-ta- rests, remains. From *paxa.
Ch. pue- soot; A. hwa- soot. From *puwa.
Ka. scrapes; E. pilak-, pilaa- cuts with knife. From *pila.
Ch. ple-z- flat; I.
pilgi-r-
Ch. pa2-at-feels tired; pinpali- rib. From *pili-pa-ri.
A. hani-na-
difficult,hari-ta- harm, har-u-ga- pities.
From Ch.*pari.
pir-i- carries; A. hid-u-sa- carries away, hid-u-la- pulls
out. From
*pedi.
Ch. picik-ew-
-k-and I. -q- may be becomesaccustomed; I. pitq-u-siq custom. From *pici; Ch.
variant forms of the same suffix.
(*p-). Ka. p= Ch. p A. h= E. p.
p is a low-frequencyphoneme, and only the following two pos-
sibleGlottalized
direct cognates have been found. Additional oblique cognates are given
elsewhere.
Ka. pag-cuc boy; I. pag-nIq buck deer. From
Ka.pazin son, Ch. *pagi.
theKa. perhaps panwar- young deer; E. panik daughter. From *pani;
dissimilated from *pansin, with asuffix.
Ka-Ch. w = A. h=E. p.
(*b-).
Ch. A. hag-sa-la-
Ch.waY-division; divides. hag-s- comesapart. From *bagi.
crosses; I. paq- meets. From *baqi.
Ka.waqit- rows; A. hau-pi- tows, I. pauq-
wau-z-
TheI. doublet papuq- rows and A. rows, paut paddle. From *bago.
haqa-dgu-sip small paddle are based on
variantroots: *bapo and *baqa.
Ch. gets wet through; A. hgu- washes self. From
Ch.wiYunt-works; A. *bigo.
wan-e- hana-sa- achieves. From *bara.
(*w-). Ka. Ch. w = A-E.
#= #.
Ch. protects,winr-et- helps; A. ila-gu- helps. From *wera. Ch. winr-
wirir-
maybe from with transposition and assimilation of the nasal.
Ch.win- *wire winwip-
Ch. trail, follows trail; I. inik sled trail. From *wini.
waqo- sits; A. aq-adagun afterwards, Y.aqumga-, aqumI-sits, I. aqu
behind,aqumuk-moves backwards. From
*waqi (Aleut) and *waqi-wo (Ch.
andE.).
Ka.
an-za-goes out, angte- engenders;A. aniqdu- baby, small child,anii-da-
pup,E. ani- goes out, is born. From *wani, with initial *w if these forms are
related
Ka.with Ch. winu- outside.
uneg guest, ong- rests, has lodging; I. unik- remains. From
*wune-gi or
*yune-gi.
Ch. pierces
wiYant- through; A. ga-lgi- smoke hole, I. iga-lak window. From
*wiga.
Ka.oza-z saw, Ka. uzgumtu-, Ch. urw-et- cutswood; A. usa- divides. From
*woza.Note that *w is lost before rounded vowels in Chukchi as well
as in the
other
languages.
(*m-).Ka-Ch. m = A. h=E. m.
mil- leaf,
Ka. milkec- fishscales; E. mitqu-, mipqu-, milqu- feather. From
*mill.
Ch.
maneg- Ka. palm of hand; A. hani- lake, hanika-l(ig)-
lies still cloth,
floats, water, E. manzin is smooth,
on manig- manipa- terrain. From *mane.

This content downloaded from 200.54.94.206 on Fri, 10 Jan 2014 11:08:42 AM


All use subject to JSTOR Terms and Conditions
1276 American Anthropologist [64, 1962
Ch. mayo- hill; I. mayuq- climbs slope. From *mayo.
Ka. mik-z- sleeps; I. mik- bird lights. From *meki.
Ch. maglal- rider; A. hagu- load. From *magu.
Ka. mazamaz vernal; E. masa-, maca- sun. From *masa.
(*t-). Ka-Ch. t = A-E. t.
Ka. togla- rubs, togtona- prow strikes; A. tupa- strikes, I. tuppalak shot.
From *topa for both stocks, but *topi for the I. form.
Ka. tak-u- measures; E. takI- long. From *taki; the suffix in Ka. has the
meaning uses, and the combination, uses or takes the length.
Ka. tg-lo big, tall, old; Y. txI- hard, diffcult. From *tigi.
Ch. timk- clod, timkicin muscle, biceps; E. timI body. From *tiim.
Ka. tamk thumb, Ch. tam- tumor, tampa-liberal, A. tama-daga each, ordi-
nary, E. tama- all. From *tami, with a basic meaning of stout, plentiful.
Ch. talqiw- flashes; A. talu-lap transparent. E. tatqi-, tanqi- moon, month.
From *talu-qi.
Ka. ton-ze- is late; E. tunu- behind, nape, backfat. From *tono. Also note
Ka. tono arm, if from shoulder,nape.
Ka. tato- is afraid; I. tatawyi- is suddenly afraid. From *tata-wi.
(*f-). No examples found.
(*d-). Ka. z= Ch. r - A-E. t
Ka. zomklo- cuts out; I. tum-nit tatooing (of woman), tumik trace. From
*domi.
Ka. zin- (also zin-), Ch. ri0e- flies, Ka. zirewirnikin bird; E. ti0i- flies,
tirmiaq duck, ti0nilpautak sail. From *di1i, with variant *diri in Ka.
Ka. zitec (also ziz) needle, Ch. rit- horn, tooth; I. titiq- makes mark, titipaq-
writes, titipa-ut pen, pencil, etc. From *diti and derivative *ditiyed (and vari-
ant *dedi in Ka.). A. ta-di- is split may involve another variant of the same
root, namely *dita, which loses the first vowel and contracts both consonants
into one.
Possibly Ka. zato- dips; A. tatu- hole in ice, I. tatat-tuq is full. From
*dati-wo, but with some other suffix in the second Eskimo form.
Ch. rim- bites, holds in teeth, rimru- gnaws; E. tamuq- chews. From two
variants, *dimi and *dami-wo. Because this is a low-frequency phoneme in
initial position, we have included this instance of oblique correspondence. We
also add a few of the medial cases, with d in Aleut in this position: Ka. ziza-
zin flow of sea, A. sida- flood, from *seda; Ka. kuzgegu- takes, A. kudmaci-
catchesin net, from *kodi; Ch. ar-e- takes away, A. ad-u- debt,from *xadi; etc.
(*n-). Ka-Ch. n = A. t= E. n.
Ka. nak-c misery, evil; A. taki-pi-, taki-gta- unwell. From *nake.
Ka. niri-1 sea; A. tana- water, drinks, I. ni0u-yuq thick liquid. From *ni0i.
Ch. nig-a southwind; E. nigi-q southeast wind, east wind, north wind (dif-
ferent winds according to dialect). From *nigi. Compare also, with variant
vocalism, Ka. nik-kil wind, weather, Ch. nikeye- north wind, nikeyerqec- east.
Ka. nikene- waits; I. nikpaak- waits at seal hole. From *niki.
Ka. nani eye, nan-la- calculates; A. tani-shines, tani-gta-sip, lighted lamp,
nani- finds. From *nani.
(*1-). Ka-Ch. 1 = A. t=E. n.
Ka.lkar- stands up; I. nikuwik- is standing, nikuwyaq- is on tiptoe. From
*liko, with suffix-contraction of the second vowel in Ka.

This content downloaded from 200.54.94.206 on Fri, 10 Jan 2014 11:08:42 AM


All use subject to JSTOR Terms and Conditions
Linguistic Relations Across Bering Strait 1277
SWADESH]
lk-legin I. nig-lik-tuq cold. From
Ka. cold; *liki. Ch. l?ale- winter evidently
contains a variant
Ka. lagl-, root *liqa.
i-lk- looks; I.
nag-waq- finds. From *lagi, with tense-variant
*i-liki in Ka. Other variants on both sides are involved in: Ch. liY- (from
*ligi) known, Ch. l?-u- looks, finds, Ka. l?u eye,Ka. naki-infamiliar; I. nakatak
rifle
sight, nautsiq-
Ka. lere- looks; Y. (from *lagu) notices, A. *tapa- tastes, tests.
nalaaq- finds. From *nara.
Ch.
lir-let- takes revenge; I. ninaq- gets angry. From *nina.
-
(*r-). Ka-Ch. r A. t = E. n.
Ch.rinni- ties; I. nimi-,
nimi-ut tie, bandage.From *remi, with assimilation
of*m to n of the suffix in Ch.
Ch.ramk-
crowd;A. tami- ties in bundles. Initial uncertain: from *rame if
related to the
Ka. rac-e- preceding.
covers; E. nasa-,
naca-, hat, cape, covers self with hat or cape.
From *raca.
This protophoneme is of low frequency, and the above
stances so far identified among normal arethe only in-
roots. An additional example, *ra-
what,will be found among the structural elements.
Ka-Ch. c = A. c=E. s, c
(*c-). (dialectvariants).
Ka.cug- rain: A.
cpu-gi- washes, cpu- (also cgu-) sponge. From *cupo
(andKa.
*cugo).
cak-e- visits; saqiq- suddenly visible. From
Ch.cakiget man's I. is *caqi.
Ka.cak-i- gnaws;I. brother-in-law; sagqiaaq brother-in-law. From *caki-pi.
sak-u- scrapesI.skin, harpoon point. From *caki.
cg-i- holds; A. sak-ut
Ka. cpa- steals, thief. From
Ka.cgantel prepared cpa-pi- *cipa.
cin-lu skin;A.I. sig-liq small dry skin. From *cigi.
Ka. fountain-like; cin-lap diarrhea. From *cin.
(*c-). c= Ch. r = A. c= E. s, c (dial.
Ka.. variants).
Ka.ca-k, Ch. A.
From *ca, or possibly ri-ra-q four; ca-rj five, literally myhand, A. ca- hand.
in all three *caqi, if we may assume that the second stem-consonant
has languages been interpreted as grammatical
thatthe hand a ending. Note
has five fingers counting the thumb but only four if the thumb
is
omitted.
Ka.co-k, Ch
ni-ro-qthree; A. points with finger,
index finger, possibly also Y. cuxxa-,cuga-pu- cuxxi- beak cuga-pu-sip
I. sigguk, siguk).
Chukotan forms have been trimmed to match four. (in The
be or The protoform may then
and the Y.for
*coga *cogi,
of
the meanings would appear tobeak may be from *cogi-ya/cogi-ye. The source
be point, applied either to the third
because it projects more than the others, or to the finger,
index, because it is used in
thepointing
No gesture. of
other instances proto *c
have asyet been identified.
Ka. r
(*z-). z=Ch. = A. c= E. s, c (dial. variants).
Ch.
runm-
Ch.rag-l-breaks,upperarm and shoulder; A. cunu-lag nape. From *zono.
A. cagi-
Ch. scratches,bursts; digs; A. cutsup,tears apart, splits. From *zagi.
Ch. rig-
re-t road; A. caga- ditch, trench, hole. From *zigi.
footprint. From *ziye.
ci-mip
zitz, zitta,Ch. ricit belt;A. cacpu- belt,girdle. From *zici for the second
Ka.
form and the A., variant in
Ka. *zici the other Chukotan forms.

This content downloaded from 200.54.94.206 on Fri, 10 Jan 2014 11:08:42 AM


All use subject to JSTOR Terms and Conditions
1278 American Anthropologist [64, 1962
(*s-). Ka. z= Ch. c = A-E. s.
Ka. zlolma- fires; A. sulu- thunder. From *sulo and *solu.
Ka. zal- behind; I. saliaq- follows aside, salpusi- passes without stopping.
From *sali.
Ka. ziza- flow of sea; A. sida- flood, I. sipqa- is on the move, skates. From
*seda.
Ka. sit- gets well, Ch. citewn it is well; A. sit-gu-pi- pardons, I. sitiksi- hard.
From *seti.
Ka. zutu- hand, zuti- manages, zu- left; A. su- takes. From *so, perhaps
reduced from a bisyllabic *suwo; note E. sau-mik left, from *sawu.
(*y-). Ka. #= Ch. y = A-E. #.
Ch. yut- protects; A. ut-gu-mayu- hides. From *yoti.
Ch. yuw- wind, yuet- wind blows; A. um-ta-breathes in, A. um-pi- lung,
Y. uwag-nap-tuq northwestwind. From *yowl.
Ch. y7u- reaches; I. upnik- comes close. From *yuqu.
Ka. ug, u? tree; I. up-pik tree, Y. uqfigaq willow, shrub. From *yopi and
*yoqi.
Ka. aka-no- hearth, Ch. yayaq-at- heats red hot; I. aqa-gu morning. From
*yaqa.
Ka. upg log, Ch. yup- pokes in; A. um-ta- throttles,squeezes, I. upsiq- stops
up. From *yopi.
Ka. ip-t-kalan bar for pushing door; E. ipuk- uses lever, rows, I. ipyutak
lever,E. ip-ut, ip-un oar. From *yepi.
(*k-). Ka-Ch. k=A-E. k.
Ka. kuk-e- cooks, Ch. kettle; Y. kukuk- kindles. From *koko.
Ka. kak-cel- fights; I. kaki- pierces, kak sharp, point. From *kaki.
Ch. kamagra- cuts out, breaks up; I. kamupayaq molar. From *kami.
Ka. kami-1 head; A. kamgi- head. From *kami.
Ka. kle- rubs, Ch. kil-kwin whetstone; E. kilik- scrapes. From *kile.
Ch. kane- closes net, ka0et- bends; Y. karigaq corner in house. From *kaii.
Ka. kalet- arrives, Ch. kal-t bottom; I. kalik- touches. From *kali.
Ka. kiw-gu-ncic river source; A. kimni- flows into, A. kimi-, I. kiwi- sub-
merges.From *kebe.
(*k). Ka. k=Ch. k A =E. k.
Ka. kgu- nail, claw; Y. xu- arrow, shot. From *kipo, with regular Yuk loss
of k when it comes into initial cluster with *p, unvoicing the latter. The Yuk
form is not directly cognate to Inuk qaqyuk, of the same meaning, whose Ka.
cognate is kag arrow, kagak- wounds.
Ka. kuk-u- chin (also l[ogkat); I. kukik fingernails, kukili-ut toothpick.
From *koke.
Ka. kan-kaz winter; I. kana0nak north wind. From *kana.
Ka. kaz-i-nen front; A. kada-mi in front, kadamga- meets. From *lkada.
(*g-). Ka-Ch. Y = A-E. k.
Ch. yiwu- tooth; A. kig-u-sip, Y. x-un tooth, E. kIgI-, kii- bites. From *gigu.
Ch. yin- fire; E. kiniq- dry, Y. kinix-c-un fire drill, kinix-pilaq lightning.
From *gini.
Ch. yaywe-1 orphan, yaywac- feels pity; Y. kaynuuyuk timid, I. kayigaq
hunger, Y. kaik-, I. kaak- goes hungry, I. kayai- abandons. From *gayi.

This content downloaded from 200.54.94.206 on Fri, 10 Jan 2014 11:08:42 AM


All use subject to JSTOR Terms and Conditions
SWADESH] Linguistic Relations Across Bering Strait 1279
(*p-). Ka-Ch. g = A-E. #.
Ka. gike-leg,
gka-lag hot; A ig-sga-, I. ikI- burns. I. ig-nIq, i~-nIq spark,
flint. From *piki.
Ka. gag tail; I. aqqu hind part. From
Ka. gan paddle; A. *papi.
Ch. gink- fat; A. an-gu- sticks in, anulya-n stakes. From *panu.
fat, robust. From
Ch. ana-tup *pini.
Ka. gamga- I.
gcenmineach; I. amisut isuk many. From *pami.
end, isuuyuk
Ch. iliac; shoulder articulation. *pici.
gitin- pretty; A. atigusan (also acipusan)straight. From *pite.
Ka. g=Ch. A-E. #.
(*x-). #
weather, Ch.
Ka. galc ala- summer; A. alagadigap pale. From *xala.
Ch.anqa- sea; A. anu- tide, current. From
*xanu.
Ka.galk-cuc old man, Ch. alii3 very;A. ali- old man,
alpu-nap glory, emi-
nence.
Ka.From twig,*xali. Ch.
giz- irg- sharp; A. plant. From *xida.
Ch.ragnatco- begs; A. asks aida-sa-pi-
favor. From *xira-pu.
lapu-
(*q-). Ka-Ch. ? - A-E. #.
Ka.?i water; A.
Ch. tal-aw-ties; A. iga-da-, iga-ta- flows. From *qiga.
Ch. day, ali- harpoonline. From *qali.
?iliw tilwi-rpassage of day; I. ilik- chars.From *qili.
Ch. tii-, nose; I. ilpi- mountain. From *qiii.
Ch. tinirj
splinters, sawdust; I.
Ka. tilerw- ilap-nmqshavings, remains. From *qila.
tel split, crack,ditch, Ch. talgatga- cutswith ax; I. alik- tears. From
*qali.
Ch. urine; I. urine. From
tittit gives itipu-
Ch. taywat- *qiti.
charity, teyweey charity; I. ai-tuq makes gift, aimiq-
visitsexpectingfood. From *qaye.
(*q-). Ka. k=Ch. q = A-E. q.
Ka. Ch. deep; A.
Ch. kimen-lu, qimca- qam-da- deep. From *qimi.
qirgat- shines, qirg- light; A. qila-, qilaga-n morning. From *qira-ga.
Alsocf. Ka. klgi-1 day,
Ka. arrow; klninin dawn; from variant *qili.
From
kalga
kipilg- Y. qalqa-paq ax. *qalipa.
I.
Ka.
Ka. covers,extends;
sharp; A.
qipik
quilt.
From *qepe.
kicg-lu sharp. From *qeca.
Ch.
qitli- qica-pi-
lazy, qitli-w- grown old; A. qada- tiredness. From *qidi.
Ka.
kwa-1, kow stone; E. qua-q frozen meat. From
*quwa.
-- A-E. an.
(*xiu-). Ka-Ch.(i)u
From the many elements beginning with in
thatthis consonant also occurred rj Chukotan, one might guess
initially in the proto language, but cor-
respondences in show preceded by a. We conclude that i
between initial *xEskaleutian
and *rjbecame a in rj
otherwise Eskaleutian early enough toescape the
operating reduction.
Ch. male Kamchadal loses *x in these cases. Examples:
raraq-aw- is industrious, satisfies sex desires, K a. r-im-zg, Ch.
rj-aw-wife, woman (presumably from has husband); E.
ar-u- gets game, arj-ut,
a~-un man.From *xiri.
Ch.
Ch.inw aw much, extremely; A. anu, E. From *xinu.
~uyr- rump, tail,~ira- inil) tail, ~ut-behind, slow;ani-big.
Y. artupi- last,
a utmost,
end, edge. From *xiru. InCh. irjir, the isi not lost because
~taa was fol-
lowed by consonant cluster in the reduplicated it
*xi3xir.

This content downloaded from 200.54.94.206 on Fri, 10 Jan 2014 11:08:42 AM


All use subject to JSTOR Terms and Conditions
1280 A merican Anthropologist [64, 1962
Ch. rane- inner space; A. ara-la- penetrates, enters. From *xira.
Ch. riwu- begins; A. ani- begins, sets out. From *xini, the Ch. with a suffix
-gu.
Ch. ra-cg- left side, 1a-rg- outside, ear- breaksoff on the sides; A. arjaa
half,
aga-maadan on the side. From *xina.
PHONEMICINTERCHANGE
Three lines of evidence lead to the conclusion that phonemic interchange
much more extensive than that found in contemporary Chukotan must have
existed in proto Chukoto-Eskaleutian: (a) doublets, with the same or nearly
the same meaning, found in individual languages or stocks; (b) oblique con-
cordances between the stocks; (c) large sets of forms in both stocks, with simi-
lar forms and meanings, which add up to fragmentary paradigms. A final test
of the conclusion would be a study aimed at determining the meaning of each
alternating consonant and vowel, as in the author's paper on Altaic doublets
(1962), but this remains for the future. We give here the formal outlines of the
system, with only broad suggestions about the possible meanings, and sam-
plings of the evidence.
In the old system all the vowels were evidently in alternation with each
other. As in modern Kamchadal, the different vowels may have in part ex-
pressed verbal voice and nominal case, but apparently also something else,
perhaps the shape of objects referred to or involved in actions. This recalls the
Altaian situation in which the rounded vowels are related to round and cylin-
dricalobjects, the low vowel (a) to broad and flat, the front unrounded to thin
and sharp. However, the shape classification may have been already vestigial
inthe terminal common period:
Among the consonants the three oral obstruents of each position were in
interchange, for example, k with 1k with g. In addition, the velar was inter-
connected with back palatal (that is, the k-set and the q-set), and the pure
dental with the sibilant (the t-set with the c-set). Again, the meanings may
have been roughly comparable with the Altaian doublets, marking contrasts
between "light" and "heavy" and "bright" and "dark." The diminutive
consonantism, still used in Chukotan, may be a semantically and formally
narrowedvestige of the old "light" category. It may be significant that the
commonest words for childin Kamchadal, pec in singular, nenelkecin plural,
contain glottalized occlusives; and theexamples of p are confined to this and
oneother element of similar meaning. This suggests that diminutive may have
beenone of the values of glottalization. The rarity (or absence) of p and fare
perhapsdue to avery old coalescence of these consonants with some other type
(possiblyb and d), and the subsequent restoration of p in certain diminutives.
Thesamplings of evidence for old phonemic alternations begins with a brief
listingof typical doublets on the level of language, family, or stock, with recon-
structionsin ancient terms:
Ch. ruy0- rump, tail, agoygik-emcacokalg- weasel (tailed-ermine). From
*xirjoyirji/xarjoyin.The weasel is presumably described thus because of his
heftiertail.

This content downloaded from 200.54.94.206 on Fri, 10 Jan 2014 11:08:42 AM


All use subject to JSTOR Terms and Conditions
Linguistic Relations Across Bering Strait
SWADESH] 1281
Ka. Ch. middle. From
kinnin, ginun *qenu/pinu.
Ka. alo-, nu-, Ch. iniw- eats, Ch. row-lg- food. From prefix *a/i/i and stem
Note also A. inu-
*liwi/niwi/rowi.
Ch.ri-let- puts into mouth, from *i-niw.
flames, ana star. From *xiri/xara. Also cf. A. ara-lip daylight,
I. agya star, from *xina/xiL.
Ka. nuku-lu dark, Ch. niki- nocturnal. From *noko/nike. Also cf. E. unuk
night,
Ka.evening, from
pec, periyan*o-noki. pen children, pen-k infantile, panagca boy, Ch.
-pi diminutive, i-plu- child, small, i-pinmi close, Ch. panwar- young deer.
Ka. c From
a contraction of *cc from *nc,
wasdropped in inflexion as thoughpresumably
*pani/pani/pana/peli/pili;
hla- son, and E. panik daughter, it were the absolute ending. Note also A.
child, from *pila/pani.
Ka. ilgce-
believes, inekecul sincere, Ch. yirul- learns. From *yelo/yena/
yero. Also cf. A. lu-nap certain,true, faithful; from *yela/yilo.
Ka. kiwguncicila- truly, Ch. ip-aw- drinks. From
A.kimni- river source, *kebi/xipi. Also cf.
flowsinto, I. ip-tiq juice, iwyiq rain, kiwi- submerges,amna-pi- river
Ka.from *kebi/xabi.
source;
skin, zileg,
Ka. zru-n zilgfishskin. From *siro/sela.
kuk-e-, uwi-/oye-/cooks, heats. From *koki/xugi/xoge. Also cf. A.
ug-su-warms self at fire, ug-sig,
uu-t- A. ig-sga-, I. iki- ug-nap withered, I. uu-nap-si- warms self, Y.
cooks; burns. From *xogi/xogo/xeki.
A.qamti-
brow,imli- hair, I. qimipiak eyelashes. From *qami/xemi/
qeme.A.kig-u-, E. kIgI- bites, I. qaqyuk, Y.
A. xuq arrow. From *gigi/qapi/gipo.
cna- animal hair, I. sagga- short hair. From
I.mitiq-, misiq-, micik- jumps. From *cira/cari-ga.
tweent ands in Inuk Y. can be explained *mete/mece. The variation be-
as dialect
enceof the preceding front vowel, but the Yuk variation under the influ-
Ch.naqam no, cannot have this cause.
-nk-at refuses. From *naqa/nki. Cf. also A. -laka, Y. -np-i-t-
not,I. nakka, nagga no; from
A.cmalu-, camluuka- *laka/npi/naki/nagi.
chin. From *cimalo/camilo. Cf. also Ch. tamlo-
wal-gipu- whale jaw, from *tamilo.
A.tana-, E. nuna land. From
fromKa-Ch. nute- *nana/nuna. Similar but perhaps distinct
land, which is better compared with I. nati- floor, and
reconstructed*nota/nati.
Therefollow some interesting instances of
phonetically oblique corre-
spondences
Ka. between the two stocks.
kig river, sea; I. kuug, kuuk, Y.kuik, kuweegit river. From
kugi. *kipi/kugu/
Ka. gike-leg, gka-lag, keka-lu hot; I.
uqqu- hot. From
xuqi. *piqa/piqa/qiqa/
Ka. aka-no- hearth; I. kig-luk outside
Ka.form is mis-recorded for initial or fireplace. From *qaqa/kigi, if the
Ch.
am ?, perhaps the Ka. comes from *a-kika.
0iroot-eight; A. qamcii0 eight. From *xam/qami.
Ka.gamtala- grinds bones; A. qimuga- grinding, I.
From kamupayaq molar.
*xamo/qemo/kamo.
Ch. goes I.
ritu- out; Ch.
gulg-, lurulur, iriapsi- goes beyond. From *xini/xene.
Ka. lilirj heart; E.uumat, uruuw anheart, uuma-,
unuuw an lives, A. is seated. From
Inuk umuci-
Eskimo is evidently a contraction *xolopo/xilono/xlini/xolno; the ofm
of *lrw, possibly based on
tion
of *lrj to or and subsequent combination of conctrac-
*rj *n *rw or *nw to m.
though theunderlying meaning may be live, sound-imitative aspects may Al-
affected the development on the Chukotan side. have

This content downloaded from 200.54.94.206 on Fri, 10 Jan 2014 11:08:42 AM


All use subject to JSTOR Terms and Conditions
1282 A merican Anthropologist [64, 1962
Ka. kira-ze- nostril, ka~a-z, ka~e-, kekaj, Ch. ?in- nose; E. gioa- nose,
qiru- root of nose. From *qerja/qarja/qetja/qeno.
Ka. gla-te- rubs; A. uli-pi- rubs. From *pila/pole.
Ch. gilgil sea ice; Y. qinu- slush ice. From *pelo/qeno.
Ka. gil- drinks; A. unu-pi. gives to drink. From *xili/xono.
Ch. anqa- sea; A. alapu- sea, salt. From *xana-qa/xala-po.
Ch. kalt- bottom;A. qala bottom.From *kala/qala.
Ch. gintew- escapes; I. anak- escapes. From *pini/pana.
Ch. gipi-lin worker, servant; I. kiwgak servant. From *pipi/gipi.
Ch. gitka- leg; E. itiga-, A. kita- foot. From *piti/piti/geta.
Ka. giti-1 skin; A. qacpi- skin, meat, I. qisik skin of sea mammal. From
*xiti/qaci/qici.
Ch. yit- drips; A. icapi- drips. From *geti/peca.
Ch. Pure- straight, open; A. qudu- straight, narrow. From *qode/qodo.
Ch. umr- strong; A. am-nip strength,solidity. From *xomi/xami.
Ch. wiYal snowstorm; A. igadu- snowstorm. From *wiga/wega.
Ka. cakic changed; I. suqusiq- changes. From *caqe/coqo.
Ka. tgime- kidney; A. dagtu-, E. taqtu, taxtu kidney. From *tige/i-dagi/
daqi.
Ch. timr-iw- gets lost; E. tamaq- loses. From *timi/tama.
Ka. tinu- descendant; Y. tangu-pa- boy. From *tino/tanu.
Ka. cne-zi- whets; E. sili-, cti- whets. From *cine/ceXe.
Ka. cpa-1 wind; supI-, cupI- blows. From *cupa/copi.
Ka. zit-lg coal; A. suda-nap graphite. From *seti/soda.
Ka. zuna left; I. saumik left. From *siwi/sawo.
Ka. lkar- stands up; I. nikuwik- stands. From *lika/niko.
Ka. lane girl; I. nirimnap-tuq small girl. From *lai)e/lere.
Ch. lawt- head, hair; E. nuyaq head hair. From *lawi/liwu.
Ka. pina hot springs; A. hangu- hot springs. From *pina/pani.
Ch. pirjku- jumps; I. paralik- gallops. From *perja/pal3a.

If the consonantic and vocalic alternations were ever simultaneously active


at some old stage of Chukoto-Eskaleutian, there would then have existed a
formidable total paradigm for many of the elements. A root such as *tiki must
have admitted six variants for the first consonant, *t/f/d/c/c/z; seven for the
second consonant, *k/l~/g/q/q/p/x, and six for each of the vowels. The total
number of possibilities is 1,512. There is reason to suspect that there were
norms of concordance or contrast between the consonants and vowels of both
syllables, so that the actual number of possibilities employed was much less
than the mathematical product. The limiting rules may have varied from dia-
lect to dialect.
Turning to the actual material of comparison, there are various instances
in which 20 or more presumptive variants of a root have been found. We cite
an archimorph for which about 50 possible variants have been found, namely
*tiki, whose primitive sense may have been claw, but which gave rise to mean-
ingslike finger, hand, touch, grasp, take, scratch, cut,reachfor, askfor, point, foot,
nose, etc.:
*tiki: I. tikit- arrives, tikiq- points, tik-pi- canoe rib, A. tik-lap long finger,
tigla- eagle.
*teko: I. tikuq- is straight, tikuaq- points, aims, tikuut arrow.

This content downloaded from 200.54.94.206 on Fri, 10 Jan 2014 11:08:42 AM


All use subject to JSTOR Terms and Conditions
Linguistic Relations Across Bering Strait 1283
SWADESH]
*teka: Ka. tekte-, teka- bone.
*toke: I. tukiq- Ch. tukke- harpoonpoint.
kicks,
*toka: Ka. toka-1- bow(shoot-instrument).
*tiki: Ch. -tk- point, upper part.
*i-tika: Ch. itke- takes away.
*a-tika:I.Ka. atkaw-outside
foot.
*taki: tak-pak of nose, nostril, Ka. takzi- pierces, takin- breaks,
tak-1-attacks.
tuq ice chisel.Ch.
*toqi:I.I.tig-u-
*tegi: siezes, tig-giri- desires, Ka. tigotina- pierces, strikes, tigcpi-
gnaws,I. tigluk- hits with fist.
*taga:I.A. taga- bird lights.
*togi: tugli-tuq coiffure stick, A. tugla- gropes.
*toga:A. tuga- strikes, whips.
*togo: I. tuuga- tusk, ivory, tuuq- works with chisel, tuugaut- punches.
*tipi: Ka. tig- strikes, Ch. tig-i- steals.
*tiga:Ka. tganmanon, tgilmat hip, buttock,tgemt breast.
*tapi:Ka. tagtu-cic canoe.
*topa: I. tugaq- aims.
*topi:tuppalak- is hit bybullet.
*a-tigo:A. atgu- finger.
*dipi:Ch. rig- digs, scratches.
*daka:Ka. zaka- penis.
*i-deka:A. dika-sip knife.
*a-diga:E- ayga, agsa-, aixa- finger.
*i-dogi:A. dugta- hook.
*ceko:Ch. cikuci- points.
*ceke:Ka. cikikc forehead.
*ceka:Ka. ceka-1 head, cek-z mouth.
*cike:Ka. cki- foot, leg, paw, forefin.
*caka:Ka. ckanafoot, leg.
*cake:I.Ka. caki-li-
*coki: sukit-tuq gnaws. sole; Ch. cuk-et- asks of host.
*coqe:A. cuqi- hill,wounds
mound, cuqi-da- root ends (diminutive of preceding).
*coqa:I. suqa- whalebone.
*cegi: I.sigit-tuq
*cagi:A. cagi- breakable,
to
I. sigguk beak.
cuts pieces; cagli- tailors, cagta- crack.
*caga:A. caga- ditch, pit.
A. cuga-pu- points,
*coga: cuga-pu-sip index finger.
Y. cuxxa- beak.
*cogi:
*cepi:I.sip-quq- shoots, Ka. cig-cag quarrelsome.
Ka. cigi-n hump, cgi- holds, cgecg mouth.
*cipe:
*ciga: Ka.cga- penis, cgap, zgap tentacle.
*capi:I.sap-pik whale tail.
Ka. caga- meets.
*capa:
*copa: I.supaq- breaks.
Ka.
*caqi: ca-k, Ch.-ra-q four, A. ca- (my hand), A. ca- hand.
five
Ka. co-k, Ch. -ro-q three.
*coqi:
STRUCTURALELEMENTS
keeping with the other evidence already cited for common
In
comparisonof inflections and other essential grammatical origin, the
apparatus in the two
stocks
reveals many evidently cognate elements,
involving sometimes direct

This content downloaded from 200.54.94.206 on Fri, 10 Jan 2014 11:08:42 AM


All use subject to JSTOR Terms and Conditions
1284 AmericanAnthropologist [64, 1962
andsometimes oblique phonological correspondences. Ina number of instances,
thestudy has brought to light fossilized elements in one or the other stock, and
hasclarified details of the probable structure of the common period. What fol-
lowsin not intended to bean exhaustive survey of agreements, but mainly to
roundout the case for relationship.
Theplural in Chukchi is commonly marked by-t(i), evidently cognate with
-t in Eskimo. In the pronouns of the several Chukotan languages, one finds
endingslike Ch. -ri, Ko. -yu, y, Ka. -za, which must derive from *-di/di/do,
presumably variants of *-ti. Aleut -n or -s, according to dialect, perhaps also
goesback to *-di. Thus, we seem to have one archimorphemeback of all these
plural endings in both stocks. However, there are other signs used in the
differentlanguages which do not correspond.
The k-plural of two Chukotan pronominal elements, -mik we and -tik ye
may be in origin not plural but dual, corresponding with A. -g/-k- and E. -k
dual. The protoform would be *-ki. Interestingly the complex -tik is found in
the Eskimo transitive paradigm for both dual and plural of second person.
Chukotan -ti-k parallels A. -di-g in tgi-di-g ye both.These correspondences
supportthe idea that the dual is an old category of Chukoto-Eskaleutian.
A. -rin they, subject suffix of verbs, seems to be somehow related to the in-
dependent pronominal forms Ia-n to him and rjii-nto them, whose final -n is
evidently not the plural but an old case ending (see below). Plurality is some-
how carried by the long vowel -ii-. It may be compared with -i, the element
that pluralizes the third person possessor in E. -ai their, and which may go back
to *-yi. The etymology of rii-n may then be *ni-yi-di, consisting of pronoun-
plural-dative. The verb ending-nin they perhaps has a somewhat different
origin, *irji-di,the -n representing the common plural ending. The ending -tgin
ye is obviously related to the independent pronouns tgi-n thou, tgi-di-g ye two,
tgi-ci ye. To explain why it apparently has -n of the singular second person
rather than -ci of the plural, perhaps there is involved an old *-t or *d plural at
the beginning, contracted with the initial consonant of the pronoun.
The common substantive plural of Koryak -w(i) is perhaps also of pro-
nominal origin. It could be a contraction of *wi-yi with the plural suffix *-yi.
Similarly, the -n plural of Kamchadal may be compared to the pronoun
*na/ini, presumably with some original but now lost plural, as perhaps *iny.
Noun and adjective suffixes and the absolute case have to be treated
together in Chukoto-Eskaleutian, because apparently there has been consid-
erable interchange of functions. This is not strange, since the absolute is in
effect a noun-marker whose use is confined to the singular when it lacks a
specific case or pronominal ending. Both stocks use a variety of these elements,
both simple and complex in origin, but in each of them and in the component
families or individual languages there are preferred forms; that is, certain
items have apparently taken over a very large proportion of total use. Com-
paring the suffixes used in the two stocks one finds much agreement in the total
inventory along with notable differences in the specific function and the fre-
quency of employment of each element.

This content downloaded from 200.54.94.206 on Fri, 10 Jan 2014 11:08:42 AM


All use subject to JSTOR Terms and Conditions
SWADESH] Linguistic Relations Across Bering Strait 1285
The most widely used archimorphs are evidently *-ni and *-ki, which, alone
or in combination with each other, are found in such occurrences as the follow-
ing'
*-ni: Ch-Ko. -n absolute; E. -n, -t in -un, -ut instrument, formed on -u- uses
(see below). The medial form in Y., -utI-, may indicate that a cluster
*-nt- is involved; this could equally well have given rise to Ch. -n.
*n-qi: Ch-Ko. -g absolute; A. -na-p, E. -nI-q general noun marker with abso-
lute. The Ch-Ko. from a reduced form without vowels, *-nq by assimila-
tion of the nasal to the stop and loss of the latter.
*-li: Ka. -1 agentive, Ch. means of, -li-n participial; E. -li-k- having.
*-le: Ch-Ko. -li-n adjectival; Y. -li adjectival for color terms, I. -li- produces,
-li-q- begins to, -lipi- is busy with, -li-q thing in given location, A. -li-p
noun marker (e.g. qam-li-p stone knife, qar-li-p shoulder.
*-lo: Ka. -lu adjectival; A. -lu-p place or instrument, I. -lu- subject participial,
I. -lu-p- having bad, also occasional noun marker (e.g. qit-lu-q carcass), -luk
marker of body-part and other nouns (e.g. quw-luk thumb).
*-li-pi: Ch-Ko. -lgi-n at times noun marker, at times merely absolute, Ka. -lg
adjectival; Siberia Yuk -lpi adjective ending, Y. -lp-ia vehicle, instrument.
*-ri-pi: Ch-Ko. -rgi-n noun marker or absolute, similar in use to -lgi-n, each
being associated with given nouns.
*-li-qi: Ch. -lqi-1, Ko. -lq(il) pertaining to.
*-li-ge: A. -lgi- employs, obtains.
*-li-go: A. -lgu-p great, augmentative noun marker.
*-re-li: Ch. -ril- equipment.
*-la-pi: Ka. -lag adjectival and noun marker (like -lg); A. -la-p occasional noun
or adjective marker (e.g. qig-la-p coral, cucpu-la-p prickly).
*-li-ga: A. -lga participial.
*-qi: Ka. -k, Ch. -q(i) adverbializer; A. -p, E. -q noun and adjective absolute,
third person intransitive of verb.
*-qe: Ka.-ki-n, Ch. -qi-n adjectival.
*-qa: I. -qa-q- has.
*-gi: A. -gi- has.
*-gi: Ka. -y noun absolute; E. -k absolute or noun-marker.
The common absolute suffix of Kamchadal is -c, a form without close paral-
lels within Chukotan. It may be an oblique cognate to the *-ti found in Yuk
-(s)tI agent, -ta(q) thing, and apparently also in -u-n/-utI- instrument, from
*-o-ndi, and a direct cognate to -ca-p, found in a few Aleut substantives, as
aca-cap companion, cig-cap hole, qag-cap dark, gloomy, black.
Noun formative are often based on verbalizers with added noun-markers or
absolutes. This mode of derivation is probably involved in a few of the com-
plexes listed above, where old verbal elements cannot be distinguished from
the noun or adjective formatives. We list below what appear to be primarily
verbalizers:
*-la-: Ch. -la-t- repetitive or intensive; A. -la continuing, varied, I. -ku-la-
frequently.
*-ro-: Ch. -ru- inceptive; perhaps I. -lug- does a little.
*-se-: Ka. -z- verbalizer, Ch. -i-cgi-n equipment, instrument; A. -si-p instru-
ment; I. -si- gets, meets.
*-wo-: Ka-Ch. -u-, -w- handles, uses; A-E. -u- does, employs, E. -u-t, -u-n/
-u-tI- instrument.

This content downloaded from 200.54.94.206 on Fri, 10 Jan 2014 11:08:42 AM


All use subject to JSTOR Terms and Conditions
1286 A merican Anthropologist [64, 1962
*-ye-: Ka-Ch. -i- frequent verb final; A-E. -i- verb final, E. -yi agentive,
-i-q uses up.
*-ti-: Ka-Ch. -t- applies and transitive; E. -t- transitive, -ti-t- transitive
causative.
*-ti-: Ch. -tw- removes,undoes; E. -it- lacks, I. -00i-t- negative.
The two stocks evidently use the same negative element in different com-
binations.
The various case formations in both stocks appear to involve different per-
mutations of perhaps four original endings, mixed with new forms of adverbial
origin. Thus, we have:
*-mi/-ma/-mi: Ch. -m, Ch. -ma with (comitative); A. -m(i), E. -p, -m of
(superordinate).
*-ni/-ni-k-e/-gi-ni: Ka. -n-kfrom, -n-k-e to; A. -ga-n in, at, E. -ni in, at, -ni-k
by, Y.-ku-n via, from, I. -ku-t via.
*-ki: Ch-Ko. -k at, in; Siberian -wi-k toward (with locatives)
*-gi: Ch. -Y-pi, Ko. -Y-pi-0 by, Ko. -Y-ti-rj to; A. -ga-n in, at, to.
*-ta/-ti/-di: Ko. -ta with, Ch. -i-ti, Ko. -Y-ti-0 to; I. -ni-tfrom, -nu-t to, -tu-t
like, A. -gaa-n toward.
In Chukotan, the comitative has an alternate form expressed with the help
of a prefix, Ka. 1K-,Ch. ga-, Ko. ga(wun)-. These prefixes are not unlike
certain independent locative forms found both in Chukotan and in Eskaleu-
tian, and used with or without locative endings as adverbs, or with the appro-
priate inflections as nouns and verbs. Structurally parallel and quite similar in
meaning are Ko. gawun- with and A. kugan on, at, which can be reconstructed
respectively as *ka-gu-ni and *ku-ga-n. A similar meaning, together,is found in
E. in a stem kati/katI-, e.g. kati-ma- cometogether,kati-ti- unites, katuti- are
associated, kati-nIq point of meeting, confluence, katgi- gathering place. It is
notable that the first syllable of this stem may be traced to ka-, just as the
Chukotan elements of like meaning. An allophonemic variant occurs in Aleut,
asi- together,from *xaze.
The bulk of locative words in both stocks have some allophonemic variant
of *ki in the first syllable, usually accompanied by some second syllable. The
bisyllabic forms occurring in both stocks sometimes correspond, either directly
or obliquely, but often also differ in the second syllable. Some examples follow,
arrangedapproximately by similar meanings:
Ka. kata-1 behind, ic back, ata-ko-1late, itgiak down, utelman lower, ici- falls,
Ch. atyew- goes down, atcat- falls to one side, katinka- base of shrub,arat- falls;
A. atmu-da-gan down, idi-, utgali-falls, I. katak- drops, I. atqu- bottomrear of
boat, at- lower part, atsik- is below, Y. aci-, asi- below; from variants of *kiti.
Ch. yanka- down on ground, kalt- bottom; A. qala bottom; from *kill. Ka.
kinnin, Ch. ginun- middle, Ka. gun meanwhile, Ch. qala- interior; A. ka0i-n
inside, I. ka0i- interioror country. A. a0a-gan inside, nu-for, Y. nnI indoors;
from *ki0i and *kin. Ka. kazinen front; A. kadami in front, E. qaa-, qai0(a)-
top; from *kiti. Ka. i0kuni- rises, Ch. girgol- above; A. ka0a-n, I. ka0a-ni
above,on top; from *ki0i, perhaps reduced from the *ki0ki implied by Ka.
and ultimately related to the *kiniki implied by the Chukchi. A. ku- on,
kugi- upper, kaki- rises; from *ki and *kiki in Aleut only.

This content downloaded from 200.54.94.206 on Fri, 10 Jan 2014 11:08:42 AM


All use subject to JSTOR Terms and Conditions
Linguistic Relations Across Bering Strait
SWADESH] 1287
The list be extended considerably. The evidence
could suggests that bisyllabic
forms in part existed in common
Chukoto-Eskaleutian,
bic *ki, perhaps a demonstrative, was the base of but that a monosylla-
most of the formations. The
second element must have been originally a
directional element in its own
right. If it had two syllables, these were often reduced to
one, blurring the
differencebetween some of the formations.
Verbalinflection in all the
Chukotan languages and in Aleut on the Ameri-
can side may optionally be achieved
by the use of an auxiliary. One of the
auxiliariesis common to both stocks, A.
a-, Ch-Ko. wa-, traceable by regular
phonology to proto *wa-. The Eskimo indicative mode sign in its transitive
form-wa- is therefore probably a suffixed version
of the same thing. The in-
transitive-wu- is an ablaut variant. The future is
expressed in Ch-Ko. by pre-
fixingra-, ya-. This has the appearance of being an
oblique cognate of Ka.
suffixedfuture -a/o-l- and the E.
optative -la/i-, but the picture is complicated
bythe fact that the auxiliary stems in
Kamchadalare cgi- for present and li- for
past,that Ch-Ko. has aprefix g- to mark the past, and the
further fact that
varioustemporal elements, not directly
inflected, are used in Eskaleutian,in-
cluding:A. -duka-, E.-la-q- in the
future, I. -la-0a-, -la-si- is about to, A. -qa
already,-na- in the past, E.-sa-q- recently, -si-q- at this
behindtime, etc. Thesituation is more difficult to moment, -na-sa-q-
explain within either stock
separatelythan taking them together.
sistsin recognizing Thekey to the problem evidently con-
two classes of elements, distinct in origin but partly con-
fusedby their close association in use: verbal
particles and tense-aspect affixes.
Whatwe call present was probably expressed
by thesimply particle. Past or
future,however, were expressed by adding to the particle
one of various tem-
poralsuffixes. This gave originally bisyllabic
formations, which were subse-
quentlyreduced by contraction.
Forexample, Ch. ra- future may well come
from*ri-la reduced to
*rla and finally to its present form. If this analysis is
correct,we apparently have to deal with three
conjugated auxiliary stems,
*wi, *ni and *ti, each with its allophonemic variations. They all coincide with
andare probably derived from demonstrative
stems, in keeping with the fact
thatChukotan still inflects its demonstratives.
Asin other sets of structural elements there
are many parallels between
Chukotan and Eskaleutianin the department of
and pronominal, demonstrative,
interrogative-indefinite elements. We treat some of the more important
instances.
Theprincipal first person marker in
mma. Ko. Chukotan is -m-, asfound in Ka.ki-
gu-mo, Ch.gi-m I,Ka. mu-za, Ch. mu-ri, Ka. -mi0(k),
Ko, mu-yuwe,
Ch-Ko.-mikus.This element is evidently also present A.
in tu-ma-n, ti-ma-s
we.InEskimo there are various bits of morphological material which
is may re-
it. E.-ma my not clear
flect a cognate, since we expect an -m- of the super-
ordinate case in all pronominal possessives and hence
must assume here con- a
tractionof superordinate -m- with a personal suffix.
In the past, it has been
supposed that the hidden element is
but a -r-, asin the frequent first person ending
-na, lost -m- is more
a likely explanation because the usual direction of

This content downloaded from 200.54.94.206 on Fri, 10 Jan 2014 11:08:42 AM


All use subject to JSTOR Terms and Conditions
1288 A merican Anthropologist [64, 1962
assimilation in Eskimo is the preceding element to the following rather than
vice versa. I. -ma me, in transitive indicative (with second person subject), may
will be a direct cognate of the Chukotan forms and not an analogical transfer of
the possessive complex, as heretofore supposed. Likewise, Y. -mti-, I. -wti we in
transitive complexes is much better taken as cognate of the Chukotan, than an
obscurely transformed version of -wu-t.
The case of the second person is much simpler. Common forms of all the
languages are traceable to regular allophonemic variants of *ti, as: Ka. tu-, -t,
-c thou, -cg ye, Ch. gi-t, gi-r, -t I, -tik ye; A. -n, -s thou, -di-g ye two, -ci ye,
E. -t, -n thou, -ti-k ye two, -si, -ci ye.
There are of course other forms of the personal pronouns found in the total
inflectional apparatus of the several languages. A first person *n occurs in Ka.
n- we, Ko. ine- me, A. -n, E. -na I, my. To explain the back nasal in Eskaleutian
one must suppose contraction with a k-sound, that is, -na from *-nka from
*ni-ka with loss of the k by virtue of the accentually weak position of the
suffixed elements. Other variant personal forms appear to be due to contraction
of the typical elements with demonstratives. Thus, Ko. -gi thou is evidently a
reduction of *-giy from *-gi-d, equivalent to Aleutor -git and Ch. -gir. Similarly,
Y. -ka I may well be reduced from some combination like *kami and therefore
an oblique cognate of Ka. ki-mma, Ch. gim, etc. Similarly, E. -gut and -wut we
could be from *gumt and *wumt. Unaalik independent first person wii is evi-
dently based on a variant of the pronominal element in I. uwa-ra I, the recon-
struction may be *uwi-gimi, reduced to *uwigi, then contracted to uwii. In the
closely related Kukoskwim dialect, the form is uwiira, presumably with a late
addition of the common -na suffix. For the sake of brevity, we omit a number of
other anomalies of the pronouns and other formatives which apparently have
their cause in similar survivals of demonstrative elements from older combina-
tions.
In the discussion of the foregoing problems we have already seen several
demonstrative elements. We add a few of the available comparisons of specifi-
cally demonstrative forms:
Ka. nin, Ko. -nin him, Ka. nike- that thing, nirki- that person; A. ni- this,
E. -ni his, him (recurrent third person). From *ini.
Ka. ana he; A. na- that. Perhaps from *inina, the resulting double nn
reduced on both sides but reflected in the conserved vowel in Ka.
Ka. la- how, nakc what, something, Ch. raq- what, something, Ch. raq-
what; E. na- in words for how, where, etc. From *na/la/ra.
Ch-Ko. na- that; A. a~ yes (perhaps from thus), ara-gta- affirms, ara-gan
inside (perhaps in that) ara-gaan past him, ara-dan before, toward, ara-daa
accordingto that, I. a[-i-q- affirms (from does or says yes), ar-na that (out of
sight). From *arja/ani.
Ka. lki-, Ch. mi-kina-, Ko. me-ki who; A. ki-n, E. ki-na who. From *Ui.
N. Ka. zu-/zi- pronominal base in zu-ze ye and zgu- something; E. su-c,
ca-, sa- what, something.From *za/zu/zi.
Ka. ti-nu- that, t-, tu- pronoun bases; A. qata what, tapa then, well, t-, tu-,
ti- pronoun bases, Siberian tawa-ni there, I. taima thus, ta(s)- that, tiatna,
tiasa so. From *ti/te/to/ta.

This content downloaded from 200.54.94.206 on Fri, 10 Jan 2014 11:08:42 AM


All use subject to JSTOR Terms and Conditions
SWADESH] Linguistic Relations Across Bering Strait 1289
Ch. wa-y that wai0qan there, Ko. wu-tku here; E. -wa- in u-wa-, wa- pro-
noun base in first person and in uwa, wa-ni here, ta-wa there, Siberian sa-wa
what, awa-ni(t)- hasn't any. From *wa/wu.
SUMMARY
From the evidence examined here, it is seen that Chukotan and Eskaleutian
share many general and specific structural features, a large number of specific
structural elements, and many lexical items. The agreements are in accord
with a consistent phonology, each component equation of which is demon-
strated by numerous examples.
Although there is evidence for old phonemic interchanges, reflected in a
large number of oblique concordances, it has been possible to present here over
120 direct sound-for-sound equivalences in lexical items and some 40 in struc-
tural items. These represent somewhat more than half of my present collection
of resolved problems, but probably a minor fraction of the forms still to be iden-
tified even in my present limited materials. This is far more than the normal
fall of chance, leading to the conclusion that an old historical identity is in-
volved. This is further supported with considerable force by the fact that
structural anomalies in each stock are often clarified by elements or formations
existing in the other.
As explained in an earlier section, the measure of the chance factor was
based on a due consideration of the number of phonemes and their structural
permutations, the equations of correspondence, and the total number of dis-
tinct items in the corpus of comparison.
It is not claimed that all the problems of structural and phonological rela-
tionships have already been worked out. In fact many years of research for
various scholars lie ahead. Still, enough clarity already exists to justify publica-
tion of these first results.
Historically, the material goes beyond indicating some undefined form of
historical connection. Instead, the number and kind of agreements is consistent
only with an ancient identity of the two languages. This implies that their
speakers in the common period formed a single speech-community, although
probably a complex one, covering considerable territory and involving various
dialectal forms by the time of their final separation. On the other hand, this
community need not have been exclusively pre-Chukotan and pre-Eskaleutian.
Other evidence, still under study, suggests that at least Uraltaian, Gilyak,
Ainu, and Wakashan are also connected.
REFERENCES AND PRINCIPAL SOURCES
AA American Anthropologist
BSLP Bulletin de la Societe de Linguistique de Paris
IJAL International Journal of American Linguistics
JSAP Journal de la Societe des Americanistes de Paris
SSUF Sprakvetenskapliga Sallskapets i Uppsala Forhandlingar
SWJA Southwestern Journal of Anthropology
YPNS Y. P. Alykor ed., Yaziki i Pyismennosty Narodov Severa, part 3, Moskva-Leningrad
1934 Languages and Literatures of the Peoples of the North).

This content downloaded from 200.54.94.206 on Fri, 10 Jan 2014 11:08:42 AM


All use subject to JSTOR Terms and Conditions
1290 American Anthropologist [64, 1962
ANKERIA,
JUHO
1951 Das Verhaltnis der tschuktschischen Sprachgruppe zu dem uralischen Sprach-
stamme. SSUF 1949-51:109-50.
BERGSLAND,KNUT
1951 Aleut demonstratives and the Aleut-Eskimo relationship. IJAL 17:167-79.
1953 Aleutisk-Eskimoisk Kobenhaven, Nordisk sprog og Kulturforlag.
1955 A grammatical outline of the Eskimo language of West Greenland. Oslo, Skrive-
maskinstua.
1956 Some problems of Aleut phonology, In For Roman Jakobson.
1958 Aleut and Proto-Eskimo. Proceedings of the Thirty-Second International Congress
of Americanists. Pp. 624-31.
1959 The Eskimo-Uralic hypothesis, Suomalais-Ugrilaisen Seuran Aikakauskinjasta
61:1-29.
BOGORAZ,V. C.
1934 Luoravetlanskiy (Chukotskiy) Yazik. YPNS 5-46. (The Chukchi Language).
1937 Luoravetlansko-Russkiy Slovary. Moska-Leningrad. (Chukchi-Russian Dic-
tionary.)
COLLINDER,BJORN
1934 Indo-uralische Sprachgut.
1948 La Parente linguistique et le calcule des probabilites SSUF 1-24.
GEOGHEGAN, RICHARD HENRY
1944 The Aleut language Washington, United States Department of the Interior.
HAMMURICH, L. L.
1957 Can Eskimo be related to Indo-European? IJAL 17:217-23.
ROMAN
JAKOBSON,
1942 The Paleosiberian languages. AA 44: 602-20.
HANS
JENSEN,
1936 Indogermanisch und Gronlandish, Germanen und Indogermanen (Helmut Arntz,
ed), Heidelberg, 2:151-58. Carl Winters Universitatsbuchhandlung.
MARSH,GORDON, and M. SWADESH
1951 Eskimo Aleut correspondences. IJAL 17:209-16.
MARTfNEZ DELRfo, PABLO
1952 Los Orfgenes Americanos, Mexico, pp. 200-25, Editorial ARS.
MENOVSHCHIKOV, G. A.
1960 Eskimossko-Aleutskiye Paralleli (Eskimo-Aleut Parallels). Ucheniye Zapyiski
167:171-92.
MOLL,T. A., and P. I. INENLYIKEY
1957 Chukotsko-Russkiy Slovary (Chukchi-Russian Dictionary). Gosudarstvennoe
Uchebno-Pedagogicheskoe Izdatelystvo.
RADLINSKY, IGNACY
1892 Stowniki Ludow Kamczackich. Krakow, Rozprawe widziat u filologicznego Academii
Umejetnosti.
RIVET,PAUL
1925 Les Australiens en Amerique. BSLP 36:623-63.
1926 Les Malayo-Polynesiens en Amerique. JSAP 18:141-278.
SAPIR,EDWARD
1925 The similarity of Chinese and Indian languages. Science 67 (1607) 16 October,
Supl. p. XII.
SAUVAGEOT, AURaLIEN
1924 Eskimo et Ouralien. JSAP 16:279-316.
SHAFER,ROBERT
1952 Athapaskan and Sino-Tibetan. IJAL 18:12-19.
SHIMKIN,D. B.
1960 Review of Aleut Dialects of Atka and Attu, by Knut Bergsland. American Anthro-
pologist 62:729-30.

This content downloaded from 200.54.94.206 on Fri, 10 Jan 2014 11:08:42 AM


All use subject to JSTOR Terms and Conditions
Linguistic Relations Across Bering Strait 1291
SWADESH]

SKORYIK, P. Y.
1958 K
Voprosu o Klassyifyikatsii Chukotsko-Kamchatskikh Yazikov, Voprosi Yazikoz-
nanyiya, pp. 21-35.
Voprosio Stravnyityelynom Izuchenyii Chukotsko-Kamchatskikh Yazikov.
STYEBYINTSKIY, S. N.
1934a Nimilanskiy (Koryatskiy) Yazik. YPNS 47-84.
1934b Itelymyenskiy Yazik. YPNS 85-104.
SWADESH,M.
South
1948a Greenlandic paradigms. IJAL 14:29-36.
A structural trend in
1948b Nootka. Word 4:106-19.
1951-2 Unaaliq and Proto Eskimo. IJAL 1:25-34, 69-76, 166-71, 241-56.
Sino Tibetan.
1952 Athapaskan and IJAL 18:178-81.
1956a Contributions, Section
B, Comparative Linguistics (on quantitative criteria for
provinglinguistic
relationship). Proceedings of the Seventh International Congress
ofLinguists, London 1952, pp.38-39.
1956bProblems of
long-rangecomparison in Penutian. 32:17-41.
1959 The mesh principle in comparative linguistics. Anthropological
Linguistics 1:7-14.
1960aOn
interhemisphere linguistic connections. In Culture in History, S. Diamond, ed.
New York,pp. 894-924. Columbia University Press.
1960b
Tras la HuellaLingiiistica de laPrehistoria. Suplementos delSeminario de Prob-
lemas
CientSficosy Filos6ficos, 2a. Serie, No. 26.
1961 Los
supuestos australianos en America. Homenaje aPablo Martmnez delRfo, pp.
401-25.
1962 Archaic doublets
in Altaic. InAmerican Studies in Altaic, N. Poppe ed.
WILLIAM
THALBITZER,
1921 The Aleutian language compared with
1928 Is there Greenlandic. IJAL 2:40-57.
anyconnection between the Eskimo language and the Uralian? Atti del
Congresso Intern. degli
XXII
1952 Possible Americanisti 2:551-67.
early contacts between Eskimo and Old World languages.
of In Indian Tribes
Aboriginal America, S.
Tax. ed. Proceedings of the 29th International Congress of
Americanists3: 50-54.
ARTHUR
THIBERT,
1954
English-Eskimo Eskimo-EnglishDictionary. Ottawa, University of Ottawa Re-
searchCenter of Amerindian Anthropology.
C.
UHLENBECK,C.
1905
Uralische Anklange in den Eskimosprachen. Zeitschriftder deutschen morgen-
landischen
1935 Eskimo Gesellschaft 59: 757-65.
en
Oer-Indogermaansch. Mededeelingender Koninklijke Akademie van
Wetenschappen, Afdeeling Letterkunde Deel 77,Ser. A. 6:179-96.
TOR
ULVING, Consonant gradation in Eskimo.
1953 IJAL19:45-52.
WORTH,
DEAN STODDARD
1959Paleosiberian etymologies. IJAL25:32-40, 105-13.

This content downloaded from 200.54.94.206 on Fri, 10 Jan 2014 11:08:42 AM


All use subject to JSTOR Terms and Conditions

Das könnte Ihnen auch gefallen