Boas, Franz. Race, Language and Culture – T23

[Language]

Classification of American Indian
Languages 11

[The author points out cases in which contiguous languages,
though different in structure and vocabulary, exhibit in common
striking morphological peculiarities that must have spread by borrowing
from language to language. A simple genealogical classification
cannot therefore adequately represent the development,
but “hybridization” must also be taken into account.]

In a paper published in 1920 22 I discussed the problem of the interrelation
of American Indian languages. I pointed out that morphological
types are distributed over large areas and that in these morphological
groups differences representing the character of the vocabulary
occur which make it difficult to assume that the languages, as now
spoken, are derived from the same “Ursprache.” I pointed out that in
the small linguistic units of early times, the conditions of mixture were
quite different from those found in languages spoken over large areas
and by many individuals. A further consideration of the problem led
to the conclusion that an answer to the fundamental question must be
sought through an investigation of mutual influences and the extent to
which they may modify languages; particularly, in how far one linguistic
type may influence the morphology of another.

I believe everybody will agree that words may be borrowed and may
modify the vocabulary of a language; perhaps also that the phonetic
character of one language may influence that of its neighbors. I have
given a few general instances in the paper mentioned before, and today
I will add one example that seems to be particularly instructive. The
Nez Percé, an eastern Sahaptin language, has rigid rules of vocalic
harmony according to which vowels may be divided into two classes:
a and o as one group; all the others as a second group. In the system of
consonants occurs an s with raised margin of the tongue and the dental t
series. Another characteristic sound is a voiced affricative, something
219like dl. During the eighteenth century a large group of the Sahaptin
penetrated into the State of Washington and some of them crossed the
Cascade Mountains where they intermarried with the Salishan tribes
resident there. The phonetic elements of the present dialect of this
region are practically identical with those of the neighboring Salishan
tribes. The vocalic system is the same. There is no trace of vocalic
harmony.

We recognize that a comparison of vocabularies of languages the
history of which is unknown offers serious difficulties, and that the
changes brought about by the shifting of sounds, by semantic modification,
and by new formations, may be so numerous that identification
becomes possible in exceptional cases only. Languages behave differently
in these respects. Some, like the Eskimo, are so conservative that
even now the differentiation between Alaskan and Greenland dialects is
slight, although the two groups have been separated for more than a
thousand years. The more striking is the divergence of the vocabulary
of the probably related Aleutian. Aztekan has changed in so far as
the higher literary style has disappeared and as old ideas have vanished
and new ones have been introduced with concomitant change of vocabulary.
The syntactic subordination and co-ordination of phrases have
yielded to Spanish types. In all other respects the modern language has
not changed. It seems even possible to recognize the dialectic differences
of various areas which may be reconstructed from the grammars of the
early sixteenth century. On the other hand, the Salishan languages of
British Columbia and Washington illustrate a great instability in morphology
and lexicography. We can only guess what the causes of the
difference in behavior of different languages may be. The often expressed
opinion that “primitive languages” undergo very rapid changes
is true to a very limited extent only.

There is no doubt that in many cases languages sprung from the same
source and changing by internal forces only may have become so different
that without historical data their relation cannot be established.

Nevertheless the question remains whether hybridization of languages,
not only in phonetics and vocabulary, but also in morphology,
may have occurred.

So far as I know the actual process of a transfer of grammatical
categories from one language to another has never been observed, although
minor changes, like the adoption of a form here and there, and
syntactic influences are known to occur. The syntactic modification of
220American languages under Spanish influence offers a good example
of the latter type of change. The proof of the diffusion of morphological
forms can be only indirect, based on facts of distribution and partial
conformity by the side of fundamental differences.

In some cases of far-reaching similarity of morphology, like that of
Athapascan and Tlingit, we may feel that an assimilation of the structure
of an older language by Athapascan is quite unlikely; and that, if
no safer correspondence of vocabulary can be found than has been presented
up to this time, we may suspect that an older vocabulary has
been taken over by the invading Athapascan. Until definite phonetic
shifts can be proven by a sufficient number of parallel forms, and until
an exhaustive comparison of vocabularies has been made, we have to
admit that a vast array of stems in the two languages cannot be identified,
including pronouns, numerals, and most other stems; and we must
leave open the question whether all, or most of the lexicographic material
can be derived from a common source.

More difficult are those cases in which a partial agreement in morphological
traits exists between neighboring and apparently distinct languages,
and disagreement in the dialects of obviously related languages.
I may give an example of this kind. I mentioned before the vocalic
harmony of the Nez Percé. So far as I am aware only the Coos of
Oregon exhibit a similar, consistent phenomenon. It is not known
whether the neighboring Molala and Kalapuya have it. Other Sahaptin
dialects do not show it.

Chinook possesses pronominal gender. There are not only pronouns
of three genders — or more strictly speaking five nominal categories,
for dual and plural belong to the same system — but every noun has
prefixed one of the five pronouns. None of the languages of the adjoining
groups have sex gender except a number of dialects located in close
proximity to the Chinook, particularly all the dialects of Salish tribes
that live along the coast northward and southward, and the Quileute.
In the Salish dialects of the interior, gender does not occur. If the Quileute
should prove to be related to Wakashan, to which it shows
morphological resemblances, it will be the only language of this group
which has gender. In all these dialects gender is confined to the pronoun.

Chinook expresses diminutives by consonantic changes. Voiced and
unvoiced consonants become glottalized and š changes to s. Velar fricatives
become midpalatal fricatives. The neighboring Sahaptin groups,
which differ fundamentally from Chinook, use consonantic changes for
221the same purpose. Some of the changes are the same as in Chinook;
š changes to s, velars to midpalatals, and besides these a change from
n to l occurs.

We find sporadic, fossilized use of the same process in the Salish
dialect spoken just north of the Chinook area, in Coos on the coast of
Oregon, 13 and as a living feature in Wiyot in Northern California. Geographical
contiguity for the last example cannot be established.

It will be noticed that while gender exists in a coastwise direction
north and south, the formation of the diminutive by consonantic changes
occurs in a territory extending eastward.

Another curious resemblance may be traced between Quileute,
Kwakiutl, and Tsimshian, which are spoken in an area extending from
the State of Washington to the Alaskan boundary. In these three languages
the pronominal representation of the noun (or article) is treated
differently for proper names and for common nouns. These form
throughout two distinct classes. In Quileute and Kwakiutl a further
correspondence is found in so far as the article used with proper names
is also used for indefinite, that is unknown objects. For instance, “I look
for a whale”, indefinite; “I found a whale”, definite.

Many American languages draw a clear distinction between possession
by the subject and possession by another person, like the Latin
suus and ejus. A small group, including the Eskimo, Algonquian, and
Kutenai, express these relations by special verbal forms, the so-called
obviative of the missionaries who wrote on Algonquian, the fourth person
of Thalbitzer. The phenomenon is most pronounced in Kutenai,
for even in the case of the simple transitive verb with third person subject
and nominal object the presence of the two third persons is indicated
by the obviative suffix following the nominal object. It is interesting to
note that the western Sahaptin languages, which as a whole group adjoin
the Kutenai, make the same distinction for the subject of the sentence
for sentences containing only one third person and those in which
the sentence contains two third persons. In both Kutenai and western
Sahaptin there is a differentiation between the forms in a sentence like,
“the man saw me,” and “the man saw the woman.” In Kutenai the
difference is found in the object, in Sahaptin in the subject. In some of
the Sahaptin dialects this trait is found only in the pronoun, not in the
noun. The general usage, in the group of languages just discussed, is
alike notwithstanding the difference of devices used.222

Another interesting feature may be observed in the languages of the
North Pacific Coast. Demonstrative pronouns are often very elaborate.
They not only distinguish between the person near the speaker, near the
person addressed, and near the person spoken of, but more exact locations
are often added. The Tlingit of Alaska differentiate between what
is near him but nearer than you, and what is near him but farther than
you; or positions in front, behind, above, or below the speaker may be
designated. Among the tribes extending from Columbia River northward
to Alaska — the same group which differentiates between proper
names and common nouns — a different demonstrative concept is introduced,
namely that of visibility and invisibility. The Chinook has
demonstratives designating, for instance, “near the speaker, visible.”
The same occurs in Quileute and Coast Salish, but not in the Salish dialects
of the interior. It is a characteristic feature of Kwakiutl. I do not
know of its occurrence in any other group of neighboring languages.

Still another feature characteristic of part of the same group is the
separation of pronominal subject and object in transitive verbs. The
verb unaccompanied by what we should call an adverb, takes a suffix
consisting of pronominal subject and object combined. When a qualifying
adverb accompanies the verb, the subject is attached to this qualifier
which takes the form of an intransitive verb, while the object remains
attached to the primary verb. “I did not see him” would be expressed
by “not-I see-him.” This tendency occurs in exactly the same form in
Quileute, Coast Salish, and Wakashan. In Tsimshian it is less fully
developed, in so far as in subjunctive forms the pronominal subject precedes
the verb and is phonetically united with the preceding adverb.
The analogy, however, is not strict.

Another interesting comparison may be made between Chukchee
and Eskimo. In regard to the general form, these two languages are
quite distinct. Chukchee employs terminal reduplication, prefixes, suffixes,
and vocalic harmony. Besides this there are rigid rules regarding
initial consonantic clusters which bring about important modifications
of stem form. Eskimo has nothing of the kind. There is no reduplication,
no prefixes whatever, no trace of vocalic harmony. Whatever
changes occur in the stem are due to the influence of suffixes. On the
other hand, a number of categories occur which are common to these two
neighboring languages. The plural forms are alike; both Eskimo and
Chukchee form the plural by a suffix t. The nominal subject in Eskimo
is treated differently in the case of transitive and intransitive verbs. The
223subject of the transitive verb has what might be called a relational form,
common to both the genitive and the transitive subject. The subject of
the intransitive verb has the same form as the object of the transitive
verb. This feature occurs also in other languages, as in Sahaptin, and it
is found in the pronominal forms of many other languages. But in the
circumpolar area only the Chukchee and Eskimo have this differentiation
of the nominal forms. The processes by means of which this differentiation
is made in Eskimo and Chukchee are quite distinct, for the
object in Chukchee is formed by terminal reduplication; in Eskimo the
subject is differentiated by a suffix. Furthermore we find in both languages
a considerable number of postpositions which express local relationships,
such as “at,” “towards,” “from,” and so on. The analogy in
the modal development of the verb is also quite striking. A remarkable
variety of participial forms occur which may take personal pronouns
and the group of concepts expressed by the modalities shows marked
similarity.

Considering these data as a whole, we may say that in a considerable
number of native languages of the North Pacific Coast we find, notwithstanding
fundamental differences in structure and vocabulary, similarities
in particular grammatical features distributed in such a way that
neighboring languages show striking similarities. The areas in which
similar features are found do not coincide in regard to the various
traits compared.

It seems to me almost impossible to explain this phenomenon without
assuming the diffusion of grammatical processes over contiguous areas.

Stress must be laid here upon the contiguity of distribution, because
comparative grammar shows clearly that similar features may develop
independently in different parts of the world. Sex categories, phonetic
similarity between the Northwest Coast and Chile, the application of
reduplication, and many other traits appear in such distribution that
historical connection is excluded. On the other hand the distribution
of the same particular grouping of concepts, or of the same methods of
expression over contiguous areas can hardly be explained on the basis
of independent origin.

So far as I can see an attempt to bring together the different languages
of contiguous areas which have similar processes is not feasible on account
of the fundamental differences in conceptualization, in grammatical
processes, and in vocabulary.

The phenomena here discussed lead to a result analogous to that
224reached by Lepsius in his study of African languages. He concluded
that a large number of mixed languages occur in Africa. His conclusions
are largely corroborated by more recent investigations, particularly
of the Sudanese languages. It is also parallel to the results obtained by
von der Gabelentz in his study of the languages of New Guinea and
Melanesia, and his inferences are substantiated by the recent investigations
of Dempwolff. The problem has been well formulated by Professor
Prokosch, who demands a detailed comparison of the European
languages with all their neighbors, no matter to what linguistic stock they
may belong. It also agrees with the view of Schuchardt, who points out
that there is a gradation beginning with a slight amount of borrowing
and extending through more intensive intermingling, to a complete
change of language. The question in which we are interested is not that
of the theoretical definition of relation of languages as defined by Meillet,
but merely a question of historical development.

If the view expressed here is correct, then it is not possible to group
American languages rigidly in a genealogical scheme in which each linguistic
family is shown to have developed to modern forms, but we
have to recognize that many of the languages have multiple roots.225