CHAPTER XVII
LINGUISTIC CLASSIFICATION
While, in the broadest sense, language is a trait of culture, its characters are so distinct as to require very different methods of investigation. The study of the different forms of speech in the Old World has become a separate learned pursuit and one in which the problems are so complex as to demand great specialization. The same is true of primitive languages in the New World, as well as in the less cultured parts of the Old. Consequently we have, under the head of New World linguistics, a fairly distinct division of our subject in which the most important investigations have been made by those who have specialized in it.
One of the first tasks in primitive linguistics is the classification of the existing forms of speech. The reader may need to be reminded that the national uniformity of language in Europe is a correlate of close political organization, one type of speech having been selected by the governing authority and its use perpetuated by enforced education. In the more primitive states of society, where political unity exists only for a single community and no legislative recognition of languages is taken, we may expect each such political unit to show some individuality in speech. In fact, so far as can be judged from the data at hand any separation of such a political unit into two or more parts will sooner or later result in different forms of speech.
Under such conditions, very unequal differences will exist. Between some groups we shall find but a small difference in vocabulary; in others, an additional phonetic change; and finally, variations in the grammatical structure. All degrees of these differences are found when we consider a very large number of political units, giving us an intergradation analogous to that we observed in the cultures of the same units. This inequality of difference renders any classification more or less arbitrary. So long as the tongues of two or more groups do not diverge beyond the possibility of communication, they are usually considered as dialects, though the degrees of mutual intelligibility connoted by that term may vary greatly. When mutually unintelligible tongues are found to possess consistent similarities in vocabulary or grammatical structure, particularly the latter, they are said to be of the same family, or stock. Thus, all the languages of the New World may be placed in stock groups, the conception of a stock being a group of related languages with their dialetic subdivisions.
The determination of these stocks for the native tribes of the United States was initiated by Gallatin[1] in 1826 and brought to a definite form in 1891 by J. W. Powell,[2] who organized the Bureau of Ethnology at Washington. Powell is credited with the system of nomenclature now used for all the languages of the New World, which is to take some native unit name and give it the adjective termination an. Thus, Siouan stock is from the term Sioux, Caddoan from Caddo, etc. In the main, this system is still followed, though the clumsiness of the termination in some instances has led to modification.
Powell confined his classification to the tribes of the United States and Canada, which he grouped under fifty or more stocks and prepared a map showing their distribution. It is not too much to say that this classification and map is the very foundation of American anthropology. The work was so well done that very few changes have been made and as it was, in the main, based upon mere vocabularies, its excellence stands as a worthy memorial to Powell and his able associates. (See map, Fig. 85, and list of stocks for United States and Canada, pp. 369-378.)
The general value of this classification arose from the fact that it was then the only classification of native tribes based upon scientific principles and, therefore, afforded the point of departure for the investigation of culture and racial anatomy. Thus, its importance arises from historical conditions, for it has very little to contribute to the problems of culture and anatomy aside from the practical matter of tribal designation. 
Fig. 85. Linguistic Stocks in the United Stales and Canada.
J. W. Powell
On the other hand, we should note that the basis of classification employed for language is different from that used for culture. As stated in the preceding chapters, cultures were grouped according to their objective similarities and their geographical associations, but such a gross grouping was not in any sense expressive of genetic relations between the cultures concerned. We saw, however, that the ultimate grouping of cultures was destined to be chronological, which, in a way, would express genetic relationships. Now, the conception of a linguistic stock is a group of languages all the members of which have a common ancestor, and between whom it is possible to establish degrees of relationship. Hence, the moment we identify a language as belonging to a definite stock, we automatically assert its genetic relationship. In this sense, the classification of the linguistic characters of man is superior to the present classification of his cultural characters.
We have now a considerable number of specialists in North American languages, but the task before them is so great that the investigations of the several stocks have not progressed far enough to make possible a comprehensive statement of the whole subject. In scarcely a single case are we able to designate the type tongue for a stock or to indicate the historical relationships of its divisions. All this is for the future. At present, the refinement of classification seems to be the chief interest. As we have stated, even the stock designation is somewhat arbitrary, for greater familiarity with these languages has brought to light new similarities between what have heretofore been regarded as distinct stocks. For instance, the
List of Stocks for Fig. 85
|
1. Algonquian |
20. Keresan |
39. Tanoan |
Fig. 86. Linguistic Stocks in Mexico and Central America. Thomas and Swanton
Shoshonean and Nahuatl have been combined; the Piman stock is also considered as a member of this group by Swanton and Kroeber; Natchesan and Muskhogean were grouped together by Swanton;[3] Sapir[4] claims Wishoskan (Wiyot) and Weitspekan (Yurok) to be Algonquian; Kroeber and Dixon[5] group the Shasta, Chimariko, Karok (Quoratean), Pomo (Kulanapan), Esselen and Yuman of California as one stock under the new name of Hokan, and in like manner the Wintun (Copehan), Maidu (Pujunan), Yokuts (Mariposan), Miwok (Moquelumnan) and Costanoan under the name Penutian.[6] All such changes have met with opposition and are accumulating a large amount of controversial literature.
Aside from these specific attacks upon the Powell classification it has been recognized that stocks show a geographical grouping of some kind. Boas designates a southeastern type of language to which also the Iroquois and Caddoan seem to belong, other students have noted a similar grouping of the Shoshonean, Kitunahan, and Kiowan stocks. Boas and later Sapir[7] find similarities between the Tlingit, Haida, and the Athapascan. It is quite probable that the further investigation of these relations will result in the elimination of some stocks, and in any event establish some kind of historical relationship. When this comes to pass, we may find that languages have also a geographical grouping not unlike that for other culture traits. Incidentally, we note that the existence of even such geographical affinities as have been so far established suggest long and permanent residence in one area.
List of Stocks for Fig. 86
|
1. Athapascan |
11. Lencan |
21. Tamaulipecan |
Unlike culture, language has no true archæology, but a few stocks have become extinct since their discovery, as indicated in the tabulated lists. Yet, it is truly surprising that so few have so far passed out of existence, though it is probable that their extinction will proceed from now on with increasing rapidity.[8] This tendency of speech to survive long after great culture change must be taken into account when we attempt to correlate our classifications. It also complicates the problem of linguistic origins.
To summarize, then, we find a genetic basis for linguistic classification, expressed by the term stock. All the native languages of the United States and Canada have been identified, or we may say there is no historic tribe in this territory whose linguistic stock is not known.[9] For Mexico and Central America[10] we can not be sure, but it is improbable that more than two or three have been overlooked. In South America,[11] we are still less certain of the completeness of our knowledge. In the appendix we have tabulated the recognized stocks for these geographical divisions: for the United States and Canada there are fifty-six stocks; for Mexico and Central America, twenty-nine; and for South America, eighty-four. Distribution maps have been prepared which we present in outline: Figs. 85, 86, and 87. Upon these, the stocks are designated by numerals which stand for the corresponding stocks in alphabetical order (see pp. 369-385.)
COMPARATIVE PHONETICS
Linguistic studies fall into two rather distinct groups: phonetics and structure. Of these, phonetics is still quite undeveloped, the greater effort being placed upon structure, or word and sentence formation. Yet, some progress has been made in phonetics. For a time, attention was given only to the necessary practical ways of recording these new languages, each field-worker devising his own system. A certain initial uniformity was secured by the mechanical limitations of printing, but even this proved unsatisfactory. Recently the American Anthropological Association appointed a committee to formulate and standardize the methods of transcribing and printing.[12] The report of this committee is essentially a pioneer work on New World phonetics, but no such practical system, however perfect, can take the place of independent phonetic investigation. A beginning has been made by Goddard who introduced the objective mechanical methods of phonetic analysis devised by European students, but as yet the whole subject is before us.[13]
In a very tentative way a few of the characteristics of North American phonetics have been suggested. Perhaps one of the most obvious differences from English, for example, is the small use made of the stress accent. Another peculiarity of American speech is a curious stopping of the breath before or after vowels, the mechanism for which is a closure of the glottis. Some of the native consonants have proved particularly difficult to European recorders. Among these are the stopped consonants, b, d, g, k, p, q, and t, which have many forms, but sometimes seem to stand between sonants and surds. These sounds are usually designated as intermediate stops. Again, there are glottalized forms resulting from a closure of the glottis and a quick release, giving an explosive sound not found in European speech. Yet the present enumeration of such assumed phonetic differences can serve no very good purpose, for until the study of New World phonetics comes into its own, we can not state what characters, if any, are peculiar to it.
In passing, note should be taken of a few pioneer regional surveys. In California, where we find a large number of different native languages, there appears to be a fair degree of phonetic uniformity.[14] For one, the vowels tend to be open rather than closed and are given with a greater breath impulse than in European languages. The intermediate stops, just mentioned, are numerous and there is a strong tendency to use two t's, but on the other hand a single k.
In how far these phonetic traits are limited to California can not be stated, but such phonetic peculiarities as have been reported for the North Pacific Coast peoples suggest some differences; for example, there is a somewhat unique voiceless l found all the way from Alaska to northern California, and 
Fig. 87. Linguistic Stocks in South America. A. E. Chamberlain
along with it, in the main, a few fortis consonants. These suggest that a study of phonetic distribution will not only reveal some definite New World characters, but show localized geographical types as well. From a much broader point of view, it appears that in phonetics, we have, perhaps more than anywhere else, an instinctive formative culture factor and an organic basis for the unconscious modification of a trait, all of which leads into psychological problems of the very greatest importance to anthropology.
COMPARATIVE MORPHOLOGY
While, as stated above, the structure or morphology of languages has received the most attention, yet when we consider the speech of the New World as a whole, even this
List of Stocks for Fig. 87
|
1. Alikulufan |
29. Coconucan |
57. Otomacan |
subject is quite undeveloped. In the main, the time of investigators has been taken up with the necessary recording of texts and the working out of translations, and though their published papers often close with summaries of the structural systems employed, this has in no case gone far enough to admit of extensive comparative studies for the continents at large. However, the reader will sometimes find in the older literature certain characterizations of New World languages, cited as distinguishing them from languages in other parts of the world. These should be ignored, for the different tongues of the New World show very wide divergence in their structure, in fact, presenting about all the known varieties.
Yet there is one aspect of morphology in which the New World shows some distinction. While, in general, the morphological classification of all languages is rather difficult, yet from the point of view of the internal coherence of the word they can be comprehended under three heads: isolating, agglutinative, and inflective.[15] Other bases of classification have been proposed, but are found less satisfactory; hence, they may be ignored here. Applying the above criterion to the speech of the New World, we find it chiefly agglutinative, in fact, as far as we know, exclusively so. On the other hand, agglutinative speech occurs in the Old World so that this character can not be taken as peculiar, we can only say that perhaps in this respect greater morphological unity exists among New World languages. Further, there seem to be some indications that certain processes, as the incorporation of the object in the verbal phrase, the polysynthetic formation of terms, etc., are of more frequent occurrence here than in the Old World, but it remains for future students to definitely establish the facts. In fact, as the case stands at present, there are no known morphological characters strictly peculiar to New World speech as a whole.
If we attempt a review of comparative studies in morphology, we must again turn to California and the Pacific Coast, for as we shall see presently, this portion of the continent is the home of a large number of small linguistic families. The most important single contribution to the subject is by Dixon and Kroeber[16] who showed that, as a whole, the languages of California differ in the non-use of reduplication of the noun to indicate plurality, the method of incorporating the pronoun and even the noun into the verb, and in contrast to these, the employment of syntactical cases, a method almost unknown in other parts of the continent. These of themselves suggest that we have here a close parallel to the grouping of political units according to culture, for these linguistically independent stocks still show sufficient traits in common to form a geographical area.
In like manner, the languages of the North Pacific Coast form another area. Though no convenient summary is available for this group of stocks, Sapir[17] has proposed the term Na-dene for the Tlingit, Haida, and Athapascan on the assumption that they have a genetic relation. However this may be, this author shows that they do form a group with some distinctive characters, as the absence of reduplication, the very frequent use of freely compounded stems, modes, and tenses indicated by internal phonetic changes in the verb stem, the somewhat general tendency to loose synthetic structure in forming words, etc. Of the other languages, Tsimshian on the one hand and Wakashan (Kwakiutl-Nootka) and Salish on the other, have some characters in common, as initial reduplication in both nominal and verbal forms, and the use of suffixes in numerals as classifiers of the objects designated. Again, all the above languages have certain phonetic similarities. Thus, while we can not group these languages so readily as those of California, yet we are able to distinguish them with respect to certain characters. Further, they fall into contiguous geographical positions.
In an analogous manner Speck[18] shows that Muskhogean of the lower Mississippi has affinities with its neighbors and recently Swanton[19] reports evidences of grouping among the several small stocks of the Gulf Coast. There is, therefore, a presumption that Caddoan, Iroquoian, Muskhogean, Uchean, etc., all possess certain characters that may be taken as linguistic indices of the region, or culture area 8 (p. 205).
Thus, in general, we seem to have a grouping of stocks around geographical nuclei and this raises the question as to the nature of this larger group. Is this somewhat vague, but still definite, group of stocks to be considered as of the same sort as the strictly genetic cluster within a single stock? If the answer proves to be negative, then we must seek for some external causes contributing to this result; but, if positive, then the stocks forming such a group are themselves of common origin or descent. Between these two positions, the anthropologists of the United States are about equally divided. If we examine the case upon its gross merits, there appears no discernible difference between the kinds of similarities by which stock affiliations are determined and those making these geographical groups. It is merely a matter of degree. That being the case, the probabilities favor the genetic interpretation. On the other hand, very vague similarities may be possible as the result of mere social contact, in which case the genetic relation could only apply to one or two elements of the language. However, speculation upon these points is futile, and we may conclude by noting that along this line the greatest advances in future linguistic researches are to be expected.
The best illustration of the above is found in the investigations of Californian languages by Kroeber and Dixon,[20] to which we several times referred. In Fig. 88 we see the positions of the original twenty-two stocks identified by Powell. Later, Kroeber and Dixon[21] discovered three large morphological groups for these stocks as indicated by the shadings on the map. Recently,[22] these investigators have become convinced that these larger groups represent single stocks, two of which are not found elsewhere (Hokan and Penutian), while the third is a member of the Algonquian stock. While other linguists have not yet accepted this genetic interpretation of the observed grouping, the facts of similarity are not in dispute. In our discussion of the Californian culture area we noted a central group of tribes possessing the most typical culture (p. 212), and we now see that these are almost exclusively members of the Hokan group. Thus, if the genetic relationship of these Powell stocks is denied, we must assume a correlation of some kind between language and culture. On the other hand, if the genetic interpretation is valid, what shall be said of the similarities between Muskhogean, Uchean and Iroquoian, previously cited?
Fig. 88. The Proposed Consolidation of Stocks in California.
Kroeber and Dixon
Finally, irrespective of these puzzling questions, there is a problem in mere word distribution. For example, the term for dog is approximately the same in many California stocks; the root term for hoofed animals is the same for several Algonquian, Siouan, and Caddoan languages and there seems to be a continental tendency to use n and m as the roots of pronouns. Some of these observations may have historical value since the word for dog must have traveled along with the animal. The probabilities are that such studies will lead to historical discoveries and establish time relations in culture.[23]
DISTRIBUTION OF STOCKS
We may summarize the previous discussion by the statement that the chief general result of linguistic investigation in the New World has been the identification and location of the several stocks. Yet, the perfecting of this classification is no mean achievement and is destined to play an increasingly larger rôle in the development of our subject. The important point is that such a classification is based upon the idea of genetic relationship, and so stands in much the same relation to our subject as does evolution to zoology. Thus, there can not be the least doubt that all the Algonquian languages had a common ancestor and the further study of the several divisions of that stock promises to reveal the general outline of this initial type. Further, we have a right to expect that when the comparative studies we have noted are more advanced, definite genetic relations will appear between many of the now recognized independent stocks. For it is apparent that the cultural and somatic unity of the New World necessitates some kind of genetic relationship between the surprisingly large numbers of stocks now enumerated in its linguistic classifications. The pursuit of these important problems will be, in the main, empirical, and as such offers one of the most enticing fields for the scientifically inclined. Yet, this is for the future, our chief consideration here being the distribution of linguistic stocks.
This is shown upon the maps. We see that the most widely distributed stocks in the northern continent are the Athapascan, Algonquian, Iroquoian, Muskhogean, Caddoan, Siouan, Salishan, and Shoshonean-Nahuatlan. The Arawakan, Tupian, Tapuyan, Cariban, Puelchean, and Tsonekan are the largest of the southern continent. In North America, the eight large stocks enumerated occupy practically all of the United States and Canada save the Arctic and Pacific Coast belt and a small part of the Gulf Coast. The lower part of Mexico and southward to the Isthmus is also the home of many small stocks. Thus of the entire eighty-five stocks in North America, all but eight are crowded into less than one-eighth of the continental area where they occupy marginal positions. In South America there is some confusion arising from the interspersion of many stocks in the Amazon Basin, but again we seem to have the smaller stocks on the Pacific side, though their marginal positions are not so prominent. One curious fact is that in the regions of higher culture we find great linguistic diversity while the very large, widely extended stocks are met with chiefly in the regions of lower culture, though not exclusively. It does not follow, however, that the populations speaking the stock languages of the latter were much larger, because they were, in the main, hunting peoples and would require a large range for the support of each family. Whatever else these facts may signify, they indicate that the development of higher culture was not a linguistic phenomenon.
Another observable tendency of the large stocks is to spread over a single geographical area. Thus, while the Siouan stock has a few straggling remnants on the coast, the great main body is found in the open plains west of the Mississippi. The Athapascan, Algonquian, Salishan, and Shoshonean stocks show similar tendencies. In South America this is not quite so clear, but still seems to be the prevailing tendency. This suggests that the ancestors of these stocks took up their abodes in these respective areas and that their later distribution is the result of normal expansion, a common cultural bond tending to hold them to the same area. The detached groups appear to have modified their culture in response to their change of habitat. This does not signify that these detached tribes were always the migrants for it may sometimes have been otherwise. Thus, we have some very important problems in these larger stocks, for example the Athapascan. Here we have a right to expect that future linguistic research will reveal the oldest language group and its relations to the others and that from this their relative movements can be deduced. As it now stands, we can form no positive idea as to their original home, whether it was in Arizona, Oregon, or Canada. The most worthy suggestion comes from Sapir who states that the observable resemblance of Athapascan to the languages of the adjoining Pacific Coast is a strong argument in favor of a northern cradle land.[24]
Of almost equal importance are the Algonquian, Siouan, and Shoshonean-Nahuatl problems. The Algonquian and Siouan stocks have somewhat similar distributions, a large compact group with a few outlying detached fragments. In the former the detached Arapaho and Blackfoot speak very widely divergent Algonquian languages and if we admit Sapir's identification of Wishoskan (Wiyot) and Weitspekan (Yurok), we have representatives still farther removed geographically and correspondingly divergent. The Cheyenne appear somewhat less divergent than the Arapaho but we have historical reasons for believing their separation from the main body to be recent.[25] On general grounds, it has been proposed that the ancestral home of the Siouan stock was on the Atlantic Coast, where representatives were found, but it appears that the Biloxi of the Gulf Coast are nearer the central linguistic types than those farther east.[26] We see here a tendency for the outlying groups to be more divergent from the main body than those nearer, a relation favoring the view that these detached groups represent stragglers, but, whether laggards or true wanderers, is difficult to decide. In the author's opinion the probabilities favor the latter. Reasoning from the marginal phenomenon of faunistic distribution, it has often been assumed that the very small stocks on the coast belts represent the survivals of the more primitive groups. This view seems to have some justification, but it has not yet been demonstrated that these stocks are the more archaic forms of language. Hence, this interpretation so far as it applies to language, must be regarded with caution.
- ↑ Gallatin, 1836. I.
- ↑ Powell, 1891. I.
- ↑ Swanton, 1915. I.
- ↑ Sapir, 1913. I.
- ↑ Dixon and Kroeber, 1913. I.
- ↑ Dixon and Kroeber, 1913. I.
- ↑ Sapir, 1915. I.
- ↑ Goddard, 1914. I.
- ↑ Powell, 1891; Boas (editor), 1911. I.
- ↑ Thomas and Swanton, 1911. I.
- ↑ Chamberlain, 1913. I; Brinton, 1891. I.
- ↑ Boas, 1916. II.
- ↑ Goddard, 1905. I.
- ↑ Kroeber, 1911. I.
- ↑ Sapir, 1911. I.
- ↑ Dixon and Kroeber, 1903. I.
- ↑ Sapir, 1915. I.
- ↑ Speck, 1907. I.
- ↑ Swanton, 1915. I.
- ↑ Dixon and Kroeber, 1903. I.
- ↑ Dixon and Kroeber, 1913. I.
- ↑ Dixon and Kroeber, 1913. I.
- ↑ Sapir, 1916. I.
- ↑ Sapir, 1915. I.
- ↑ Michelson, 1912. I.
- ↑ Dorsey and Swanton, 1912. I.