Popular Science Monthly/Volume 78/May 1911/The Languages of the American Indians

THE LANGUAGES OF THE AMERICAN INDIANS
By Dr. A. L. KROEBER

CURATOR OF ANTHROPOLOGY, UNIVERSITY OF CALIFORNIA

THE day is past when educated people believed that the Indian languages were only random jargons of a few inarticulate sounds, without grammar or order, and so badly in need of supplementary gestures to make them intelligible that the Indians could not converse in the dark. Still farther have we got beyond the point of speaking of the Indian language, as if all tribes used essentially one and the same idiom. Such notions may yet linger among the uncultured, and now and then reflections of them still crop up in books written by authors whose knowledge is not first hand. But the progress of science has been so great in the last half century that the world now looks upon the tongues of the native Americans with newer and sounder ideas.

Probably the most important and most surprising fact about American Indian languages is their enormous number. On the North American continent there were spoken probably 1,000, and possibly even more different languages and dialects. Of South America we know less, but everything points to an equal linguistic variety on that continent. The tremendous total is astounding because the aboriginal population in both continents certainly numbered fewer millions than are to-day found in many single European countries in which only one language prevails. The twenty-five or fifty millions of American Indians possessed as many different languages as the billion or more inhabitants of the old world.

Language and History

To the historian and the ethnologist this linguistic diversity is of the utmost consequence, because it affords him his most important means of classifying the native peoples of America, and ascertaining their connections, their migrations and in part even their origins.

To the student of old-world history and ethnology, philology is also a serviceable handmaid, though to a less degree than in America. This happens, in the first place, because the languages of the eastern hemisphere are, on the whole, each more widely spread; and secondly, because history and archeology carry our knowledge of many peoples of Europe, Asia and Africa back for thousands of years—as compared with the bare four centuries since the discovery of America. History is, therefore, much more able to stand on its own feet in the old world than in the new. Nevertheless, when the historian goes back to origins. he has always been compelled, even in Europe and Asia, to call in the aid of language, and sometimes with the most fruitful results.

Starting, for instance, with our own language, English, the tongues nearest of kin to it are Dutch, German and Scandinavian. Next in closeness of relationship are the various Romance languages, evolved from the decay of ancient Latin—such as French, Italian and Spanish. Still more different, but yet with sufficient similarities to make relationship and ultimate common origin absolutely certain, are Russian and the other Slavic languages, Greek, Armenian, Persian and the various Hindu dialects. The Englishmen who first heard Hindu speech certainly did not suspect that the languages of these dusky people were similar to their own, and that a direct connection or community of origin must at one time have existed between the Englishman and the Hindu. Yet philology has shown such to be a fact, which is now a matter of common knowledge, the entire group of languages spoken from England to India being known as the Indo-European family or Aryan stock.

When a student of Hebrew examines Arabic, it is very quickly evident that the languages have much in common. The speech of the ancient Phoenicians, Syrians and Babylonians, and of the modern Abyssinians, is also similar. This group of languages constitutes what is called the Semitic family. Every dialect within the family possesses obvious similarities to every other Semitic dialect, just as all Aryan languages possess certain words and features among themselves. But no Aryan language has any resemblance to or connection with any Semitic language. It is therefore clear that the ancestors of all the Semitic-speaking nations must have had, at some far distant time, a single common origin, and that at this period they were entirely separate and distinct from the progenitors of the peoples that belong to the Aryan family.

The Turkish language is entirely unconnected with either Aryan or Semitic and belongs to a stock of its own. We know from history that the Turks are recent immigrants in Europe and that they came not very long ago, as the historian reckons, from central Asia. But if the Turkish migrations and invasions had taken place 2,000 years earlier than was the case, we should in all likelihood have had no historical record of the fact, and the historian would erroneously classify the Turks as related to the neighboring Aryan nations—unless he called upon philology to aid him.

It has often been asserted that languages are readily learned and unlearned, and that races put them on and off as a man dons or doffs a garment. But in reality there is probably nothing, not even physical type, that is as permanent as a people's speech.

Thus, even to-day Breton, a pure Celtic speech, maintains itself in France as the every-day language of the people in the isolated province of Brittany—a sort of philological fossil. It has withstood the influence of 2,000 years of contact, first with Latin, then with Frankish German, at last with French. In the same way, its Welsh sister tongue flourishes in spite of the Anglo-Saxon speech of the remainder of Great Britain. The original inhabitants of Spain were mostly of non-Aryan stock. Celtic, Roman and Gothic invasions have successively swept over them and finally left the language of the country Romance; but the original speech also survives the vicissitudes of thousands of years and is still spoken in the western Pyrenees as Basque. Ancient Egypt was conquered by the Shepherd, the Assyrian, the Persian, the Macedonian and the Roman, but whatever the official speech of the ruling class, the people continued to speak Egyptian. Finally, the Arab came and brought with him a new religion, which entailed the use of the Arabic language. Egypt has finally become Arabic-speaking, but until barely a century ago the Coptic language, the daughter of the ancient Egyptian tongue of 5,000 years ago, was kept alive by the native Christians along the Nile; and even to-day it survives in literature. While nations, like individuals, can learn and unlearn languages, as a rule they do so only with the utmost reluctance and with infinite slowness. Speech tends to be one of the most persistent and permanent ethnic characters.

Indian Linguistic Families

The seemingly endless Indian idioms are by classification reducible to about 150 groups or families, almost equally divided between North and South America. The first problem of American ethnology, after determining and mapping these families, is to deduce the probable migrations of peoples that can be inferred and the connection which existed between different tribes. The second task is to carry out similar inquiries within the bounds of each group or family, and in this way to ascertain the minor or more recent affiliations and movements.

The number of languages is large; the aboriginal population was relatively sparse; the necessary consequence is an unusually small number of people per distinct language. In California, where the linguistic diversity reached its height, there were spoken about 135 idioms belonging to 21 families. The total Indian population was 150,000 or a little less—an average for each dialect of almost exactly 1,000 souls, and only 7,000 for each linguistic family. There is something incongruous in comparing the tongue of a paltry 7,000 uncivilized people with, for instance, the whole group of Aryan languages that are the birthright of hundreds of millions of people of the most important nations. Yet to the ethnologist such comparisons are a necessity, for each group of related languages, whether extending only over a little valley, or spreading from continent to continent, is an ultimate unit in itself, which can not be brought into connection with the other or with any other group. Historically the small family may be as significant as the large, for it represents just as separate an origin.

The Great Uto-Aztekan Stock

Perhaps the best known and most important single tribe in America were the Aztecs, who founded and held the city of Mexico and ruled from there over a large part of the modern republic of that name. Excepting perhaps the Incas of Peru, they were the most powerful nation in the new world at the time of its discovery and conquest. Their

PSM V78 D513 North american ute aztecan language family.png

civilization, though for the most part borrowed from other tribes rather than invented, was also of the highest. As to their own origin the Aztecs had certain traditions, according to whose testimony they came from a point in the north, called Aztlan, less than a thousand years ago; in other words, some four or five centuries before the overthrow of their empire by Cortez.

While historians have usually accepted this native tradition, philological evidence renders it very improbable. The Aztec language, more properly called Nahuatl, is the southernmost of a trailing chain of related dialects extending through the length of Mexico and the Great Basin region of the United States. Being at the southernmost extremity of this chain, we have every reason to believe that the Aztecs have moved southward—just as it is natural that the Hindus, who are the easternmost of the Aryans, entered India from the west, and the Celts, who are the westernmost, came into their territories from the east. But if the Aztecs had come from Sonora or adjacent parts of northern Mexico as late as four or five centuries before the discovery, their language should still be very similar to the dialects of those districts. This is not the case. Aztec and the languages of northern Mexico are related, but the relationship is undoubtedly distant. In other words, the Aztecs separated from the Indians of northern Mexico so long ago that their language became considerably changed, and there is every reason for believing that they have maintained a separate existence for very much more than 500 years, just as it is a moral certainty that the ancient people speaking Sanskrit, Greek, Latin and Gothic broke loose from one another more than five centuries before we first hear of any of them. Languages do not change over night. In other words, because Aztec is a member, but a detached and divergent member, of the great Uto-Aztekan family, it is necessary to conclude that the Aztecs came from the north indeed, but came at a very ancient period.

Cliff Dwellers and Pueblos

In New Mexico and Arizona there formerly lived the Cliff Dwellers, and have lived in historical times the Pueblo Indians, builders of large villages of stone, and constructors of irrigation ditches and other remains of a monumental character. These relics so far surpass anything else found in the United States that the superiority of the Pueblos over all their neighbors is unquestioned. This superiority has led to their being brought into connection with the Aztecs, as the nearest nation that had risen to a similar proficiency in arts and industries.

The Cliff Dwellers and the Pueblos are, however, known to be practically identical in their arts, implements, architecture and even religion—so far as idols and symbols and other visible remains can make the nature of an ancient religion evident. The two peoples are clearly only ancient and modern strata of one race. If, therefore, the ancient Cliff Dwellers were Aztecs, the Pueblos should still show in their language close kinship with the Aztecs. This is not the case, the Pueblo Indians, as a class, not being in any way related in speech to the Uto-Aztekan family. It accordingly follows that the popular identification of Cliff Dwellers and Aztecs is based only on ignorance or imagination, and that the weight of historical evidence is adverse to this view.

The historic development of the great Uto-Aztekan family has been determined still farther. One branch comprises a number of tribes in California. Until recently all these tribes were believed to have been the result of a single immigration into the state. It is now clear that they represent three distinct strata. One mass of them has been resident in southern California for a very long time, long enough for the originally uniform language to divide into several dialects. Another body came at a different time, or by a different route, into the Sierra Nevada Mountains of central California. Whether this movement was earlier or later than the first mentioned we can not yet tell, but it is certain that it was distinct. The third stratum is represented by a recent movement from Nevada westward into the eastern parts of California; but even this was entirely prehistoric.

The Algonkin Family on the Atlantic

Another of the great linguistic families of North America is the Algonkin, one of the first to be known. To this large stock belonged Powhatan, Pocahontas and the other Indians among whom the English settlers of Virginia formed their colonies. Other Indians of the same family formed their treaty with William Penn, sold Manhattan Island to the Dutch, met the Pilgrims from the Mayflower, and learned to read Eliot's bible. Most of eastern Canada, the Ohio Valley, the Great Lake region and the country north to Hudson Bay, were also occupied by Algonkin tribes. Separated from all these, and far to the west of the Mississippi in the great plains at the base of the Rockies, lived three groups of Algonkins that at one time or another had evidently made their way there from the original eastern home. These were the Blackfeet, Arapaho and Cheyenne.

In historic times the Cheyenne and Arapaho have usually been allies and closely associated. They are to-day on the same reservation. But all the inferences made as to a joint migration of the two tribes from their original eastern home have proved mistaken. The Cheyenne language is closely similar to the dialect of the Ojibway and other tribes of the Great Lake region. The Arapaho is more different—so much so, in fact, that when vocabularies of it were first recorded, its essentially Algonkin character was not recognized. It follows that the Arapaho represent an ancient and the Cheyenne a recent separation from the tribes farther east. The third group in the plains, the Blackfeet, have specialized their dialect to about the same extent as the Arapaho, but in different ways. While they, therefore.
PSM V78 D516 North american inuit algonquin and athabascan language map.png

branched off at about the same time as the Arapaho, it is clear that they have been distinct from them ever since.

Conservatism of Indian Languages

It has often been said that the languages of Indians and other uncivilized peoples, in fact all languages that are not fixed by writing, change very rapidly. It has been declared that in the course of a generation or two such idioms alter to an extent that men could not understand the talk of their grandfathers, and that in consequence a very few centuries would suffice to alter the features of a language bo thoroughly that its original relationship with kindred languages could no longer be ascertained. All such statements are utterly wild, and there is a mass of evidence to contradict them.

Immediately after the Spanish conquest the Aztec language was written down. Documents were recorded in it and extensive grammars and dictionaries prepared. These grammars and dictionaries are perfectly correct and entirely applicable to the Aztec language as it is spoken to-day. The same is true of the various Maya dialects of Yucatan. We possess records going back two centuries and more of Eskimo, Algonkin, Iroquois and other languages of the United States and Canada as well as of South American tongues. In no instance is any notable change observable. It may in fact be doubted whether most Indian languages have changed as much in pronunciation in the last three hundred years as English has since the time of Skakespeare.

Of course the vocabularies recorded some centuries ago and those written down recently are often far from identical, but the principal differences of this sort must be laid to the imperfect and often curious systems of orthography used. Almost all Indian languages contain at least some sounds that do not occur in the languages of Europe, The Spanish conqueror or the French explorer would represent these unfamiliar sounds with different letters than the subsequent English settler or German scientist. In fact differences fully as great as those between old and modern vocabularies can be found in lists of words taken down in the same period in recent times, by different observers, particularly if these observers were of different nationality. It is probable that the superstition as regards the alleged rapid change of Indian languages is due largely to this cause.

The conservatism of American languages is brilliantly illustrated by the Athabascan family, another of the great linguistic stocks of North America. All the Athabascan dialects are remarkably close, so that a person acquainted with one could learn to understand another in a very short time. The same grammatical processes continue through all of them with almost no change. Yet some of the Athabascan tribes occupy the interior of Alaska and the northwestern parts of Canada. Two branches are in the great plains: the Sarsee, closely affiliated with the Blackfeet, and the Kiowa-Apache, almost amalgamated with the Kiowa though retaining their own speech. In New Mexico and Arizona are the Navaho and Apache. In the interior of British Columbia, just south of Puget Sound in Washington, along the coast of Oregon, and in northwestern California, are other areas, each separated from the other, in which Athabascan was spoken. The tribes belonging to the family are scattered over parts of an area measured by more than forty degrees of latitude and sixty of longitude and embracing at least half of North America. Their original center of dispersion is unknown, but wherever they came from in the first place it is clear that it must have taken them a very long time to force their way individually over thousands of miles, over mountains and rivers, and constantly crowding aside hostile tribes as they moved from one residence to a new home. Here again, as in all the historical conclusions which it is possible to draw from linguistic conditions in America, we are dealing with periods measurable at least by thousands of years; and yet in all this long lapse of time the Athabascan dialects have changed but slightly and superficially.

The Eskimo

The Eskimo have often been proclaimed as an Asiatic people. While confined to the shores of Arctic America, their east and west range is tremendous. If one follows the coast, as they must have done in their migrations, the distance between their eastern and western outposts in Greenland and Alaska is at least 5,000 miles. Yet over this whole stretch the language is so uniform that any one dialect is almost entirely intelligible to the people of regions thousands of miles away. The only divergent language belonging to the Eskimo stock is that of the Aleutian Islands. Where the Eskimo came from is still a moot problem, but as there is nothing in Asia to which their language bears any relationship, their Asiatic origin must at best be viewed as doubtful.

How the Languages Sound

Many popular misconceptions are still prevalent as to the nature of Indian languages. It is commonly supposed that they are characterized by strange and harsh sounds such as "clicks" and "gutturals." On examination the so-called clicks turn out to be nothing but a form of l produced more with one side of the tongue than the other and sounding nearly like tl or hi. This sound is perfectly well known in Welsh and in many other languages of the old world. The guttural sounds also are generally not abnormal, and often less conspicuous than in Hebrew and Asiatic languages. As a rule we may state that no native American language possesses any sound formations that can not be exactly paralleled and duplicated in one or more languages of the old world. What is more, it need hardly be said that among a thousand or more languages and dialects there is opportunity for every range of variation, and any attempt to characterize the phonetics of all Indian languages by one term or by a single description must necessarily be fallacious. As a matter of fact there are many forms of native speech that are exceedingly smooth, harmonious and pleasing even to English ears. On the whole the American Indian finds English as full of strange sounds and difficult sound-combinations as we think the Indian languages to be when first we hear them.

Writing of Indian

No American language was written in a native alphabet. So far as the Indians possessed a means of visible communication, it was by picture writing. In the highest development of this, in Mexico, the picture writing took on to a certain degree, but only partially, a phonetic character. Pictures and symbols were sometimes interpreted as such, and at other times read as sounds, almost exactly as in the rebuses with which we amuse idle moments. Even then, however, the characters usually represented whole words, or at best syllables, and as they did not stand for individual sounds they were never true letters, and did not form an alphabet properly speaking.

All Indian philology accordingly rests on an. oral learning of the languages, and all writing of them has had to be in systems applied by the investigator from other languages, or specially devised by him. The former was the earlier and less satisfactory method. The Spaniard used the Roman alphabet with its Spanish values, the Englishman and American the letters of English. Where sounds were encountered which are not present in these languages, they were usually either omitted, or represented by a character whose customary value somewhat resembled the sound in question.

More recent studies have generally been based upon a systematic and scientific modification of the Roman alphabet. In this certain principles have now been universally accepted for half a century. The most important of these are three.

First, every character or letter must represent one and only one sound. Second, each sound, whenever it occurs, must be denoted by one and the same character. Third, single sounds must be written by single letters, and vice versa, double letters are used only for combinations of sounds. If these principles are strictly adhered to, it does not much matter what characters or modifications of the Roman letters are employed, as long as the investigator is sufficiently conversant with the language not to confuse those sounds which are somewhat similar; and provided also that he furnishes a key or explanation giving the exact phonetic value of every character employed by him. In the choice of characters there are, however, certain preferences. English k and c, for instance, are usually only two different ways of writing the identical sound. In any scientific system of orthography k is preferable because it has the same value in every European language that uses the Roman alphabet, as well as in Greek and the alphabets derived from it. The letter c, however, stands for a great variety of different sounds. In English and French it represents not only the sound of k, but of s, in Spanish th, in German ts, and in Italian, in certain cases, ch. K, which can not be misunderstood, is therefore always used in scientific systems.

In the same way the five vowel characters are pronounced in almost exactly the same way in the great majority of the languages of Europe. Philology, therefore, uses these letters exclusively with their "continental" values rather than with the English sounds, which are quite specialized and which sometimes require two letters, like ee or oo, to represent a single sound, and in other cases express a diphthong or double sound, such as a-i, by the single letter i.

In general very few students of American languages employ precisely the same set of modifications of the Roman alphabet, for the reason that the great majority of them are working with different languages, whose sounds are unlike, so that precisely the same set of diacritical marks would be inappropriate and even inaccurate. The foundation of the system is, however, universally accepted, and may be roughly described as consisting of the vowel characters with their continental values, the consonantal characters with their English values, plus diacritical and typographical modifications to meet particular requirements.

Number of Words

There has been particularly great misapprehension as to what may be called the extent or size of Indian languages—the range of their vocabulary. This is not surprising in view of the fact that similar misstatements are still current as to the number of words actually used by single individuals of civilized communities. It is true that no one, not even the most learned and prolific writer, uses all the words of the English language as they are found in an unabridged dictionary. All of us understand a great many words which we habitually encounter in reading and may even hear frequently spoken, but of which our speech faculties for some reason have not made us master. In short, every language, being the property and product of a community, possesses more words than can ever be used by a single individual, the sum total of whose ideas is necessarily much less than those of the whole body. Added to this are a certain mental sluggishness which restricts most of us to a greater or less degree, and the force of habit. Having spoken a certain word a number of times, our brain becomes accustomed to it and we are apt to employ it to the exclusion of its synonyms.

The degree to which all this affects the speech of the normal man has, however, been greatly exaggerated. Because there are, all told, including technical terms, a hundred thousand or more words given in our dictionaries, and because Shakespeare in all his writings used only fifteen thousand different words, and Milton only six thousand, it has been concluded that the average man, whose range of thought and power of expression is immeasurably below that of Shakespeare and Milton, must use an enormously smaller vocabulary. It has been stated that the average English peasant goes through life without ever using more than six or seven hundred words, that the vocabulary of Italian grand opera is only about three hundred words, and that most of us do well if we know a couple of thousand words. If such were the case it would only be natural that the uncivilized Indian, whose life is so much simpler, and whose knowledge more confined, should be content with an exceedingly small vocabulary.

It is, however, certain that the figures just cited are very erroneous. If any one who considers himself an average person will sit down and make a list or rough estimate of his speaking vocabulary, he will find it to be far above a thousand. It may safely be said that the so-called "average man" knows, and on occasion uses, the names of at least a thousand different things; in other words, that his vocabulary possesses more than a thousand nouns alone. To these must be added the verbs, of which every one employs at least several hundred; adjectives; pronouns; and the other parts of speech, the short and familiar words that are absolutely indispensable to all communication in any language. It may be safely estimated that it is an exceptionally ignorant and stupid person in any civilized country that has not at his command a vocabulary of at least two thousand words, and probably the figure in the normal case is a great deal higher.

When any one has professed to declare on the strength of his observation that a particular Indian language consists of only a few hundred terms, he has displayed chiefly his ignorance. He has either not taken the trouble to exhaust the vocabulary, or has not known how to do so. It is true that the traveler or settler can usually converse with natives to the satisfaction of his own needs with a knowledge of only two or three hundred words. Even the missionary can do a great deal with this stock if it is properly chosen. But of course it does not follow that because the white man in most cases has not learned more of a language, that there is no more. On this point the testimony of the philologist or student, who has made it his business to learn all the language as nearly as may be, is the only evidence that can be considered.

If now we review the Indian languages that have been most thoroughly explored, so to speak, and of which dictionaries are in existence that are even tolerably representative, as of Aztec, Maya, Algonkin, Eskimo, Sioux and several other idioms, it is found that all of these contain 5,000 words, and some considerably exceed this number. What is more, we discover that professions of an incomplete knowledge of a language usually come from the very men who have compiled these dictionaries or who have given years to the study of a language. It is the old story that it is only by increased information that one obtains a perception of one's ignorance. The words are there in the Indian languages; it is only when we have learned several thousand that we begin to realize how many there must still be which are unrecorded. It may safely be said that every American Indian language, whether or not it has yet been studied, possessed before coming in contact with white civilization a vocabulary of at least 5,000 different native words.

How the Grammar is Ascertained

Just as the Indian speaks sounds without being able to represent them in writing, and just as he possesses thousands of words without suspecting it, he also follows complex and intricate rules of grammar without being in the least aware of the fact. There is of course nothing strange in this. We are so accustomed to being taught grammar in school that we often allow ourselves to slide into the hasty opinion that we speak and write grammatically on account of this training. There are, however, perfectly illiterate and uneducated people, who, merely through association with those who talk grammatical English, speak with entire correctness. The first grammarians among the Greeks and Hindus did not invent the rules governing speech in their tongues, but only perceived and set down in systematic shape the grammatical forms and constructions already existing in those languages. So it is only a hasty judgment that would conclude that Indian languages are without grammar or form, merely because the Indian does not know that there is such a thing as grammar.

The Indian's ignorance, however, brings it about that the structure of no Indian language can be learned ready made, but has to be gradually explored and worked out step by step. With good interpreters this is a fascinating pursuit, and with proper philological training it is often not as difficult as might at first seem, though it is always a laborious and lengthy task on account of the wealth of the languages and the intricacy of their structure.

For instance, when forms like the following are obtained:

l-emlu-i I eat
m-emlu-i you eat
l-emlu-ya I ate
m-emlu-hi you will eat
emlu-hi he will eat

it is obvious on comparing the Indian forms with their English equivalents that the stem emlu is the only element that occurs in every one of these Indian words, and the word eat the only one that is common to all the translations. There can, therefore, be no doubt that emlu means "to eat." In the same way comparison shows that wherever we have the English pronoun "I," the Indian language in question possesses the prefix l-. Similarly "you" is the equivalent of the prefix m-, while "he" does not seem to be expressed. A suffix -i occurs when the English rendering is in the present tense, -ya for the past, and -hi for the English future. These five phrases, if we can rely on their having been accurately translated, therefore reveal not only a verb stem, but three pronominal elements and three tense elements. They show, furthermore, that person in the verb is expressed by prefixes, instead of by independent words, as in English, or by endings, as in Latin; and that tense is denoted by suffixes, as in most other languages. In other words, we have derived from these examples a partial idea of that most difficult element in all grammars, the conjugation of the verb.

It is, however, not always as plain sailing as this. The average Indian, even if he has been an official interpreter, has been accustomed to give only the gist or substance of what he has to translate. He has never been troubled with the finer distinctions of tense, mode, number and case, some of which are quite abstract. He is very apt to slur these distinctions over, and to give an approximate instead of an exact translation; so that it is usually necessary to obtain a great number of examples, and patiently compare them, before any positive deductions can be made with safety. In many tribes even the best interpreter's power of expressing himself accurately in English is quite limited, even though he may understand an ordinary conversation perfectly well. If his own language makes no distinction between singular and plural, as not infrequently happens, he uses the English plural and singular indiscriminately. Many Indian languages lack gender and express "he" and "she" by the same pronoun. Most Indians, unless they have gone to school for some time, fail to observe this distinction, and even the school graduate in his unguarded moments is apt to relapse into the habit of calling a woman "he." When "he," "she," "him," "her," "it," "they" and "them" are all expressed by the one general pronoun "him," the investigator has met a serious difficulty.

His only recourse in such an event is to desist from the attempt to obtain exact translations of individual phrases or detached sentences, and to write down from dictation narratives or other continuous texts, of some length, subsequently getting these translated as nearly as may be word for word. Even if the translations are inaccurate in detail, they will be enough to give the drift of the story. Then, by knowing the context, the student is often able to correct the faulty expression of his interpreter. By the context he will know whether the pronoun refers to a man or a woman, to one person or several, and whether it is in the subjective or objective case. A single narrative or description may be of but little aid, but when a considerable series has been obtained, and has been carefully analyzed, he has in hand sufficient material to determine almost any point, provided he gives it proper time and consideration. It is for this reason that the collecting of texts in Indian languages has been carried on to so great an extent of recent years, and is justly looked upon as a basis of all analysis of Indian languages that pretends to any thoroughness or completeness.

The Phonograph

Great hopes have often been placed in the phonograph, but except as an indirect accessory, the instrument has proved of no service at all to the student of Indian languages, invaluable though it may be for recording aboriginal music. The phonograph still reproduces sound with too great imperfection. When we hear a record in our own language we do not observe this fact, because we are listening for what we can recognize rather than for those parts of the diction which we fail to recognize. Just as we can understand a person who mutters or whispers or speaks with indistinct articulation, simply because we succeed in hearing the majority of the sounds which he utters, and our imagination and familiarity with the language enable us to supply the missing sounds, until we think we have actually heard the whole—so we do in listening to a speech record from the phonograph. We can follow the whole of a record made in our own language, even if it is mechanically only tolerable; but we can hardly write down correctly a single word of a record made in an entirely foreign language. This may seem strange, but can easily be verified by experiment.

The only value of the phonograph to the student of Indian languages is the indirect one of assisting him in the procuring of texts. The Indian informant has every opportunity to speak as naturally and rapidly as he wishes. When a body of such records has been obtained, they can be gone over sentence by sentence, and if need be, word for word, with an interpreter, who speaks as slowly as may be necessary for correct dictation. By this double method the most satisfactory texts can be obtained. Though the labor is increased, and the instrument serves only for the first step of the process, the final product is a perfect written text.

"Gluing Together"

Many attempts have been made to describe briefly and generally the grammatical structure of Indian languages. It has been commonly said that the languages, as a class, are agglutinating, that they "glue" one element to another to form words. But just such pasting together of word elements into words occurs in many of the Aryan languages, in fact in forms of speech all over the world. It is hard to see why on account of some subsidiary difference the same process should be called "inflection" when it takes place in our own language, and "agglutination" when it occurs in Indian or other idioms. It is probably only a desire to set off ourselves from all other people that is at the bottom of the distinction between "inflecting" and "agglutinating" languages.

Polysynthetis

A different description of American languages is contained in the word "polysynthetic," meaning a high degree of combination. There is no question but that many Indian languages are extremely polysynthetic, uniting into a single word, especially in connection with the verb stem, many elements of expression which in English and even in Latin and Greek have to be expressed by a number of separate words. Thus the English sentence "I will roll it there with my foot" would be expressed in the Washo language, from which the preceding illustrations have also been drawn, by a single word containing eight syllables, and divisible into six distinct elements.

di-liwi-  lup-  gic-  ue-  hi
I- foot-with-roll-thither-will

What is particularly characteristic of the polysynthetic process as exemplified by this word, is that most of the elements as used here can not stand as separate words. They are thus more like our prefixes and suffixes and are more properly word-elements than words in themselves. Thus if the Washo wishes to say "I," as in answer to the question "Who is it?," he says le; whereas in composition, as in the above long word, "I" is expressed by the prefix di-. The word for "foot" is mayop, yet the element or prefix meaning "foot" in a polysynthetic compound shows no relation whatever to mayop, being liwi. In the same way there or thither as a separate word, as in answer to the question "to where?" is di; in a compound word the suffix -ue is used.

It is necessary to observe that some American languages do not show this peculiar polysynthetic character, but it is true that the majority of them do possess it, and that some carry it to an extreme degree, so that with references to the languages as a class, it can not be denied that they tend to be polysynthetic.

Every variety of grammatical form can, however, be found in the native languages of America, just as they possess a tremendous diversity of words and of phonetic characters. Some of the languages are very simple, others very complex. Some can be readily learned and analyzed, others present great obstacles. In spite of all the work that has been done by ethnologists, missionaries and others, the great majority of languages are still practically unknown. They offer a tempting and almost unlimited field of philological research. Their study is urgent because many have become extinct and most of the remainder are fast perishing before the inroads of English or Spanish; and it is of the utmost importance on account of the aid which it furnishes to history and archeology. Our future knowledge of the history and prehistory of the American Indian will depend more largely on our knowledge of his languages than on any one other thing.