Essays of an Americanist / I. Ethnologic and Archæologic. II. Mythology and Folk Lore. III. Graphic Systems and Literature. IV. Linguistic.

The last example is not fully explained by my authorities; but it shows the verbal change.

Something like this occurs in the Pame dialects. They reveal a manifest indifference to the integrity of the theme, characteristic of polysynthetic languages. Thus, our only authority on the Pame, Father Juan Guadalupe Soriano, gives the preterit forms of the verb “to aid:”

So, of “to burn:”

Knu aum, I burned.

Kuddu du taum, they burned.^[310]

A large number of such changes run through the conjugation. Pimentel calls them phonetic changes, but they are certainly, in some instances, true syntheses.

All these traits of the Othomi and its related dialects serve to place them unquestionably within the general plan of structure of American languages.

THE BRI-BRI LANGUAGE.

The late Mr. Wm. M. Gabb, who was the first to furnish any satisfactory information about it and its allied dialects in Costa Rica, introduces the Bri-Bri language, spoken in the highlands of that State, by quoting the words of Alexander von Humboldt to the effect that “a multiplicity of tenses characterizes the rudest American languages.” On this, Mr. Gabb comments: “This certainly does not apply to the Costa Rican family, which is equally remarkable for the simplicity of its inflections.”^[311]

This statement, offered with such confidence, has been accepted and passed on without close examination by several unusually careful linguists. Thus Professor Friedrich Müller, in his brief description of the Bri-Bri (taken exclusively from Gabb’s work), inserts the observation—“The simple structure of this idiom is sufficient to contradict the theories generally received about American languages.”^[312] And M. Lucien Adam has lately instanced its verbs as notable examples of inflectional simplicity.^[313] The study of this group of tongues becomes, therefore, of peculiar importance to my present topic.

Since Mr. Gabb published his memoir, some independent material, grammatical as well as lexicographical, has been furnished by the Rt. Rev. B. A. Thiel, Bishop of Costa Rica,^[314] and I have obtained, in addition, several MS. vocabularies and notes on the language prepared by Prof. P. J. J. Valentini and others.

The stock is divided into three groups of related dialects, as follows:—

I. The Brunka, Bronka or Boruca, now in southwestern Costa Rica, but believed by Gabb to have been the earliest of the stock to occupy the soil, and to have been crowded out by later arrivals.

II. The Tiribi and Terraba, principally on the head-waters of the Rio Telorio and south of the mountains.

III. The Bri-Bri and Cabecar on the head-waters of the Rio Tiliri. The Biceitas (Vizeitas) or Cachis, near the mouth of the same stream, are off-shoots of the Bri-Bris; so also are the small tribes at Orosi and Tucurrique, who were removed to those localities by the Spaniards.

The Bri-Bri and Cabecar, although dialects of the same original speech, are not sufficiently alike to be mutually intelligible. The Cabecars occupied the land before the Bri-Bris, but were conquered and are now subject to them. It is probable that their dialect is more archaic.

The Bri-Bri is a language of extreme poverty, and as spoken at present is plainly corrupt. Gabb estimates the whole number of words it contains as probably not exceeding fifteen hundred. Some of these, though Gabb thinks not very many, are borrowed from the Spanish; but it is significant, that among them is the pronoun “that,” the Spanish ese.

Let us now examine the Bri-Bri verb, said to be so singularly simple. We are at once struck by Mr. Gabb’s remark (just after he has been speaking of their unparalleled simplicity) that the inflections he gives “have been verified with as much care as the difficulties of the case would admit.” Evidently, then, there were difficulties. What they are, becomes apparent when we attempt to analyze the forms of the eighteen brief paradigms which he gives.

The personal pronouns are

These are both nominative and objective, personal and, with the suffix cha, possessives.

The tenses are usually, not always, indicated by suffixes to the theme; but these vary, and no rule is given for them, nor is it stated whether the same theme can be used with them all. Thus,

Here are three forms for the present, not explained. Are they three conjugations, or do they express three shades of meaning, like the three English presents? I suspect the latter, for under ikiana, to want, Gabb remarks that the form in -etke, means “he wants you,” i. e., is emphatic.

The past aorist has two terminations, one in -na, and one in -e, about the uses and meanings of which we are left equally in the dark.

The future is utterly inexplicable. Even Prof. Müller, just after his note calling attention to the “great simplicity” of the tongue, is obliged to give up this tense with the observation, “the structural laws regulating the formation of the future are still in obscurity!” Was it not somewhat premature to dwell on the simplicity of a tongue whose simplest tenses he acknowledges himself unable to analyze?

The futures of some verbs will reveal the difficulties of this tense:—

In the last example mia is the future of the verb imia, to go, and is used as an auxiliary.

The explanation I have to suggest for these varying forms is, either that they represent in fact that very “multiplicity of tense-formations” which Humboldt alluded to, and which were too subtle to be apprehended by Mr. Gabb within the time he devoted to the study of the language; or that they are in modern Bri-Bri, which I have shown is noticeably corrupted, survivals of these formations, but are now largely disregarded by the natives themselves.

Signs of the incorporative plan are not wanting in the tongue. Thus in the objective conjugation not only is the object placed between subject and verb, but the latter may undergo visible synthetic changes. Thus:

In the latter sentence na is the sign of the past aorist, and the verb in synthesis with it drops its last syllable. The wai Gabb could not explain. It will be noticed that the negative precedes the whole verbal form, thus indicating that it is treated as a collective idea (holophrastically).

Prepositions always appear as suffixes to nouns, which, in composition, may suffer elision. This is strictly similar to the Nahuatl and other synthetic tongues.

Other examples of developed synthesis are not uncommon, as—

The opinion that the Bri-Bri is at present a considerably corrupted and worn-down dialect of a group of originally highly synthetic tongues is borne out by an examination of the scanty materials we have of its nearest relations.

Thus in the Terraba we find the same superfluous richness of pronominal forms which occurs in many South American tongues, one indicating that the person is sitting, another that he is standing, a third that he is walking.^[315]

The Brunka has several distinct forms in the present tense:

Although Bishop Thiel supplies a number of verbal forms from this dialect, the plan of their construction is not obvious. This is seen from a comparison of the present and perfect tenses in various words. The pronouns are—

For instance:

These examples are sufficient to show that the Brunka conjugations are neither regular nor simple, and such is the emphatic statement of Bishop Thiel, both of it and all these allied dialects. In his introduction he states that he is not yet ready to offer a grammar of these tongues, though well supplied with lexicographical materials, and that “their verbs are especially difficult.”^[316]

The Cabecar dialect, in which he gives several native funeral poems, without translations, is apparently more complicated than the Bri-Bri. The words of the songs are long and seem much syncopated.

THE TUPI-GUARANI DIALECTS.

Several writers of the highest position have asserted that these dialects, spoken over so large a portion of the territory of Brazil, are neither polysynthetic nor incorporative. Thus the late Prof. Charles F. Hartt in his “Notes on the Lingoa Geral or Modern Tupi,” expressed himself: “Unlike the North American Indian tongues, the languages of the Tupi-Guarani family are not polysynthetic in structure.”^[317] With scarcely less positiveness Professor Fredrich Müller writes: “The objective conjugation of the Tupi-Guarani does not show the incorporation usually seen in American languages, but rather a mere collocation.”^[318]

It is, I acknowledge, somewhat hazardous to venture an opinion contrary to such excellent authorities. But I must say, that while, no doubt, the Tupi in its structure differs widely from the Algonkin or Nahuatl, it yet seems to present unmistakable signs of an incorporative and polysynthetic character, such as would be difficult to parallel outside of America.

I am encouraged to maintain this by the recent example of the erudite Dr. Amaro Cavalcanti, himself well and practically versed in the spoken Tupi of to-day, who has issued a learned treatise to prove that “the Brazilian dialects present undoubtedly all the supposed characteristics of an agglutinative language, and belong to the same group as the numerous other dialects or tongues of America.”^[319] Dr. Cavalcanti does not, indeed, distinguish so clearly between agglutinative and incorporative languages as I should wish, but the trend of his work is altogether parallel to the arguments I am about to advance.

Fortunately, we do not suffer from a lack of materials to study the Tupi, ancient and modern. There are plenty of dictionaries, grammars and texts in it, and even an “Ollendorff’s Method,” for those who prefer that intellectual (!) system.^[320]

All recent writers agree that the modern Tupi has been materially changed by long contact with the whites. The traders and missionaries have exerted a disintegrating effect on its ancient forms, to some of which I shall have occasion to refer.

O Selvagem i Curso da Lingua Geral. By Dr. Couto de Magalhaes (Rio de Janeiro, 1876).

Turning our attention first to its synthetic character, one cannot but be surprised after reading Prof. Hartt’s opinion above quoted to find him a few pages later introducing us to the following example of “word-building of a more than usually polysynthetic character.”^[321]

Such examples, however, are not rare, as may be seen by turning over the leaves of Montoya’s Tesoro de la Lengua Guarani. The most noticeable and most American peculiarity of such compounds is that they are not collocations of words, as are the agglutinative compounds of the Ural-Altaic tongues, but of particles and phonetic elements which have no separate life in the language.

Father Montoya calls special attention to this in the first words of his Advertencia to his Tesoro. He says:—“The foundation of this language consists of particles which frequently have no meaning if taken alone; but when compounded with the whole or parts of others (for they cut them up a great deal in composition) they form significant expressions; for this reason there are no independent verbs in the language, as they are built up of these particles with nouns or pronouns. Thus, ñemboé is composed of the three particles ñe, mo, e. The ñe is reciprocal; mo an active particle; e indicates skill; and the whole means ‘to exercise oneself,’ which we translate, ‘to learn,’ or ‘to teach,’ indeterminately; but with the personal sign added, anemboe, ‘I learn.’”

This analysis, which Montoya carries much further, reminds us forcibly of the extraordinarily acute analysis of the Cree (Algonkin) by Mr. James Howse.^[322] Undoubtedly the two tongues have been built up from significant particles (not words) in the same manner.

Some of these particles convey a peculiar turn to the whole sentence, difficult to express in our tongues. Thus the element é attached to the last syllable of a compound gives an oppositive sense to the whole expression; for example, ajur, “I come” simply; but if the question follows: “Who ordered you to come?” the answer might be, ajuré, “I come of my own accord; nobody ordered me.”^[323]

Cavalcanti observes that many of these formative elements which existed in the old Tupi have now fallen out of use.^[324] This is one of several evidences of a change in structure in the language, a loss of its more pliable and creative powers.

This synthesis is also displayed in the Tupi, as in the Cree, by the inseparable union of certain nouns with pronouns. The latter are constantly united with terms of consanguinity and generally with those of members of the body, the form of the noun undergoing material modifications. Thus:

Postpositions are in a similar manner sometimes merged into the nouns or pronouns which they limit. Thus: tenonde, before; guenonde, before him.

It appears to me that the substratum, the structural theory, of such a tongue is decidedly polysynthetic and not agglutinative, still less analytic.

Let us now inquire whether there are any signs of the incorporative process in Tupi.

We are at once struck with the peculiarity that there are two special sets of pronouns used with verbals, one set subjective, and the other objective, several of which cannot be employed in any other construction.^[325] This is almost diagnostic of the holophrastic method of speech. The pronouns in such cases are evidently regarded by the language-faculty as subordinate accessories to the verbal, and whether they are phonetically merged in it or not is a secondary question.

The Tupi pronouns (confining myself to the singular number for the sake of brevity) are as follows:

The verbal affixes are united to the theme with various phonetic changes, and so intimately as to form one word. The grammars give such example as:—

In the first person singular, the two pronominal forms xe and a are usually merged in the synthesis xa; as xamehen, I love.

Another feature pointing to the incorporative plan is the location of the object. The rule in the old language was to place the object in all instances before the verb, that is, between the verb and its subject when the latter was other than a personal suffix. Dr. Cavalcanti says that this is now in a measure changed, so that when the object is of the third person it is placed after the verb, although in the first and second persons the old rule still holds good.^[326] Thus the ancient Tupis would say:

But in the modern tongue it is:

With the other persons the rule is still for the object to precede and to be attached to the theme:

Many highly complex verbal forms seem to me to illustrate a close incorporative tendency. Let us analyze for instance the word,

which means “him whom I teach” or “that which I teach.” Its theme is the verbal mboe, which in the extract I have above made from Montoya is shown to be a synthesis of the three elementary particles ñe, mo, and e; xe is the possessive form of the personal pronoun, “my”; it is followed by the participial expression temi or tembi, which, according to Montoya, is equivalent to “illud quod facio;” its terminal vowel is syncopated with the relative y or i, “him, it”; so the separate parts of the expression are:—

I shall not pursue the examination of the Tupi further. It were, of course, easy to multiply examples. But I am willing to leave the case as it stands, and to ask linguists whether, in view of the above, it was not a premature judgment that pronounced it a tongue neither polysynthetic nor incorporative.

THE MUTSUN.

This is also one of the languages which has been announced as “neither polysynthetic nor incorporative,” and the construction of its verb as “simple to the last degree.”^[327]

We know the tongue only through the Grammar and Phrase-Book of Father de la Cuesta, who acknowledges himself to be very imperfectly acquainted with it.^[328] With its associated dialects, it was spoken near the site of the present city of San Francisco, California.

Looking first at the verb, its “extreme simplicity” is not so apparent as the statements about it would lead us to expect.

In the first place, the naked verbal theme undergoes a variety of changes by insertion and suffixes, like those of the Quiche and Qquichua, which modify its meaning. Thus:

Again:

The author enumerates thirty-one forms thus derived from each verb, some conjugated like it, some irregularly. With regard to tenses, he gives eight preterits and four futures; and it cannot be said that they are formed simply by adding adverbs of time, as the theme itself takes a different form in several of them, aran, aras, aragts, etc. In the reflexive conjugation the pronoun follows the verb and is united with it: As,

where ca is a suffixed form of can, I; ne represents nenissia, oneself; the g is apparently a connective; and the theme is ara. This is quite in the order of the polysynthetic theory and is also incorporative.

Such syntheses are prominent in imperative forms. Thus from the above-mentioned verb, oio, to catch, we have,

in which mit is apparently the second person men, with a postposition tsa, mintsa; while yuts is a verbal fragment from yuyuts, which the author explains to mean “to set about,” or “to get done.” This imperative, therefore, is a verbal noun in synthesis with an interjection, “get done with thy gathering.” It is a marked case of polysynthesis. A number of such are found in the Mutsun phrases given, as:

In this compound cannis, is for can + huas, me + for; yuts is the imperative interjection for yuyuts; the remainder of the word is not clear. The phrase is given elsewhere

Without going further into this language, of which we know so little, it will be evident that it is very far from simple, and that it is certainly highly synthetic in various features.

CONCLUSIONS.

The conclusions to which the above study leads may be briefly summarized as follows:

1. The structural processes of incorporation and polysynthesis are much more influential elements in the morphology of language than has been conceded by some recent writers.

2. They are clearly apparent in a number of American languages where their presence has been heretofore denied.

3. Although so long as we are without the means of examining all American tongues, it will be premature to assert that these processes prevail in all, nevertheless it is safe to say that their absence has not been demonstrated in any of which we have sufficient and authentic material on which to base a decision.

4. The opinion of Duponceau and Humboldt, therefore, that these processes belong to the ground-plan of American languages, and are their leading characteristics, must still be regarded as a correct generalization.

[ADDENDUM.

Shortly after the above essay appeared in the Proceedings of the American Philosophical Society, its arguments and conclusions were vigorously attacked by M. Lucien Adam in the Revue de Linguistique et de Philologie Comparée, Tome XIX (Paris, 1886). He begins by pointing out that examples of incorporation may be found in tongues of the Old World—which has never been denied (see above, pp. 353–4). Having acknowledged the incompleteness of his own definitions, he intimates that those I give are calculated rather to sustain my theory than to prove a linguistic trait. He then proceeds to lengthy and minute criticisms of the analyses I have made of the examples given under the several languages discussed. I am quite willing to concede that with the imperfect grammars and lexicons of these tongues so far published, I may have tripped at times in such analyses; but I am far from acknowledging that all those of M. Adam are correct, and I am quite certain that in some he is mistaken. The question, however, is one not possible to discuss in this place, and I must leave it; but I would refer the earnest student to the acute and learned article of M. Adam, which is much the most thorough yet written on the negative side of the debate.]

Archæologists tell us that the manufacturers of those rude stone implements called palæoliths wandered up and down the world while a period of something like two hundred thousand years was unrolling its eventless centuries. Many believe that these early artisans had not the power of articulate expression to convey their emotions or ideas; if such they had, they were confined to inarticulate grunts and cries.

Haeckel proposed for the species at this period of its existence the designation Homo alalus, speechless man. Anatomists have come forward to show that the inferior maxillary bones disinterred in the caves of La Naulette and Schipka are so formed that their original possessors could not have had the power of articulation.^[330] But the latest investigators of this point have reached an opposite conclusion.^[331] We must, however, concede that the oral communication of men during that long epoch was of a very rudimentary character; it is contrary to every theory of intellectual evolution to suppose that they possessed a speech approaching anything near even the lowest organized of the linguistic stocks now in existence. By an attentive consideration of some of these lowest stocks, can we not form a somewhat correct conception of what was the character of the rudimentary utterances of the race? I think we can, but, as I believe I am the first to attempt such a picture, I offer it with becoming diffidence.

The physiological possibility that palæolithic man possessed a language has, as I have said, been already vindicated; and that he was intellectually capable of speech could, I think, scarcely be denied by any one who will contemplate the conceptions of symmetry, the technical skill, and the wise adaptation to use, manifested in some of the oldest specimens of his art; as for example the axes disinterred from the ancient strata of San Isidro, near Madrid, those found forty feet deep in the post-glacial gravels near Trenton, New Jersey, or some of those figured by De Mortillet as derived from the beds of the Somme in France.^[332] We have evidence that at that period man made use of fire; that he raised shelters to protect himself from the weather; that he possessed some means of navigating the streams; that he could occasionally overcome powerful and ferocious beasts; that he already paid some attention to ornamenting his person; that he lived in communities; and that his migrations were extensive.^[333] In view of all this, is it not highly improbable that he was destitute of any vocal powers of expressing his plans and desires? I maintain that we should dismiss the Homo alalus, as a scientific romance which has served its time.

More than this, I believe that by a judicious study of existing languages, especially those which have suffered little by admixture or by distant removals, we can picture with reasonable fidelity the character of the earliest tongues spoken by man, the speech of the Palæolithic Age.

This primitive utterance was, of course, not the same everywhere. It varied indefinitely. But for all that it is almost certain that in all localities it proceeded on analogous lines of development, just as languages have everywhere and at all times since. By studying simple and isolated languages, those which have suffered least by contact with others, or by alterations in conditions of culture, we can catch some glimpses of the character of man’s earliest significant expression, the “baby-talk of the race,” if I may use the expression. I have gleaned a certain number of such traits in the field of American linguistics, and present them to you as curiosities, which, like other curiosities, have considerable significance to those who will master their full purport.

The question I am about to consider, is, you will observe, quite different from that which concerns itself with the origin of linguistic stocks. Many of these unquestionably arose long after man had acquired well-developed languages, and when the cerebral convolutions whose activity is manifested in articulate expression had acquired a high grade of development through hereditary training. How such stocks may have arisen has been lucidly set forth by my learned friend Mr. Horatio Hale. He demonstrates by many examples that in the present cerebral evolution of man, infants develop an articulate language with the same natural facility that any other species of animal does the vocal utterances peculiar to its kind.^[334]

But in this essay I am contemplating man as he was before hundreds of generations of speaking ancestors had evolved such cerebral powers.

I begin with some observations on the phonetic elements. These are no other than what we call the alphabet, the simple sounds which combined together make up the words of a language. In all European tongues, the mere letters of the alphabet, by themselves, have no meaning and convey no idea; furthermore, their value in a word is fixed; and, thirdly, arranged in a word, they are sufficient to convey its sound and sense to one acquainted with their values.

Judged by certain American examples, all three of these seemingly fundamental characteristics of the phonetic elements were absent in primitive speech, and have become stable only by a long process of growth. We find tongues in which the primary sounds are themselves significant, and yet at the same time are highly variable; and we find many examples in which they are inadequate to convey the sense of the articulate sound.

As exemplifying these peculiarities I take the Tinné or Athapascan, spoken widely in British America, and of which the Apache and Navaho in the United States are branches. You know that in English the vowels A, E, I, O, U, and the consonants, as such, F, S, K, and the others, convey to your mind no meaning, are not attached to any idea or train of ideas. This is altogether different in the Tinné. We are informed by Bishop Faraud,^[335] a thorough master of that tongue, that its significant radicals are the five primitive vowel sounds, A, E, I, O, U. Of these A expresses matter, E existence, I force or energy, O existence doubtful, and U existence absent, non-existence, negation or succession. These vowels are “put in action,” as he phrases it, by single or double consonants, “which have more or less value in proportion as the vowel is more or less strong.” These consonantal sounds, as we learn at length from the works on this language by Father Petitot, are also materially significant. They are numerous, being sixty-three in all, and are divided into nine different classes, each of which conveys a series of related or associated ideas in the native mind.

Thus, the labials express the ideas of time and space, as age, length, distance, and also whiteness, the last mentioned, perhaps, through association with the white hair of age, or the endless snowfields of their winter. The dentals express all that relates to force terminating, hence uselessness, inanity, privation, smallness, feebleness; and also greatness, elevation, the motor power. The nasals convey the general notion of motion in repetition; hence, rotation, reduplication, gravitation, and, by a singularly logical association, organic life. The gutturals indicate motion in curves; hence, sinuousness, flexibility, ebullition, roundness, and by a linear figure different from that which underlies the Latin rectitudo, justness, correctness. The H, either as an aspirate or an hiatus, introduces the ideas of command and subjection, elevation and prostration, and the like.^[336]

You will observe that in some of these cases the signification of a sound includes both a notion and its opposite, as greatness and smallness. This is an interesting feature, to which I shall refer later.

Turn now to another language, the Cree. Geographically it is contiguous to the Tinné; but, says Bishop Faraud, who spoke them both fluently, they resemble each other no more than the French does the Chinese. Nevertheless, we discover this same peculiarity of materially significant phonetic elements. Howse, in his Cree Grammar, observes that the guttural K and the labial W constitute the essential part of all intensive terms in that language, “whether the same be attributive, formative, or personal accident.” Indeed, he maintains that the articulate sounds of the Cree all express relative powers, feebleness or force, independent of their position with reference to other sounds.

You may inquire whether in the different groups of American tongues the same or a similar signification is attached to any one sound, or to the sounds of any one organ. If it were so, it would give countenance to those theories which maintain that there is some fixed relation between sound and sense in the radicals of languages. I must reply that I have found very little evidence for this theory; and yet some. For example, the N sound expresses the notion of the ego, of myself-ness, in a great many tongues, far apart geographically and linguistically. It is found at the basis of the personal pronoun of the first person and of the words for man in numerous dialects in North and South America. Again, the K sound is almost as widely associated with the ideas of otherness, and is at the base of the personal pronoun of the second person singular and of the expressions for superhuman personalities, the divine existence.^[337] It is essentially demonstrative in its power.

Again, in a long array of tongues in various parts of the world, the subjective relation is expressed by the M sound, as has been pointed out by Dr. Winkler; and other examples could be added. Many of these it is impossible to attribute to derivation from a common source. Some writers maintain that sounds have a subjective and fixed relation to ideas; others call such coincidences “blind chance,” but these should remember that chance itself means merely the action of laws not yet discovered.

You might suppose that this distinction, I mean that between self and other, between I, thou and he, is fundamental, that speech could not proceed without it. You would be mistaken. American languages furnish conclusive evidence that for unnumbered generations mankind got along well enough without any such discrimination. One and the same monosyllable served for all three persons and both numbers. The meaning of this monosyllable was undoubtedly “any living human being.” Only after a long time did it become differentiated by the addition of locative particles into the notions, “I—living human being,” “Thou—living human being,” “He—living human being,” and so on. Even a language spoken by so cultured a people as the ancient Peruvians bears unmistakable traces of this process, as has been shown by Von Tschudi in his admirable analysis of that tongue; and the language of the Baures of Bolivia still presents examples of verbs conjugated without pronouns or pronominal affixes.^[338]

The extraordinary development of the pronouns in many American languages—some have as many as eighteen different forms, as the person is contemplated as standing, lying, in motion, at rest, alone, in company, etc., etc.—this multiplicity of forms, I say, is proof to the scientific linguist that these tongues have but recently developed this grammatical category. Wherever we find overgrowth, the soil is new and the crop rank.

In spite of the significance attached to the phonetic elements, they are, in many American languages, singularly vague and fluctuating. If in English we were to pronounce three words, loll, nor, roll, indifferently as one or the other, you see what violence we should do to the theory of our alphabet. Yet analogous examples are constant in many American languages. Their consonants are “alternating,” in large groups, their vowels “permutable.” M. Petitot calls this phenomenon “literal affinity,” and shows that in the Tinné it takes place not only between consonants of the same group, the labials for instance, but of different groups, as labials with dentals, and dentals with nasals. These differences are not merely dialectic; they are found in the same village, the same family, the same person. They are not peculiar to the Tinné; they recur in the Klamath. Dr. Behrendt was puzzled with them in the Chapanec. “No other language,” he writes, “has left me in such doubt as this one. The same person pronounces the same word differently; and when his attention is called to it, will insist that it is the same. Thus, for devil he will give Tixambi and Sisaimbui; for hell, Nakupaju and Nakapoti.”^[339] Speaking of the Guarani, Father Montoya says: “There is in this language a constant changing of the letters, for which no sufficient rules can be given.”^[340] And Dr. Darapsky in his recently published study of the Araucanian of Chile gives the following equation of permutable letters in that tongue: