POLYSYNTHESIS IN THE LANGUAGES OF THE
AMERICAN INDIANS.

BY J. N. B. HEWITT.

In the early part of this century Peter S. Duponceau announced his conviction, obtained from a cursory study of the scanty and imperfect linguistic material accessible to him, that the grammatic phenomena of the known tongues of the American Indians are characterized by a common ground plan, or, adopting a phrase of Maupertuis, a "plan of ideas." This plan he called polysynthetic or syntactic, and defined it as follows:

"A polysynthetic or syntactic construction of language is that in which the greatest number of ideas are comprised in the least number of words. This is done principally in two ways. 1. By a mode of compounding locutions which is not confined to joining two words together, as in Greek, or varying the inflection or termination of a radical word, as in most European languages, but by interweaving together the most significant sounds or syllables of each simple word, so as to form a compound that will awaken in the mind at once all the ideas singly expressed by the words from which they are taken. 2. By an analogous combination [of] the various parts of speech, particularly by means of the verb, so that its various forms and inflections will express not only the principal action, but the greatest possible number of the moral ideas and physical objects connected with it, and will combine itself to the greatest extent with those conceptions which are the subject of other parts of speech, and in other languages require to be expressed by separate and distinct words. Such I take to be the general character of the Indian languages." *

He elsewhere says:

"I am inclined to believe that these forms are peculiar to this part of the world, and that they do not exist in the languages of the old world." †

In an essay, which won, in 1833, the Volney prize of the Institute of France, he says:

"À l'aide d'inflexions, comme dans les langues grecque et latine, de particules, affixes et suffixes, comme dans le copte, l'hébreu et les langues dites sémitiques, de la jonction de particules significatives, comme

*Transactions of the Historical and Literary Committee of the American Philosophical Society, held at Philadelphia, for promoting useful knowledge, vol. i, p. xxx. † Loc. cit., p. 370.

dans le chinois, et enfin de syllabes et souvent de simples lettres intercalées à l'effet de réveiller une idée de l'expression de laquelle cette lettre fait partie, à quoi il faut ajouter l'ellipse, qui fait sousentendre, les Indiens de l'Amérique sout parvenus à former des langues qui comprennent le plus grand nombre d'idées dans le plus petit nombre de mots possible. Au moyen de ces procédés ils peuvent changer la nature de toutes les parties du discours; du verbe, faire un adverbe ou un nom; de l'adjective ou du substantif, un verbe; enfin, tous les auteurs qui ont écrit sur ces langues avec connaissance de cause, depuis le nord jusqu'au sud, affirment que, dans ces idiomes sauvages, on peut former des mots à l'infini.”

If a general principle of the kind here described could be established it would be of the utmost importance to the students of comparative grammar. This, however, can be done only by a careful and thorough analysis by the modern methods of linguistics of every language concerned, an analysis which has not yet been made. For such an analysis trustworthy and sufficient data must also be at hand.

The lexic and syntactic material relating to these languages is, in some instances, quite extensive, consisting mostly of short vocabularies, translations of the Holy Scriptures or portions thereof, and more or less pretentious lexicons and grammars; but, for the purpose of comparative or other study, these are so faulty and misleading and so warped by erroneous theories and misapprehensions that they are of small value and of precarious utility in morphologic study. The learned Father Cuoq, equally well-versed in Iroquoian and Algonquian speech, says:

"Que penser de certaines traductions des Stes. Écritures? Ceux qui ont tant soit peu étudié les différentes portions de la Bible traduites dans les langues indiennes de l'Amérique par les soins de certaines Sociétés Bibliques, en trouvent la traduction-il m'est pénible de le dire-vraiment pitoyable. Ce n'est rien moins qu'une profanation de la parole de Dieu; et je suis assuré pour ma part que les membres eux-mêmes de ces sociétés seraient les premiers à répudier leurs pauvres publications et à les condamner aux flammes, s'ils connaissaient les incorrections, les inexactitudes, les solécismes, les barbarismes, et les contre-sens dont elles fourmillent."†

Duponceau had no ready means of testing the work of his chief authorities, and so was compelled to accept their unsupported state

*Mémoire sur le système grammatical des langues de quelques nations indiennes de l'Amérique du nord. Paris, 1838, p. 89.

"Jugement erroné de M. Ernest Renan sur les langues sauvages," p. 105.

ments and deductions. He drew his information of the Iroquoian language from the works of Zeisberger and Pyrlaeus, chiefly those of the former. A careful and unbiased examination of Zeisberger's work shows that the worthy missionary had at best only a superficial and precarious knowledge of that language, for he lacked the very elementary acquaintance with it which would have enabled him invariably to distinguish its words from their derivatives and from its sentences and phrases.

The method of inflections, which is common to European and other tongues, need not detain us; the method of intercalation or interweaving vocal elements claimed to be peculiarly characteristic of the polysynthetic scheme demands some consideration. Had it a substantial basis of fact it would indeed serve to mark off from all others those languages in which it was found to prevail. The use of a process so singular and abnormal in its operation can be established only by the evidence of unequivocal facts. The data adduced as proof that such a method of combining vocal elements is one of the most characteristic traits of all known Indian tongues are of the most questionable character. This process is not a part of Iroquoian grammar, nor has a satisfactory example of it been cited from Algonquian speech, and Rev. J. Owen Dorsey states that it does not find a place in the Siouan grammatic processes; hence it follows that the languages of these three great stocks are not polysynthetic within the meaning of this term as used by Duponceau, because they do not use the so-called "artificial elements" nor the alleged process of "interweaving together" or "intercalation" of vocables, which alone constitute the characteristic traits of the supposed "polysynthetic construction." This raises the presumption that careful study will show that other less-known Indian tongues, which, like the three named above, have been classed as polysynthetic by Duponceau and his disciples, are not founded on that theoretic plan; because wherever the syntactic and morphologic processes have been ascertained from accurate and sufficient data they have been found at variance with the polysynthetic processes, and they likewise differ greatly among themselves in their ground plans. It has, in fact, been found that those Indian languages whose lexic and syntactic phenomena have been thoroughly analyzed have not, as Duponceau maintained, a peculiar construction of language, in which the greatest number of ideas are comprised in the least number of words," which is the motive or object of his conjectured ground plan or "plan of ideas."

Duponceau further says:

"Les Indiens, surtout ceux qui sont chasseurs et nomades, n'ont pus une tête bien analytique. Ils se sont bientôt embrouillés dans la formation de leurs mots : recevant leurs idées en groupes, ainsi que la nature nous les présente, ils ont voulu les exprimer à la fois avec toutes leurs parties, telles qu'ils les apercevaient.* Ont-ils voulu, par exemple, donner un nom à un certain arbre, ils n'ont pas pensé à le désigner simplement par le fruit, ou par quelque autre apparence unique; mais ils ont dit: l'arbre portant tel fruit et dont les feuilles ressemblent à telle chose, et ils ont cherché à exprimer tout cela par un seul mot. Mais comment faire? S'ils joignaient tous ces mots ensemble, ils en auraient un nouveau d'une longueur énorme; et puis, leur nouvelle langue, abondant en consonnes, n'etait pas heureusement formée pour une pareille jonction. Alors ils ont pris quelque chose de chaque mot, et par la réunion et l'intercalation des syllables, et même de sons simples tirés de la phrase qu'ils avaient choisie, ou plutôt des mots incohérens qui la présentaient à leur esprit, ils ont formé un nom propre composé de ces différentes parties d'idées; et pour celles qu'ils n'ont pu y faire entrer, l'ellipse est venue à leur sécours. * * Ce qui nous paraît le plus probable, est que les langues, comme le monde, ont commencé par le chaos, et ont acquis de la régularité plus tôt ou plus tard, sous une forme ou une autre, selon le génie des peuples, leurs situations ou leurs besoins. Celles des Indiens de l'Amérique du nord ont retenu beaucoup de ce genre chaotique qui a dû présider à leur formation. Les parties du discours y sont entremêlées d'une manière qui fait croire qu'elles n'ont pas toujours été soumises aux règles qui les gouvernent actuellement et qui, introduites peu à peu, n'ont pu que modifier, sans le détruire, le système de formation des mots qui paraît avoir prevalu dès le commencement.

*

"Ce système polysynthétique est ce qui caractérise les langues algonquines, ainsi que toutes celles de l'Amérique, et influe nécessairement sur leurs formes grammaticales, qui ne diffèrent que dans les détails."

To this he adds the following foot-note:

"La plus forte preuve qu'on puisse donner du mélange d'idées qui a existé au temps de la formation de ces langues, est le nombre de mots qu'elles ont pour exprimer la même chose, selon les circonstances qui l'accompagnent. Il y a un verbe pour dire 'j'ai envie de manger de la viande,' et un autre pour 'j'ai envie de manger de la soupe ou de la bouillie;' un mot, pour une plaie faite avec un instrument tranchant; un autre, pour une plaie faîte avec un instrument contondant; ces langues généralisent rarement." †

In support of these striking statements Duponceau has produced no trustworthy proofs. He has adduced only the most fanciful

*This is in substance the doctrine of holophrasis, to which attention will be given hereafter.

Mémoire, pp. 118-120.

reasons to support his conviction that the Indian languages still preserve the "chaotic style" which "seems to have prevailed from the beginning." The intermixture of the parts of speech does not follow from the fact that a language can in a word-sentence say, "I desire meat," or "I desire soup," and can distinguish between a "cut" and a "bruise." Such word-sentences are governed by certain fixed laws of position and sequence of stems.

The usual method of obtaining a vocabulary from an unlettered people is largely responsible for the doctrine that Indians rarely generalize. A savage is asked, How do you say "I eat meat," or "I drink soup?" and, if he understands the question, he replies by the appropriate sentences (not words, as many think), meaning, in his own vernacular, "I eat meat," or "I drink soup." He can distinguish between a cut and a bruise, and shows it by his language, but must it be inferred from this that he cannot generalize, or that he does it but rarely?

The materials of the language of the Iroquois consist of notional words, namely, nouns, verbs, and adjectives; representative words, namely, prefixive and independent pronouns; relational words, adverbs, conjunctions, and suffixive prepositions; and derivative elements, namely, formatives and flexions.

The distinctive nature and characteristic functions of these elements cannot be changed at will by any speaker, for the good and sufficient reason that a language does and can do only what it is in the habit of doing. In the category of notional words, the class of elements called noun-stems may not indifferently assume the functions and the flexions peculiar to either the verb-stems or the adjective-stems, neither can the verb-stems nor the adjective-stems indifferently assume the functions and the flexions peculiar to either of the other two classes of elements in that category; hence Duponceau's sweeping statement concerning the general character of the American Indian languages, that "they can change the nature of all parts of speech; of the verb, make an adverb or a noun; of the adjective or substantive, a verb," is not true of the Iroquoian tongue. The elements of its lexicon have acquired their individual values by virtue of a series of historical changes, and they severally retain these values solely at the behest of conventional usage, being subject at all times to further mutations of form and signification as this usage may decree.

The stems of words and word-sentences are not divided for any