Perhaps the writer may be allowed to add in closing that he has noticed the various specimens figured not because of any remarkable peculiarities they possess, but because they show something as to the nature and style of the objects used by the Iroquois and Algonkin occupants of the Champlain Valley. Specimens very nearly like those figured here have been found, and often in far greater number, in the west and south, as anyone may see by examining the collections of some of our museums or by consulting the numerous figures given in Mr Moorehead's most valuable Stone Age. From this it seems probable that most, at least, of the problematical forms had their origin outside of New England, and that either the objects themselves were imported from farther west, or those made elsewhere were imitated by the Indians of the Champlain Valley. Probably some of the forms originated here but the close resemblance of most to those found far more abundantly elsewhere suggests importation to a considerable extent. It may be well to add that some of the tubular pipes and the last mentioned bird-stone are now in the American Museum of Natural History, New York City, UNIVERSITY OF VERMONT, BURLINGTON, VT. THE PROBLEM OF NOUN INCORPORATION IN THE AMERICAN LANGUAGES By EDWARD SAPIR HE term "incorporation" has been much used in discussion devoted to the structure of American languages. Despite the steadily growing mass of American linguistic material, a good share of the data presented in the last few decades being distinctly superior from the point of view of critical analysis to much that served as illustrative material in earlier days, it can not be asserted that the term is always clearly understood or satisfactorily defined. This paper is not at all concerned with whether the linguistic stocks of America are or are not as a whole characterized by a process that may be called "noun incorporation," but aims merely to give a usable definition of the term and to show that several of these stocks actually make use of the process. This may not seem a very revolutionary attempt, nor is it intended to be. As, however, Dr Kroeber has undertaken in a recently published paper1 to demonstrate the mythical or, at any rate, theoretically unlikely character of noun incorporation, it seems in order to accept his implied challenge and to present some new data by way of rebuttal. On two or three negative points all must be in hearty agreement with Dr Kroeber. In the first place so-called pronominal incorporation and noun incorporation stand in no necessary relation to each other. A very large number of American, as of non-American, languages make use in the verb of affixed elements of pronominal signification; they are, as regards their syntactical use, very commonly subjective, less frequently, though by no means rarely, also objective, and still less commonly they indicate also dative, ablative, or other case relations (thus, in Wasco, "him" and "me" in "I give it to him" and "he takes it from me" are as thoroughly "incorporated" into the verb-complex as are the 1A. L. Kroeber, "Noun Incorporation in American Languages," XVI. Internationaler Amerikanisten-Kongress, 1909, pp. 569-76. subjective "I" and "he" and objective "it"). As Dr Kroeber points out, it is incorrect to consider these pronominal elements as truly "incorporated" forms of independent personal pronouns; being either simpler in form than the latter or, often enough, etymologically unrelated to them, they are best considered as formal or inflectional in character. Whether or not they may, in particular cases, be thought to have been originally independent elements that have, through an intermediate proclitic or enclitic stage, coalesced with the verb stem into a morphologic unit, matters not at all; historical considerations should not interfere with a descriptive analysis, otherwise morphologic change in language ceases to have a meaning. In the case of the Wasco1 sentences referred to before, the "incorporated" elements -n- "I, me," -- "it," and -i"him," are evidently not actually incorporated forms or secondary developments of the corresponding independent personal pronouns náika, táxka, and yárka, while -tc- "he" (as subject of transitive verb) is quite unrelated to the independent pronoun. Few more striking cases can be found than that of Takelma. Here we have no less than eight distinct affixes to indicate the first person singular ("my, I, me") in the noun and verb (wi-, -t'ek', -t'k', -t'e, -t'e", -în, -n, -xi), yet not one of these is etymologically related to the independent pronoun gi. Clearly, then, the incorporation of a noun or noun stem into the verb is not in most cases analogous to pronominal "incorporation." It may even be argued on general grounds that nominal and pronominal incorporation tend to be mutually exclusive processes. The main purpose of a pronominal affix is to refer to or replace a substantive, in the former case often determining also its syntactic relation; hence a pronominally incorporating language should find noun incorporation unnecessary, and vice versa. The fact that this theoretical conclusion is by no means entirely borne out by the facts shows how little reliance is to be placed in a priori considerations. We shall find, however, that noun incorporation can indeed exist without true pronominal incorporation or rather inflection. In the second place it is clear that verbal affixes that refer to nouns, in other words, convey a substantival idea, are not instances 1 Of Chinookan stock. |