Page:IJAL vol 1.djvu/12

4 wide continuous distribution. This is true of morphological processes as well as of particular psychological aspects of American languages. Thus the incorporation of the, nominal object, which in former times was considered one of the most characteristic features of American languages, is confined to certain areas, while it is foreign to others. The tendency to qualify generalized verbal terms by means of elements which express instrumentality is characteristic of some areas. The occurrence of various specific elements that define locality of an action, as affecting objects like “hand,” “house,” “water,” “fire,” or other special nominal concepts, is characteristic of other regions. Classification of actions or of nouns according to the form of the actor or of the object also belong to several groups of languages. Nominal cases are present in some languages, absent in others. In a similar way we find present in some regions, absent in others, processes like that of reduplication or of vocalic or consonantic modification of stems.

Attempts to classify languages from these distinct points of view do not lead to very satisfactory results. Not only would the purely morphological classifications be contradictory, but in many cases where a close morphological agreement exists, it remains highly unsatisfactory to co-ordinate vocabularies and the phonetic equivalents of similar morphological ideas. On the basis of Indo-European experience, we should be very much inclined to seek for a common origin for all those languages that have a far-reaching morphological similarity; but it must be acknowledged, that, when the results of classifications based on different linguistic phenomena conflict, we must recognize the possibility of the occurrence of morphological assimilation. The problem is analogous to that of the relation between Finnish and Indo-European languages, which Sweet assumed as established, while the observed relations may also be due to other causes.

Owing to the fundamental importance of these questions for the solution of the problem of the historical relationship between American languages, it seems particularly important to attempt to carry through these classifications without prejudging the question as to the genealogical position of the various groups. It is quite inconceivable that similarities such as exist between Quilleyute, Kwakiutl, and Salish, should be due to a mere accident, or that the morphological similarities of Californian languages, which Kroeber and Dixon have pointed out, should not be due to a definite cause. The experience of Aryan studies might induce us to agree that these must be members of single linguistic stocks; but this assumption leaves fundamental differences unaccounted for, and neglects the possibility of morphological assimilation, so that at the present time the conclusion does not seem convincing. We ought to inquire, first of all, into the possibility of mutual influences, which will be revealed, in part at least, by lack of correspondence between lexicographic, phonetic, and detailed morphological classifications.

We do not mean to say that the investigation may not satisfactorily prove certain genealogical relationships; but what should be emphasized is, that, in the present state of our knowledge of primitive languages, it is not safe to disregard the possibility of a complex origin of linguistic groups, which would limit the applicability of the term “linguistic family” in the sense in which we are accustomed to use it. It is certainly desirable, and necessary, to investigate minutely and carefully all suggestive analogies. The proof of genetic relationship, however, can be considered as given, only when the number of unexplained distinct elements is not over-large, and when the contradictory classifications, to which reference has been made before, have been satisfactorily accounted for.