NO. I
��REVIEWS
��crossing, the classification of possessed nouns into separable versus inseparable, there is to be recognized an independent classification of possessed nouns into terms of relationship versus all others. Uhlenbeck does not take this view. He prefers to consider such languages as Takelma, Yuki, Porno, and Mutsun as survivals of an earlier condition, in which both terms of relationship and body- part nouns constituted a separable class of possessed nouns ; and that, as they grew more analytic in character, the body-part nouns gradually yielded to the analogy of the vast majority of nouns. Such a language as Haida, according to Uhlenbeck, represents a transition stage.
So long as we look at the facts in a purely schematic way, Uhlenbeck's historical theory seems plausible; but further consideration of the facts tends to cast doubt on the correct- ness of his view. Leaving Chimariko aside, it certainly seems suggestive that the funda- mental difference between the separable and inseparable pronominal affixes of such lan- guages as recognize the distinction merely lies in the presence of an affix of separable significance. The example of Algonkin, further, strongly suggests that this type of affix is a morphological element that has per se nothing to do with pronominal classifica- tion. On the other hand, the pronominal relationship-term affixes of Takelma, Yuki, Pomo, and Mutsun form a morphologically distinct class of elements. In other words, the two types of classification of possessed nouns (separable versus inseparable, and terms of relationship versus other ncuns) work, on the whole, along quite distinct lines; whence we must conclude that they are historically distinct phenomena, and merely intercross in certain languages (Haida, Siouan).
That our point of view is sound (i.e., that the concept of separability or inseparability is generally, directly or at last analysis, indicated by an affix, and that, on the other
��hand, the terms of relationship generally owe their distinctness as a class to the factor of pronominal classification), is further indicated by other linguistic data, in part not accessible to Uhlenbeck. In Southern Paiute there is no real classification of possessed nouns into separable and inseparable, nor any classifica- tion of possessive pronominal affixes; but there are two suffixes of not infrequent use that bear on the concepts of acquirement and inseparability, i'ni- ACQUIRED BY, OWNED BY (e.g., qani-i'ni- HOUSE OWNED BY ONE, qani- HOUSE, HOUSE ONE LIVES IN); and -'a- INSEPARABLY BELONGING TO, chiefly used with body-part nouns that in ordinary experience often occur disconnected from the body, like
BONE, SALIVA, ' SINEW, FAT, HORN (e.g., 00- BONE, 00- a- BONE IN ONE'S BODY).
In Nootka, again, there is, with certain interesting exceptions to be presently noted, but one series of possessive pronominal affixes; but before the possessive suffix proper nor- mally appears one of two suffixed elements, -uk-, -'ak-, indicating that the possessor and the object possessed are physically separable (hence including terms of relationship) ; or -'at-, indicating that they are not physically separable (hence applying, above all, to parts of the body). The latter element is morpho- logically identical with the passive suffix in verbs. The Nootka -'at- forms suggest that, in any reduction of the range of the insepa- rable class of possessed nouns, it would be the terms of relationship not, as Uhlenbeck assumes, the body-part nouns that would be levelled out by analogy. From another point of view, however, the Nootka terms of relationship stand in a class by themselves. Not only are most of them provided with a distinctive relationship-term affix -qso (cf. the corresponding -mp of Kwakiutl), but the second person singular possessive is either formed in the regular manner (-qso plus separably possessive -'ak plus pronominal -'itqak, contracted to -qsak'itqak) or, far more frequently, by using the bare stem without
�� �
