290
��INTERNATIONAL JOURNAL OF AMERICAN LINGUISTICS
��VOL. I
��xa-na TO DRINK, Atakapa ka; Atsugewi -ima-, Achomawi -(ii)iiiia-, Comecrudo imdy TO SEE : Pomo ma-, Yana ml-, Chimariko -mam-, Karok inah-, Coahuilteco mas, Comecrudo mdhe. Even the dialectsof a single group vary on this point, as could be abundantly illustrated from Pomo and Yuman.
Statistics based on the comparative vocabu- lary are of little significance at present, owing to the fact that most of the languages are but sparsely represented, some far more sparsely than others. Thus, the fact that Pomo, Chima- riko, Yuman, and Yana offer the greatest num- ber of cognates to the CoahuHtecan languages, while Chumash, Esselen, Shastan, Seri, Karok, and Chontal offer the least loses nearly all its significance when we remember that there was less matori.il available for comparison in the latter group than in the former. In proportion to the amount of material to chose from, indeed, Esselen, Karok, Seri, and Chontal seem to offer more similarity to the Coahuiltecan languages than Yana, which, in manuscript form, is by far the best known to the writer of all the lan- guages compared '. The relatively small num- ber of Yana-Coahuiltecan cognates found is probably the only significant point that could at present be made on statistical evidence. It is doubtless closely related to the fact, abun- dantly proven by other evidence, that of all Hokan languages Yana is the most specialised and therefore the least typical. Turning to the
i . This I consider a most encouraging fact. If the resemblances here discussed were entirely explainable as due to accident, the Yana-Coahuiltccan parallels should have been several time as numerous as for any other pair, whereas, as a matter of fact, there are only a trifle over half as many Yana-Coahuiltecan parallels as Pomo- Coahuiltecan ones.
��Coahuiltecan languages, we find that the order of degree of similitary to Hokan is Tonkawa, Comecrudo, Coahuilteco (including one example each from San Francisco Solano and Maratino), Karankawa, Atakapa, and Coto- name, the number of Tonkawa-Hokan cognates being somewhat greater than of Pomo-Coa- huiltecan. This, if significant at all, is as it should be, for Tonkawa is an interior language and, geographically speaking, relatively nearest the Hokan languages of California.
A glance at Powell's linguistic map, so far from creating dismay at the hazardous nature of our attempt, rather serves to render it intel- ligible.. True, there is an enormous distance separating Tonkawa and Yuman, or Coahuil- teco and Seri. But is it an accident that practi- cally the whole of the vast stretch of country separating the Coahuiltecan from the Yuman tribes is taken up by the Southern Athapascans (Lipan, various Apache tribes, and Navaho) ? That these last are intrusive in this area has always been felt probable by both ethnologist and linguist. The relationship of Athapascan to Haidaand Tlingit, which I have demonstrat- ed in another paper 2 , raises this feeling to a certainty. I venture to put forward the hypo- thesis that the Hokan-speaking and Coahuilte- can-speaking tribes formed at one time a geo- graphical continuum and that at least one of the factors in their disruption was the intrusion of Athapascan-speaking tribes from the north. An earlier intrusion of Uto-Aztekan (more par- ticulary Sonoran-Shoshonean) tribes from the south may eventually also have to be taken account of.
2. The Na-dene Languages, a Preliminary Report, Ame- rican Anthropologist, n. s., vol. 17, pp. 554-558, 1915.
�� �
