NO. 2
��TAKELMAN, KALAPUYAN, AND CHINOOKAN LEXICOGRAPHY
��177
��compiled by Dixon and Kroeber, and while my superficial examination of this lexical material impressed me with the probable soundness of Dixon's and Kroeber's conclusions, I still must refuse to accept them as final, as long as these vocabularies are continued to be withheld from publication and until more morphological evi- dence is brought into play. Nothing is more dangerous and unsatisfactory in an investi- gation of this sort than to arrive at so-called final conclusions that are seemingly based solely upon lexicographical material. In the same way it would be wrong to deny the exist- ence of a relationship between two languages, merely because the evidence of the lexical material is negative. It is well to bear in mind that in trying to establish genetic relationships between languages that seem to be, at first sight, non-related, lexical and morphological evidence must be treated separately, and that morphological evidence must be accorded greater weight. I believe it to be a fact, es- tablished by investigations in other linguistic fields, that lexicography is more easily subject to borrowing, to loss of words and stems, and to new additions; and that the formative elements and structure of a language are more stationary and less influenced by those of some neighboring tongue. 1 Of course, I am per- fectly aware of the fact that instances may be cited where the morphology of one language has undergone changes due to borrowing. But these instances are so few that they, in no way, affect the correctness of my statement.
The absence of conclusive evidence concern- ing Penutian and Hokan is the more unfortu- nate, as there exist strong reasons to believe that the Takelman, Kusan, Siuslawan, Yakon- an, Kalapuyan, and (perhaps) Chinookan lan- guages spoken in Oregon may be proven to be Penutian sister-tongues. For that reason, the
1 A paper dealing with this question and particularly showing how dialects may undergo considerable lexical changes and still retain their full original structure, is in the course of preparation.
��additional collection of material from the Pe- nutian field in California and the immediate publication of the data thus far collected would seem to constitute one of the most im- portant tasks that confront the investigators in the California area. Without such material the Penutian theory must, for the time being, be held in abeyance, and the establishing of a relationship between California-Penutian and Oregon-Penutian must be deferred to the future.
Undoubtedly the strongest and best evi- dence adduced by Dixon and Kroeber is that upon which they base their conclusion con- cerning the genetic relationship between Yu- rok and Wiyot. The amount of lexical cor- respondences, the existence of phonetic shifts, and the presence of structural similarities are too numerous and too regular to be accounted for as due to accident or to borrowing, al- though it would be highly desirable to produce more evidence in the near future. But suf- ficient proof has already been furnished to jus- tify a belief that additional material would rather increase than decrease the certainty of a genetic relationship between Yurok and Wiyot.
A very weak case of reduction is found in Sapir's previously mentioned attempt to clas- sify Yurok and Wiyot as Algonkin languages, which, on the face of the evidence presented, is far from conclusive. The difficulties encount- ered by him were twofold: First, inadequate Yurok and Wiyot data, both grammatical and lexical; and, secondly, unfamiliarity with the intricate and complicated structure of the Al- gonkin languages. To the first are probably due the unsatisfactory and irregular corre- spondences quoted by him as based upon phonetic shifts, while the second has been re- sponsible for the numerous comparisons of wrong morphological elements. Sapir's paper, more than any other effort, demonstrates the imperative necessity of basing all attempts at establishing relationships upon exhaustive and (phonetically) sound lexical material and upon
�� �
