224 michelson: American Indian languages 



the entire world of which we have records going back continu- 

 ously for more than a thousand years. The percentage of the 

 total stocks so transmitted is altogether too small to afford 

 a firm and sure foundation for such a mode of action. If we 

 could establish from a minute study of the dialects of some dozens 

 of stocks that the kind of differentiation, etc., was on the whole 

 of a similar nature in these stocks, we would be entirely justified 

 in applying the principles derived from such a study to the 

 determination of the limits of stocks in American Indian lan- 

 guages and other stocks as well. Unfortunately such a study 

 has not been made, nor is there any prospect of it being done in 

 the immediate future. 



We have a similar difficulty in the reconstruction of parent- 

 languages of American Indian linguistic stocks. In the case of 

 Indo-European languages we again can take advantage of 

 principles derived from a study of the historical development 

 of the separate members of the stock, and apply the results to 

 the prehistoric period. We can not do this in the case of Ameri- 

 can Indian languages. The nearest approach to this would be 

 a very minute study of the dialects of known stocks. In some 

 cases there is no doubt that this would even largely counter- 

 balance the difficulty spoken of. For example, most of the 

 dialects of the Algonquian stock are so closely related that it 

 can readily be ascertained in at least many cases what is archaic 

 and what is secondary. Thus it is certain that the Fox e and i 

 vowels are more primitive than the Ojibwa i vowel, and that the 

 terminal vowels preserved in Fox, Sauk, Kickapoo, Shawnee, 

 and Peoria, but not appearing in Ojibwa, Ottawa, Potawatomi, 

 etc., are archiac. 2 Hence these features are to be ascribed to the 

 Algonquian parent language. However, we can not know that 

 precise quality of the prehistoric e and i vowels. Similarly the 

 combination of a sibilant followed by a surd stop in Cree is 

 more archaic than the correspondents in many of the related 

 languages, and so is to be likewise ascribed to the parent lan- 

 guage. (The actual proof that the Cree combination is more 



2 Amer. Anthrop. N. S., 15: 470. 1913; this Journal, 4: 403. 1914; Ann. Rep. 

 Bur. Amer. Ethnol., 28: 247. 1912. 



