32 BUEEAU OF AMERICAN ETHNOLOGY [bull. 40 



ing phonetic modifications occur. Thus the elements ta-t-ru^n combine 

 into ta'hu^n i make (because tr in a word changes to Ji) ; and ta-t-riks- 

 ru^n becomes tahtkstu^n i make an arrow (because r after s 

 changes to t) . At the same time riks arrow occurs as an independ- 

 ent word. 



If we follow the principle laid down in the preceding remarks, 

 it will readily be seen that the same element may appear at one time 

 as an independent noun, then again as a part of a word, the rest of 

 which has all the characteristics before described, and which for 

 this reason we are not inclined to consider as a complex of independ- 

 ent elements. 



Ambiguity in regard to the independence of parts of the sentence 

 may also arise either when in their significance they become depend- 

 ent upon other parts of the sentence, or when their meaning is so 

 vague and weak as compared to the other parts of the sentence that 

 we are led to regard them as subordinate parts. Words of this 

 kind, when phonetically strong, will generally be considered as inde- 

 pendent particles; .when, on the other hand, they are phonetically 

 weak, they will generally be considered as modifying parts of other 

 words. A good example of this kind is contained in the Ponca 

 texts by the Rev. James Owen Dorsey,^ in which the same elements 

 are often treated as independent particles, while in other cases they 

 appear as subordinate parts of words. Thus we find ^eama these 

 (p. 23, line 17), but jdbe amd the beaver (p. 553, Hne 7). 



The same is true m regard to the treatment of the grammar of the 

 Sioux by the Rev. S. R. Riggs. We find in this case, for instance, 

 the element 'pi always treated as the endmg of a word, probably 

 owing to the fact that it represents the plural, which in the Indo- 

 European languages is almost always expressed by a modification 

 of the word to which it applies. On the other hand, elements like Jcta 

 and sni, signifying the future and negation respectively, are treated 

 as independent words, although they appear in exactly the same 

 form as the pi mentioned before. 



Other examples of this kind are the modifying elements in Tsim- 

 shian, a language in which innumerable adverbial elements are 

 expressed by fairly weak phonetic groups which have a definite 

 position. Here, also, it seems entirely arbitrary whether these 

 phonetic groups are considered as separate words, or whether they 



1 Contributions to North American Ethnology, vi. 



