BOAS] HANDBOOK OF AMERICAN INDIAN LANGUAGES 35 



that it would be entirely arbitrary to designate the one group as 

 words modified by the other group, or vice versa. 



The importance of this consideration for our purposes lies in the 

 fact that it illustrates the lack of definiteness of the terms stem 

 and affix. According to the ordinary terminology, affixes are 

 elements attached to stems or words, and modifying them. This 

 definition is perfectly acceptable as long as the number of modifying 

 ideas is limited. When, however, the number of modifpng elements 

 becomes exceedingly large, we may well doubt which of the two is 

 the modifier and which the modified, and the determination finally 

 becomes entirely arbitrary. In the following discussions the attempt 

 has been made to confine the terms prefix, suffix, and affix entirely to 

 those cases where the number of ideas expressed by these elements 

 is strictly limited. Wherever the number of combined elements 

 becomes so large that they can not be properly classified, these 

 terms have not been used, but the elements have been treated as 

 co-ordinate. 



Discussion of Grammatical Categories 



From what has been said it appears that, in an objective discus- 

 sion of languages, three points have to be considered: first, the con- 

 stituent phonetic elements of the language; second, the groups of 

 ideas expressed by phonetic groups; third, the methods of combining 

 and modifying phonetic groups. 



It seems desirable to discuss the second of these points somewhat 

 more fully before taking up the description of the characteristics of 

 American languages. 



Grammarians who have studied the languages of Europe and 

 western Asia have developed a system of categories which we are 

 inclined to look for in every language. It seems desirable to show 

 here in how far the system with which we are familiar is character- 

 istic only of certain groups of languages, and in how far other systems 

 may be substituted for it. It seems easiest to illustrate this matter 

 by discussing first some of the characteristics of the Indo-European 

 noun, pronoun, and verb, and then by taking up the wider aspects of 

 this subject. 



