BOAS] HANDBOOK OF AMERICAN INDIAN LANGUAGES 433 



I have recorded as equally admissible — 



g'o'x^g'in and g'o'x^gun my visible house here 

 g'o'x^g'aEn and g'd'x^gwaEU my invisible house here 



While the rule just described is founded entirely on the phonetic 

 influence of the stem element upon its suffix, we have also a class of 

 phonetic changes which are due to etymological causes, and can not 

 be brought entirely under phonetic rules. 



When a word ending in a consonant is followed by a suffix beginning 

 with another consonant, there is a strong tendency to elision of the 

 initial consonant of the suffix, although the combination may be 

 admissible according to the general phonetic laws. Thus the com- 

 pound of the stem qds- to walk, and the suffix -x'Hd to begin, would 

 result in the phonetically admissible combination qd'sx'^id, which we 

 find in a word like ^wdlasx'e' lynx. Nevertheless, the resulting form 

 is qa'sHd. The elision of the initial sound of the suffix is therefore not 

 entirely due to phonetic causes, and must be treated in detail in a 

 discussion of the suffixes. It is quite likely that the suffixes in ques- 

 tion may be compounds of two suffixes, the first of the combination 

 being dropped. The question will be discussed more fully in § 18 

 (p. 449). 



Another ante-active change which is not entirely due to phonetic 

 causes is the transformation of a into wd after n and vowels, which 

 occurs in a few suffixes : for instance — 



t!e'p-d to step off Id'-wd to be off (the right line) 



sop-d'la to chop off dd'wd to fail to hold 



Tc'at-d'la long thing on water han-wd'la hollow thing on 



water 

 TYiEX-dfla canoe drifts on water g'V-wdla to be on water 



("2) Retroactive Chanyes 



The changes just mentioned are best explained as an effect of the 

 stem upon the suffix. We find, however, also others, indicating 

 an action of the suffix upon the stem. These consist in a hardening 

 or weakening of the terminal consonant of the stem, and can not be 

 explained by phonetic causes, but must be founded on etymological 

 processes. 



The following examples illustrate these processes which were men- 

 tioned before in § 2. In the first column the stems are given, the 

 terminal sounds of which are modified by the addition of suffixes. 

 In the second column hardened forms are given, in the third weakened 

 44877— Bull. 40, pt 1—10 28 § 4 



