PHONOLOGY. 223 



lungiuiges, but only as far as the transition of words from one dialect to 

 another is concerned. But in illiterate languages the same interchange, often 

 a more extensive one, talces place within one and the same dialect. 



So much did this fact contradict the time-honored, ancient ideas of 

 grammar lodged in the heads of missionaries and school-teachers, and so 

 little did it conform to Latin, Greek, and Hebrew models, that the puzzled 

 grannnatical writers on American, African, or (Iceanic languages bluntly 

 denied the existence of certain sounds wliirh tliey knew to be in the lan- 

 guage, but found to alternate with others for no apparent causes. This 

 relieved them from the necessity of accounting for this puzzling phonetic 

 fact. The existence of the sonant unites \\;\a flatly denied to many Amer- 

 ican Indian tongues, and the Mohawk-Irocpiois* alphabet was proclaimed 

 to possess sixteen sounds (or "letters", as they were called) only, while in 

 reality it has over twenty-four, all of which are easily expressed by the 

 Roman alphabet. 



I have observed alternation of sounds in all tlie North American 

 languages which I have studied personally with the aid of natives, and 

 have also hinted at one of its hidden causes, viz. the laryngeal or pectoral 

 pronunciation of the red man. Even those Indians whose languages have 

 been reduced to waiting for fifty or one hundred years back, and in whose 

 books all traces of this interchangeability were suppressed by the mission- 

 aries, etc., as the Creeks, Cha'hta, and Iroquois, permute their consonants and 

 vowels with the same libei-ty as if these books had never appeared in print. 

 It would be exactly so with us if our ancestors had not had a literary 

 training foi' the last thousand years at least. 



I have recorded the alternations observed by me in the Kdyowe (or 

 Kiowa) language in a monograph published in the American Antiquarian, 

 IV, pp. 280-285, under the title: "Phonetics of the Kayowe Language", 

 the results obtained there being almost identical to those to be given below 



This permutability of cognate sounds forms one of the prominent jilio- 

 netic features of Klamath, and occuis in initial as well as in medial or final 

 sounds. Still there are words in which certain sounds do not interchano-e 

 with others. This is especialh^ observed in homonyms, where jiermutation 



* This dialect of Iroquois lacks 1>, p, and f. 



