105 



These more or less variable tendencies are all that the 

 phonetician has to depend on when he wants to determine the 

 nature of the vowels. To describe a vowel is merely to describe 

 that position of articulation with which it is usually produced 

 in the language under investigation. The alphabetical symbol 

 of the sound is merely an abstraction or the type of a whole 

 group of shades, in determining these it is generally convenient 

 to take the standard from some language which we know from 

 hearing (as French). 



Direct observation of the positions of vowel-articulation by 

 looking at the mouth of another speaker is both difficult and 

 uncertain. [ have therefore principally used the indirect method, 

 which is to imitate the sounds myself and to try to keep con- 

 trol of the distances in my mouth and the shape and positions 

 of my tongue by means of the usual methods of self-observation 

 (by observation in a mirror, by whispering and breathing in, 

 by the tongue's feeling of its own movements). The following 

 analphabetical indications and the classification of the sounds 

 include merely the most essential sound-shades, for I have 

 found that it would be impossible to find expressions for all 

 the shades. So the results which I am going to present are 

 rather to be taken as boundary-marks or mile-stones than as 

 mathematically correct expressions for all the vowel sounds of 

 the language. 



My experience with vowel-systems has been that every time 

 1 have tried to arrange the Greenlandic vowels under the usual 

 heads, they have burst the bounds of the system. Jespersen's 

 method (analphabetical symbols which aim to give an exact 

 quantitative expression of the articulation) is no doubt the only 

 one that can give satisfaction whenever the object is to dis- 

 tinguish fine shades of speech-sounds, but even that is not 

 entirely infallible in the case of the vowels, for its method of 

 valuation is even for experts partly dependent upon individual 

 judgment. Since in the case of the uvular sounds the system 



