Abstract
The traditional classification of Kazakh vowels, the most complicated objects in Kazakh phonetics, is somewhat inadequate in that it does not uniquely identify each vowel. Moreover, the formant structure of the vowels has not been investigated yet, which would potentially be useful in many applications regarding speech processing. In this work a formant analysis of Kazakh vowels is performed. This analysis suggests a new natural classification system which precisely describes each vowel. The proposed classification is supported by the membership function introduced and by the geometric interpretation derived from the constructed vowel space. Further, from this mathematical model some linguistic statements are inferred, which are related to the real phonological processes.