A Fuzzy Model for Analysing Perceptual Vowel Space: The Case of Modern Persian

The article proposes a fuzzy approach to Modern Persian perceptual vowel space. The theory of fuzzy sets is based on multi-valued logic, in which a category has no sharp boundary, so a membership function plays a key role. To achieve our goal, we carried out an acoustic experiment. A number of 96*2 synthetic stimuli of the six vowels in Modern Persian (/i,e,a,ɑ,o,u/) were examined through an Identification test. Each of the stimuli was randomly played twice for twenty participants (10 males and 10 females) and the values of the vowels’ first two formants (F1 and F2) were recorded. Based on the collected data, not only were the “crisp values” and “fuzzy values” of vowels tabulated, but the areas of “auditory clarity” and “auditory vagueness” around each vowel were also identified. The results showed that the vowel continuum between two adjacent vowels were vaguely perceived by the participants. It  was concluded that fuzzy vowels have no clearly defined boundaries. In other words, they may contain partial degrees of membership denoted by a real number that ranges from zero (meaning no membership) to one (meaning complete membership). Moreover, the  fuzzy sets theory proved to have the ability to calculate and explicate the overlapping vowels in the vowel space.