Abstract
Acoustically a naturally-spoken vowel is composed of five formants. However, the acoustic quality of a vowel is known to be mostly determined by $F_1\;and\;F_2$. The main purpose of this study was to examine how synthesized vowels with $F_1\;and\;F_2$ are perceived by Korean native speakers. In addion, we are interested in finding whether the synthesized vowels are perceived differently by standard Korean speakers and Kyungnam regional dialect speakers. In the experiment 9 Seoul standard Korean speakers and 9 Kyungnam dialect speakers heard 536 vowels synthesized in vowel space with $F_1\;by\;F_2$ and categorized them into one of 10 Korean vowels. The resultant vowel map showed that each Korean vowel occupies an unique area in the two-dimensional vowel space of $F_1\;by\;F_2$, and confirmed that $F_1\;and\;F_2$ play important roles in the perception of vowels. The results also showed that the Seoul speakers and the Kyungnam speakers perceive the synthesized vowels differently. For example, /e/ versus /$\varepsilon$/ contrast, /y/, and /$\phi$/ are perceived differently by the Seoul speakers, whereas they were perceptually confused by the Kyungnam speakers. These results might be due to the different vowel systems of the standard Korean and the Kyungnam regional dialect. While the latter uses a six-vowel system which has no /e/ vs /$/ contrast, /v/ vs /i/ contrast, /y/, and /$\phi$/, the former recognizes these as different vowels. This result suggests that the vowel system of differing dialect restricts the perception of the Korean vowels. Unexpectedly /i/ does not occupy any area in the vowel apace. This result suggests that /i/ cannot be synthesized without $F_3$.
