The Imitation of Synthetic Vowels and Some Implications for Speech Memory

Publisher Website
Google Scholar
Abstract
The ability of American English speakers to imitate synthetic vowels was assessed by spectral comparisons of the computer-generated vowels and the corresponding human imitations. The stimulus vowels formed two trajectories in the (vowel formant) F1-F2-F3- space: one trajectory from /u/ to /i/ and the other from /æ/ to /i/. It is concluded from the imitation responses that the vowels in the /u/ to /i/ series are represented in memory as a continuous transformation of the acoustic signal, whereas the vowels in the /æ/ to /i/ series are represented as a classification transformation.

This publication has 0 references indexed in Scilit: