Speaker-Listener Dialect Error in Discrimination Testing

Abstract

Articulation curves were obtained from a group of listeners who listened to speech discrimination-test words in their own dialect and a dialect different from their own. The speakers were from the General American dialect and the Southern American dialect regions. A male and a female speaker represented each dialect. The listeners were all from the General American dialect. The curves were derived from the mean values of 20 listeners for each of the nine sensation levels. Each listener responded to the test stimuli only at one sensation level, and to both dialects. The responses to the female speech sample showed a consistent difference in favor of the listener's own dialect. The responses to the male speech sample did not yield consistent differences in favor of either dialect. The difference between the responses to the male and the female speech sample is discussed.