Doctoral Dissertation Research: Dissimilation in Phonological Response-Priming

University Of California-Berkeley, Berkeley CA

Investigators

Abstract

A given speech sound is not always pronounced the same way. This dissertation research investigates what factors constrain variation in pronunciation. Previous speech research has identified many sources of variation in speech sounds: speaker age, gender, dialect, socio-economic background, physical differences between speakers, stylistic differences, psychological and emotional condition, etc. All these sources of variation could easily make speech differ wildly from person to person, but normally the sounds people actually say are relatively similar from utterance to utterance. One common source of variation in the pronunciation of speech sounds is the phonetic nature of nearby sounds. Preliminary experimental results for this dissertation have shown that in some circumstances, there are tendencies for a produced sound to be less like a previously planned sound. This suggests that there is a cognitive mechanism restraining variation. With NSF support, Mr. Tilsen will conduct additional experiments in order to further investigate this phenomenon. The experiments will address important details, including whether the cognitive mechanism responsible for restraining variation is specific to certain types of speech sounds, and whether the mechanism is specific to certain languages or universal. These experiments will contribute to our understanding of speech in a number of ways. Clinicians will be better able to recognize and diagnose speech pathologies; speech recognition and speaker identification technologies will work more reliably. Suggestive similarities between this speech phenomenon and related phenomena in other domains of behavior - such as hand and eye movement - will encourage researchers to look for deep principles governing all types of movement, including speech.