Exploring the movement dynamics of manual and oral articulation: Evidence from coarticulation

Abstract

This project explores three classes of human action through an investigation of long-distance coarticulation, defined here as the articulatory influence of one phonetic element (e.g., consonant or vowel) on another across more than one intervening element. Our first experiment investigated anticipatory vowel-to-vowel (VV) coarticulation in English. The second experiment was patterned after the first but deals instead with anticipatory location-to-location (LL) effects in American Sign Language (ASL). The sign experiment also incorporated a non-linguistic manual action, permitting a comparison of effects not only between spoken and signed language, but also between linguistic and non-linguistic manual actions.

For the spoken-language study, sentences were created in which multiple consecutive schwas (target vowels) were followed by various context vowels. Eighteen English speakers were recorded as they repeated each sentence six times, and statistical tests were performed to determine the extent to which target vowel formant frequencies were influenced differently by different context vowels. For some speakers, significant effects of one vowel on another were found across as many as five intervening segments. In the sign study, motion-capture technology was used to investigate LL coarticulation in the signing of five ASL users. Some evidence was found of significant LL coarticulatory influence of one sign on another across as many as three intervening signs. However, LL effects overall were weaker and less pervasive than the VV effects found in the spoken-language study. The outcomes in the sign study's linguistic and non-linguistic conditions did not substantially differ. Collectively, these results support and complement previous research which has found that in comparisons of linguistic and non-linguistic manual and oral actions, movement patterns associated with oral and manual actions differ appreciably, while the linguistic vs. non-linguistic distinction appears to show little or no influence.