Dynamic specification of speech and sign  

Lindblom, B., Mauk, C.* & Moon, S.-J.~

*University of Pittsburgh, ~Ajou University, Suwon, Korea

In Dynamics of Speech Production and Perception (in press). Edited by P. Divenyi. IOS Press.

Abstract

Experimental findings show that perception is more sensitive to changing stimulus patterns than to purely static ones. In speech articulatory movement and spectral changes are pervasive whereas steady-state intervals are rare. In view of those observations, it is curious that the traditional frameworks for describing speech tends to emphasize static attributes. For instance, the IPA framework specifies consonants and vowels using many terms that refer to speech production steady-states e.g., labial, high, voiced, stop, nasal etc. If perception thrives on change, why is phonetic specification built mainly around steady-state attributes?

The present chapter discusses the above issue in the light of new experimental data on hand movements in American Sign Language and tongue shape variations in VCV sequences. A target-based interpretation is proposed for both the manual and the phonetic data. The proposed analysis is compatible with current evidence on the physiological modeling of non-speech movements.

The proposed account resolves the static-dynamic paradox by suggesting that target control of speech movements arises from a 'least action' strategy which derives unique and smooth inter-target trajectories from optimization. The response characteristics of the production system turn the motor commands, that is a series of dicrete and static positional targets, into a continuous time-varying modulation of vocal tract aerodynamics and geometry thus creating a speech signal rich in perceptually important dynamic information. While this proposal acknowledges the 'phonetic gesture' as a significant aspect of perceptual processing, it strongly questions assuming that the input motor commands are "gestural". Thus, according to the present interpretation, the paradox is resolved.


PERILUS | All issues | 2005 issue | Previous | Next | Phonetics at Stockholm University