The variation of absolute and relative measures of speech activity

Abstract

The task dynamic approach to speech motor control proposes that the variability of articulatory gestures can be reduced by calculating proportions relative to the duration of a gestural cycle. This hypothesis was examined in the light of a sizeable data base of temporal, spatial and velocity measures of vertical tongue movements in the production of the syllable /ka/ ( N subj = 13, N obs = 2178). The absolute-measure variation of gestures, gesture segments and inter-articulatory delays was compared to the relative-measure variation for variables that were calculated as proportions either of the movement or of the cycle in two conditions, across speech production tasks vs. within tasks. It was expected that the calculation of gesture durations and inter-articulatory delays as a proportion of the cycle would yield the greatest reduction in variation, and that this reduction would be cumulative with reductions of variation from the across-task to the within-task condition. However, all variables showed greater variation reductions when calculated as proportions of the movement, and only the inter-articulatory delay between onset of lingual and laryngeal motor activity showed clearly cumulative variation reductions. Further, velocity-related variables showed particularly low variation, as did the longer absolute temporal variables, given a specification of speech production task. These differences in variability are in better accord with a notion of active control over temporal variables of speech than with a notion of temporal, spatial and velocity measures of vertical tongue movements constraints, as suggested by the task dynamic approach.