A composite model of speech production

A model of the final stages of speech production, implemented on a VAX 11/780 computer, is described. Flexibility and a capacity for development in line with published data on real speech are achieved by means of a large number of parameters. The model is used to investigate the extent to which observed patterning of selected speech sounds is predictable from production processes and constraints. For sequences containing 'open' or 'close' vowels combined with voiced or voiceless fricatives, the simulation yields the kind of covariation of consonant and vowel segment features that is observed in real speech.