Review of the Literature

In the study of speech there is a well-known dichotomy between features that are planned by the speaker (i.e., present in the input to speech production) and mechanical features that emerge due to vocal tract constraints. Those aspects of speech which result from elasto-inertial, biomechanical, or aerodynamic constraints are not targeted by the speaker and can be explained by general phonetic principles. Examples include (a) the few milliseconds’ delay in the onset of voicing in vowels and sonorants following unaspirated stops due to the time needed to release the oral pressure-build up for the stop and initiate the transglottal flow needed for voicing ( J. Ohala 1983b) and (b) coarticulation with adjacent sounds which can be described in terms of the production constraints of the sounds involved (e.g. the ‘‘Degree of Articulatory Constraint’’ model of Recasens et al. 1997). The aspects of speech under the control of the speaker minimally include parameters used to signal segmental and prosodic contrasts, to mark linguistic boundaries and constituency, and language-specific coarticulatory and timing features. Planned effects also include language-specific articulatory maneuvers to inhibit (or facilitate) intrinsic influences on articulation (e.g. active expansion of the oropharyngeal cavity to preserve voicing in obstruents; J. Ohala 1997a). In recent years a great deal of evidence has suggested that fine-grained phonetic detail, such as language-specific preferences for certain ranges of voice onset time (VOT) values for prevoiced and