In order to complete a computational model of speech production incorporating the relations between neuromotor, biomechanical and aerodynamic levels of description, a means for assigning linguistically relevant input strings is needed. This paper describes a two-phase curve-fitting and template matching method for mapping between phonemes and articulator movement behavior based on via point estimation. Among other things via point estimation provides a good means of data compression and the estimation software has proved to be an excellent empirical tool.