ABSTRACT
Linguists, like all scientists, differ in the relative importance which they accord on the one hand to the description and classification of observable data, and, on the other hand, to attempts to develop a formal model, on the basis of which some of the observed data become predictable and hence understandable. Although most accounts of intonation can be classified in the former category without further argument that the latter approach is at least as legitimate. A reasonable constraint on the input to the phonetic component might be summed up by saying that representations at this level must be "pronounceable". The output of the phonetic model can be taken to be a set of physiological and/or acoustic parameters which vary as a function of time. The measurable Fo curve of an utterance can be thought of as an interaction between the intrinsic characteristics of the P-segments and the continuously evolving tension of the vocal-folds.
