ABSTRACT
The problem of definition is of some significance when we consider the development of the writing systems that claim to be reflecting salient features of spoken language for it becomes necessary to ask what the features might be. In the visual modality drawings serve as good representations of objects. We might guess, therefore, that, for speech, an equally effective representation would have to be presented in the auditory modality. And indeed, such 'shorthand' speech has been produced – taking the form of highly schematic spectograms which, when converted back to sound, produce recognizable speech. If we cannot define speech easily in acoustic terms, is it possible to isolate the psychologically relevant variables? In other words, what information does the hearer actually use when speech is decoded? Which of the innumerable acoustic variations we have identified are actually significant, and which are not.
