Volume 37, Issue 1-2

The long-term task of phonetics is to contribute towards the construction of a theory of the use and form of spoken language. The choice of this objective is primarily scientifically motivated but is reinforced by the prospect of using phonetic knowledge for educational, clinical and technological purposes. Phonetics is relevant to the general study of spoken language in two ways: It enters into describing how language is used by speakers, listeners and learners (the psychology and physics of speech behavior) and accounting for how language is formed by the social and biological constraints on their vocal/auditory performance (the phylogeny and ontogeny of language form). Such a program is today served by phonetics oriented towards both natural science and behavioral experimentation and by general linguistics dominated by its quest for autonomy and a place in the tradition of speculative philosophy. While the immensity of the subject matter of spoken language certainly makes a multiplicity of empirical approaches mandatory, it is equally true that a broader paradigm lending theoretical unification to the mosaic of research cultures must be sought. This appears to be the case particularly with regard to the full exploitation of the theoretical and practical relevance of phonetics. The main point of the present paper is that such unification can be achieved by rejecting all attempts at making linguistics autonomous and hostile to ‘extralinguistic’ explanation and by embedding the study of language and speech, and thus phonetics, within the framework of biological functionalism.

The blood flow to a given brain region increases as the level of neural activity is augmented. Hence mapping of variations in regional cerebral blood flow affords a means of imaging the activity of various brain regions during various types of brain work. The paper summarizes the patterns of cortical activity seen during various language functions, emphasizing the practically symmetrical involvement in both hemispheres. A case of auditive agnosia (with complete cortical word deafness but preserved pure tone thresholds) is presented. The patient’s normal speech constitutes evidence that auditory feedback (absent in his case) is not a prerequisite for speaking.

Methodologies of speech research with respect to the production processes are discussed, with an emphasis on the recent development of new instrumental techniques. It is argued that systematic studies of large amounts of speech data are necessary to understand the basic characteristics of speech. The traditional notion of phoneme-size segments seems inappropriate for interpreting multidimensional articulatory movements by a concatenative model. Experimental means such as a computer-controlled X-ray microbeam technique and advanced statistical processing, in combination with a new theoretical framework of phonetic description, promise future development.

To derive speech wave data from area function specifications and the reverse, to predict the area function from the speech wave, are fundamental problems of acoustic theory of speech production. Deviations from ideal resonator theory in terms of vocal tract boundary conditions and source filter interactions are discussed. Perturbation theory is related to special problems of male-female vocal tract scaling. Shortcomings of the inverse transforms are discussed. Merits of lossy transmission line theory over standard linear prediction procedures are emphasized. The use of bandwidths for removing ambiguities is illustrated in simple models. A limited amount of bandwidth data supplementing formant frequency data and model related vocal tract constraints appears to be optimal.

This paper gives a tutorial review of the linear prediction method in its application to acoustical analysis of sampled speech. The paper explains how the formant frequencies and the fundamental frequency can be estimated from sampled speech waves by the use of linear prediction. Advantages and disadvantages of the method are discussed together with the problems in estimating the above parameters. Application examples are also given.