Three Areas of Experimental Phonetics. By Peter Ladefoged.
Oxford University Press. 1967, vi + 180 pp.

The three sections of this book are: Stress and Respiratory Activity. The Nature of Vowel Quality, and Units in the Perception and Production of Speech.

The first section describes experiments aimed at determining how the lungs and respiratory muscles are used in speech. They lend support to the assumptions of Stetson, Daniel Jones and others that ‘it is possible to make simple statements about the nature of linguistic stress’ and more particularly incline the author to the confident assertion that ‘stress is a gesture of the respiratory muscles which ... can be specified in terms of the amount of work done on the air in the lungs’ and is a ‘measurable bodily activity ... more than something we hear; it is something we do’.

Measurements were made if subglottal pressures, of the electrical activity within various muscles, into which electrodes were inserted (‘electromyography’), and of the variations in volume of the air in the lungs. This last involved the use of a 'plethysmograph', a rigid airtight container enclosing the subject entirely except for head and neck.

Although Stetson's general intuitive suggestions were borne out by Ladefoged's findings his more precise assertions in various ways were often in fact disproved. The general conclusion was that ‘there is certainly insufficient basis for a chest pulse theory of the syllable in normal speech’. However, the results suggest that ‘stress is best described in physiological rather than acoustic terms’ for although there is ‘no single, simple acoustic event that always occurs in all stressed syllables in spoken English — every stress is accompanied by an extra increase of subglottal pressure’.

The long middle section of over 90 pages is based on half a dozen articles and papers and the author's doctoral thesis. As elsewhere in this book, some of the findings were made in collaboration with other scientists — physiologists, communications engineers and a psychologist, D. E. Broadbent, who is responsible for most of the chapter on ‘Adaptation to Different Personal Characteristics’, which discusses the psychological mechanism responsible for the adjustments made when listening to voices which differ only in their personal characteristics. These findings correlate largely with those in other fields of sensory judgment where the theory of ‘adaption level’ was formulated.

There is much in this section of great interest to language teachers as regards the use of visual aids in the teaching of pronunciation for it includes an important critique of the now so widely used International Phonetic Association's cardinal vowel diagram which Daniel Jones originated. The first chapter draws attention to the curious but very little recognised fact that in speech sounds we perceive two kinds of quality — phonetic and personal — in a way quite distinct from our perception of the different timbres of musical instruments. The second chapter deals with the history of vowel description from Robert Robinson (1617) onwards. It makes the point that our present custom of describing tongue postures in terms of two articulatory dimensions — general height of the tongue and advancement of its highest point — was an innovation of Alexander Melville Bell's in his Visible Speech (1867) and has never been substantiated by experimental observation. We learn that the first step away from Bell's ‘boxes’ was taken by Passy nearly thirty years before Jones settled on his first form of the cardinal vowel diagram.

Chapter 3 describes acoustic analyses of sets of the primary cardinal vowels recorded by ‘eleven experienced phoneticians’. (Incidentally the way these 11 subjects are referred to does not make it perfectly clear that the eleventh is Daniel Jones himself.) The remark ‘none of these cardinal vowels are in any way unusual sounds ... each of these qualities might easily have occurred in the normal everyday speech of an individual’ is a little hard to interpret. Surely cardinals 1, 5 and 8 are at least relatively much less usual than the others. The tense voice qualities — no doubt reflecting effort to place the vowel correctly — used by some of the phoneticians in the recordings make it a little inadvisable to relate them too freely to the articulations of everyday speech. Another surprising comment is that auditory equidistance ‘may be a property ascribed to the cardinal vowels solely by their originator’ followed as it is by ‘most of the phoneticians with whom the subject was discussed considered that the interval between each of the first five vowels is greater than that between each of the last four’. Is this to be interpreted as suggesting that Jones's own ear was less reliable than the collective impressions of his trainees? Or that Jones was guilty of preferring to ignore a fact which disturbed the elegance of his formulation? We are not told who or even how many phoneticians were consulted on this matter. It would have been interesting to compare their impressions in this respect with their vowel plottings in the last chapter, but judging from that chapter, such variety of impressions is not surprising. Chapter 3 offers confirmation of the generally accepted view that the first two formants are usually most significant in determining the acoustic quality of a vowel but that the third formant is essential in some cases.

Chapter 4 deals with ‘The Relative Nature of Vowel Quality’ describing experiments employing a speech synthesiser. The interesting conclusion arrived at is that ‘the linguistic information conveyed by a given vowel is partly dependent on the relations between the frequencies of its formants and the frequencies of the formants of other vowels occurring in the same auditory context’.

The last and in some ways most fascinating chapter of all is based on an experiment in which 18 phoneticians were asked to plot on the cardinal vowel diagram 10 vowels of a speaker of a Gaelic dialect not familiar to any of them. ‘Each subject listened to the recording by himself, playing it back as often as he wished and in any way that he found convenient’ (It is not completely clear what this very last remark means). It was intended to correspond ‘as much as possible to the typical situation in which a phonetician needs to be able to describe vowels for purposes of linguistic research. The plottings are presented in the form of several admirable diagrams. There were some startling results. Firstly the three subjects described as ‘good phoneticians with a knowledge of many different languages and experience of dialectology’ who had not, however, received training in recognition of the cardinal vowels were adjudged by the author as ‘nevertheless relatively unable to communicate in writing in an unambiguous way about the quality of a vowel sound’. In terms of peripheral versus central qualities the group of phoneticians of Edinburgh were found to understand each other well. And so were the London group amongst themselves. But each group understood ‘very significantly’ less well the phoneticians of the other university. When it came to unrounded back vowels the trained subjects proved to be little or no better than the three phoneticians with no cardinal vowel training.

There is an interesting table showing the minimum areas that would contain the points representing the judgments of 14 out of the 15 trained subjects: it gives the number of square millimetres of the area and the percentage of the total diagram this amounted to. These percentages range from 0.6 for the [i] to 26.7 for the one near to (secondary) cardinal vowel 16. A welcome addition to this table would have been the maxima of latitudinal and longitudinal placing contrast: for instance, although [a] had less area than [o] it had more than twice the lateral spread of plottings. One set of the discrepancies — the lip-posture judgments — is taken to suggest that this feature is ‘not easy to assess in auditory terms alone ... although 11 out of the 18 subjects considered that the vowel in word J (gaoth) had close lip-rounding [here again it is unfortunately not specified who or what backgrounds], all the subjects who met the informant after the experiment then considered that this vowel had a spread or neutral lip-position. As the author suggests, we can only presume that there are certain vowel qualities which may be produced with either of two or more positions of the tongue but differential lip postures. The whole question of how far each subject after seeing the results was willing to concede error and how far he felt confident or otherwise of his original judgment would have been a most interesting further pursual of the experiment. Clearly the phoneticians from Scotland seem likely to have been helped in some of their judgments by some degree of familiarity with Gaelic vowel systems. A sterner test would have isolated the vowels and mixed speakers and languages so that there was no consciousness of a system of contrasts to influence the plotting. As it was ‘the mean minimum area for the phoneticians' judgments’ of the seven vowels nearest to primary cardinals was ‘under 2 per cent of the total vowel area’. Anyway the practical value of the cardinal vowel system is handsomely vindicated, even if the notation of three always independent variables for vowel specification is shown to be a myth. As Ladefoged says, to abandon it would be ‘to abandon the only internationally known method of specifying vowels at all accurately’.

In the last section the first chapter deals with ‘Temporal Characteristics of Speech’. It describes experiments designed to throw light on the ‘stored patterns’ involved in the perception of speech. It is suggested that the smallest units are not likely to be the size of phonemes. Listeners were shown to have had difficulty in determining the physical order of arrival of individual speech sounds’ — in these cases when clicks and s-sounds were superimposed at points in speech sequences, the subjects being asked to identify those points. Another far from obvious conclusion occurs in the other chapter of this section. ‘Motor and Auditory Characteristics of Speech’, viz that ‘correct pronunciation ... almost always precedes perception’. The final suggestion is that ‘speech production probably consists of stored units or target values corresponding to vowels and initial and final consonant allophones’.

This book is copiously and excellently illustrated. It has valuable bibliographies to each section. The Oxford University Press deserve our gratitude for making it available in their so reasonably priced Language and Language Learning series.
J. Windsor Lewis
The above review appeared in ENGLISH STUDIES 53. The Netherlands 1 February 1972.