Sounds clear and riveting

The declared aim of this book is to be "an advanced tutorial introduction for students with a basic grounding in phonetics who are interested in acquiring a foundation for ... research in the phonetic sciences".

The obvious comparison is with the 1968 Manual of Phonetics edited by Bertil Malmberg. This work, however, is much longer, reflecting growth in speech science over three decades. Techniques for probing articulation and processing acoustic signals have been developed, computers are now used to decode air-waves into words, and the advent of "human-centred information technology" has expanded the applications of the discipline.

This handbook offers an excellent way of catching up with these advances. Its list of contributors could hardly be more distinguished. Most chapters offer admirably clear and up-to-date surveys of their topics. The first chapter, by Maureen Stone on laboratory techniques for investigating articulation, took me in 22 pages from initial ignorance to a satisfying grasp of the fundamentals of techniques such as computed tomography and magnetic resonance imaging. Its quality is typical of the chapters that follow (though some make larger assumptions about the readers' knowledge of relevant aspects of mathematics or physics).

The handbook does not render the earlier manual entirely obsolete. I tested it on an issue that cropped up in my teaching recently, about formant-frequency differences between the sexes. A chapter by Gunnar Fant in the Manual included two or three pages on this topic. The Handbook index led me only to a bare mention of the existence of such differences, in Francis Nolan's chapter on speaker recognition and forensic phonetics. I found a more explicit, unindexed statement in Kenneth Stevens's chapter, "Articulatory-acoustic-auditory relationships", but this said far less than Fant said in 1968. The main Handbook chapter on acoustic phonetics, by Osamu Fujimura and Donna Erickson, seems less successful than many other contributions, reading as if it were put together in a hurry - unfortunate, in view of the central importance of this topic. Most contributors are conscientious about surveying the main features of their assigned field in an even-handed fashion, but quite a lot of the chapter deals with Fujimura's idiosyncratic analysis of English phonology, which makes English look more like Japanese than it normally does, and seems out of place here.

Although editors and publisher are British, the "centre of gravity" of the handbook is American - a fact which reflects shifts in the subject itself. More than half of the 26 chapters are contributed by authors working in the US. A generation or so ago, phonetics was a subject in which Britain played a leading - perhaps the leading - role, and British phonetics had a distinctive agenda emphasising detailed observation of the sounds found in all varieties of human speech, including exotic languages and nonstandard dialects.

That agenda is largely absent from this book. Most chapters assume a world in which people speak standard varieties of American English or a handful of other well-known languages. The few exceptional chapters are mostly by cisatlantic authors (such as the chapter on voice quality by Ailbhe Ni Chasaide and Christer Gobl of Dublin), or by authors such as Peter Ladefoged who work in the US but are British by training. (A 1996 book by Ladefoged and Ian Maddieson, The Sounds of the World's Languages, complements the handbook in this respect.) This emphasis on familiar languages is partly a consequence of the increased technical sophistication of instrumental and computational phonetics, that inevitably draws mainly on languages spoken in the vicinity of expensive laboratories; it is reinforced by the modern pressure on academics to make their research responsive to the needs of the taxpayers who fund it, which is only fair. Nevertheless, from a humanistic point of view this retreat from engagement with the full diversity of human speech seems regrettable. Much of the liberal educational value of language studies used to lie in the way that they revealed astonishingly subtle and diverse organisation in the speech even of obscure tribes.

Ladefoged's chapter on linguistic phonetic descriptions is an exception to another trend in this volume. Most contributors give no credence to the once-fashionable idea that speech sound encodes a system comprising a few neat patterns embedded in the brains of our species as part of an inherited "language organ". Bertrand Delgutte's chapter on the auditory neural processing of speech lists numerous animal species which have been found to hear speech sounds in much the same way as us.

Ladefoged (whose stature within the discipline is justly reflected by his being the only author allotted two chapters) takes a line which has always seemed oddly schizophrenic. On one hand, he more than perhaps anyone else draws attention to the fact that sounds unlike any in familiar languages sometimes occur with clearly phonemic status in one obscure language. Pirahi, spoken by some people in Brazil, has a double flap sound which involves protruding the tongue so that it almost touches the chin. On the other hand, Ladefoged continues to pursue a programme, introduced in his 1971 book Preliminaries to Linguistic Phonetics, of finding a system of binary or few-valued features adequate for describing the sound patterns of all languages.

If speech sounds were all definable in terms of a few distinctive features, how could the Pirahi oddity fit in? Some linguists have argued that diverse sounds are shown to be related abstractly by behaving phonologically as "natural classes", but Ladefoged rarely discusses data of that sort.

John Ohala, in a handbook chapter which queries the validity of phonology as an independent subject, concludes that natural classes of sounds may play no role in speakers' grammars. Ladefoged merely says that although strange sounds do not fit the theory, if they are rare enough that does not matter. "Any addition complicates the theory. It is just a matter of the price. We always have to decide whether the extra coverage is worth the cost."

Surely this takes a Thatcherite approach to science beyond reason? Scientific theory should be true, not cheap. It is one thing to advocate a theory despite a few awkward phenomena that one has not yet managed to reconcile with it; it is another to claim that a theory is true even though it is contradicted by evidence that one has no intention of reconciling.

Ladefoged's distinctive-feature programme harmonises with ideas that exerted a powerful hegemony in America for many years, but (on the evidence of this book) the rest of the speech science community has now definitively junked.

The production of the handbook is attractive, though the editing could be better. The index is dire: again and again, for instance, it omits mentions of personal names, defeating one of the easiest ways for readers to locate elusive passages. There are few misprints, but I found several wrong phonetic symbols.

In the context of an achievement as large as this publication, however, my criticisms are little more than quibbles. This is an important book, which does much to make accessible the current state of knowledge in an increasingly significant discipline.

Geoffrey Sampson is at the school of cognitive and computing sciences, University of Sussex.