The sound in the symbol

When Red Thunder Cloud died at the beginning of this year, the language of the North American Catawba tribe ceased to exist. He was its last speaker, and The Times honoured his passing with a leader on January 17. It is a ruthless fact that spoken languages become extinct, and will continue to do so. But this fact has little or nothing to do with a much more durable phenomenon: the use and perpetuation of writing systems.

It has taken a while to reach a clear definition of a writing system, as opposed to the false notion of a "written language". No one has yet invented a purely written language, that is, a set of symbols capable of expressing any or all thought that could be written and read without any concern for language as it is pronounced. But it is still sometimes naively proposed that there have been such systems.

Even Jean-Francois Champollion (1790-1832), the decipherer of Egyptian hieroglyphs, once believed that the hieroglyphs were not meant to be read aloud. His change of mind was crucial to his eventual success. His extraordinary achievement was founded on inscriptions such as the Rosetta stone which bears a Ptolemaic decree in three scripts, Hieroglyphic, Demotic and Greek. The Greek text of the Rosetta stone revealed how to pronounce the hieroglyphs for royal names, such as Ptolemy, which are distinctly located in the Hieroglyphic text by their surrounding cartouches. Champollion had assumed that the phonetic use of hieroglyphs was unique to foreign (in this case Greek) names, but in 1822 he realised that nearly all the hieroglyphs could be assigned phonetic values. This helped him to identify the sounds and significance of whole words.

Despite the popular frenzy caused by Champollion's decipherment, more than a century passed before a majority of scholars accepted the importance of basic phonetic principles underlying all writing systems. Studies of writing systems have eschewed many fallacies over the years, but certainly none more obstructive than the blindfold that separated sound from symbol.

It was not until the present century that studies of the writing of particular languages began to mature into a concern with the theory of writing systems, which underlies the three books under review. The late Ignace Gelb, who contributed to deciphering Hittite Hieroglyphic, pioneered this approach by establishing the study of writing as the science of grammatology. In A Study of Writing (1952), Gelb proposed a rather strict principle of evolution that has since been questioned but nevertheless provides a useful tripartite typology of writing systems, consisting of logographic, syllabic and alphabetic.

Logographs are isolated graphs signifying complete words, such as Sumerian signs, many (but not all) Egyptian hieroglyphs and Mayan glyphs, and Chinese characters. With the exception of Chinese characters, these early logographic systems developed by re-employing logographs as symbols for syllables within polysyllabic words or names. Modern Japanese uses a writing system that combines logographs borrowed originally from China, called kanji (the word means Chinese characters) with two Japanese-invented syllabaries of 51 signs each. The third system in Gelb's typology, an alphabet, employs signs that represent the isolated phonemes of the spoken language.

Today, the system that contrasts most strongly with all others is the Chinese. Chinese writing was once subject to the same misconstrual as Egyptian hieroglyphs. Chinese characters were thought to be ideographs, that is signs representing things or concepts that worked on the mind of a reader without any kind of intervention from the sounds of words. A reader, on sighting the character for "sheep", for instance, was thought somehow to push a "sheep recognition" button in his or her brain, irrespective of how the word for sheep was pronounced. What partly occasioned this misguided thinking was the fact that the pronunciation of Chinese varies so radically throughout China, particularly in the south. If two or more readers could pronounce the same character very differently, yet have the same meaning in mind, then surely the link between sound and symbol was exceedingly weak or even, according to the most facile explanation, nonexistent.

This deceptive reasoning about Chinese characters failed, most amazingly, to recognise that by extension of its own logic one ought to be able to learn to read Chinese characters without speaking or studying the Chinese language - a wholly untenable position of course. Chinese is monosyllabic and the Chinese writing system is logographic. Each syllable in Chinese is a meaningful word, and each of these words is written with a single character, although hundreds of words are compounds of two or more words, for example shankou "pass" (shan "mountain" + kou "mouth").

The oldest characters display their origin, written on "oracle bones" and turtle shells, more than 3,000 years ago, as pictographs (see facing page). But such pictographic origins are quite distinct from the actual functioning of the modern descendants of these pictographs today, which, if it were purely pictographic, would entail that the modern reader instantly recognise the "sheep" character as a picture of a sheep. Moreover, and even more important, to expand a writing system by creating pictographs to identify separately every known object, phenomenon and sensation would have been impossible.

The most significant role of these early Chinese characters was actually as symbols of sound, and herein lies the explanation for the origin of the Chinese writing system, an origin remarkably similar to the development of, say, cuneiform, Egyptian hieroglyphs and Mayan glyphs - despite the later divergence of Chinese characters from other writing systems. At some stage in the history of all these systems it was recognised that a symbol representing a word/sound could be re-employed to represent that sound in any number of words, regardless of the original word's meaning. The process is called rebus writing (properly paranomasia), and is probably familiar to most of us as the "sounds like..." tactic in games of charades: "sun" (in the sky) for "son" (of his father), "belief" as a combination of "bee" and "leaf". The Chinese "sheep" graph is used in six words all pronounced yang, none of them meaning sheep, each of them given meaning by the addition of a signifier/determinative (see facing page). Chinese signifiers play the same role as rebus writing in all the other ancient writing systems.

Probably no one reviewer has the knowledge to test the accuracy of every contribution to The World's Writing Systems, a collection written by 79 experts and edited by Peter Daniels and William Bright. The book contains a section covering the Near East that includes the Mesopotamian and Old Persian Cuneiform scripts, the Hieroglyphic, Demotic, Hieratic and Meroitic scripts of Egypt and the Sudan, as well as the Iberian, Berber and Aegean scripts. The Middle Eastern section covers the Jewish scripts and the Aramaic scripts used for writing the Aramaic, Iranian and Altaic languages, in addition to Arabic and Ethiopic scripts. In the European section comes the Phoenician script, the Greek, Anatolian, Roman, Coptic, Gothic, Slavic, Georgian and Armenian alphabets, as well as the Runic script and Ogham. The South Asian section deals with the Brahmi and Kharoshthi scripts and their derivatives, namely Devanagari, Gujarati, Gurmukhi, Bengali, Oriya, Sinhala, Kannada, Malayalam, Tamil and Tibetan. Further derivatives of Brahmi script, such as those for writing Burmese, Khmer, Thai and Lao, are surveyed in another section on Southeast Asia. The East Asian section surveys writing in China, Japan, Korea and the writing of the Tangut, Kitan and Jurchin peoples, that is the principal Siniform scripts adapted from Chinese characters. The seal script from the Indus Valley and the Rongorongo script from Easter Island - both of them undeciphered - and the Mayan with other Mesoamerican scripts, are discussed in a section entitled "Decipherment".

In a section devoted to scripts invented in modern times, we have the Cherokee and West African scripts, while another section concerning the adaptation of the Roman and Cyrillic alphabets discusses, for example, the use of the Roman alphabet for writing Vietnamese. There is also a section on secondary notation systems, such as music and numbers. Finally, a section on sociolinguistics deals with the coexistence of scripts and their dominant religious or political colouring. (Children in Yugoslavia were taught to write in both the Cyrillic and Roman scripts, but the recent war has diminished the teaching of Cyrillic in Croatia and the Roman script in Serbian areas.) The editors state that their survey deals primarily with how the sounds of languages are represented in writing. However, the book does not provide consistently successful answers to the question. Unsurprisingly, not every one of 79 contributors shows an equal concern with the issue; some hardly tackle it at all. Still, each section contains a useful sample of writing accompanied by phonetic values, glosses and a translation, all of which provide a good background before one reads the relevant section.

There is also sometimes a slight lack of coherence. The exciting story of how sound values were ascribed to the glyphs of Mayan writing, for instance, and thence the successive breakthroughs in interpretation, is given in the section on decipherment by Daniels. But this information is inconvenient to integrate with the Mayan subsection proper by Martha Macri, since it is presented by a different author and is located separately. Although the scholarship looks impeccable and the illustrative data provide insights second only to actually learning the system in question (as the data do throughout the book), this particular arrangement is somewhat cumbersome.

Together, the various sections present an almost overwhelming amount of material, particularly since the ones dealing with the earliest writing systems, those of the Near East and Egypt, involve some highly technical linguistic detail. However, any section repays careful reading, and this book is without question an invaluable source for introducing the phonetic, linguistic and orthographical principles governing the writing on tablets, papyri, stones and bones that we see in museums and libraries around the world.

The Blackwell Encyclopedia of Writing Systems is the sort of delightful encyclopedia that once opened is hard to put down. It is an A-Z of ancient and modern writing systems with further entries on leading archaeologists, linguists and the terms used in linguistics and grammatology. The book is richly illustrated in black and white with specimens of writing, diagrams and charts. In this respect, Florian Coulmas does not stint on including whatever is entertaining and evocative: for instance, a toy jug of the 7th century bc in the shape of a cockerel and inscribed with the Etruscan alphabet, now located in the Metropolitan Museum in New York.

The encyclopedia does not offer the scholarly depth of The World's Writing Systems, but it plugs a few of the gaps in the larger book. The writing systems of Vietnam - a not under-populated region of the world - receive fuller treatment from Coulmas, including consideration of the Vietnamese use of Chinese characters. Nguyen Dinh-hoa, the relevant author in The World's Writing Systems, describes only the romanised writing system of Vietnamese that was initially introduced into Vietnam by the Jesuits in the 17th century. Vietnamese characters used in the preromanised system called Chu'nom ("southern writing") were borrowed or adapted from the Chinese system according to several principles, one of which entailed borrowing Chinese characters for their phonetic values but not their meaning, so as to represent the sounds of Vietnamese words. Thus, the Vietnamese character for the verb an,"to eat",uses the Chinese character for an, "peace"; the original meaning "peace" is redundant.

Sometimes, to maintain a distinction, a signifier is added. In the case of an, it is a small square enclosure, the "mouth" signifier. The resultant character would not be familiar to any reader other than a Vietnamese speaker (no Chinese speaker would recognise it). Vietnamese was able to borrow from Chinese in this way because it too is a monosyllabic language. Though Coulmas does not comment on the point, the Vietnamese case demonstrates an extremely interesting repetition of linguistic history. The process by which Vietnamese scribes borrowed homophonous Chinese characters for the words they needed to write follows exactly the same principles of rebus writing observed in the early stages of the formation of the Chinese writing system.

In The Origin and Early Development of the Chinese Writing System, William Boltz discusses at length the development of the Chinese writing system in the widest possible context of early writing in other parts of the ancient world. It is a very careful and intelligent discussion, much of which avoids intricate Sinology or linguistics too involved for those who do not know Chinese.

Boltz's theories are published in the Chinese subsection of The World's Writing Systems, and his contribution there is one of the notable few to confront squarely the question of how the sounds of a language are expressed in written form. In his book, Boltz challenges the persistent misapprehension that Chinese is an ideographic script, and shows that ideographs are an entirely false concept in the context of writing systems. He also argues that, again contrary to popular formulations of the origins of writing systems, pictographs belong to a precursor stage in the emergence of a writing system. He is quite right in urging a distinction between, say, a drawing of a sheep and a visually exactly similar zodiograph (from Greek zodion, "small picture") of a sheep standing for yang. This distinction emphasises the need to understand graphs as representatives of words rather than things: it is this intervention of speech between thing and word that is such an important step in the maturing of a full writing system.

The story of how Chinese writing developed is known in particularly rich detail, and Boltz relates it with great vividness. His final chapter concerns the standardisation that began in the third century bc, and addresses the absorbing question of why the Chinese did not allow an alphabet to emerge. One of the most fascinating comments by Boltz is based on highly convincing evidence that Chinese writing at one point, about two millennia ago, reached a stage of development when it seemed to be tending towards a syllabary. Several semantically distinct words sharing the same phonetic value during this period were written with the same graph. That matters did not run their natural course was a result, Boltz thinks, of a philosophy of preserving at all costs the myriad one-to-one correspondences between things and words. To the Chinese, unlike the West, the idea of writing the name of a thing with more than one graphic sign intimated chaos.

Oliver Moore is a curator, department of oriental antiquities, British Museum, with special responsibility for the current exhibition, Mysteries of Ancient China.