Visible marks of the move from sound to stone

The spur to invent written language was so strong it occurred more than once. Andrew Robinson reflects on bone tags, runes and other inscriptions

The reason the first writing will continue to intrigue us," writes Stephen Houston in the final sentence of this state-of-the-art collection of articles on the origins of writing edited by him, "is that it forms an elastic and untameable frontier of knowledge, made wonderful and mysterious and enticing by the limits of evidence." How, where, when and why human beings first learnt to make their thoughts visible are certainly all those things, while the material evidence of the earliest writing is unquestionably patchy and far from conclusive despite some two centuries of archaeological excavation.

Until the 1960s, as epitomised in I. J. Gelb's A Study of Writing , the consensus was that writing - meaning a script capable of expressing at least some of the complexity of spoken language - began in Mesopotamia (Iraq) around 3300BC with inventories of numerical and pictographic signs scratched on to clay tablets. These signs rapidly developed into a more abstract script known as cuneiform, able to express the entire range of human thought, which was used to write various spoken languages (such as Babylonian, Assyrian and Old Persian) in the Near East for three millennia until the time of Christ.

The motivation for writing's invention was said to be accountancy; the elite needed to keep track of the increasing economic complexity of society by creating written records with a degree of permanency. Writing apparently was invented in Egypt a little after it was in Mesopotamia, in about 3100BC, in the Indus Valley around 2500BC, in Europe (Crete) around 1900BC, in China (Anyang) around 1200BC, and it did not exist in the New World, since Aztec and Maya writing were thought not to represent language, while the Inca civilisation had no writing at all, only a complex form of counting (the knotted quipu). Whether the idea of writing - but not of course the particular sign systems, which were as various as cuneiform wedges, Egyptian hieroglyphs and Chinese characters - spread to these other places by borrowing from Mesopotamia, or was instead invented independently in, say, Mesopotamia and China, was not settled. But a majority of scholars favoured the monogenesis of writing in Mesopotamia, in tandem with the notion of the Fertile Crescent as the "cradle of civilisation".

Today, much of that 1960s consensus is in flux. Scholars divide quite sharply on the question of how to define writing. Must writing represent spoken language, as in John DeFrancis's Visible Speech: The Diverse Oneness of Writing Systems (1989), or can it include systems of entirely non-phonetic notation such as those used in mathematics, music and road signs, as in Roy Harris's The Origin of Writing (1986) - not to mention Aztec picture language and Inca quipus?

Another question is how much phonetic information was encoded in the earliest Mesopotamian scripts, and did they really evolve quickly into "full" writing? Or could "full" writing have developed through the idiosyncratic increase of proto-writing's expressive range in different ways (and at different rates) according to the needs of different societies? The philologists who have dominated the study of scripts, such as Gelb, always tend to view the origin and evolution of writing purely in terms of the linguistic structure of a script and its language. Today, the trend is away from this, in the direction of seeing a writing system as part of a whole culture. This view of writing undoubtedly provides a better explanation of why the ancient Egyptians chose not to reduce their complex hieroglyphic script to their simple set of about 24 uniconsonantal signs in the 2nd millennium BC - comparable to our use of the 26 letters of the English alphabet - and also why the Japanese today hang on to thousands of kanji (Chinese characters) in their Chinese-derived writing system, despite its mind-boggling complexity, rather than resorting purely to their long-standing simple kana syllabary. Egyptian hieroglyphs and Chinese characters enjoyed great prestige in their cultures.

Moreover, if accountancy was so important a motivation for inventing writing, why is there such meagre evidence of accountancy in the earliest Egyptian and Chinese writing? How much of this writing was on perishable materials such as papyrus and bamboo - as opposed to baked and fired clay tablets - that have simply rotted and disappeared? Are the exceptionally early primitive marks and pictograms that have been discovered in recent decades in Egypt and in China, evidence of Egyptian and Chinese writing before 3100BC and 1200BC?

With the exciting decipherment of the Maya script beginning in the 1970s, we now know for sure that the Maya had a full writing system with phonetic representation. Furthermore, it too shows little evidence of accountancy but appears to have been religiously motivated. Even earlier Mesoamerican systems, pre-Maya, appear to date the appearance of writing in the New World to about 600BC. Since it strains credulity to think that these New World systems were stimulated by the diffusion of the much older writing of the Old World, the majority of scholars now accept that writing was invented at least twice - in the Old World and in the New. The majority probably now favour independent invention of writing in the Old World, too, especially in China but possibly also in Egypt, rather than monogenesis in Mesopotamia.

The First Writing , which originated in a conference organised by Houston in the US in 2000, explores all these issues and more in scholarly depth. The ten main articles, which are copiously illustrated and referenced, cover the definition of writing and its cultural evolution in many societies, and focus on the earliest cuneiform, on the partially deciphered proto-Elamite script of Iran, on Egyptian writing (including a detailed discussion by John Baines of the discoveries of bone tag inscriptions in tomb U-j at Abydos dating from about 3200BC), on Chinese writing, on European runes and on the Mesoamerican scripts. (The important undeciphered scripts of the Indus Valley and Easter Island are understandably omitted.) Most of the articles are readable by non-specialists in these scripts, though some skipping will be required, especially in Robert Englund's highly technical, if excellent, account of proto-Elamite and in Houston's survey of the Maya script and its possible Olmec, Zapotec and Isthmian origins.

As editor, Houston, who has strong views of his own, admits to some frustration at being unable to corral all the contributors into framing key questions about writing in his own particular way. But he is right to conclude that a heavier editorial hand would have been self-defeating. For the study of writing is not a science with established universals; it is shaped by the very different societies it embraces. Thus, at the most fundamental level, there is severe disagreement between most of the contributors and the final contributor, Elizabeth Hill Boone, who maintains that writing does not have to represent language but can be defined as "the communication of relatively specific ideas in a conventional manner by means of permanent, visible marks." As for the controversial question of monogenesis versus independent invention, the two contributors on Chinese, Robert Bagley and Françoise Bottéro, differ radically: Bagley thinks that the writing of Anyang probably developed independently and slowly out of earlier Chinese cultures, only now being excavated and understood, while Bottéro favours Chinese borrowing of writing with rapid invention of the characters, probably under the original influence of Mesopotamian writing.

Nevertheless, a few statements in the book should have been challenged by the editor before publication. John Robertson, in his article on "The possibility and actuality of writing", constructs a theory of the relative importance of sound and vision in writing: that "acoustic signs do not emphasise the iconic relationship between sign and object" - unlike visual signs. As evidence, Robertson cites Roman Jakobson's contention that one cannot identify "railroad stations and trains, streets, harbours, sea, wind, (or) rain" from tape-recordings of such things - which is plain wrong. Boone, in "Beyond writing", claims that "the computer, with its vast storage capabilities, is also changing the fundamental concepts of writing, reading and thinking". She gives no evidence for this astonishing claim. Personally, having written books both in longhand and on computer, and done research for them in traditional libraries and on the internet, I feel that the computer's effect on the fundamentals of writing, reading and thinking is nil, or at most very superficial. As for Boone's belief that "mathematics and science have largely outgrown (alphabetic writing)", one can only ask her to spend an hour or two reading scientific papers in journals such as Nature or Science . Even high-level mathematics books require considerable amounts of alphabetic writing to gloss the mathematical symbols.

If I were a Japanese or Chinese, I might agree somewhat more with Boone's statement about computers and writing - though for reasons quite different from hers. Computers have posed a major challenge for the word processing of Japanese script ever since the 1980s, because of the difficulty of creating a keyboard that can deal with several thousand kanji , instead of 26 alphabetic letters. The jury is still out as to whether the problem can be solved or whether the Japanese must adapt to inputting their language in a romanised form - which would be equivalent to reducing their script to the syllabic kana for the purposes of word processing.

J. Marshall Unger, an American professor of Japanese with a Japanese wife, has long been interested in this issue. In 1987, he published The Fifth Generation Fallacy , a prescient book about the Japanese love affair with artificial intelligence, and declared that the development of computers that could conveniently process Japanese kanji was unlikely. In Ideogram , his latest contribution on the subject, Unger, with characteristic depth of knowledge and breadth of experience, precision of exposition and original analogies (for example, a comparison of Chinese characters with American Gregg shorthand), analyses the nature of Chinese and Japanese writing and the confused notions of it still common in the West. The target readership appears to be beginners in these languages, with some knowledge of the characters, but others willing to make some intellectual effort could certainly cope with all but a few sections.

Commenting on computer and internet use in Japan in recent years, Unger concludes that as Japanese consumers have discovered the limitations of kanji word processing, they have increasingly turned to romanised word processing of Japanese. "When doing transcriptive input," says Unger, "users must constantly check for the wrong kanji and make style choices. They must refrain from accepting whatever the program offers first, especially when writing extemporaneously, because most programs simply suggest whatever kanji was most recently used for a particular pronunciation. Things are not made easier by manufacturers who vie to cram ever more kanji and archaic forms into their look-up dictionaries; this practice just increases the need for users to be cautious and make choices."

Unger's account is a world away from the utopian European Enlightenment myth of the Chinese characters as a "universal" communication system, independent of the babel of spoken languages, capable of uniting all intellectuals in the Republic of Letters through a set of written signs that did not employ phonetic symbols - the disembodied ideograms of Unger's title. This myth has never entirely dissipated, as Boone's article yet again confirms. Ideogram and The First Writing should help to lay it to rest. For if there is one thing that unites all the ancient scripts despite their visual diversity, it is that full writing cannot be divorced from speech; words, and the scripts that employ words, involve both signs and sounds.

Andrew Robinson, literary editor of The Times Higher , is the author of The Story of Writing: Alphabets, Hieroglyphs and Pictograms and Lost Languages: The Enigma of the World's Undeciphered Scripts .