How Writing Was Invented: From Clay Tokens To The Alphabet

AWriting is usually presented as the moment at which speech acquired a durable form, an invention whose purpose is taken to be self-evident: people wished to preserve what would otherwise be forgotten. That account, which survives in most textbooks, carries a second assumption, namely that scripts improved steadily as their designers learned to represent sound more economically. On this reading the history of writing is a ladder, ascending from cumbersome pictures through syllabic signs to the twenty-odd letters of an alphabet. The ladder is tidy. It is also hard to reconcile with what excavation has produced, because the earliest legible documents from Mesopotamia, Egypt and the Yellow River valley are overwhelmingly administrative, and nothing resembling a story, a prayer or a law appears in the record until several centuries after the first signs were pressed into clay. A survey completed in 1998 of the roughly 5,800 published tablets from the earliest phase at Uruk classified 87 per cent as records of commodities, personnel or land; fewer than 40 could be read as continuous prose.

BThe most sustained challenge to the ladder comes from Miriam Delacroix, a palaeographer at the Brantwood Institute for Ancient Economies, who maintains that writing was not devised in order to record language at all, but in order to settle quarrels about ownership. Her starting point is a class of object that predates the tablets by millennia: small clay counters, shaped as cones, discs and tetrahedra, which occur at farming sites across the Near East and appear to have stood for measured quantities of grain, oil and livestock. Sealed inside hollow clay envelopes, the counters could not be examined without breaking the seal, and pressing each counter into the soft outer surface before enclosing it produced, in effect, a list on the outside. Once that outer list was trusted, Delacroix argues, the contents became redundant. Between 2004 and 2011 her team re-examined 42 sealed envelopes in three museum collections; in 19 of them the impressed marks did not correspond to the counters within, a discrepancy she reads as evidence of amendment after sealing rather than of carelessness.

CA list of impressed shapes is not yet writing, since it fixes amounts without fixing words, and the step that matters is the one that ties a sign to a sound. Delacroix locates that step in a narrow and rather unromantic place: the recording of personal names. A sign depicting a reed could be borrowed for any name containing the syllable with which the word for reed began, and once scribes accepted such borrowing for names they had, without intending anything of the sort, invented phonetic writing. The pattern holds, she concedes, only in archives where goods moved between parties who did not know one another personally, since a village storekeeper had no need to write a neighbour's name down. Within the same Uruk corpus, signs used for their sound value cluster in the columns reserved for individuals in 71 per cent of the disputed consignments, against 17 per cent elsewhere on the same tablets. The mechanism, on this account, is a response to distrust rather than a search for expressiveness.

DAnton Reisz, an archaeologist attached to the Kestrel Foundation, is sceptical of the whole construction, and his objection concerns where the evidence comes from rather than what it says. Temples and palaces were excavated first because their walls could be found, and their storerooms yield tablets by the thousand; the houses of ordinary people, dug rarely and published still more rarely, cannot be expected to produce a comparable picture. Writing may look administrative, he contends, merely because administrators occupied buildings worth digging. To test the point Reisz has spent years walking ploughed fields at 34 rural sites, collecting inscribed potsherds, a surprising number of which carry marks that are neither numerals nor names. The difficulty is that surface material cannot be dated within less than about two centuries, and his published corpus amounts to 63 fragments, a figure he himself calls thin. Delacroix, unsurprisingly, dismisses so small a sample as no threat to an archive of thousands.

EThe alphabet occupies a comparable position in popular accounts, being credited with bringing reading within reach of ordinary people by cutting hundreds of signs down to a couple of dozen. The reduction is real enough; the consequence is not. Working through school exercise tablets and legal witness lists, Ceren Aydin of the Halloway Centre for Comparative Literacy has compared communities that took up alphabetic scripts with neighbouring communities that kept syllabic ones, and finds no difference in the proportion of documents signed rather than sealed for nearly six hundred years after the change. Literacy widened, on her evidence, only where writing surfaces became cheap, which in the eastern Mediterranean means only after supplies of imported papyrus and prepared skins had been secured. A simple script with nothing affordable to write on, she insists, buys very little. Aydin accepts that her measure is crude, since a signature demonstrates the ability to write one name and nothing more.

FWhat survives of the argument, once the sharper claims are set aside, is a modest and rather uncomfortable proposition. Writing appears to have emerged not from a wish to speak across time but from the need to state an entitlement that a stranger would respect, and phonetic signs seem to have entered by a side door, carried in on the names of individuals; yet Reisz is right that the record is skewed, and any statement about what ordinary households did with clay remains a statement about what has not been found. Both pressures were probably at work, in proportions that differed from one valley to the next. The undeciphered scripts sharpen the discomfort. About a dozen systems, among them the seals of the Indus cities recovered since the excavations of 1922, have resisted every attempt at decipherment, and several of them are short, repetitive and bureaucratic in appearance, which is exactly what the accounting hypothesis predicts and what makes them impossible to read.