Empires of the Word - Nicholas Ostler [8]
But there was always a subjective element in learning derived from recitation, as well as a practical limit on the amount that could be retained—unless perhaps complementary teams of record-keepers could be organised. Moreover, speaking now from the anachronistic point of view of the modern historian, there would always be a tendency to inauthenticity in ancient records held in memory. In use, there was always a pressure to update them little by little to meet the needs of the contemporary world: otherwise, as gradual changes accumulate in social institutions and in the language too, really ancient records would tend to become both irrelevant and incomprehensible. Even today, when oral traditions can be found intact, it is seldom possible to gain clear, unambiguous information about the past from the testimony of rememberers. Recall is an act of disciplined reimagination, and the remote past may be beyond anyone’s ken.
All this is resolved through the miracle of writing. Writing traditions usually begin in some kind of process of accounting records—at least tallies and tokens are often the earliest clear predecessors of written documents to survive—the intent being to provide objective proof of the quantities involved in some transaction. But with practice it often became clear that the symbols were in principle capable of recording any message, and as facility in handling the symbols grew they became usable as a direct aide-mémoire even for fluent speech.
Once a culture has written documents, the first traces begin to be laid down which will later enable the history of the language to be written. If the writing system has a clear link to the language as spoken (and, despite the usual symbolic start in numbers and concepts, in practice it is impossible to develop a fully functional writing system without reference to words in spoken language), then the mute stones or clay tablets or preserved animal skins—whatever—begin to reveal to us something we might have thought quite evanescent—how the language was actually spoken, perhaps thousands of years ago.*
All the languages whose careers we shall consider have written histories that extend back over a thousand years, and sometimes two or three times this long. In almost every case, literacy is a skill that was learnt from visitors or neighbours, and then became part of a language’s own tradition. As it happens, with the exception of Chinese, even the languages that originated writing, and so made the earliest use of it, have dropped their original system, and borrowed another,†
The past careers of languages are as diverse as the worlds that each language has created for its speakers. They have suffered very different fates: some (like Sanskrit or Aramaic) growing to have speaker populations distributed across vast tracts, but ultimately shrinking to insignificance; others (such as the languages of the Caucasus or Papua) twinkling steadily in inaccessible refuges; others still yielding up their speakers to quite different traditions (as in so many parts of North and South America, Africa and Australia). Some (such as Egyptian and Chinese) maintained their speakers and their traditions for thousands of years in a single territory, defying all invaders; others (such as Greek and Latin) spread by military invasion, but ultimately lost ground to new invaders.
Often enough, one tradition has piggybacked on another, ultimately supplanting it. One big language parasitises another, and in a coup de main takes over the channels built up over generations. This is a common trick as empires succeed one another, in every time and continent: Persia’s Aramaic made good use of the networks established for Lydian in seventh-century Asia Minor; in the sixteenth century, Spanish usurped the languages of the Aztecs and Incas, using them to rule in Mexico and Peru; and in the early days of British India, English and Urdu gained access to power structures built in Persian. But the timescale on which these changing fortunes have been played out is astonishingly varied: