The attractions of French in early modern Europe stemmed from the abundance delivered by the French economy, and much the same is true of English today. Somehow, speakers feel that they can share in the wealth by accepting the language. But the prestige of Latin, in the years of the Renaissance and after, and of Greek, in the heyday of the Roman empire, came not so much from abundance of wealth as from wisdom-perhaps itself a result of positive a.s.sociations, since it is only when wealth is in good supply that that luxury product, education, can also be afforded. This underlay the attraction of Chinese and Akkadian: the sheer inaccessibility of written competence in these languages, paid for through a decade or more of study, added greatly to their prestige, and hence curiously their attractiveness.

And certain sorts of knowledge also offer greater access to wealth: the Greek sophists of the fifth century BC showed this, making the power of persuasive speech available at a fee; modern governments are buying into the same motive when they see competence in English as the road to economic development.

kai toto plen e murion est' axion stateron

hairoumenon tous hettonas logous epeita nikan

And this is worth more than 10,000 staters to take the weaker argument and then win Aristophanes, Clouds, 11. 1041-2 (Athens, 423 BC) All prestige languages give access to a special enjoyment, because they all have extensive literatures, and the first purpose of literature is to give enjoyment to the people who can appreciate it. Usually, knowing that not many others can share the appreciation has been part of the pleasure. This has been a charm of cla.s.sical languages down the ages, from the Akkadian epics recited in the Hitt.i.te tablet-house of the thirteenth century BC to the Persian poetry quoted in seventeenth-century India and the French novels read in nineteenth-century Russia. But in the present age the charm of the prestige language, English, is somewhat reversed-perhaps as a side effect of the first globalised market in culture. English, especially as the American entertainment industry promotes it, is meant to convey that its culture is universally accessible, that it gives a release from other languages' traditions and restraints. The purported special link with freedom is part of this, but as we have just argued, this is hard to sustain as serious politics. All the same, if you are rich, it's much easier to be free and irresponsible.

Despite the current vogue for English as the language of the young and free--as well as the learned and the rich-ultimately the a.s.sociation of a language with profound religious truth gains the most loyal adherents, creating a reputation that may last for thousands of years. This is the only basis for a language to claim a value above simple a.s.sociation with some historic success. Hebrew, Sanskrit and Arabic all claim a mystic force which goes beyond the mere expression of meaning, or exchange of information among speakers.* As such, they may vanish from everyday discourse, but can never be demeaned as old fashioned or irrelevant, as long as there are believers to revere and treasure them.

Languages sometimes spread without any of these forms of prestige, of course. Brute military force can be powerful too, and it is difficult to see any charisma arising from the spread of Turkish to Anatolia, Spanish to Peru, Russian to Siberia, j.a.panese to Korea, or indeed English to Ma.s.sachusetts. This is not to say that the conquerors will not have found their own behaviour impressive; especially in pre-literate societies, they may celebrate their conquests in word and song. Such heroic poetry is good for an indigenous literature, but it is unlikely to appeal to outsiders, let alone the conquered.

A prestige language, in general, is any foreign language that is learned for cultural advantage. Sumerian, Akkadian, Chinese, Sanskrit, Greek, Latin, Arabic, Turkish, Persian, Italian, French, German and English have all been such languages in their time. But the time will not last for ever. To be a prestige language, its native speakers-or the written records they have left-must somehow impress, and so attract imitators. This impact will depend on the cultural development of the recipients, as well as the merits of the originals. As potential recipients grow in wealth, knowledge and self-confidence, and begin to distinguish themselves, the attraction of a foreign model will shrink. It is unsurprising, therefore, that the charms of what was available in Latin and Greek diminished in the nineteenth century while the speakers of English were taking the world by storm with their own technical innovations and a global empire that left the achievements of the cla.s.sics in the shade. Likewise, the charms of French and even the recently developing German faded before the self-confident speakers of English.

What makes a language learnable

There are three different ways in which languages are learnt.

Every native language is learnt by small children almost without effort, from their families and older siblings. For this to happen, there has to be a reasonably stable environment, where most of the community around the child speaks the given language.

If this is absent, so that the surrounding people do not share a common language, a language may still be learnt, but it will be a new formation, distinct from all the languages that the adults knew, a mixture of them reconstructed on first principles. When a group of children learn such a language, a creole comes into being. If the learners are older, adults looking for some common means of de facto communication, the result is a pidgin.

The third possibility is that the language is consciously studied and learnt, either through daily exposure to it, or through formal instruction, perhaps at a school. This process does not depend on the native capacity for language-forming that is active in the minds of small children; in fact, it can be put into effect whatever the age of the learner. In this case, the learner must already speak another language, and use it-explicitly and implicitly-in acquiring the new one.

The first two ways are not in any way dependent on the structure of the languages being learnt. It is generally accepted by linguists that any natural language, of whatever structure, can be learnt by any normal child-regardless of the child's ancestry or its parents' own linguistic background. Some sounds, and some linguistic structures, may take longer to get implanted than others, but everything will come in time. This fact is almost the definition of what it is to be a natural language. As for the genesis of creoles, the matter is controversial, but it appears that they all tend to have a common structure, which emerges naturally as the language is formed. The structures of the contributing languages, from whose parts the learners are constructing the creole, have no effect on the structure of the new language as it comes together.

The third case, which is common when a language is spreading in a territory new to it, may, however, have some interesting consequences for the possible succession of languages. In this case, learners of a language will retain in their minds some background formed by the language or languages they knew before, what is called the substrate. This substrate may impose a constraint on the kind of language that can then be successfully learnt.

Such a constraint may be of two kinds. It may cause the learners to come up with a new version of the language, influenced by their old speech. English spoken in India has lost its characteristic diphthongs: the words gate and boat, [geyt] and [bewt] in England, are p.r.o.nounced [get] and [bot] in India. Likewise, the stress-timed tempo of English has been replaced by a more even, syllable-timed, pace. But more radically, the constraint may act as a major block on the learners ever gaining effective command of the new language. An example of this might be seen in the widespread failure of English Language Teaching (ELT) in j.a.pan for several decades after the Second World War, despite Herculean efforts on all sides to give the next generation competence in this new skill.

The idea that there might be this kind of structural constraint on the adoption of a language is highly controversial; it will be difficult to demonstrate in a particular case because there will always be a mult.i.tude of non-linguistic reasons which might be inhibiting take-up of the language. But the perspective of this book, where language dynamics have been surveyed over centuries, gives some new arguments to show that it may be a real factor limiting the spread of certain languages into certain territories, or rather among certain populations.

Consider, then, the curious retrenchment in the domain of Arabic, which seemed to roll back from its farthest limits in the east and the west, about three centuries after its first spread following the death of Muhammad. (See Chapter 3, 'Arabic-eloquence and equality: The triumph of 'submission", p. 93.) It settled permanently only in the territories that had previously spoken an Afro-Asiatic language, i.e. one that was structurally close to Arabic itself. First of all, Arabic took over the whole of the Aramaic-speaking world, modern Syria and Iraq. Here Arabic could have replaced Aramaic almost word for word. It then overran quickly, and subsequently pervaded, the countries in North Africa, whose vernacular was Egyptian (now known as Coptic) and Berber, although in these cases the spread was far slower, and-at least in the case of Berber-is by no means complete. But in al-Fars (Persia) and in el-Andalus (southern Spain), despite their early reputation as centres of Arabic scholarship, the language was expelled, except in the liturgy of Islam. These are precisely the countries where the substrate language was Indo-European, respectively Persian and Spanish; perhaps it is not so easy for a population speaking an Indo-European language to pick up an Afro-Asiatic one.* Certainly, the next major language spreaders to come through, the Turks, did not pick up Arabic, although they did accept, and even spread into Europe, the religion of Islam. The Turks' language is even less similar structurally to Arabic than Indo-European is. Islam continued to spread in the second millennium, into the Far East; but never again did it carry Arabic with it outside the mosques.

Consider the varying success of Greek in western Asia and Egypt, after Alexander's epochal conquests in 332-323 BC. In principle the administration was everywhere converted from Aramaic to Greek, and there were Greek settlements all over, at least within bigger cities; but Greek only became pervasive in Asia Minor, the great peninsula of Anatolia. (See Chapter 6, 'Kings of Asia: Greek spread through war', pp. 247ff.) In other words, Greek was most successful in the old domain of the Phrygian language in the centre (known from inscriptions to have been closely related to it), and of the Lydian and other Anatolian languages, Indo-European tongues whose structure was also fairly similar to Greek. Greek was unsuccessful, except in planted communities of native speakers, in Syria, Palestine and Mesopotamia (where people spoke Aramaic), in Egypt (where people spoke Aramaic and Egyptian), and in Persia (where people spoke Aramaic and Persian). It is most surprising structurally that Greek did not take root in Persia, since Persian is a fairly similar Indo-European language (and was famously learnt in a year by an ageing Greek Themistocles-see the Plutarch quote on p. 5); but perhaps there are non-linguistic reasons why an alien language should be particularly resented and so resisted in the heartland of what had been an independent and mighty empire for over two centuries.

A third example where language structure seems to have been crucial in the life prospects of a language is almost total absence of Mongolian from central and western Asia and from Europe, since the far-reaching conquests of the Mongols under Tamerlane across Iran in the fourteenth century, and previously under Genghis Khan and his successors in the thirteenth. The Golden Horde which sacked Kiev in 1240 was a Mongol army; and even Babur's dynasty, which dominated India from the sixteenth century, rejoiced in the name 'Mughal', that is to say Mogul or Mongol, although his language, as we have seen, was Turkic. (See Chapter 3, "Third interlude: Turkic and Persian, outriders of Islam', p. 106.) None of the Mongol invasions was soon undone or rolled back: what had happened to all these successful Mongolians?

A crucial feature of the Mongol-led invasions was the fact that they largely recapitulated earlier conquests of Turks (such as Huns and Khazars). Furthermore, they were conducted predominantly with contingents of Turkic-speaking warriors. Now Turkic and Mongolian, even if they are not genetically related, have become highly similar to each other structurally. (See Chapter 4, 'Northern influences', p. 145.) It was very easy, therefore, for a Mongol speaker to pick up Turkic, so to speak-and no doubt often quite literally-on the trot. Outside Mongolia, Mongols tended to be in a minority, and so their language was submerged in the language of their Turkic allies.

A fourth example has been suggested by a student of the Roman takeover of Gaul. We have seen that this led to a rapid, and surprisingly thorough, spread of Latin in place of Gaulish. Brigitte Bauer presents evidence that Latin and Gaulish were in many respects highly similar languages.10 This similarity would have allowed the kind of word-for-word language replacement that we have just posited in the spread of Arabic over Aramaic. By contrast, the structure of British-still perhaps bearing the influence of a pre-Celtic substrate-was rather different, above all being a verb-initial language: verbs come first in the sentence. It would have been harder for Britons to learn to express themselves in Latin than it was for Gauls, and this stubborn fact may be at the root of why France today speaks a Romance language, but Great Britain does not.

Overall, it seems that-despite the received wisdom of linguists over two centuries and more-there may be circ.u.mstances in which the very essence of a language, its structure, can play a role in its viability. Languages, we suggest, are more easily learnt by a new population, and hence spread more easily, when they are structurally similar to the old language of that population. No particular structure is preferred in this process, just similarity of new with old. Otherwise learning a new language is an uphill struggle, perhaps too difficult for many who are already grown.

Vaster than empires

If this book has shown one thing, it is that world languages are not exclusively the creatures of world powers. A language does not grow through the a.s.sertion of power, but through the creation of a larger human community.

Clearly, military or economic might can act as strong inducements to community growth. Rome's irresistible armies, Spain's adelantados and Britain's Royal Navy have all played essential roles in the wider projection of Latin, Spanish and English in their eras. But failures of naked conquest or commercial development to cause linguistic spread have been too frequent to ignore: the political success of Turkic, Mongol and Manchu in mastering northern China did not extend their languages, nor did the Germanic invasions of the Roman empire dislodge or even reduce Latin. The Netherlands' commercial success for two centuries in South-East Asia did nothing for Dutch. Those universal traders, the Phoenicians in the Mediterranean and the Sogdians up and down the Silk Road, may have spread literacy skills along with their merchandise; but they did not convert their customers to wider use of their languages.

It appears that military conquest or economic domination will usually spread a language only if the conquerors come in overwhelming numbers, either through long-term immigration or a collapse of the native population. A less brutal alternative is for the conquerors to enlist the conquered into what is clearly a more technically developed, and potentially enriching, civilisation, as the Romans did in Gaul, and the British in India.

But this is just another aspect of the fundamental point about language spread, namely that it depends on community growth. This is how Sanskrit was able to spread in South-East Asia for a millennium without a conquest, how Quechua took over the domains held by the Inca, and how French was taken up at the eastern end of Europe by the elite of foreign powers who simply wanted to imitate the highest culture they knew. Aramaic too was first spread through a widely scattered community of bilingual scribes capable of originating and interpreting messages written in it, even as communities that spoke it were being uprooted and resettled over the entire a.s.syrian empire. There are more ways, and indeed more effective ways, than violence by which to enlarge a community. A common language is what enables ever more members to partic.i.p.ate in it. As Anderson puts it: 'Much the most important thing about language is its capacity for generating imagined communities, building in effect particular solidarities. After all, imperial languages are still vernaculars, and thus particular vernaculars among many.'11 Each community is differentiated with its own particular approach. Each is given its character by the traditions of its past, and many or most of them are conveyed by narratives and rituals shared in its own language. Contrary to the a.s.sumption of most twentieth-century Western philosophy, a language is never simply 'language'. Each language has its own colour and flavour. In this book, we have glimpsed some of the distinctive traits of the various traditions: Arabic's austere grandeur and egalitarianism; Chinese and Egyptian's unshakeable self-regard; Sanskrit's luxuriating cla.s.sifications and hierarchies; Greek's self-confident innovation leading to self-obsession and pedantry; Latin's civic sense; Spanish rigidity, cupidity and fidelity; French admiration for rationality; and English admiration for business ac.u.men. These manifold qualities can sometimes be seen in the languages' literatures. But they leap out when the languages' histories are told.

It is a paradox that this book, which has told the stories of languages that have so vastly extended their reach, often at the expense of others, is above all a tale of diversity. After all, the kinds of developments recounted here are what have led to the modern crisis of language endangerment, a situation so serious that it is reasonable to believe that, for half the world's languages, their last speakers may already be alive today.12 But there are still over six thousand languages in the world, even if the dozen or so whose tales have been told in this book now account for about two-fifths of the world's speakers.

It is worth asking whether the diversity of consciousness and ident.i.ty that each language represents can or should survive in the modern world. Past industrial and scientific revolutions argue that there is a single, unified path to valid knowledge and industrial organisation, and boast a display of seemingly magical achievements to prove it. Nevertheless, at least until the mid-nineteenth century it was the interplay of research in half a dozen different languages which kept up the pace of intellectual advance. And even today, a penetrating observer of the role of English in the modern world can remark: 'in 500 years' time...if [English] is by then the only language left...it will have been the greatest intellectual disaster that the planet has ever known'.13 But we should not be too overwhelmed by forecasts of impending unity. Half a dozen spiritual revelations have offered themselves as universal truths in the past 2500 years, and most of them are still in contention. Likewise the languages whose histories this book has reviewed have been spreading in increasing circles for twice that period of time. Despite all this rampant compet.i.tion, almost all of them-or their successors-are still in existence at the beginning of the twenty-first century.

A lingua franca is a convenience: for speakers to convey a message across the world, certainly, but also for listeners when one language community appears to have more than its fair share of useful knowledge. But despite the myth of the Tower of Babel, and its vulgar interpretation as a cautionary tale, language diversity is not a liability for the human race. Most people in the world are multilingual, and everyone could be; no one is rigorously excluded from another's language community except through lack of time or effort. Different languages protect and nourish the growth of different cultures, where different pathways of human knowledge can be discovered. They certainly make life richer for those who know more than one of them.

In writing this book, I have consciously been embarking on a new approach within the general field of linguistics. Instead of looking at the current status of the world's major languages, I have taken a historical view. But instead of comparing words in different languages systematically, with a view to reconstructing their past, as a historical linguist usually does, or comparing the overall structures of different languages, like a language typologist, I have considered the evolving status of each language over the centuries of its career. Where any comparison has been attempted, it is the comparison of those careers. This kind of work might be called the study of language dynamics.* It is an approach, previously little explored, to understanding human societies: how language, in all its evolving variety, organises not just the human mind but also the large groups of human minds that const.i.tute themselves into societies, which communicate and interact, as well as think and act.

From this point of view, our focus on large languages has been above all a convenience. All languages have their own histories, but few are well enough doc.u.mented to reveal very much about them. It is the large and famous languages that typically have the most adequate doc.u.mentation. This is where we needed to start, to lay down the outlines of this new field. And this we have done. But ultimately language dynamics must encompa.s.s the history of human language in all its diversity.

kva suryabhavo vasa kva calpavi;aya mati t.i.tirur dustara mohad uupenasmi sagaram here the lineage born of the Sun, and here my weakly endowed mind: would I in folly cross the impa.s.sable ocean in my canoe?

Kalidasa, The Line of Raghu, i.2 * Everywhere the situation was complicated by the simultaneous surge in the use of third parties, mostly black Africans, as slaves; to an extent they, or a mix of them and the indigenous population,

became the representatives of a new minority community, with the immigrants now the majority. But this slave-a.s.sociated minority was never divided by language from the majority community, since they had become a community themselves only by adopting some version of the slave-owners' language.

* A little-noted spin-off of this was the first use of printed books as language tutors, initially of Latin. This in turn led to the development of missionary linguistics, originally as an aid to preaching in exotic places (Ostler: 2004).

This can be expected soon to benefit small language communities, as well as the great languages that have been the subject of this book.

* However, the nature of the home community is changing, partly under the influence of English. Rising levels of female education, more and more including English, and the prevalence of domestic media such as radio and television, mean that the 'mother-tongue' situation for learning a first language in the home will increasingly include English.

* Of the forty-eight most heavily used intercontinental flows of telephone calls in 1994, 46.9 per cent (53 billion minutes) were between English speakers. Another 50.4 percent (57 billion) were between English speakers and countries of other languages (figures from TeleGeography Inc., as cited in Graddol 1997: 37).

* See Chapter 13, p. 532.

* Seventeen per cent (mostly Azeris) are in Iran; 7 per cent are in China (mostly Uyghurs); and 7 per cent (made up of Tatars, Chuvash and Bashkirs, and a variety of tiny groups) are in Russia.

* Thai, too, means 'free': so the ideal can also be found outside the European tradition.

* There is some irony in that this claim is also made by many small indigenous communities for their own languages, which never spread at all.

* And on the other side of the coin, when such a language is successfully picked up, but then effaced by another Indo-European language, the evidence may still be seen in deeply alien features surviving three millennia later. This is what was suggested to account for quirks of British Celtic ( see Chapter 7, 'Run: The impulsive pre-eminence of the Celts', p. 292).

* Or, more explicitly and technically, diachronic sociolinguistics. Another recent example, largely focused on Africa, is Mufwene (2001 ).

NOTES.

Prologue: A Clash of Languages

1. As such, it was recorded in many near-contemporary chronicles. For Motecuhzoma's words in Nahuatl, I here quote from the contemporary encyclopedia of Aztec civilisation, General History of the Affairs of New Spain (xii.16), compiled by Fray Bernardino de Sahagun, and for Cortes's in Spanish from the eyewitness account of his serving soldier, Bernal Diaz del Castilla, True History of the Conquest of New Spain (ch. lx.x.xix).

1 Themistocles' Carpet

1. Sykes (2001, chs 7, 10); Weale et al. (2002). See Chapter 7, 'Against the odds: The advent of English', p. 310.

2. Anderson (1991) is a good guide to the short but fraught history of the concept of the nation, and its transplantation for use all over the world.

3. Sahagun, vi. 13.

4. Karttunen (1990: 291-4).

5. Quotations from three Nahuatl speakers, cited in King (1994: 136-7).

6. Aulus Gellius, Noctes Atticae, xvii.17: Quintus Ennius tria corda habere sese dicebat, quod loqui Graece et Osce et Latine sciret.

2 What It Takes to Be a World Language; or, You Never Can Tell

1. e.g. in Lipiski (1997:46).

2. Firth (1964: 70-1). This is a reissue of works originally published in 1937 and 1930.

3. Whitfield (1999: 36).

3 The Desert Blooms: Language Innovation in the Middle East

1. Tablet II, II. 36-48; text from Lambert (1960: 40); trans. W. G. Lambert in Pritchard (1969: 596-600), slightly modified.

2. II. 70-8; text from Lambert (1960: 148); trans. W. G. Lambert in Pritchard (1969:601).

3. Lipiski (1997: 42-4).

4. The Words of Ahiqar, col. xiv, 208-23; text from Lindenberger (1983: 209), with vowels supplied by Peter T. Daniels; trans. from Pritchard (1969: 430).

5. The evidence is in the Elamite p.r.o.noun system, and some features of noun and verb morphology; Diakonoff (1985: 3); McAlpin (1981). But the attribution is still controversial.