Gujarati vaniyan, 'merchants', Hindi dubhaiya, 'bilingual' (Yule and Burnell 1986 [1903], s.vv.).

The Mughals had brought Persian to India in the sixteenth century as their language of culture, although their ordinary sipahi ('sepoy') spoke Turkic. There is something strangely a.n.a.logous to the Norman conquest of England here, with Persian in the role of French, and Delhi's vernacular, developing into 'Urdu' under Persian influence, in the role of English. In this sense Urdu, literally 'language of the camp', was the distinctive linguistic creation of the Mughals in India. And it was this, not English, which was to become the major language of the British Indian Army. (See Chapter 5, 'Sanskrit no longer alone', p. 222.) Arabic 'educator, composer' (Yule and Burnell 1986 [1903], s.v.).

* Comparing this with the role of missions in the spread of Spanish points up another irony. For as noted in Chapter 10 ("The Church's solution: The lenguas generales', p. 364), the Spanish missions had served to r.e.t.a.r.d the spread of Spanish, while the state was inclined to encourage it. In Brazil, something similar had occurred (see Chapter 11, 'Portuguese pioneers', p. 392). But in British India, the effects of Church and state-or state monopoly-were the reverse of this.

* This was the very period when British academic studies of India's history were making giant strides: between 1835 and 1837 James Prinsep, a.s.say Master at the mint, and secretary of the Asiatic

Society of Bengal, succeeded in deciphering the Brahmi writing of the emperor Asoka's third-century BC inscriptions, and so unlocked the central story of the Maurya dynasty. (See Chapter 5, "The character of Sanskrit', p. 188.) James's brother Henry Thoby, then Chief Secretary to the government, had spoken out eloquently against Macaulay's minute, possibly even leaking it and so providing the basis for a pet.i.tion from eight thousand Muslims and another from Hindus. James, in an editorial in the Asiatic Society's Journal, condemned 'a measure which has in the face of all India withdrawn the countenance of the Government from the learned natives of the country, and p.r.o.nounced a verdict of condemnation and abandonment on its literature' (Allen 2002: 166-7).

* Motto: 'A la.s.s and a lakh a day' (Dalrymple 2002: 33).

* English law, especially as applied in Australia, has a revealing quasi-synonym for this: terra nullius, literally 'land belonging to n.o.body'.

* The company had attempted early on (1612-22) to set up agencies for spice trading at Patani (in Halmahera, the far east of Indonesia) and Ayutthaya, then capital of Siam, and in 1669 for tin at Kedah in the Malay peninsula, but they had always been expelled by the Dutch.

* Even today, location in the UK provides the best medial point from which to understand speakers of English from all over the world: US, South African, Caribbean, Indian, Singaporean and Australian varieties are all frequently heard on the British media, together with a range of UK regional dialects (notably Scots, Ulster, Newcastle, Liverpool, Yorkshire, Birmingham and c.o.c.kney); all are a.s.sumed to be intelligible to a British audience. The USA, by contrast, has for over thirty years already applied dubbing or subt.i.tles to films in the English of Australia.

* It is difficult to attribute this directly either to British or US influence; English was already widely used as a (then neutral) working language of the European Community before UK accession in 1971. But British English remains the majority option when English is taught in Europe.

* The 42 million Continentals capable of taking part in an English conversation in 1950 grew to 60 million (18 per cent) over the thirty years to 1980; the figure had reached 80 million (21 per cent) by 1990 and 105 million (31 per cent) by 2000. Taking account of differing competence at different ages-in 1994, 10 per cent of the over-fifty-fives knew some English, but 55 per cent of those between fifteen and twenty-four-Graddol expects the numbers of English-speaking Continentals to peak around 190 million in 2030.

An allusion to Matthew Arnold's memorable remark, in the preface to Literature and Dogma, that 'Culture is to know the best that has been said and thought in the world.' But we are now less committed than Arnold (or Macaulay) to the view that one language can offer privileged access to the whole sweep of human culture.

* Phoenician and Hebrew, though neither achieved great expansion, and both were as languages highly alike, are cla.s.sic cases of language communities on opposite sides of this divide. As for languages such as Akkadian or Aramaic, Nahuatl or Quechua, we know too little about their contemporary societies to place them in this framework.

PART IV.

LANGUAGES TODAY AND TOMORROW.

Ohe, iam satis est, ohe, libelle,

iam peruenimus usque ad umbilicos.

Tu procedere adhuc et ire quaeris,

nec summa potes in schida tenari,

sic tamquam tibi res peracta non sit,

quae prima quoque pagina peracta est.

lam lector queriturque deficitque,

iam librarius hoc et ipse dicit

Ohe, iam satis est, ohe, libelle.

Whoa there, that's enough, whoa there, my book.

Now we've the reached the endpapers.

You want to keep going on and on,

And can't be stopped on the last page,

As if your subject was not exhausted

As it actually was on the first.

The reader is complaining and flagging,

even the publisher is saying:

'Whoa there, that's enough, whoa there, my book.'

Martial, Epigrams, iv.89 (December AD 88)1

13.

The Current Top Twenty

The simplest, biological, criterion for success in a language community is the number of users the language has. In setting the boundaries for such a community the linguist's main guideline is 'mutual intelligibility': the community is, after all, the set of people who can understand one another using the language.

There are many difficulties with this definition. There are practical difficulties, having to do with the impossibility of actually testing whether populations can understand one another, one to one. How much understanding counts as knowing the language? And what if people typically know the language of their neighbours, and so can understand them even when they are speaking a different language? This is a common situation in Aboriginal Australia, but also in many other multilingual parts of the world. Then there are political difficulties, having to do with people's desired or imagined membership of one community rather than another, and the tendency of census data to confuse members of an ethnic group with speakers of its traditional language. And there are, of course, many theoretical difficulties. Importantly, how many languages should be counted when they fade off at the edges into the next language, something they often do. Sometimes speakers of A can talk to B, and B to C, but A can't talk to C. This is a common situation in the northern plain of Pakistan and India, extending up into Nepal, Panjabi gradually merging into Hindi and then Nepali. Sometimes speakers of A can understand B, but not vice versa, as in the notorious case of Portuguese and Spanish: intelligibility is not always mutual.

A further difficulty comes when the languages are considered historically. Mutual intelligibility has no doubt always been a.s.sured in each generation as between parent and child, but this is not enough to guarantee that the language has stayed the same down the centuries. We can't easily understand what was written in English before the sixteenth century, and if we could hear their speech, we should probably have difficulty with our ancestors in the eighteenth. In fact, languages, even those spoken in the most standardised and widespread communities, almost always change. Should this have an impact on our a.s.sessment of language ident.i.ty, and so of the success of a language over time?1 Consider, for example, the case of Latin. Should this language be considered dead, a n.o.ble tradition sadly ended, because it has no native speakers whose words are close to what we find in the texts of the Roman empire? Or should it rather be considered to have gone to language heaven? Its texts from every period are still read, and its modern forms, collectively called the Romance languages-Spanish, Portuguese, French, Italian, Romanian, Catalan, Occitan and many more-are spoken worldwide, with a total speaker population of over 660 million, making it up to the present the second-most successful language in the world (after Mandarin Chinese). O death, where is thy sting?

Still, it is possible to construct the league table of the world's most widespread languages as they are spoken today, even if it is necessary to make a few arbitrary decisions to do so. The table, once revealed, gives useful hints on the factors behind a large body of people coming to speak the same language. It is also a useful corrective to the linguistic bias that tends to be created by our usual reliance on Eurocentric media.

These figures2 are based on use of the languages as first and second languages, i.e. not just native speakers but also people who have acquired the language for some other purpose and use it actively. Such 'secondary' speakers are clearly part of the language's community. But we must be cautious about the numbers and hence the detailed ranking. The figures here are based ultimately on census returns, which may be subject to distortions with political intent. And English particularly has a large tail of 'foreign language' learners who are quite competent in it and use it frequently, even if it plays no official role in their countries, and may be unrecorded in census figures.3 However, the ident.i.ty of the mega-languages is in practice uncontroversial.

The size distribution of the world's languages is a lesson in itself. Adding together the native-speaker communities of these top twenty languages, we already have 57 per cent of the world's population. Indeed, the top twelve alone account for 50 per cent of the world, hinting at how tiny the populations of most of the other six and a half thousand languages still spoken must be.

In the world's top twenty, all the languages have their origins in the south or east of Asia, or in Europe. There is not one from the Americas, from Oceania or (most surprisingly) from Africa.* But quite naturally, and conversely, these absent areas are precisely where the world's remaining linguistic diversity is concentrated.

The languages can be divided into two sets: those that have grown 'organically' and those that have been put together through processes of 'merger and acquisition'. Organic growth is princ.i.p.ally through population increase in the area of origin, but it can also include encroachment on neighbouring areas. Merger and acquisition spreads a language to discontinuous areas of the world, princ.i.p.ally through seaborne invasion and settlement. All the languages that have spread in this latter way, English, Spanish, Portuguese and French, had their origins in western Europe, and indeed are daughter languages of Latin, or profoundly influenced by it. Although the other three European languages in the list, Russian, German and Italian, are not known in recent history for their a.s.sociated governments' attachment to peaceful methods of expanding their domains, their linguistic growth has been in practice predominantly organic. It is worth noting, as an early antidote to any militaristic presumptions about the causes of language growth, that outside the activities of the European colonists in the second half of the second millennium AD, very little of the growth of these giant languages in the top twenty can be set down to imperial aggression.*

What does account for their growth, then? It is noticeable that a great many of the languages (nine out of twenty) are spoken in civilisations sustained by rice as a staple crop (Bengali, j.a.panese, Korean, Wu and Yue Chinese, Javanese, Tamil, Marathi, Vietnamese). Evidently, rice is capable of supporting dense and extensive populations, and its cultivation, through controlled flooding, requires a high level of organisation. Other languages which are not predominantly in the rice area are spoken in neighbouring areas that have a.s.sumed political control of the rice areas (Mandarin Chinese, and Hindi and Urdu, which are linguistically in a dialect continuum if they are distinct at all). It is also inescapable that, outside the European languages, the list is predominantly made up of languages of the two cultural giants of Asia, China and India.

Looking farther down the list (to the top fifty), many of the same patterns obtain: more variants of Chinese (Jinyu, Xiang, Hakka, Min, Gan), more Indian minority languages (Gujarati, Kannada, Malayalam, Oriya, Panjabi, Bhojpuri, Awadhi, Sindhi), more rice economies (Burmese, Sundanese (of western Java), Thai), more large European languages, grown organically (Polish, Serbo-Croat), despite (in one case) a colonial past (Dutch).

Politically, it is noteworthy that almost all these languages have been under a centralised authority for at least a millennium: large-scale languages do not flourish in areas of small-scale political units, although curiously the languages that have grown organically in western Europe, Italian and German, are exceptions to this. Evidently, the longer-term history of Italy gives something of an explanation: there had been political unity until the breakdown of the western Roman empire in the fifth century AD; the political union achieved in the last two centuries, the basis for its apparent linguistic unity today, harks back to that golden era. German is likewise an artefact of the politics of the last two centuries; but the fact that the speakers of the various German dialects have stayed close enough over the last two millennia to accept a common literary standard is surprising and impressive, for there is little overall political unity at earlier stages of the language community's history. (See Chapter 11, 'Curiously ineffective-German ambitions', p. 446.) Another question concerns the choice of the language that spreads out in these favourable environments. Is there a criterion that predicts which language in a group will spread out to eclipse its neighbours? In a centralised kingdom, this is often a matter of policy, conscious or unconscious: unsurprisingly, the standard chosen for promotion is usually the variety used in the national capital. Hence Mandarin Chinese is historically the form of the language closely a.s.sociated with the city of Beijing,4 as j.a.panese is with Tokyo.5 In Middle English, the dialect that came to predominate and so set the standard was that of London.6 Among the various middle Indian dialects, Hindi/Urdu was characteristic of the Delhi region.7 Russian is by origin the Moscow variant of eastern Slavonic;8 Vietnamese is based on the region of Hanoi;9 and French is the Romance speech of Paris.10 Sometimes, the national capital has moved: so standard Spanish derives from the speech of Toledo, capital of the kingdom of Castile in the mid-thirteenth century;11 and Korean is believed to have originated in the region of Silla in the south of the Korean peninsula, which was dominant in the seventh to tenth centuries.12 Grossly, then, one could claim that, in the political economy of languages, it pays to be the dialect of a city that becomes a national capital; it pays to be in a tropical plain, especially if it grows rice; and above all it pays to be in East or South Asia. But all these criteria have exceptions: indeed, English started out with none of these advantages. As in business, it is evident that merger and acquisition can outpace organic growth.

A disadvantage in basing our observations on the properties of the most widespread languages in the world, as they happen to be today, is that the top twenty list gives no intrinsic sense of the dynamics of language growth: which, for example, are the new entrants, and which languages are moving up, which down? How would the list have looked a century ago, and how will it look a hundred years hence?

To answer these questions we need to combine demographics with some sense of language prestige as it waxes and wanes.

The demographics are a matter of common knowledge: the Asian languages in the list, whose growth has been organic, will continue to grow, with the exceptions of China (by policy) and j.a.pan (through falling natural fertility). The Asian languages are in the list because of their countries' huge populations, which by their very nature did not develop overnight. Therefore, they must be permanent members of the list, until and unless these populations show some tendency to change their language (for example, to adopt the local national language in place of their own regional speech), or are hit by some immense catastrophe, so frighteningly huge that it would have to be comparable to the epidemics that devastated the Americas after the advent of the Europeans from the sixteenth century.13 Some jockeying and rebalancing among languages in China and the Indian subcontinent is likely in the next fifty years: the fertility rate per woman in China in the period 1995-2000 has been 1.8, in j.a.pan 1.4, whereas in India and Bangladesh it has been 3.1, in Pakistan 5.0. On these trends, India's overall population is projected to overtake China's by 2050; in the same period, Pakistan (majority language Panjabi, spoken by perhaps a third of the population) is set to become the third most populous country in the world (overtaking the USA), but Bangladesh (speaking Bengali) will just hold its position (as eighth most populous). Applying the population growth percentages to these languages, the main effect should be a jump in the speaker numbers for Urdu and Panjabi, while the Indian regional languages Telugu, Marathi and Tamil may overtake the regional forms of Chinese, Wu and Yue. Mandarin Chinese is so far ahead (with three speakers for every one of English, Hindi-Urdu or Spanish today) that it will still be by far the most widespread language in the world, though perhaps in fifty years only double the size of its nearest rival.