Babel, page 22
Bengali as a beauty
As for the Bengali abugida, its basic workings are of an elegance that borders on the boring – at first sight at least. Once you’ve mastered its thirty-five consonant characters and eleven vowel signs, you just start clicking the vowels on to the consonants. Take the character called na, which in its barest form looks like this: When it appears in a word, it’s usually pronounced /n/. Next, take the signs and , which stand for the sounds /a/, /u/ and /e/ (as in bar, book and bay). Tacking the signs on to the character gives us for /na/, for /nu/ and for /ne/. Note that in , the vowel sign is positioned to the right of the consonant, as we would expect in a script read from left to right, but the vowel is placed underneath it in and to its left in – indeed, some vowel signs are placed around the consonant. This corroborates the idea that the vowel signs are not to be considered as separate characters. Each consonant-plus-vowel combination has to be read as a whole, as one compound character. Otherwise, it would be tempting to pronounce as /en/ rather than the correct /ne/.
But this apparent consistency starts to crumble on closer inspection, because Bengali has a few intricacies up its sleeve, some useful, others less so. The best of them is the fact that the consonant characters have what you might call an ‘inbuilt vowel sound’. When I said just now that is pronounced /n/, I hedged the statement with ‘usually’, and that was because when appears not covered with a vowel but in the nude, so to speak, it is actually pronounced /no/ (as in not without the t). The greeting NOMOSKĀR, for instance, is written as with the bare as its first sign. That’s a neat trick. Syllables with /o/ are very frequent in Bengali, so the default option saves a lot of time and ink.
But what to do when the sound you want to write is actually just /n/, not /n/ plus a vowel? There’s a solution for that: you add something that looks like a grave accent (as in learnèd) to the bottom right of the consonant: The sign goes under several names, among them the evocative term ‘vowel killer’. Killing vowels is an excellent solution to the problem just mentioned, but here we encounter the first failing in the Bengali system: in practice, the sign is often omitted, thereby creating ambiguity.
There are other shortcomings too. The pronunciation of the inbuilt vowel is not as consistent as I made it sound just now. It’s usually the regulation /o/ sound, but in quite a few cases it’s /ō/ (as in know), even though there is a separate vowel sign for this sound. And as we just saw, it may even be silent, namely when the vowel killer is omitted. So whenever you see a consonant such as in the nude, there’s no way of telling whether it represents /no/, /nō/ or /n/. Ah, and by the way, the Roman transliteration of the Bengali inbuilt vowel is not o, as would make sense, but a, because that used to be the pronunciation centuries ago.
The great poet Rabindranath Tagore – who wrote in Bengali and English – with fellow Nobel laureate Albert Einstein, 1930.
Next, in another surprise move, Bengali occasionally uses full characters to write vowels instead of vowel signs tacked on to consonant characters. This occurs when a vowel is not preceded by a consonant to carry it, which is to say at the beginning of a word. The designers of the script could have created a silent character to carry any of the vowel signs, which is exactly how some abugidas work (Limbu for one), but not Bengali. Instead, eleven full vowel characters have been created, in addition to the more frequent eleven vowel signs. And while we might hope for, indeed expect, these vowel characters to be similar in shape to the vowel signs, most aren’t. The sound /u/, to give one example, looks like as a full character, but like when it’s a vowel sign.
Bengali as a behemoth
The designers of the Bengali abugida could have stopped here, and they probably should have. What we’ve discussed so far is all that any abugida needs: a complete, perhaps somewhat overequipped, toolkit for writing consonants and vowels, both in pairs and when occurring separately. But I said it before and I’ll say it again: history has a way of messing up writing systems. While any language is always changing, writers tend to resent this and try to keep it stable on the page. They don’t mind frilly spellings that reflect old pronunciations, underlying grammatical patterns and foreign word origins. They get a kick out of exceptions, complications and subtleties that are beyond the ken of mere mortals. Recall what a member of the French Academy wrote: it’s a good thing if a spelling ‘distingue les gens de Lettres d’avec les ignorans et les simples femmes’.
Bengali, like French, is a case in point. So what bells, whistles and hidden defects does this particular script have? It has two separate signs for the same /i/-sound (as in beat) and another two for /u/ (as in book), but none for /e/ (as in rest, not as in bay) – because, well … history. For the same reason, the /sh/ sound can be represented by three different characters, two of which moonlight as /s/. (Of course, it’s no worse than aspects of English, where the /i/ sound – or to be precise /ī/ – can be spelt as ee, ea, ei, ie, y, ey, e or even i.)
More troubling even than these inconsistent spellings is the issue of the so-called conjunct signs or ligatures, which fuse several consonant characters into one. Authentically Bengali words rarely if ever have two consonants at the beginning of a syllable, but words borrowed from Sanskrit, Persian or English do. The language regulators could have decided to use vowel killers to write these. Take the word SKRU for ‘screw’. They could have ordained that this be spelt with three characters: an s equipped with a vowel killer, a k with a vowel killer, and an r with a u sign attached. Instead, they decided that the correct spelling consists of just one character, a monster ligature fusing s, k and r, with a u sign tacked on. I can’t even reproduce this thing in MS Word, because the software rips it apart into three separate characters plus two vowel killers and a u sign, exactly as I suggested.
Now, if these ligatures were a marginal phenomenon, like the French œ, the Dutch capital IJ or the more frequent Danish-Norwegian Æ, anyone could live with them; I would have discreetly ignored them. But there exist no fewer than 285 different ligatures, and while many follow simple rules, quite a few don’t. Between them, they bring the sum total of Bengali characters to 331, and that’s without counting the vowel signs, some historical characters and a handful of essential diacritics. This means that the ligatures really play havoc on Bengali as a system: it’s complexity bordering on chaos. Imagine what schoolchildren have to go through. Imagine the problems that printers and typewriter manufacturers used to face. Bengali should have done as MS Word and I suggested: use vowel killers to take care of consonant clusters. Impossible? Ugly? In Tamil, they do just that, and don’t go telling the Tamilians there’s something wrong with their language.
The Phoenician connection
While Tamil deals with this cluster issue rather cleverly and while Limbu, as I mentioned earlier, does a more efficient job writing word-initial vowels, the take-home message about India’s scripts is that most of them are similar – not visually, but structurally speaking. Of the big ones, only Arabic and Roman do not fit the pattern, and they were introduced from outside rather than being home-grown. With these two exceptions, all of India’s major scripts are abugidas, and while each one of them has different character shapes and many have this or that individual foible, they all share many features. For this, there are two historical reasons: they’ve all sprung from the same common source and their users have a long tradition of writing Sanskrit, the classical language of South Asia.
The common source, the great-grandparent, so to speak, of all later Indian abugidas, is called Brahmi, which is why all the offspring – dozens upon dozens of them – are collectively referred to as ‘Brahmic’. The family tree would be huge and full of arcane names, so I’ve pruned it to the bare essentials, leaving only the four branches that are represented in this book. If we were to study every individual branch of the whole family tree of Brahmic scripts, we would note that the modern scripts have maintained most of the original characters, and it’s principally their shapes that have morphed through the centuries. Only occasionally were old characters jettisoned or new ones created from scratch. This implies that most of the modern characters in each Brahmic script are historically linked to their counterparts in the other Brahmic scripts. It’s just that all these centuries of gradual divergence have obscured the old links – which is why no longer looks anything like (that’s bhaarat for ‘India’, if you remember).
Dates indicate first known occurrence; older scripts may continue to exist alongside newer ones for a long time.
In the severely pruned family tree above, we see Brahmi perched on top, yet this shouldn’t be taken as an indication that it had no ancestor. Most experts, especially those outside South Asia, trace Brahmi to the Middle East, though they disagree over the details. The most likely scenario seems to be that Brahmi was an improved successor of a slightly earlier, more westerly abugida called Kharosthi, used in and around today’s Pakistan. This in turn was inspired by, but far from a carbon copy of, the Aramaic script. The scenario makes geographical and chronological sense, since the Aramaic script was used at the time by South Asia’s powerful westerly neighbour, the Persians. If the Indian scripts are indeed the offshoots of Aramaic, this implies that they and our Latin scripts share a common origin:
In this family tree, the middle column consists of abjads, the right column of abugidas and the left column of alphabets
How on earth could this one common origin result in systems as dissimilar as the Indian, European and Arabic scripts – in abugidas, alphabets and abjads? The thing is that Phoenician and Aramaic are Semitic languages, just like Arabic and Hebrew, and they all use abjads because they actually prefer them. It sits easily with their structure, in which vowels are not all that important. In a word, Semitic and abjads are a happy marriage.
But when these scripts spread west to Greece and east to South Asia people in those regions soon discovered that an abjad just wouldn’t work for their own tongues, which were all Indo-European, not Semitic*. To make writing work, vowels sounds had to be represented. The Greeks chose a simple solution: they took a couple of old Phoenician consonant characters they had no use for and turned them into vowels, thereby creating the first alphabet. Unsophisticated though the idea was, it has worked remarkably well for nearly three thousand years.
The South Asians tackled the problem differently, perhaps more ingeniously. A rich scholarly tradition of linguistic analysis gave them keen insight into phonology. Like the Greeks, they realised that writing their languages without vowels would make the resulting texts unintelligible. And since their syllables were typically of the ‘one consonant, one vowel’ type, they hit on the novel idea of equipping the consonant characters with vowel signs. Seeing how the /a/ (as in Iraq) was more frequent than others, they chose to make it the default vowel (later to be pronounced /o/ in Bengali). Additionally, their phonological acumen enabled them to substitute a more logical order of the characters for the haphazard one that Middle Eastern and European scripts have always had: they clustered consonants that share the same ‘point of articulation’, the place in the mouth where they’re produced; they also made the vowels a separate group. It’s this rearrangement of the characters that explains why Bengali’s list of consonants begins with ka, kha, ga, gha; and that of the vowels begins with short a, long a (ā or aa), short i, long i. As I mentioned a few pages ago, this new order doesn’t make for a sonorous coinage like alphabet, abjad or abugida; however, the organising principle behind it makes admirable linguistic sense.
For Greek, an abugida system like this wouldn’t have worked as smoothly, because like most members of the Indo-European family (remember Russian’s VZGLYAD), it thinks nothing of multiple consonants at the beginning of a syllable – think of Greek names such as Plato, Strabo and Ptolemy (with non-silent p!). When in the second millennium BCE people of Indo-European stock first reached South Asia, their language, an early form of Sanskrit, was much like that. But mingling with the Dravidian aboriginals of the subcontinent, their language gradually changed. In Dravidian languages, consonant clusters were non-existent, and as a result, the Sanskrit speakers gradually began to avoid them too. When some time around about 500 BCE writing was reintroduced to the subcontinent, the living languages of the day had few or no consonant clusters left. The abugida concept fitted them like a glove.
The script of Lao, Laos’s national language, is one of many in South East Asia that are derived from an old Indian script. Colonial French is still sticking it out, but for how much longer?
Writers and their banal pursuits
If it required an extraordinary leap of the imagination to turn the Aramaic abjad into an abugida, yet another aspect of early Indian writing was more exceptional still. Typically, wherever writing takes root, the scribes and their scriptures carry high prestige. There’s something supernatural about this skill of catching the ephemeral spoken word and fixing it on some surface from which it can be retrieved and made to sound and echo and echo again. In societies with a low degree of literacy – all societies until not so long ago, that is – capable writers were generally not mere craftsmen, but could climb to lofty positions in the spheres of religion and state. In China, the script was a defining feature of civilisation. In Europe, the Roman alphabet became the signature of Western Christianity (which is why the pagan runes were pushed into disuse). In the Muslim world, the Arabic script attained almost divine status.
Not so in South Asia. The elites stubbornly clung to the oral tradition, which had preserved for centuries not only the ancient Hindu Vedas but also more recent works of Buddhism, philosophy, poetry, drama and even science and technology. The underlying reason for their resistance to writing may have been to protect their position: memorising all those venerable texts was part of what justified their privilege. Or it may have been because they didn’t trust the new-fangled medium to be sufficiently durable – a worry that was eventually borne out, as most texts from the era, unless carved in stone, succumbed to South Asia’s hot and humid monsoon climate.
Such writing as was done served commerce and other quotidian pursuits. It briefly played a more exalted role when the Buddhist King Ashoka (mid-third century BCE) had his social and moral precepts published on pillars, boulders and cave walls throughout his extensive empire. But on the whole, South Asian elites would drag their feet for over half a millennium before taking to writing in the mid-second century CE.
This was to have two major consequences for the scripts themselves. Firstly, when the oral literature was finally committed to paper (or rather, to dried leaves), an acute spelling problem made itself felt. The abugidas had been tailor-made to suit the vernaculars, but the whole cultural tradition was in Sanskrit, which still had all those Indo-European consonant clusters. Whereas the daughter languages of Sanskrit had KISĀN for ‘farmer’, TĀRĀ for ‘star’ and TIN for ‘three’, (Old) Sanskrit itself had KRSHĀNA, STR and TRI. If the South Asians had begun to write in Sanskrit right from the moment they put pen to leaf, centuries earlier, their scholars might well have decided against an abugida, developing something more akin to an alphabet instead, just like the far-away Greeks. But by this time, the abugida was well ingrained in the Indian literary mind.
Rather than being replaced, the system was merely adapted to Sanskrit by creating those ligatures that today torment schoolchildren, printers, typewriter manufacturers and software developers. They need to learn them not because they read and write the Sanskrit language itself – they don’t, any more than the average European reads and writes Latin. But the modern languages have adopted a huge number of Sanskrit words, in much the same way that English and other European languages have borrowed words from classical Greek and Latin: in modern Bengali, KRSHĀNA is once again used for ‘farmer’, after centuries of absence from the vocabulary. And just as our spelling has kwiksotikally maintained the expendable but classical characters c, q and x, so nearly all Brahmic scripts have maintained the equally superfluous ligatures.
The other consequence of the belated acceptance of writing by the elites is India’s present multiplicity of scripts. If they had embraced the new invention as soon as it hit the market and applied it to the ancient holy texts, Brahmi (or even its predecessor) would have become a revered standard throughout the region. But they didn’t, and by the time writing was finally used for purposes of religion, scholarship and literature, a tradition of freely designing new script types had already taken hold. After all, writing for a long time was just this thing merchants and humble desk-workers did, and just as human languages are continuously diverging into dialects, so scripts will likewise diversify unless there’s a centripetal force to stop them. That unifying role can be fulfilled by a holy book – the Koran, Bible and Tanakh all prove this – but in terms of prestige, the Hindu scriptures never eclipsed the oral tradition. Alternatively, an empire of long standing can have the same effect of preserving the script as it is, but in South Asia this never happened, as for most of its history the region has been either divided or ruled by elites wielding foreign scripts.
In the absence of a holy book and a solid empire, the script is bound to hit the fan, resulting in numerous varieties scattered all over the place. This happened in several parts of Europe in its pre-empire and pre-Bible period: there were not only several Greek and Roman alphabets, but also Etruscan, Oscan, Iberian, Tartessian, Ogham, Runic and others. The same happened in South Asia, resulting in a whole family of abugidas. And when the Hindu and Buddhist scriptures spread to Southeast Asia, along with a culture that held the oral tradition in high esteem, local writers would soon adapt the script to the needs of their own tongues. The national languages of Burma, Thailand, Laos and Cambodia are written in as many abugidas to this day; many languages of Malaysia, the Philippines, Vietnam (though not Vietnamese) and Indonesia (including Javanese) used to have their own varieties before adopting the Latin alphabet. To the north of South Asia, the Tibetan script is yet another Brahmic offshoot, and even Japanese writing drew some inspiration from India. In the end, all these Brahmic apples rolled away so far from the Brahmi tree that the ancient texts at some point became inaccessible. They wouldn’t be deciphered until the nineteenth century, by outsiders.
