Comparing writing systems

New writing systems for old and new languages

Designing a spelling system from scratch - and a language from scratch

If you were designing a spelling system from scratch, what would you do?

What would you choose?

Some people have the opportunity to design writing systems for illiterate societies, or, even better fun, they decide they would like to design a language and its writing system both de nova.

What do they do? What has been learnt from the lessons of history? What mistakes do they manage to avoid? What do they choose from all the options that are so varied and available?

Writing systems invented within the last fifty years show the fruits of experience and experiment. They also show how practical issues affect ideals of theoretical and logical perfection.

Language Planning is a rapidly developing as an international discipline. The design of writing systems is becoming an important part. A fifth of the world's languages have writing systems that were only invented in the last forty years or so, mostly by Americans, Britons and Germans. They are almost without exception alphabetic. Since they do not have to cope with the weight of tradition and outmoded speech forms, sound/symbol relationships can be regular, although this may be at the cost of homographs for words that are pronounced the same, and changed visual appearance of related morphemes that change their pronunciation.

European missionaries first set out to invent scripts for illiterate tribes so that they could translate the Bible and teach the people to read it. They worked on what seemed natural to them - simply find out the sounds of the language and then match letters to each sound. This seemed to work well enough - and many of their writing systems still operate in Oceania, Africa and elsewhere.

J G Paton's story, written by his wife in 1891, gives a fascinating account of his amateur approach in Vienatu, in what were then the New Hebrides.

J G Paton ran a competition for chiefs to teach their peoples to read, with prizes of red cotton shirts.
The former cannibal chief who won the competition used tribal imagery as mnemonics to help his people to remember the alphabet - F was a man with two clubs, for example.

His tribe apparently liked this approach, for they learned quickly and well.

Translations were made most successfully and easily for languages that were were not heavily inflected, irregular or had phonemes that were novel to the translators, which they might not recognise properly. For example, Hawaiian has only thirteen phonemes including an unfamiliar glottal stop - English-speaking missionaries devised a twelve letter alphabet but omitted the glottal stop.

A language that is not written down until late in its development may become simpler and more regular over time rather than more irregular. Some African languages are still smoothing out complex and complicated systems of syntax and changes of word shape, and O'Halloran (1980) hoped that scholars would not tie these languages down in print to these linguistic immaturities. The English language lost almost all its inflections while it was an illiterate vernacular immediately following the Norman Conquest - but perhaps it could have had a better spelling if it had been left out in the cold a little longer.

Today modern linguists can ensure that essential phonemic distinctions in a language are represented, even when they are different from those in other languages. Experts are now also more able to recognise when phonemic distinctions may be better overlooked, to keep things simple when perfectionism could be a pedantic nuisance. Approaches differ as to whether to register tone markings or other suprasegmental distinctions. Grimes & Gordon (1980) note however that if the designers of an orthography leave them out, the actual users of the writing system are unlikely to invent their own, even if they are really necessary.

New writing systems are usually designed for the roman keyboard, for practical reasons, even though various forms of diacritics, letter combinations, or redesignations of letters may be necessary. When scripts based on the International Phonetic Alphabet have been designed for some African languages whose speech sounds seem far removed from the original Latinate of the roman alphabet, the more esoteric symbols may be discarded because they impede mass communication.

As new writing systems settle in, users tend to start representing morphemes as lexical units in print, even when the phonemes have changed (e.g. nation-national) and they also start to make graphical distinctions between words that sound the same, if spelling them the same is likely to cause confusion (as it can with English letter and letter).

When local readers are familiar with neighboring prestige orthographies, they may want some compatibility with them in any new writing system for their own language, partly because they are familiar, as well as to aid intercommunication and multilingual learning (e.g. African tribes wanting some Frenchification for their own new writing systems.) Previously illiterate people often respect and follow traditions of teaching reading that are imported, regardless of the fit to their own orthography. These influences can introduce irregularities, or reduce the capacity of the new writing system to represent its language adequately. The Otomi people of Central Mexico rejected diacritics in their customised script designed for them because these were not also found in Spanish. A compromise had to be worked out that could still represent essential distinctions while letting noncritical ones lapse. Tone marks have been rejected by users of some Ghanaian languages, because prestige English has none, even though this means that some sentences may have to be read twice or more in order to guess at the words with the help of their context.

A paradoxical example is the Yao people of northern Thailand. They maintain their traditional writing system and literature in Chinese characters which are given the pronunciation of Cantonese, a language which they do not know, but which comes out as a phonemic equivalent of their own language. That is, the Chinese characters bear information about the pronunciation of the syllables, not about the meaning of the words. These, plus a few symbols of their own, are preferred to either a Thai-based system or romanisation, although either of these would be easier to learn and more suited to their own language.

O'Halloran (1980, 1982) was commissioned to design a script for the Mandinka language in Gambia in 1947. He supposed it would do well enough to use the Africa Script, or Westermann Script, a variant of the International Phonetic Alphabet designed by German and English scholars for African languages. But he found that Gambian readers always mispronounced baio and most other vowel combinations - but they read bayo with no trouble. He found that he was turning the alphabet into a syllabary, and even agreeing to a few inconsistencies that seemed to suit the Mandinka language better. He also found that children learned to read this syllabic language 'at great speed'. 'In the Gambia we set a period of two months for the attainment of complete fluency in reading. Hardly any children failed to achieve it.' He attributed this to the combination of an open-syllable language, and the modified orthography that took note of this. O'Halloran also described the Mende orthography in East Africa, which the Mende developed for themselves with only three vowels, although Western linguists would want to make it seven. O'Halloran thought that the Mende were on the right track by economising.

If English spelling had been invented only this century, it would be very different from what we have! Just imagine. The recently invented writing systems show how much improvement might have been made - but also the continuing importance of social as well as linguistic factors in the design of spelling.

Sales appeal is necessary as well as user appeal.


Writing systems for invented languages

More than 700 artificial languages have been invented, each with the hope of creating a universal language. When the languages themselves are invented, the writing systems invented for them can be theoretically perfect because the language itself will be simple. Invented languages have no historical accretions and residues, no random inputs from other linguistic systems, no complex vowels or consonants. Most invented orthographies are based on alphabetic rather than logographic or syllabic principles. They have pure 'morphophonemic' sound/symbol correspondence since the morphemes (units of meaning) are stable and the inflections are regular.

Some invented written languages are extremely ingenious, such as that of the classical scholar Lepsius (1863). The most well known is Esperanto which was seriously considered and experimented with for the League of Nations in 1922 before the advent of simultaneous translation techniques - a secretariat report commented how easily it could be mastered sufficiently for communication between twenty-eight countries. It has Middle-European grammatical forms, but its only orthographic weakness appears to be the circumflexed consonants.

Recently there has been the American-created Interlingua, and the French Uropi.
The script of Tolkien's fictional Middle Earth has a cult following which demonstrates how fascinating leximanufacture can be. And how fascinating learning an ideal orthography for English might be

Esperanto English'

Hamlet in Esperanto spelling

 Tu bi or not tu bi; dagt iz di kuestjn:
Uetho tiz nouble in di majnd tu sofo
do slingz and arouz ov autrejgs fortjun
or tu tejk armz agejnst a si ov troblz
And baj opouzing end dem? To daj; tu slip;
Nou mor; and baj a slip tu sej jui end
Di hart-ejk and di tauzand natjuural soks
Dat fles iz er tu, tiz a knsjumejsn
divaautli tu bi uist.

Stuart Campbell (1982) translated this Hamlet soliloquoy into Esperanto spelling, with the aims of showing that English in both language and spelling has basic defects which make it unsuitable as an international language, and also to show how Continentals might prefer to spell English. His spelling, though odd to the English, would look familiar to many Europeans. Esperanto diacritics have been omitted in my version.