Localisation in South Asia

South Asia has been slow to move to localisation. The elite of the region, those who can afford to buy and use computers, are very fluent in English. This is only 5% of the total population, but in this region of well over a billion people this still adds up to many more people than in the UK and Ireland. The current use of computers in South Asia, including e-Content, is overwhelmingly in English. But look at this web page from the site of ICICI bank, (www.icicicommunities.org/communities /index.asp) India's second largest bank with over 500 branches across India. The whole site uses a lot of red, a colour of prosperity and happiness, much used by brides; it also uses a lot of saffron, an auspicious religious colour.

This particular page offers the facilities to make donations to charity. Giving to charitable donations is an important part of Hindu culture, and other Indian banks, but not all of them, include similar facilities. The previous version of this page was relatively sparse, with Indian images, offering not just charitable donations but also the ability to make pujas. Often when people are involved in financial transactions they will stop at a shrine and undertake a small ritual or puja, seeking blessing and good fortune in this transaction. This site offers simple facilities, but some specialist puja sites enable a full ritual, for example, of offering rice or lighting candles. The important thing to note is that the website has been used to express important aspects of Hindu culture. But why is it in English, and not in an Indian language, when a language is a deeply important component of a culture? Why isn't this website offered in Hindi or some other Indian language?

Some sites do now offer a choice, but historically this has been problematic. There are some 500 distinct languages in South Asia, though the exact total is contestable as the division between dialect and language is not that clear. These languages fall mainly into the three language families of Indo- European across Pakistan, North India and the foothills of the Himalayas, through to Bangladesh; Dravidian in South India, and Tibeto-Burmese in the Himalayas. The languages of Pakistan and Afghanistan are written with extended Arabic writing systems, arising from Islamic influences. All the other languages, as well as those across South-East Asia, are written with alphabetic writing systems that came from a common ancestry, Brahmi - some 15 distinct writing systems in all. We show some examples below, taken from banner lines on web pages.

As soon as PCs became available, and it became easy to create new Roman fonts for PCs, the facilities were used to create fonts for South Asian scripts. But it was not that easy. Each alphabet has some 50 or more distinct letters. These are often combined in writing as 'conjunct' characters composed of two or more letters.

Each writing system has many hundreds of distinct characters composed as conjuncts from the underlying alphabet, and this means that there are many hundreds of distinct characters that could be used for printing and display. They could not fit into a single font table, particularly when this local language character set was required to exist alongside the roman alphabet. Compromises had to be made, and each font creator made their own different compromises, aiming to sell fonts on the basis of their visual appearance. Font creators gave little regard to the actual internal encoding for the characters, and often these encodings arose as an accidental byproduct of how the keyboard is laid out. The result is that the needs of word processing and desktop publishing are met, more or less, but data cannot be shared across the Internet unless all parties have the same font from the same supplier.

This is bad for localisation in South Asia, with poor quality desktop publishing, and it makes data sharing nearly impossible. The resolution to quality came with the realisation that computers could be smart enough to work out what compounds were needed to present writing to people, and all that was needed to represent any Brahmi writing system was the base alphabet of around 50 letters. A rendering engine would then combine letters as needed, even moving letters backwards or forwards if needed. This is how the original Indian ISCII standard, developed during the 1980s, worked and this is how Unicode works.

The National Government of India has long mandated that all government information should be delivered to them in the official languages of India, making available funds to support this, and this has fuelled some developments. But it is not sure how much they actually use this information, and what the real need is. An accounting package working in Hindi failed, and there are very few websites with information in local languages. What has sustained developments is not the commercial market, but rather it has been the passionate enthusiasm of many people believing that this must come about. In particular, people believe that in order to make the benefits of IT available to everybody, IT must spread beyond the 5% competent in English and must work in local languages.

Will this gradual development of local language support slowly generate the market for local software? India has a vibrant software industry, but only about 20% of its software development effort is focused on internal markets - the rest is focused on external markets, often in local branches of major US software companies. This is contrasted with China, where some 80% of their software development effort is focused internally. Could it be that India's very competence in English has inhibited the development of its own local language markets? Nevertheless, we are very optimistic about the development of these local language markets, with a lot of the needs met by internal software development, but also a considerable amount of imported software localised, at least to the most important of South Asian languages.

Adapted from Omniglot, 2004

chinese translation | japanese translation | korean translation