Wednesday, April 1, 2009

Indic languages : is transliteration the simplest solution ?

NLP researchers at Google and other places have done a great job in transliteration from Roman script to Indic scripts. Now transliteration feature is available in GMail.

http://googleblog.blogspot.com/2009/03/email-in-indian-languages.html


However, there is a simpler solution available.




Just click on on the menubar (or some keyboard shortcut) and change you keyboard layout to India. Your keyboard then starts emitting Unicode (ISCII) symbols for Indian scripts (eg: Devanagari). These layouts have been standardised long back. Only thing missing in the present setup is that Indian characters are not printed on the keyboards. ( Of late, you can find Indian charecters printed on some cell phones though)


The Japanese,Russian, German, French and many other coutries have such keyboards. (Its not a difficult task anyways). Hindi, for example, is the 4th largest spoken language in the world. Yet, why there is no demand for such things ?

I personally believe Devanagri is a better script for Hindi (and Sanskrit and Marathi) as it is very precise. What is the point of writing these languages in the not-so-precise Roman script such that your transliterating software produces the correct Devanagari (or for every word you have to select from drop-down box) ? Why not write directly in Devanagari ? Would n't it be much simpler ?