Our transliteration scheme
Our transliteration has been done for easy assimilation by any young person, or non-Tamil speaker who might casually pick up a songbook or song sheet at a Saturday temple satsang, having had very little exposure to any formal language studies of any kind in his or her life.
"Indologists" may prefer to see diacriticals, and Tamil purists will complain that some subtleties of Tamil have been lost. But our experience in the field shows that if we put a diacriticalized version into print, only those who are trained in what the marks mean and how they are pronounced can use it. For the uninitiated, it is "opaque" and is no aid at all. Even those who do take the time to study out and learn a diacritical scheme, unless they are very strict with themselves, invariably "regress" back to their native language habits if they do not take the time to learn the native script. A long "aaahh" with a mark over just gets pronounced as a short "a" over time. Use of capital letters was avoided for similar reasons.
We have opted for fully lower-cased style that preserves explicit vowel forms (long and short) while sacrificing some subtleties in consonants. In actual practice, slight loss of perfection on consonant pronunciation is forgivable but if long and short vowels are not maintained, it sounds quite "terrible." Hopefully "vaakya" will clearly indicate that the first "aa" is long. This style has been carefully working out with several experts and Tamil speakers who are familiar with issues of teaching children. With one single exception, all the roman letter-syllable equivalents we use are already in use by the Tamil diaspora community and well understood. This transliteration has been "field tested." One example: we have given it to an 11-year-old Canadian born Sri Lankan whose first language is English and asking him to read it for the first time in his life. He did remarkably well and did not stumble at all.
Of course improvements can always be made, but if someone wants youth or students to sing more correctly, there are limits to any roman transliteration. Our advice is simple: learn Tamil from a competent teacher. Meanwhile we need to make it easy for the next generation to pick up and "go for it!" and sing the songs.
Sandhi, "Joining" or "Liason" Characters
As many of you know, written Tamil frequently introduces extra characters at the beginning and end of words to make the transition smooth, phonetically. These are then often rendered in the Tamil script, making it difficult to discern the root words. Ancient Tamil used no word breaks at all. A verse or sloka was written as a continuous stream of characters. Beginnings and endings of words were freely modified by truncation of certain vowels and the introduction of sandhi (joining or liason) in the interests--and by very strict rules--of making a smooth vocal transition from the end of one word to the beginning of another. The challenge of this system is that only the very highly trained are able to break the words properly to ascertain the meaning.
For example: "ennai ennukku arivithan" "about me - to me - he introduced" could be written (and is usually sung as) "ennaiyennukkarivitthan" where liason "y" is introduced and suffix "-ku" becomes "kk". But singers are often heard on one repetition to sing the two words together in a single unbroken phrase, then on the next repetition the same singer break apart the two words, enunciating each one separately with a pause in between.
This can be quite frustrating for someone trying to learn the songs and has made the study and learning of Tamil songs for the current generation a daunting challenge. Since our "prime directive" is to understand what we are singing, our response to this challenge has been, in the interests of clarifying meaning, to drop the liason letters in the transliteration. This has been done wherever it does not change pronunciation or force the introduction of new syllables. This aids reading and translation.
The sandhi's are based on natural vocal position of the lips and tongue. The two most common (and natural for the tongue) are "y" and "v" between a word ending in vowel and a second one starting with a vowel. For example, to transition from "i" or "ai" to another vowel, "y" is usually used. "Nallai endru" is sung "nallaiyendru" But if lips are forward where the word ends in "o" then typically the letter "v" will be introduced as in "poh endran" becomes "pohvendran" when sung together. This means simply you may, and often should, freely pronounce the sandhi letters you hear in singing, but don't emphasize them. Tamil is very soft. The "v" in "pohvendran" is very lightly enunciated and not at all like the hard "V" in "Victory." The best way to train oneself is to sing along with a tape or your iPod. Practice while reading and listening to the audio will help you get it right.
Consonants
The challenge with many Tamil consonants is that they have no English equivalent and are not easily represented in plain roman script. Our simple transliteration scheme makes no attempt to distinguish between three different "L" or different "N" sounds (dental, palatal and cerebral). The subtle differences in "th" "d" "thth" and "dd" are virtually impossible to present in simple roman letters. One may be forgiven use of a dental "N" in place of a palatal "N." But if long and short vowels are observed, the Tamil comes out sounding reasonably good, even if all consonants are not enunciated perfectly. The one consonant we felt we need to not sacrifice was the "high L," sometimes represented by the infamous "zh" as in "Tamizh." We have made an innovation to use the form of "l." -- the letter "l" followed by a dot, which requires no diacritical mark, is subtle enough not to be too visually distracting and for those who know Tamil well enough, an indicator for this difficult sound. Others will pronounce a simple "l" which is the best they can do. This was felt better than "zh" which gets wrongly pronounced by the uninitiated as "z" in "zoom". In Mauritius today, we have young men with the name "Pazani" and pronounced as such when their name sake is Lord Murugan. "Pal.ani" might have saved them from the lifelong mistaken pronunciation that has arisen from the use of "Pazhani."
This single Tamil character "k" is pronounced three ways in different contexts. It is a hard "k" in some contexts, a soft "g" in others and even pronounced as a subtle "h"in others. Because of the English speaker's tendency toward strong aspirated "h" we have decide not to us "h" between vowels so as to avoid the strong "ha!" where the Tamil is a very soft "h" between vowels. We opted for "g" as in "Murugan." Of course, many English speakers will also hit the "g" too hard... when it should be pronounced softly and even as "Muruhan". A similar "controversy" arises around "t, th, tt, thth, dth, dd" etc. We have done our best to follow a middle path between accuracy and simplicity within the limitations of the 26 letters of the roman alphabet.
Transliteration Notes Log
The following is a collection of notes, questions to experts and answers on specific issues, out of which we evolved the algorithms for a programatic, automatic transliteration process. We wrote our own software to create the first rough draft of the roman transliteration, based on these rules. Songs were subsequently edited by hand to take care of specific issues.
[this section coming later]
