Alphabet soup

Different romanised spellings of place names are a common sight throughout Thailand. Sign makers in Bangkok's Thong Lo and Chit Lom districts can be particularly creative.

Assistant Professor Wirote Aroonmanakun has just finished updating the Thai romanisation programme.

Rocking back in his chair, Wirote asks how a non-native speaker might be expected to pronounce the word "mountain" had he never heard it before. He may have a point. The connection between English spelling and pronunciation is famously tenuous: the result of years of borrowing from other languages, ham-fisted Anglicisations, simple misunderstandings, pronunciation fads, smudged transcripts and (possibly) drunken lexicographers.

You see, in comparison to English, would-be Thai speakers would appear to have it easy. A largely phonetic language, anyone who's made the effort to learn Thai's 44 consonants and 32 vowels (plus those pesky tones) could hazard a guess at the pronunciation of most words. Sure, there are certain tricky factors to consider - for one thing, it would help to recognise whether a certain word has come to Thai from Pali or Sanskrit - but, overall, it's a pretty user-friendly language.

So why is it that changing Thai into romanised spelling seems to be such a mess? If you were to wander around my neighborhood in Bangkok (which is really not advisable, unless you have a thing for vicious dogs and traffic jams), you would see at least three different spellings of its main street: Thong Lo, Thong-lor, Thonglaw. Similarly, a visitor to the North might be forgiven for writing home to ask for more money to be sent to Chiang Mai, Chiangmai, or even Chiengmai.

Alphabet soup

The Thai romanisation programme is a reliable resource point for standardised spelling.

But need it really be such a clutter? Isn't there a "right way" to spell Thai words so a non-Thai speaker can say them correctly? Well, yes and no.

Here, it's worth delving into the murky waters of linguistics and looking at the subtle, but important, differences between transliteration, transcription and romanisation.

Transliteration is primarily concerned with writing a word with a different alphabet, e.g., representing Thai words with Roman letters. Not a bad idea, but it should be pretty obvious that the alphabet we use in English (among other languages) is rather ill-equipped to match many of the sounds in Thai. How exactly would you write a falling tone in the Roman alphabet, for example?

Transcription, on the other hand, is more concerned with representing sounds. The most common device used to do this is the International Phonetic Alphabet (IPA), which contains enough phonetic symbols to cover most sounds. Unfortunately - and despite the name - the IPA is a motley collection of squiggles, dashes and dots unrecognisable to all but the most dedicated of dictionary buffs and confirmed linguists.

And so, if you haven't already torn your hair out, we come to romanisation, a kind of hybrid system that tries to find a middle ground between transcription and transliteration, by using the Roman alphabet to try and represent Thai sounds.

Unfortunately, the Roman alphabet's not really the best tool to be working with. Not if you consider how many languages actually use the thing. The chances of a Cockney and a Parisian pronouncing the word "water" the same way, for example, are practically nil. As Sumner Spalding said in his 1977 paper Romanization Revisited, "the phonetic values of the letters of the Roman alphabet have about as much firmness as sponge rubber and are as riddled with ambiguities as a Swiss cheese is with holes."

Adding tone markers to romanised words may seem like the simple answer, but isn't necessarily going to plug Spalding's holes and get around the basic incompatibility of the two systems (the Roman and Thai alphabets). Several Thai phrasebooks use tone markers with romanisations, but these can only really serve as general guidelines, rather than accurate representations.

It's interesting to note, however, that the Chinese have had a go at doing just this. The Pinyin system - where pin means "spell(ing)" and yin means "sound(s)" -uses the Roman alphabet to transform Standard Mandarin, tones and all. But, alas, it still doesn't quite work: our Cockney and Parisian pals are still going to pronounce words differently unless they learned pinyin pronunciation.

It's official

And so we come back to our friend Wirote, or Assistant Professor Wirote Aroonmanakun, to give him his full name. He spends most of his time tucked away in a corner of Chulalongkorn University's linguistics department trying to make sense of all this linguistic lunacy.

Wirote has just finished fine-tuning the second version of a computer programme that could have considerable impact on how accessible Thai is to non-native speakers. The programme, sponsored by Thailand's Royal Institute, enables the user to type in Thai script and instantly receive a standardised transcription of the words in the Roman alphabet.

The standard it uses, naturally enough, is that of the Royal Institute. They first devised the Royal Thai General System (RTGS) of transcription in 1954 as a way to write Thai place names in the Roman alphabet and a little more than 10 years later, the system was adopted by the government. The RTGS has undergone several revisions since those early days, but it remains the only official method of writing Thai place names.

The RTGS is not foolproof (it ignores tones altogether and fails to distinguish between long and short vowels) but it does provide an official reference point. It even addresses one of the most confusing aspects of Thai writing by telling you where breaks should be between words. This means, for example, that not only should we now know that the traffic is a nightmare on Thong Lo, but that Chiang Mai makes for a nice weekend, and the beaches in Surat Thani aren't as good as they used to be.

"The programme recognises one or two hundred thousand syllables and uses a probabilistic approach to select the appropriate ones," Wirote says. "But it's dependent on probability, making the most likely [combination of letters]. That's why this system is not 100 percent accurate yet." It might not be perfect - Wirote reckons on something like 97 percent compatibility with the RTGS - but it's still pretty good. Especially when you consider the English-to-Thai version he is working on stalled at around 60 percent accuracy.

"In Thai," he says, "pronunciation is relatively easy to predict, because we have rigid spelling rules. But in English…there are so many exceptions."

Wirote's exasperation brings to mind an anecdote often associated with Irish playwright George Bernard Shaw, a champion of English spelling reform. Legend has it that Shaw gave up on the notion of simplified spelling when an enthusiastic supporter suggested "ghoti" to be a more serviceable representation of the word "fish" (taking the "gh" from cough, the "o" from women and the "ti" from station). Shaw is understood to have given up on the whole idea there and then, and instead backed proposals for a whole new alphabet.

He may have had a point.

To download the romanisation programme, visit www.arts.chula.ac.th/%7Eling/tts/ . For more information about the Royal Institute, visit www.royin.go.th .