"romanization, Transliteration, And Transcription"

[DavidHouston]

The Journal of The Royal Institute of Thailand published an article in 2006 entitled "Romanization, Transliteration, and Transcription for the Globalization of the Thai Language." The article is in English. The author reviews the history of the various attempts at rendering Thai script into a format more accessible to foreigners and presents some ideas with which I was not familiar. This article should serve either as fodder for more discussion regarding transliteration or close down some. I suspect that the discussions will continue in this forum and others.

See http://www.royin.go.th/upload/276/FileUpload/758_6484.pdf.

(I tried to search for a discussion of this topic in this forum but could not find the relevant discussion; if the issue has been presented and discussed, I ask the moderator to please remove this topic. Thanks.)

[DurianChips]

Interesting article. After reading it though, I'm really not sure of the purpose of the romanization system described.

The following two paragraphs caught my attention (I apologize for any typos; the highlights are mine):

"This Romanized transliteration system is a means of converting the Thai writing system into a Roman alphabet writing system. Since there are fewer Roman characters than Thai characters, diacritical marks, punctuation marks, and a combination of two Roman characters, or a digraph, are needed to represent one Thai character. The aim of this system is to provide a means of international communication of written messages in a form which permits the automatic transmission and reconstitution of either written script by humans or machines. This system of conversion is intended to provide complete and unambiguous reversibility. It is recognized that the transliterated form resulting from this system does not always provide the correct pronounciation of the original Thai text. However, the system serves as a means of finding the Thai graphisms automatically and thus to allow those with a knowledge of Thai to pronounce the Thai text correctly.

In this system 87 Thai characters, including numerical symbols and traditional symbols can be transliterated into the Roman alphabet and re-transliterated back into Thai characters. For example เตะ "to kick" is transliterated as "eta" where "eta" gives the wrong pronounciation but it can be re-transliterated back to เตะ, either by humans or machines."

I thought that the purpose of most Romanization systems was to give non-Thai speakers a close approximation (not exact of course) to the pronounciation of the Thai word. If it can't do this, it obviously seems to have little value as a teaching tool. Someone could learn the new rules for pronounciation of the Roman characters, but if they are going to go to all that trouble, they might as well just learn the Thai characters.

Is it intended for professional linguists? Then why not just use IPA?

I sure hope it's not intended as a tool for Romanizing the Thai names of people and places, because it seems to me such a system should strive to reflect the Thai pronounciations, and not have absolute transliteration reversibility as the goal.

Who would make use of "a means of international communication of written messages in a form which permits the automatic transmission and reconstitution of either written script by humans or machines"?

[Richard W]

I thought that the purpose of most Romanization systems was to give non-Thai speakers a close approximation (not exact of course) to the pronounciation of the Thai word. If it can't do this, it obviously seems to have little value as a teaching tool.

Neither the appalling ISO 11940:1998, which is the scheme that converts เตะ to eta, nor the RTGS is intended as a teaching tool.

Who would make use of "a means of international communication of written messages in a form which permits the automatic transmission and reconstitution of either written script by humans or machines"?

ISO 11940 is horribly dated in its concepts, and order of vowels and tone marks it prescribes is the opposite of the industry standard. Let me deal with machine (i.e. computer) application first.

The scheme was designed on the assumption that a computer that could not handle Thai could handle the whole of the Roman script, or Latin script as the typographers have chosen to call it. But for Microsoft's decision to render a tone mark plus sara am beautifully, that would be completely false. I suspect it is false anyway - a system that cannot handle Thai is unlikely to be able to handle Vietnamese. It is a fact that it is easier for a modern computer system to handle Thai than the ISO 11940 transliteration of it! (The problem is in the fonts.) It needs the back-up of a method of further transliterating the Roman transliteration to ASCII (or perhaps just to Latin-1). You may think I'm unfair, but it includes such exotic features as accents normally restricted to vowels (grave, circumflex, acute, caron, breve, horn) on consonants, as well as dot below, macron and underline on many consonants. (SAMPA and X-SAMPA, discussed on another thread, are basically schemes for transliterating the IPA to ASCII.)

Automatic reconstitution has no problem so long as you only use the scheme for transliterating the Thai language. (There is even a problem with using it for Pali!) Minority Thai script writing systems start presenting complications, e.g. with their fondness for using underline as a diacritic on Thai letters. To be fair, Microsoft (who're putting a lot of effort into supporting minority languages) hadn't considered this issue either, even though some English-Thai dictionaries use underlining in the phonetic representation of English pronunciation using the Thai script. (I've supplied Microsoft with examples from the dictionaries.) Reconstitution becomes quite difficult when you consider these details, especially if the dot below the middle of a consonant with a vowel is encoded as a dot below rather than as phinthu.

The one place where it might have a role is in human transmission. One application - the only one I've actually seen any evidence of, and then only in Estonia - is to specify the title of a book in Thai. One cannot reasonably expect library staff to process titles in Thai script, Devanagari, Arabic, Cyrillic, Tamil and Bengali, but one might hope to train them to process accented letters.

Reversible transliteration does have a role - try back-converting a Thai name and address in the countryside from the RTGS!

[DavidHouston]

Richard,

Thank you so much for the explanation and for the reference to http://en.wikipedia.org/wiki/Royal_Thai_Ge...f_Transcription (through ISO 11940:1998) and for the referenced downloadable article, "AUTOMATIC ROMANIZATION FOR THAI" (ThaiRomanizeCOCOSDA99). That information should be exhaustive enough for all readers of this forum. What should impress us the most is the fact that scholars have been dealing with the issue of Thai Romanization for many decades; the current computer environment solves some problems but raises others.

Richard makes the point that the library and publishing contexts are important forums for Romanization; this is not just an issue for new learners of the language.

And, the most important lesson of all may be that the easiest way for new students to access the Thai written language is to learn the Thai script itself. As with Darwinian biology, the Thai writing system has evolved over may generations to be the most effective manner of conveying the spoken language in written form.

Good luck.

[dominique355]

I was always amazed by the "official" transliteration of Thai names and words and how wrong they appeared to me.

Take His Majesty the King's name, transliterated into "Bumiphol", while by sound it should rather be "Bumiphon".

Same for the new airport, and I guess we all have numerous and some even hilarious examples of such transliterations.

I found it always quite funny to invite my friends from Europe the a small restaurant named "Mae Porn". You should see the average European lady's eyes when she sees the restaurant's name outside . OK the guys are sometimes disappointed when they are inside.

I just wonder who was behind all these mistakes. And then, the English language obviously serves as a basis for these transliterations. However, the English language is very ill suited for this (although I admit, it's the most commonly spoken foreign language here).

But the explanation given here is frankly (IMO) total nonsense. If "to kick" is transliterated into "eta", while in reality it sound "dte", what does a person who does not speak Thai do with this? And what for does a Thai speaking person need these transliterations?

I would hope they find a way of transliterating Thai words in such a way that people who do not speak or read Thai, can pronounce the Thai word so that Thai people understand him.

[DavidHouston]

I was always amazed by the "official" transliteration of Thai names and words and how wrong they appeared to me.

Take His Majesty the King's name, transliterated into "Bumiphol", while by sound it should rather be "Bumiphon".

Same for the new airport, and I guess we all have numerous and some even hilarious examples of such transliterations.

I found it always quite funny to invite my friends from Europe the a small restaurant named "Mae Porn". You should see the average European lady's eyes when she sees the restaurant's name outside . OK the guys are sometimes disappointed when they are inside.

I just wonder who was behind all these mistakes. And then, the English language obviously serves as a basis for these transliterations. However, the English language is very ill suited for this (although I admit, it's the most commonly spoken foreign language here).

But the explanation given here is frankly (IMO) total nonsense. If "to kick" is transliterated into "eta", while in reality it sound "dte", what does a person who does not speak Thai do with this? And what for does a Thai speaking person need these transliterations?

I would hope they find a way of transliterating Thai words in such a way that people who do not speak or read Thai, can pronounce the Thai word so that Thai people understand him.

A challenge for my European friends: is there anyway a native English speaker can pronounce Hungarian or Czech correctly using only characters found in English? Seems to me that this is not a problem only for the Thai language.

[dominique355]

You are right David. That's why they have invented phonetic language. Seems to be unknown here

[meadish_sweetball]

With 'here', do you mean Thailand or this forum?

If by 'phonetic language' you mean IPA, it is difficult to reproduce it on computers.

Thais who study linguistics will learn IPA. As a pronunciation aid for learning English, they use Thai lettering with some extra markers to approximate English sounds.

The RTGS is particularly unphonetic (aphonetic? non-phonetic?) when it comes to Indic-based words like His Majesty's name or the new airport. The system aims at approximating Sanskrit instead of Thai pronunciation.

[DavidHouston]

With 'here', do you mean Thailand or this forum?

If by 'phonetic language' you mean IPA, it is difficult to reproduce it on computers.

Thais who study linguistics will learn IPA. As a pronunciation aid for learning English, they use Thai lettering with some extra markers to approximate English sounds.

The RTGS is particularly unphonetic (aphonetic? non-phonetic?) when it comes to Indic-based words like His Majesty's name or the new airport. The system aims at approximating Sanskrit instead of Thai pronunciation.

How about the efficacy of Glenn Slayton's keyboard-friendly phonetic system used in the Thai-language.com dictionary? He distinguishes between long and short vowels with doublings and marks tones via a single capital letter at the end of each syllable. The letters and tones are "M" for mid; "H" for high; "L" for low; "R" for rising; and "F" for falling. Seems to me that this is similar to and perhaps inspired by the Chinese Pinyin system where tones are noted by number. In addition, consonants are rendered fairly easily, using, for example, b for บ; bp for ป and ph for พ and ผ.

No system is perfect as so many on this board have pointed out. However, if one takes as guidance one objective: a simple system for new learners to get a sound which approximates Thai in as many respects as possible, Glenn's may be adequate for the purpose. His has little relationship to the Thai spelling, like some systems attempt to do and is useful for speech but not for writing. Alternatively, there is the Mary Hass system on Thai2English, but this is not easily keyboard accessible.

If the objective is purely to have a phonetic approximation of Thai in all its grandure, consider Glenn's contribution. As you suggested, Meadish, perhaps moderators could request that posters use a predefined system to render Thai in a consistent fashion. Many would not but at least you made an attempt. I agree with your suggestion that the transliteration key could be attached for reference. The other benefit is that new learners and readers might be more inquisitive in asking what the correct sound of the Thai word they wish to express is when they post. The phonetic system may be so challenging, however, that the posters will opt to learn real Thai. Now, that would be a significant achievement!

[Sateev]

I was always amazed by the "official" transliteration of Thai names and words and how wrong they appeared to me.

Take His Majesty the King's name, transliterated into "Bumiphol", while by sound it should rather be "Bumiphon".

Same for the new airport, and I guess we all have numerous and some even hilarious examples of such transliterations.

I found it always quite funny to invite my friends from Europe the a small restaurant named "Mae Porn". You should see the average European lady's eyes when she sees the restaurant's name outside . OK the guys are sometimes disappointed when they are inside.

I just wonder who was behind all these mistakes. And then, the English language obviously serves as a basis for these transliterations. However, the English language is very ill suited for this (although I admit, it's the most commonly spoken foreign language here).

...

Often, and certainly in the case of 'Bumiphol', the mistake lies in making a letter-by-letter transliteration. Such a process ignores the rules regarding consonants the change at the end of a syllable. ล, which ordinarily has the 'L' sound, changes to an 'N' at the end of a syllable. By translating each Thai consonant as an English equivalent, the change in sound is missed. That is why you see, variously, Cholburi/Chonburi, etc.

People doing the transliterating probably are under the impression the we, also, change the 'L' to an 'N' under the same rule.

Sateev

[Richard W]

The RTGS is particularly unphonetic (aphonetic? non-phonetic?) when it comes to Indic-based words like His Majesty's name or the new airport.

The RTGS transcription of the airport name is Suwannaphum. The current name comes from a graphic system by dropping the diacritics, though the best documented whole-language version I've seen would have given Suvarrnabhumi after diacritic dropping.

The system aims at approximating Sanskrit instead of Thai pronunciation.

Just like Thai!

One aim of the graphic system is that the same sequence of letters should be transliterated the same, whether the word is embedded in Thai, Pali, Sanskrit or even, I believe, Khmer. In the case of the airport, it's only the -รรณ- sequence that tells you the word is Thai. Sanskrit has -รณ- and Pali has -ณณ-. This is particularly useful when transliterating inscriptions. (Scholars have priority over tourists.)

Secondly, before diacritic stripping, the graphic system is just as phonetic as the Thai script itself. After diacritic stripping, it still gives you a better idea of the tone than the RTGS.

[Richard W]

As you suggested, Meadish, perhaps moderators could request that posters use a predefined system to render Thai in a consistent fashion.

We tried using an extension of the RTGS, but it was found to be too much effort. One problem, of course, was that it required one to truly know how a word was pronounced, and that can be tricky.

One innocent example is เมตร. The RID gives [F]meet; the thai-language.com transcription currently gives [F]meet ('maeht[r]F' in its notation) but the sound-clip gives [H]met; thai2english.com gives [F]met; Nit Tongsopit's dictionary gives [H]met (though length marks often go missing from that dictionary); and so on. As far as we can make out, the actual pronunciation seems to vary between [H]met and [H]meet.

[meadish_sweetball]

Thanks for the correction; my mistake for assuming the RTGS was behind the Suvarnabhumi spelling.

Secondly, before diacritic stripping, the graphic system is just as phonetic as the Thai script itself. After diacritic stripping, it still gives you a better idea of the tone than the RTGS.

Are these diacritics you mention, tone diacritics? The ^ ´ ` etc. of IPA, or other ones?

I guess you are right, but obviously that would require the prior knowledge necessary to deduct the tone from this graphic system in Roman script - knowledge that might be scarce even among scholars, unless they have specialized in Thai and Indic linguistics?

As a contrast, literate Thai people and even foreign students who have learned to read and write Thai typically acquire the more common pronunciation rules for Indic-based words represented in Thai script quite quickly, so I do not think there is as much of a problem there.

Using the RTGS would ensure smoother communication here, I believe. But then again, we are probably arguing from different angles.

[meadish_sweetball]

As you suggested, Meadish, perhaps moderators could request that posters use a predefined system to render Thai in a consistent fashion.

We tried using an extension of the RTGS, but it was found to be too much effort. One problem, of course, was that it required one to truly know how a word was pronounced, and that can be tricky.

One innocent example is เมตร. The RID gives [F]meet; the thai-language.com transcription currently gives [F]meet ('maeht[r]F' in its notation) but the sound-clip gives [H]met; thai2english.com gives [F]met; Nit Tongsopit's dictionary gives [H]met (though length marks often go missing from that dictionary); and so on. As far as we can make out, the actual pronunciation seems to vary between [H]met and [H]meet.

Yes, this word is tricky. The tone is high without a doubt - I've never heard a native speaker pronounce it with a falling tone, but the vowel length seems to fluctuate.

[DavidHouston]

Let's see if we can deliniate specific needs as defined throughout this thread:

1. The need for language scholars to have a common method of expressing sounds throughout the spectrum of languages;

2. The need for non-academic Thais to be able to use non-Thai characters to express words and names to a foreign audience (Tourism Authority of Thailand, traffic sign writers, etc.);

3. The need for new students of Thai to gain a rudimentary understanding of the range of Thai sounds before they learn the Thai writing system;

4. The need of librarians and publishers to catalog Thai literary works;

5. The need for two-way, mechanical transliterations using computer models and rules.

I am sure that there are others as well. Seems to me that our need in these forums is limited to need 3; while the others are of academic interest and other active, inquiring minds, can we agree on what is most useful in this context?

[Richard W]

Thanks for the correction; my mistake for assuming the RTGS was behind the Suvarnabhumi spelling.

Secondly, before diacritic stripping, the graphic system is just as phonetic as the Thai script itself. After diacritic stripping, it still gives you a better idea of the tone than the RTGS.

Are these diacritics you mention, tone diacritics? The ^ ´ ` etc. of IPA, or other ones?

In the graphic system as defined by Rama VI and recounted by Griswold, there are several diacritics:

• Numeric superscripts (¹²³⁴) corresponding to the written tone marks of Thai.
  
• Macron for the high vowels and /a:/ - I don't know how the syllabic liquids are transliterated. (Standard practice for Indic scripts)
  
• Breve for mai hanakat and maitaikhu above <e>
  
• Dot below for the retroflex consonants and visarga (sara a, written aḥ when not part of another vowel). (Standard practice for Indic scripts)
  
• Grave for the low vowels (è, ò) not represented in most Indic scripts.
  
• Umlaut for non-low central vowels (ö, ï).
  
• Underline for some of the consonants added for Tai - ฏ ต ป ฮ.
  

Something must be missing in either my understanding or in Griswold's account, for there seem to be some holes in the system.

I guess you are right, but obviously that would require the prior knowledge necessary to deduct the tone from this graphic system in Roman script - knowledge that might be scarce even among scholars, unless they have specialized in Thai and Indic linguistics?

I was comparing it with the RTGS. In the RTGS, ฐาน ([R]thaan) and ทาน ([M]thaan) merge as than. In the graphic system, they are distinguished as ṭhāna and dāna, which still contrast as thana and dana when the diacritics are stripped off.

[Richard W]

3. The need for new students of Thai to gain a rudimentary understanding of the range of Thai sounds before they learn the Thai writing system;

Shouldn't this be, 'The need to concisely and unambiguously represent the sounds of Thai words;'? Learning the Thai script won't help you know whether ห้อง has a long vowel or a short vowel. On the other hand, it will hep you use a Thai dictionary to find the tones of the final syllables of วิษณุ (high) and กฤษณา (rising).

[DurianChips]

3. The need for new students of Thai to gain a rudimentary understanding of the range of Thai sounds before they learn the Thai writing system;

Shouldn't this be, 'The need to concisely and unambiguously represent the sounds of Thai words;'? Learning the Thai script won't help you know whether ห้อง has a long vowel or a short vowel. On the other hand, it will hep you use a Thai dictionary to find the tones of the final syllables of วิษณุ (high) and กฤษณา (rising).

I think the first goal may be more appropriate for a beginning student, and the second for a more advanced student.

[DavidHouston]

I agree with both Richard and Durian. However, the Thais use a standard phonetic system of their own langauge to express the correct pronunciation of ambiguous words very clearly. For example, the Matichon Dictionary gives the following phonetic equivalents of Richard's two examples: วิษณุ - "[วิด-สะ-นุ]" and กฤษณา - "[กฺริด-สะ-หนา]". Now isn't that very clear? A foreign student need not have a very broad knowledge of the Thai written language to pronounce these two words correctly.

[mangkorn]

I agree with both Richard and Durian. However, the Thais use a standard phonetic system of their own langauge to express the correct pronunciation of ambiguous words very clearly. For example, the Matichon Dictionary gives the following phonetic equivalents of Richard's two examples: วิษณุ - "[วิด-สะ-นุ]" and กฤษณา - "[กฺริด-สะ-หนา]". Now isn't that very clear? A foreign student need not have a very broad knowledge of the Thai written language to pronounce these two words correctly.

Thanks for clearing this one up: วิษณุ - "[วิด-สะ-นุ]

I was going batty trying to understand why the final syllable has a high tone.

[Richard W]

I agree with both Richard and Durian. However, the Thais use a standard phonetic system of their own langauge to express the correct pronunciation of ambiguous words very clearly.

Unfortunately, the standard phonetic system does not show the length of such words. An extreme case is showing the vowel length of [MS]ngoen. While I have seen forms analogous to เงิน็ used for transliterating Northern Thai, most Thais think this notation is just plain wrong.

For example, the Matichon Dictionary gives the following phonetic equivalents of Richard's two examples: วิษณุ - "[วิด-สะ-นุ]" and กฤษณา - "[กฺริด-สะ-หนา][/size[". Now isn't that very clear? A foreign student need not have a very broad knowledge of the Thai written language to pronounce these two words correctly.

Which is why I said that understanding the Thai writing system is all you need to find this information from a Thai dictionary.