Sanskrit-Thai transliteration issues

This post is meant to elicit feedback, please leave comments below if you think something in this article is factually incorrect.

Common problems

Without going too far into history, Thai alphabet is part of a Brahmic family of scripts and as such, like Devanagari itself, was originally perfectly equipped for writing down Sanskrit. Many modern Thai words have Sanskrit origins, too. With time, however, correct pronunciation has been lost even though spelling in many cases remained the same. Pronunciation standards were not maintained and rather the way the common people say these words became official, and so modern Thai has many rules to regulate this “corruption”, for the lack of a better word.

In Hindi, for example, people drop final “a” so that Bharata becomes Bharat and Prabhupada becomes Prabhupad. Modern Thai goes further than that. Take Thai word for Lord Indra – พระอินทร์. First part is “Pra”, which is a honorific and doesn’t concern us here. “Indra” itself is อินทร์ – it has all the letters “i”, “n”, “d/t”, and “r”, but if you drop final “a” saying “Intr” becomes impossible so “tr” is dropped as well and pronunciation becomes just “In” – Pra In. In some cases full pronunciation is retained, however, like part of the official Thai name for Bangkok is “Mahintara”, which is basically “Mahindra”. In this case “dra/tra” is retained but a vowel is inserted in the middle to make pronunciation easier – “intara”. Sometimes it’s spelled “Inthara” – there are resorts and hotels with this name (to be fair, there’s also “Ramintra”).

Thus there are many cases where letters are part of the word but are silent and there are rules for when they are read aloud and when not.

Another way of simplification concerns final consonants – whatever consonant appears at the end of the word it must make one of the eight permitted sounds. Thus words ending in “r” actually end in “n” and words ending in “s” actually end in “t”. Sometimes, to an English speaker, this affects letters in the middle of words, too, like letter ล – “l” in the middle of ชลบุรี becomes “n” – ChoNburi. There are rules to govern these changes as well.

Implicit vowels – one of the first things to learn when reading Devanagari, are present in Thai but their pronunciation is not fixed. Depending on the position in the word they become “a” or “o”. In the word ถนน – which is written as “th-n-n” implicit vowels are read as thAnOn. There are rules to govern these changes, too.

Another major area is consonant clusters. In Sanskrit there are hundreds and possibly thousands of them but in Thai there are only fourteen. Any combination outside of these fourteen requires inserting a vowel, similar to “Intara” above. Every school child is taught English nowadays but without sufficient training and discipline even common words like “school” and “spa” (there’s spa on every street corner in Bangkok now) are pronounced as “sa-kool” and “sa-pa”. This rule isn’t fixed, however – proper pronunciation of fashionable words is a sign of social status so many Thais will say “blueberry” perfectly, but those who don’t care enough would still say “baluberry”. So, the rule for inserting “a” is there but it’s not applied when there’s a need for trying to sound like a foreigner. In any case, consonant clusters transliterated from Sanskrit need to indicate that they are, indeed, clusters and that inherent vowels need to be dropped.

In some cases Thai consonants become silent precisely because they are part of clusters that are difficult to say, notably in ศร combination which is important to us because it’s part of “Sri”. Second letter, ร, becomes silent and so words like Srinakarin are pronounced Sinakarin. Full Thai spelling is ศรีนครินทร์ – Srinakarindra, and “Sinakarin” is what is left of it after all the rules are applied.

Finally, and this is probably the biggest and the most obvious issue – some Sanskrit sounds have been totally lost. Five of them are in the “Gha .. Bha” group of Sanskrit letters, and there are no Thai sounds for ś and ṣ either. As there’s no Thai “sh” for Krishna, Thai speakers say “s” instead – Krisana (-sana because “sn” is not an allowed consonant cluster). These missing sounds need to be taught – how to make them, how to position one’s mouth, what they should sound like etc. This isn’t an issue with transliteration per se – because letters for these sounds still exists, but we need to teach people how to make totally new sounds and we need to include these techniques in our pronunciation guide.

Good news is that “sh” is common in English and so Thai people are very familiar with it already. Nobody has been a better teacher than “Share” button on Facebook. Otherwise, Thai pronunciation of “share” and “chair” is about the same – with soft “ch”.

A few words need to be said about “ch” as well – there appears to be confusion among Thai speakers whether letter จ should be written as “ch” in English (and so as “ca” in Caitanya). Bangkok’s most famous market is Chatuchak and it’s written like this on every map and in every travel guide, but when abbreviating many would write it as JJ market instead. “JJ Market” is also the name of a huge mall they built there. One English language newspaper in Thailand would write someone’s name as Chakrit but another would write Jakrit instead. However, all agree that Thai ruling dynasty should be spelled Chakri – not Jakri. I mention this because original Thai letters for “ca” and “ja” have been swapped from the time of Sanskrit and any possible confusion needs to be cleared.

Thai language also doesn’t have “v” and so “va” is pronounced as “wa”. In some cases this “wa”, which is technically a consonant, will be made into an official part of a diphthong – there are rules for that. As as result, “deva” will be read as “dewa” (or actually as “thewa” because of d-th sound shift). Words with “sva” will be read as “sawa” – because of v-w and insertion of “a” between “sv”.

When devising transliteration scheme we should keep in mind not only how to reverse all these rules but remember that many of the rules were often introduced specifically to accommodate wishes of common people. In effect, we have to go against what people want. Here’s a list of what needs to be accounted for:

  • teach production of new sounds
  • insist on pronunciation of consonant clusters
  • insist on pronunciation of correct final consonants when these sounds are not allowed in Thai
  • indicate correct vowel when a Thai speaker would instinctively read differently

We should also decide how strict we need to be with enforcing all these new rules. BBT standard for transliteration into English has been “100% accuracy” from the start and with so many ISKCON devotees studying Sanskrit it’s not going to be relaxed. In real life, however, English speaking devotees make no differentiation between na-ṅa-ña-ṇa, ta-ṭa, and da-ḍa groups of Sanskrit letters, for example. At most, they are aware of ś and ṣ but only vaguely about a and ā or u and ū. What seems important is “bha” type of sounds because every ISKCON devotee can make them and they form the core of our most common words like “bhakti”, “Bhagavan”, and “Prabhupada”. So, we need to decide which of these sounds and rules should be a priority and introduced forcefully.

Acceptance by Thai public is another area we should pay attention to. It can be of two kinds – impressions of people on the street and opinions of authorities when people ask for clarification. They are rarely the same, though eventually authoritative opinions prevail over initial public reaction. If the authorities (academics or influential monks in our case) do not approve of our selected method we will be allowed public space to preach whatever we want and even collect our own following but we will never be accepted as a genuine article by people who actually control Thai society. That will be ironic because these people spill gallons of blood to preserve old Vedic traditions against onslaught of modernization and new ideas. This is an important subject outside the scope of this article, but despite being known as a Buddhist country, Thai Royal court is governed by “Hindu” rules, Royal ceremonies are performed by caste Brahmins, and King’s main duty is to govern the country according to rules of “Dharmarajya”, making Thailand into a de facto last Hindu/Vedic kingdom in the world. These are the same principles that are promoted very heavily in our books so we should not be seen as a corrupting influence instead.

One final word – solutions to each of these challenges should include appropriate markup in the transliterated text and this could create an additional problem. I don’t think we can say that we’ve done a good job if our transliteration method cannot be used on digital devices like phones and computers and devotees cannot possibly type “Prabhupada” into their Facebook posts. Just imagine the outcry if every time English speaking devotees would try to type “Prabhupada said” and it would look like garbage on their screens or the screens of their recipients. They will quickly find a way around it and stop using our proposed transliteration. Then we will have no control if their workarounds are correct or how it affects their pronunciation. These days hardly anyone reads printed books and there’s little use for transliteration scheme that is good only for paper but is useless for computers, phones, and the internet. Our English transliteration is already affected by this problem, for example.

In the next section I’ll give an overview of several possible transliteration schemes.

Pali Sanskrit

This method is the preserved writing scheme for the original Pali and Sanskrit and I will refer to is as PS.

Pros – 100% accuracy, official status and therefore full compliance with ISO standards and full support on all digital platforms.
Cons – reading rule is different from modern Thai, over a dozen letters sound differently from modern Thai, and so the text looks unintelligible to a reader without pronunciation guide.

As I mentioned earlier, originally Thai script was meant to write down Sanskrit perfectly. Over the time, however, pronunciation of many letters have changed and now many of original Sanskrit letters duplicate more common modern ones. This has made them obsolete even though they are still part of Thai alphabet and are taught to children in schools. Thai name for Bhagavad Gita, for example, is kept almost unchanged – ภควัต-คีตา now vs ภควทฺ-คีตา in PS, it still uses ภ letter in front but now instead of original “bha” sound it makes “pha”, and original ค is now not “ga” but “kha”, making it “Phakhawat” (ทฺ-d from PS was replaced by ต-t). Same holds for the word “Krishna” itself. In modern Thai it’s กฤษณะ and in PS it’s กฺฤษฺณ – all the core letters are the same but their pronunciation has changed (to “Krisana”).

Looking at the list of problems to overcome – first is the “new sounds” – in PS original Sanskrit sounds are there, part of the alphabet, and they can be found in a variety of sources, starting from Wikipedia, but many Thai Sanskritologists reportedly do not use them when speaking. In this transliteration scheme new sounds need to be taught in the way they were pronounced before. Techniques for adding “h” sound to “b” to make “bha”, for example, need to be invented and explained in the pronunciation guide and people need to be reminded which old letters they apply to.

The other three issues are dealt with automatically because PS does not recognize modern Thai rules at all. Take consonant clusters, for example – in Sanskrit they are indicated by ligatures merging two consonants together and there are hundreds of these unique combinations. No keyboard, however, can have so many letters. What people actually do is use “virama” extensively instead. When two consonants are meant to be merged “virama” is put between them and then it’s up to the font to display a correct ligature. In the first words of Bhagavad Gita – “Dhritarashtra uvaca”, there’s a consonant cluster ś-ṭ-r. It is typed as (using Devanagari symbols, of course) ś – virama – ṭ -virama – r. The font then merges these three into one ligature.

In PS it works the same way. Virama in Thai is Pinthu, a little dot underneath the consonant, but consonants do not need to be merged and so no ligatures need to be learned. That śṭra combination is typed as ษ-Pinthu-ฏ-Pinthu-ร. Pinthu is not displayed on its own so I didn’t type it here, but it is shown underneath and it becomes “ษฺฏฺร”. Here’s with a bigger font ษฺฏฺร

In PS letters sound always the same, there are no changes for any reasons applied in modern Thai and inherent vowel is always “a” so Thai rules for occasionally changing it to “o” are not applied either.

Consider example of the word “mantra” – มนฺตฺร (used more for something like “incantation” or “prayer”, for regular “mantra” there’s shorter version of the same word). Let’s try to read it. First is ม-m, there’s no Pinthu, so “ma”. Second is นฺ-n with Pinthu, so just “n” – “ma-n”. Then ตฺ-t with Pinthu, so “t” – ma-n-t. Finally there’s ร-r, no Pinthu, so “ra”. And now we have ma-n-t-ra.

What I demonstrated here is the only reading rule and pretty short one at that, but it needs to be taught. Without it people won’t know what to make out of the text. They know about existence of Pinthu in principle but not how it is used in PS. Good news is that people can google how to read Pali Sanskrit and there are plenty of pages explaining it. Bad news is that the way people parse their own language into words is very difficult to change and even if there’s only one rule to learn ignoring all the old ones is going to be difficult, plus there are about a dozen letters where sounds are different, too. Thai letter ช – cha, for example, is very very common but in PS it makes sound “ja” instead of “cha” and memorizing this change requires effort.

Switch to using PS also needs to be indicated in the text. In our English books most of the Sanskrit is put in italics but not always. There are plenty of Sanskrit words that appear as part of the normal text, usually proper names. Readers recognize them by the use of diacritics and if there are no diacritics, like in “Veda” or “yoga”, the words are still read the same way. In PS, however, there are only two symbols that could be considered as equivalent of English “diacritics” (Pinthu and Nikhahit – anusvara in oṁ) to indicate that reading rules have changed and without them the same word could be pronounced very differently. Even something as simple as “yoga” won’t have an expected “a” at the end and would look like a misspelling to a Thai reader – unless there’s indication that PS reading rule should be applied.

Good news is that we can use italics for all PS words, even in verse translations (where italics are never used in our English books). The bad news is that italics cannot be copy-pasted between many apps. Facebook does not allow them, for example, and none of the messenger apps like Line, Whatsapp, or Viber support them. Good news is that simply surrounding words with spaces would indicate to Thai readers that there’s something special about them (spaces are not used between words in Thai, only between sentences).

Regarding accuracy – it will become similar to English. All the information necessary for perfect pronunciation is there and it’s up to individuals if they want to differentiate between sounds like n, ṇ, and ñ in speech and writing. Even better than English – all the symbols needed for typing PS are on standard Thai keyboard and so Thai devotees will quickly learn and indicate the difference between “varnashrama” and “varṇāśrama” – the difference which is often lost on English speakers.

Regarding acceptance by the public – no academic could ever say that PS deviates from the tradition and for many of them it would probably be even closer to tradition that their own stands. General public will find it, at least at first, as cumbersome, but this is also their opinion of “Rajasap” – language used in the Royal court, which is so high class and  flowery that it’s incomprehensible to ordinary people. It is still accepted out of deep respect for the institution and it is not expected to change to suit the common taste. In this sense, we can argue that Bhagavad Gita should be no less respectable and deserve an effort to read it.

Modern Pali

There’s no widely known term for it but it’s the way Buddhist Pali texts are recorded in Thai and then used by monks and laymen for chanting prayers on all ceremonial occasions. It tries to address the same problems as we have with transliterating Sanskrit and it generates the same kind of conflicts – what’s correct? what’s better for the people? shouldn’t it be made easier? how strict should we be? etc. The situation there roughly as follows. The ideal is Pali Sanskrit I described above but it is deemed to be too difficult  and is hardly followed. Google search gives ten times more results for “Modern Pali” spelling of popular prayers than for their “Pali Sanskrit” spellings.

However, when I searched google for the first line of Tipitaka, the main Pali Buddhist text, number of results was roughly the same for PS and modern spelling. Interestingly, the same situation exists for our common prayers as well: “nama oṃ viṣṇu-pādāya” gives 500 results and “nama om vishnu padaya” gives 50,000 while properly transliterated ślokas from Bhagavad Gita easily outnumber their versions typed without diacritics. That is to say – scriptures are preserved in their correct form but common usage isn’t, and this demonstrates importance of the ability to use transliteration in our own typing. I bet not many devotees can answer whether correct spelling of “sh” in Vishnu is “ś” or “ṣ” because they have never had to type it themselves.

Anyway, top results for “Tipitaka” are sites that usually offer alternative transliterations for the readers and Pali Sanskrit is always there among the choices.

Pali is studied by every monk wishing to advance in the hierarchy – it’s part of their “bhakti-shastri”, so to speak, but old (and so correct) pronunciation is not part of the exam so very few monks pay attention to it even if it’s the same textbooks written by the same Prince Vajirananavarorasa who institutionalized Thai Buddhism. There are also two “advanced” Pali courses which are far superior in content and demand correct pronunciation but completing them does not give one any social advantages like adding a special honorific to one’s name or rights to royal cremation so even fewer monks take them. When Westerners transliterate these Thai Pali texts into English they invariably use correct pronunciation and invariably note that Thais themselves do not read them correctly. It’s a known issue.

Here’s an example from Wikipedia, I won’t post Thai script, only transliteration of how it would be read. First line in Pali Sanskrit, second in modern Pali:

arahaṃ sammāsambuddho bhagavā
arahang sammasamphuttho phakhawa

The advantage of the second line is that every Thai speaker can read it right away without any training but it looks very different from the original. “Bha” has been replaced with “pha”, “d” and “dh” with “t” and “th”, “ṃ” with “ng”, but what is not immediately obvious is that long “ā” has been lost, too, and become short “a” instead.

Here’s another common prayer, this time I’ll post only Thai:

นโม ตสฺส ภควโต อรหโต สมฺมาสมฺพุทฺธสฺส
นะโม ตัสสะ ภะคะวะโต อะระหะโต สัมมาสัมพุทธัสสะ

Second, modern Pali version, has a lot of “ะ” which force the explicit “a” vowel and that’s how they solve one of the problems.

Consonant clusters are not as common in Pali as they are in Sanskrit but “sva” I mentioned earlier is very very popular and pronounced as “sawa” even if written correctly.

Pali has the advantage of being primarily oral and, more importantly, inviolable as language of liturgy, so monks can force what they think is correct pronunciation regardless of Thai rules, while people will learn to read it as they first heard it. Every Thai household has a book, a brochure, or a leaflet with these prayers somewhere around and they read them just as they heard the monks recite them.

I haven’t seen any enthusiasm among devotees for adapting these method, however. Nor will it sound anything close to what the rest of ISKCON devotees would expect.

Some new alternative

Our existing Bhagavad Gita is the obvious example here but the process would be the same anyway. New sounds (bha, sha etc) need to be taught and need to be indicated in the transliterated text. In English diacritic marks are used in printed books but almost never when devotees write or type themselves. So, if we introduce some new markings not easily available people won’t use them. In English it results in no difference between na and ṇa or śa and ṣa – it’s always na and sha, but in Thai there will be no difference between bha and pha and that wouldn’t sound right to an average ISKCON ear so new markings need to be not only taught in the pronunciation guide but also be usable.

For consonant clusters – Thai language has a character called Yamakkan to indicate beginning of the cluster but it is not included in common keyboard layouts and so impossible to use. The idea was that Pinthu, which is on every Thai keyboard, serves the purpose just as well.

I don’t know how correct pronunciation of final consonants can be enforced or indicated. One would just have to state in the pronunciation guide the rule that their sound never changes and hope people don’t forget to follow it while otherwise reading the text in their familiar way.

Inherent vowels can be made explicit and there are ways to achieve it in ordinary Thai, one of them is adding “ะ” after every consonant like the example above.

Here’s my analysis of the transliteration method in the existing Bhagavad Gita:

New sounds – bha, gha etc are taken care of by adding Pinthu or Nikkhahit above or below the consonant – wherever there’s space left. Ś and ṣ are not represented, however, and soft “ch” is used instead. Thus, the word “cakṣur” has the same ช for both “c” and “ṣ” – ชัคชุร. Same “cha” – ช is used in Kṛṣṇa as well whereas Thai word for Kṛṣṇa has “s” sound (but correct Pali Sanskrit ษ letter). “Va” vs “wa” is not mentioned.

Because Nikkhahit is being used to indicate “h” in “bha” group of sounds there’s no letter to indicate “ṁ” left so it’s simple “m” instead.

Consonant clusters are not indicated (partly because Pinthu has also been used for adding “h” to “bha” instead of its usual role) and so people would, reportedly, read kṣetre in dharmakṣetre as “ka-shetre”, or even “ka-chetre”.

Final consonants are not enforced so the word “cakṣur” would be read as “chakachun” (but with softer “ch” than in English).

Final ‘ḥ” is indicated by “ฮ”, which appears to be correct, but without giving the rule about its pronunciation it would be read as “ho” so that we have “māmakāhO pāṇḍavāś caiva” or “mucyante sarva-kilbiṣaihO”. It’s never used in this way in Thai but “ho” would be the expected sound at the end of words.

Inherent vowels are indicated properly – “man-manā” is มัน-มะนา. The way they are indicated differs for the first and second “a”, however. Long “ā” at the end is indicated by “า” so that’s been taken care of, too.

Regarding acceptance by the public – the first issue people notice is that spelling of “Krishna” is different from Thai, both modern or Pali Sanskrit (which have almost no difference between them). When someone asked about it on, the largest discussion board in Thailand, the answer was that Thai language still uses original Sanskrit but in our Bhagavad Gita it’s a transliteration from English – which is not the effect our books are supposed to produce. Beyond that, it needs to be investigated how an inquisitive Thai person will read transliterated ślokas after reading the supplied pronunciation guide (which explains only the production of “bha” sounds and not much else). Thai devotees have grown fond of this method, however, and do not see the need for any changes.

The prospects of adopting this method for the internet are bleak – because Pinthu and Nikkhahit are used in a way different from ISO standard for Thai language and any parsing software treats these combinations as invalid typing errors. How it displays these errors varies by exact combination and digital platform. On one end of the spectrum Chrome hides them all, on the other end iPhones and Apple software almost always  display them as clear errors. Typing these combinations is impossible on iPhones, possible on Android, and tricky on Windows – to produce “Prabhupada” one would need to type “PrabUHpada” and so enter symbols that go above and below “b” in a reversed order. They would still display as errors on iPhones, though. This is regardless of fonts or of whether it’s a website, an app, or an e-book – it’s considered an error on the level of the text parser itself.

Another consideration is that if we propose a new, non-traditional transliteration method people won’t be able to read our text online if they don’t have access to printed pronunciation guide. Traditional methods, on the other hand, everyone can google on the internet.

Finally, my personal opinion is that we should present 100% accurate Sanskrit transliteration and then leave it to people to raise their own pronunciation to this standard if they Robbing them of the opportunity to ever read Sanskrit correctly seems unfair. This has been done in English and it’s even easier to implement in Thai because it’s already present in the language itself. Not preserving 100% accuracy would also go against the spirit of “as it is” in the name of our Bhagavad Gita.


Thai Bhagavad Gita

I needed a quick place to show the state of the digital version of Thai Bhagavad Gita to people and thought that WP would be easy to post and access but the book is too big for it to handle. On my mobile phone it wouldn’t load at all and even on the computer it takes forever, and since WP doesn’t trim posts on the front page this whole blog became unusable, so Gita had to be moved – to git(a)hub, as it turned out.

Right now it can be found here. Need a picture of the real book for thumbnail and there are lots of characters displayed improperly on mobiles so it’s work in progress. Once done, will send it to

Actually, any website that wants to display it can use it, too, I mean if it’s presented correctly etc and if local BBT doesn’t mind. This is not an invitation to scrap it off the github without saying a word, though.