Problem With Thai Syllabification

[bytebuster]

Hi everyone,

I would like someone to point me on some more or less formal rules of Thai syllabification.

There are plenty of loanwords that have rather odd syllabification, and they are difficult to syllabify by a program.

Here are some examples of what i mean:

สุรวงศ์ [sù - ráʔ - woŋ], while สุ-รวงศ์ [sù - ruaŋ] or even สุร-วงศ์ [sùn - woŋ] would be not only grammatically valid, but seem to be more preferable from the point of common sense.

For instance, the word of จรวด [jà - rùat] completely conforms the grammar rules; it is not จะระวด, จร-วด or จระวด.

Here's another example: นวยนต์ [náʔ - wáʔ - jon]. The logic is like that:

1. นวย-นต์ is impossible because นต์ is not a valid syllable;
   
2. นว-ยนต์ - see below, let's just assume it is invalid;
   
3. So, the 1st syllable is นะ, and continue to analyze the rest;
   
4. วย-นต์ is invalid, similarly to item #1;
   
5. So, the second syllable is วะ;
   
6. So, the entire word would be นะ-วะ-ยนต์;
   

But wait a minute, wo-wen (ว) is a valid final consonant! All Thai grammar books say that words like เร็ว [rew] - it is not "Sara-ew", it is sara-e with the final consonant wo-wen.

If this is true, นว [now] would be a valid syllable! I'm quite aware it is a dictionary word containing two syllables, but grammatically it seems valid as single syllable as well.

My question is what formal rules can I apply so that these words syllabified properly?

Also, please note that I don't consider adding these words as exceptions unless it is absolutely necessary. See here for my background.

Thanks for your help.

[Richard W]

I can't offer any general rules, just specific examples.

For สุรวงค์, the clue lies in the karan, which says Sanskrit or English, and which rules out -ua- in most cases. That leaves sù-rá-woŋ, sǔn-woŋ and sǔn-rá-wonɡ. However, if you say sǔn-woŋ, how will people tell that you know that it has รʔ I suppose you might reduce the odds in this case by remember that Sanskrit -rv- yields Pali -bb-, which makes sù-rá-woŋ more likely than sǔn-rá-wonɡ.

For นวยนต์, there are two approaches. First, /now/ is *not* a possible Thai syllable. There are restrictions on the co-occurrences of vowels and final /w/ and /y/ in Thai. Secondly, have you forgotten your Sanskrit? *navyantra is unlikely - one would expect *noyantra instead, though I suppose *navyantra might arise from pre-Classical Sanskrit *naviantra.

[bytebuster]

I can't offer any general rules, just specific examples.

Richard, thank you very much; I've been waiting for your response.

The idea of karan is excellent. I will certainly take it into account. There is a limitation of core logic making syllables independent to a large extent, so I'm not sure if I can do something is karan lays several syllables after, but it worth thinking.

And yes, thank you for reminding me of possible consonant reduplication resulting [sǔn-rá-woŋ].

Sanskrit -rv- yields Pali -bb-

I've never heard of such rule and failed to google anything. Can you please elaborate?

There are restrictions on the co-occurrences of vowels and final /w/ and /y/ in Thai.

I realize there are no such words. However, it looks rather strange that [-aw -ew -iw]are possible but [-ow], or, more specifically, default vowel + wo-wen is not. I mean, it is hard to know this rule in written form.

Yes, Sanskrit has very few diphtongs (-ai-, -au-, and -ar-, actually), but an automatic transcription tool can't guarantee that a certain word is Sanskrit, not proto-Khmer, Mon, Mid Chinese, or Malay.

Secondly, have you forgotten your Sanskrit?

It's rather hard to forget something I've never studied

[Richard W]

For Sanskrit -rv- yielding Pali -bb-, see http://en.wikipedia....i#Assimilations and look for 'sarva'. (I suspect the English page is the most complete on the topic.) I first became aware of it through นิพพาน for nirvana.

The restrictions on the co-occurrences of vowels and final /w/ and /y/ in Thai are that Front + /j/ does not occur (เอย represents /əəj/), and rounded + /w/ does not occur. It so happens that /oj/ is also missing, while /oːj/ does occur.

...an automatic transcription tool can't guarantee that a certain word is Sanskrit, not proto-Khmer, Mon, Mid Chinese, or Malay.

Only surprise is guaranteed when readinɡ Thaiǃ However, sometimes the information is given.

[bytebuster]

Only surprise is guaranteed when readinɡ Thaiǃ However, sometimes the information is given.

Wow, what a useful info! I guess it worth a deeper look into Pali, and, most likely, some statistical rules will arise. Once again, thanks a lot!

I'm just a bit aware of ending up with a monstrous universal transcriber for a whole family of Indic languages...