ศาลาชุมนุม

[Chris Pirazzi]

Hello all,

I would like to understand more about the phenomenon whereby certain syllables of Thai words, particularly those with sara "a" that are not word-final, get pronounced with a mid tone (or some might say low tone) instead of the tone predicted by the tone rules, e.g. "ma" and "ta" (edited out "ya": it is long) in มหาวิทยาลัย ma-haa-wit-ta-yaa-lai.

For this post, I am specifically not referring to the "irregular tone rule" concerning pairs of spoken syllables where the tone of the second is sometimes determined by the class of the first, e.g. ถนน, เฉลียง, and sometimes by the class of the second, e.g. แสดง, เกษียณ, เฉพาะ. That rule has been studied to quite satisfactory detail on this and other forums, e.g.:

http://www.thai-language.com/id/830221 (part 3d)

Instead, I am referring to cases where a single syllable gets pronounced mid/low, seemingly because it occupies an "unimportant" position at the start or middle of a syllable.

I cautiously refer to these as "unstressed" syllables but I am aware that that is a very dangerous term to use, since "stress" in English is really about inflection and vowel length, and these parameters are already "used up" in Thai to determine the meaning of a word!

I would like to better understand this "unstressed" phenomenon. In particular:

What vowels can it happen to?

• only short ones?
• only sara a?
• only implicit (unwritten) sara a?

In what contexts does it happen?

• is it even remotely consistent across different Thai speakers?
• is it predictable from the adjacent sounds?
• is it predictable from the adjacent written letters?
• is it predictable from the word's etymology?
• does it ever happen word-final?
• are there any other constraints on where it can happen in a multi-syllabic word?

What effect does it have on the syllable?

• does it always make the tone low? mid?
• can it shorten a long vowel?
• are there any other effects it can have?

Do educated Thai people consider it to be

• a "bad, sloppy" pronunciation ?
• one possible correct pronunciation (another one being the pronunciation predicted by the tone rules) ?
• the only correct pronunciation ?

What words (in Thai and English) do people use to refer to this phenomenon? "Unstressed" seems like a poor choice of word, or at least one that needs to be carefully re-defined in the context of Thai language. Very recently someone mentioned a Thai linguistic concept gaan-pra-wi-san-cha-nii but I have not yet tried to guess the Thai spelling and google it, nor do I have any idea where to start looking for a book or reference about it on the academic side.

I am of course open to the possibility that there are several, similar phenomena operating here, and I'd like to understand as many of them as possible!

I figured the esteemed crowd here would already know quite a lot about this subject and be able to give me some info or pointers to info. Up here in Pai in the mountains of Mae Hong Son, I don't have access to any academic libraries, but if anyone happens to have any insights on this, PDF or online papers, books about this that can be ordered (e.g. from Amazon), or emailable scans of relevant papers, it would be greatly appreciated!

Thanks!

- Chris Pirazzi

[David and Bui]

Please provide some specific examples of Thai words and their pronunciations in your context. Is it possible that some of what you are hearing is a function of the local dialect? Your examples would allow us readers to compare what you are hearing with the sounds of Central Thai pronunciation to see how wide-spread the phenomenon of which you speak is.

Reflecting on your statement: "I cautiously refer to these as 'unstressed' syllables but I am aware that that is a very dangerous term to use, since 'stress' in English is really about inflection and vowel length, and these parameters are already 'used up' in Thai to determine the meaning of a word!"

I am not sure that all of the intonational possibilities are "used up" in making tones. Intonation is as much a part of Thai speech patterns as is tone, in my experience. Of course, Thai intonation is dramatically different than what we use in English.

Thanks so much for your interesting and careful observations. I hope you get a lot of thoughtful responses.

[Chris Pirazzi]

After many hours of Googling I managed to find one source that discusses this question, though it's by no means a complete answer. It addresses the effects of "unstressed"ness but only says a little about the environment for it. Here's what it has to say:

The Thai Writing System By Nantana Danvivathana
Published 1987
Buske Verlag
345 pages
ISBN 3871187534

Many pages of it are on Google Books.

Searching for the word "stressed," we find many interesting passages including this passage on p.160-161:




So some effects of "unstressed"ness include shortening, conversion to mid tone, and reduction/centralization (e.g. a to ə,
-ะ to เ-อะ).

On p.160 the book also asserts that

• all stressed short syllables with no final consonant actually end in a glottal stop (I'm not convinced of this, e.g. อะไรนะ, but anyway let's assume it's true)
• when such a syllable is unstressed, it often loses its glottal stop, e.g. the ศิ in ศิริ does not have a glottal stop; same with อุษา and other words given.

So another effect of being "unstressed" can be the loss of final glottal stop.

As far as the environments where "unstressed" syllables can occur, the book clearly includes the first "syllable" of initial consonant clusters that cannot cluster together, e.g. ขยาย, ฝรัง, etc. (p.150), and in that case only in "running connected speech." The book gives a few examples of other environments too, like ศิริ, but doesn't really address the issue of how to predict whether a syllable will be "stressed" or "unstressed."

Well it's something, but definitely not the whole picture!

[David and Bui]

Are you talking about linguistic changes over time or of inconsistencies between the current written language and the current spoken language?

For example, as you noted, the current word for "carpentry nail" in is " ตะปู [N] nail; spike Class. ตัว, ดอก". However, a synonym is ตาปู. There is an island off the cost of Phangnga which looks like a nail (referred to by tour guides as James Bond Island). The sign at the location calls this place "เกาะตาปู". Since a nail does look like the eye of a crab, this is a pretty reasonable metaphor which entered the common parlance. The longer word does take more effort to pronounce than the shorter word.

The Royal Institute Dictionary merely says "ตาปู น. ตะปู." for the longer word and
"ตะปู น. สิ่งทําด้วยโลหะ มีปลายแหลม หัวมนแบน ขนาดต่างๆ กัน สําหรับตรึงสิ่งอื่นให้แน่นโดยใช้ค้อนเป็นต้นตอกลงไป, ตาปู ก็เรียก."

Perhaps the longer word is in the process of becoming archaic?

Do linguists call this shortening of words "contraction"?

In any event, is the phenomenon to which you refer in your original posting current pronunciation practice or a migration in spelling and pronunciation over time?

Thanks.

[Chris Pirazzi]

Please provide some specific examples of Thai words and their pronunciations in your context. Is it possible that some of what you are hearing is a function of the local dialect?

Sure, no problem. Here's a few examples:

มหาวิทยาลัย - "ma" and "ta" seem to be mid-tone /ə/

(note "ya" was erroneously included in my original post; I edited that out since it's really the long syllable "yaa").

All of the "Non-conforming 'Leading Consonant' Clusters" (as Glenn calls them in http://www.thai-language.com/id/830221) are examples where the initial syllable's vowel quality seems to change from /a/ to /ə/ (again for this post I am interested only in the sound of the /a/ syllable, and not in the magical tone effects on the second syllablle):

สนาม
ขยาย
ฝรัง

As far as the change in tone, it's a bit hard to hear in the syllables above since they have high initial consonants, so we expect a low tone, and it's hard to distinguish a low tone from a mid tone when it is spoken so quickly. If we look at similar clustered words with a low initial consonant, which should have a high tone, the contrast is greater and we can indeed hear that the tone doesn't match that predicted by the tone rules:

ทหาร - I tend to hear mid-tone /tə/ rather than /a/ or high-tone
คติ - "The Thai Writing System" states that this has a mid tone cuz it is unstressed (p.153)
คดี - similar case, which I've definitely heard with a mid tone

As far as dropping glottal stops, here are some examples from "The Thai Writing System":

ศิริ first syllable lacks its "normal" glottal stop
อุษา first syllable lacks its "normal" glottal stop

As I think of more examples, I will post them.

In general, what we see here is a wide range of effects which fall under the general category of "unstressed." I would definitely put out the possibility that that is too vague a characterization, and what we actually have are a set of unrelated effects that happen in different, but overlapping, environments. So we shouldn't assume that we have to define a single term "unstressed" as such: instead, it's more interesting to figure out what actually are the effects and their environments. So apologies from me on the name of this thread

I'd love to hear anyone's thoughts on any of these effects, their environment, and the other questions I posed in the first post.

Thanks!

- Chris Pirazzi

[Chris Pirazzi]

Are you talking about linguistic changes over time or of inconsistencies between the current written language and the current spoken language?

My intent was definitely just the current spoken language (say even Bangkok TV news dialect, to simplify it even more) but it's a very interesting idea that these cases might arise from archaic spellings/pronunciations!

[Chris Pirazzi]

I found another tempting morsel of information in the 1998 AUA Thai Language Center Reading Course (J. Marvin Brown, apparently written in 1979!) p.56:



This is interesting, if depressing, because it shows that basically any pattern of stressed and unstressed is possible for two bare consonants in a row, and that sometimes there is more than one acceptable reading.

But, again, the AUA book doesn't offer any advice as far as predicting which readings are correct and which are not, for a given word. That would be of real value for a Thai learner! Anyone have any insights here?

Also, The AUA book doesn't clearly define "weak syllables." There is a section on p.29 which suggests simply that a weak syllable is a syllable which is pronounced mid but should be something else, and gives several examples including single bare consonants like สบู่ and also non-clusterable pairs of consonants like ถนน. But it doesn't give us any clue as to which single bare consonants will be "weak" vs which are not. Any ideas here folks?

Thanks

- Chris Pirazzi

[Rikker]

Chris, you've already covered some of this yourself, but here are my observations:

Notice that not only does the distinctive tone get lost, but in the case of สระอะ, the vowel changes from true อะ to more like schwa (uh). In linguistic terms, the vowel is centralized, much like American English unstressed vowels (America = uh-mer-uh-kuh).

What vowels can it happen to?
* only short ones?

Yes, only short vowels. But a vowel may be shortened from a long vowel and thus become weak, e.g. ตาปู > ตะปู. This actually demonstrates how the 'weak' syllable retains its etymologically correct tone--it's not 'meant' to be pronounced with the low tone of ตะ. Thus inserting a glottal stop after the vowel อะ and enunciating the low tone may may be very artificial.

* only sara a?

No, it happens with other short vowels, like ทุเรศ, ศิลา, etc.

* only implicit (unwritten) sara a?

This is irrelevant, particularly because spellings for many words have changed, and a word spelled with ะ at one point may have it removed later. This has no effect on the pronunciation. For example ชะนิด is nowadays spelled ชนิด.

*open syllables (no final consonant).

Yes, this is right. Only in true open syllables, though. The glottal stop seems to force a tone one way or the other. For example, สะ in สะพาน is a 'weak' syllable, but สะ in สะใจ is not, because you can't say สะใจ without a glottal stop.

* is it even remotely consistent across different Thai speakers?

Yes, I'd say so. Anyone who is speaking normal connected speech who is not speaking self-consciously (i.e. not a newsreader, who is intentionally speaking ชัดๆ), this is a natural linguistic process.

* is it predictable from the adjacent sounds?

From the environment, yes, the actual adjacent sounds may not be relevant. It has to be a syllable not at a morpheme boundary, as in the first syllable of กระต่าย, but not กระเต็มหน้า (where กระ here mans 'freckles' and is pronounced with a clear low tone and glottal stop). Since this is a phenomenon of 'weak' (i.e. unstressed) syllables, it must by definition be within a polysyllabic word, because any monosyllabic word must be stressed (although it's probably possible that this also happens in longer fixed expressions consisting of several monosyllabic words--I just can't think of one).

* is it predictable from the adjacent written letters?

See above--this is irrelevant.

* is it predictable from the word's etymology?

Since most polysyllabic words come from other languages, many/most instances are probably not native words, but this is not a requirement, since we see it in the first syllables of ตะปู (reduced from ตาปู), ฉะนั้น (reduced from ฉันนั้น), ทหาร (reduced from, by my theory, ทวยหาร = ทวยหาญ). Basically, it doesn't matter how how it became polysyllabic, it just matters that it is.

* does it ever happen word-final?

See above--basically, no.

* are there any other constraints on where it can happen in a multi-syllabic word?

Non-stressed syllables, which usually means not the last syllable, but there are other factors involved.

What effect does it have on the syllable?
* does it always make the tone low? mid?

It basically makes it a mid-tone.

* can it shorten a long vowel?

See above. Being an unstressed syllable causes long vowels to shorten, yes, which is how ตาปู becomes ตะปู, and conveniently keeps its mid tone.

Do educated Thai people consider it to be
* a "bad, sloppy" pronunciation ?

I'd say no, it's pretty universal.

* one possible correct pronunciation (another one being the pronunciation predicted by the tone rules) ?

It's a matter of the difference between 'strong' form (aka citation form), which is how it is pronounced in isolation, and 'weak' form, which is how it is pronounced in running connected speech.

[Rikker]

This is interesting, if depressing, because it shows that basically any pattern of stressed and unstressed is possible for two bare consonants in a row, and that sometimes there is more than one acceptable reading.

But, again, the AUA book doesn't offer any advice as far as predicting which readings are correct and which are not, for a given word. That would be of real value for a Thai learner! Anyone have any insights here?

I think AUA is overanalyzing some of these words. The words สหรัฐ, นวนิยาย, ปฏิเสธ, in my experience, are very commonly pronounced with both the first two syllables weak.

They're right about กรกฎาคม. Even RID recognizes this one, giving [กะรักกะ-] as an alternate pronunciation. It's kind of arbitrary which ones get ไม้หันอากาศ and which don't. See ศัตรู [สัดตรู], which gets one, and สตรี [สัดตรี], which doesn't. In the original Sanskrit for both, the first syllable is open syllable /sa/, and Thai turns it into a closed syllable /sat/.

Also, The AUA book doesn't clearly define "weak syllables." There is a section on p.29 which suggests simply that a weak syllable is a syllable which is pronounced mid but should be something else, and gives several examples including single bare consonants like สบู่ and also non-clusterable pairs of consonants like ถนน. But it doesn't give us any clue as to which single bare consonants will be "weak" vs which are not. Any ideas here folks?

In my reading, Thais sometimes pretend that there's a difference between how they pronounce the first syllables of words from at least three categories:
*'eroded' words (คำกร่อน), like สายดือ > สะดือ
*clusters in foreign loanwords, like English /st/ start > สตาร์ต, or Khmerថ្នល់ /tnɑl/ > ถนน.
*inherent vowels in Indic words, like in สระ 'vowel'.

I would contend that the vowels of the first syllables of these are all pronounced the same, 'weak' with mid-tones and schwa-like vowels. Thais call this reduced version of อะ 'เสียงอะกึ่งเสียง' (Google กึ่งเสียง to see plenty of references to this). I don't think that most understand this wholly phonological phenomenon.

Oh, and by the way, การประวิสรรชนีย์ is just the formal term for 'writing the symbol ะ'. วิสรรชนีย์ is the formal name of the symbol. It's not relevant to this discussion, since orthography doesn't affect the phonology here.

Sorry my posts are so slapdash and hurried. Hope they make sense.

[Chris Pirazzi]

Wow, thanks for your response...this is incredibly useful!

A few questions, taking your text out of order:

Chris
* are there any other constraints on where it can happen in a multi-syllabic word?

Non-stressed syllables, which usually means not the last syllable, but there are other factors involved.

I think AUA is overanalyzing some of these words. The words สหรัฐ, นวนิยาย, ปฏิเสธ, in my experience, are very commonly pronounced with both the first two syllables weak.

This gets to the heart of one of my main questions. Ideally I'd like to arrive at a set of rules that Thai learners, and/or computer code, can use to predict the unstressed syllables in any given word, or, if there is no deterministic set of rules, at least some rules that will steer us in the right direction in most cases!

In other words, the texts I've found so far do a great job of showing us the various possibilities, but I have a somewhat harder task in mind, one that is more relevant to the Thai learner or computer algorithm: predicting which of these possibilities a given word will use! It may not be possible in all cases, but I'd like to see how close we can get.

With that in mind, let's start here: say a new Thai learner (or a computer algorithm) is presented with a new Thai word such as กรณี, สหรัฐ, นวนิยาย, ปฏิเสธ, which definitely contains more than one syllable. Say we first make the following assumptions:

• there are no morpheme boundaries (as in กระเต็มหน้า)
• all of the syllables with (implicit and explicit) short vowels are open syllables. So, this assumption omits cases like:
  
• วิทยาลัย where some syllables (วิท, ลัย) have final consonants
• สะใจ where there is a "hidden" glottal stop final consonant,
• กรกฎาคม where there is a "hidden" −ั− that again genererates a final consonant.

Under those assumptions, can we then conclude that all the non-final short syllables are weak and will be pronounced mid (and that [a] becomes [ə])?

AUA clearly thinks no, but I'm wondering about your opinion. What are the "other factors" you allude to that determine which short, open syllables are unstressed? Are the factors you had in mind all covered by our assumptions above or are there others?

Now, say we relax those assumptions a bit. Let's say we still assume no morpheme boundaries, but we allow some of the short syllables to be closed (via an explicit consonant วิท, a hidden glottal สะใจ, or a hidden −ั− กรกฎาคม).

First, is it possible for one of those closed, short syllables to be "unstressed"? In particular,

• Will they always have the tone predicted by the tone rules, or are there cases where they also get shunted to a mid tone?
• Will they always have their natural phonetic value (e.g. [a] for อะ), or are there cases where they get shunted (perhaps centralized) to another value (e.g. [ə]) because of their "unstressed" position? So far I have not been able to think of any words where closed, short syllables change their phonetic value in a pattern that resembles the "unstressed" positions, but I'm wondering if anyone else out there has.
  
  There do seem to be cases where closed, short syllables change their phonetic value in general. For example, I hear Thais use the same /e/ sound in เปล and เป๊ะ but a distinctly different sound in เม็ด, and I hear many (but not all) Thais use the same /i/ sound in หมี and ติ but not in ปิด. But since these effects happen even in single-syllable words, they don't seem to be related to "unstressed" syllables.

Second, does anyone know any tricks or techniques we can use to predict when syllables in ambiguous multi-syllabic words like วิทยาลัย, กรณี, สหรัฐ, นวนิยาย, ปฏิเสธ, สะใจ, and กรกฎาคม are closed? In other words:

• predict that วิทยาลัย is wit-ta and not wi-ta
• predict that สะใจ is saʔ- (low) and not sa- (mid)
• predict that กรกฎาคม has a "hidden" −ั− at รก

I wouldn't expect that we'd be able to predict these in all cases, but I am wondering if anyone out there knows a way to predict any of these cases with reasonable probability.

Chris
What vowels can it happen to?
* only short ones?

Yes, only short vowels. But a vowel may be shortened from a long vowel and thus become weak, e.g. ตาปู > ตะปู. This actually demonstrates how the 'weak' syllable retains its etymologically correct tone--it's not 'meant' to be pronounced with the low tone of ตะ. Thus inserting a glottal stop after the vowel อะ and enunciating the low tone may may be very artificial.Just curious, as far as 'etymologically correct' theory, the are there any cases of long syllables with high consonants getting shortened, and do they have a rising tone?

In my reading, Thais sometimes pretend that there's a difference between how they pronounce the first syllables of words from at least three categories:
*'eroded' words (คำกร่อน), like สายดือ > สะดือ
*clusters in foreign loanwords, like English /st/ start > สตาร์ต, or Khmerថ្នល់ /tnɑl/ > ถนน.
*inherent vowels in Indic words, like in สระ 'vowel'.

I would contend that the vowels of the first syllables of these are all pronounced the same, 'weak' with mid-tones and schwa-like vowels. Thais call this reduced version of อะ 'เสียงอะกึ่งเสียง' (Google กึ่งเสียง to see plenty of references to this). I don't think that most understand this wholly phonological phenomenon.

Interesting! Can you suggest any English or Thai references that further discuss the stressed/unstressed topic, either books or papers (ideally ones available in the boonies or emailable )?

Thanks again for this hugely useful information,

- Chris Pirazzi