Brassica for Thai

[Richard W]

I'm not really sure where this thread belongs - and it probably belongs as part of another thread.

Bradn asked for a suggestion to showcase Brassica's handling of suprasegmentals. It strikes me that the development of Thai (a.k.a. Siamese) might be a stressing example, perhaps taking it from the 13th to the 21st centuries. It's got a number of interesting phenomena.

Firstly, there are tonal correspondences that even seem to go back to Proto-Tai-Kadai, though of course these might just be coda features and phonemes that later became tones. Li Fang-Kuei reconstructs 13th century Thai as having an extensive set of voicing contrasts, though I suspect that was a register (breathy vowels v. modal) contrast as in Khmer and Mon. However, early accounts of Mon suggest that the onset voicing and breathiness were then correlated features in Mon, with the change from voicing contrast to register contrast not being completed until more recent times.

This onset voicing or breathiness contrast has induced a tone split apparently creating a system of six contour tones. This breathiness has induced aspiration on initial consonants. Thai orthography principally uses a 3-way tone indication plus pairs of consonants indicating the former consonants; the tone sets are designated A, B and C in Tai-Kadai philology. Two of these six tones (breathy B and modal C) then merged, and Thai orthography often writes modal C as breathy B. (There are a couple of examples in the name of the letters of the Thai alphabet.) Li also posits a sequence of preglottalised consonants; Pittayapon treats the stops amongst them as implosive. For Thai tone development, these had the same influence as voiceless consonants. These preglottalised consonants had their own influence on tone splitting in Northern Thai, patterning with initial glottal stop.

Checked syllables originally lacked a tone contrast; conditioned by vowel length and original voicing of the onset consonant. they now have three possible tones, in contrast to the five possible tones of unchecked syllables. Phonetically, a fourth tone is also possible on unstressed syllables. This tone can survive changes of vowel length.

Bangkok Thai had a sixth tone, apparently now extinct. I suspect it is related to some typically unstressed words written as though in the rising tone actually normally being pronounced with the high tone.

The glottal stop presents a problem of phonemic analysis. At the start of a syllable, there is no contrast between a glottal stop and zero, though it is quite audible at the start of non-initial syllables. Indeed, the surname of a previous Thai prime minister, Prayut Chan-ocha, was transliterated (in accordance with the standard) with a hyphen to show the glottal stop at the start of the second syllable. Syllable final glottal stops are also non-contrastive, being the automatic coda for a short 'open' vowel.

The sequence of phonemes in a Thai syllable is significantly correlated, which led to some descriptions erroneously denying the existence of the falling tone on checked syllables with short vowels. Some of the distributional gaps or non-gaps are due to recent history - the combinations could not arise from early Thai by regular sound changes. (It makes me cringe when I see analyses that ignore history.)

There have been shifts in Thai tones since the middle of the 20th century. We have description of these tones in terms of tone numbers. One is that the falling tone now seems best described as a mostly level high pitch. Another is that a native cue to recognising the rising tone is that it initially dips to the lowest pitch of any tone, a feature not recognised in the old description as '24', contrasting with '41' for the falling tone. Li records that the high tone as having an allophonic distinction of '453' in unchecked syllables and '55' in checked syllables.

The phonetic development of Thai has probably been complicated by literacy. Thai is full of loans from Sanskrit with a fair few from Pali, and, less recognisably, Khmer. Quite a few words from Pali seem to have been Sanskritised, and the authoritative dictionary gave up on attempting to say whether a word came form Pali or came from Sanskrit. I get the impression that early elites were fluent in Khmer. It's difficult to know whether tone-spreading is a borrowing from Khmer 'consonant governance'. It is not unknown for Thai and Khmer to disagree on whether it happens - it has been lexicalised in both languages.

Thai has also recovered the category of sesquisyllables from Khmer. For Thai, these also known as words with impure consonant clusters. They contain an anaptyctic vowel, and some Thais have told me that this vowel is phonetically distinct from the vowel of a separate syllable.

[bradrn]

Many thanks for this!

The ideal would be to get a listing of sound changes which is comprehensive enough to evolve at least a few sample words from an earlier form to modern Thai. Do you know of any references which might help me in this?

[jal]

with a fair view from Pali,

I suppose you mean "few"?


JAL

[Richard W]

with a fair view from Pali,

I suppose you mean "few"?

Yes. 'few', not 'view'. Now corrected in the OP. 'Fair few' actually means 'close to many'.

[bradrn]

The ideal would be to get a listing of sound changes which is comprehensive enough to evolve at least a few sample words from an earlier form to modern Thai. Do you know of any references which might help me in this?

Any response to this?

(I don’t normally bump questions like this, but it’s sort of key for what I want to do. Though I guess a reference isn’t strictly necessary, if you’re willing to write out all the details yourself.)

[Richard W]

Any response to this?

I'll try populating a table piecemeal. Unfortunately, it will take a little while to verify the the word's existence in Sukhothai inscriptions, especially as I'm not firing on all four cylinders at the moment.

For the transcriptions I've used the following system of phonetic vowel components:

	Front	Back unrounded	Back rounded
High	i	ɯ	u
Mid	e	ɤ	o
Low	ɛ	a	ɔ

In modern Thai, all 9 vowel sounds occur both short and long. For the front vowels, the long vowels are significantly higher than the short vowels, but the difference is trifling for back round vowels - using English [ʊ] for /u/ risks being misinterpreted as /o/. For [a], the difference is usually short [ɐ] versus long [ä], a distinction preserved in Thai dialects that do not contrast vowel length.

I write the semivowel codas as /i/ and /u/ rather than /j/ and /w/. This makes the Sukhothai-era diphthong /aɯ/, now merɡed with /ai/ in most Tai dialects, less peculiar. There are three diphthongs in Thai - /ia/, /ɯa/ and /ua/. For Thai as normally described, there is only a length contrast if one interprets a final glottal stop as an emanation of a short vowel, rather than seeing the shortness of the diphthong as being conditioned by the (underlying) final glottal stop. Some Thais claim there is an (unwritten) length difference in indubitably closed syllables, and Northern Thai and Lao have mechanisms for marking the short diphthong in such syllables.

Some sources use different symbols for the non-low back unrounded vowels, and some sources may describe them as central. The official transliteration once used the Vietnamese letters with horns, and these horns are sometimes written as apostrophes.

No.	Gedney cell	Siamese	ɣloss	Modern transcription	Sukhothaii transcription	PT?	ST?	Remarks
1.	A1	หู	ear	huu	huu
2.	A1	ขา	leɡ	kʰaa
3.	A1	หัว	head	hua	hua
4.	A2	ปี	year	pii	pii
5.	A2	ตา	eye	taa	taa
6.	A2	กิน	eat	kin	kin
7.	A3	บิน	to fly	bin	ʔbin
8.	A3	แดง	red	dɛɛŋ	ʔdɛɛŋ
9.	A3	ดาว	star	daau	ʔdaau
10.	A4	มือ	hand	mɯɯ	mɯɯ
11.	A4	งู	snake	ŋuu	ŋuu
12.	A4	นา	field	naa	naa
13.	B1	ไข่	egg	kʰai
14.	B1	ผ่า	to chop	pʰa	pʰaa
15.	B1	เข่า	knee	kʰaau
16.	B2	ป่า	wood, forest	paa	paa
17.	B2	ไก่	chicken	kai	kai
18.	B2	แก่	old	kɛɛ	kɛɛ
19.	B3	บ่า	shoulder	baa	ʔbaa
20.	B3	บ่าว	servant	baau	ʔbaau
21.	B3	ด่า	to scold	daa	ʔdaa
22.	B4	พ่อ	father	pʰɔɔ	bɔɔ
23.	B4	แม่	mother	mɛɛ	mɛɛ
24.	B4	ไร่	dry field	rai	rai
25.	C1	ข้าว	rice	khaau
26.	C1	เสื้อ	cloth	sɯa	sɯa
27.	C1	ห้า	five	haa	haa
28.	C2	ป้า	aunt	paa	paaa
29.	C2	กล้า	rice seedlinɡ	klaa	klaa
30.	C2	ต้ม	to boi	tom	tom
31.	C3	บ้า	mad	baa	ʔbaa
32.	C3	บ้าน	villaɡe	baan	ʔbaan
33.	C3	อ้า	open	ʔaa	ʔaa
34.	C4	น้ำ	aunt	naa	naa
35.	C4	น้อง	younɡer siblinɡ	nɔɔŋ	nɔɔŋ
36.	C4	ม้า	horse	maa	maa
37.	D1L	ขาด		khaat
38.	D1L	เหงือก	ɡum (anatomy)	ŋɯak	ŋ̥ɯak
39.	D1L	หาม	to carry	haap	haap
40.	D2L	ปอด	lung	pɔɔt	pɔɔt
41.	D2L
42.	D2L
43.	D3L
44.	D3L
45.	D3L
46.	D4L
47.	D4L
48.	D4L
49.	D1S
50.	D1S
51.	D1S
52.	D2S
53.	D2S
54.	D2S
55.	D3S
56.	D3S
57.	D3S
58.	D4S
59.	D4S
60.	D4S
61.

[bradrn]

Thanks so much for endeavouring to put this together! I’m slightly surprised this isn’t described already… but it shouldn’t be too hard to extract some sound changes from this.

(Also, what’s a ‘Gedney cell’?)

[Richard W]

Thanks so much for endeavouring to put this together! I’m slightly surprised this isn’t described already… but it shouldn’t be too hard to extract some sound changes from this.

I did give a reference to Li Fang-Kuei's A Handbook of Comparative Tai (1977), which for some reason seems to be out of print. I'm surprised the publishers don;t run off copies from the typescript master from time to time - I've seen exorbitant prices asked for copies in good condition. It's a paperback, and I suspect most copies are worn out. I picked my copy up, new, at an English language bookshop on Sukhumvit Road in Bangkok, fairly close to Soi Cowboy. Most of their stuff was coffee-book style, so perhaps they'd stocked it by mistake. (Sukhumvit Road has been described as a 'Farang-infested slice of Bangkok.) The post has disappeared, so perhaps I didn't actually post it. Most of what I'm saying comes from there, but obviously not the stuff about 21st century tones.

(Also, what’s a ‘Gedney cell’?)

A cell in the Gedney box, or to be more specific, Gedney's tone-box, a very useful concept for talking about or analysing the development of Tai-Kadai tones. It's a bit flexible - the vowel lengths for checked syllables are those at some relevant time, which varies from dialect to dialect, and where what Li reconstructs as Proto-Tai *tr became /tʰ/, it may be classified as an aspirated onset. Pitayapon reconstructs Li's clusters as sesquisyllable's onsets, and reconstructs Li's *tʰ as *tr. The Great Tai Consonant Shift spread through the region as a wave, and hasn't reached some parts of Vietnam.

[Richard W]

… but it shouldn’t be too hard to extract some sound changes from this.

I can give you the onset changes:

Unchangedː

Symbol type	Thai spelling	Sukhothai	Modern	Remark
letter	ก	k	k
letter	ข	kʰ	kʰ
letter	ง	ŋ	ŋ
letter	จ	t͜ɕ	t͜ɕ
letter	ฉ	t͜ɕʰ	t͜ɕʰ	A change to [ɕ] is reported to be in progress.
letter	ต	t	t
letter	ถ	tʰ	tʰ
letter	น	n	n
letter	ป	p	p
letter	ผ	pʰ	pʰ
letter	ฝ	f	f
letter	ม	m	m
letter	ย	j	j
letter	ร	r	r
letter	ล	l	l
letter	ว	w	w
letter	ส	s	s
letter	อ	ʔ	ʔ

Devoiced with aspiration

Symbol type	Thai spelling	Sukhothai	Modern	Remark
letter	ค	ɡ	kʰ
letter	ช	d͜ʑ	t͜ɕʰ	A change to [ɕ] is reported to be in progress.
letter	ท	d	tʰ
letter	พ	b	pʰ

Neglecting tone, it seems that the change was, for example, [ɡam] > [ɡa̤m]> [ka̤m] > [kʰam].

Devoiced

Symbol type	Thai spelling	Sukhothai	Modern	Remark
letter	ซ	z	s
letter	ฟ	v	f

Voiced:

Symbol type	Thai spelling	Sukhothai	Modern	Remark
digraph	หง	ŋ̊	ŋ
digraph	หญ	ɲ̊	j
digraph	หน	n̥	n
digraph	หม	m̥	m

Deglottalised

Symbol type	Thai spelling	Sukhothai	Modern	Remark
letter	ด	ʔd	d
letter	บ	ʔb	b
digraph	อย	ʔj	j	Apart from four words, the modern spelling is in ย- or หย-.

Miscellanous

Symbol type	Thai spelling	Sukhothai	Modern	Remark
letter	ฃ	x	kʰ
letter	ฅ	ɣ	kʰ
letter	ญ	ɲ	j
cluster	มล	ml	m or l or mal The last is claimed to be a spelling pronunciation	CAUTION: This may be an overbold synthesis of mine.

For onset clusters other than /ml/ or /sr/, the change is restricted to the initial consonant of the cluster. It seems that Thai poetry has a cluster /sr/ not reported by foreign analysts.

[bradrn]

Li Fang-Kuei's A Handbook of Comparative Tai (1977)

It appears that my university library has a copy. I’ll have a look at it.

The Great Tai Consonant Shift

What’s this?

… but it shouldn’t be too hard to extract some sound changes from this.

I can give you the onset changes:

Thank you!

On the other hand, having a look at your earlier table, it appears that the rhymes have undergone minimal change. Is that really correct?

[bradrn]

Reading through Li now. Starting with chapter 2, the tone changes, for Thai they are:

A1 → 24 / after aspirated stops and voiceless continuants
A1 → 33 / otherwise
A2 → 33
B1 → 22
B2 → 41
C1 → 41
C2 → 453 (a development →55 is also mentioned but no examples are given)
D1S → 22
D1L → 22
D2S → 55
D2L → 41

Plus some irregular tone changes which I don’t think I need to worry about.

[bradrn]

Continuing to read… here’s the labial clusters (which are all I have time to transcribe right now):

pr → t (but modern /pr/ got reintroduced from loanwords)
pʰl, pʰr → pʰ
ʔbl, ʔbr → d
ml → m, l, mal (no conditioning factor reported, appears sporadic; already mentioned by Richard)
vr → pʰr
vl → tʰ (only one example, unreliable)
fr → pʰr (only one example, unreliable)

While pl, bl, br remained clusters in the modern language.

[bradrn]

Continuing on… dental clusters:

tl, tr → t
tʰl → tʰ
tʰr → h (plus one instance of /r/)
dl → l
dr → r
ʔdl, ʔdr → ʔd → d (plus one instance of /r/)
nl, nr → n

Velar clusters:

kl → kl
kr → kr, kl (no obvious conditioning factor)
kʰl, kʰr → kʰ
gl → kʰl
gr → kʰr
ŋl, ŋr → ŋ (only one example)
xr → h

Labiovelars:

kw → kw
kʰw → kʰw
ŋw → ŋ, w (‘but the rules are difficult to formulate’)
xw → kʰw
ɣw → kʰw

Liquids, sibilants, velars, laryngeals: all as mentioned by Richard, except also l̥ → l, r̥ → h (plus one instance of ŋ→h)

[bradrn]

The vowel system is more intricate, but here’s the monophthong developments, as best as I can transcribe them:

(NB. I believe the symbol ⟨ɩ̈⟩ corresponds to IPA /ɨ/, unless it’s /ɯ/)

V → Vː / in open syllables
{a, ɛ, ɔ} → {aː, ɛː, ɔː} / always
u̯i → iː (dubious)
ɩ̈ə, ɩ̈u → ɩ̈ː
uə, uo, ɩ̯̈u → uː
i̯o → u
e → o / _ {m,p}
i̯e → e
ə → a
u̯ə → ə → a (dubious, only one example)
u̯ɩ̈ → o
iɛ, i̯ɛ → ɛː (apparently these two protoforms must be distinguished as other languages show different correspondences)
ɩ̯̈a, u̯a → aː
uɔ → ɔː
ɩ̈ɔ, ɩ̯̈ɔ → ɔː

[bradrn]

Opening diphthongs:

ie → ia
ɩ̈e → ɩ̈a
ɩ̈o → ɩ̈a
ue → ua
uɩ̈ → ua

(Also note from the previous post: ɩ̈ə, ɩ̈u → ɩ̈ː, iɛ → ɛː, uɔ,ɩ̈ɔ → ɔː.)

Closing diphthongs:

əi → ai
ei → iː
ɛi → ai
ai → aːi
ɩ̯̈ai → aːi
ɔi → ɔːi
u̯əi → ɔːi (uncertain)
əɩ̈ → ai (late merger, not reflected in orthography)
eɩ̈, ɛɩ̈ → ai
oɩ̈ → ai
əu → au
eu → uː
ou → au
i̯əu → au (but note irregular kuu ‘I’)
au → aːu
ɛu → ɛːu (few examples)
ɩ̈i → əi (only one example)

(Plus a diphthong of uncertain reconstruction which goes to iː in Thai.)

Some of the above are probably subsumed by the previously detailed changes to single vowels, but I haven’t gone through and worked out which.

Triphthongs:

iəu → iau (or possibly *ieu)
uəi → uai (uncertain reconstruction)

And… this is it for the rules, I believe! It shouldn’t be too hard for me to convert this into Brassica format (and thence get it into the Index Diachronica), though I don’t have time to do it just right now.

The bigger annoyance could be example protoforms to evolve. I think there’s sufficient information in the book for me to infer the protoforms for their various example words, but it would be better if I could get something more authoritative.

[Richard W]

c
The Great Tai Consonant Shift

What’s this?

The.massive set of phonation shifts affecting initial occlusives.

On the other hand,, having a look at your earlier table, it appears that the rhymes have undergone minimal change. Is that really correct?

Apart from length, Thai seems to have the stablest Tai rimes.

[Richard W]

Reading through Li now. Starting with chapter 2, the tone changes, for Thai they are:

A1 → 24 / after aspirated stops and voiceless continuants
A1 → 33 / otherwise
A2 → 33

He distinguishes the two A1’s as A1H and A1M. That’s why having 20 labels is generally more convenient if flitting from dialect to dialect.

C2 → 453 (a development →55 is also mentioned but no examples are given)

See his discussions of Siamese in the early sections of the work. I think he has a Siamese index. Marvin Brown notes that the C4 pronouns spelt as A1 are 55 as though checked.

[Richard W]

Continuing to read… here’s the labial clusters (which are all I have time to transcribe right now):

pr → t (but modern /pr/ got reintroduced from loanwords)
pʰl, pʰr → pʰ
ʔbl, ʔbr → d
ml → m, l, mal (no conditioning factor reported, appears sporadic; already mentioned by Richard)
vr → pʰr
vl → tʰ (only one example, unreliable)
fr → pʰr (only one example, unreliable)

While pl, bl, br remained clusters in the modern language.

Here you‘ e definitely changed the starting point from Sukhothai Thai to Li’s Proto-Tai. Several of these clusters are now seen as sesquisyllabic onsets.

[bradrn]

Continuing to read… here’s the labial clusters (which are all I have time to transcribe right now):

pr → t (but modern /pr/ got reintroduced from loanwords)
pʰl, pʰr → pʰ
ʔbl, ʔbr → d
ml → m, l, mal (no conditioning factor reported, appears sporadic; already mentioned by Richard)
vr → pʰr
vl → tʰ (only one example, unreliable)
fr → pʰr (only one example, unreliable)

While pl, bl, br remained clusters in the modern language.

Here you‘ e definitely changed the starting point from Sukhothai Thai to Li’s Proto-Tai. Several of these clusters are now seen as sesquisyllabic onsets.

Well, what reason do I have to start at Sukhothai if the source starts from Proto-Tai?

[Richard W]

Continuing on… dental clusters:
…

Velar clusters:

kl → kl

Probably not clusters. And one of them seems to depend on extra-Tai cognates.

Also, a uvular series is now recognised. Some of the evidence is presented in the chapter on velars.