Zompist Bboard Again

Richard W wrote: ↑Mon Oct 07, 2024 1:49 am

bradrn wrote: ↑Wed Oct 02, 2024 9:24 pm Continuing on… dental clusters:
…

Velar clusters:

kl → kl

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

If not clusters, then what are these? And which depends on extra-Tai cognates?

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

Where?

bradrn wrote: ↑Wed Oct 02, 2024 9:24 pm Liquids, sibilants, velars, laryngeals: all as mentioned by Richard, except also l̥ → l, r̥ → h (plus one instance of ŋ→h)

Resorting to XSAMPA, /l_0/ and /l/ merge and /w_0/ and /w/ merge late in, I think, all dialects.

Li’s.vowel reconstruction strikes me as less well-founded than his reconstruction of onsets. He himself complained of insufficient data. It also looks as though the change from sesquisyllabic to monosyllabic happened within the individual branches. Li’s correspondences may be sound, but the reconstruction is something of an artlang.

Richard W wrote: ↑Mon Oct 07, 2024 6:52 am Li’s correspondences may be sound, but the reconstruction is something of an artlang.

It’s still something I can use which is reasonably realistic and supported by real data. That’s all I ask for, for this.

Real data is why I proposed starting with Sukhothai Thai, though solid suprasegmentals take a little longer tii hi o kick in. Unless Siamese-Chinese dictionaries can help.

Richard W wrote: ↑Mon Oct 07, 2024 7:22 am Real data is why I proposed starting with Sukhothai Thai, though solid suprasegmentals take a little longer tii hi o kick in. Unless Siamese-Chinese dictionaries can help.

But the suprasegmentals are the point of all this, though. If I start with Proto-Tai then I can simulate the splitting and merging of the old A/B/C/D tones to their modern Siamese values, plus there’s the changes to the vowel qualities underneath those. Sukhothai postdates those changes, to my (limited) understanding.

bradrn wrote: ↑Mon Oct 07, 2024 8:59 am But the suprasegmentals are the point of all this, though. If I start with Proto-Tai then I can simulate the splitting and merging of the old A/B/C/D tones to their modern Siamese values, plus there’s the changes to the vowel qualities underneath those. Sukhothai postdates those changes, to my (limited) understanding.

My understanding is quite different. All phonetic Tai writing systems postdate the original formation of the original 3-way tone system - the presumed conditioning finals are nowhere to be seen. However, the conditioning onsets for the tone splits can be seen in the writing system of Sukhothai, and the contrast is preserved in Viet Tai. What is noteworthy about careful inscriptions is that they also show the A/B/C contrast, which has a tendency to get dropped. Thus for Sukhothai tones on live syllables, the initial consonant letter is the only additional information required.

There are a few length changes which might be included. Final glottal stop shortens the preceding vowel, which becomes relevant as length becomes phonemic on [E] and [O]. There is a good percentage guess for modern Thai that tone B shortens these vowels in closed syllables. It looks like a slow sound change. The length marker on these sounds only came into use after the sack of Ayutthaya. It’s quite possible that /ee/.and /oo/ were foreign phonemes when the inscriptions were incised; sometimes the letter for the latter was used for /OO/.

Leaving aside for now the matter of inadequacies in Li’s reconstruction, I collected some words for which he gives sufficient information for me to infer a reconstruction. Here’s what I have so far:

*ʔbɩ̈n → bon (A1) ‘above’
*drɩ̈o → rɩ̈a (A2) ‘boat’
*ňuŋ → juŋ (A2) ‘broom; rice stubble’
*siu → siu (B1) ‘chisel’
*bɔ → phɔɔ (B2) ‘father’
*ha → haa (C1) ‘five’
*rɔi → rɔɔi (C2) ‘string’
*ʔi̯et → ʔet (D1S) ‘one’
*pruɔk → tɔɔk (D1L) ‘bamboo strip’
*gu̯ot → khot (D2S) ‘bent’
*muək → muuk (D2L) ‘mucus’

I‘ve just read a plausible argument that the word for father had tone C, and that SW Tai switched to B to match mE B ‘mother’. (Sorry for the XSAMPA. I’ve been stuck in hospital for a week with an iPhone for communication , and couldn’t get automatic password recovery to work for me.)

Richard W wrote: ↑Tue Oct 08, 2024 6:58 am I‘ve just read a plausible argument that the word for father had tone C, and that SW Tai switched to B to match mE B ‘mother’.

Ah, interesting. I’ll probably keep on looking for more examples anyway.

(Sorry for the XSAMPA. I’ve been stuck in hospital for a week with an iPhone for communication , and couldn’t get automatic password recovery to work for me.)

Oof. Sorry to hear that — hope things improve!

Richard W wrote: ↑Tue Oct 08, 2024 6:58 am (Sorry for the XSAMPA. I’ve been stuck in hospital for a week with an iPhone for communication , and couldn’t get automatic password recovery to work for me.)

That's gotta suck. Hope things get better.

bradrn wrote: ↑Mon Oct 07, 2024 1:25 am Well, what reason do I have to start at Sukhothai if the source starts from Proto-Tai?

The Sukhothai inscriptions are real: Proto-Tai is inferred and disputed. But perhaps using Li’s recnstruction is as valid as using Brugmannian PIE for PIE.

Richard W wrote: ↑Tue Oct 08, 2024 1:10 pm

bradrn wrote: ↑Mon Oct 07, 2024 1:25 am Well, what reason do I have to start at Sukhothai if the source starts from Proto-Tai?

The Sukhothai inscriptions are real: Proto-Tai is inferred and disputed. But perhaps using Li’s recnstruction is as valid as using Brugmannian PIE for PIE.

Yep, pretty much. It may be a bit outdated, but at least it involves natlangs and describes a plausible set of sound changes. And honestly, by this point my main priority is to get Brassica v1.0.0 finished and released ASAP.

(When we get around to putting this stuff in the new Index Diachronica, then we can have these discussions. But for a mere example file for a sound change applier, I think it suffices.)

Looks like I forgot one: mw → m.

I’ve transferred these changes to Brassica. They all seem self-consistent, in that my reconstructions produce the results I expect them to. But when I double-checked my words with SEAlang’s dictionary, it seems that a couple (juŋ and ʔet) are either very rare or non-existent, so I’ll continue looking for replacements and further examples.

Most importantly, the exercise revealed some nasty edge cases in Brassica which I hadn’t previously noticed. It also revealed some non-obvious contortions which are required when dealing with two tone systems simultaneously. But the edge cases have now been patched up, and once I finish documenting all of this, Brassica 1.0.0 should be ready for release!

bradrn wrote: ↑Thu Oct 10, 2024 10:20 am But when I double-checked my words with SEAlang’s dictionary, it seems that a couple (juŋ and ʔet) are either very rare or non-existent, so I’ll continue looking for replacements and further examples.

เอ็ด [ʔet] 'one' is very much there; it's just that SEALang doesn't recognise an initial glottal stop as phonemic.

For the 'stubble' morpheme you should have tried looking up the full Thai word ยุงปัด [juŋpat], but though it's in the authoritative Thai-Thai dictionary, it's certainly rare. However, as this standard Siamese form is grossly irregular in its evolution, I suggest you use Li Section 9.6 no. 14 [juŋ] 'mosquito' instead of no. 13.

(As you can see from my character sets, I'm now back home with my desktop PC and my books.)

bradrn wrote: ↑Thu Oct 10, 2024 10:20 am Most importantly, the exercise revealed some nasty edge cases in Brassica which I hadn’t previously noticed. It also revealed some non-obvious contortions which are required when dealing with two tone systems simultaneously.

What phonemic system did you use for the three Proto-Tai tones? Evidence for or against creakiness and glottal constriction for older Thai may be hard to come by; glottal constriction is under-recorded for modern Tai dialects. It's surprising that we have evidence in stone for glottal co-articulation of final stops in the Sukhothai inscriptions.

Switching systems is quite common, e.g. from Latin vowel length to Romance vowel quality, and in much of English, quantity and quality are concurrent. I can't remember how significant quality is for native interpretation of Thai front vowels; as an Englishman, I perhaps surprisingly rely on the quality differences.

Richard W wrote: ↑Thu Oct 10, 2024 5:40 pm (As you can see from my character sets, I'm now back home with my desktop PC and my books.)

Very glad to hear this! Hope you’re doing better.

bradrn wrote: ↑Thu Oct 10, 2024 10:20 am But when I double-checked my words with SEAlang’s dictionary, it seems that a couple (juŋ and ʔet) are either very rare or non-existent, so I’ll continue looking for replacements and further examples.

เอ็ด [ʔet] 'one' is very much there; it's just that SEALang doesn't recognise an initial glottal stop as phonemic.

Thanks!

For the 'stubble' morpheme you should have tried looking up the full Thai word ยุงปัด [juŋpat], but though it's in the authoritative Thai-Thai dictionary, it's certainly rare. However, as this standard Siamese form is grossly irregular in its evolution, I suggest you use Li Section 9.6 no. 14 [juŋ] 'mosquito' instead of no. 13.

OK, I’ll have a look.

bradrn wrote: ↑Thu Oct 10, 2024 10:20 am Most importantly, the exercise revealed some nasty edge cases in Brassica which I hadn’t previously noticed. It also revealed some non-obvious contortions which are required when dealing with two tone systems simultaneously.

What phonemic system did you use for the three Proto-Tai tones?

Same as in Thai script — A/D, B and C. The annoyance comes from the simple fact that I do need to switch tonal systems. Most of the time switching systems is fine, but in this case it means I need to juggle two separate sets of tonal correspondences. (For instance, ⟨a⟩ could refer to {a, aᴮ, aᶜ}, or it could be {a, ā, à, â, á, ǎ.) It all seems to have turned out reasonably comprehensible, though.

Final set of input and output words:

ʔbu̯ɩ̈n → bōn [บน ‘above’]
ʔblɩ̈en → dɩ̄an [เดือน ‘moon, month’]
drɩ̈o → rɩ̄a [เรือ ‘boat’]
ɣɩ̈ən → khɩ̄ɩ̈n [คืน ‘night’]
ňuŋ → jūŋ [ยุง ‘mosquito’]
siuᴮ → sìu [สิ่ว ‘chisel’]
bɔᴮ → phɔ̂ɔ [พ่อ ‘father’]
rueᴮ → rûa [รั่ว ‘leak’]
haᶜ → hâa [ห้า ‘five’]
rɔiᶜ → rɔ́ɔi [ร้อย ‘string’]
ʔi̯et → ʔēt [เอ็ด ‘one’]
pruɔk → tɔ̀ɔk [ตอก ‘bamboo strip’]
dlək → lāk [ลัก ‘steal’]
muək → mūuk [มูก ‘mucus’]

Richard, could you double-check that the Thai script is correct? Then this should be ready to add to Brassica’s set of examples.

(Yes, I know the transcription is somewhat eccentric, but honestly I can’t find one that isn’t, so I thought I might as well stick with Li’s.)

Richard W wrote: ↑Thu Oct 10, 2024 5:40 pm Switching systems is quite common, e.g. from Latin vowel length to Romance vowel quality, and in much of English, quantity and quality are concurrent. I can't remember how significant quality is for native interpretation of Thai front vowels; as an Englishman, I perhaps surprisingly rely on the quality differences.

The switch from quantity to quality is complete in much of NAE, but then new quantity has been innovated to replace the old quantity (and which in some dialects like my own is arguably phonemic in some more shallow phonemic analyses).

Hope you don't mind me jumping in. I might be missing some details from earlier posts

bradrn wrote: ↑Thu Oct 10, 2024 8:42 pm Final set of input and output words:

ʔbu̯ɩ̈n → bōn [บน ‘above’]
ʔblɩ̈en → dɩ̄an [เดือน ‘moon, month’]
drɩ̈o → rɩ̄a [เรือ ‘boat’]
ɣɩ̈ən → khɩ̄ɩ̈n [คืน ‘night’]
ňuŋ → jūŋ [ยุง ‘mosquito’]
siuᴮ → sìu [สิ่ว ‘chisel’]
bɔᴮ → phɔ̂ɔ [พ่อ ‘father’]
rueᴮ → rûa [รั่ว ‘leak’]
haᶜ → hâa [ห้า ‘five’]
rɔiᶜ → rɔ́ɔi [ร้อย ‘string’]
ʔi̯et → ʔēt [เอ็ด ‘one’]
pruɔk → tɔ̀ɔk [ตอก ‘bamboo strip’]
dlək → lāk [ลัก ‘steal’]
muək → mūuk [มูก ‘mucus’]

Richard, could you double-check that the Thai script is correct? Then this should be ready to add to Brassica’s set of examples.

The Thai script is correct, however some of the transcribed output words do not have the tones I would expect (from learning modern Thai). I'm assuming the tones of modern Thai are what's intended, and mid tone does not occur in stop-final syllables. What are you verifying the outputs against?

ʔi̯et → ʔēt [เอ็ด ‘one’] -- should be ʔèt (low tone)
dlək → lāk [ลัก ‘steal’] -- should be lák (high tone)
muək → mūuk [มูก ‘mucus’] -- should be mûuk (falling tone)