Zompist Bboard Again

Ser wrote: ↑Sun Oct 11, 2020 7:07 pm

aliensdrinktea wrote: ↑Sun Oct 11, 2020 6:30 pmThe addition of /θ/ was a nice touch to distinguish Spanish's phonology from the other Romance languages, but <z> and <c> are completely nonsensical choices for it. <j> and <g> for /x/ are hardly better. Sure, there may be historical reasons behind the spelling, but there's a point where orthographic changes are needed to reflect the sound change. Spanish has long passed that point.

I like the way someone who used to come here in the past liked to put it:

"Hi, I'm the Spanish language, and I'ma use <j> for [x], because fuck everyone else."

…which assignment has since become standard in American languages, along with ⟨x⟩ /ʃ/, except for the languages using ⟨x⟩ /x/ or ⟨j⟩ /h/, causing much confusion to everyone involved…

bradrn wrote: ↑Thu Sep 17, 2020 7:00 am My new favourite misplaced phoneme: a pharyngeal fricative /ħ/… found in Teiwa, a Papuan language of Pantar in eastern Indonesia. This phoneme is strange enough that the grammar felt obliged to single it out specifically as being unusual.

Not too unusual - some other Papuan languages are reported to have /ħ/, although in these cases other sources for the same language frequently have /h/ instead. More out of place is /ʁ~ʕ/ in Kusunda, spoken in Nepal... then again, uvulars and even epiglottals aren't uncommon in Tibetic.

Other cases of arguably out-of-place phonemes:
- /ʕ/ in Alyutor
- Tibetic epiglottals, I guess
- strident vowels (otherwise known mostly from Khoisan) in a few Sino-Tibetan languages, especially Bai (imo Tangut grade 2 could've been strident as well)
- /ɴ/ in Mapos Buang (Papuan) and maybe Kusunda and Greenlandic... afaik it's otherwise only attested in Sino-Tibetan
- back unrounded vowels in Celtic

Nortaneous wrote: ↑Sun Oct 11, 2020 11:12 pm

bradrn wrote: ↑Thu Sep 17, 2020 7:00 am My new favourite misplaced phoneme: a pharyngeal fricative /ħ/… found in Teiwa, a Papuan language of Pantar in eastern Indonesia. This phoneme is strange enough that the grammar felt obliged to single it out specifically as being unusual.

Not too unusual - some other Papuan languages are reported to have /ħ/

Which ones?

Other cases of arguably out-of-place phonemes:
- /ʕ/ in Alyutor

Not too strange, given that the Pacific NW is practically next door!

- Tibetic epiglottals, I guess

All right, you have to be joking here… source, please?

- strident vowels (otherwise known mostly from Khoisan) in a few Sino-Tibetan languages, especially Bai (imo Tangut grade 2 could've been strident as well)

And this is just unbelievable. Again: source?

- /ɴ/ in Mapos Buang (Papuan)

Austronesian, surely… but that’s not too strange, given that Proto-Oceanic had /q/.

- back unrounded vowels in Celtic

Celtic in general has a whole bunch of misplaced phonemes. Irish, for example, has a consonant system which seems more Micronesian than anything else. And Welsh has its voiceless sonorants, more commonly found in Sino-Tibetan. Though I must admit that I didn’t know about the back unrounded vowels in Irish (though it looks like they’re only in Ulster dialect?).

bradrn wrote: ↑Sun Oct 11, 2020 11:28 pm

Nortaneous wrote: ↑Sun Oct 11, 2020 11:12 pm Not too unusual - some other Papuan languages are reported to have /ħ/

Which ones?

Kobon, Huli, Managalasi

- Tibetic epiglottals, I guess

All right, you have to be joking here… source, please?

Pick five papers by Hiroyuki Suzuki and you'll find one - he does a lot of fieldwork but unfortunately tends to publish in moon runes. Here's one in English, but unfortunately it's on dialect comparison rather than phonology; he's also written a phonology of Sharkhog, but it's in Chinese or something. (Japhug /ʁ/ is apparently an epiglottal in syllable-final position.)

- strident vowels (otherwise known mostly from Khoisan) in a few Sino-Tibetan languages, especially Bai (imo Tangut grade 2 could've been strident as well)

And this is just unbelievable. Again: source?

Uh, like, the Wikipedia article on strident voice, man. They don't have a source but it's probably Edmondson & Esling 2006, which reveals that I might be wrong anyway; don't have the time at the moment to reread it and see if they report aryepiglottal trilling from Bor Dinka etc.

(actually don't strident vowels have aryepiglottal trilling *and* pharyngealization? much to be fixed on la wik)

Either way, Sino-Tibetan is very phonologically weird! Suzuki insists that there's such a thing as ɧ and even ɧʰ (aspirated fricatives aren't too surprising of course), Japhug has the tɕ/c/kj contrast, and velarized or uvularized vowels are everywhere. (imo Tangut grade 1 was probably uvularized) Extremely large segmental inventories are common, including contrasts like /xʰ ʰx ʰxʰ/ (dGudzong Tibetan, probably), although here preaspiration is probably better analyzed as a consonant cluster beginning with /h/.

The whole region is underrated as a source of weirdness. Wikipedia again:

Taa has at least 58 consonants, 31 vowels, and four tones (Traill 1985, 1994 on East ǃXoon), or at least 87 consonants, 20 vowels, and two tones (DoBeS 2008 on West ǃXoon), by many counts the most of any known language if non-oral vowel qualities are counted as different from corresponding oral vowels.

The most what? Vowels? There's a dialect of Chong (Austroasiatic) with 56, not counting diphthongs - there's a series with no marked phonation, a creaky series, a breathy series, and a both series, and the Sino-Tibetan language 'Bo-skad has 52: 12 base qualities plus nasalization, creaky voice, and both. (What's the language with the most vocalic base qualities? If it's not Kensiu, it could be one of those Tibetic dialects that Suzuki reports a massive inventory for, but presumably these can be analyzed down somewhat.) Consonants? Under the maximal defensible analysis (prenasalized and preaspirated initials are units), Lhagang Choyu has at least 92.

- /ɴ/ in Mapos Buang (Papuan)

Austronesian, surely… but that’s not too strange, given that Proto-Oceanic had /q/.

A little strange given that uvular nasals are extremely rare in general. But uvulars are surprisingly common in New Guinea!

Celtic in general has a whole bunch of misplaced phonemes. Irish, for example, has a consonant system which seems more Micronesian than anything else. And Welsh has its voiceless sonorants, more commonly found in Sino-Tibetan. Though I must admit that I didn’t know about the back unrounded vowels in Irish (though it looks like they’re only in Ulster dialect?).

I was thinking of Scottish Gaelic and Welsh (where they're written as central, but details)

The Irish consonant system is much more Slavic than it is Micronesian; it's just that convention says you write things with ˠ.

Nortaneous wrote: ↑Mon Oct 12, 2020 3:46 am The most what? Vowels? There's a dialect of Chong (Austroasiatic) with 56, not counting diphthongs - there's a series with no marked phonation, a creaky series, a breathy series, and a both series, and the Sino-Tibetan language 'Bo-skad has 52: 12 base qualities plus nasalization, creaky voice, and both.

I think you overlooked Chong's tense open-mid vowels, taking the tally to 60. However, if register can multiply vowels up, surely tone can too (tone can include register effects), so by the same rules Standard Thai has 9 qualities × 2 lengths × 5 tones = 90 non-diphthongal vowels. (Thai length and quality interact significantly for front vowels.) For historical reasons, some of the 90 possibilities may be missing.

Richard W wrote: ↑Mon Oct 12, 2020 9:05 am

Nortaneous wrote: ↑Mon Oct 12, 2020 3:46 am The most what? Vowels? There's a dialect of Chong (Austroasiatic) with 56, not counting diphthongs - there's a series with no marked phonation, a creaky series, a breathy series, and a both series, and the Sino-Tibetan language 'Bo-skad has 52: 12 base qualities plus nasalization, creaky voice, and both.

I think you overlooked Chong's tense open-mid vowels, taking the tally to 60. However, if register can multiply vowels up, surely tone can too (tone can include register effects), so by the same rules Standard Thai has 9 qualities × 2 lengths × 5 tones = 90 non-diphthongal vowels. (Thai length and quality interact significantly for front vowels.) For historical reasons, some of the 90 possibilities may be missing.

If we're going to compete, I would prefer to just compare vowel positions on the traditional IPA chart, not tones, coarticulations etc. If we do go all the way though, Taa could up the ante again by also including its own length contrasts, as well as nasalization and pharyngealization.Just to give an idea of how insanely huge it is, I quote from Wikipedia:

A long, glottalized, murmured, nasalized o with falling tone is written ôʼhõ.

.

Richard W wrote: ↑Mon Oct 12, 2020 9:05 am

Nortaneous wrote: ↑Mon Oct 12, 2020 3:46 am The most what? Vowels? There's a dialect of Chong (Austroasiatic) with 56, not counting diphthongs - there's a series with no marked phonation, a creaky series, a breathy series, and a both series, and the Sino-Tibetan language 'Bo-skad has 52: 12 base qualities plus nasalization, creaky voice, and both.

I think you overlooked Chong's tense open-mid vowels, taking the tally to 60. However, if register can multiply vowels up, surely tone can too (tone can include register effects), so by the same rules Standard Thai has 9 qualities × 2 lengths × 5 tones = 90 non-diphthongal vowels. (Thai length and quality interact significantly for front vowels.) For historical reasons, some of the 90 possibilities may be missing.I

In languages where phonation isn't an epiphenomenon of tone, phonation is generally on a different phonological level from tone, isn't it?

Also, you're right, it's 60 - I read the 8 in the database as a 6 (really need to clean my screen) and /ɛe ɛ̰ḛ/ are phonetic variants of underlying /ɛː ɛ̰ː/ so probably shouldn't count as diphthongs in this sense.

Nortaneous wrote: ↑Mon Oct 12, 2020 11:45 am Also, you're right, it's 60 - I read the 8 in the database as a 6 (really need to clean my screen) and /ɛe ɛ̰ḛ/ are phonetic variants of underlying /ɛː ɛ̰ː/ so probably shouldn't count as diphthongs in this sense.

As /ã̤ ː/ is [ʌ̤a̰] (breathy tense), you might be able to argue the count down to 57ǃ I wonder if /iu/, /aɪ/ and /ao/ are real diphthongs, or should count as having semivowel codas. /pa̤a̰j/ ʹtwoʹ is counted as CVVN in the syllable structures.

Nortaneous wrote: ↑Mon Oct 12, 2020 3:46 am

bradrn wrote: ↑Sun Oct 11, 2020 11:28 pm

Nortaneous wrote: ↑Sun Oct 11, 2020 11:12 pm Not too unusual - some other Papuan languages are reported to have /ħ/

Which ones?

Kobon

Davies’ grammar says not — [x] and [h], yes, but not [ħ].

Huli

This one doesn’t have /ħ/ either.

Managalasi

I can’t find a source for this one, though. (Although Wikipedia does say it doesn’t have /ħ/.)

- Tibetic epiglottals, I guess

All right, you have to be joking here… source, please?

Pick five papers by Hiroyuki Suzuki and you'll find one - he does a lot of fieldwork but unfortunately tends to publish in moon runes.

I assume ‘moon runes’ is a quaint colloquialism for the Japanese writing system, correct?

- strident vowels (otherwise known mostly from Khoisan) in a few Sino-Tibetan languages, especially Bai (imo Tangut grade 2 could've been strident as well)

And this is just unbelievable. Again: source?

Uh, like, the Wikipedia article on strident voice, man.

Oops, that should have been the first place I checked, sorry!

Suzuki insists that there's such a thing as ɧ and even ɧʰ (aspirated fricatives aren't too surprising of course)

In Sino-Tibetan? Swedish, yes, but surely nowhere else?

Japhug has the tɕ/c/kj contrast

Ah yes, Japhug, about which I have complained before.

and velarized or uvularized vowels are everywhere.

OK, now I’m just confused. How exactly does one velarise a vowel?

(What's the language with the most vocalic base qualities? … )

Danish, surely?

The Irish consonant system is much more Slavic than it is Micronesian; it's just that convention says you write things with ˠ.

Hmm, true, I suppose… (Though I seem to remember some sort of interaction between the consonants and the vowels, of the sort present in Micronesian but not in Slavic.)

bradrn wrote: ↑Mon Oct 12, 2020 3:56 pm

Nortaneous wrote: ↑Mon Oct 12, 2020 3:46 am Kobon

Davies’ grammar says not — [x] and [h], yes, but not [ħ].

Davies consistently describes it as pharyngeal here and here.

Huli

This one doesn’t have /ħ/ either.

This one (PDF) does. Unlike for Kobon and Managalasi, it looks like there was some attempt to render an actual <ħ> for the phonetics.

Managalasi

I can’t find a source for this one, though. (Although Wikipedia does say it doesn’t have /ħ/.)

Here.

I assume ‘moon runes’ is a quaint colloquialism for the Japanese writing system, correct?

Japanese and Chinese.

Suzuki insists that there's such a thing as ɧ and even ɧʰ (aspirated fricatives aren't too surprising of course)

In Sino-Tibetan? Swedish, yes, but surely nowhere else?

In Amdo Tibetan dialects.

OK, now I’m just confused. How exactly does one velarise a vowel?

Jackson T.-S. Sun: (PDF)

Syllables containing velarized vowels are pronounced with the dorsum of the tongue arched toward the soft palate. I first discovered velarized vowels in the Horpa dialects of Rangtang County (J. Sun 2000b); subsequent field research turned up velarized vowels in Zhongre, Dawei (Kangshan Township), and Mulang (Ribu Township)varieties of Showu rGyalrong and, most recently, in the Luoxi variety of Lavrung (J. Sun forthcoming-b). A preliminary comparison shows that while distinctive vowel velarization is not reported in any other Tibeto-Burman language, such vowels may be a feature of Proto-rGyalrongic. When Professor Huang Bufan joined me in my Luoxi Lavrung sessions in Fall 2002, she commented that the Luoxi velarized vowels sounded like what she had previously described as ‘tense vowels’ in Muya (see Huang 1991a:101).

Other Rgyalrongic languages (e.g. Nyagrong Minyag) are reported to have uvularized vowels (which are probably extant allophonically in American English due to spreading from coda -l, but I don't know if anyone's done the necessary phonetic studies to confirm that), but IIRC Xun Gong has an argument somewhere in here that a distinction between vowel velarization and uvularization is necessary. Other Qiangic varieties (Hongyan?) are described as having pharyngealized vowels, but that's the thing that's in IPA so who knows what they actually are.

(What's the language with the most vocalic base qualities? … )

Danish, surely?

Wikipedia has /a ɑ ɒ e œ ɔ e ø o i y u/ for Danish, ignoring unstressed vowels. 12 isn't that many - probably not the highest even in Germanic. (Wikipedia says Colognian has /a ɛ œ ɔ e ø o ɪ ʏ ʊ i y u/, for 13, but the near-high vowels can only appear short and it's possible that some of these could be analyzed away.) Kensiu has /a ɛ ʌ ɔ e ə o e̝ ɚ o̝ ɪ i ɯ u/ for 14 - 13 if you ignore the r-colored vowel, but r-coloring can be either a secondary articulation or a base. (In American English, /ɚ/ arguably fills the high central vowel slot: like /i u/, but unlike other vowels, it has an associated semivowel and can form closing diphthongs.)

The Irish consonant system is much more Slavic than it is Micronesian; it's just that convention says you write things with ˠ.

Hmm, true, I suppose… (Though I seem to remember some sort of interaction between the consonants and the vowels, of the sort present in Micronesian but not in Slavic.)

The claims that the Irish short vowel system is vertical?

I'm skeptical of the claims of a VVS in Marshallese (seems like it could probably just be closed syllable neutralization), but you still get consonant-vowel interaction in Slavic - i/ɨ, Russian ë, the infrequency of nonpalatalizing /e/, etc.

Nortaneous wrote: ↑Mon Oct 12, 2020 4:43 pm here and here. … This one … Here.

Huh, so they do! (Though the contradiction between your source and mine for Huli is a bit confusing…)

Suzuki insists that there's such a thing as ɧ and even ɧʰ (aspirated fricatives aren't too surprising of course)

In Sino-Tibetan? Swedish, yes, but surely nowhere else?

In Amdo Tibetan dialects.

Interesting article, thanks! One slight omission: he forgot to refute Ladefoged’s contention (on articulatory grounds, if I remember correctly) that such a sound doesn’t actually exist.

OK, now I’m just confused. How exactly does one velarise a vowel?

Jackson T.-S. Sun: (PDF)

Syllables containing velarized vowels are pronounced with the dorsum of the tongue arched toward the soft palate. I first discovered velarized vowels in the Horpa dialects of Rangtang County (J. Sun 2000b); subsequent field research turned up velarized vowels in Zhongre, Dawei (Kangshan Township), and Mulang (Ribu Township)varieties of Showu rGyalrong and, most recently, in the Luoxi variety of Lavrung (J. Sun forthcoming-b). A preliminary comparison shows that while distinctive vowel velarization is not reported in any other Tibeto-Burman language, such vowels may be a feature of Proto-rGyalrongic. When Professor Huang Bufan joined me in my Luoxi Lavrung sessions in Fall 2002, she commented that the Luoxi velarized vowels sounded like what she had previously described as ‘tense vowels’ in Muya (see Huang 1991a:101).

Another fascinating article, thank you! I had thought that aspectually-conditioned stem variation was restricted to Papuan languages (e.g. Komnzo, Abui), so it’s interesting to know it pops up in Sino-Tibetan as well.

And actually, now that you give a definition, I think I may have velarised vowels in my own speech! In particular, I think I may have [ɯˠ] as an allophone of syllabic /l̩/ (though I’m not completely certain that it’s really velarised).

The Irish consonant system is much more Slavic than it is Micronesian; it's just that convention says you write things with ˠ.

Hmm, true, I suppose… (Though I seem to remember some sort of interaction between the consonants and the vowels, of the sort present in Micronesian but not in Slavic.)

The claims that the Irish short vowel system is vertical?

People have seriously claimed that‽ (But no, I wasn’t suggesting that; I had thought that earlier in the thread you said something about vowel neutralization in Irish, but I can’t find it now, so maybe I was misremembering.)

I'm skeptical of the claims of a VVS in Marshallese (seems like it could probably just be closed syllable neutralization)

Yes, I remember we had a long discussion about that earlier in this thread.

but you still get consonant-vowel interaction in Slavic - i/ɨ, Russian ë, the infrequency of nonpalatalizing /e/, etc.

Well, I don’t really know much about Slavic phonology, so I’m not too surprised I got that wrong. (I should really stop talking about topics I don’t know… )

bradrn wrote: ↑Mon Oct 12, 2020 6:00 pm

Nortaneous wrote: ↑Mon Oct 12, 2020 4:43 pm here and here. … This one … Here.

Huh, so they do! (Though the contradiction between your source and mine for Huli is a bit confusing…)

SIL OPDs are typically pretty bad, although the linguistic quality of sufficiently old SIL papers isn't great either.

Interesting article, thanks! One slight omission: he forgot to refute Ladefoged’s contention (on articulatory grounds, if I remember correctly) that such a sound doesn’t actually exist.

That's not strictly necessary - if it does exist, it can't not. But I'm not convinced that it's really doubly articulated, especially without instrumental evidence. The sound in question could be a sequence, an unusually backed sibilant, or a strongly velarized postalveolar.

Nortaneous wrote: ↑Mon Oct 12, 2020 11:23 pm

bradrn wrote: ↑Mon Oct 12, 2020 6:00 pm

Nortaneous wrote: ↑Mon Oct 12, 2020 4:43 pm here and here. … This one … Here.

Huh, so they do! (Though the contradiction between your source and mine for Huli is a bit confusing…)

SIL OPDs are typically pretty bad, although the linguistic quality of sufficiently old SIL papers isn't great either.

Ah, that’ll be good to know for the future.

Interesting article, thanks! One slight omission: he forgot to refute Ladefoged’s contention (on articulatory grounds, if I remember correctly) that such a sound doesn’t actually exist.

That's not strictly necessary - if it does exist, it can't not. But I'm not convinced that it's really doubly articulated, especially without instrumental evidence. The sound in question could be a sequence, an unusually backed sibilant, or a strongly velarized postalveolar.

That’s what I meant: I’d like proof that it is indeed phonetically doubly-articulated, rather than some other phoneme which was misanalysed (which I believe is the situation with the Swedish sound).

bradrn wrote: ↑Mon Oct 12, 2020 11:35 pm That’s what I meant: I’d like proof that it is indeed phonetically doubly-articulated, rather than some other phoneme which was misanalysed (which I believe is the situation with the Swedish sound).

Swedish sje has other issues as well.

Another option could be a phonetic cluster - Dayang Pumi has clusters like [sʃ ɕʃ zʒ], so some sort of [ɕx] that's treated as a unit could be within the realm of possibility.

For other instances of units that aren't, /st/ is claimed for a few languages I can't remember and reconstructed for PNEC, retroflex trilled affricates have been reported for Ersu, labial-alveolars show up in Yeli Dnye and NWC, and Wela is alleged to have "an affricate consisting of a voiceless, unaspirated, alveolar stop plus a lenis voiced, alveolo-palatal, grooved fricative" - as an allophone of /k/, but it has a phonemic prenasalized counterpart where the stop compoennt is still allegedly voiceless. Very weird! Yamdena apparently has an "alveolar-velar double stop", although it's not clear to me why this isn't analyzed as a cluster. There's also the Somali uvular-epiglottal stop, whatever that is...

Nortaneous wrote: ↑Tue Oct 13, 2020 2:28 am For other instances of units that aren't, /st/ is claimed for a few languages I can't remember and reconstructed for PNEC, retroflex trilled affricates have been reported for Ersu, labial-alveolars show up in Yeli Dnye and NWC, and Wela is alleged to have "an affricate consisting of a voiceless, unaspirated, alveolar stop plus a lenis voiced, alveolo-palatal, grooved fricative" - as an allophone of /k/, but it has a phonemic prenasalized counterpart where the stop compoennt is still allegedly voiceless. Very weird! Yamdena apparently has an "alveolar-velar double stop", although it's not clear to me why this isn't analyzed as a cluster. There's also the Somali uvular-epiglottal stop, whatever that is...

And don’t forget Hmong’s laterally-released plosives!

(But what reason is there to analyse Yélî Dnye /t͡p/ etc. as being non-unitary? I find it even easier to pronounce as a single consonant than /k͡p/…)

bradrn wrote: ↑Tue Oct 13, 2020 2:49 am And don’t forget Hmong’s laterally-released plosives!

(But what reason is there to analyse Yélî Dnye /t͡p/ etc. as being non-unitary? I find it even easier to pronounce as a single consonant than /k͡p/…)

Aren't the laterally released plosives argued for on phonotactic grounds? I don't see anything wrong with positing that the only permissible clusters are labial + /l/. There's a paper somewhere out there arguing for unit rhotacized peripheral plosives in Pumi on articulatory grounds.

WRT Yeli Dnye, the main issue is that Henderson isn't very principled. Prenasalized and nasally released stops, but [t̪p t̠p kp n̪m n̠m ŋm] are units, as is [lβ]... but [βʲ lʲ] are clusters. Unit prenasalized stops are well-attested elsewhere, but labial-coronals aren't (except allophonically in NWC) - why not either minimalism (no 'complex' units) or maximalism (only Pj Pw Nj Nw are clusters)?

Nortaneous wrote: ↑Fri Oct 30, 2020 7:17 pm

bradrn wrote: ↑Tue Oct 13, 2020 2:49 am And don’t forget Hmong’s laterally-released plosives!

(But what reason is there to analyse Yélî Dnye /t͡p/ etc. as being non-unitary? I find it even easier to pronounce as a single consonant than /k͡p/…)

Aren't the laterally released plosives argued for on phonotactic grounds? I don't see anything wrong with positing that the only permissible clusters are labial + /l/.

Yes, but apparently they’re sometimes analysed as unit /pˡ bˡ/ etc. (According to Wikipedia, at least.)

There's a paper somewhere out there arguing for unit rhotacized peripheral plosives in Pumi on articulatory grounds.

Yes, you liked to it earlier in the thread.

WRT Yeli Dnye, the main issue is that Henderson isn't very principled. Prenasalized and nasally released stops, but [t̪p t̠p kp n̪m n̠m ŋm] are units, as is [lβ]... but [βʲ lʲ] are clusters. Unit prenasalized stops are well-attested elsewhere, but labial-coronals aren't (except allophonically in NWC) - why not either minimalism (no 'complex' units) or maximalism (only Pj Pw Nj Nw are clusters)?

Well, I only read the Wikipedia article rather than the original analysis, so I’m not surprised I missed this stuff. This SIL analysis seems slightly saner — it proposes a simple consonant inventory of only 13 consonants, which can then be palatalised and labialised, with suprasegmental nasalisation and simultaneous bilabial closure.

Nortaneous wrote: ↑Tue Oct 13, 2020 2:28 am For other instances of units that aren't, /st/ is claimed for a few languages I can't remember and reconstructed for PNEC,...

English, perhaps. There's evidence from learning to read, phonology (paste, post), having its own rune, and even Anglo-Saxon alliteration. One learned paper on the subject is SPecial STatus: Presigmatised Stops.

Minnan Chinese has some pretty elaborate tone sandhi for syllables in non-pausal position. For example, let's look at southern/standard Taiwanese Hokkien, which has seven tones, numbered from #1 to #5 plus #7 and #8. (There is no tone #6 because the conventional numbers are based on Late Middle Chinese tone categories, and historically the Late Middle Chinese 陽上 voiced shǎng = #6 and 陽去 voiced qù tone = #7 merged into one tone, and so "#7" is used by convention.) Some grammatical/function words are also basically toneless.

Now, sandhi.


Source: Wikimedia. https://en.wikipedia.org/wiki/File:Taiw ... dhi_01.svg

As you can see in this diagram, in this dialect of Minnan, in non-pausal position (i.e. when a tone-bearing syllable is not the last word of an intonation phrase), tone #2 becomes #1, #3 becomes #2, #4 becomes #8 when ending in /p t k/ but #2 when ending in /ʔ/ (written -h in the standard romanization), #5 becomes #7, #7 becomes #3, #8 becomes #4 when ending in /p t k/ but #3 when ending in /ʔ/, and finally #1 becomes #7.

Oh, but it gets better. In some key grammatical or near-grammatical monosyllabic words, the sandhi chain is applied twice:
欲 beʔ4 'to want sth; be about to [do]; if', 佮 kaʔ4 'with; and [NP]', 閣 koʔ4 'again', 才 chiaʔ4 'only [VP]; only then; just a moment ago' plus 去 khi3 'to go'. They have the cited tones in pausal position, but in non-pausal position these particular tone 4 words and the tone 3 'to go' go through tone #2 and arrive at tone #1.

Oh, but there's more. In the triplication of a monosyllabic adjective (i.e. reduplication but you do it thrice), the rules are slightly different for the first syllable, as #5 stays unchanged (instead of becoming #7), and #7 becomes #1 (instead of #3), and also #8 becomes #5 if followed by /ʔ/ (instead of #2), with the usual pattern applying otherwise. And a syllable followed by the -á suffix follows a quite different diagram altogether.

If Hokkien is not a prime example of a terrible conlang, made by a half-asleep guy off work at midnight after downing some two bottles of strong 60% alcohol-content kaoliang, then I damn don't know what is.

I think it was just an engelang to make fun of every existing phonological theory