Zompist Bboard Again

Travis B. wrote: ↑Sat May 23, 2026 11:46 am(...) I can convert it to mean 'contain' by changing the word order and, when needed, adding an inverse morpheme to the verb (e.g. the containing NP is not lower on the personhood/animacy/topicality hierarchy than the contained NP).

That's interesting. Does that mean that regardless of S/O word order, S is always the most salient NP (based on whatever hierarchy you have), unless you mark the verb as "inverse", and the actual S/O word order determines the topicality? So e.g.:

me pesca etam - I eat a fish
pesca me etam - it's a fish I eat
pesca me du-etam - the fish eats me
me pesca du-etam - it's the fish that eats me
?


JAL

jal wrote: ↑Sun May 24, 2026 5:21 am

Travis B. wrote: ↑Sat May 23, 2026 11:46 am(...) I can convert it to mean 'contain' by changing the word order and, when needed, adding an inverse morpheme to the verb (e.g. the containing NP is not lower on the personhood/animacy/topicality hierarchy than the contained NP).

That's interesting. Does that mean that regardless of S/O word order, S is always the most salient NP (based on whatever hierarchy you have), unless you mark the verb as "inverse", and the actual S/O word order determines the topicality? So e.g.:

me pesca etam - I eat a fish
pesca me etam - it's a fish I eat
pesca me du-etam - the fish eats me
me pesca du-etam - it's the fish that eats me
?

Yes. Note, though, that when both core arguments are otherwise of equal salience, the first one is considered to be more salient because it is more topical.

I would translate your fourth sentence as "I am eaten by the fish". And yes, while there is a passive, that is used primarily as a valence-reducing operation rather than to change argument topicality.

Reminds me of Malay / Bahasa Indonesia

In my old notes for Proto-Dwarvish (temporary name) there are notes for a proposed consonant inventory and sound changes from a protoprotolanguage. So in proposal 2 for example, we have sound changes like:
p → pʰ → pɸ → pf
b̥ → b → p
b̬ → ɓ → b
And so on.

I'm wondering which proposal is the most plausible? Is there any proposal that is unlikely?

Qwynegold wrote: ↑Wed May 27, 2026 9:45 am I'm wondering which proposal is the most plausible? Is there any proposal that is unlikely?

The only thing is I'd expect ejectives, especially dorsal ejectives, and implosives, especially labial implosives, to be more stable than your options indicate, which have them only as intermediates. Also, if they are lost I would expect them to become tenuis or, in the case of ejectives, pharyngealized or velarized/uvularized rather than affricated. Also, a common pattern is when there is a contrast between voiceless plosives and ejectives is for the voiceless plosives to become aspirated to increase the contrast with the ejectives.

All right, so proposal 4 is the least good out of these. Thanks for the input! I'll think about which one of the remaining three I want to use.

I came across an interesting paper, Articulatory possibilities for implosives, which implicitly argues by describing how to produce them yourself that the reason for the lack of attested implosive affricates, fricatives etc. may be coincidental¹ rather than due to being impossible to produce. Wikipedia gives Swahili as an example where j can be either an implosive stop [ʄ] or a non-implosive affricate [d͡ʑ] in free variation, but Swahili lacks a phonemic voiced affricate [d͡ʑ] from which one would need to maintain a distinction when realising j.

¹ Coinci-dental consonants sounds like a cool idea for a surrealist conlang.

The current version of Ithkuil seems way simpler than what I remember. The original version had like eighty consonants and twenty vowels and half a dozen tones whereas this one only has forty phonemes. It seems even conlangs have embraced the ozempic craze and slimmed down.

malloc wrote: ↑Wed Jun 17, 2026 5:58 pm The current version of Ithkuil seems way simpler than what I remember. The original version had like eighty consonants and twenty vowels and half a dozen tones whereas this one only has forty phonemes. It seems even conlangs have embraced the ozempic craze and slimmed down.

Yes, John Quijada has simplified it very much, apparently because he had realized that it was too complicated for others to learn it. It still is a quite complex language, though.

I am having troubles deciding on the "vocalism" of the person marking prefixes. The prefixes are nV-, mV, and ʔV-, but I don't know if the vowel should be /ɑ/ or /ɪ~i/. Part of the difficulty is that I can't decide which vowel is more "marked" or less "marked" (and more likely to be a placeholder/epenthetic vowel).
(current) Relatedly, because my irrealis stem is differentiated through gemination, I don't have any favorable ways of fitting quadriliteral roots into the paradigm (because both the realis and irrealis forms would be identical). I might have to rip off of Semitic languages (ablaut and different vocalism of person markers).

I would go with whatever vowel is more common as less marked and conversely what vowel is less common as more marked. As crosslinguistically /a/ tends to be the most common 'full' vowel I would consider it less marked without any further information.

Travis B. wrote: ↑Sat Jun 20, 2026 7:43 pm I would go with whatever vowel is more common as less marked and conversely what vowel is less common as more marked. As crosslinguistically /a/ tends to be the most common 'full' vowel I would consider it less marked without any further information.

I was reading this paper
https://www.semanticscholar.org/paper/M ... e27f4a0cad
https://roa.rutgers.edu/files/578-0203/ ... DI-0-1.PDF (pdf link)
According to this paper, in the absence of /ə/ or /ɨ/, the most likely candidates to be epenthetic vowels are /a/ or /i/. But to me, it is hard for me to determine which of the two is least marked.

Ahzoh wrote: ↑Sat Jun 20, 2026 10:08 pm

Travis B. wrote: ↑Sat Jun 20, 2026 7:43 pm I would go with whatever vowel is more common as less marked and conversely what vowel is less common as more marked. As crosslinguistically /a/ tends to be the most common 'full' vowel I would consider it less marked without any further information.

Well, /a/ maybe, but I am talking about /ɑ/ which in this language generally manifests as [ɑ], which feels very strong and prominent.

I was reading this paper
https://www.semanticscholar.org/paper/M ... e27f4a0cad
https://roa.rutgers.edu/files/578-0203/ ... DI-0-1.PDF (pdf link)
And the problem is that I cannot decide if nonlow vowels are least marked or if low vowels are.

There is this paper I have not yet finished reading which claims to investigate why languages with the same vowel system have different epenthetic vowels and suggests that its results "indicate that frequency does not predict the epenthetic vowel"
https://www.semanticscholar.org/paper/A ... 24733539f4
https://scispace.com/pdf/abstract-inves ... 6ndw3x.pdf (pdf link)

"/ɑ/" and "/a/" are often interchangeable, with contrasts between the two that do not involve a secondary contrast such as length being relatively uncommon, and a language will generally have one or the other, or if it has neither, it will generally have /ɒ/.

Travis B. wrote: ↑Sat Jun 20, 2026 10:23 pm "/ɑ/" and "/a/" are often interchangeable, with contrasts between the two that do not involve a secondary contrast such as length being relatively uncommon, and a language will generally have one or the other, or if it has neither, it will generally have /ɒ/.

Still, under many circumstances /ɑ/ feels equally as unmarked as /i/ and sometimes it is /i/, especially word-finally (unless I pronounce /ɑ/ more like schwa, which it's not supposed to be pronounced like)

And I do think this matter is relevant for determining the quality of hyper-reduced person-marking morphemes (just lik a schwa would).

Ketsuban wrote: ↑Wed Jun 03, 2026 2:54 pm I came across an interesting paper, Articulatory possibilities for implosives, which implicitly argues by describing how to produce them yourself that the reason for the lack of attested implosive affricates, fricatives etc. may be coincidental¹ rather than due to being impossible to produce. Wikipedia gives Swahili as an example where j can be either an implosive stop [ʄ] or a non-implosive affricate [d͡ʑ] in free variation, but Swahili lacks a phonemic voiced affricate [d͡ʑ] from which one would need to maintain a distinction when realising j.

¹ Coinci-dental consonants sounds like a cool idea for a surrealist conlang.

Thanks for the input. Hmm, I still think that my alternative 4 is the least likely though.

Ahzoh wrote: ↑Sat Jun 20, 2026 10:08 pm I was reading this paper
https://www.semanticscholar.org/paper/M ... e27f4a0cad
https://roa.rutgers.edu/files/578-0203/ ... DI-0-1.PDF (pdf link)
According to this paper, in the absence of /ə/ or /ɨ/, the most likely candidates to be epenthetic vowels are /a/ or /i/. But to me, it is hard for me to determine which of the two is least marked.

Awesome! I was looking for something like this a few months ago and didn't find anything good. Instead, I had long conversations with AI about it — and it didn't find me papers.

It annoys me when I'm just starting to reading a paper and I spot something like this though:

For example, when borrowing from English, Korean epenthesizes the vowel [ɨ] in order to avoid structures such as CVC as in (1): (1) Korean epenthesis: Bus → [pəsɨ] Gas → [kasɨ] Rove → [lopɨ]

Korean is perfectly comfortable with CVC. The reason "bus" and "gas" get an epenthetic ㅡ /ɯ/ (weird to represent it as ⟨ɨ⟩ ... and also very weird to represent ㅓ/ʌ/ as ⟨ə⟩) is that there is no syllable final /s/ as it merges with /t/ in that position. English syllable final plosives also tend to get an epenthetic ㅡ because they tend to be released when spoken clearly (in most dialects), whereas in Korean, a CVC syllable ending where the final C is a plosive is unreleased. There is no issue with CVC as in 팀 [tʰim] "team" and 풀 [pʰul] "pool".

I also cannot believe "Rove" would be borrowed as 로프 [lopɯ]. That's how "rope" is borrowed. I can't find evidence "Rove" borrowed into Korean and I'm not sure what context is meant (a personal name?), but it would be 로브 [lobɯ] (which is also how "robe" is borrowed). The phoneme represented by ㅍ is [pʰ] initially and [p] medially. The phoneme represented by ㅂ is [p] initially and [ b ] medially. The author has used square brackets and used the initial phone of the phoneme in question in a medial position and given "rope" rather than "rove". It's pretty disheartening to see such sloppiness right at the beginning of a paper.

Imralu wrote: ↑Fri Jun 26, 2026 5:44 pm /ɯ/ (weird to represent it as ⟨ɨ⟩ ... and also very weird to represent ㅓ/ʌ/ as ⟨ə⟩)

Don’t know about /ʌ/, but there’s hardly any languages (if indeed there are any at all) which contrast /ɯ/ and /ɨ/. This wouldn’t the first time I’ve seen one symbol where other authors use the other.

Imralu wrote: ↑Fri Jun 26, 2026 5:44 pmBus → [pəsɨ]

So the Koreans create their own version of the busy


JAL

Now my "repost stuff I like from Bluesky on the ZBB" habit is spreading to other forums than Ephemera:

https://bsky.app/profile/opinionhaver.b ... nf3dev5c2p

Voynich Manuscript.
Voynich Desk Reject.
Voynich Lower Impact Manuscript.
Voynich Revise and Resubmit.
Voynich Revise and Resubmit.
Voynich Research Note.

I don't know what to do or how to decide what vowel should come between the root stem and the case endings.
I don't want to pull a Semitic and have the vowel indicate case, but I don't want to clone PIE either. And I'm no longer satisfied with the vowels indicating gender.

The case endings Possible plural endings
-ā (found on the verb as well)
-n (found on pronouns)
-H (laryngeal of uncertain value, lengthens preceding vowel)

I have thought maybe having ancient obviative marking be fossilized into a theme vowel, but the result isn't very satisfying. But it would satisfy why all nouns in construct state end in the same vowel.