Conlang relay [relocated] (aka "The Cursed Relay")

[Zhen Lin]

I found a SIL keyboard layout for IPA, but that wasn't really X-SAMPA so I decided against it.

[Corumayas]

An issue I've been mulling over in Eige-Isthmus is that while the Eige Valley langs reversed the voicing of stops and affricates in general, there's a root *tsimp- which begins with a voiceless consonant in both branches (Faraghin sempan, 'twist, stunt', Eastern Miwan timpi:za 'necklace').

I thought about declaring that word irregular, or an early borrowing in Proto-Eige Valley from a variety that didn't have the voicing shift; but it would be more satisfying to make it regular. So then I considered adding aspirates to Proto-Eige-Isthmus, which would allow a whole set of words that were voiceless in both branches. I added them to the wiki page about a month ago.

However, I've just realized that *tsimp- is currently the only secure root with cognates in both branches that begins with *ts. (There's also one tentative root where Faraghin /s/ corresponds to Ndak Ta /s/, which could come from Ngauro /s/ or /z/.) So it could just as well be that all *ts remained voiceless, and PEI *ts *dz both became PEV *ts.

This would mean no *dz in Proto-Eige Valley. That's not a problem for Miwan, which would've merged *dz into /d/ anyway; nor for Ngauro, which has another source of /z/ in *j. But it might make a difference for Meshi, so I wanted to check with boy#12 before making it official. (Of course, the merger of *ts and *dz could occur in Ngauro and Miwan but not in Meshi if you want; that might change the structure of the family a bit.)


Another consequence of the change is that Ngauro's early consonant inventory (/p t k b d g m n ŋ ts f s r l j w/) looks more like Ndak Ta's. The only differences are the presence of /f j/ in Ngauro, and phonemic /kw bw ŋw/ in NT. (Later, Ngauro loses /ts r w/ to sound changes, and gains /v z/.) Which suggests to me (fwiw) that it'd be very easy for the Ndak to adopt the Ngauro writing system...

[Basilius]

Another upload: Proto-Isles sources of Ppãrwak sounds.

That is, sound correspondences grouped by Ppãrwak reflexes, then by Proto-Isles sources for each reflex. (It would be more informative the other way round - I'll do that, too.) Just (complete) lists of comparanda, no attempt to explain or stratify the sound changes, as of now.

A provisional thing again, sort of working records, please don't blame me for errors in wiki markup conversion &like.

[...] it has some unnecessary repeated entries and some question marks.

That's because there seem to be typos etc. in the sources, so it is sometimes important to know that a form occurs more than once, or has no reliable etymon, etc. I'll make it more compact when I have time.

[Radius Solis]

Another consequence of the change is that Ngauro's early consonant inventory (/p t k b d g m n ŋ ts f s r l j w/) looks more like Ndak Ta's. The only differences are the presence of /f j/ in Ngauro, and phonemic /kw bw ŋw/ in NT. (Later, Ngauro loses /ts r w/ to sound changes, and gains /v z/.) Which suggests to me (fwiw) that it'd be very easy for the Ndak to adopt the Ngauro writing system...

That would be realistic enough, I should expect: looking at it the other way, given that Ngauro came first, it may well be that Ndak Ta's development from proto-Talo-Edastean was quite influenced by Ngauro, which for them could have been a prestige language early on, and at minimum a source of loans. This could have been a central motivator for NT's loss of the Talo-Edastean uvular series, in particular, at the time of the Ndak settling of Kasadgad and their language being taken up by the larger existing Ngauro population who probably would have had much more trouble pronouncing the uvulars than the labiovelars. So yes, a general similarity between Ngauro's and NT's phoneme sets seems not only realistic but desirable to me.

Also:
This gives me an idea about a possible source of further hints about Ngauro: if you're interested, you might consider having a couple of the Eastern Edastean shared developments be a result of Ngauro influence on eastern dialects of colloquial NT. The tendency towards auxiliary verbs, for example.

[Basilius]

Corumayas > The Faraghin reflex has a spirant, and I notice that Faraghin has no voiced spirants.

Besides, even if it had them, spirants don't need to follow the same pattern as stops/affricates.

Therefore, it's just a matter of the relative timing of fricativization. If it happened early enough, the spirant could remain voiceless.

Just one more option that may help to avoid positing an extra series because of just one pair of cognates.

(Affricates, too, can be somewhat special in terms of VOT-based contrasts, so your scenario looks quite natural as well.)

[dunomapuka]

This would mean no *dz in Proto-Eige Valley. That's not a problem for Miwan, which would've merged *dz into /d/ anyway; nor for Ngauro, which has another source of /z/ in *j. But it might make a difference for Meshi, so I wanted to check with boy#12 before making it official. (Of course, the merger of *ts and *dz could occur in Ngauro and Miwan but not in Meshi if you want; that might change the structure of the family a bit.)

I would like Meshi to do the same thing as Miwan. Merge away!

[Corumayas]

It shall be done! I've just finished editing the wiki article, with some ambivalent comments about Meshi's behavior; I'll go clean those up now.

Radius: Nice ideas! I will definitely consider Eastern Edastean features when working on Ngauro (as well as general Ndak ones maybe, given your first paragraph).

Basilius: As it happens, the regular Faraghin reflex of *dz is actually /r/ (from Western Isthmus *ð).

In an attempt to get back into Eige-Isthmus grammar, I'm considering something analogous to the Muskogean infix/ablaut system for marking aspect. I need to think about how that could work and what it might look like in the daughter langs.

[dunomapuka]

Hooray, Naidda's Tsinakan text is up! Since you have put this on a separate "Texts" page, Radius, does this mean you're thinking of doing other texts? I've been thinking I want to do another one for Namɨdu, but not sure what. 'Course, my language description isn't finished yet as it is...

[Radius Solis]

Yes, actually. The Tsinakan text is really a rather poor one for showing Naidda's more usual narrative style which leans heavily toward poetry, parallelism and parataxis, not the syntax-heavy prose style of the Tsinakan text. But I had to stick to the original syntax as much as I could because 1. it's supposed to be a Naidda translation of the original NT, not a native tale; and 2. for comparison purposes with the other Edastean languages.

So, to illustrate what a native Naidda tale would be like, I have been working on a short passage more suited to the task. The trouble is that it has turned into a dark little story recounting in religious terms an unfortunate historical incident with a grisly end (sacrifices to the Blood God). I'm not sure yet if I want to keep this and run with it or try for something else. It's not that I don't want to do it, I'm just wary of giving the impression that such dark and powerful material is standard fare for these people. Maybe I should come up with a third and happier text to counterbalance it...


Texts aside, though, the Naidda grammar is essentially done. I might add one or two more modal adverbs (I need a way to say "able to X") and/or other such things (the Tsinakan Text leaves me wishing for a preposition "against") and I'm still correcting minor infelicities as I run across them and a doing little visual cleanup, but there's no further major work coming.

But, I am thinking of splitting the page up. That is one reason I made a subpage for /Texts. The page is so damn big, I'm thinking to move elsewhere any material I can. I'm still deciding how to go about that.

[Zhen Lin]

If anyone is interested, I have written an account of the sound changes of Ayāsthi. It should be complete, but may be inaccurate. (I'm translating it from a program specification.)

[Basilius]

If anyone is interested, I have written an account of the sound changes of Ayāsthi. [...]

Yes, at least for me in particular, this kind of stuff is probably the most tasty part of what is going on here

Which seems to mean that I'll ask too many questions, too, unless somebody stops me
I am asking them without really knowing the lang, so some may sound dull, sorry...
I suppose that some bizarre points are explained by the way your algorithm was built, but anyway.

---
> [ɑ ɛ o u] → [æ e œ ɯ] / _j

That is, before [j] the sound loses its rounded quality but remains back? Remaining opposed to [ɨ] for long? Meanwhile, next thing that happens to [ɯ] seems to be the following:

> [ɯ ɨ] merger
> * [ɯ] → [ɨ]

(immediately followed by "Delabialisation of "). Then, why not > [ɨ] right away?

- - -
> Vowels are nasalised before [n m].
> [m] → [ɴ] / [õ ũ õː ũː]_

Does this mean that [m] is dropped after rounded nazalized V's? Or is [ɴ] a consonant?

---
> → [e o] / V_, when [ɑ æ œ ɔ o] is in the next syllable

is this recursive? e. g. ...uCuCo... > ...uCoCo... or ...oCoCo..?

---
> [ɹə lə mə nə ɲə] → [əɹ əl əm ən əɲ] / #_#

Does this mean "only in monosyllabics with zero codas"?

---
> Geminate consonant shifts [...]
> * [θː zː ʒː z̞ː] → [sθ sː ʃː ɹː]

IMHO [θː] → [sθ] does not look especially natural. Is this just how the program works, or is there a motivation behind this change? (Ultimately this cluster seems to yield just [θ] anyway?)

---
> # [ə] → ∅ / _V, unless the following vowel is also [ə], but never if nasalised or long.
> # [ə] → ∅ / V_V, V_C, except if nasalised or long

Does "nasalised or long" apply to [ə] or the adjacent V's? or perhaps to both?

---
> [ɦ] → [ʔ] / C_, _C, _#
> [ʔ] → ∅ / _#
> [ʔ] → ∅ / _C, geminating the next consonant.

Are these applied consecutively? That is, ultimately [ɦ] > ∅ except when followed by a V?

---
> [ɴm ɴn ɴɲ] → [mm nn ɲɲ]

What exactly is [ɴ] here?

---
> If the first vowel was stressed before synaeresis, then the resulting long vowel has circumflex accent; otherwise it has acute accent.

I assume "otherwise" = "if the second V was stressed". Correct?

---
> Paragoge
> * ∅ → [ɨ̆] / VːCː_#, VːCC_#

IMHO this one would look more natural if made an exception to the earlier word-final apocope, without an extra to-and-fro development. I don't think the accompanying gemination will be an issue.

***********
Also, it would be nice if you supply each change with a few examples sometime later. From personal experience, a good format is smth. like the following:

(ProtoLangForm > ) ImmediatelyBeforeChange > ImmmediatelyAfterChange ( > DaughterLangForm)

Besides making it easier to read and understand, sometimes it helps to find bugs. Also, sometimes a search for potential examples reveals that very few or none is actually available, leading to unexpected gaps in the resulting phoneme inventory.

BTW, what software do you use? I suspect it can help with supplying such examples.

[Zhen Lin]

Some of it is an attempt to make it have stages which could potentially be developed as intermediate languages, or to provide branch points; essentially, the focus is not on getting from the start to the end with as few rules as possible.

> Vowels are nasalised before [n m].
> [m] → [ɴ] / [õ ũ õː ũː]_

Does this mean that [m] is dropped after rounded nazalized V's? Or is [ɴ] a consonant?

Overall, /m/ is eventually dropped after rounded vowels, yes. [ɴ] is the uvular nasal.

> → [e o] / V_, when [ɑ æ œ ɔ o] is in the next syllable

is this recursive? e. g. ...uCuCo... > ...uCoCo... or ...oCoCo..?

This is not recursive.

> [ɹə lə mə nə ɲə] → [əɹ əl əm ən əɲ] / #_#

Does this mean "only in monosyllabics with zero codas"?

Monosyllables with zero initials.

IMHO [θː] → [sθ] does not look especially natural. Is this just how the program works, or is there a motivation behind this change? (Ultimately this cluster seems to yield just [θ] anyway?)

I thought it was difficult to pronounce [θː], so I decided to change it to [sθ]. It certainly survives - simply look at the name of the language: [ɑjɑːsθɨ̆].

> # [ə] → ∅ / _V, unless the following vowel is also [ə], but never if nasalised or long.
> # [ə] → ∅ / V_V, V_C, except if nasalised or long

Does "nasalised or long" apply to [ə] or the adjacent V's? or perhaps to both?

[ə] itself.

> [ɦ] → [ʔ] / C_, _C, _#
> [ʔ] → ∅ / _#
> [ʔ] → ∅ / _C, geminating the next consonant.

Are these applied consecutively? That is, ultimately [ɦ] > ∅ except when followed by a V?

Consecutively, yes. Phonetically, [ɦ] (or [ɣ] or other allophones) only appears intervocalically; phonemically, it can also appear after consonants, but as [ʔ], for instance, in the phrase as’-géċoṁ /ɑsɦɛːkɔ̃/ [ɑsʔɛːkɔ̃].

> If the first vowel was stressed before synaeresis, then the resulting long vowel has circumflex accent; otherwise it has acute accent.

I assume "otherwise" = "if the second V was stressed". Correct?

Of course.

IMHO this one would look more natural if made an exception to the earlier word-final apocope, without an extra to-and-fro development. I don't think the accompanying gemination will be an issue.

Perhaps. But I prefer the sound of [ɑjɑːsθɨ̆] to [ɑjɑːsθə], which would what it would have been otherwise.

Also, it would be nice if you supply each change with a few examples sometime later. From personal experience, a good format is smth. like the following:

(ProtoLangForm > ) ImmediatelyBeforeChange > ImmmediatelyAfterChange ( > DaughterLangForm)

Besides making it easier to read and understand, sometimes it helps to find bugs. Also, sometimes a search for potential examples reveals that very few or none is actually available, leading to unexpected gaps in the resulting phoneme inventory.

Hmm. Maybe. I suspect some of the rules I have are strictly extraneous as the pattern they match will never arise. (For instance, I don't think there's anything which would result in a [ɦ] appearing before a consonant.)

BTW, what software do you use? I suspect it can help with supplying such examples.

A sound change program which I wrote myself.

[Basilius]

Some of it is an attempt to make it have stages which could potentially be developed as intermediate languages, or to provide branch points; essentially, the focus is not on getting from the start to the end with as few rules as possible.

I thought I understood this

I thought it was difficult to pronounce [θː], so I decided to change it to [sθ].

[θː] is exactly like [θ], only longer Shouldn't be a problem in a lang that otherwise observes quantity distinctions on consonants. I don't see how it can be more difficult than a cluster involving two adjacent but distinct POA's.

It certainly survives - simply look at the name of the language: [ɑjɑːsθɨ̆].

Actually, there's something really cool about the change [t] > [sθ]. (I thought it was simplified because I misinterpreted the conditions of [sθ] > [θ] in Ayāsthi, sorry...)

But, to undestand how it all works... It seems that the sole source of that [θː] ( > [sθ]) is the following change:

> [ə] → ∅ / VC_#, with ɡemination of the preceding consonant.

(sole, since Adāta doesn't seem to allow geminates or clusters, and other changes in the same section don't produce any).

Also, the same change seems to be the only one that creates the conditions for your paragoge rule (the other point that exacerbated my naturalistic paranoia). Besides, [sθ] seems to be the only cluster (if geminates aren't considered clusters) which can trigger the paragoge.

Is the above correct? (If it is, there seems to be an alternative interpretation of the [θː] > [sθ] change available, with somewhat different intermediate stages; dunno if you welcome such unbidden proposals...)

A sound change program which I wrote myself.

Something I can only envy

[Zhen Lin]

But, to undestand how it all works... It seems that the sole source of that [θː] ( > [sθ]) is the following change:

> [ə] → ∅ / VC_#, with ɡemination of the preceding consonant.

More or less. This is a holdover from the original Ayāsth, which didn't have much (morpheme-internal) syncopation going on. In theory, however, word-internal geminates and clusters could arise from borrowings, contractions, compounds, etc.

Also, the same change seems to be the only one that creates the conditions for your paragoge rule (the other point that exacerbated my naturalistic paranoia). Besides, [sθ] seems to be the only cluster (if geminates aren't considered clusters) which can trigger the paragoge.

Geminates are covered by that rule - there's /tʰaːlo/ → [θɑːllɨ̆], /haːra/ → [ɦɑːɹɹɨ̆] etc.

Is the above correct? (If it is, there seems to be an alternative interpretation of the [θː] > [sθ] change available, with somewhat different intermediate stages; dunno if you welcome such unbidden proposals...)

That should be interesting to hear.

[julianallees]

Okay, I have hardly any idea why or what this post is for, but I am just replying so I can state that I was part of it.


Yours truly,


julianallees


8th February, 2009

[Basilius]

[...] so I can state that I was part of it.

While it was not me who conjured your spirit to re-appear before peoples of Akana, may I ask what was your nickname back then?

For I don't seem to find your current one among those of original contributors...

Welcome, anyway

[Basilius]

That should be interesting to hear.

The idea is that it would look much better if the actual change were [θt] > [st] > [sθ] rather than [θː] > [sθ].

[θt] > [st] would be quite expectable in a lang that has no phonemic [θ] yet but has [s], also in codas (available in Adāta at least word-finally). Sort of unification of POA's.

That [θt] could be a realization of preaspirated ['t]. Later on [st] > [sθ], similar to [t] > [θ] in some other environments.
The development ['t] > [θt] > [st] > [sθ] would be parallel to ['p] > [ɸp] > [ɸɸ] = [ɸː] and ['k] > [xk] > [xx] = [xː] in the same position.
Now, the position.
Other consonants seem to become geminates under the same conditions.
This kinda reminds me of Fennic consonant gradations.
Which makes me think of some peculiarity that arises on the boundary of syllables differing in some prosodic quality, of which (syllables) one is word-final.
The prosodic quality that triggers the whole chain of changes could be vaguely termed "reduced" (I believe phonetic precision is unnecessary if not harmful here - it could mean extra-short, or carrying lower tone, or displaying some lax phonation, or a combination of somesuch...)
No matter what precisely it was, it existed on word-final syllables contrasting with some "normal" quality. And it triggered some sort of anticipatory lengthening (realized as preaspiration on plain voiceless stops) of the preceding consonant.
The above would mean that word-finally the vowel reduction produced not just the usual schwa but some vowel of that very "reduced" quality, [ə̆].
And that vowel reduction preceded preaspiration (and subsequent shifts).
The whole set of changes:

* [ɑ ɛ e ɔ o œ] → [ə] / #_#, #_C#, _#, but word-finally further → [ə̆]
* [p] [t] [k] → ['p] ['t] ['k] → [ɸp] [θt] [xk] /_[ə̆]# ; otherwise C > Cː /_[ə̆]#
* [θt] → [st]
* (various changes of aspirates producing, among other things, phonemic [ɸ] [θ] [x])
* [p t k] (not part of affricates) → [ɸ θ x] /V [ɸ s θ x]_ except word-finally ([ɸɸ] = [ɸː], [xx] = [xː]).

As for [ə̆], you can either drop it with subsequent paragoge in the same position or make it directly → [ɨ̆] at any suitable point. (In the latter case, if you wish to branch off a sister with final clusters, just delete final [ə̆]'s as the first sound change in the sister branch. In the main branch everything, including the resulting alternations, remains as is if the "reduced" vowel blocks dropping the schwa in the preceding syllable - cf. the rules for deleting reduced vowels in Slavic, as a natlang precedent.)

[Zhen Lin]

Very interesting reimagining. The main problem would be how to rearrange the intervocalic lenitions so that the plain stops don't shift too soon. Perhaps I will make a note of it on the page.

This kinda reminds me of Fennic consonant gradations.

Hmmm... I suppose it might, however, the conditions are quite different - for example, we have the verb ésṫı [ɛːsθɨ̆], which becomes ésṫ-əṅ [ɛːsθə̃] and é’ṫ-şeı [ɛːθʃe] when followed by clitics. A more regular verb such as peılàrr becomes peılàrr-əṅ and peılàr’-şeı with the same clitics.

[Basilius]

Zhen Lin > I'm glad you liked it
Yes, the morphology promises to be fanciful, I anticipate

[Zhen Lin]

Anticipate? Ayāsthi was "completed" quite some time back, actually. See here.

[Radius Solis]

It might be nice to have a brief catalog of some of the probable areal features that are known in Akana. Actually there isn't much outside the Eige valley, but within it I can think of a few.

- Development of nasal+stop clusters into aspirates. Adāta and Ndok Aisô both do this (albeit by different paths), and Tlaliolz did it too in its early stages. The three languages formed a contiguous area. Coincidence? Well, yes, technically, from an external standpoint ghur's and dewrad's decisions were coincidence. But internal to Akana that's just got to be an areal trait. A further development of the aspirates into ejectives occurs in Tlaliolz (for dorsal POAs) and in northern dialects of Ndok Aisô, so this too might count.

- Pronominal prepositions... I started that with Naidda, but I then did it in Pencek too, and IIRC Legion has opted to do something similar in Komejech. So that's three languages, but with a distance gap and a thousand years separating Naidda from Pencek. Still, it might well be an areal feature, especially if the Miwan languages do it as well. Any thoughts about that, Corumayas? Either way, it should probably count.


So, before haring off to start a new wiki page, let me ask: what else can we think of that might be an areal feature? Anything shared by two or more adjacent languages, especially if common genetic descent doesn't provide an easier explanation?

[Basilius]

Anticipate? Ayāsthi was "completed" quite some time back, actually. See here.

I thought you were redesigning it (which was a projection from my own method: usually I don't have a finalized morphology until I understand every detail in historical phonetics).
So I failed to realize that the sound changes must be consistent with the morphology which is already fixed
Since I tried to mostly arrive at the same resulting forms, my proposal must still work, in principle. I must check.

[Basilius]

Zhen Lin > The actual problem with my proposal is palatalization (producing the alternations [θ] :: [s] and [sθ] :: [ss], AFAICT).

In your original sound changes it worked on [θ] ( → [ɕ]) before the [θː] → [sθ] change and before the vowel reduction (including [ɛ e] → [ə]).

In my proposed reinterpretation, too, it must work before the reduction (for otherwise palatalized reflexes won't be available before schwas). But in my scenario [θ] emerges only after reduction.

Therefore, a correction is needed: the palatalization affects [t] making it [t_j], and that [t_j] -
(1) spirantizes to [ɕ] in the position of spirantization;
(2) in the position of preaspiration, develops to ['t_j] then [ɕt_j] then [ɕː] (to produce the same reflexes as your palatalization of [θ] and [sθ]);
(3) in the positions where the plain voiceless stops are preserved (i. e. only word-initially?), it ultimately loses palatalization (perhaps parallel to [ɕ] → [s]).

Dunno if the other dentals ([d], [tʰ]) should be treated similarly in this setting.

[Zhen Lin]

Anticipate? Ayāsthi was "completed" quite some time back, actually. See here.

I thought you were redesigning it (which was a projection from my own method: usually I don't have a finalized morphology until I understand every detail in historical phonetics).

Same here. I've already revised it once - once again may be too much!

Zhen Lin > The actual problem with my proposal is palatalization (producing the alternations [θ] :: [s] and [sθ] :: [ss], AFAICT).

Ah, of course. How did I forget about that.

[Corumayas]

So, before haring off to start a new wiki page, let me ask: what else can we think of that might be an areal feature? Anything shared by two or more adjacent languages, especially if common genetic descent doesn't provide an easier explanation?

I've been looking over our known protolangs for precisely this recently, looking for hints for PEI morphosyntax.

Head-initial syntax is a big one across northeastern Peilaš: not only do all the early Macro-Edastean langs show it, so do Proto-Núalís-Takuña and Proto-Eigə-Isthmus. (Although the Isthmus langs then switch to head-final, as does Qedik later on... maybe another areal effect there?)

Another I've been looking at is the categories marked on verbs: among langs of eastern Peilaš, voice and aspect tend to be the most fundamental, while mood is more peripheral (and generally involves a much larger number of possibilities). Both voice and aspect involve three or more possibilities in each protolang (but not exactly the same ones). Tense is often mixed in with aspect or mood, or both (e.g., Ndak Ta has a basic tense/aspect system, but the future is a mood; in Proto-Núalís-Takuña, tense/mood is expressed by auxiliaries, but aspect is marked on the main verb).

I'm sure we can think of others, too...

Edit:

- Pronominal prepositions... might well be an areal feature, especially if the Miwan languages do it as well. Any thoughts about that, Corumayas?

Sure, why not?