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

[Corumayas]

Well, I'll try to help a little.

But you may be looking for more detailed reconstructions than are entirely reasonable. These are conlangs that were made fairly quickly for a relay, by conlangers with different amounts of knowledge and experience... Besides, even for real language families it seems that such detailed reconstructions of syntax are often impossible.

Also, I'm not sure I understand why most of these can't be regular innovations. Can you explain the issues a bit more?

For example:

Otherwise I can only point to a few issues with Ppãrwak... it preserves stem-final consonants in all forms of verbs, including those based on PI non-sensory present (where stem-final consonants were ousted by -s in Ran's PI).

This sounds like a classic (maybe a bit simplistic) case of analogy to me-- the consonant simply got generalized to all forms of the verb, even ones where it used to be deleted. Not a very interesting explanation, but is there another reason why it's unsatisfactory?

On another front, the passive constructions in Mûtsipsa' and Zele look like they might represent a shared innovation... and maybe Zele's SVO order is an intermediate step on the way to Máotatšàlì's verb-first syntax?

[Radius Solis]

Well, we all need some downtime once in a while, and after spending a bunch of time on Naidda and then Kozzang Fasa I just lost steam, as has happened a few times before. So I've been spending my spare time on other things lately, playing games and such.

But I'm starting to get the conlang itch again and looking to resume work on either of two projects... 1) Tlaliolz really needs to be wrapped up, including finalizing the pTE etymon list and putting up a Tl. lexicon, and hopefully at least a short text; and 2) Kozzang Fasa still needs lots of stuff done and wikified, including working out a bit more of its contact language Kennan Kei.

Anyone have a preference for which they'd like me to attack first?

[Corumayas]

I'd suggest working on the one that's closer to completion first.


I'm vacillating between a couple of projects too, both of which have gotten kind of stuck: one is trying to get PEI settled enough that we can create a lexicon for it and start deriving daughters. The immediate stumbling block for that right now is the dozen or so cognate proposals that don't quite fit the phonological developments as written. Should I revise the PEI consonant and vowel systems to accommodate them? Or declare them not cognate after all? Given the small number of cognates so far (15 on the "fairly certain" list), I'm hesitant to throw any plausible matches out; besides, the sound correspondences probably could use some more convolutions to make them more interesting and realistic. But revising what I've already done seems like a bit of a pain in the ass, and bending the reconstruction every which way to fit in a few oddball correspondences is not very aesthetically pleasing to me.

The other thing I'd like to do is revise Ghaf. The old Ghaf webpage finally got deleted a little while ago (it was on a server I no longer pay for, in a state I no longer live in), and since its ancestor Ayasth(i) was heavily revised I'd really like to revise it accordingly before putting it back online. I don't know if Legion and jmcd (authors of the intervening langs in the chain) are still reading this, or if they're at all interested in redoing their langs; I think the last thing Legion said about it was that it would be too much work. I would be willing to put Ayasthi through the sound changes et al. for both their langs-- in fact I've already made a partial attempt at doing that, but I ran into a lot of questions, both about what they did the first time around and about how they'd deal with the revisions in Ayasthi. So I'm not sure how to proceed with that either.

[Legion]

Still reading the thread, not much time to work on stuff right now :/

Yeah, I'm not very willing to redo something with the Ayasthi revision, because frankly Zhen Lin's langs are a nightmare to work with, diachronically. Can't we just retcon Ayasthi into becoming extinct without daughter (I'd happily let Agaf dying, it's way too ugly anyway) ?

[Corumayas]

I guess we could. I kind of like Ghaf, though... if only because it's still my most complete conlang. I'd like to try to make it better. And I was hoping to get the Ayasthi lexicon into it.

I could do a very sketchy update of Agaf myself, if you don't mind. I wouldn't publish it, just use it as an ancestor.

I don't know if we'll ever get the conworld history filled in up to the fourth millenium YP (or whenever Ghaf was supposed to be), so maybe this is becoming more of an apocryphal side project of mine at this point, rather than a canonical part of Akana.

[Legion]

I guess we could. I kind of like Ghaf, though... if only because it's still my most complete conlang. I'd like to try to make it better. And I was hoping to get the Ayasthi lexicon into it.

I could do a very sketchy update of Agaf myself, if you don't mind. I wouldn't publish it, just use it as an ancestor.

I wouldn't actually mind if you published it under your name and it officially replaced the existing version of Agaf — if you got the will and energy to do that, go ahead with my blessing.

[Basilius]

I wouldn't actually mind if you published it under your name and it officially replaced the existing version of Agaf [...]

I believe this means that you change the status of the project to "open"; I think such an option was established precisely for this type of cases. BTW, I'd suggest using it as often as possible, that is, assigning the "open" status to all projects that aren't especially dear to you.

Yeah, I'm not very willing to redo something with the Ayasthi revision, because frankly Zhen Lin's langs are a nightmare to work with, diachronically.

Can you spell this out in some detail? I looked through Ayāsthi and found nothing especially troublesome, except perhaps very limited vocabulary (which is in fact a much more common problem). But that's me...

[...] and since its ancestor Ayasth(i) was heavily revised I'd really like to revise it accordingly before putting it back online. [...] but I ran into a lot of questions, both about what they did the first time around and about how they'd deal with the revisions in Ayasthi.

I think this is a good illustration of why revising a project that has been already published, not to mention the daughters already published by others, is not especially constructive.

From personal experience, whenever I attempted a revision of a conlang of mine, the result was a completely new project.

But revising what I've already done seems like a bit of a pain in the ass, and bending the reconstruction every which way to fit in a few oddball correspondences is not very aesthetically pleasing to me.

IMHO cross-dialectal loans can always save Progress from being halted :)

I don't know if we'll ever get the conworld history filled in up to the fourth millenium YP (or whenever Ghaf was supposed to be) [...]

I'd say that it would be nice to delay "the End of History" as much as possible. After all, the main fun with Akana seems to be all about the various ways in which languages can change and evolve. But profound changes require a lot of time, while the most prolific protolanguages are all relatively recent.

Is there any reason why 2000 CE shouldn't be eqivalent to, say, 4500 YP?

[Cedh]

I don't know if we'll ever get the conworld history filled in up to the fourth millenium YP (or whenever Ghaf was supposed to be) [...]

Is there any reason why 2000 CE shouldn't be eqivalent to, say, 4500 YP?

The only piece of historical information about a modern technology level that we have is found in the Erhadzy grammar, where

The personal computer revolution reached Lesan approximately nine centuries ago

Based on the amount of linguistic change, we have estimated this to a rough equivalence of 2000 CE = c. 2800 YP. There are a few centuries of flexibility in this, and I guess if we needed to, we could even decide to remove the futuristic background from Erhadzy if Zhen Lin so approves. However, what I believe Corumayas really meant with the above statement is that there's so much history to be filled in in earlier periods, which would be needed as a background for later events, that we'll likely keep focusing on the first two millennia for years to come. Or something like that.

[Basilius]

Well, I'll try to help a little.

Thank you :) I think actually we understand each other, so my comments below are just... comments :)

And I copied my list of issues to the Wiki (a bit expanded, but still relating to morphosyntax only).

But you may be looking for more detailed reconstructions than are entirely reasonable.

That's true :) I got interested in the region around Ttiruku mainly because it looked like a crossroads area promising various curious contact scenarios. Now I reap the fruits: it seems that most if not all of the Isles-speaking peoples migrated via Ttiruku, so I need to know how their languages looked in the past to understand what kind of traces they could leave.

These are conlangs that were made fairly quickly for a relay, by conlangers with different amounts of knowledge and experience...

OK, the descriptions are indeed very incomplete, but I don't think they must remain so forever :)

Besides, even for real language families it seems that such detailed reconstructions of syntax are often impossible.

My concerns are about conlanging, not comparative studies. I think with actual reconstruction, the main problem is how to choose among too many probable scenarios basing on too little data. The issues in my list are different: it is supposed that we already have a protolanguage, but I cannot invent a natural-looking scenario for certain changes (which I need to figure out the intermediary stages etc.).

This sounds like a classic (maybe a bit simplistic) case of analogy to me-- the consonant simply got generalized to all forms of the verb, even ones where it used to be deleted. Not a very interesting explanation, but is there another reason why it's unsatisfactory?

I think that analogy needs an example to follow. We have a language that builds a particular tense by adding -s which must substitute for the stem-final consonant if present - *without any exceptions*. Now imagine somebody using a form that does not follow this rule: it just won't sound like "non-sensory present" for the native speakers, so it has no chance to be generalized (IMHO). (It would be a bit different if the daughter lang used a form with additional marking, in whose derivation the original non-sensory present would work just as an intermediary stem; but this isn't the case with Ppãrwak.)

So I was thinking of obtaining such example by an additional change (not supported by any attested material), like VhV(s) -> Vh and VqV(s) -> Vq, with vowel assimilated and -s dropped, but now I'm in doubts.

On another front, the passive constructions in Mûtsipsa' and Zele look like they might represent a shared innovation... and maybe Zele's SVO order is an intermediate step on the way to Máotatšàlì's verb-first syntax?

Maybe. The main problem with non-final V in Zele and Mûtsipsa' is about serial verbs (innovative extraposition of *single* arguments to the position behind the *main* predicate looks less problematic: clefts or somesuch). With Máotatšàlì, OTOH, it seems easier to imagine some optional V-fronting, but this looks rather like an archaism not retained in "Ran's PI".

In general, I believe that every change in WO necessitates finding the answers to two questions: (1) How the new WO became available as an option? and (2) Why the older default WO got marginalized and ultimately banned?

With Zele and Mûtsipsa', I see an easy way to answer the first question, but not the second one. In particular, it is unclear how V-last WO was abandoned with non-final verbs in serial constructions which must have become much more difficult to parse during the intermediary stage.

[Basilius]

However, what I believe Corumayas really meant with the above statement is that there's so much history to be filled in in earlier periods, which would be needed as a background for later events, that we'll likely keep focusing on the first two millennia for years to come. Or something like that.

I understand :)

However, do you really believe this project can ever be finished? Or rather, do you believe it should? And what is "finished", in this case? Even when all the languages spoken from YP 2000 on have been sketched, with decent level of detail for their spread and external history, there'll remain tons of stuff to add, although the focus will inevitably drift to historical events, literature, &like.

OTOH just designing the languages for a dozen dominant nations of the 5th millennium can be finished in a limited time.

[Radius Solis]

I don't think there is ever a desirable "finished" state that Akana should aim towards (let alone achievable!). I don't know what world-level goals the rest of you have, but mine is very simple: keep working on Akana things however much and for however long we find enjoyable.

[Legion]

Okay so. The only projects I think I'll be realistically able to finish some day are Komejech, the Ndak Ta and Komejech scripts, and Nanxútayi (the revision of Thokyunèhòta).

Which means are now officially open for others:

-Kozado (daughter language of Adata, mostly complete, but the lexicon needs expansion).
-Koyek (daughter language of Adata, closely related to Kozado, lacks syntax, lexicon and sample text).
-Proto-Coastal-Western (daughter language of Proto-Western, so far only has sound changes, phonology, and the basis of nominal morphology).
-Agaf (daughter of Ayasthi, needs complete revision since the source language has itself been deeply revised, I have a possible list of sound changes for that revision, but using it is not compulsory)

You'll find all this stuff here: http://thelegion.free.fr/open/

If anyone is willing to work on any of those, and finds anything unclear, do not hesitate to ask.


Basilius > basically, Zhen Lin's conlangs are all around the lines of "über-complex-morpho-phonology", they have large phonemes inventories, intricate sandhi rules, Chechenesque aglutination-inflection systems, and so on. Even with a modest list of sound changes, it's very time-consuming to run all the language through it, to count the buttloads of irregularities it generates, to decide what to change with analogy, and so on.

[jmcd]

If someone redoes Agaf, I'll redo Yiaf (with less vowel phonemes for a start). If noone else redoes Agaf, I'd put myself in for that too. I'm also willing to take on Kozado vocab expansion.

[Basilius]

OK. There are two questions that seem to need comments and approval from the community, and especially from Zompist.

(1) The ephemeral colony of the Affanons somewhere on the coast of Huyfárah, having its hayday around -200 YP.
Does its existence intervene with anything?

(2) Canonization of the name for what we still call "the NE continent".

The Affanons have preserved the memory of their original homeland (i. e. the home of Proto-Isles, which was near the SW corner of the said continent).
The native Affanonic form of its name is Tysaffai (IPA, 3 syllables, stress on -saf-).
It has been suggested (by Radius) that the word was first borrowed into Fáralo, so later it stuck (as the name of the whole continent) in its Fáralo form.
It remains to figure out how specifically it was adapted to Fáralo phonetics.
The following points seem to allow for more than one option:

- vowels, especially the -y- in the first syllable, which (syllable) can in principle become Ti-, Tu-, Tuy-, Tiw-, or Twi- (initial tw- is permitted, cf. e. g. twæno 'avoid, prevent');
- the cluster -ff-; coda -f is permitted in Fáralo at least word-finally, as e. g. in epaf 'light (not heavy)' or labaf 'ignore', which may mean that it could also appear internally in compounds and therefore was not forbidden word-internally in fresh loans.

However, that's Fáralo of around 0 YP, and the word was borrowed some 200 years earlier.

OTOH Ttiruku and Zeluzh(ia) seem to retain their native form (Ppãrwak and Zele respectively, as its seems), so perhaps this was a deliberate practice among the scholars of later epochs? And therefore the Affanonic word can be used as is?

It is mainly the lack of a canonized name for the continent that prevents me from Wikifying some important stuff, e. g. Cedh's theory of three waves.

Zompist: ping?

[Basilius]

[...] basically, Zhen Lin's conlangs are all around the lines of "über-complex-morpho-phonology", they have large phonemes inventories, intricate sandhi rules, Chechenesque aglutination-inflection systems, and so on.

I see. This explains why our perceptions are so different: my previous experience with diachronic conlanging was mostly about well-attested natlangs taken as protolanguages. So Ayāsthi didn't look even moderately complex to me :)

Even with a modest list of sound changes, it's very time-consuming to run all the language through it, to count the buttloads of irregularities it generates, to decide what to change with analogy, and so on.

Sure. IMHO less trivial morphologies just cannot be treated this way. I always try to fit sound changes very cautiously, not to kill too much of the morphology, or to kill exactly what I want to get rid of.

I believe that with natlangs, too, the sound changes do depend on how the morphological machinery works - for example, the sound correspondences between Semitic langs are very different from what one gets used to with IE.

jmcd > Hi, it's nice to see people coming back :)

[zompist]

I don't mind the colony, though you'd better have a good story for how they got there. Back in the original contest, each Isles creator had a specialty as a source of borrowing, and Fáralo was included; it supplied terms for open sea navigation. As no one ever knew where the PI homeland was, this left open the question of how the Isles languages got to their islands in the first place! But they shouldn't have had open sea navigation at the time the languages were described, or they wouldn't have needed Fáralo terms.

(This doesn't preclude coastal navigation, of course.)

As for Tysaffai, I'd say the Fáralo adaptation would be Tuysáfa.

[Corumayas]

I.

I think that analogy needs an example to follow. We have a language that builds a particular tense by adding -s which must substitute for the stem-final consonant if present - *without any exceptions*. Now imagine somebody using a form that does not follow this rule: it just won't sound like "non-sensory present" for the native speakers, so it has no chance to be generalized (IMHO). (It would be a bit different if the daughter lang used a form with additional marking, in whose derivation the original non-sensory present would work just as an intermediary stem; but this isn't the case with Ppãrwak.)

So I was thinking of obtaining such example by an additional change (not supported by any attested material), like VhV(s) -> Vh and VqV(s) -> Vq, with vowel assimilated and -s dropped, but now I'm in doubts.

Ah, I see. Your newer idea, if I understand right, is that Ppãrwak is descended from a dialect of Proto-Isles that didn't delete the stem-final consonant? Since the deletion is presumably the result of a sound change, it seems perfectly reasonable to say that one dialect lacked that change.

On the other hand, if the non-sensory present was a relatively rarely-used form, maybe speakers could forget to delete the consonant and rebuild the tense on the full stem? (It sounds like the kind of mistake language learners might make, to me...)

In general, I believe that every change in WO necessitates finding the answers to two questions: (1) How the new WO became available as an option? and (2) Why the older default WO got marginalized and ultimately banned?

With Zele and Mûtsipsa', I see an easy way to answer the first question, but not the second one. In particular, it is unclear how V-last WO was abandoned with non-final verbs in serial constructions which must have become much more difficult to parse during the intermediary stage.

This is probably a little over my head. I do think that, once any pattern becomes available, it can spread by analogy and take over; but the details of how and why a particular pattern spreads at the expense of another (especially when it comes to syntax) are mostly beyond me.


II.

OK. There are two questions that seem to need comments and approval from the community, and especially from Zompist.

(1) The ephemeral colony of the Affanons somewhere on the coast of Huyfárah, having its hayday around -200 YP.
Does its existence intervene with anything?

Where do the Affanons live? My concern is that, if they still live on the eastern continent, -200 seems too early for a transoceanic colony, however shortlived. (A migration, ok; but a colony in any kind of regular contact with its motherland seems unlikely. I think sailing technology shouldn't be reliable enough for that till sometime in the second millenium YP. Otherwise-- in addition to zompist's concerns-- we may be in danger of having easterners interfering in Peilaš's history too early.)


III.

I'm in the midst of making some revisions to the Eige-Isthmus diachronics. The oddball vowel correspondences are starting to fall into place (and they're actually not quite as random as I thought). Sorting out the consonants should be next.

One thought I've had: to fit in cedh's suggestion of adding PEI *ð *z (giving Miwan v z but Proto-Isthmus *d *dz), maybe there's a full set of voiced fricatives that align with the stops, viz. *β *ð *z. This could hint at an earlier three-way contrast *aspirated - *tenuis - *voiced for stops, with lenition shifting them each down a step to *voiceless stop - *voiced stop - *voiced fricative in PEI. And that would let me leave *f alone (moving it to *θ really doesn't appeal to me), since there's no need for the voiced and voiceless fricative sets to line up if their origins are separate.

I'm not entirely convinced that these fricatives are necessary, but I think I'll add them in parentheses (like the velar nasal and the glottals).


IV.

A related project is a typological comparison of the known protolangs of Eastern Peilaš. Proto-Talo-Edastean, Proto-Eige-Isthmus, and Proto-Hitatc seem to form a loose areal grouping with some shared features, which is nice (and which I'll keep in mind as I work on PEI); Proto-Xoronic, Proto-Peninsular, and Proto-Nualis-Takuña appear somewhat more diverse. I started working through the WALS list of features the other day; maybe the data could go online at CALS. I won't add other people's conlangs to that site without their permission, though.


V.

If someone redoes Agaf, I'll redo Yiaf (with less vowel phonemes for a start). If noone else redoes Agaf, I'd put myself in for that too. I'm also willing to take on Kozado vocab expansion.

I've already tried putting the Ayasthi lexicon and morphology through the original Agaf sound changes, so I probably have a bit of a head start. But maybe we could work on Agaf together somehow?

[Radius Solis]

I don't mind the colony, though you'd better have a good story for how they got there. Back in the original contest, each Isles creator had a specialty as a source of borrowing, and Fáralo was included; it supplied terms for open sea navigation. As no one ever knew where the PI homeland was, this left open the question of how the Isles languages got to their islands in the first place! But they shouldn't have had open sea navigation at the time the languages were described, or they wouldn't have needed Fáralo terms.

All true, but there's open-sea navigation and then there's open-sea navigation. The Fáralo were the ones to figure out how to do it reliably and anywhere, whereas the Isles groups probably just had a technology level comparable to the Polynesians, or even a bit less. So they will still have had a great need for the Fáralo technology, because the Polynesian method is risky and limited to knowledge of regional weather and bird patterns. It is adequate for basic island-hopping across modest distances, and suffices to explain how the Isles diaspora happened, but doesn't let them go running off to the Xsali empire on a whim.

It also suffices to get them to Huyfárah. Since Isles groups clearly have already established themselves in places within visual distance of Siixtaguna (at least the Mûtsinamtsys did), from there all you need is coast-hugging to get to Huyfárah. Slow and tortuous coast-hugging, with all those little islands and jaggy peninsulas, but it will do the job. But I agree with the others that the Affanons settlement would have to be independent, not a colony of a distant group.

[Cedh]

I'd suggest to place the Affanons just outside the Fáralo sphere, for example in the area marked red on the map below. In order for the Fáralo version of the name Tuysáfa to become standardized, it suffices if some Fáralo captain of the Silver Age period (4th century YP) would record the legends of the Affanons. (The map shows the extension of Huyfárah around this time.)
This way, we can also increase the likelihood of daughter languages of Affanonic surviving...

[jmcd]

If someone redoes Agaf, I'll redo Yiaf (with less vowel phonemes for a start). If noone else redoes Agaf, I'd put myself in for that too. I'm also willing to take on Kozado vocab expansion.

I've already tried putting the Ayasthi lexicon and morphology through the original Agaf sound changes, so I probably have a bit of a head start. But maybe we could work on Agaf together somehow?

OK but how would that work though? Would you do some aspects and then send it to me who'd do other aspects? Or would we be PMing each other over the timefor every aspect?

[Legion]

NB: if you're gonna use the old Agaf sound changes list, you should also have a look at the phonology page of Agaf, which contains various allophonic "suggestions" that imo should be included in the sound changes.

[Corumayas]

jmcd: I was thinking we could do it on the wiki. That way it'd be easy for us both to edit it, and we could discuss on the talk page rather than sending tons of pms.


My idea is more or less as follows:

1. run Ayasthi through both Legion's old and new Agaf sound changes and compare them (I can do this)

2. decide on a final set of changes based on what we like best from 1.

3. fill out the morphology and syntax, trying to preserve the features of old Agaf where possible

4. build a lexicon, including the words inherited from Ayasthi but also some newly derived ones and ideally some loans (the loans could be tricky though, since we don't necessarily know what languages would be spoken nearby-- this would require inventing some history)

5. translate the sample text

6. it's all yours to derive "new Yiaf" from


I don't know how long this will all take; I'm definitely interested in doing it, but I'm not in a big hurry, and I'm working on other things too so I don't want it to monopolize my time. Step 1 might take me a week or two, give or take. Some of the other steps could take a lot longer (especially 3 and 4, but they can be worked on simultaneously). And progress may go either faster or slower with two people working on it, it probably depends on how well we communicate...


Legion: I'll include your allophonic rules in the list of old sound changes.

Similarly, the starting point for step 1. should include all the allophony Zhen Lin describes for Ayasthi. Do your new sound changes include those allophonic rules in the changes? Or should they start with those already applied?

Also, are you willing/able to answer questions about how you derived Agaf originally? There were some rules in the old changes I wasn't sure how to apply (and it looks like the same may be true of the new rules... I don't understand some of your notation).


I may have questions for Zhen Lin too...


Hopefully the end result will be a richer, more realistic chain of languages in our branch.

[Legion]

Corumayas > yes, my new sound changes do include Ayasthi allophonic rules. In fact, the changes are divided in three parts: the entire first part is dedicated to Ayasthi allophony (with maybe one or two small simplification of my own), that should help you decipher the (standard but quite advanced) formal notation I used.

For instance the rule:

[+sib] > [αfront βback γvoiced] / _C[αfront βback γvoiced]

Corresponds to:

"Sibilants underwent regressive assimilation for point of articulation and voicing."

Greek letters denote an identical value in both parts of the equation. Thus instead of writing, say:
[+sib] > [+voiced] / _C[+voiced]
[+sib] > [-voiced] / _C[-voiced]

I just write:

[+sib] > [αvoiced] / _C[αvoiced]

And that allows me to cover "regressive vocing assimilation of sibilants" with a single rule.


In short: the allophony is included in the new sound changes (but obviously not in the old ones; I don't know how the sound changes will react if the new allophonic rules are applied, but do as you wish).

Also, don't force yourself to preserve too much of Old Agaf. I was quite happy with the sandhi and umlaut rules I had managed to put together, but they're insanely complex, so I won't be angry if you do not try too hard to replicate them (adding to the fact that were an extention of Ayasth's already quite complex sandhi rules, with additional sound change and analogy). The only part I'd like to see preserved was the peculiar verbal morphology, but if you don't like it, feel free to take it in another direction.

And of course yes, I'm avalaible to answer any precise question you have.

[Corumayas]

Thanks, Legion, that makes it much clearer.

I have a few questions about the old Agaf sound changes. First, can you clarify how the hiatus reduction rules work?

Hiatuses reduction : A short vowel is ellided when in contact with a long vowel. When two long vowel are in contact, the most further away from the accent is deleted. Shwas are ellided. Semi vowels are ellided. Front vowels are ellided. High vowels are elided.

-Can an accented vowel be elided? If so, what happens to the accent?

-Schwas etc. are elided only when in hiatus with another vowel, right?

-Since [w j] were already changed to fricatives, what semivowels would be elided here?

-If two vowels which both fit into the elided categories are in contact, which one gets elided? Do front vowels get elided first, then high ones? If they're both front or both high, what happens?

-If two short vowels which are neither front nor high are in contact, do they both remain?


The rest are smaller questions:

Accent moves on the first long vowel of a word

What happens if there is no long vowel? Does the accent stay where it was?

Some speakers drop word-final unstressed /ø/. Some also drop intertonic (neither in an initial or final syllable) unstressed /ø/.

How would the second rule work, e.g. if there were several unstressed syllables in a row with /ø/?

Though this is regarded as higly incorrect and colloquial by purists, an increasing number of speakers tends to merge the aspirates occlusives with the voiceless ones.

Does this merger produce the pattern you describe in the Appendix (aspirated except at word boundaries)?

The other major sound-change issue is deciding what to do with all the phones that didn't exist in old Ayasth-- [ɔː], [ɯ], [ʉ(ː)], [ɐ(ː)], [ɰ], the ultrashort [ɨ̆] with its post-sibilant allophones [z̩] and [ʒ̩] (which for some reason are mentioned under "timing and syllabification" but not under "allophony"), the rhoticized vowels... I have some ideas for dealing with them, but I'm open to suggestions.


Zhen Lin:

I also have questions about a few of the Ayasthi allophony rules.

High vowels adjacent to lower vowels had "implied" glides:

* [j] was inserted between /i iː/ and another vowel.
* [j] was inserted between /y y:/ and a front vowel.
* [ɰ] was inserted between [ɯ] or /ɨ ɨː/ and another vowel.
* [w] was inserted between or /uː/ and another vowel.
* [w] was inserted between /y yː/ and a back vowel.

-Do the glides only occur after high vowels, or both before and after? E.g., is aúr [ɑ.u˞ ː] or [ɑ.wu˞ ː]?

-Is it right that they only occur between a high vowel and a non-high one, so there's no glide at all in words like èchwy [ɛ.kxu.ɨ]?

-How does this apply to [ʉ] < /wɨ/, if at all?

Otherwise, [j] drops out before /i iː/, [ɰ] drops out before /ɨ ɨː ɯ/, and [w] drops out before and /uː/. (Consequently, /wɯ/ is in isolation.)

What does this mean, in relation to the previous rule? It seems to imply that high vowels could have glides inserted before them, and that such glides would stay but others were deleted. Can you clarify?

Short /ɹ/ in final position elided, leaving a rhotacised vowel.

This means word-final position, right? Are all vowels allophonically rhoticized before /ɹ/, or just word-final ones where it's been elided?


I'll probably have questions about umlaut and sandhi eventually, but this is plenty for now.

[Legion]

Hiatuses reduction : A short vowel is ellided when in contact with a long vowel. When two long vowel are in contact, the most further away from the accent is deleted. Shwas are ellided. Semi vowels are ellided. Front vowels are ellided. High vowels are elided.

-Can an accented vowel be elided? If so, what happens to the accent?

I don't remember, but I'd say: 1) yes and 2) it is transfered on the remaining vowel.

-Schwas etc. are elided only when in hiatus with another vowel, right?

Yes; those rules should be read as a priority order "elide this; if it's not possible, elide this; if it's not possible, elide this, etc".

-Since [w j] were already changed to fricatives, what semivowels would be elided here?

That's probably a redundant rule, you can ignore it.

-If two vowels which both fit into the elided categories are in contact, which one gets elided? Do front vowels get elided first, then high ones? If they're both front or both high, what happens?

Again it's a priority order. However I admit thas way it's imprecise, it should probably be formulated this way: the frontest one is ellided, the highest one is ellided, etc; so if you have /oa/, /o/ is higher than /a/ and so is ellided. Which brings us to...

-If two short vowels which are neither front nor high are in contact, do they both remain?

With the modification above, the only problematic case would be with two identical short vowels... but Ayasth had a Sandhi rules stating that two identical short vowels in contact merge as a long vowel. I don't remember if Ayasthi revision has that rule, but we can import it if so.

The rest are smaller questions:

Accent moves on the first long vowel of a word

What happens if there is no long vowel? Does the accent stay where it was?

Yes.

Some speakers drop word-final unstressed /ø/. Some also drop intertonic (neither in an initial or final syllable) unstressed /ø/.

How would the second rule work, e.g. if there were several unstressed syllables in a row with /ø/?

All of them would be dropped (but knowing that again, this rule is a priority order: if the final /ø/ is dropped, a preceding one would no longer be intertonic and wouldn't be dropped). Including this in sound change obviously require to create some rules to deal with consonants clusters formed this way. I'll leave that to you unless you want specific advices.

Though this is regarded as higly incorrect and colloquial by purists, an increasing number of speakers tends to merge the aspirates occlusives with the voiceless ones.

Does this merger produce the pattern you describe in the Appendix (aspirated except at word boundaries)?

The appendix only describes an alternate *phonemic analysis* of Agaf, based on the standard pronounciation, while this sound change would create an effective *phonetic merger* between the aspirates and plain occlusives. Of course there's a relation between the two ("since aspirate and plain occlusives don't contrast phonemically, let's merge them phonetically as well"), but that's not to be confused.

The other major sound-change issue is deciding what to do with all the phones that didn't exist in old Ayasth-- [ɔː], [ɯ], [ʉ(ː)], [ɐ(ː)], [ɰ], the ultrashort [ɨ̆] with its post-sibilant allophones [z̩] and [ʒ̩] (which for some reason are mentioned under "timing and syllabification" but not under "allophony"), the rhoticized vowels... I have some ideas for dealing with them, but I'm open to suggestions.

If you have ideas I'm fine with it, particularly since I don't have any idea myself :p