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

[Radius Solis]

Huh. It must have slipped my attention when that was posted, if indeed it was, as I've never seen that list or had any idea it existed. Not that it was really a waste of time... the wikification is worthwhile, as you say... and besides which point, word category and noun class will be important for morphological comparison.

[dunomapuka]

I have another complication to throw into the general puzzle of this part of the world and its chronology.

I noticed that the Mutsipsa words borrowed into Fáralo have undergone ALL of the sound changes from NT to Fáralo. This would not, on the surface of it, make sense, because Muts. and Fá. are supposed to be roughly contemporaneous. I would like us to resolve this issue, in case we want to borrow more Mutsipsa or Takuña words into Fáralo.

PROPOSED RETCON: These words were written according to NT rules of pronunciation. Much later, a spelling pronunciation was adopted.

ALTERNATIVE: Mutsipsa would need to be, like, a millennium earlier.

[Corumayas]

Huh. It must have slipped my attention when that was posted, if indeed it was, as I've never seen that list or had any idea it existed. Not that it was really a waste of time... the wikification is worthwhile, as you say... and besides which point, word category and noun class will be important for morphological comparison.

It was about 9 posts ago.

It might be good if you keep working independently though, in case you come to different conclusions about the source of a few words.

The words cedh didn't find etymologies for mostly fall into a few categories: there are placenames (which seem to be compounds, but some are quite opaque), words for topography (e.g. pxakelihi 'sand dunes'-- which must be a loan, Mûtsipsa' doesn't have /l/!), plants and animals, metals, military and political terminology, some philosophical terms (including dux 'master, sensei'), and a few basic words-- daxa 'strong', tsuh 'push, move, knock' (maybe < PI datuh 'hit'?), and û 'kick'.

I'm guessing that there are loans from Thokyunèhotà (a few words seem to be from PI roots but with different sound changes), and some might be from Takuña. There might also be more from Máotatšàlì, though cedh already identified several of those. Hell, there could be some from Fáralo for that matter...


boy #12: Oh dear... the loans might've taken place a few centuries ago, when contact was first made between Siixtaguna and Huyfárah; and maybe the rate of change in Fáralo has picked up since then. But all the sound changes? Seems like that would put the loans at 1500 years ago or more.

[Radius Solis]

It was about 9 posts ago.

***double take***

I must be blind.

I noticed that the Mutsipsa words borrowed into Fáralo have undergone ALL of the sound changes from NT to Fáralo.

Well... it's theoretically possible (wayyy out on a limb but possible) for all of those sound changes to remain synchronically active as phonological rules of Fáralo... not that I have any idea what exposure it would have had to most of the conditioning environments once the originals were changed, to keep the rules active. Usually a sound change sticks around as an active rule only so long as there are morphophonological alternations that result from it, which provide a continuing source of the material that triggers the rule. Occasionally it happens instead by language contact - a rule remains active because loanwords having the sounds to be changed keep arriving from a contact language. But I can think of no such explanations for the bulk of Fáralo's rules remaining in force.

[zompist]

I noticed that the Mutsipsa words borrowed into Fáralo have undergone ALL of the sound changes from NT to Fáralo. This would not, on the surface of it, make sense, because Muts. and Fá. are supposed to be roughly contemporaneous. I would like us to resolve this issue, in case we want to borrow more Mutsipsa or Takuña words into Fáralo.

Not really; the Mûts. words were all modified by hand, not treated as NT words.

If you look at the timeline, Etugeist monks were brought to Huyfárah 600 years before the time of the Fáralo grammar. That's plenty of time for change, especially on strange foreign words.

[Basilius]

While the discussion of the Isles languages is going on and zompist is here, I'd like to ask a question that bothers me.

The Proto-Isles phoneme h is classified as pharyngeal in the Grammar and is supplied with a phonetic transcription /ħ/.

However, I observe that:

- in Ppãrwak, PI h uniformly yields a velar [k] syllable-finally;
- in Zele at least one of its reflexes is a uvular r /R/, including word-initially (ruzh rurune 'the wind blows' etc.);
- Máotatšàlì (always?) reflects word-final h as š irrespective of the quality of the preceding vowel.

To be fair, Máotatšàlì š can also reflect the PI glottal stop q (intervocalic and sometimes word-initial), so its appearance will sometimes need a tricky explanation anyway. Mûtsipsa' reflexes raise less suspicion (a couple instances of word-initial x seem to be due to a former palatalization), and I don't know (but am really curious) what Legion is planning for his revision of Thokyunèhotà (Nanxútayi / Sošunanxútayi).

Anyway, for me a development like pharyngeal -> uvular -> velar is much less probable than exactly the other way round. If I were *reconstructing* Proto-Isles from its daughters, I would postulate a velar x (if not š or something more exotical). Or perhaps I'd assigned different qualities to syllable-initial and syllable-final h. But I wouldn't even consider /ħ/ for such a series of reflexes, honestly.

What do other people think (and zompist in particular)?

(And why do I have the feeling that I've already seen this issue discussed somewhere?)

[dunomapuka]

I think there should be lots more Isles daughters in the long run, so we can design some that actually have /ħ/.

In keeping with this spirit, the Thokyunehota loans in Mûtsipsa' can be assigned to some other Isles language, since Thok. is being remodeled.

And as for Mûtsipsa'/Fáralo, I stand corrected... I think things might be a little smoother, then, if we move the date of the Mûts. grammar back a few hundred years. Any real reason it needs to be 0-200 YP? On the other hand, the language need not change very much in the course of those few centuries, so maybe the difference between the dates doesn't matter much.

I would like to do my own Isles language at some point.

[Basilius]

I think there should be lots more Isles daughters in the long run, so we can design some that actually have /ħ/.

No, absence of attested /ħ/ is not the problem which I'm trying to point to.

Examples of this PI phoneme reflected as /h/ or zero are not lacking in the "attested" daughters. The trouble is that while /x/ > /χ/, or /x/ > /h/ > zero (with or without /ħ/ as an intermediary step) is commonplace and can affect all or most of languages in a group (cf. Germanic), the reverse development (except via palatalization or somesuch process of narrow scope) looks difficult and dubious. It appears still more bizarre that it happened in more than one daughter lang.

So my question was, in fact, whether people feel it to be important that PI h is exactly /ħ/.

[Basilius]

I would like to do my own Isles language at some point.

I believe there's a lot of possibilities for that.

The Isles-speaking peoples known to date fled from some invasion, but it can be doubted if all the Proto-Isles speakers who stayed in their original homeland were assimilated by the invaders. Besides, some could flee in a different direction - e. g. to the north, along the western coast of their continent.

Also, I'm inclined to think that all the known Isles-speaking peoples migrated via Ttiruku (at least nobody objected to this last time I asked), and Ttiruku with surroundings is planned to become a natural reserve of linguistic diversity. So any of the known languages of the Isles stock can have sisters spoken in a few villages on Ttiruku or nearby.

[zompist]

Anyway, for me a development like pharyngeal -> uvular -> velar is much less probable than exactly the other way round. If I were *reconstructing* Proto-Isles from its daughters, I would postulate a velar x (if not š or something more exotical). Or perhaps I'd assigned different qualities to syllable-initial and syllable-final h. But I wouldn't even consider /ħ/ for such a series of reflexes, honestly.

Proto-Isles is Ran's language. However, the discrepancy between expected /x/ and actual /ħ/ is just the sort of surprise one might find in comparing a reconstruction to an actual attested parent language.

The easiest solution is to posit an ħ x change early enough that it affected the entire family; I think this would be natural enough as it makes the phonology more symmetrical (consolidating two POAs).

[Basilius]

The u-uu discrepancy, however, I can think of no mechanism to account for. Surely it wouldn't depend on sibilancy of the preceding consonant?

Doesn't seem unimaginable (e. g. a somewhat fronted articulation, which caused an early monophthongization, which affected the resulting vowel quantity), but cedh's list also has:

súy 'who, what' > suu
tuy 'leg' > tu

(BTW, with this addition the list of relevant material seems to be complete.)

Monophthongization after dentals/alveolars only where the tone was HL?

Looking now at polysyllables, we also see two outcomes. But not the same two! Instead we see u and zero.

dúduy - duutu (lake)
pumuy - pwu (fish)
wánuy - wen (bad)
hunuy - hun (blow)

I notice that both zero reflexes are after n. And, given the rather peculiar consonant development in "fish", this can actually mean "after any resonant". For example, the vowel was deleted at a stage where resonants were allowed word-finally while t wasn't.

Contrary to what has already been suggested, I would force everything I could into one set of sound changes, no matter how narrow some contexts will be. It will never be too late to resort to cross-dialectal loans, but elaborating consistent sound changes for several dialects can prove even more complicated :)

[Basilius]

The easiest solution is to posit an ħ :> x change early enough that it affected the entire family; I think this would be natural enough as it makes the phonology more symmetrical (consolidating two POAs).

I know! I know! It was a *pharyngealized velar*!
The time-machine owner who did the fieldwork for her PhD thesis just wasn't a professional phonetist, so she misheard it :)
(And I already know how to invent a historical explanation for such a sound in a lang with no pharyngealization contrast...)

[Radius Solis]

You don't need an explanation, really. Many languages make some use of "minor" features that are not systematically contrastive; sometimes it's there merely to help out an existing contrast. Examples: many American English varieties have a consistently pharyngealized /r/ (especially in the West), and most everybody consistently preglottalizes fortis plosives in coda position. These things just happen.

[Radius Solis]

Map stuff.

Looking at the Mûtsipsa' culture/history material, some things stand out to me about the island group:

- There is a larger "main" island, two small nearby islands, and one distant one;
- One of the islands, Dûkejdih, was small enough and the area geologically active enough that it was possible for an earthquake to sink "over half" of the island;
- explicit mention is made in the philosophy section about how they view earthquakes and volcanoes, strongly suggesting that the island group is subject to both
- the archipelago is not, by all prior mappings that I'm aware of, located near any tectonic boundaries

Putting two and two together, I arrive at the conclusion that this can't be just any old type of island configuration, it has to be a chain of islands caused by a mid-plate hotspot. The classic example of such a situation is Hawaii, which consists of a large island at one end of the chain hosting several active volcanoes, with several smaller, older islands that host extinct volcanoes stretching in a line from the big island in the direction of plate motion. The Hawaiian Islands are also prone to earthquakes that can be quite large (8+ in magnitude), and in extremely rare events, major chunks of the islands have sunk. They're just big piles of volcanic detritus, after all, not completely stable land. The largest known example: the northern half of the island of Moloka'i once slid into the sea all in one go; the debris bed from this landslide is scattered over a hundred miles of Pacific ocean floor (google a bathymetric map, it's impressive). Smaller but similar events happen every few hundred thousand years in the islands, apparently.

Sound like just about what we want for the Mûtsinamtsys islands? If so, I would like to redraw the island chain's map to better fit this scenario.


edit: google maps has bathymetry now, I just remembered. Here's the Hawaiian Islands. Molokai is the unlabeled long narrow island in the middle of the central group; the rough-looking part of the ocean floor to its north, which is dead-center on that link, is the debris bed.

[Cedh]

I'll have a try at the most important sound changes for Mûtsipsa'. Note that some exceptions or further changes with limited scope may be missing, and that the order of things is likely not completely accurate:


m, j, w > 0 / _# (with a few exceptions)
h, ? > 0 / VV_#
? > 0 / #_

dz > ts

h > x / _i

wa, aw > o
wi, iw > y
ja, aj > e
ju, uj > û

á, áj, áw > e, ej, y
í, íw > ii, yy
ú, új > uu, ûû

ji, wu > ii, uu
ij, uw > ii, uu (sporadically?)

jy > iy

m, p, t ,k, ? > w, f, s, x, h / V_V
d, g > t, k / V_V
0 > ? / V_V

V > 0 / VC_# (conditions unclear, possibly after sonorants only)
V > 0 / #(p,t,ts,k,?)_(w,f,s,x,h)V, #t_tsV, #(t)s_NV

Long vowels remain intact with both syncope rules.

----

Map stuff.

The hotspot thing is a good observation. I'll include it in the next version of the Peilaš map. It requires only minor changes, the general direction of plate movement (SSW) is already about right for the islands in question (currently aligned WSW).

[Salmoneus]

Er... the collapse of half an island is a major historical event. All coastal and near-coastal cities on the bay will have been devastated. The last major Hawai'ian collapse put debris down 1000ft high on nearby islands, and caused a 15m-high tsunami on the west coast of Australia.

We talk about this now mostly in the context of the western flank of Cumbre Vieja, which is expected to collapse in the (geologically) near future. According to the predictions, if the whole western flank of this volcano, one of several on the island, were to collapse, it would create a tsunami 900m high. Even by the time it reached North America (having crossed the Atlantic faster than a jumbo jet), the tsunami would still be FIFTY METRES HIGH.


It's hard to imagine anybody surviving on the island, and there would be amazingly high death tolls on neighbouring islands - look at the indian ocean tsunami a few years ago, and consider that that was only a little tsunami caused by a minor landslide.

I think that either this account should be changed or it should be specified that the island was extremely small all along. Either way, it should show up in other histories.



[People tend to forget just how large islands are, and that volume rises faster than area. As a demonstration, I read somewhere that if we wanted to move the unstable rock on Cumbre Viejo's west side, it would take us around 30 million years to do it by truck. It's a mind-boggling amount of rock]

[Radius Solis]

The hotspot thing is a good observation. I'll include it in the next version of the Peilaš map. It requires only minor changes, the general direction of plate movement (SSW) is already about right for the islands in question (currently aligned WSW).

Well, I'd like to remap the islands, as hotspot-generated chains take fairly specific forms. Taking a roughly SSW plate motion, something like this is what I've got in mind.

[Radius Solis]

Er... the collapse of half an island is a major historical event. All coastal and near-coastal cities on the bay will have been devastated. The last major Hawai'ian collapse put debris down 1000ft high on nearby islands, and caused a 15m-high tsunami on the west coast of Australia.

I know. I ignored that because it's Rory's work and he is unlikely to wish to be reached for comment on the matter. I can think of three ways to address this; we probably need at least two of them:

1. the natives conflated the amount of the island that sunk (big but far less than half) with the amount inundated and denuded by the resulting tsunami (easily more than half), the tsunami explaining why only two villages were left on an island that's supposed to be substantial.

And 2. it is possible for the other islands and the Siixtaguna coast to escape the main force of the tsunami, provided the landslide was directed toward the open ocean, away from all other nearby land. With landslide-generated tsunamis, as opposed to megathrust earthquake tsunamis like 2004's, the great bulk of their force tends to be directed to one side (the direction the landslide fell) and areas "behind" this can get away with onlap heights a tenth what they'd for coasts on the landslide side. So it's entirely possible for a landslide tsunami from Mûtsinamtsys to cause greater damage a thousand miles away in Ttiruku than on the other islands of its own chain. Nowhere nearby is going to escape at least some slosh, but it doesn't have to be armageddonlike.

3. Rugged coastlines where elevation rises quickly from the water are greatly less susceptible to tsunami damage.

[Corumayas]

? > 0 / #_

I don't think this one is correct. Comparing your wordlist with Rory's original, I found that the spreadsheet software deletes <'> at the beginning of entries. At least some word-initial /?/ survived.

Er... the collapse of half an island is a major historical event. All coastal and near-coastal cities on the bay will have been devastated. The last major Hawai'ian collapse put debris down 1000ft high on nearby islands, and caused a 15m-high tsunami on the west coast of Australia.

I know. I ignored that because it's Rory's work and he is unlikely to wish to be reached for comment on the matter. I can think of three ways to address this; we probably need at least two of them: ...

I don't see why we can't incorporate this as a major historical event, though. It probably happens fairly early in the first millenium BP-- a period when our history is still not very detailed. An event like this wouldn't have to hurt much in the long term; but it might add some interest if it affects Huyfarah and Kasca-- maybe destroying a coastal city or three and contributing to the "dark age"?


EDIT: Also:

I think there should be lots more Isles daughters in the long run....

I agree.

In keeping with this spirit, the Thokyunehota loans in Mûtsipsa' can be assigned to some other Isles language, since Thok. is being remodeled.

Of course, that requires being able to identify which words were originally Thokyunehota loans. There's a couple words with Lotoka etymologies which were probably from there, and I can see hints of what might be Thok. changes to Proto-Isles roots; but it'd be great if Legion could confirm which Mûtsipsa' words were from his language.

And as for Mûtsipsa'/Fáralo, I stand corrected... I think things might be a little smoother, then, if we move the date of the Mûts. grammar back a few hundred years. Any real reason it needs to be 0-200 YP? On the other hand, the language need not change very much in the course of those few centuries, so maybe the difference between the dates doesn't matter much.

Maybe it's not essential, but I'd prefer not to push the Isles languages very much earlier... otherwise we might have to move other things too (e.g., moving the Isles people's migrations would mean also moving Proto-Nualis-Takuna earlier).

The Isles-speaking peoples known to date fled from some invasion, but it can be doubted if all the Proto-Isles speakers who stayed in their original homeland were assimilated by the invaders. Besides, some could flee in a different direction - e. g. to the north, along the western coast of their continent.

Also, I'm inclined to think that all the known Isles-speaking peoples migrated via Ttiruku (at least nobody objected to this last time I asked), and Ttiruku with surroundings is planned to become a natural reserve of linguistic diversity. So any of the known languages of the Isles stock can have sisters spoken in a few villages on Ttiruku or nearby.

This makes sense (though I wouldn't expect very close sisters of Mutsipsa' et al. in Ttiruku-- I'd guess that all the western Isles langs might form a genetic subgroup).

Presumably there are Isles langs spread all along the long island chain from east of Ttiruku to Ik'im, as well as on the lone island in the middle of the sea north of Ttiruku, and probably on at least some of the islands scattered along the plate boundary between Peilash and Zeluzhia to the southwest.
(I'm looking at cedh's latest map: )

Contrary to what has already been suggested, I would force everything I could into one set of sound changes, no matter how narrow some contexts will be. It will never be too late to resort to cross-dialectal loans, but elaborating consistent sound changes for several dialects can prove even more complicated

Well, sure. I think we should do as rigorous a reconstruction as possible, taking note of exceptions and alternate developments wherever they occur (like the one I did for Adāta).

[Cedh]

? > 0 / #_

I don't think this one is correct. Comparing your wordlist with Rory's original, I found that the spreadsheet software deletes <'> at the beginning of entries. At least some word-initial /?/ survived.

Damn it, you're right. I didn't even look at the lexicon part of Rory's grammar once I had copied the words to the spreadsheet, so I completely missed this software bug.
(Corrected in the SC list above.)

[Ran]

Basilius has asked for my comments, so here goes:


1) I don't claim any exclusive right to edit Proto-Isles. If you guys want to change it, flesh it out, etc., feel free.

2) The quality of /h/: when I first made up the three codas /h ? s/ I had in mind East Asian tonogenesis and was expecting these to be turned into tones, phonation, or some other suprasegmental thing, or simply gotten rid of. I didn't expect them to be strengthened into /x/, /S/, etc. I realize that h > x and h > S are slightly unusual. In the interest of realism, I suppose it would make more sense to reconstruct, say, /x k s/ instead of /h ? s/. Perhaps a parent language had a richer system of finals, with /p t c k ? f s S x h/ etc., that collapsed, with /p t c k ?/ > /k/ or /?/, /s S/ > /s/, and /f x h/ > /h/. That would explain the paucity of allowed codas as opposed to onsets.

3) With regards to reduplication, certainly different dialects of PI could have used different pitch patterns. IIRC, my version of PI had only the HH pitch pattern for reduplicated words; so another way to look at it would be to say that the pitch distinction is neutralized in reduplicated words, with different PI dialects using different surface realizations.

4) As for derivatives, there really aren't any systematic derivations for the derivate-like words. I made up all the words on the fly (since it's supposed to be a protolanguage with no known parent languages, after all). Of course, there are a fair number of roots that were reused, like hapa- and duy-.

[TzirTzi]

Yay, Takuña grammar! ^^

I'm sure there are lots of holes - indeed, I realise that the lexis section is only partially finished and the derivational morphology needs an overhall - but that's a good core of it there.

Plus I've added quite a bit to the lexicon.

[Basilius]

Ran > Thank you so much! (and sorry for not responding earlier!)

The quality of /h/: when I first made up the three codas /h ? s/ I had in mind East Asian tonogenesis and was expecting these to be turned into tones, phonation, or some other suprasegmental thing, or simply gotten rid of.

I felt that :) I suspect Ppãrwak could have some suprasegmental reflexes of PI syllable-final -s. (Or at least it can be re-interpreted this way.)

In the interest of realism, I suppose it would make more sense to reconstruct, say, /x k s/ instead of /h ? s/.

Since you are giving us a carte-blanche for revisions, I hurriedly propose that we preserve some sensible conservatism :) Altering the coda-to-onset mappings looks like a major rearrangement, so let's see if we have a real reason for that in the first place.

As for my problem with the reflexes of h, I think a pharyngealized velar /xˤ/ is already a solution. While I was kinda joking when I proposed it, there was a reason why this idea made me so sprightly. The matter is that pharyngealization can have both velarizing (/uvularizing) and palatalizing effect on adjacent sounds, sometimes it the same dialect. So /xˤ/ can easily produce a uvular reflex in one daughter language and š in another. Besides, the same type of palatalizing pharyngealization can help to explain other strange instances of š in a dialect for which such pharyngealization has already been postulated.

With regards to reduplication, certainly different dialects of PI could have used different pitch patterns. IIRC, my version of PI had only the HH pitch pattern for reduplicated words; so another way to look at it would be to say that the pitch distinction is neutralized in reduplicated words, with different PI dialects using different surface realizations.

What is important for me right now is that therefore in the common ancestor of such dialcts the reduplicates were prosodically different from the other words. While the exact difference remains to be determined, it can already be used to explain e. g. the inverted tonal reflexes in Ppãrwak.

As for derivatives, there really aren't any systematic derivations for the derivate-like words. I made up all the words on the fly (since it's supposed to be a protolanguage with no known parent languages, after all). Of course, there are a fair number of roots that were reused, like hapa- and duy-.

Therefore, we have a nice challenge of reconstructing some derivation patterns basing on the existing material :) I already have a couple ideas (will post later)...

[Basilius]

You don't need an explanation, really.

"Need" is not the right word here. Rather, a detail like that can shed light on still earlier stages. And I am already missing some "Proto-Macro-Isles": the last wave of the NW-to-SE migration that brought Proto-Isles speakers to the NE continent must have left a trace on Ttiruku. Besides, the people who ousted PI speakers from their homeland could speak a related language, and they are among first candidates for "semi-successful invaders" in my scenario about Ttiruku and vicinities.

Many languages make some use of "minor" features that are not systematically contrastive; sometimes it's there merely to help out an existing contrast. Examples: many American English varieties have a consistently pharyngealized /r/ (especially in the West), and most everybody consistently preglottalizes fortis plosives in coda position. These things just happen.

Didn't know about that pharyngealized /r/, thank you! But such minor features are interwoven with others... Thus, glottalization helps to distinguish voiceless from voiced, labialization and pharyngealization on [r] may stress the difference from e. g. flapped intervocalic allophones of [d] and [t] (don't the areas overlap, BTW?)... Also, they seem to have a long (pre-)history in this particular language, and parallels in related tongues. So they can tell something important, still.

This makes sense (though I wouldn't expect very close sisters of Mutsipsa' et al. in Ttiruku-- I'd guess that all the western Isles langs might form a genetic subgroup).

What if we try to *discover* such groupings first, based on purely linguistic things? Like, earliest sound changes that happened to each language? It would be a lot of fun if we discover that Mutsipsa' 's closest relative is e. g. Zele :)

Since earliest dialectal splits must have taken place still on the NE Continent, and then all the migrations passed (consecutively?) through the bottleneck of Ttiruku...

[Basilius]

TzirTzi >
A question... as I understand, PNT (or somewhat more recent) vowels developed to Takuña as follows: /ii uu/ > [iː uː], /ia ua/ > /ii uu/ (phonetically [ɪj ʊw] or [ej ow]). That is, a bit different from what comes to mind first :) Perhaps, this point needs some explication...

Otherwise, thank you for a very pleasant reading: so many features I'd never think of myself, assembled into something logical and coherent :) also, references to "closely related varieties" sound promising...

Not really related... I came across a real-world orthography that uses ñ for [ŋ]: the official Romanization of Kazakh. It is much less used than cyrillics presently, but the official website of Government of Kazakhstan still supports a version in such orthography.