zompist bboard

Lots of people have diachronics questions, so let's put all the sound change quickies into one thread.

What's the best way to derive a set of affricates along the lines of /tsʰ ts tɬʰ tɬ tʂʰ tʂ tʃʰ tʃ/?

Nortaneous wrote:What's the best way to derive a set of affricates along the lines of /tsʰ ts tɬʰ tɬ tʂʰ tʂ tʃʰ tʃ/?

From what kind of initial inventory?

Nortaneous wrote:What's the best way to derive a set of affricates along the lines of /tsʰ ts tɬʰ tɬ tʂʰ tʂ tʃʰ tʃ/?

Palatalization.

roninbodhisattva wrote:

Nortaneous wrote:What's the best way to derive a set of affricates along the lines of /tsʰ ts tɬʰ tɬ tʂʰ tʂ tʃʰ tʃ/?

From what kind of initial inventory?

Something along the lines of /pʰ p tʰ t kʰ k m n ŋ s h l r j w/. I don't really know what I'm going to do with that protolang. Might have a series of uvulars in there also, and I'll probably end up dropping the aspirated series and deriving them from geminates.

I was thinking of using something like:
l r j > tɬ tʂ cç
ll rr jj > tɬʰ tʂʰ cçʰ
and either starting with /ts/ or getting it from some variation of /s/, but I'm not sure how realistic any of that is.

MadBrain wrote:

Nortaneous wrote:What's the best way to derive a set of affricates along the lines of /tsʰ ts tɬʰ tɬ tʂʰ tʂ tʃʰ tʃ/?

Palatalization.

How? Seems to me that that only takes care of the postalveolars.

Nortaneous wrote:I was thinking of using something like:
l r j > tɬ tʂ cç
ll rr jj > tɬʰ tʂʰ cçʰ

Yeah, this second part make sense to me along the lines of something like rr > tr > tʂ or tʂʰ. I would think the first one would be conditioned? Unless you're going to fill in those open sonorants with some other change, you'd have a weird inventory.

I have a rule in Kohanese that between vowels, consonants lenite (such as t>þ) but that they do not do so after a stressed vowel. Plausible?

dhokarena56 wrote:I have a rule in Kohanese that between vowels, consonants lenite (such as t>þ) but that they do not do so after a stressed vowel. Plausible?

Yes.

That's pretty much the exact condition for Verner's Law, except initial consonants resisted Verner's Law, so it's definitely plausible.

roninbodhisattva wrote:

Nortaneous wrote:I was thinking of using something like:
l r j > tɬ tʂ cç
ll rr jj > tɬʰ tʂʰ cçʰ

Yeah, this second part make sense to me along the lines of something like rr > tr > tʂ or tʂʰ. I would think the first one would be conditioned? Unless you're going to fill in those open sonorants with some other change, you'd have a weird inventory.

Geminated consonants would already all be fortis.

The first one would probably only occur in codas, and I might have something similar happen with nasals: m n ŋ > bm dn gŋ > b d g / _%. (Is that the symbol for a syllable boundary?)

I'd probably also roll up some other clusters into that. I'm thinking maybe something like: (P = plosive)
Pl Pr Pj > tɬ tʂ cç
P:l P:r P:j > tɬʰ tʂʰ cçʰ

Nortaneous wrote:

roninbodhisattva wrote:

Nortaneous wrote:I was thinking of using something like:
l r j > tɬ tʂ cç
ll rr jj > tɬʰ tʂʰ cçʰ

Yeah, this second part make sense to me along the lines of something like rr > tr > tʂ or tʂʰ. I would think the first one would be conditioned? Unless you're going to fill in those open sonorants with some other change, you'd have a weird inventory.

Geminated consonants would already all be fortis.

The first one would probably only occur in codas, and I might have something similar happen with nasals: m n ŋ > bm dn gŋ > b d g / _%. (Is that the symbol for a syllable boundary?)

I'd probably also roll up some other clusters into that. I'm thinking maybe something like: (P = plosive)
Pl Pr Pj > tɬ tʂ cç
P:l P:r P:j > tɬʰ tʂʰ cçʰ

Yeah, now I see this as perfectly reasonable and fine!

I have a language that doesn't allow /w j/ in codas; if they end up there, they become vowels and diphthongize the preceding vowel, so /aj/ becomes /a͡i/. However, diphthongs only form when the second vowel is higher, and so I'm not sure what to do with /iw/ and /uy/. Are /ju/ and /wi/, respectively, plausible developments?

Definitely. Middle English /ew/, /iw/ > Modern English /ju/, for example.

Would it be reasonable to change a vowel pharyngealization contrast into tone?

Yes, definitely.

Nortaneous wrote:

roninbodhisattva wrote:

Nortaneous wrote:What's the best way to derive a set of affricates along the lines of /tsʰ ts tɬʰ tɬ tʂʰ tʂ tʃʰ tʃ/?

From what kind of initial inventory?

Something along the lines of /pʰ p tʰ t kʰ k m n ŋ s h l r j w/. I don't really know what I'm going to do with that protolang. Might have a series of uvulars in there also, and I'll probably end up dropping the aspirated series and deriving them from geminates.

I was thinking of using something like:
l r j > tɬ tʂ cç
ll rr jj > tɬʰ tʂʰ cçʰ
and either starting with /ts/ or getting it from some variation of /s/, but I'm not sure how realistic any of that is.

MadBrain wrote:

Nortaneous wrote:What's the best way to derive a set of affricates along the lines of /tsʰ ts tɬʰ tɬ tʂʰ tʂ tʃʰ tʃ/?

Palatalization.

How? Seems to me that that only takes care of the postalveolars.

/t<C>/ clusters. /ts tl tr tj/ maybe to /t^s t^K t^s` t^S/. /ths thl thr thj/ to /t^sh t^Kh t^s`h t^Sh/. No idea if this is plausible, but I have considered it for a descendant of Chambren. Using canSAMPA.

Canepari wrote:

Nortaneous wrote:

roninbodhisattva wrote:

Nortaneous wrote:What's the best way to derive a set of affricates along the lines of /tsʰ ts tɬʰ tɬ tʂʰ tʂ tʃʰ tʃ/?

From what kind of initial inventory?

Something along the lines of /pʰ p tʰ t kʰ k m n ŋ s h l r j w/. I don't really know what I'm going to do with that protolang. Might have a series of uvulars in there also, and I'll probably end up dropping the aspirated series and deriving them from geminates.

I was thinking of using something like:
l r j > tɬ tʂ cç
ll rr jj > tɬʰ tʂʰ cçʰ
and either starting with /ts/ or getting it from some variation of /s/, but I'm not sure how realistic any of that is.

MadBrain wrote:

Nortaneous wrote:What's the best way to derive a set of affricates along the lines of /tsʰ ts tɬʰ tɬ tʂʰ tʂ tʃʰ tʃ/?

Palatalization.

How? Seems to me that that only takes care of the postalveolars.

/t<C>/ clusters. /ts tl tr tj/ maybe to /t^s t^K t^s` t^S/. /ths thl thr thj/ to /t^sh t^Kh t^s`h t^Sh/. No idea if this is plausible, but I have considered it for a descendant of Chambren. Using canSAMPA.

Oh for heaven's sake. I was going to say this, basically, but do you really have to ruin it by making up a transcription system that's pointless and nobody else uses?

Even then I think you're using the wrong brackets.

Canepari wrote:/t<C>/ clusters. /ts tl tr tj/ maybe to /t^s t^K t^s` t^S/. /ths thl thr thj/ to /t^sh t^Kh t^s`h t^Sh/. No idea if this is plausible, but I have considered it for a descendant of Chambren. Using canSAMPA.

Sounds realistic enough to me. Also, can* is for phonetic detail; X-SAMPA and IPA work well enough on the phonemic level. Don't hammer in the screw.

(Although, having said that, canSAMPA does have a few features that would be useful for writing the phonemic level; ^ is definitely something that I wish X-SAMPA had, and it could easily be extended to, say, /t^s^h/ for IPA /t͡sʰ/, so as not to limit the amount of diacritics available to write additional phonetic segments contained in the same phonemic segment. And yeah, that can be done with X-SAMPA <)> or CXS/Z-SAMPA/whichever one I'm thinking of <+>, but those are ugly and rather unintuitive, and the X-SAMPA tiebar doesn't handle strings of more than two characters well. <ts)h)>?)

Nortaneous wrote:What's the best way to derive a set of affricates along the lines of /tsʰ ts tɬʰ tɬ tʂʰ tʂ tʃʰ tʃ/?

I don't know whether it is the best, but in Brishkan (an Albic language under development) I have these (and some other) changes:

tʰs > tsʰ
ks > tʂ
kʰs > tʂʰ
gz > dʐ
tj > tʃ
tʰj > tʃʰ
dj > dʒ
kl > tɬ
kʰl > tɬʰ
gl > dl
tl > tɬ
tʰl > tɬʰ
tʰr > tr̥ʰ
tr > tr̥
kʰr > tr̥ʰ
kr > tr̥
gr > dr

All the affricates you want, plus some rhotic ones.

How realistic is */r ʀ/ /d g/? What about */ð/ /ɣ/?

What about */ð/ /ɣ/?

This happened in Irish. They don't use the spelling <dh> to be perverse. Well, not only to be perverse.

ná'oolkiłí wrote:How realistic is */r ʀ/ /d g/? What about */ð/ /ɣ/?

/ð/ > /ɣ/ happened in all three Goidelic languages. /r/ > /d/ makes sense, especially if the rhotic is flapped (a flap is basically a very brief stop), and /ʀ/ > /g/ is much the same farther back in the mouth, so I see no problem with that either. I have seen weirder changes.

Nortaneous wrote:

Canepari wrote:/t<C>/ clusters. /ts tl tr tj/ maybe to /t^s t^K t^s` t^S/. /ths thl thr thj/ to /t^sh t^Kh t^s`h t^Sh/. No idea if this is plausible, but I have considered it for a descendant of Chambren. Using canSAMPA.

Sounds realistic enough to me. Also, can* is for phonetic detail; X-SAMPA and IPA work well enough on the phonemic level. Don't hammer in the screw.

(Although, having said that, canSAMPA does have a few features that would be useful for writing the phonemic level; ^ is definitely something that I wish X-SAMPA had, and it could easily be extended to, say, /t^s^h/ for IPA /t͡sʰ/, so as not to limit the amount of diacritics available to write additional phonetic segments contained in the same phonemic segment. And yeah, that can be done with X-SAMPA <)> or CXS/Z-SAMPA/whichever one I'm thinking of <+>, but those are ugly and rather unintuitive, and the X-SAMPA tiebar doesn't handle strings of more than two characters well. <ts)h)>?)

That's why I will use it from now on, to preach the Gospel according to Saint Lucian...

ná'oolkiłí wrote:How realistic is */r ʀ/ /d g/?

I think r >: d is attested in Malayo-Polynesian (there's a large area where *d and *r merge, sometimes to /r/, sometimes to /d/).

For Nort's affricate issue, consider deriving some of the affricates secondarily from the others. Maybe first *t *tʲ *ts by some sort of palatalization, or *ks ts, then assimilations: *ts-l tɬ-l, or *lts *rts tɬ tʂ.

(It is of course entirely acceptable to also start with a bunch of affricates in the protolang and have them merge into something or the other elsewhere.)

Is it at all plausible to have a series of palatals but no velars (except allophonically)? If so, how might it come about? Fronting of velars?