Sound Change Quickie Thread

[Ælfwine]

I am wondering what I should do with clusters involving /r/ in my Vinlandic conlang.

Right now I have a sound change which turns most /r/ into /l/, mimicking the development in several native american languages of the region, perhaps with an intermediate stage of /ɽ/. Compare many Scandinavian dialects which have /ɽ/ from /l/ and /r/, particularly when preceding historical /ð/.

However as Old Norse was a cluster heavy language, this creates a situation where a lot of /tr/, /kr/ etc. become /tl/ and /kl/. Clusters like /tl/ would probably not be tolerated in Vinlandic. So what should I do? Perhaps I could palatalize /tr/ to /t͡ʃ/? Or change /tr/ > /tl/ > /kl/ (I *think* the latter cluster is tolerated in Algonquian languages, but I am not sure?)

[cedh]

I initially thought about suggesting ʔl, but most eastern Algonquian languages (unlike, say, Ojibwe, or Iroquoian languages like Mohawk) do not have a phonemic glottal stop in the first place. Therefore (*pr) *tr *kr > tʃ is probably the most interesting option, especially if your language would have some but relatively few instances of tʃ without this.

Another possibility would be *r > w or j / C_

[Whimemsz]

.

[Ælfwine]

That's what I was thinking.

My hidden worry is that I am going too unrealistically far in fitting ON into a Native American language. Outside of turning the language into a creole I may not need to turn every /kl/ into /tʃ/, or make the language perfectly in line with the languages around it. I'll have to think on this.

[Nortaneous]

kl > tl > ɬ? the first step has happened a few times in Germanic anyway

[bradrn]

kl > tl > ɬ? the first step has happened a few times in Germanic anyway

I know that I have {kl, tl} → /tɬ/ and {gl, dl} → /dɮ/, which I imagine could work well in a North American context.

[Knit Tie]

Can both ʂ and ɕ merge into s when following k or p, but not t?

[WeepingElf]

I have a question on vowel harmony. What do you think is more likely (or has a natlang precedent), a shift from palatal harmony to root retraction harmony, or vice versa?

[Nortaneous]

I have a question on vowel harmony. What do you think is more likely (or has a natlang precedent), a shift from palatal harmony to root retraction harmony, or vice versa?

Vice versa. It's easier for back rounded vowels to front than for front rounded vowels to back. The only proposed case of retraction harmony > palatal harmony that I know of is Mongolic, and that's questionable - and root retraction harmony appears in at least Old Korean and Chukotko-Kamchatkan, so maybe it's a Paleo-Siberian feature that spread. Qiangic sometimes has pharyngealization spreading, which looks a lot like the Chukotko-Kamchatkan dominant/recessive system, but in CK it probably wasn't pharyngealization because IIRC it doesn't interact with pharyngeals.

[WeepingElf]

Thank you, Nortaneous. The direction from root retraction to palatal harmony seems more natural to me, too.

[Darren]

Can both ʂ and ɕ merge into s when following k or p, but not t?

I'm not sure of any natlang precedent, but I think so. /t/ would be able to assimilate in POA (i.e. /ts tʂ tɕ/ → [ts ʈʂ t̠ʲɕ]) while /p k/ couldn't. That would make the t+S more distinct than the k+S or p+S (where S = /s ʂ ɕ/) and thus less likely to merge.

[anteallach]

kl > tl > ɬ? the first step has happened a few times in Germanic anyway

I know that I have {kl, tl} → /tɬ/ and {gl, dl} → /dɮ/, which I imagine could work well in a North American context.

So do you say Dlory to God?

[bradrn]

kl > tl > ɬ? the first step has happened a few times in Germanic anyway

I know that I have {kl, tl} → /tɬ/ and {gl, dl} → /dɮ/, which I imagine could work well in a North American context.

So do you say Dlory to God?

Depends on what you mean by ⟨dl⟩, but yes, although the normative pronunciation is still /gl/ rather than /dl/ for me. When I made the connection I did consider posting there, but was wary of necroing, so I didn’t.

[bradrn]

The protolanguage I’m currently working on has a fairly elaborate set of pronouns: 3 persons and 3 numbers (singular, dual, plural), plus an inclusive-exclusive distinction in the first person non-singular, resulting in 11 pronouns. It also has polypersonal agreement with the subject, direct object and one indirect object of the verb. By my count, this gives 1463 different agreement forms — not counting the reflexive forms, since I haven’t quite figured out all the details of those. At the moment, this is fine, since the forms are constructed regularly. However, I’m worried that when I make a child language, any sound changes I will apply will distort these forms, making them irregular. Since humans are usually not capable of memorising 1463 different irregular forms, what will happen? I know one possibility is that the forms will analogise to become regular again, but I’m sure there are other possibilities.

Also, for the same language, does the following vowel shift look reaosnable?

1. Starting vowels: /a e i o aː eː iː oː/
2. /eː/ → /iː/ (vowels: /a e i o aː iː oː/)
3. /oː o/ → /uː u/ (vowels: /a e i u aː iː uː/)
4. /e/ → /ə/ (vowels: /a ə i u aː iː uː/)

[Pabappa]

yes that makes sense. could throw in a few conditions, but even if you want all three shifts to be unconditional i think it is perfectly justified, and /a i u ā ī ū/ with unpaired /ə/ makes sense as well.

also, it occurred to me you could compress it into two steps if you wanted:
1) /e/ > /ə/
2) /ē ō o/ > /ī ū u/.

The outcome is the same, but the order is different ... the only flaw is that since #2 raises mid vowels to high vowels, one might expect the /ə/ to raise to /ɨ/, so this idea is perhaps best if you want the schwa to be [ɨ] at least as an allophone.

[Richard W]

i don't think you'd have 1463 semantically distinct forms for long, even accepting your combinatoric arithmetic.

For example, if you dug out a table of Hebrew polypersonal perfective forms, you'd find all sorts of semantic neutralisations. I think the contrast between 2pm and 2pf subjects gets lost when an object is added, and I think also 1s v. 2sf subject distinction also goes missing.

Another restriction I recall is that second person arguments have to be the same. Hebrew has the additional complication of having reflexive stems. Finally, comparing different grammars, I got different forms for the same combinations of stem and arguments. Possibly that just shows that synthetic polypersonal agreement was dying in Biblical Hebrew.

[bradrn]

i don't think you'd have 1463 semantically distinct forms for long, even accepting your combinatoric arithmetic.

I’m sure you can, as long as they are made completely regularly. (Which isn’t so difficult: I would say that probably most languages with polypersonal agreement have separate affixes for subject, object etc., so you can easily get to 100+ forms in ‘normal’ polypersonal systems.) My question was: what happens to these forms when they become irregular?

For example, if you dug out a table of Hebrew polypersonal perfective forms, you'd find all sorts of semantic neutralisations. I think the contrast between 2pm and 2pf subjects gets lost when an object is added, and I think also 1s v. 2sf subject distinction also goes missing.

Another restriction I recall is that second person arguments have to be the same. Hebrew has the additional complication of having reflexive stems. Finally, comparing different grammars, I got different forms for the same combinations of stem and arguments. Possibly that just shows that synthetic polypersonal agreement was dying in Biblical Hebrew.

I didn’t know that Hebrew had polypersonal agreement! Do you have any sources I could look at for these ‘semantic restrictions’?

yes that makes sense. could throw in a few conditions, but even if you want all three shifts to be unconditional i think it is perfectly justified, and /a i u ā ī ū/ with unpaired /ə/ makes sense as well.

also, it occurred to me you could compress it into two steps if you wanted:
1) /e/ > /ə/
2) /ē ō o/ > /ī ū u/.

The outcome is the same, but the order is different ... the only flaw is that since #2 raises mid vowels to high vowels, one might expect the /ə/ to raise to /ɨ/, so this idea is perhaps best if you want the schwa to be [ɨ] at least as an allophone.

I quite like this approach as well! I’d be fine with /ɨ/ as an allophone of /ə/, so I’ll definitely consider using this approach.

[linguistcat]

Might mess around, might take all /s z/ in Old Japanese and turn them into /ɬ ɮ/ in most places for my cat lang. Maybe not before /u i/ though, not 100% on those.

[Nortaneous]

could develop new sibilants from raising i > ɿ > ɨ, ɿ > zɨ / V_, ɨ > 0 in various environments; u would parallel this but it's a little more complicated and the output could just merge back into u. high vowels are very unstable in Japonic. Japanese preserved i but probably had u > ɿ > ɯ

[linguistcat]

could develop new sibilants from raising i > ɿ > ɨ, ɿ > zɨ / V_, ɨ > 0 in various environments; u would parallel this but it's a little more complicated and the output could just merge back into u. high vowels are very unstable in Japonic. Japanese preserved i but probably had u > ɿ > ɯ

Very true and I did want to change up the syllable structure, maybe throw in some aspects from the Ryukyuan side of things, including some syllabic consonants.