Zompist Bboard Again

New forum, new thread. Post your sound change questions here folks. Old thread by dhok here.

Does it make sense to collapse /l/ in the coda with /n/? Some people have cited Thai as an example, but I have found little discussion of this in linguistic papers.

Yes it does. Codas are very collapsible and /l/ and /n/ aren't very far from each other (both are coronal sonorants)

Yes, it does. It happened in Hittite too, albeit word initally*. PBase mentions Koromfé as a language where /l/ >[n] happens as a synchronic change in phrase final coda's.

*Apparently Wiktionary thinks the dissimilation was *m -> l. Judge for yourself.

l > n also happens in some Algonquian languages, I believe.


How about unconditional /g/ > /ʒ/ > /j/?

How about unconditional /g/ > /ʒ/ > /j/?

Arabic famously has unconditional /g/ > /dʒ/ and leniting that to /j/ seems pretty straightforward.

How about unconditional ʃ ʒ > ç ʝ?

Definitely can see that

If a language undergoes a sound change that collapses /a/ and /ɒ/ into a single phoneme, how would the new phoneme best be written? Or would it depend on what the new "default" for the phoneme is?

Most people would write /a/. I generally use what I imagine the most common realisation of a phoneme is (not necessarily the prototype) unless if I'm lazy.

From what I've seen, if a language has just one open vowel, it will be transcribed /a/

How about unconditional /g/ > /ʒ/ > /j/?

[/quote]Im not sure I believe in an unconditional shift like this. Semitic languages have shifts that are tied to their morphology, and are unlikely to occur in all environments in other languages. This would probably work best if the lanbga\\age has only a few types of environments ... e.g. it's either all CVCV, or allCVCVC but with grammar that can expose the final C's to vowerls as well.

but if the language has a syllable structure comparable to that of Standard Average European, i would advise against this. the first step /g/ > /ʒ/ is possible, but i dont think an unconditional /ʒ/ > /j/ can happen if it would result in consonant clusters flipping sonority. e.g. /glas/ > */jlas/. So if you dont have these types of clutsrs to begin with i htink it would be OK, or if you allow exceptions for rare caseslike these clusters it would be OK.

mèþru wrote: ↑Wed Jul 11, 2018 2:57 pm Most people would write /a/. I generally use what I imagine the most common realisation of a phoneme is (not necessarily the prototype) unless if I'm lazy.

Ars Lande wrote: ↑Wed Jul 11, 2018 3:37 pm From what I've seen, if a language has just one open vowel, it will be transcribed /a/

I was leaning toward /a/ myself and will just make notes about it.

Pabappa wrote: ↑Wed Jul 11, 2018 4:04 pm

How about unconditional /g/ > /ʒ/ > /j/?

Im not sure I believe in an unconditional shift like this. Semitic languages have shifts that are tied to their morphology, and are unlikely to occur in all environments in other languages. This would probably work best if the lanbga\\age has only a few types of environments ... e.g. it's either all CVCV, or allCVCVC but with grammar that can expose the final C's to vowerls as well.

but if the language has a syllable structure comparable to that of Standard Average European, i would advise against this. the first step /g/ > /ʒ/ is possible, but i dont think an unconditional /ʒ/ > /j/ can happen if it would result in consonant clusters flipping sonority. e.g. /glas/ > */jlas/. So if you dont have these types of clutsrs to begin with i htink it would be OK, or if you allow exceptions for rare caseslike these clusters it would be OK.

Actually, Central Franconian (or at least some varieties of it) do have /g/ (or /ɣ/?) > /j/ resulting in clusters like /jl/.

Pabappa wrote: ↑Wed Jul 11, 2018 4:04 pm

How about unconditional /g/ > /ʒ/ > /j/?

Im not sure I believe in an unconditional shift like this. Semitic languages have shifts that are tied to their morphology, and are unlikely to occur in all environments in other languages. This would probably work best if the lanbga\\age has only a few types of environments ... e.g. it's either all CVCV, or allCVCVC but with grammar that can expose the final C's to vowerls as well.

but if the language has a syllable structure comparable to that of Standard Average European, i would advise against this. the first step /g/ > /ʒ/ is possible, but i dont think an unconditional /ʒ/ > /j/ can happen if it would result in consonant clusters flipping sonority. e.g. /glas/ > */jlas/. So if you dont have these types of clutsrs to begin with i htink it would be OK, or if you allow exceptions for rare caseslike these clusters it would be OK.

Well, if it helps the language in question is indeed triliteral and thus highly analogized, and it also has a syllable structure of CVC, thus no initial clusters.

Another issue in my project is glide hardening. I have long planned on the glides /v j/ hardening to obstruents in certain positions. It seems obvious that /v/ would harden to /p/ given their shared place, but it seems less clear what obstruent /j/ should become. Presumably it should harden into a palatal stop, but the language has no such phoneme. Thus far, I have considered hardening it to /s/ since that seems the closest equivalent and grouping /s/ with /j/ greatly simplifies my plans for a featural script. Yet that seems difficult to justify without invoking tenuous similarities (both are laminal coronals, /j/ often turns other consonants into sibilants, and so forth).

No, I think that works.

Consider PIE *ty > proto-Greek *ts > Attic s

/s/ works; /j/ > /z/ is certainly attested. /k/ could also work, I think.

I found a book preview on Google Books that mentions some connections between /s/ and palatals in passing. Apparently some languages in South America even have [s] as an allophone of /j/, although the preview doesn't include the name of the reference for this.

*v *j > b ɟ is attested in Nganasan. /d g/ only occur natively from consonant gradation.

In Nivkh, there's initial lenition in certain environments. /tʰ t/ lenite to /r̥ r/, and /cʰ c/ lenite to /s z/.

There doesn't have to be a firm distinction between fricatives and affricates; some languages in East Asia don't make this distinction, and have word-initial clusters that look like /rb-/, /jl-/, etc.