Sound Change Quickie Thread

[bradrn]

Would it be reasonable to have ʔ → k between vowels but ʔ → Ø everywhere else? e.g. maʔ → ma, amasʔe → amase, but paʔel → pakel.

I've never seen the glottal stop to fortite to /k/ so not sure.

Ah. I figured that since k → ʔ seems reasonably plausible, so would ʔ → k. I'll redo that one then.

[Whimemsz]

Yeah, debuccalization is extremely common but the reverse essentially never happens. (For glottal stops anyway; [h] can change to a glide or take on some of the features of neighboring vowels so become a fricative at a different POA, and a glide or non-glottal fricative can then undergo fortition to an occlusive, though that's not an especially common change, and seems to be most common word-initially when it does happen.)

[Pabappa]

Would it be reasonable to have ʔ → k between vowels but ʔ → Ø everywhere else? e.g. maʔ → ma, amasʔe → amase, but paʔel → pakel.

I've never seen the glottal stop to fortite to /k/ so not sure.

Ah. I figured that since k → ʔ seems reasonably plausible, so would ʔ → k. I'll redo that one then.

I wouldnt say /ʔ/ > /k/ is impossible, but it's certainly more rare than /k/ > /ʔ/ because it's not a symmetric change .... /k/ has one articulator, /ʔ/ has none.

There are some languages with /Ø/ > /k/, but i dont know offhand what the conditioning environments are. Probably its most common in langauges that dont have voicing distinctions.

[bradrn]

Yeah, debuccalization is extremely common but the reverse essentially never happens. (For glottal stops anyway; [h] can change to a glide or take on some of the features of neighboring vowels so become a fricative at a different POA, and a glide or non-glottal fricative can then undergo fortition to an occlusive, though that's not an especially common change, and seems to be most common word-initially when it does happen.)

Do you have any ideas as to why this is the case? (Then again, Pabappa's symmetricity argument could provide a reason: in debuccalization, we're losing an articulator, so it's not symmetric.)

I wouldnt say /ʔ/ > /k/ is impossible, but it's certainly more rare than /k/ > /ʔ/ because it's not a symmetric change .... /k/ has one articulator, /ʔ/ has none.

There are some languages with /Ø/ > /k/, but i dont know offhand what the conditioning environments are. Probably its most common in langauges that dont have voicing distinctions.

From the Index Diachronica I can find ∅ → k / #s_l and ∅ → k / s_(h)w. It looks like it's most common between consonants to break up clusters.

[Whimemsz]

I'm comfortable saying that a pure change of Ø > k outside of that sort of epenthesis in clusters is impossible, and ditto for any other Ø > obstruent [including nasals] change. The only sort of Ø > consonant change you can have at all is either epenthesis in clusters like that, or epenthesis of a glide or glottal consonant at word or syllable boundaries -- which can later undergo fortition to something more substantial, which is how you get, for example, a few Austronesian languages with word-initial fricatives or /l/ or similar where in an earlier stage of the language the word was vowel-initial (e.g. Proto-Oceanic *qatop "sago palm" > *atop > */jatop/ > Toqabaqita /θao/ [with /θ/ via */ð/ via */j/ via *Ø]; Proto-Oceanic *api "fire" > */japi/ > Motu /lahi/)

[Knit Tie]

Speaking of /h/, I am looking to eliminate it from my conlang entirely. What can I do, aside from merging it with /x/ and eliding it? Turning it into /j/ after /i/ and into /w/ after /u/, followed by turning it into a glottal stop intervocalically and just eliding it in all other environments?

[Salmoneus]

I'm comfortable saying that a pure change of Ø > k outside of that sort of epenthesis in clusters is impossible, and ditto for any other Ø > obstruent [including nasals] change. The only sort of Ø > consonant change you can have at all is either epenthesis in clusters like that, or epenthesis of a glide or glottal consonant at word or syllable boundaries -- which can later undergo fortition to something more substantial, which is how you get, for example, a few Austronesian languages with word-initial fricatives or /l/ or similar where in an earlier stage of the language the word was vowel-initial (e.g. Proto-Oceanic *qatop "sago palm" > *atop > */jatop/ > Toqabaqita /θao/ [with /θ/ via */ð/ via */j/ via *Ø]; Proto-Oceanic *api "fire" > */japi/ > Motu /lahi/)

Austronesian languages also have epenthetic /t/, /s/, /x/ and /N/. Apparently /t/ is probably morphological in origin, and /s/ and /x/ may be derived from /h/, but /N/ looks more difficult, since it's added even in recent loanwords. Palauan 'kuabang' (guava), 'toktang' (doctor), 'biang' (beer), etc. It also has epenthetic initial /N/ (not found in loanwords). Blust admits that a 'careful internal analysis' suggests there may be a morphological cause, but it's evidently not something that's transparent, so probably good enough for conlanger emulation...

Also, even with just glottals and glides, />k would seem possible. There are austronesian languages with epenthetic /g/ and /x/ (the former from /w/, the latter presumably from /h/), which are probably close enough.

Blust also notes, for comparative purposes, that epenthetic /k/ and /t/ are generally accepted for Maru, a Tibeto-Burman language. Again, though, in peripheral locations only.

[dhok]

Unconditional epenthetic word-initial /ŋ/ is also attested in some varieties of Chinese.

[akam chinjir]

In Cantonese you get cases of hypercorrection where ŋ gets inserted before a word-initial vowel (reacting to the widespread loss of initial ŋ). Didn't know there were cases of genuine epenthesis.

[Whimemsz]

I'm comfortable saying that a pure change of Ø > k outside of that sort of epenthesis in clusters is impossible, and ditto for any other Ø > obstruent [including nasals] change. The only sort of Ø > consonant change you can have at all is either epenthesis in clusters like that, or epenthesis of a glide or glottal consonant at word or syllable boundaries -- which can later undergo fortition to something more substantial, which is how you get, for example, a few Austronesian languages with word-initial fricatives or /l/ or similar where in an earlier stage of the language the word was vowel-initial (e.g. Proto-Oceanic *qatop "sago palm" > *atop > */jatop/ > Toqabaqita /θao/ [with /θ/ via */ð/ via */j/ via *Ø]; Proto-Oceanic *api "fire" > */japi/ > Motu /lahi/)

Austronesian languages also have epenthetic /t/, /s/, /x/ and /N/. Apparently /t/ is probably morphological in origin, and /s/ and /x/ may be derived from /h/, but /N/ looks more difficult, since it's added even in recent loanwords. Palauan 'kuabang' (guava), 'toktang' (doctor), 'biang' (beer), etc. It also has epenthetic initial /N/ (not found in loanwords). Blust admits that a 'careful internal analysis' suggests there may be a morphological cause, but it's evidently not something that's transparent, so probably good enough for conlanger emulation...

Also, even with just glottals and glides, />k would seem possible. There are austronesian languages with epenthetic /g/ and /x/ (the former from /w/, the latter presumably from /h/), which are probably close enough.

Blust also notes, for comparative purposes, that epenthetic /k/ and /t/ are generally accepted for Maru, a Tibeto-Burman language. Again, though, in peripheral locations only.

Right, I didn't mean to suggest you couldn't end up with a chain of Ø >>> k, just that a single one-step change of Ø > k can't happen.

The [ŋ] epenthetic processes you and others have mentioned are interesting, though I'm still inclined to think they ultimately derive either from the change of another consonant to ŋ (say, [j] > [ɲ] > [ŋ]), or maybe something like #ØV > #hV > (rhinoglottophilia) #hV~ leading to epenthetic [ŋ]. (Though I realize neither of those specific scenarios work for standard Cantonese.)

[StrangerCoug]

Is it attested to have phonemic /ŋ q/ but lack phonemic /ɴ/?

[bbbosborne]

if i want to merge 2 similar sounds, is one more likely to emerge dominant than the other? for example, if i want to merge /x^w ʍ/ which is more likely to remain?

[Whimemsz]

Is it attested to have phonemic /ŋ q/ but lack phonemic /ɴ/?

Yes, that's attested in multiple languages (take Jakaltek as one random example), because both /ŋ/ and /q/ are vastly more common phonemes than /ɴ/ is; /ɴ/ is virtually non-existent.

UPSID and PHOIBLE and WALS are all unreliable in details, but using them to make broad generalizations about frequencies is probably okay. So. In UPSID's database, 52.55% of languages have /ŋ/, 11.53% have /q/, and /ɴ/ is in only one language out of the 451 in their database (0.22%), Japanese (and Japanese is debatable). In PHOIBLE's database, 56.47% of languages have /ŋ/, 8.59% have /q/, and only 4 out of 2,155 languages (0.19%) -- Japanese, Kinyarwanda, Kusunda, and Kalaallisut (and at least Japanese and Kalaallisut are debatable, see below) -- have /ɴ/. In WALS's database, 50% of the languages for which they have data have /ŋ/, and 15.4% of the languages for which they have data have a uvular stop, though not necessarily /q/. (And they have no information on /ɴ/* but you probably get the picture about /ɴ/.) I'm encouraged by the fact that these percentages are relatively similar among the different databases (with the lack of voicing and secondary feature specification of the uvular stop in WALS probably accounting for the higher "uvular stop" percentage there), so they're likely reasonably close to the real cross-linguistic numbers.


*Actually, in the chapter on feature 6, Ian Maddieson writes:

One further kind of uvular continuant is found in a few languages, a voiced uvular nasal (written in phonetic transcription with the small capital letter [ɴ]). In the present survey only Japanese is considered to have a uvular nasal in its inventory of consonants, but it is a special case and other interpretations are possible. This nasal is the only consonant which can appear in word-final position in Japanese, as in the word /hoɴ/ ‘book’. Since this nasal has a clearly different distribution and pronunciation from the syllable-initial nasal sounds /m, n/, it is usually regarded as a distinct sound in discussions of the sound system of Japanese. Uvular nasals also occur in the dialects spoken by the Inuit peoples of Alaska, Canada and Greenland. The facts are complex but, as described by Rischel (1974) for West Greenlandic, in most cases the uvular nasals seem to be the result of a nasal consonant and a uvular stop "fusing" together into a single sound. For this reason, the uvular nasal is not considered a basic sound of this language although uvular stops and fricatives are.

[linguistcat]

I'm working from Old Japanese and I want to add more nasal sounds and fricatives mostly from prenasalized and unvoiced stops, respectively, to make it sound more "cat-like". But I don't want it to all sound the same once I've done this. Is there a way to derive new stops from other things? There isn't much to work with in OJ's phonology it seems.

[Nortaneous]

UPSID and PHOIBLE and WALS are all unreliable in details, but using them to make broad generalizations about frequencies is probably okay. So. In UPSID's database, 52.55% of languages have /ŋ/, 11.53% have /q/, and /ɴ/ is in only one language out of the 451 in their database (0.22%), Japanese (and Japanese is debatable). In PHOIBLE's database, 56.47% of languages have /ŋ/, 8.59% have /q/, and only 4 out of 2,155 languages (0.19%) -- Japanese, Kinyarwanda, Kusunda, and Kalaallisut (and at least Japanese and Kalaallisut are debatable, see below) -- have /ɴ/

Kusunda is also debatable - it's probably a /ŋʕ/ sequence. Phonemic /ɴ/ has been reported for some dialects of Bai, but I'm not sure how far to trust the data there. Wikipedia says it's also been reported in Mapos Buang.

I'm working from Old Japanese and I want to add more nasal sounds and fricatives mostly from prenasalized and unvoiced stops, respectively, to make it sound more "cat-like". But I don't want it to all sound the same once I've done this. Is there a way to derive new stops from other things? There isn't much to work with in OJ's phonology it seems.

Are you working from Old Japanese or Proto-Japonic? The two aren't the same - some Ryukyuan languages probably conserve vowel contrasts that were lost in OJ. (The short version is OJ merges PJ *e *o into *i *u, but it might have been more complicated than that.)

Either way, there are a few things you could do. Yonaguni has *w- *y- > b- d-. Many Ryukyuan languages develop glottalized consonants from somewhere - I forget, but I think it involves the loss of a mora from an earlier high vowel preceded by a non-nasal consonant.

Apparently there's a dialect of Okinawan where /k/ is glottalized before high vowels, except there's an additional vowel /ɨ/ that's realized identically to /i/ but doesn't condition the glottalization - so [tɕʼi ki kʼu] /ki kɨ ku/ - and then /k/ is [h] elsewhere. That's something you could do. See here and here.

If you do a Ryukyuan and end up needing more vowels, /r/ and vowel sequences tend to be unstable. /r/ can fortite to /d/ (Okinawan sometimes has -rj- > -d-) or drop, and vowel sequences can create new vowels (Okinawan -ee- < *-ai-).

I'm assuming you don't want to develop and lose fricated vowels, so probably the easiest way to do it is to develop conditional glottalization and merge the mid and high vowels, and then drop -r- to create new long mid vowels, and then shorten them somehow.

I don't know of precedent for shifting the OJ 'voiced' series to nasals (aside from OJ /g/), but that doesn't mean it doesn't exist.

Thomas Pellard has a lot of papers on Japonic historical phonology. That might be a good place to start.

[dhok]

Isn't Proto-Bantu supposed to have had partially fricated *i *u, which conditioned fricativization on a preceding consonant and then merged with *ɪ *ʊ?

[linguistcat]

I'm working from Old Japanese and I want to add more nasal sounds and fricatives mostly from prenasalized and unvoiced stops, respectively, to make it sound more "cat-like". But I don't want it to all sound the same once I've done this. Is there a way to derive new stops from other things? There isn't much to work with in OJ's phonology it seems.

Are you working from Old Japanese or Proto-Japonic? The two aren't the same - some Ryukyuan languages probably conserve vowel contrasts that were lost in OJ. (The short version is OJ merges PJ *e *o into *i *u, but it might have been more complicated than that.)

Either way, there are a few things you could do. Yonaguni has *w- *y- > b- d-. Many Ryukyuan languages develop glottalized consonants from somewhere - I forget, but I think it involves the loss of a mora from an earlier high vowel preceded by a non-nasal consonant.

Definitely Old Japanese. There's a lot more to work with and works better with my timeline. Plus from what I've read, Proto-Japonic probably had *b- *d- where Old Japanese had w- j- and Yonaguni was the conservative one of the two. I'd have to go back and find the things I'd read on it.

Apparently there's a dialect of Okinawan where /k/ is glottalized before high vowels, except there's an additional vowel /ɨ/ that's realized identically to /i/ but doesn't condition the glottalization - so [tɕʼi ki kʼu] /ki kɨ ku/ - and then /k/ is [h] elsewhere. That's something you could do. See here and here.

If you do a Ryukyuan and end up needing more vowels, /r/ and vowel sequences tend to be unstable. /r/ can fortite to /d/ (Okinawan sometimes has -rj- > -d-) or drop, and vowel sequences can create new vowels (Okinawan -ee- < *-ai-).

I'm assuming you don't want to develop and lose fricated vowels, so probably the easiest way to do it is to develop conditional glottalization and merge the mid and high vowels, and then drop -r- to create new long mid vowels, and then shorten them somehow.

I don't know enough about fricated vowels to say whether or not I'd want to do that. So I should probably look into them!

I could definitely use vowel qualities that get lost in Japanese to produce more allophones. I might want it to affect things after the vowel, since that might add a bit of unique flavor where most consonant changes in Japanese are either intervocalic or conditioned by the next vowel.

I don't know of precedent for shifting the OJ 'voiced' series to nasals (aside from OJ /g/), but that doesn't mean it doesn't exist.

I thought it was pretty well accepted that the "voiced" stops + /z/ were prenasalized in Old Japanese, but that "unvoiced" stops and /s/ voiced allophonically between vowels, so it might be better to call them plain and nasalized in this context.

Thomas Pellard has a lot of papers on Japonic historical phonology. That might be a good place to start.

Indeed. I read his paper about the differences between Proto-Japonic *e and *o in Eastern and Western Old Japanese. I don't think I've read too many others of his. I look into more of his work. You did give me some things to think about.

[akam chinjir]

I've been relying on a sound change that I'm not sure about, thought it'd be worth asking here. It's ur → au̯ with the r in coda. I'm imagining something like ur → ɨrʷ → əu̯ → au̯---with rounding shifting off of the vowel and onto the coda. Does that seem plausible?

(There's a parallel ar → ai̯ that presumably involves an intermediary aɚ → aə̯; I guess r→ɚ→ə̯ would happen in parallel to rʷ→u̯. It's not a disaster if the changes are stress-sensitive. It's also fine if r → rʷ / u_ is bled by, say, a following coronal. In case it makes a difference, the language will also be losing all other coda consonants.)

[Nortaneous]

Isn't Proto-Bantu supposed to have had partially fricated *i *u, which conditioned fricativization on a preceding consonant and then merged with *ɪ *ʊ?

Yes, cf. Angami P > PF / _ə; presumably ə < *ʮ

[Pabappa]

I've been relying on a sound change that I'm not sure about, thought it'd be worth asking here. It's ur → au̯ with the r in coda. I'm imagining something like ur → ɨrʷ → əu̯ → au̯---with rounding shifting off of the vowel and onto the coda. Does that seem plausible?

(There's a parallel ar → ai̯ that presumably involves an intermediary aɚ → aə̯; I guess r→ɚ→ə̯ would happen in parallel to rʷ→u̯. It's not a disaster if the changes are stress-sensitive. It's also fine if r → rʷ / u_ is bled by, say, a following coronal. In case it makes a difference, the language will also be losing all other coda consonants.)

those make perfect sense. Is your r like the English r, German, or something else? I think this would work best with a sound that s already liquid...Spanish R's would need to change to a liquid before they could start the change.