Sound Change Quickie Thread

[WeepingElf]

What's something interesting I can do to finish off this set of changes?
b d g > β ɦ ɦ /V_V
Nb Nd Ng > m: n: ŋ:
Fb Fd Fg > p t k /#_
Fb Fd Fg > p: t: k: /V_V
b d g > ? ? ?

I want to get rid of voiced stops completely, but I don't want them to change into unvoiced stops again because that's boring.

In one conlang (a descendant of Old Albic) I have the following development.

1. Everywhere except after nasals, and, in case of /d/, after liquids:

b > w
d > r
g > 0

The result is a system like Quenya (the source of my inspiration), in which /b/ and /g/ occur only after homorganic nasals, and /d/ only after /n/, /l/ and /r/.

2. Loss of the remaining voiced stops:

a) After nasals:

mb > mm
nd > nn
ŋg > ŋŋ

b) After liquids:

rd > rr
ld > ll

And the voiced stops are gone.

[8Deer]

While nasals causing nasalisation of preceding vowels is well attested diachronically, is it possible for a nasal to cause following vowels to becoming nasalised, e.g. /arma/ > [ar.mã], /uni/ > [u.nĩ], etc.

Its definitely attested somewhere, I just can't remember for the life of me where. I thought it might have been Scottish Gaelic, but I can't seem to find anything about it.

[----]

Yoruba does that with /m/; cf. /mu/ [mũ˧]

[sangi39]

I just found a source that states that the Uma Juman dialect of Kayan, spoken in northern Sarawak, has progressive nasalisation of vowels after syllable-initial nasal consonants. It also states:

According to the cross-linguistic overview of vowel nasalization in Schourup (1973), progressive vowel nasalization does not appear to be significantly less frequent than regressive vowel nasalization (see also Blust 1997). However, it is certainly the case that nasal effacement is far more typical for syllable-final nasals. It is reasonable to assume that this asymmetry is due to the fact that the syllable-final position is a weak position, and as such disfavours segmental complexity and the presence of consonantal material

So it seems, to me at least, that while phonetic vowel nasalisation might be equally likely to be progressive or regressive, the loss or denasalisation of the consonant which would lead to nasalised vowels becoming phonemic, might be common when that consonant is syllable-final than when it is syllable-initial, which probably explains why VN > V~N > V~ is so much more common than, say, NV > NV~ > DV.

I'd definitely love to include something like /armada/ > [ar.mã.da] > [ar.bã.da] > [ar.bãn.da] > [ar.ban.na] in one of my future conlangs, though, especially where a related language might have the simpler /armada/ > [ar.ma.ta] or something

[zyxw59]

I'd definitely love to include something like /armada/ > [ar.mã.da] > [ar.bã.da] > [ar.bãn.da] > [ar.ban.na] in one of my future conlangs, though, especially where a related language might have the simpler /armada/ > [ar.ma.ta] or something

And the great thing about the /armada/ > /arban.na/ change is that it could repeat, pushing the nasal even further along in the word, i.e. /arban.na/ > [arbana] > [arbanã] > [arbadã] > [arbadãn]

[Buran]

.
Unstressed short /e/ and /o/ are raised to /i/ and /u/ respectively. Unstressed short /a/ is lost (probably becoming [ə] before disappearing entirely). No long vs. short vowel minimal pairs remain; long vowels become short.
Examples:
ˈbeːkʰad > ˈbeːkʰəd > ˈbekʰd
doˈkʰapʰ > duˈkʰapʰ > duˈkʰapʰ
geˈkʰeb > giˈkʰeb > giˈkʰeb
kʰtʰeːs > kʰtʰeːs > kʰtʰes
ˈgtʰeːbkʰe > ˈgtʰeːbkʰi > ˈgtʰebkʰi
leːsˈkʰob > leːsˈkʰob > lesˈkʰob
saːbkʰ > saːbkʰ > sabkʰ

This generates clusters of mixed voicing (as in your first example /ˈbekʰd/. Also what would happen to a word like /pataˈka/ -> /ptka/?

Clusters like that are what generate the /i/ /a/ epenthesis. Besides, it's nothing compared to what I have in Hvaltic, which is sort of like Nuxálk in that every non-plosive can be a nucleus (I was actually inspired by Proto-Indo-European syllabic consonants; I hadn't heard about Nuxálk's madness until about a week ago).

[Nortaneous]

Its definitely attested somewhere, I just can't remember for the life of me where. I thought it might have been Scottish Gaelic, but I can't seem to find anything about it.

Something around there, I think Scottish Gaelic, has /knV/ [kɾṼ].

Not the same thing, but Basque has palatalization of onset coronals after /i/: egin [egin] but egina [egiɲa].

[----]

Weeeeell, that's not entirely a rule. It does happen in like 99% of cases though. Apparently they just put an <h> after it for non-palatalization (e.g. Ainhoa [ainoa])

[Linguist Wannabe]

.
Unstressed short /e/ and /o/ are raised to /i/ and /u/ respectively. Unstressed short /a/ is lost (probably becoming [ə] before disappearing entirely). No long vs. short vowel minimal pairs remain; long vowels become short.
Examples:
ˈbeːkʰad > ˈbeːkʰəd > ˈbekʰd
doˈkʰapʰ > duˈkʰapʰ > duˈkʰapʰ
geˈkʰeb > giˈkʰeb > giˈkʰeb
kʰtʰeːs > kʰtʰeːs > kʰtʰes
ˈgtʰeːbkʰe > ˈgtʰeːbkʰi > ˈgtʰebkʰi
leːsˈkʰob > leːsˈkʰob > lesˈkʰob
saːbkʰ > saːbkʰ > sabkʰ

This generates clusters of mixed voicing (as in your first example /ˈbekʰd/. Also what would happen to a word like /pataˈka/ -> /ptka/?

Clusters like that are what generate the /i/ /a/ epenthesis.

But from your examples, the epenthesis comes after the clusters e.g.

Epenthetic /a/ (probably originally [ɐ], then [a]) is inserted to break up word-initial consonant clusters. Epenthetic /i/ (probably originally [ɨ], then ) is inserted to break up coda stop-consonant clusters.
Examples:
ˈgtʰebkʰi > gɐˈtʰebkʰi > gaˈtʰebkʰi

We still have a cluster of mixed voicing.

[Astraios]

We still have a cluster of mixed voicing.

So??

[jmcd]

It is possible to have clusters of mixed voicing; I'm sure there are examples of such in English for example. But if you really want to avoid them, you can have the epenthetic vowel appear between the two consonants, like what happened in Persian.

[Linguist Wannabe]

I thought that clusters of mixed voicing were extremely unstable cross-linguistically.

[Click]

True, but that doesn't mean they are unattested and if you read through all the sound changes Adjective Recoil provided, you could've seen that he got rid of them.

[Pole, the]

We still have a cluster of mixed voicing.

So??

So…

DIEEEEE INFIDEL

…

…

… kidding.

So your language will look a bit wierdo.

[Aili Meilani]

Given the basic stops /t c k/, is /t c/ > /p t/ plausible?

[Whimemsz]

It depends. c > t doesn't present any problems, but t > p is a really bizarre change. HOWEVER you could add some intermediaries like t > θ > f > p, and that's perfectly plausible. But if you're just picturing it as a one step chain shift, c > t and meanwhile t > p, that's hard to justify.

[Also note Slavey-Hare-Dogrib (Athabaskan), where the dental consonant series changed to a labial series, so you've got correspondences like *ts > kʷ/p, *z > w, etc. I don't know what the intermediary steps there were, but I seriously doubt whether the changes did NOT involve intermediary steps! A number of Athabaskan languages feature a fronting of the dental series, so *ts > tθ, etc., so I would propose one possible sequence in the Slavey-Hare-Dogrib case would be *ts > *tθ > *pf > p; *z > *ð > *v > w; etc.]

[[EDIT: Apparently Cedh already suggested this sequence a little while back]]

[Aili Meilani]

I was afraid that's the case.

So I guess that if I want to fill the gap, I'd have to do e.g. f v θ ð → p b f v, or conditional k g → kʷ gʷ → p b, or find a sensible way to denasalize m, or…

What are some plausible changes that yield p?

[ObsequiousNewt]

Aside from debuccalization and assimilation, how many examples of POA shift can you actually name?

[Whimemsz]

Aside from debuccalization and assimilation, how many examples of POA shift can you actually name?

What do you mean? There's a fair number of examples. This thread lists some, scattered in various posts. t > k is actually kind of common (most common when the language lacks /k/, normally due to earlier debuccalization). It happened independently in various Austronesian subgroups over a dozen times, and it also happened in Blackfoot. p > k is known from Arapahoan. Sa'ban has g > p/ʤ. There's further examples in that thread and in the Correspondence Library. I already mentioned Slavey-Hare-Dogrib in my last post. Many other Athabaskan languages also show POA shifts. Spanish had ʃ > x. Rhotics often change POA. θ > f style changes are very common. You could count the extremely common kw > p, w > g etc. type changes, but I suppose those could also fall under your "assimilation" heading, since they involve, basically, feature coalescence or loss.

Basically, there's nothing unusual about POA shifts, though certainly unconditioned POA shifts are less common than conditioned ones of the "k > tʃ before front vowels" type. They are probably most likely either where a gap exists in the phonological system (e.g., /k/ is missing, so the realization of /t/ expands to include [k], leading to the possibility of a shift in the dominant realization), or where there's an articulatory/auditory basis for the change (e.g., the interchange of θ~f, or interchange of labials and velars).

[Pole, the]

Aside from debuccalization and assimilation, how many examples of POA shift can you actually name?

What do you mean? There's a fair number of examples. This thread lists some, scattered in various posts.

Maybe he meant those with their own names?

[ObsequiousNewt]

Naw, they just seemed rare, from what I saw, and I was wondering whether I could get away with them in the first place.

[Nortaneous]

I was afraid that's the case.

So I guess that if I want to fill the gap, I'd have to do e.g. f v θ ð → p b f v, or conditional k g → kʷ gʷ → p b, or find a sensible way to denasalize m, or…

What are some plausible changes that yield p?

If you have a voicing contrast, you can leave a gap for p and just get b through w or g around rounded vowels or something like that.

[----]

Another kind of interesting POA shift is Burmese /r/->/j/ unconditionally. The /r/ wasn't a trill but instead an approximant, however I am too lazy to copy-paste the IPA symbol.

[Pogostick Man]

Another kind of interesting POA shift is Burmese /r/->/j/ unconditionally. The /r/ wasn't a trill but instead an approximant, however I am too lazy to copy-paste the IPA symbol.

That happened in some forms of Mayan too (Ch'olan, Chujean, and Q'anjob'alan if Wikipedia is to be believed).

[gach]

Also in Chukchi, though only to dissimilate two /r/s in consecutive syllables:

jara- < *ra-ra- ~ ra-
"house"