Question re Origin of Uvulars

[2+3 clusivity]

Does anyone know of a good source on uvular stop diachronics?

My general impression is that uvular approximants and fricatives are commonly formed by backed coronals and velars. Does this also hold true for stops? Is secondary articulation a factor? Any insight greatly appreciated.

[Pabappa]

Yeah that's a good question. All I can give for advice is that the common Semitic /q/ is believed to be a reflex of proto-Semitic /k_>/; that is, a voiceless velar ejective stop. This method works well in a conlang if you only need a /q/, but not so well if you're trying to mirror an entire series of velar consonants over to the uvular side. Japanese uvular nasal is from syllable-final /m/ and /n/, but what made it go all the way to uvular is anyone's guess. I suspect it was actually just vowel nasalization that "hardened" back up to a uvular consonant.

[Hallow XIII]

Japanese doesn't have a uvular nasal, it's just overloading the symbol for underspecification. See also Burmese.

Anyway, for fricatives, velar fricatives often back if there's no contrast. If you want an entire series you can generally do backing before low vowels, which I am sure has happened in a bunch of natlangs but I couldn't give you an example right now.

[vokzhen]

Most of what I know is:
k > q type changes triggered by vowel quality or ʁ (by far the most common afaict)
k' > kˤ > q type changes (Semitic, first stage is in Adyghe-Kabardian)
Unconditioned k' > q' potentially, in Ingush /k'/ is between /k q/ while /q'/ is further back still
r > ʁ type changes, which if they are the only uvulars in the language can (probably rarely) progress to > q at least in some positions to make the system more symmetrical (that one crazy Upper Saxon dialect... Chemnitz).
ʁ > ɢ after nasals
One of the Qiangic languages has a weird change of pre-initial s- > χ- which turns s-ʁ into [ʁɢ] if I'm reading it right. There's only three instances of [ɢ] in the paper, all originating from /sʁ/.
The same paper notes Proto-Tibeto-Burman *kr is reflected as South Qiangic χ(tʂ), and and many ʁ go back to PTB *ŋ(w)a. And less helpfully that PTB velars split into uvulars and velars, with *k ending up as all of *k *q *qh without any clear conditioning factors, but that *q is twice as common as *qh.

Given q > ʔ, you could probably have hypercorrection of ʔ > q to innovate uvulars in new positions. I've used ʔ > q word-initially in parallel to other initial fortition (r > d, w > b) but I have no idea how realistic it is.

Japanese doesn't have a uvular nasal, it's just overloading the symbol for underspecification. See also Burmese.

I thought in Japanese it's actually uvular finally (at least in careful pronunciation)?

[Pogostick Man]

I read awhile ago that there's at least one language (can't remember which or where I read it) that had k > q / _{aʊ,ɔ,o} as a conditioned change.

[2+3 clusivity]

Great! Thanks y'all.

[Tropylium]

Selkup and Ob-Ugric have uvulars that originated from *k + central/back vowel (in the former, not if before close *ɨ *u). This was phonemicized in some of the languages by later sound changes such as:
(Northern Mansi) æ > ɑ
(Western Khanty) ɯ > i
(Selkup) w > kʷ, then kʷ > k / _o
In other varieties, the contrast was only phonemicized by the introduction of Russian loanwords.
Lots of Turkic languages have something similar to this, too.

I recall seeing a seemingly weird change k > q / _V (including front vowels) posited for one Tibeto-Burman language, Lisu IIRC? And after this, clusters like kr kj > k.

(This sounds like a thing that one should be also able to do some Oceanic-like chainshift stuff with: *t *k > *t *q > *k *q?)

Armenian, then, had *ɫ > ʁ.

[Buran]

I read something somewhere (which, most frustratingly, I can't seem to find now) which said that Australian English is in the process of backing /k/ to post-velar before /oː/ and /ɔ/.

[Pabappa]

It says it on Wikipedia, but in that case, by postvelar they dont mean uvular, but just a velar that's farther back then average.

https://en.wikipedia.org/wiki/Voiceless ... currence_2

[Nortaneous]

I recall seeing a seemingly weird change k > q / _V (including front vowels) posited for one Tibeto-Burman language, Lisu IIRC? And after this, clusters like kr kj > k.

Lahu. PLB *k *kj *kr *kl *kw > q c k k p.

Mongolic developed uvulars from velars around low vowels, then phonemicized them somehow -- probably through syncope.

You could also shift l or r to q, since both of those can become ʁ and Chemnitz German has ʁ > q. Maybe even d > ð > ɣ > ʁ > q.

In Truku Seediq: "Diachronically *k in Proto-Atyalic is assimilated to [q] in the Seediq dialects due to the following /h/ or /?/ in the root".

[Pabappa]

Wow, Chemnitz German merged /b ~ p/ and /d ~ t/ but kept /g ~ k/ distinct as /k/ vs /kʰ/ and now it even has /q/! Still though, it does seem to suggest that [q] is the least common pronunciation of the four /r/'s, and I would imagine it'd be most common at the beginning of a word and least common between vowels. Someone wityh access to https://en.wikipedia.org/wiki/Journal_o ... ssociation maybe could check to see if it's a true 100% free variation or not, since they apparently wrote a whole article just about the Chemnitz dialect's phonology.

[Nortaneous]

it's online somewhere, linked from the wiki article on chemnitz german i think

[gach]

Mongolic developed uvulars from velars around low vowels, then phonemicized them somehow -- probably through syncope.

Mongolic has a lot of reduction and loss of unstressed vowels in many of its branches. Here are some examples from Khalkha taken from Svantesson's Mongolian syllable structure (1994, explicitly showing schwas in the phonemic notation for clarity):

<bag> = /pag/ ("team")
<baga> = /paɢ/ ("small")

<ülger> = /ulgər/ ("story")
<argalaa> = /arɢəla/ ("dried dung-REFL.POSS")

[vokzhen]

Fourth down on google search for "Chemnitz phonology"
http://academic.reed.edu/linguistics/kh ... ialect.pdf

[sirdanilot]

I'd want to hear some recordings to believe that uvular stop thing. It sounds to me that we're just dealing with a flapped pronunciation of the uvular fricative which makes it appear like a uvular stop. But hey I am not an expert in German let alone this dialect (I never heard of it until now).

In any case, even if [q] does exist in this dialect, it is not phonemic, not even remotely so. It would be interesting if some other distinction were somehow lost so that this would become phonemic but I doubt it'll happen as [q] here occurs in free variation rather than in specific contexts.

I like the pharyngealization bit though, people are often a bit scared at describing European languages with such sounds but German definitely has some pharyngealization going on around /R/. The entire thing sounds more pharyngeal than uvular to me sometimes.

[sirdanilot]

I think pretty much all possibilities to create /q/ in your conlang have been covered, I couldn't think of another way barring some kind of bizarre chain shift or something. Just k -> q before back vowels, then get rid of your back vowels. If you don't want to assume that your conlang had /q/ since always, then I also see no reason to assume it had a series of ejective consonants including /k'/ or something (which feasibly turns into /q/).

The most simple possibility is that the /q/ was just always there since time immemorial. I mean, why could /k/ and /t/ 'have always been there' but not /q/? It is not such an exotic sound and it occurs in many of the world's languages. If it's not directly apparent where it comes from, then just assume it was there. I mean nobody asked where the emphatic consonants in proto-semitic came from. They were just there, and produced different results in different semitic languages (pharyngealization, ejective, /q/, completely loss of the contrast etc.)

In most western Dutch varieties (including what is commonly deemed 'standard Dutch') /x/ is actually uvular /X/, But I couldn't see how this would ever turn into /q/ even in a thousand years. In my dialect, the /x/ has become /h/ in all positions, but the pronounciation of this /h/ varies. In some position (such as in acht 'eight') the pronounciation comes close to a pharyngeal fricative. Perhaps such a change could with some fantasy ever lead to /q/? At least a pharyngeal is not so far away from /q/.

[Qwynegold]

I thought in Japanese it's actually uvular finally (at least in careful pronunciation)?

Sentence finally is what I've heard.

[Hallow XIII]

The most simple possibility is that the /q/ was just always there since time immemorial. I mean, why could /k/ and /t/ 'have always been there' but not /q/?

Because that view is naive and does not take into account the question of distribution. Sound changes don't act on inventories of phonemes, and generally their effect is less radically changing the phoneme inventory than redistributing phonemes in words. Look at the various Slavic languages: their phonologies are all very similar, but any given segment will by no means necessarily have the same segment corresponding to it in a cognate. So even if your language has had /q/ since "time immemorial", it is very unlikely that Yourlang /q/ always (or even at all!) corresponds to Proto-Yourlang /q/.

[jal]

In most western Dutch varieties (including what is commonly deemed 'standard Dutch') /x/ is actually uvular /X/, But I couldn't see how this would ever turn into /q/ even in a thousand years.

The fact that your imagination is limited, doesn't mean it can't happen. /ɣ/ has become /g/ in most Germanic languages (save Dutch, where it became voiceless and backed in the Western dialects) so fricative -> stop is attested in the language family.


JAL

[vokzhen]

In most western Dutch varieties (including what is commonly deemed 'standard Dutch') /x/ is actually uvular /X/, But I couldn't see how this would ever turn into /q/ even in a thousand years.

The fact that your imagination is limited, doesn't mean it can't happen. /ɣ/ has become /g/ in most Germanic languages (save Dutch, where it became voiceless and backed in the Western dialects) so fricative -> stop is attested in the language family.


JAL

I'm not sure /g ɣ/ is the best example of stop hardening, as far as I can tell the two phones tend to have a strange relationship in a lot of languages. In Germanic most languages have either /ɣ/ (Dutch) or /g/ (most of the others) to the exception of the other, despite otherwise being symmetrical in stops and fricatives, parts of Slavic have a change of g>ɣ (then changed again often, back to /g/ (and inconsistently to /v/ in Russian) or weakened to /ɦ/), again despite stop-fricative symmetry elsewhere, languages without a voice contrast having /p t k q/ [b d ɣ ʁ] as the primary voiced allophones, and similar things.

[sirdanilot]

The most simple possibility is that the /q/ was just always there since time immemorial. I mean, why could /k/ and /t/ 'have always been there' but not /q/?

Because that view is naive and does not take into account the question of distribution. Sound changes don't act on inventories of phonemes,

Then what is a chain shift?

and generally their effect is less radically changing the phoneme inventory than redistributing phonemes in words. Look at the various Slavic languages: their phonologies are all very similar, but any given segment will by no means necessarily have the same segment corresponding to it in a cognate. So even if your language has had /q/ since "time immemorial", it is very unlikely that Yourlang /q/ always (or even at all!) corresponds to Proto-Yourlang /q/.

Why would /q/ be a less stable segment, than say, /t/ or /k/ or /n/? I do not know very much about comparative indo-european linguistics but I thought that /n/ in most languages corresponds to PIE *n. Sound change happens, but it doesn't have to happen only because you are using a non-european sound like /q/.
Yes, /q/ can change (to stuff like a glottal stop or just /k/ or so), but why would it necessarily have to in all positions?

[KathTheDragon]

I'm not sure /g ɣ/ is the best example of stop hardening, as far as I can tell the two phones tend to have a strange relationship in a lot of languages.

Just after Verner's Law, Proto-Germanic would have have /β ð ɣ ɣʷ/ as its voiced obstruents. These all became stops after nasals, and in the case of /β ð/, initially as well. In West Germanic, /ð/ ended up as a stop in all positions. So hardening is very well attested in Germanic.

[gach]

Why would /q/ be a less stable segment, than say, /t/ or /k/ or /n/?

The thing is uvulars and other "laryngeals" or "back the throat" sort of sounds seem demonstrably less stable than many other phones. Janhunen specifically goes to say this about them:

As far as diachrony is concerned, a common feature of laryngeals is that they tend to be short-lived. Laryngeal consonants are conspicuously easily lost from the paradigm, but, at the same time, they are easily replaced by new laryngeals arising from other segments.

Since this family of phones is easily lenited out of existence, there's a clear need to identify the conveyor belt of processes creating more of them.

[Zaarin]

Why would /q/ be a less stable segment, than say, /t/ or /k/ or /n/?

The thing is uvulars and other "laryngeals" or "back the throat" sort of sounds seem demonstrably less stable than many other phones. Janhunen specifically goes to say this about them:

As far as diachrony is concerned, a common feature of laryngeals is that they tend to be short-lived. Laryngeal consonants are conspicuously easily lost from the paradigm, but, at the same time, they are easily replaced by new laryngeals arising from other segments.

Since this family of phones is easily lenited out of existence, there's a clear need to identify the conveyor belt of processes creating more of them.

As far as uvulars go, though, /q/ seems to be relatively stable, especially if it is an areal feature--nearly all the indigenous languages of the Pacific Coast, especially in the Northwest and Arctic, have it, for example--even the ones whose only velar stop is /kʷ/.