Sound Change Quickie Thread

[Emily]

Actually, this isn't really a "quick question" but I also feel like it's too small to be worth making a separate thread for.

So I want to have a sort of "consonant gradation" in my tricon language, except I want it to be non-productive and only occur with the "weak" consonants /w j h ħ/.

In general, it should work like this:

Code: Select all

weakened < normal > hardened
/iː/     < /j/    > /k/~/g/
/uː/     < /w/    > /p/~/b/
/ɒː/     < /h/    > /ŋ/
/ɛː/     < /ħ/    > /ŋ/

I don't really know how to go about justifying this, it's just that for some roots they have derived forms that are different than the rest of the verbal system. I have a general gist of how to achieve what I want. The weakening is a fairly simple diphthongization or compensatory lengthening. As for the hardening of these sounds, well I'm sure that's achievable through gemination e.g. /w: (> β) > b/ but there are also languages where /j w/ become /k p/ or /c k/ when they occur before other consonants.
Nonetheless, I feel like there are a lot of loose-ends that makes this system not feasable.

Anyways, I shall demonstrate with detail how this looks in practice:

For example, the causatives. Normally it's usually either taPRaḪ- or PaRRaḪ-, but for certain root sets you have this:
ī- (√ʾ-y-y) "go" > āg- (√ʾ-h-g) "drive away, cause to flee"
kū- (√k-w-w) "be" > kāb- (√k-h-b) "create, cause to exist, beget"
sā- (√s-h-h) "come" > sāñ- (√s-h-ñ) "cause to come, lead"
sē- (√s-ḥ-ḥ) "punish" > sēñ- (√s-ḥ-ñ) "cause to punish, offend
All of the double final-weak roots, few that they are, behave this way.

malī- (√m-l-y) "take" > malag- (√m-l-g) "cause to take"
malā- (√m-l-h) "heal" > malañ- (√m-l-ñ) "cause to heal"
Although final-weak roots usually use the taPRaḪ- pattern.

ūś- (√ʾ-w-ś) "burn, cook" > abaś- (√ʾ-b-ś) "burn cook"
zūl- (√z-w-l) "be red" > zabal- (√z-b-l) "redden"
ḫūn- (√ḫ-w-n) "be blue" > ḫaban- (√ḫ-b-n) "make blue, bluen"
Although, usually middle-weak roots simply reduplicate the second radical such as:
tīl- (√t-y-l) "perish > talal- (√t-l-l) "destroy, lay waste, make perish"

The aforementioned weak consonants are also vocalized when they are "middle radicals" or occur before strong consonants. Of course this is solely due to diphthongization or compensatory lengthening after elision.
dūl- (√d-w-l) "dig up, reveal"
rīb- (√r-y-b) "plant [seeds]"
ṣāb- (√ṣ-h-b) "kill"
lēb- (√l-ḥ-b) "yield, give up"

Nouns derived from weak roots also show a hard grade, but they are not voiced. These words are mostly doublets of words that didn't have this happen to them:
zūlum "soil, dirt"; zaplum "mud"
ūśim "fire"; apśum "heat, warmth"
ūlum "date"; aplum "fruit"
ḫūnum "sky"; ḫapnum "shroud, covering"
tīśum "tongue"; takśum "word"

There is also pitūm "daughter, girl" and kitīm "son, boy" which I want to have some relationship to the morphemes *wa "female" and *ya "male"

the two things that make this tricky are 1) three of the four "hardened" consonants are velars, and 2) two of the hardened consonants are stops but two are nasals. my first instinct was to say that they should all derive from earlier voiced fricatives (it seems pretty straightforward to have /β/ harden into /b~p/, but also weaken into /w/, and then for /w/ to further lenite to /uː/), but it's hard to come up with an original set that results in what you have here. i would expect earlier /ɣ/, for example, to harden into /g~k/ and soften into /h/ and /ɒː/ (ɣ > h is widespread in slavic languages for example); turning into /ŋ/ seems like it's softening rather than hardening. and in the first row, i would expect /i/ and /j/ to alternate with palatals rather than velars. i guess the answer to this is how much are you willing to change to make it work?

[Eti]

I've been tearing my hair out for way too long for what this issue is worth, so I'm looking to you folks for a hand. I can't figure out the best way to describe the sound that plurals in my language make. To form plurals, a /j/ is appended to the end, but it has a rather abrupt glottal stop. So far, I believe the best way to describe it is "cutting /j/ short with /ʔ/". Take for example the word pronounced /kinj/ - plural of /kin/ - The /j/ at the end is cut off early before the latter part of the sound it makes has time to be heard.

Basically, do a glottal stop about half way through pronouncing /j/. Another crude way to describe it is if you were reading a story, "Y- Y- You did what?" - That cutoff on the word "you" when our unnamed character stutters happens to get the sound sort of right. Put more effort into stopping the "Y-"s and you just about have it.

How would I write that sound? Would something as simple as /jʔ/ work or is that the IPA equivalent of keyboard mashing? Naturally, there may also be a fundamental problem with ending words with /j/ (is there? I wouldn't think so) and so if that's the case then I'll deal with things as needed. Thanks a bunch ahead of time.

[Rounin Ryuuji]

Usually, a supershort vowel is written with a breve, so I don't see why you couldn't transcribe it as [j̆] or the whole thing as [j̆ʔ], though I admit I find [kinj̆ʔ] a bit difficult to articulate as a monosyllable, if I'm understanding how it's supposed to go.

[Pabappa]

i would just recommend going with some other sound, to be honest ..... a word-final cluster of /jʔ/, if it occurs after a consonant, is a trip up and down the sonority hierarchy in a tenth of a second. it begs the question of how you would handle it after stops, too .... how would /pjʔ/ be pronounced, for example, and how it would be distinct from /pj/?

Romanian has word-final palatalization to mark plurals, and it is superficially similar to what youre doing .... it might be a more stable alternative.

[Eti]

Usually, a supershort vowel is written with a breve, so I don't see why you couldn't transcribe it as [j̆] or the whole thing as [j̆ʔ], though I admit I find [kinj̆ʔ] a bit difficult to articulate as a monosyllable, if I'm understanding how it's supposed to go.

i would just recommend going with some other sound, to be honest ..... a word-final cluster of /jʔ/, if it occurs after a consonant, is a trip up and down the sonority hierarchy in a tenth of a second. it begs the question of how you would handle it after stops, too .... how would /pjʔ/ be pronounced, for example, and how it would be distinct from /pj/?

Yeah I did think the difficulty of its pronunciation was a bit of a flaw. In a good score of cases that I've run into, if something is so complicated that I need to put a huge amount of brain power into it, it's absolutely not out of the question to just nuke it and try something else with a different approach.

As far as the question about distinction goes, I don't think it would be very distinct to be honest. After playing with it for a little while, stops at the end would basically need to be aspirated to really say it (/p^hj/) but I'd need to consider it all more.

Romanian has word-final palatalization to mark plurals, and it is superficially similar to what youre doing .... it might be a more stable alternative.

I'll have a look at this, then.

Thanks to you both for your time.

[Jonlang]

I have a conlang, whose current name is just L. Initially, I decided to rid it of voiced plosives because its sister, P has them everywhere. Then, I realised that having no voiced plosives was rather boring and no decided I would allow them only after nasals, so nd, mb, ŋɡ. If I, therefore, insert a nasal before a voiced plosive in some environments, let's say VdV > VndV (where V is any vowel) what would this process be called? In my notes I refer to it as "strengthening" or "plosive strengthening" and, in once instance, "fortifying" which to me is too similar to "fortis" and "fortition".

[bradrn]

Am I alone in thinking there’s nothing wrong with final /jʔ/? It would probably end up pronounced as an ultrashort creaky [ḭ̆], which is perfectly plausible. Admittedly it plays havoc with the sonority hierarchy, but that’s hardly unprecedented — e.g. Japhug has initial /jt/, and even English has /s/+stop clusters which violate it.

I have a conlang, whose current name is just L. Initially, I decided to rid it of voiced plosives because its sister, P has them everywhere. Then, I realised that having no voiced plosives was rather boring and no decided I would allow them only after nasals, so nd, mb, ŋɡ. If I, therefore, insert a nasal before a voiced plosive in some environments, let's say VdV > VndV (where V is any vowel) what would this process be called? In my notes I refer to it as "strengthening" or "plosive strengthening" and, in once instance, "fortifying" which to me is too similar to "fortis" and "fortition".

I’d call this ‘prenasalisation’.

[dɮ the phoneme]

I have a conlang, whose current name is just L. Initially, I decided to rid it of voiced plosives because its sister, P has them everywhere. Then, I realised that having no voiced plosives was rather boring and no decided I would allow them only after nasals, so nd, mb, ŋɡ. If I, therefore, insert a nasal before a voiced plosive in some environments, let's say VdV > VndV (where V is any vowel) what would this process be called? In my notes I refer to it as "strengthening" or "plosive strengthening" and, in once instance, "fortifying" which to me is too similar to "fortis" and "fortition".

I'd just call it prenasalization. Voiced stops spontaneously prenasalizing is attested at least in the environments V_, NV_ and _VN (where N is a nasal). I believe an unconditioned shift of voiced stops > prenasalized stops is also known, but I can't remember for sure. In any case it's basically just another way to realize a fortis/lenis contrast so it makes sense.

[bradrn]

I believe an unconditioned shift of voiced stops > prenasalized stops is also known, but I can't remember for sure.

It must have happened at least once between Proto-Malayo-Polynesian and Proto-Oceanic, since PO has prenasalisation on voiced stops whereas PMP doesn’t.

[Ahzoh]

the two things that make this tricky are 1) three of the four "hardened" consonants are velars, and 2) two of the hardened consonants are stops but two are nasals. my first instinct was to say that they should all derive from earlier voiced fricatives (it seems pretty straightforward to have /β/ harden into /b~p/, but also weaken into /w/, and then for /w/ to further lenite to /uː/), but it's hard to come up with an original set that results in what you have here. i would expect earlier /ɣ/, for example, to harden into /g~k/ and soften into /h/ and /ɒː/ (ɣ > h is widespread in slavic languages for example); turning into /ŋ/ seems like it's softening rather than hardening. and in the first row, i would expect /i/ and /j/ to alternate with palatals rather than velars. i guess the answer to this is how much are you willing to change to make it work?

It has to start from approximants because there already exists fricatives and there shouldn't be a merger of them. Also this isn't really gradation but ordinary consonant mutation.

• /j w/ in coda-position is attested to become /c k/ or /k p/ so I imagine /j: w:/ could either directly become /b g/ or by way of /kj pw/ > /gj bw/
• /ħ h/ msimply elide (with only vowel change and lengthening as a trace of their distinction) and when doubled they both become /ʔ/ which is what becomes /ŋ/ by way of rhinoglottophilia.

[Emily]

where is /w/ > /k/ attested??

[Ahzoh]

where is /w/ > /k/ attested??

Some Sino-Tibetan language has final open /i u/ > /ic uk/
http://www.incatena.org/viewtopic.php?p ... 1#p1132681

But another thread involving me had concluded diphthongs could achieve this as well.

Edit: And I forgot that Armenian did /dw/ > /rk/

[Nortaneous]

where is /w/ > /k/ attested??

according to the index, in Mandekan; also w > kʷ prevocalically in some Salishan languages and w > g in Armenian and Tohono O'odham

[Nortaneous]

Am I alone in thinking there’s nothing wrong with final /jʔ/? It would probably end up pronounced as an ultrashort creaky [ḭ̆], which is perfectly plausible. Admittedly it plays havoc with the sonority hierarchy, but that’s hardly unprecedented — e.g. Japhug has initial /jt/, and even English has /s/+stop clusters which violate it.

I'm not convinced that use of the featural distinction between fricatives and approximants is universal to Sino-Tibetan - there are plenty of consonant charts with perfectly Cartesian fricative voicing contrasts except for holes where the approximants are, which unless there are language-internal reasons to posit separate classes seems like sacrificing descriptive parsimony for adherence to some standard model

[bradrn]

Am I alone in thinking there’s nothing wrong with final /jʔ/? It would probably end up pronounced as an ultrashort creaky [ḭ̆], which is perfectly plausible. Admittedly it plays havoc with the sonority hierarchy, but that’s hardly unprecedented — e.g. Japhug has initial /jt/, and even English has /s/+stop clusters which violate it.

I'm not convinced that use of the featural distinction between fricatives and approximants is universal to Sino-Tibetan - there are plenty of consonant charts with perfectly Cartesian fricative voicing contrasts except for holes where the approximants are, which unless there are language-internal reasons to posit separate classes seems like sacrificing descriptive parsimony for adherence to some standard model

Never mind Sino-Tibetan, even Spanish doesn’t really contrast the two. And I’ve noticed that languages tend not to have voiced fricatives and approximants at the same PoA. And people tend to use symbols like /β ɣ ʕ/ when they really mean [β̞ ɰ ʕ̞]. I’m beginning to suspect that the difference between voiced fricatives and approximants is generally a phonetic one rather than a phonemic one (except for the handful of languages in which they contrast).

[Rounin Ryuuji]

Never mind Sino-Tibetan, even Spanish doesn’t really contrast the two. And I’ve noticed that languages tend not to have voiced fricatives and approximants at the same PoA. And people tend to use symbols like /β ɣ ʕ/ when they really mean [β̞ ɰ ʕ̞]. I’m beginning to suspect that the difference between voiced fricatives and approximants is generally a phonetic one rather than a phonemic one (except for the handful of languages in which they contrast).

Am I understanding right in thinking that such a mechanism would be probably why /ʎ/ tends to merge with /j/ in languages that have both (Spanish and French both seem to have done this), and that having phonemic /j/ and /ʑ/ (or is [ʑ] possibly distinct enough?) might be unstable?

[bradrn]

Am I understanding right in thinking that such a mechanism would be probably why /ʎ/ tends to merge with /j/ in languages that have both (Spanish and French both seem to have done this)

I’m assuming so.

and that having phonemic /j/ and /ʑ/ (or is [ʑ] possibly distinct enough?) might be unstable?

This situation seems fairly stable, presumably since their PoAs are just different enough. On the other hand, it’s very rare to have both /j/ and /ʝ/.

[Nortaneous]

Never mind Sino-Tibetan, even Spanish doesn’t really contrast the two. And I’ve noticed that languages tend not to have voiced fricatives and approximants at the same PoA. And people tend to use symbols like /β ɣ ʕ/ when they really mean [β̞ ɰ ʕ̞]. I’m beginning to suspect that the difference between voiced fricatives and approximants is generally a phonetic one rather than a phonemic one (except for the handful of languages in which they contrast).

It's real in parts of Sino-Tibetan - ɮ/l contrast is occasionally attested - but in practice a phonetic alphabet could get by leaving it consistently underdetermined and disambiguating with diacritics, and for phonology that could be an improvement over IPA.

Then again, ɮ/l is mostly attested in ST from languages too poorly studied for any certainty as to whether it'll hold up, and in Stau /ɬ ɮ/ have plosive release [ɬᵗ ɮᵈ].

Is the Stau contrast real? Geshiza is pretty closely related.
ɮæ - ʑæ 'hand'
vɮi - vʑe 'neck'
vɮɛ - vʑæ 'tongue'
ɮi - ʑe 'wheat'
ɣɮɛ - wʑæ 'four'
ʁɮərko 'bamboo shoot' - wʑə 'bamboo'
zɮæ 'read' - zla 'read aloud'
nɮə - ldə 'heavy'
vɮæzæ - zælvo 'sleeve'

There's only one case I can find of Stau /ɮ/ corresponding to Geshiza /ld/, and it's not ideal, although n- could be a prefix. And Stau doesn't generally permit lC initial clusters, although there's a Tibetic variety (Zanskar Kenhat) where the only initial clusters are /lt ld ltɕ ldʑ xkʲ/ - the l-initial clusters are inherited, but supplemented by zl- > ld-. The logic in this paper could probably be extended to argue that /ɬ ɮ/ in Stau are actually prelateralized stops /ˡt ˡd/.

[Rounin Ryuuji]

I was wondering if that also might be the case. I remember, ages ago, when I looked at Spanish, thinking "yo" sounded more like [ɟjo]. I was surprised to later learn it was phonemically /ʝo/, though I had been perceiving it as /jo/, and had trouble distinguishing [ʝ] from [j] aurally (I still do, in fact); I suppose the similarity of sound must cause them to "trespass" in each-other's space increasingly (for this same reason, I wouldn't expect phonemic /ʑ ʒ/ to last very long), until one pushes the other somewhere else, or they end up merging.

[bradrn]

I was wondering if that also might be the case. I remember, ages ago, when I looked at Spanish, thinking "yo" sounded more like [ɟjo]. I was surprised to later learn it was phonemically /ʝo/, though I had been perceiving it as /jo/, and had trouble distinguishing [ʝ] from [j] aurally (I still do, in fact); I suppose the similarity of sound must cause them to "trespass" in each-other's space increasingly (for this same reason, I wouldn't expect phonemic /ʑ ʒ/ to last very long), until one pushes the other somewhere else, or they end up merging.

It seems that languages with one postalveolar PoA generally have /ʒ/, whereas those with two tend to have /ʐ ʑ/, because */ʐ ʒ/ or */ʒ ʑ/ would simply be too close together. An inventory with all of /ʐ ʒ ʑ/ is extremely rare outside Sino-Tibetan.

(This, by the way, is why I often end up removing your postalveolars in the phrase evolution game; there’s just too many of them, too close together for my liking. Hmm, now I’m wondering what other phonetic universals I’m inadvertently breaking?)