Sound Change Quickie Thread

[Vijay]

Well, some Romance and Germanic languages at least have /sk/ > [ʃ] before a front vowel.

[Travis B.]

Is the sound change /s/ + voiceless stop > voiceless fricative (e.g., /st/ > /θ/, /sk/ > /x/) attested in some natural language? Caesaraugusta > Zaragoza could be an examßple of /st/ > /θ/ (through Arabic "Saraqusta", I think), but I don't think it was a regular sound change.

I somehow would imagine that that is more like /s/ > /θ/, noting that it happens to both /s/s.

[Zaarin]

Is the sound change /s/ + voiceless stop > voiceless fricative (e.g., /st/ > /θ/, /sk/ > /x/) attested in some natural language? Caesaraugusta > Zaragoza could be an examßple of /st/ > /θ/ (through Arabic "Saraqusta", I think), but I don't think it was a regular sound change.

s > x / _k is certainly attested in Gaulish among others; s > θ / _t is not difficult to imagine. At that point you'd just have to delete the following stop. Another avenue might be st > ʦ > θ and sk > kx > x.

[Pole, the]

Don't some instances of PIE *sk- result in Slavic /x/?

[Pogostick Man]

I know in Gothic, *gs (presumably > ks) > xs. Does that work for you?

[Travis B.]

Also note that in Dutch *sk > sx.

[Atrulfal]

In a system with /p b t d k ʔ/ would it be fine to do k ʔ > g k just to fill the gap ? There was a /g/ earlier in this system, but it became /ɣ/.

[Benturi]

Thanks for your answers! I I could extend /sk/ > /sx/ as in Dutch and apparently Slavic, as Pole mentioned (see R. Matasović's Proto-Indo-European *sk- in Slavic) to /st/ > /sθ/ and then simplify the clusters.

[jmcd]

Another possibility would be creating a series of aspirated fricatives e.g. sk > sx > hx > xʰ

[Travis B.]

In a system with /p b t d k ʔ/ would it be fine to do k ʔ > g k just to fill the gap ? There was a /g/ earlier in this system, but it became /ɣ/.

Umm ʔ does not really become other plosives. If anything happens to it, it is either to disappear (possibly introducing phonemic tone in the process), or to merge with another obstruent to form an ejective or implosive.

Also mind you that [k] tends to be more favored than [g], due to voicing becoming more difficult the further back in the mouth a plosive is, except in environments where voicing would be specifically expected, e.g. intervocalically. I would probably just keep it as /k ʔ/ myself and maybe add an intervocalic allophone [g] of /k/.

[Frislander]

In a system with /p b t d k ʔ/ would it be fine to do k ʔ > g k just to fill the gap ? There was a /g/ earlier in this system, but it became /ɣ/.

Umm ʔ does not really become other plosives. If anything happens to it, it is either to disappear (possibly introducing phonemic tone in the process), or to merge with another obstruent to form an ejective or implosive.

Also mind you that [k] tends to be more favored than [g], due to voicing becoming more difficult the further back in the mouth a plosive is, except in environments where voicing would be specifically expected, e.g. intervocalically. I would probably just keep it as /k ʔ/ myself and maybe add an intervocalic allophone [g] of /k/.

Yeah, just leave it as it is, it's fine.

[CatDoom]

So, let's say you've got a system where certain kinds of phonemic onset clusters are broken up by an epenthetic schwa, such that you might have a word like /rti/ which is pronounced [rəti]. Then you have a sound change where unstressed short vowels elide, so that /rti/ is now pronounced [rti] or [r̩ti]. Would it be plausible to then have a second sound change in which a prothetic vowel is inserted to resolve clusters that violate the sonority hierarchy like that or, alternately, before sonorants which have become syllabic?

The whole sequence of sound changes would then be something like [rəti] > [rti] > [urti] (if the prothetic vowel were ).

[StrangerCoug]

I would expect the elision to be blocked in situations like that, but the answer is yes. Initial clusters starting with /s/ got an /e/ before it in Western Romance.

[Nortaneous]

So, let's say you've got a system where certain kinds of phonemic onset clusters are broken up by an epenthetic schwa, such that you might have a word like /rti/ which is pronounced [rəti]. Then you have a sound change where unstressed short vowels elide, so that /rti/ is now pronounced [rti] or [r̩ti]. Would it be plausible to then have a second sound change in which a prothetic vowel is inserted to resolve clusters that violate the sonority hierarchy like that or, alternately, before sonorants which have become syllabic?

The whole sequence of sound changes would then be something like [rəti] > [rti] > [urti] (if the prothetic vowel were ).

Something like that can happen in the coda, as in Tocharian A: PToch *rätræ > *rätr > *rätär > rtär.

[StrangerCoug]

I have a protolang with an inventory of /a e i ə ɨ o u k kʰ ɠ ŋ t tʰ ɗ n p pʰ ɓ m h s r j w/ and syllable structure of (C)(J)V(C), where J is one of /r j w/. The only restriction on the syllables is that if the J slot is filled, the C slot in the onset can't be any of /r j w/. I want to derive a daughter protolang where /j/ gets lost and leaves behind vowel fronting and lengthening and a palatalized/velarized split in the consonants. The exact changes I want, in the order I want to apply them, are:

1. Cj Cʲ (I'm thinking of having a syllable break block the change and simply drop syllable-initial /j/, but let me think it over.)
2. /e i/ cause all the consonants in the syllable to become palatalized.
3. aj ej ij oj uj eː eː iː œː yː
4. əj ɨj œː yː /[+labial]_
5. əj ɨj eː iː otherwise
6. The consonants that did not become palatalized become velarized.

(I didn't list the change because it's not actually part of the palatalization shift, but it's preceded by a change of w v except after a consonant.)

I'm looking for interesting things I can do with this split. I specifically do not want the palatalized consonants to develop into affricates/postalveolars since I have another branch where a similar environment creates these and I want them kept distinct. Any ideas?

[vokzhen]

Laminal/apical or dental/alveolar split. Soft velars front forcing the soft coronals to retroflex. I'm not aware of changes specifically to soft labials, but Cj clusters fricativized in Chinese and become labial-alveolar or labial-palatal stop clusters in Greek and Polish, which can assimilate or lose the labial part. In your language it would probably make sense for them to break back into clusters, with the offglide being labiolpalatal since you already have a similar change.

[StrangerCoug]

Laminal/apical or dental/alveolar split.

With which corresponding to which?

[vokzhen]

With an actual velarized "hard" series, more than likely the hard series is dental and the soft series alveolar. Without that it could *probably* go either way, depending on the specifics of the sounds involved, though I'm guessing at that. A sound that starts out with the tip of the tongue pointed at the root of the upper teeth might soften to a laminal-alveolar for a hard dental/soft alveolar, while an apico-alveolar might palatalize involving a shift to the tip of the tongue against the upper teeth (lamino-denti-alveolar), against the lower teeth ("lamino-dental?"), or between the teeth (interdental), for a hard alveolar/soft dental. Or the initial palatalization may just be purely an apical(hard)/laminal(soft) in terms of contact, with one or both shifting around (hard forward and/or soft back).

For apical/laminal, palatalized would be laminal.

Irish has an apico-dental /t d n l/ and a lamino-alveolar /tj dj nj lj/. Russian is lamino-denti-alveolar /t d n l/, lamino-alveolar /tj dj nj lj/, and tonguetip-behind-lower-teeth /ts s sj z zj/. Many Central Chadic and Mixe languages have phonemic palatalization, but the only grammars I have that give more detail than "alveolar" don't.

[StrangerCoug]

Thanks for the clarification.

[dhok]

n > ŋ is attested in Andalusian/Caribbean/Central American Spanish, but only in word-final position. Ganan [ˈga.naŋ ~ ɣ̞-].

Slight necro- colloquial Samoan has changed all instances of *n to /ŋ/.

[Porphyrogenitos]

When sound changes occur that lead to initial consonant mutation, do those changes necessarily (or strongly tend to) occur word-internally, as well? I'm well aware that once the word-initial changes become lexicalized/grammaticalized, they'll start occurring at the beginning of all kinds of words simply due to analogy and will cease to bear any strong relationship to still-productive sound changes, but what I mean is, say -

After an article, /p t k/ are all lenited to /f θ h/ because of the V_V environment that's created. Does the change p t k > f θ h / V_V necessarily happen word-internally, too? I've noticed that when it does - especially in a (C)V language - it can lead to some pretty massive consonant loss, especially after some mergers.

Or to rephrase one more time, are there consonant mutation sound changes that uniquely happen just at the beginnings of word?

[WeepingElf]

When sound changes occur that lead to initial consonant mutation, do those changes necessarily (or strongly tend to) occur word-internally, as well?

AFAIK yes.

[Tropylium]

When sound changes occur that lead to initial consonant mutation, do those changes necessarily (or strongly tend to) occur word-internally, as well?

That probably goes for mutation that arises thru lenition, but might not for mutation that arises thru fortition: e.g. ɬ ~ l and r̥ ~ r in Welsh, where *l and *r are the original sounds, but ɬ and r̥ are treated as the "unmutated" grades. Similarly, if you wanted to call final devoicing "final consonant mutation", then that's also an example of a sound change that only happens at the word boundary.

Welsh does seem to have word-medial ll and rh too though; no idea where those came from.

[Pogostick Man]

Welsh does seem to have word-medial ll and rh too though; no idea where those came from.

Dunno about /r̥/, but *l → ɬ / _t.

[Frislander]

Welsh does seem to have word-medial ll and rh too though; no idea where those came from.

Dunno about /r̥/, but *l → ɬ / _t.

I think geminate l might also have received the same treatment, as we have castell, presumably from Latin castellus (have I got the etymology right?)