Sound Change Critique Thread

[bradrn]

Inspired by Zju's comment from the Sound Change Quickie Thread:

I want you to poke holes at this:

Code: Select all

p t k
s ʃ
r l
m n

a o u e i

k > q / _{a, o}, {a, o}_
p t k q s ʃ r l m n > pʲ tʲ kʲ qʲ ɕ ɕ j j mʲ nʲ / _i
ɕ > sʲ
i > e / [+J]_
l > w
qʲ > cˠ > tˠ
t > ɾ / V_V
s n t > sˠ nˠ tˠ
w ʃ > wˠ ʃˠ > ʕ ħ
r q > z q > zˠ ʡ
j > ʝ > z
e o > i u
i u > e o / _Cˠ, Cˠ_
pʲ mʲ sʲ tʲ kʲ nʲ > p m s t k n

Code: Select all

m	n	nˠ
p	t	tˠ	k	ʡ
	s	sˠ		ħ
	z	zˠ		ʕ
	ɾ

a i u

This is a thread for the critiquing of whole sets of sound changes. If you have a question about the plausibility of a single sound change, approaches to get from one sound change to another, or similar questions, use the Sound Change Quickie Thread; if you already have a whole set of sound changes, and want to know if they're all plausible, you haven't included implausibly many of one type of change, or similar questions, this is the thread for you.

I'll start with my own:

Code: Select all

m  n
p  t  k  ʔ
b  d  ɡ
   ts
   dz
   s  x
   z  ɣ
w  ɹ  j
   l

i  ɯ u
e    o
æ    a

p b t d k ɡ > f v s z x ɣ / V_V
x ɣ > j
w > v / V_C or V_#
æ e i ɯ > e i ɨ ɨ
ej > i / _#
i > ɨ / j_ or _j
s z > ʃ ʒ / _i
consonants next to each other turn into geminates (e.g. inkɨ > inːɨ)
V > ∅ / _#
ts dz > s z
j > ∅ / V_C
ʔ > ∅ / next to consonants or at word end

Code: Select all

m  n
p  t     k  ʔ
b  d     ɡ
f  s  ʃ
v  z  ʒ
w  ɹ     j
   l

i ɨ u
e   o
  a

[Xwtek]

w > v / V_C or V_#

I bit suspect this is not that realistic. For example, phoneme like /iwk/ can turn in /ivk/ which is harder to pronounce. Probably it's better if w>f

æ e i ɯ > e i ɨ ɨ

i > ɨ is unrealistic unconditionally. You should restrict this.

ej > i / _#
i > ɨ / j_ or _j

æj ej > i ɨj /_# sounds strange to me. I think they should both collapse into /i/

s z > ʃ ʒ / _i

This also sounds strange unless the affricate is also affected.

consonants next to each other turn into geminates (e.g. inkɨ > inːɨ)

Please explain this. This is not extensive enough.

Code: Select all

m  n
p  t     k  ʔ
b  d     ɡ
   s  ʃ
   z  ʒ
w  ɹ     j
   l

i ɨ u
e   o
  a

[/quote]

f and v are also phonemic, too. Because p b ʔp ʔb > f v p b.

[Zaarin]

w > v / V_C or V_#

I bit suspect this is not that realistic. For example, phoneme like /iwk/ can turn in /ivk/ which is harder to pronounce. Probably it's better if w>f

b > v / _C happened in Neo-Punic. w > v unconditionally is so trivially common as to be not worth mentioning. /ivk/ is not at all hard for me to pronounce, but it could easily be allophonically [ifk] anyway.

[Xwtek]

w > v / V_C or V_#

I bit suspect this is not that realistic. For example, phoneme like /iwk/ can turn in /ivk/ which is harder to pronounce. Probably it's better if w>f

b > v / _C happened in Neo-Punic. w > v unconditionally is so trivially common as to be not worth mentioning. /ivk/ is not at all hard for me to pronounce, but it could easily be allophonically [ifk] anyway.

If w > v unconditionally, well, you're right. However, the exact changes w > v /_C is suspect.

[Zaarin]

I bit suspect this is not that realistic. For example, phoneme like /iwk/ can turn in /ivk/ which is harder to pronounce. Probably it's better if w>f

b > v / _C happened in Neo-Punic. w > v unconditionally is so trivially common as to be not worth mentioning. /ivk/ is not at all hard for me to pronounce, but it could easily be allophonically [ifk] anyway.

If w > v unconditionally, well, you're right. However, the exact changes w > v /_C is suspect.

I realize it's my Anglophone mouth, but I actually find semivowels in coda very difficult to pronounce outside of diphthongs. Your example of /iwk/ is nearly impossible for me to pronounce as anything other than /yk/ or /uk/ or at best /ikʷ/, for instance. Fortition of coda semivowels consequently seems the most natural thing in the world to me--or at least the only thing more natural would be diphthong collapse.

[mèþru]

Also things like /vk/ are no challenge for one who speaks Hebrew

[Pabappa]

uncodnitional coda /u/ > /v~f/ happenedi n Greek, but some of thw words, like Zeus becoming /zefs/ , are loans from Ancient Gk .... the inherited form of that name is in fact días (Δίας).

[bradrn]

Thanks everyone! I still have a few questions though:

w > v / V_C or V_#

I bit suspect this is not that realistic. For example, phoneme like /iwk/ can turn in /ivk/ which is harder to pronounce. Probably it's better if w>f

From the other posts there seems to be a bit of disagreement about this. Personally, I find /ivk/ very easy to say, but /iwk/ much harder. On the other hand, it does have a tendency to devoice, so you're probably right here.

æ e i ɯ > e i ɨ ɨ

i > ɨ is unrealistic unconditionally. You should restrict this.

It was originally intended to be a chain shift: æ > e > i > ɨ, with ɯ collapsing into ɨ as well. Now that you mention it though, it does seem a little strange.

ej > i / _#
i > ɨ / j_ or _j

æj ej > i ɨj /_# sounds strange to me. I think they should both collapse into /i/

Does this matter as much though? There's plenty of examples of sets of perfectly sane changes which do weird things together.

s z > ʃ ʒ / _i

This also sounds strange unless the affricate is also affected.

Good point.

consonants next to each other turn into geminates (e.g. inkɨ > inːɨ)

Please explain this. This is not extensive enough.

I'm not sure what you mean by this. If anything, this seems too extensive - it applies to all consonants next to each other!

Code: Select all

m  n
p  t     k  ʔ
b  d     ɡ
   s  ʃ
   z  ʒ
w  ɹ     j
   l

i ɨ u
e   o
  a

f and v are also phonemic, too. Because p b ʔp ʔb > f v p b.

Good point - that's just an oversight on my part. But ʔp ʔb don't do anything special - it's VpV VbV which turn into VfV VvV.

EDIT: This oversight has been fixed on my original post.

[Xwtek]

ej > i / _#
i > ɨ / j_ or _j

æj ej > i ɨj /_# sounds strange to me. I think they should both collapse into /i/

Does this matter as much though? There's plenty of examples of sets of perfectly sane changes which do weird things together.

Now that you say that, you're probably right. and I'm probably wrong here.

consonants next to each other turn into geminates (e.g. inkɨ > inːɨ)

Please explain this. This is not extensive enough.

I'm not sure what you mean by this. If anything, this seems too extensive - it applies to all consonants next to each other!

I mean your description is not extensive enough. What does the consonant collapse into.

[bradrn]

consonants next to each other turn into geminates (e.g. inkɨ > inːɨ)

Please explain this. This is not extensive enough.

I'm not sure what you mean by this. If anything, this seems too extensive - it applies to all consonants next to each other!

I mean your description is not extensive enough. What does the consonant collapse enough.

I'm not quite sure what you mean by 'What does the consonant collapse enough', but what I was trying to say was something like this:

If two consonants are immediately next to each other, the second consonant turns into the first consonant, resulting in a geminate. For instance: inkɨ > inːɨ, payt͡se > payːe, ɯɡaʔpæ > ɯɡaʔːæ

[Xwtek]

consonants next to each other turn into geminates (e.g. inkɨ > inːɨ)


I'm not sure what you mean by this. If anything, this seems too extensive - it applies to all consonants next to each other!

I mean your description is not extensive enough. What does the consonant collapse enough.

Mistyping. I meant into.
[/quote]

I'm not quite sure what you mean by 'What does the consonant collapse enough', but what I was trying to say was something like this:

If two consonants are immediately next to each other, the second consonant turns into the first consonant, resulting in a geminate. For instance: inkɨ > inːɨ, payt͡se > payːe, ɯɡaʔpæ > ɯɡaʔːæ

Is it more realistic to turn payt͡se > pat͡ʃːe or peːtse instead? Also, never in my mind this change is good: hat.su > hat.tu. It's more likely to turn into this instead hat.su > ha.t͡su or hat.t͡su. Also beware that in that case, these rule:
j > ∅ / V_C
ʔ > ∅ / next to consonants or at word end
basically just degemination and deleting glottal stop word finally.

[bradrn]

Is it more realistic to turn payt͡se > pat͡ʃːe or peːtse instead?

Sorry, that was a typo: I meant pajt͡se > pajːe.

Also, never in my mind this change is good: hat.su > hat.tu. It's more likely to turn into this instead hat.su > ha.t͡su or hat.t͡su.

This language doesn't distinguish /hat.su/ and /ha.t͡su/; in this case, it would be analysed as /ha/ + /t͡su/, and thus this change would not apply. (Actually, are there any languages which distinguish t.s vs t͡s?)

Also beware that in that case, these rule:
j > ∅ / V_C
ʔ > ∅ / next to consonants or at word end
basically just degemination and deleting glottal stop word finally.

Good catch - I'll have to think a bit more about what I wanted to do with these.

[Xwtek]

Is it more realistic to turn payt͡se > pat͡ʃːe or peːtse instead?

Sorry, that was a typo: I meant pajt͡se > pajːe.

That's not what I meant. I mean, Is it more realistic to turn pajt͡se > pat͡ʃːe or peːtse instead?

Also, never in my mind this change is good: hat.su > hat.tu. It's more likely to turn into this instead hat.su > ha.t͡su or hat.t͡su.

This language doesn't distinguish /hat.su/ and /ha.t͡su/; in this case, it would be analysed as /ha/ + /t͡su/, and thus this change would not apply. (Actually, are there any languages which distinguish t.s vs t͡s?)

Polish.

Also beware that in that case, these rule:
j > ∅ / V_C
ʔ > ∅ / next to consonants or at word end
basically just degemination and deleting glottal stop word finally.

Good catch - I'll have to think a bit more about what I wanted to do with these.

I suggest you to reorder the rule, so that that rule is executed first before gemination.

[Zaarin]

Also, never in my mind this change is good: hat.su > hat.tu. It's more likely to turn into this instead hat.su > ha.t͡su or hat.t͡su.

Attic Greek has such a change, but it has an intermediate step Tt > t͡sː > tt (a similar change in Celtic and other Greek dialects yielded /sː/). I'm inclined to agree, however, that a change whereby all medial clusters geminate the first consonant is suspect. I'm not sure a geminate glottal stop is even possible--if it were, I'd think Semitic would allow it, which it doesn't. I'd more expect either a specific consonant to yield gemination (/n/ in Canaanite, /lkʼ/ > /kːʼ/ in Phoenician, Tt > ss in Celtic, Tt > tt in Attic Greek) or else specific clusters to geminate in specific ways. For example, I agree that /jt͡s/ is more likely to become /t͡sː/ or /t͡ʃː/ than /jː/. As for the proposed geminate glottal stops, they could easily trigger vowel length instead.

(Actually, are there any languages which distinguish t.s vs t͡s?)

Many.

[Tropylium]

æ e i ɯ > e i ɨ ɨ

i > ɨ is unrealistic unconditionally. You should restrict this.

Seems perfectly fine in this chain shift context to me. (If you wanted to, you could assume a few intermediate stages: e i > e ɪ > i ɪ > i ɨ.)

ej > i / _#
i > ɨ / j_ or _j

æj ej > i ɨj /_# sounds strange to me

This is not even remotely strange, although the dissimilation of (ej >) ij > ɨj probably should happen earlier than the shift (æj >) ej > i (which would likely proceed, in more detail, as ej > ij > iː > i).

Actually, note that in combination with the chainshift e i > i ɨ, this could also be routed as assimilation rather than dissimilation: just have e > i / _j_ already before the chainshift, and you'll end up with je ej >> jɨ ɨj anyway.

[bradrn]

Is it more realistic to turn payt͡se > pat͡ʃːe or peːtse instead?

Sorry, that was a typo: I meant pajt͡se > pajːe.

That's not what I meant. I mean, Is it more realistic to turn pajt͡se > pat͡ʃːe or peːtse instead? \

Possibly that's true. Really, the problem is that I don't know what natlangs do in this situation: are there different outcomes for every different CC pair? (Although see Zaarin's comment, quoted below.)

Also, never in my mind this change is good: hat.su > hat.tu. It's more likely to turn into this instead hat.su > ha.t͡su or hat.t͡su.

This language doesn't distinguish /hat.su/ and /ha.t͡su/; in this case, it would be analysed as /ha/ + /t͡su/, and thus this change would not apply. (Actually, are there any languages which distinguish t.s vs t͡s?)

Polish.

Interesting - I didn't know that! Personally, I don't see anything wrong with /t.s/ > /tː/.

I'm inclined to agree, however, that a change whereby all medial clusters geminate the first consonant is suspect.

Maybe this is nitpicking, but are you trying to say that 'it is suspect to geminate all medial clusters', or 'it is suspect for gemination to always preserve the first consonant'', or both?

I'm not sure a geminate glottal stop is even possible--if it were, I'd think Semitic would allow it, which it doesn't.

Probably true.

I'd more expect either a specific consonant to yield gemination (/n/ in Canaanite, /lkʼ/ > /kːʼ/ in Phoenician, Tt > ss in Celtic, Tt > tt in Attic Greek) or else specific clusters to geminate in specific ways. For example, I agree that /jt͡s/ is more likely to become /t͡sː/ or /t͡ʃː/ than /jː/. As for the proposed geminate glottal stops, they could easily trigger vowel length instead.

I'll definitely have to think about doing this instead.

[Whimemsz]

I don't think any of the changes proposed are unrealistic, with one possible exception; Akangka's point that ts, dz might be expected to undergo the same palatalization process as s, z is certainly true, but it doesn't mean they must. However, the EVERY C1 becomes the geminate in C1C2 clusters change is a bit iffy; as Akangka points out, [jts] > [j:] is pretty weird compared to it winding up as [ts:] or [:ts] or [vowel.change+ts] or something similar. You might want to consider having different outcomes depending on the relative positions of the two consonants on the sonority hierarchy (though I should emphasize stuff like [nk] > [nn] is still plausible).

Geminate glottal stops are absolutely possible though -- you just close your glottis for longer... (Speaking of Semitic, some Arabic varieties in fact DO have geminate glottal stops, the ones that realize Classical Arabic /q/ as /ʔ/, e.g. Cairene baʔʔāl "grocer")

[Zaarin]

Maybe this is nitpicking, but are you trying to say that 'it is suspect to geminate all medial clusters', or 'it is suspect for gemination to always preserve the first consonant'', or both?

The latter--I have no problem imagining a language in which all internal CC sequences are geminates. Whimemsz suggestions are good regarding sonority.

Geminate glottal stops are absolutely possible though -- you just close your glottis for longer... (Speaking of Semitic, some Arabic varieties in fact DO have geminate glottal stops, the ones that realize Classical Arabic /q/ as /ʔ/, e.g. Cairene baʔʔāl "grocer")

Interesting. I didn't know this.

[dɮ the phoneme]

Ok, here's a few sound changes from my own conlang. I'm been looking for some critique of them for some time. Aside from just their plausibility, I'm curious as to how long these might be expected to take. I had estimated about 2000-2500 years, but that may be way off.

Also, this has turned out very long and I wasn't able to get the spoiler tag working (does the new board still have that?), so if there's another way to avoid jamming a wall of text like this in the faces of those who aren't interested, please let me know. edit: appearently it's the "more" button...

The starting inventory:

Code: Select all

p	t	ʈ		k	q
b	d		ɟ
f	s	ʂ		x	X	h
	l	ɻ	j	w
m	n

i		u
e		o
	a

Basically CV(C) syllables. Primary stress is penultimate, secondary stress falls on all heavy syllables (those with a coda sonorant).

Now, the changes:

More: show

1. loss of coda nasals:
VN→Ṽː→Vː/ _$

2. palatalization:
[-retroflex, -uvular] → [+palatalized] / _[+front]
[-retroflex, -uvular] → [+palatalized] / _(C)Cʲ

3. lowering adjacent to uvulars:
{i e} {u o} → æ ɔ / adjacent to a uvular

4. u-mutation:
i e {a æ} → y ø ɔ / _(C)[+vowel, +round]

5. intervocalic lenition:
Between either vowels or semivowels, voiceless stops become voiced and voiced stops become voiced fricatives:
[-voice, +obstr., -glottal] [+voice, +obstr.] → [+voice] [-stop, +fricative] / {V, j, w}_{V, j, w}

6. voiced obstruent mergers:
{ɢ ɣ} {ðʲ ɣʲ} → ʁ j

7. primary syncope:
Vowels delete in the syllable following the primary stress

8. stress reasigment:
σˈσ → ˈσσ / _#

9. secondary syncope:
Vowels delete in an unstressed word-internal syllable preceding primary or secondary stress

10. palatal assimilation:
C → [-palatalized] /[-palatalized]_[-palatalized]
[-retroflex, -uvular] → [+palatalized] / _(C)(C)(C)Cʲ (blocked by a retroflex or uvular)

11. cluster simplification:
Clusters of dorsals assimilate in place to their last element, as do clusters of coronals.
Clusters of obstruents whose elements differ in voicing, but all share the same place and manner of articulation, assimilate in voicing to their first element. Palatalized consonants depalatalize before their homorganic non-palatalized counterparts.

12. vowel reduction:
i {y ø} {u o} {e a}→ɨ ɵ ɔ æ/ in an unstressed syllable

13. pre-velar raising:
e ø æ a {o ɔ} → i y e æ u/ _[+velar, -round] except after a uvular

14. fronting of palatalized back consonants:
[+velar, +palatalized] hʲ → [+palatal, -palatalized] ç

15. debaucalization:
X → h

16. geminate reduction:
C₁C₁ → C₁

17. [h] loss:
h → ∅

18. pre-nasal lowering:
i y e o → e ø æ ɔ / _[+nasal]

19. lenition of [ɟ]:
ɟ → j in the coda

20. fortification:
j w → dʒ v in the onset

21. post-alveolarization:
{ç sʲ} zʲ c ɟ → ʃ ʒ tʃdʒ

22. loss of [ɻ]:
ɻ → {ʂ ʐ} in the onset, ɻ → ∅ elsewhere, perhaps coloring adjacent vowels/coronal consonants in the process

23. vowel coaslescence:
as yet undetermined vowel coalescence, which eliminates hiatus as well as coda semivowels, creating long vowels

Finally, the resulting inventory:

Code: Select all

p	pʲ	t	tʲ	ʈ	tʃ	k	q
b	bʲ	d	dʲ	ɖ	dɮ	g
f	fʲ	s		ʂ	ʃ	x
v	vʲ	z		ʐ	ʒ		ʁ
		l	lʲ		
m	mʲ	n	nʲ

i	y		u
e	ø		o
æ			ɔ
	a

With contrastive vowel length, and maximal syllable structure CVCCC.

[Zaarin]

Aside from just their plausibility, I'm curious as to how long these might be expected to take. I had estimated about 2000-2500 years, but that may be way off.

I don't think there's a standard or consistent rate of change, but your proposed time depth looks fine to me.

4. u-mutation:
i e {a æ} → y ø ɔ / _(C)[+vowel, +round]

I might expect /æ/ to pattern with /e/ here, but I don't think what you have is wrong per se.


The rest looks fine to me. End result looks rather Uralic sans vowel harmony.