Sound Change Quickie Thread

[Nortaneous]

I wouldn’t know; I know very little about Austroasiatic. (And why is Austroasiatic in particular relevant here?)

Large vowel systems, coda consonants, relatively well-studied.

True, I didn’t notice that /w/ is also missing. But are you sure /j w/ can only appear in codas due to the Great Vowel Shift or resyllabification? Etymonline, for instance, lists an apparent counterexample in the form of may < OE mæg.

Analyzing MidE diphthongs as vowel + semivowel sequences could be viable, but might complicate the GVS... unless the diphthongal values are original and monophthongization was secondary? Which is hard to square with the existence of UK dialects that don't have the toe-tow merger.

[Travis B.]

True, I didn’t notice that /w/ is also missing. But are you sure /j w/ can only appear in codas due to the Great Vowel Shift or resyllabification? Etymonline, for instance, lists an apparent counterexample in the form of may < OE mæg.

Analyzing MidE diphthongs as vowel + semivowel sequences could be viable, but might complicate the GVS... unless the diphthongal values are original and monophthongization was secondary? Which is hard to square with the existence of UK dialects that don't have the toe-tow merger.

I would personally interpret the LMidE diphthongs as diphthongs, which did not participate in the GVS, and which were monophthongized afterwards (except for LMidE /ei/, which could be argued to have either merged with LMidE /eː/ before the first stage of the GVS or EModE /iː/ afterwards), as LMidE /ɛi ɔu/ both survived until after the GVS (in most dialects LMidE /ɛi/ merged with LMidE /aː/ relatively early, but the two survive as distinct phonemes in at least some present-day dialects, albeit recessively, and LMidE /ɔu/ remained separate from LMidE /ɔː/ to a rather late date in most dialects, and as you mention has not fully merged in all dialects).

[bradrn]

OK, I think I’ve managed to come up with a fairly reasonable set of sound changes which gets me close to what I want. The major developments are:

1. Preliminary vowel shift: {a, aː, ə, i, iː, u} → {æ, ɔ, ə, e, i, u}
2. Final obstruent loss: [+obstr] → ∅ / _ C or _ #
3. Glottal loss: {ʔ, h} → ∅
4. Tonic umlaut: stressed V₁ → V₃ / followed by V₂ in next syllable, where V₃ is determined from V₁ and V₂ à la Northern Vanuatu:
   More: show

   V₁ \ V₂	…i	…e	…æ	…ɔ	…u	…ə
   i…	i	e	e	e	i	e
   e…	e	æ	æ	æ	e	ə
   æ…	æ	æ	æ	ɔ	ɔ	a
   ɔ…	o	ɔ	a	ɔ	o	ɔ
   u…	u	o	o	o	u	o
   ə…	e	ə	a	ɔ	o	ə

5. Posttonic vowel syncope: V → ∅ / V[+stressed] (C) _. (Note that this is blocked by consonant clusters, so e.g. *yakʰaantəwu → yæxɔntəwu not *yæxɔntwu.)

The end result of these changes is general monosyllabisation: *fawətɬ → fæw, *ʔaŋətʰ → aŋ, *suʔəɣ → sɔŋ. Other roots end up with a ‘postsyllable’: bæln̩, næym̩, tɔfŋ̍, tuŋl̩. I also quite like how the verbal paradigms come out, with just the right amount of irregularity:

(IMPF / CONT / ITV / PFV / TEL / DIM)
yusɔ / yusɔl / yusɔts / yošə / yošəŋu / yošətse
faw / fawle / fawse / fawše / fawšeŋu / fawšetse
ul / ulle / ultse / ol / olŋu / oltse
tuŋl̩ / tuŋl̩le / tuŋl̩se / tuŋle / tuŋleŋu / tuŋletse

(The protolanguage has one more aspectual suffix in addition to the above, which is originally regular but becomes mildly irregular in the descendant; I’m undecided as to whether I want to keep it or not.)

One thing I am a bit unsure about is exactly how the posttonic syncope (change (5) above) should operate. At the moment, it’s blocked by consonant clusters, so odd developments like *tamlaŋ → tæmlŋ (to make up a random example) is sensibly barred. However, this restriction also blocks changes like *taylaŋ → tæylŋ̍ and *tantaŋ → tæntŋ̍, despite those being perfectly fine. Another problem is words like *tanay — does this turn into tæn, or tæny, or tæni, or what? And I don’t even want to think about what happens to something like *tanway, but nevertheless it is a possibility I am forced to consider. Does anyone have any ideas?

[Qwynegold]

Well, first of all, what are your syllabic consonants? Make a sound change #6 that deletes vowels before those consonants that can become syllabic.

Why should t{mlN be banned when t{jlN is allowed?

[bradrn]

Well, first of all, what are your syllabic consonants?

All sonorants; that is, /m n ŋ l/.

Make a sound change #6 that deletes vowels before those consonants that can become syllabic.

Not sure I understand this suggestion… could you give an example? I’m already making those consonants syllabic by deleting the proceeding vowel (rule #5).

Why should t{mlN be banned when t{jlN is allowed?

Hmm… good point, actually, I didn’t think about it that way.

[cedh]

You could allow /j w/ to become syllabic. I actually think this would be fairly likely with changes like the ones you have.

And/or you could also just exclude jC wC Cj Cw sequences from counting as "consonant clusters that block syncope on the other side of C". There's no reason that all consonants in clusters must prevent vowel deletion in the same way.

[Qwynegold]

Not sure I understand this suggestion… could you give an example? I’m already making those consonants syllabic by deleting the proceeding vowel (rule #5).

I mean like:
V → ∅ / V[+stressed] CC _C[+sonorant]

[kodé]

One thing I am a bit unsure about is exactly how the posttonic syncope (change (5) above) should operate. At the moment, it’s blocked by consonant clusters, so odd developments like *tamlaŋ → tæmlŋ (to make up a random example) is sensibly barred. However, this restriction also blocks changes like *taylaŋ → tæylŋ̍ and *tantaŋ → tæntŋ̍, despite those being perfectly fine. Another problem is words like *tanay — does this turn into tæn, or tæny, or tæni, or what? And I don’t even want to think about what happens to something like *tanway, but nevertheless it is a possibility I am forced to consider. Does anyone have any ideas?

Yep, it’s perfectly fine to allow syncope to create some clusters but not others. For *tanay, any of tæn tæny tæni would be reasonable. For *tanway, if you follow the suggestion of make sonorants syllabic adjacent to syncope, you can get tænuy . I could see tæn and tænw as possible but less likely reflexes.

[Rounin Ryuuji]

One thing I am a bit unsure about is exactly how the posttonic syncope (change (5) above) should operate. At the moment, it’s blocked by consonant clusters, so odd developments like *tamlaŋ → tæmlŋ (to make up a random example) is sensibly barred. However, this restriction also blocks changes like *taylaŋ → tæylŋ̍ and *tantaŋ → tæntŋ̍, despite those being perfectly fine. Another problem is words like *tanay — does this turn into tæn, or tæny, or tæni, or what? And I don’t even want to think about what happens to something like *tanway, but nevertheless it is a possibility I am forced to consider. Does anyone have any ideas?

Yep, it’s perfectly fine to allow syncope to create some clusters but not others. For *tanay, any of tæn tæny tæni would be reasonable. For *tanway, if you follow the suggestion of make sonorants syllabic adjacent to syncope, you can get tænuy . I could see tæn and tænw as possible but less likely reflexes.

Building on this, a sequence of sounds /tænw/, while not pronounceable, is not really easy to pronounce, and I tend to want to make it either [tænwə] or [tænu]. If /tænw/ came to be, I would expect either epenthesis or the full vocalisation of the /w/. You could also do /nw/ > /m/ if you want more consonant gradation.

[bradrn]

OK, so I think I’ve figured out a new set of changes I’m happier with:

1. Preliminary vowel shift: {a, aː, ə, i, iː, u} → {æ, ɔ, ə, e, i, u}
2. Final obstruent loss: [+obstr] → ∅ / _C or _#
3. Sonorant assimilation: [+sonorant]₁ [+sonorant]₂ → [+sonorant]₁ ː / except when first consonant is a semivowel
4. Glottal loss: {ʔ, h} → ∅
5. Tonic umlaut: stressed V₁ → V₃ / when followed by V₂ in next syllable
6. Posttonic vowel syncope: V → ∅ / after stressed syllable
7. Semivowel vocalisation: {w, y} → {u, i} / C_C or C_#

(New or altered rules are bolded.)

The main changes I made were adding an assimilation/gemination rule to get rid of weird clusters, allowing syncope after consonant clusters, and vocalise semivowels. (I do hope these new changes are still plausible!) I particularly like kodé’s suggestion of doing *tanway → tænuy, a possibility I somehow didn’t think of. The output for those problematic cases is now as follows:

*tamlaŋ → tæmm.ŋ
*taylaŋ → tæyl.ŋ
*tantaŋ → tænt.ŋ
*tanay → tæ.ni
*tanway → tæ.nuy

As is evident from these examples, these new sound changes make the syllable structure of this descendant a bit more complex than my earlier version. Specifically, maximal syllable structure is now (C)V(sonorant)(obstruent), while minor syllables are restricted to being a single syllabic consonant. (Previously the sound changes allowed them to be more complex, but that led to postsyllables like lŋ (← *tamlaŋ), which I didn’t particularly like. In exchange, the coda of normal syllables has gained an extra obstruent.) There are now geminates, but only /mː nː ŋː lː/, which are restricted to the syllable coda. Overall, this is far more complex and realistic than the phonotactics I normally give my conlangs, so I’m fairly satisfied with this.

And finally, here’s an attempt at guessing what all of this will eventually look like:

Sæssæŋ šɔ Ælsæŋ muŋiwæyelle faw, naŋ exæyənə pæws bəyesɔl, aŋseŋæ esæwyošətse læm.

Of course, I haven’t really developed any grammar yet, so it will probably end up looking quite different by the time I’ve worked it all out.

[Pabappa]

sound shifts in tonal languages.....

offhand, would you expect péà to be more likely to shift to pʲá or to pʲà?

That is, would the original tone of the whole syllable predominate (since the high tone mora is more prominent, i expect it to be longer)? Or would the tone of the previously suppressed segment predominate as it becomes longer?

As a third option, could it be whichever is higher is the one that predominates? Since I have a rarer sequence of the type pèá to take care of as well.

Japanese could help me here, but i wasnt able to figure anything out and Im not sure we know enough about the tones of earlier stages of Japanese to really make use of the sound changes.

Thanks for any help you can provide.

[Travis B.]

sound shifts in tonal languages.....

offhand, would you expect péà to be more likely to shift to pʲá or to pʲà?

That is, would the original tone of the whole syllable predominate (since the high tone mora is more prominent, i expect it to be longer)? Or would the tone of the previously suppressed segment predominate as it becomes longer?

As a third option, could it be whichever is higher is the one that predominates? Since I have a rarer sequence of the type pèá to take care of as well.

Japanese could help me here, but i wasnt able to figure anything out and Im not sure we know enough about the tones of earlier stages of Japanese to really make use of the sound changes.

Thanks for any help you can provide.

Were the higher tone mora more prominent I would expect it to become /pé/ or /pʲé/ instead.

[Richard W]

I'd expect the result to be a falling tone, especially if you start with a 2-tone system. I believe that is a typical result from West African languages, which often have moraic tone assignment.

On the other hand, Tai contractions generally end up with the tone of the second syllable. Note that the Tai languages generally have iambic stress, and tone tends to be assigned to sesquisyllables.

[bradrn]

sound shifts in tonal languages.....

offhand, would you expect péà to be more likely to shift to pʲá or to pʲà?

That is, would the original tone of the whole syllable predominate (since the high tone mora is more prominent, i expect it to be longer)? Or would the tone of the previously suppressed segment predominate as it becomes longer?

As a third option, could it be whichever is higher is the one that predominates? Since I have a rarer sequence of the type pèá to take care of as well.

Japanese could help me here, but i wasnt able to figure anything out and Im not sure we know enough about the tones of earlier stages of Japanese to really make use of the sound changes.

Thanks for any help you can provide.

Thinking about this autosegmentally, you could get any of the following results:

tone-autosegmental.png (4.01 KiB) Viewed 13086 times

I’m inclined to agree with Richard W that /pʲâ/ seems most likely, but any of the others would be plausible as well; I suppose it would depend on the tone spreading rules for this language.

[Rounin Ryuuji]

sound shifts in tonal languages.....

offhand, would you expect péà to be more likely to shift to pʲá or to pʲà?

That is, would the original tone of the whole syllable predominate (since the high tone mora is more prominent, i expect it to be longer)? Or would the tone of the previously suppressed segment predominate as it becomes longer?

As a third option, could it be whichever is higher is the one that predominates? Since I have a rarer sequence of the type pèá to take care of as well.

Japanese could help me here, but i wasnt able to figure anything out and Im not sure we know enough about the tones of earlier stages of Japanese to really make use of the sound changes.

Thanks for any help you can provide.

Going with sound changes in Japanese, I would expect the vowel to be bimoraic: では /déwà/ (unless I misunderstand) > じゃあ /ɟʑáà/ > (later) じゃ /ɟʑa/)

I would possibly also expect the resulting vowel to be bimoraic */pʲ(j)áà/, at least in a content word (the one Japanese example I can think of is では > じゃあ, じゃ, the two forms now coexisting; if you want it to shorten from there, though, I think I would expect */pʲ(j)á/.

[dɮ the phoneme]

I would possibly also expect the resulting vowel to be bimoraic */pʲ(j)áà/, at least in a content word (the one Japanese example I can think of is では > じゃあ, じゃ, the two forms now coexisting; if you want it to shorten from there, though, I think I would expect */pʲ(j)á/.

If the language contrasts length, or has important weight considerations for assigning stress, then I would expect bimoraicity as well. Otherwise though, not necessarily. Plenty of African languages will delete vowels with no compensatory lengthening and retain the tone on an adjacent vowel.

[Rounin Ryuuji]

I should probably say that I also would at least frequently imagine length developing from that, too. A new tone could also emerge, too, I imagine.

[Pabappa]

Thank you all for the replies.

This sound shift occurs shortly after the parent language breaks up into several daughter languages, so I can easily follow more than one path here, but I have a clear favorite daughter picked out because it's the one that migrates northward and becomes a nationwide language.

I think I'm going to go with a long high vowel as the reflex of both /éà/ and the much rarer /èá/. Getting palatalization is important because the parent language had an unbalanced inventory of vowel sequences .... most arose through elision of /j/, so (neglecting tones) sequences of /e/ + vowel and /i/ + vowel were very commom, but sequences beginning with the other three vowels were very rare because hiatus had been rare in the primordial language.

I was reluctant to incorporate vowel length here, as that was also rare, but I thought about it some more and realized that it's probably a good idea, since I'm probably going to create a diphthong first and then smooth the tone over it, .... so i'll have /éà/ > /éá/ ... and then shift the diphthong into a /ʲá/ sequence. (Which i dont consider a diphthong because palatalization is common in the area this language moves into.) The resulting imbalance of having long vowels primarily occur after palatalized consonants will be corrected for as well, since most palatalized consonants will ultimately shift into various new plain consonants, and because the vowel length will be at least conditionally shortened as time goes by.

A different branch could do a shift such as /ia ie ii io iu/ > /ī ī ī ʲū ʲū/, but even here palatalization is important and all /i/ will be palatalized as well.

I haven't done much conlanging lately, and I find it hard to work on my big projects such as Moonshine, but touching up easy languages like this family .... called Oyster .... makes me feel good, and this is a step in a new direction for me, since it's a language that ends up interacting with languages that are only distantly related. In fact, this area of the world has four families in close proximity to each other .... Moonshine, Oyster, Dreamer, Play.

[Rounin Ryuuji]

This sound shift occurs shortly after the parent language breaks up into several daughter languages, so I can easily follow more than one path here, but I have a clear favorite daughter picked out because it's the one that migrates northward and becomes a nationwide language.

I think that's only natural — one language will appeal to you, so you spread it wide throughout the fiction, and find it takes the most life inside your mind. That's what's happened with me, at least.

I think I'm going to go with a long high vowel as the reflex of both /éà/ and the much rarer /èá/. Getting palatalization is important because the parent language had an unbalanced inventory of vowel sequences .... most arose through elision of /j/, so (neglecting tones) sequences of /e/ + vowel and /i/ + vowel were very common, but sequences beginning with the other three vowels were very rare because hiatus had been rare in the primordial language.

I like it when languages acquire features.

I was reluctant to incorporate vowel length here, as that was also rare, but I thought about it some more and realized that it's probably a good idea, since I'm probably going to create a diphthong first and then smooth the tone over it, .... so i'll have /éà/ > /éá/ ... and then shift the diphthong into a /ʲá/ sequence. (Which i dont consider a diphthong because palatalization is common in the area this language moves into.) The resulting imbalance of having long vowels primarily occur after palatalized consonants will be corrected for as well, since most palatalized consonants will ultimately shift into various new plain consonants, and because the vowel length will be at least conditionally shortened as time goes by.

A different branch could do a shift such as /ia ie ii io iu/ > /ī ī ī ʲū ʲū/, but even here palatalization is important and all /i/ will be palatalized as well.

Sounds like a lot of interesting ideas.

I haven't done much conlanging lately, and I find it hard to work on my big projects such as Moonshine, but touching up easy languages like this family .... called Oyster .... makes me feel good, and this is a step in a new direction for me, since it's a language that ends up interacting with languages that are only distantly related. In fact, this area of the world has four families in close proximity to each other .... Moonshine, Oyster, Dreamer, Play.

Sometimes, a small return can open the way for something bigger to follow.

[Ahzoh]

Are these changes possible?

Reflexes:
/ʃ/ > /ɬ/ (gains lateralization as a result of influence from preexisting /ɬ/)
/ʒ/ > /ɮ/ > /l/ (same as above)
/t͡ʃ d͡ʒ/ > /s z/ before another /s z/, maybe also intervocalically
/t͡ʃ d͡ʒ/ > /ɬ ɮ/ before another /ɬ ɮ/
/t͡ʃ d͡ʒ/ > /k g/ maybe before another velar, maybe also word-initially and word-finally