Sound Change Quickie Thread

[----]

yes that's very reasonable

[StrangerCoug]

What are interesting things I can do to glottalic consonants besides just merging them into their pulmonic counterparts? (I'm already aware of /s’/'s tendency to want to become /t͡s’/.)

[mèþru]

One idea: Turn them into pulmonic counterparts + glottal stop, which can then trigger tonogenesis.

[Zaarin]

What are interesting things I can do to glottalic consonants besides just merging them into their pulmonic counterparts? (I'm already aware of /s’/'s tendency to want to become /t͡s’/.)

Glottalization to pharyngealization occurred in Semitic and the Caucasus. You could turn your plosives into implosives. Fricatives, as you noted, show a tendency to become affricates. They can become simple glottal stops, which seems to be a common allophonic variation word-finally. They can induce creaky voice then de-glottalize, phonemicizing the creaky voice.

[mèþru]

The last of which can also lead to tonogenesis.

[Zaarin]

The last of which can also lead to tonogenesis.

Which can lead in either direction (high or low): see Athabaskan languages which show mirror image tonemes, or Mohawk vs. Mandarin (syllable-final glottals in Mohawk vanished and created low-tone, where they created high tone in Mandarin). It appears that if glottalization is the emphasized feature, then high tone results, as in Mandarin or Slavey; if creakiness is the emphasized feature, then low tone results, as in Mohawk or Navajo.

[Chengjiang]

Asymmetrical deaffrication by POA? I'm wondering if there are any natlang examples of affricates deaffricating at one POA while remaining affricates at another, e.g. alveolar affricates deaffricating while palatoalveolars are preserved. I know there are examples that are asymmetrical by voicing, usually (but not always) with the voiced one deaffricating.

Also, regarding glottalized consonants: I'm pretty sure glottalized > long is attested.

[smii]

According to index diachronica, 'Cayuga to Lower Cayuga' had ts > t, but nothing with tʃ

[vokzhen]

I'm wondering if there are any natlang examples of affricates deaffricating at one POA while remaining affricates at another, e.g. alveolar affricates deaffricating while palatoalveolars are preserved.

Spanish says hi There's also Taishanese tsʰ > tʰ,s>ɬ versus tɕʰ > tsʰ that appears to have happened by piecing together things from this vowel reconstruction. Lateral affricates seem able to merge into other sounds without effecting other affricates, e.g. > /t/ in most Sotho-Tswana languages and some Nahualt dialects or > /x,k,q/ in Northeast Caucasian. pf>f in German (and some Chinese dialects?) and kx>x in some Southern Bantu without effecting other affricates, but they're odd enough you may not want to use them as precedence.

[Chengjiang]

I'm wondering if there are any natlang examples of affricates deaffricating at one POA while remaining affricates at another, e.g. alveolar affricates deaffricating while palatoalveolars are preserved.

Spanish says hi.

If you're talking about the source of modern Spanish ch, I thought that wasn't an affricate at the same time [ts] and [dz] existed. Was it?

[jmcd]

To expand a little on the German, there exist affricates for all three points of articulation, depending on the language variety: the sound change boundaries for p>pf (accompanied by p>f) do not coincide with those for t>ts (accompanied by t>ts>s) or with those for k>kx (accompanied by k>x). /kx/, absent from the standard, is the rarest of the three, found only in Bavaria, Austria and Switzerland.

Wikipedia's summary indicates that there was a point at which /ts/ the affricate was accompanied by /f/ and /x/ the fricatives.

[Porphyrogenitos]

So, I've got a simple five-vowel system /a e i o u/.

The change /u/ > /i/ happens, with historical /u/ remaining distinct in some positions because of palatalization triggered by historical /i/.

/o/ then shifts to /u/. I know this could give me a stable four-vowel system /i u ɛ~æ ɔ~ɑ/, but what are my options for maintaining/regaining /a e i o u/ if I have /a/ move up to become /o/? What are some fairly simple environments for either /a/ to resist the shift to /o/, or for /e/ to undergo a shift to /a/? (I then plan on at least partially erasing those environments to make the distinction phonemic.)

[Zaarin]

So, I've got a simple five-vowel system /a e i o u/.

The change /u/ > /i/ happens, with historical /u/ remaining distinct in some positions because of palatalization triggered by historical /i/.

/o/ then shifts to /u/. I know this could give me a stable four-vowel system /i u ɛ~æ ɔ~ɑ/, but what are my options for maintaining/regaining /a e i o u/ if I have /a/ move up to become /o/? What are some fairly simple environments for either /a/ to resist the shift to /o/, or for /e/ to undergo a shift to /a/? (I then plan on at least partially erasing those environments to make the distinction phonemic.)

If you have pharyngeals or uvulars, Ho > a or eH > a are both very plausible (where H = any pharyngeal/uvular/epiglottal consonant), happening in Semitic and Indo-European, among others.

[Pole, the]

So, I've got a simple five-vowel system /a e i o u/.

The change /u/ > /i/ happens, with historical /u/ remaining distinct in some positions because of palatalization triggered by historical /i/.

/o/ then shifts to /u/. I know this could give me a stable four-vowel system /i u ɛ~æ ɔ~ɑ/, but what are my options for maintaining/regaining /a e i o u/ if I have /a/ move up to become /o/? What are some fairly simple environments for either /a/ to resist the shift to /o/, or for /e/ to undergo a shift to /a/? (I then plan on at least partially erasing those environments to make the distinction phonemic.)

There's a plenty of hanges that can create [e], [a] or [o]. A couple of ones I can come up with atm (simplified for the sake of brevity):

Slavic: [a aː] → [o a]
Germanic: [a aː] → [a oː]
Polish: [e eː] → [o a] (before non-palatal coronals)
Polish: [ĭ ŭ] → [ə] → [e] or syncope
Shtokavian: [ĭ ŭ] → [ə] → [a] or syncope
Russian: [ĭ ŭ] → [ə] → [e] (after palatals), [o] (after non-palatals), or syncope
Russian: [ĕ] → [o] (stressed, before non-palatals)
East Slavic: [ẽ] → [ʲa]
Polish: [ẽ õ] → [ʲã ã], later [ã ãː] → [eN oN]
German dialects: [ə ər] → [e o]

[vokzhen]

a>o resisted before a front vowel, or a shift of e>a before back vowel. e>a triggered by velars, nasals, or liquids could work, not just uvulars/pharyngeals. You could have a>o only trigger near those same consonants. You could have at the same time a shift of o>u *blocked* by those conditions, creating [o] as a positional merger of both /a u/. You could have general a>o, e>a, but then coronal-triggered umlaut of new u>i, o>e, a>e, restoring /e/ while making /i/ even more common and /o/ rather rare, given relatively equal starting points.

Length formed somehow, only either long or short /a/ backs (either one could work), length lost. /aj aw/ > /a:/. Or loss of intervocal /h ? j w/, with diphthongization of high vowels (ia eu > ja aw) but merger of non-high vowels (ea oa ae ao > a:). Or allophonic open-syllable lengthening, vowel shift, lengthening becomes nonproductive.

Creation of a tense-lax system, say from open versus closed syllables or the presence of certain codas (like Andalusian -h), with a slight rearrangement as it re-collapses into a plain 5-vowel system.

For options not directly based around /a/ not backing or /e/ lowering, there's, vowel reduction to schwa, schwa rapidly shifts into the open /a/ spot as it opens up. There's extensive loaning. There's getting /u/ from a different source, like unstressed /a o/ > [ə], thus /aw ow/ > [əw] > /u/, or vowel reduction to schwa and then ə>u, pre-empting o-raising in the first place.

[Porphyrogenitos]

Do ejectives ever spontaneously become affricated to increase their volume/further differentiate themselves from non-ejectives? I.e. do languages with ejectives tend to have ejective affricates with no non-ejective equivalent? I'm just curious, since I've noticed that for me, at least, /t͜s' t͜ʃ' t͜ɬ'/ are all a lot louder than and easier to distinguish from /t/ than /t'/ is.

[gufferdk]

Do ejectives ever spontaneously become affricated to increase their volume/further differentiate themselves from non-ejectives? I.e. do languages with ejectives tend to have ejective affricates with no non-ejective equivalent? I'm just curious, since I've noticed that for me, at least, /t͜s' t͜ʃ' t͜ɬ'/ are all a lot louder than and easier to distinguish from /t/ than /t'/ is.

I think a few Salishan languages (at least Nuxalk) have /t͜ɬʼ/ without /t͜ɬ/. I know that ejective fricatives have a tendency to turn into affricates, but I don't know if affricates like to deaffricativize.
Based on the limited data availible in the index diachronica (https://chridd.nfshost.com/diachronica/search?q=t%CA%BC) I think /t͜s' t͜ʃ' t͜ɬ'/ /t'/ is more likely than the ones in reverse.
The only changes goin in reverse are t' t͜s' or t͜ʃ' /#_ (in some words, conditioning factors unclear) in some yokutsan languages, whereas everone of /t͜s' t͜ʃ' t͜ɬ'/ to /t'/ are attested (each one in a different family), with seemingly no conditioning environment.

[Zaarin]

Do ejectives ever spontaneously become affricated to increase their volume/further differentiate themselves from non-ejectives? I.e. do languages with ejectives tend to have ejective affricates with no non-ejective equivalent? I'm just curious, since I've noticed that for me, at least, /t͜s' t͜ʃ' t͜ɬ'/ are all a lot louder than and easier to distinguish from /t/ than /t'/ is.

As gufferdk says, it tends to go the other way, and ejective fricatives tend to affricate (for example, sʼ > tsʼ in many Semitic languages and probably Egyptian as well). When languages feel the need to distinguish ejective from plain stops, it seems like the plain stops tend to aspirate (see: Akkadian and Phoenician). But of course many PNW languages have a three-way contrast of C Cʰ Cʼ. One Plateau or Upper Plains Native American language (I want to say Kutenai, but I could be mistaken) with weak ejectives chiefly distinguishes ejectives by creakiness in following vowels and by the regular allophonic change of Cʼ > ʔ word finally.

[gufferdk]

How could on reasonably get rid of a tone-system like /V˦ V˩ V˥˩/, while still maintaining some distinction between the types? Would going to something like /V V̰ Vː/ work?
If we are starting with a small vertical vowel system like /ɨ ə ɐ/, would it then be possible to have something like this?
ɨ˦ ə˦ ɐ˦/u o ɑ/_C[+uvular]
ɨ˦ ə˦ ɐ˦/u o ɑ/ʀ_
ɨ˦ ə˦ ɐ˦/i e æ/_
ɨ˩ ə˩ ɐ˩/u o ɑ/_
ɨ˥˩ ə˥˩ ɐ˥˩/something/

[mèþru]

How could on reasonably get rid of a tone-system like /V˦ V˩ V˥˩/, while still maintaining some distinction between the types?

Add some consonants? Changing the vowels is the most likely answer, but you could add laryngeal consonants in addition (well, I guess it would be kind of necessary as an intermediate step if you want to change the vowels' phonation). I assume that the higher pitch tones would result in close/front vowels, and the lower pitch tones would result in open/back vowels. Falling tone could swing either way or lead to intermediate vowels. The vowels are likely to be heavily coloured (allophonically at least) by consonants towards backness/frontness before the loss of tone anyway. Changes along the back/front continuum would likely be influenced by and phonemicise that allophony.

[Soap]

I would be interested in hearing about tone causing vowels to change quality since I have a major conlang with 6 tones but only 5 vowels, of which only 3 are common, and no diphthongs. I actually like setups like this, and I made it this way on purpose, but I like diversity too and I especially like designing minor languages that are only a few thousand years separate from the parent language but look as if they are completely unrelated. I asked a similar question long ago (not in this thread; this thread is newer) and got an answer which seemed to imply that vowels might change quality due to tone on a few occasions but it's not a widely reported sound change, and that tone is essentially a suprasegmental feature that doesn't affect the pronunciation of vowels or consonants. Vowel and consonant changes can create tones, but there don't seem to be many, if any, examples of tones causing shifts in vowels or consonants.

I'd be happy to see counterexamples if anyone knows some.

[Pogostick Man]

High tone can cause high vowels to diphthongize. I'll have to look to find the paper I saw that in, but I have read about this occurring.

[gufferdk]

Digging around in my bookmarks I found this book: https://drive.google.com/open?id=0B9Pul ... Wl6T3BpSW8

Apparently if we extend the concept of tone to register, these features seem to coincide:
High pitch, modal voice, unvoiced onsets, retracted tongue root (leading to lax, low vowels)
vs.
Low pitch, glottalisation, "deep pharyngeal resonance" (sometimes misleadingly labeled "breathy voice"), voiced onsets, advanced tongue root (leading to tense, high vowels)

According to the book, many closely related Mon Khmer languages have these two registers and then uses only a few of the individual characteristics of them to keep them seperate. The relationship between low tone and ATR is also seen in other places, such as Madurese.

The book says that Cantonese high tone only occurs on short lax vowels and mid tone only occurs on tense vowels, but low tone on both types, but is unclear on which way causality goes and whether +-ATR is involved.

Direct interaction between vowel quality and tone, without +-ATR being involved seems to be much rarer, adn often controversial though high vowles have a higher base frequency, even in non-tonal languages.
It says Fuzhou is claimed to have such an interaction but the data is complex and controversial. It gives this simplified subset:
Tone 12, 242, 13ʔ: ei(ŋ), ou(ŋ), øü(ŋ), aiŋ, auŋ, ɔuŋ
Tone 44, 52, 22: i(ŋ), u(ŋ), ü(ŋ), eiŋ, ouŋ, øüŋ
And points to Wright 1983 and references therein for a full account (do someone have access to that?).

Additionally, i remember reading somewhere that contour tones tend to lengthen the vowel compared to register tones


Given this I think doing something like this with my starting inventory doesn't seem too unrealistic, especially if i pretend my starting /˩ ˦ ˥˩/ wasn't purely a pitch contrast:
ɨ ə > u o /_C[+dorsal (or glottal, probably)]
ɨ ə > u o /ʀ_
ɨ ə > i e /otherwise
V˥˩ > Vː
i˦ u˦ e˦ o˦ a˦ > ɪ ʊ ɛ ɔ a
possibly a˩ > ə or ʌ
V˩ > V

Fronting of velars to alveolars of some description, and uvulars to velars could then phonemicise the front/back distinction leaving me with /i iː ɪ u uː ʊ e eː ɛ o oː ɔ a aː (ə or ʌ)/ which would then probably be unstable as hell thanks to the 5-way height contrast. Thoughts?

[StrangerCoug]

Fronting of velars to alveolars of some description, and uvulars to velars could then phonemicise the front/back distinction leaving me with /i iː ɪ u uː ʊ e eː ɛ o oː ɔ a aː (ə or ʌ)/ which would then probably be unstable as hell thanks to the 5-way height contrast. Thoughts?

Based on English, I do not find it plausible that a five-way height contrast would, by itself, render your vowel system unstable. What I would believe to be the source of instability is that you have a short tense/long tense/lax distinction in your vowels, which I am unaware of there existing in any natlang. (German contrasts /ɛ ɛː eː/ by some analyses, which could be considered close, but whether /ɛː/ is distinct from /eː/ is debated.)

[Soap]

OK so from what you both posted it looks like high tone can cause vowels to develop retracted tongue root, which means that they will be likely to have voiceless onsets if the language previously made no voicing distinction for stops and may pull their vowels slightly inward towards the center of the vowel space, making /i u/ move towards [ɪ ʊ] and possibly on down to [e o] if this would cause no major misunderstandings. Lastly, it may cause pharyngealization of the vowel, at least near the beginning.

This surprises me, as I had assumed that high tone would be linked to high vowels. However I get that the term "high" is being used for two unrelated concepts so I shouldn't really be all that surprised. I'm still surprised that pharyngealization is linked to high tone, though, since although I actually derived it from high-tone syllables myself, I had assumed that the "harshness" of the pharyngealization would quickly drag the two pharyngealized tones down to the very bottom of the pitch scale. I think I'll still do that, since I lowered other high tones too.

I think I still have a huge mess on my hands because I simply have way too many marginal phonemes. The parent language is beautiful but any possible direction of evolution causes the phonology to become far more unstable than that of the parent language.

Fronting of velars to alveolars of some description, and uvulars to velars could then phonemicise the front/back distinction leaving me with /i iː ɪ u uː ʊ e eː ɛ o oː ɔ a aː (ə or ʌ)/ which would then probably be unstable as hell thanks to the 5-way height contrast. Thoughts?

Perhaps diphthongs instead of the long vowels? If not, remember that sound shifts are usually messy. A neat happy three-tone, five-vowel system doesnt have to produce a neat happy one-tone, fifteen-vowel system. You could certainly merge some of those, and if you have syllable stress reappearing as a response to the loss of tone you could make those losses conditional, perhaps with most of them occurring only in unstressed syllables, so that you still have a fifteen vowel system, or close to it, in stressed syllables but in unstressed syllables you may just have five or even fewer. (Long vowels commonly reduce in unstressed position.)

I guess if your third tone is a falling tone, it wouldn't be out of the question to form diphthongs where the vowel starts out like the high-tone type and ends up like the low-tone type. This probably really would be unstable ... it's the sort of idea I'd consider but probably wouldn't happen in a natlang. But even I would quickly rework the resulting diphthongs into ones that wrap around more, so that instead of /iɪ/ I'd have /iə/, possibly soon leading to /jə/, and so on.