Sound Change Quickie Thread

[Nortaneous]

Velarization could be argued to be "fake" phonologically, in the sense that there arguably aren't any cases where velar constriction specifically rather than any kind of dorsal constriction, labialization, or lack of a contrasting feature is actually the essential phonetic cue involved in distinguishing sounds from each other. English is a pretty good example--the distinction between 'light' and 'dark' L isn't between plain and velarized /l/; it varies among dialects but for many American English speakers all instances of /l/ are velarized and the 'dark' ones involve other features. For the varieties where a plain [l] actually is present, many other features indicate a dark L such as less occlusion, other dorsal constriction, labialization, etc.

Does it make sense to describe AmE as having a contrast between 'light' and 'dark' L at all? There are contexts where /l/ is often vocalized (I hear things like [hɛo̯ˤp] from people who don't otherwise have l-vocalization, for example, and for people with l-restoration in -alm it's usually realized as [-ɔˤm]), but that's distinct enough from the prototypical case that I don't think anything is gained from the term.

[mae]

-

[bradrn]

Three questions, all to do with the glottal stop:

1. Is ʔ → ŋ / [+nasal+vowel] _ plausible? (e.g. tãʔe → tãŋe)
2. Is the following set of changes plausible?

   [+vowel]ʔ[+vowel] → [+long] / _ (e.g. aʔi → aː, oːʔə → oː)
   ʔ → Ø / C_

3. After applying those changes, the only glottal stops left are at the beginning and end of a word, and immediately before consonants. Is it more usual for these ʔC clusters to stay as they are, or to turn into glottalised consonants?

[Knit Tie]

That is a possibility, but I would limit it to the most stressed positions, i.e. initial consonants in stressed onsets and geminates.

Just there?

Sorry for the delay - was on a trip and did not have access to my laptop, and typing on my phone is a pain.

You could have unconditional fortition of /ɰ/ - it just is to me that it is a rather "weak" consonant to begin with, which makes it seem like that if it were to undergo fortition at all it would undergo fortition in the positions most prone to this, i.e. at the starts of stressed onsets and when geminate.

Does English even have geminates?

Also, what is a fun thing one can do with the English diphthongs diachronically other than smoothing them over?

[Xwtek]

Does English even have geminates?

Yes, but accross morpheme like back kick, this saddle, midday. There is no geminate morpheme internally. Phonemically, it's a cluster of two identical consonant from different morpheme, though.

Also, what is a fun thing one can do with the English diphthongs diachronically other than smoothing them over?

Turning it into V+/j/ or /w/ and do something with /j/+C.

[Xwtek]

ʔ → ŋ / [+nasal+vowel] _

I don't know, but if you do
Ṽ → Vŋ
ŋʔ → ŋ
Vŋ → Ṽ / _{C, #}

It should be possible. However, there's a problem if your language allows something like ãʔka as the result is ãka, not *aŋka.

[+vowel]ʔ[+vowel] → [+long] / _

Yes, just delete the glottal stop and resolve the resulting hiatus by lengthening the first vowel. On the other hand, if you have something like aʔio, probably it's not realistic.

ʔ → Ø / C_

There's the exact sound change in Proto-Siouan-Iroquoian -> Proto-Iroquoian according to index diachronica.

After applying those changes, the only glottal stops left are at the beginning and end of a word, and immediately before consonants. Is it more usual for these ʔC clusters to stay as they are, or to turn into glottalised consonants?

I don't know, but both are realistic. In Indonesia, where nonsyllable final glottal stop occurs only in loanwords, the ʔC remained cluster. However it's very easy to turn them into glottalized consonant.

[Tropylium]

Yeah, Icelandic has instances of [tj] that go back to Proto-Germanic (roughly 2000 years), e.g. *hwatjanan > hvetja 'to sharpen' (the actual English cognate is "to whet").

There's also an area in northwestern Finland that has retained /t̪j/ for the last ~2000 years, though almost everywhere across Finnic it's gone to either at least /tʲj/ (then possibly > dʲj > dʲː > ɟː) or *d̪j > /ðj/ (then mostly > ɾj > rj or > lj).

In both cases this is dental [t̪j] and not alveolar [tj] though, which could be relevant: dentals tend to resist palatalization better than alveolars.

Also, what is a fun thing one can do with the English diphthongs diachronically other than smoothing them over?

Break them into two syllables; before liquids this is common enough already (/aɪl/ > /aɪ.(j)əl/, /aʊr/ > /aʊ.(w)ər/ etc.), but you could go further with things like /aɪb/ > /ajʊb/.

[Nortaneous]

Also, what is a fun thing one can do with the English diphthongs diachronically other than smoothing them over?

Break them into two syllables; before liquids this is common enough already (/aɪl/ > /aɪ.(j)əl/, /aʊr/ > /aʊ.(w)ər/ etc.), but you could go further with things like /aɪb/ > /ajʊb/.

au > æ̝jə happens in some Southern AmE

[bradrn]

ʔ → ŋ / [+nasal+vowel] _

I don't know, but if you do
Ṽ → Vŋ
ŋʔ → ŋ
Vŋ → Ṽ / _{C, #}

It should be possible. However, there's a problem if your language allows something like ãʔka as the result is ãka, not *aŋka.

That’s certainly one solution. However, I was asking specifically about doing that rule in one go, as opposed to doing several rules in series.

[+vowel]ʔ[+vowel] → [+long] / _

Yes, just delete the glottal stop and resolve the resulting hiatus by lengthening the first vowel. On the other hand, if you have something like aʔio, probably it's not realistic.

That makes sense. Luckily I don’t have anything like aʔio; in fact, the phonotactics forbid vowel hiatuses. (Is that the correct plural?)

ʔ → Ø / C_

There's the exact sound change in Proto-Siouan-Iroquoian -> Proto-Iroquoian according to index diachronica.

Of course… I should have checked ID first! Anyway, thanks for doing it for me!

After applying those changes, the only glottal stops left are at the beginning and end of a word, and immediately before consonants. Is it more usual for these ʔC clusters to stay as they are, or to turn into glottalised consonants?

I don't know, but both are realistic. In Indonesia, where nonsyllable final glottal stop occurs only in loanwords, the ʔC remained cluster. However it's very easy to turn them into glottalized consonant.

Thank you!

[akam chinjir]

In both cases this is dental [t̪j] and not alveolar [tj] though, which could be relevant: dentals tend to resist palatalization better than alveolars.

Thanks Tropylium, that's good to know!

I wonder---could it make any sense to have dental [t̪] and [n̪] alongside alveolar [d] and [s]? (I want sj to end up just as s, maybe s → ʃ → s; and I want dj to do something like dj → d͡ʒ → ʒ → r; but for tj and nj I just want c and ɲ. I figure this divergence may be easier to justify if palatalisation hits d and s before t and n.)

[Richard W]

I wonder---could it make any sense to have dental [t̪] and [n̪] alongside alveolar [d] and [s]? (I want sj to end up just as s, maybe s → ʃ → s; and I want dj to do something like dj → d͡ʒ → ʒ → r; but for tj and nj I just want c and ɲ. I figure this divergence may be easier to justify if palatalisation hits d and s before t and n.)

I'm not sure of the cause, but 'palatalisation' of English /d/ is more advanced than palatalisation of /t/ (compare drink v. trim and duke v tube). The palatalisation of /s/ is complete in some words (e.g. sure and sugar); I don't notice any palatalisation of English /n/.

[Travis B.]

I wonder---could it make any sense to have dental [t̪] and [n̪] alongside alveolar [d] and [s]? (I want sj to end up just as s, maybe s → ʃ → s; and I want dj to do something like dj → d͡ʒ → ʒ → r; but for tj and nj I just want c and ɲ. I figure this divergence may be easier to justify if palatalisation hits d and s before t and n.)

I'm not sure of the cause, but 'palatalisation' of English /d/ is more advanced than palatalisation of /t/ (compare drink v. trim and duke v tube). The palatalisation of /s/ is complete in some words (e.g. sure and sugar); I don't notice any palatalisation of English /n/.

Depends. Here drink and trim are equally palatalized, while tube (due to /u/, not /ju/ as /j/ was lost long ago here) is more palatalized than duke.

[bradrn]

I wonder---could it make any sense to have dental [t̪] and [n̪] alongside alveolar [d] and [s]? (I want sj to end up just as s, maybe s → ʃ → s; and I want dj to do something like dj → d͡ʒ → ʒ → r; but for tj and nj I just want c and ɲ. I figure this divergence may be easier to justify if palatalisation hits d and s before t and n.)

I'm not sure of the cause, but 'palatalisation' of English /d/ is more advanced than palatalisation of /t/ (compare drink v. trim and duke v tube). The palatalisation of /s/ is complete in some words (e.g. sure and sugar); I don't notice any palatalisation of English /n/.

Depends. Here drink and trim are equally palatalized, while tube (due to /u/, not /ju/ as /j/ was lost long ago here) is more palatalized than duke.

Same with me. drink and trim are [d͡ʒɹ̠ʷɪŋk] and [t͡ʃɹ̠ʷɪm], while duke and tube are [dᶻʲʉk] and [tᶴʲʉːb].

[Nortaneous]

I'm not sure of the cause, but 'palatalisation' of English /d/ is more advanced than palatalisation of /t/ (compare drink v. trim

Other way around for me - I have tr > chr but not dr > jr.

[Xwtek]

In both cases this is dental [t̪j] and not alveolar [tj] though, which could be relevant: dentals tend to resist palatalization better than alveolars.

Thanks Tropylium, that's good to know!

I wonder---could it make any sense to have dental [t̪] and [n̪] alongside alveolar [d] and [s]? (I want sj to end up just as s, maybe s → ʃ → s; and I want dj to do something like dj → d͡ʒ → ʒ → r; but for tj and nj I just want c and ɲ. I figure this divergence may be easier to justify if palatalisation hits d and s before t and n.)

The dental [t̪] and alveolar [d̺] is realistic. It's attested in some Austronesian languages like Standard Indonesian.

[akam chinjir]

I'm with those with variable palatalisation. I have (roughly) dʒrink and trink(et), but druck and tʃruck. (The examples I've come up with so far make this seem like it might be a frequency effect?)

Thanks Akganka, that's good to know!

[Xwtek]

After I read this article, it says that this change is not natural:

*C > 0 / _#

But why?

[akam chinjir]

After I read this article, it says that this change is not natural:

*C > 0 / _#

But why?

It sounds like the issue is that word-final lenition tends to target particular classes of segments, and tends to involve a series of changes. Like, plosives might debuccalise to glottal stops, which are later lost; or you might get a similar sequence with fricatives that goes via h. So you wouldn't expect to get blanket loss of all word-final consonants unless a bunch of these separate processes occurred all at roughly the same time. Basically, you'd expect it to happen only as a coincidence.

Apparently though the evidence suggests that several times, independently, Austronesian languages have lost all word-final consonants, all at once, and with no evidence that it was a gradual process. That's the change she says seems unnatural (and maybe unattested elsewhere in the world, see her second footnote).

[bradrn]

After I read this article, it says that this change is not natural:

*C > 0 / _#

But why?

It sounds like the issue is that word-final lenition tends to target particular classes of segments, and tends to involve a series of changes. Like, plosives might debuccalise to glottal stops, which are later lost; or you might get a similar sequence with fricatives that goes via h. So you wouldn't expect to get blanket loss of all word-final consonants unless a bunch of these separate processes occurred all at roughly the same time. Basically, you'd expect it to happen only as a coincidence.

Apparently though the evidence suggests that several times, independently, Austronesian languages have lost all word-final consonants, all at once, and with no evidence that it was a gradual process. That's the change she says seems unnatural (and maybe unattested elsewhere in the world, see her second footnote).

Interestingly enough, the Index Diachronica says that something similar can happen for vowels; that is, it seems that V > 0 / _# is attested. Do you have any ideas on why this one is natural, whereas Akangka’s change isn’t?

[akam chinjir]

Interestingly enough, the Index Diachronica says that something similar can happen for vowels; that is, it seems that V > 0 / _# is attested. Do you have any ideas on why this one is natural, whereas Akangka’s change isn’t?

The tautological answer is that vowels form a natural class more easily than consonants as a group do, so that V > 0 / _ # is more analogous to [+plosive] > 0 / _# (especially if it's actually more like V > ə > 0 / _#, like [+plosive] > ʔ > 0 / _#). I'm not sure how illuminating that is, though.

(Blevins has fairly specific ideas about the phonetic conditions that make particular changes natural, but my knowledge of such stuff is pretty basic.)