Sound Change Quickie Thread

[Darren]

How realistic is ʍ → hʷ → h?

Very: I believe it happened in most English dialects.

In most it's become /w/ apart from before rounded vowels where it did become /h/, because it was perceived as allophonically rounded [hʷ] (which is why who, whom, whoop all have /h/). Some Scots dialect has probably done it. It's realistic anyway, even if it hasn't been attested.

Also, can
ʃp ʃt ʃk → f ʃt x, i.e. can the middle cluster remain as the other two simplify?

I'm not sure about this, but I think /t/ is often more stable than /p/ especially and to a lesser extent /k/. Even if it was fricated initially (e.g. to /ʃθ/) it could easily come back to /ʃt/.

[bradrn]

How realistic is ʍ → hʷ → h?

Very: I believe it happened in most English dialects.

In most it's become /w/ apart from before rounded vowels where it did become /h/, because it was perceived as allophonically rounded [hʷ] (which is why who, whom, whoop all have /h/). Some Scots dialect has probably done it. It's realistic anyway, even if it hasn't been attested

Sorry! For me, ‘who’ and ‘whom’ are the first words which come to mind when thinking about English /ʍ/, and I completely forgot about all the other words where it changed to /w/ rather than /h/.

[Darren]

Sorry! For me, ‘who’ and ‘whom’ are the first words which come to mind when thinking about English /ʍ/, and I completely forgot about all the other words where it changed to /w/ rather than /h/.

No need to apologise. It did happen in most English dialects, just under certain conditions.

[Richard W]

How realistic is ʍ → hʷ → h? Also, can
ʃp ʃt ʃk → f ʃt x, i.e. can the middle cluster remain as the other two simplify?

While reasonable, I’d give greater credence to ʍ → hw → h. I wouldn’t say that had happened in isolation in English; /hw/ > /h/ has generally happened as a result of a following long, rounded vowel.

I’m not sure about /ʃt/, but /st/ can behave like a single consonant.

[Ahzoh]

What can I do to make the construct state forms of nouns as irregular as Arabic broken plurals? I'm not sure how to go about it. Stuff like making nouns like sārb-e and ʾābz-a have construct state patterns like surūb and ʾubūz but nouns like semṭ-a and xaśr-a become something like semeṭ and xeśer

I know I can just do some simple vowel changes for biconsonantal roots' construct state forms, such as:
kap-a / kapi > kap-a / keb

[Man in Space]

What can I do to make the construct state forms of nouns as irregular as Arabic broken plurals? I'm not sure how to go about it. Stuff like making nouns like sārb-e and ʾābz-a have construct state patterns like surūb and ʾubūz but nouns like semṭ-a and xaśr-a become something like semeṭ and xeśer

I know I can just do some simple vowel changes for biconsonantal roots' construct state forms, such as:
kap-a / kapi > kap-a / keb

What's the general pattern of your regular states and how do you normally form construct states? Also, what's the full scheme of vowels we're dealing with?

[Ahzoh]

What can I do to make the construct state forms of nouns as irregular as Arabic broken plurals? I'm not sure how to go about it. Stuff like making nouns like sārb-e and ʾābz-a have construct state patterns like surūb and ʾubūz but nouns like semṭ-a and xaśr-a become something like semeṭ and xeśer

I know I can just do some simple vowel changes for biconsonantal roots' construct state forms, such as:
kap-a / kapi > kap-a / keb

What's the general pattern of your regular states and how do you normally form construct states? Also, what's the full scheme of vowels we're dealing with?

Most, if not all, noun stems have either CV(V)CC- or CV(V)CVVC- shape [with a short V gender ending (or VC for plural)] with stress on the last, heaviest syllable. The typical construct state is formed by simply removing the gender vowel, with some vowel epenthesis (usually an echo vowel) since Vrkhazhian disprefers final and initial consonant clusters.

The vowel system looks like this:

Code: Select all

/i iː 	u uː/
/ɛ ɛː 	ɑ ɑː/

[bradrn]

What can I do to make the construct state forms of nouns as irregular as Arabic broken plurals? I'm not sure how to go about it. Stuff like making nouns like sārb-e and ʾābz-a have construct state patterns like surūb and ʾubūz but nouns like semṭ-a and xaśr-a become something like semeṭ and xeśer

I know I can just do some simple vowel changes for biconsonantal roots' construct state forms, such as:
kap-a / kapi > kap-a / keb

What's the general pattern of your regular states and how do you normally form construct states? Also, what's the full scheme of vowels we're dealing with?

Most, if not all, noun stems have either CV(V)CC- or CV(V)CVVC- shape [with a short V gender ending (or VC for plural)] with stress on the last, heaviest syllable. The typical construct state is formed by simply removing the gender vowel, with some vowel epenthesis (usually an echo vowel) since Vrkhazhian disprefers final and initial consonant clusters.

The vowel system looks like this:

Code: Select all

/i iː 	u uː/
/ɛ ɛː 	ɑ ɑː/

So, to make sure I understand this, if the noun is tɑns-, and the gender ending is -ɛ for singular and -ɛt for plural, then the noun is declined as follows:

	Non-construct	Construct
Singular	tɑnsɛ	tɑns
Plural	tɑnsɛt	tɑns[ɑ]t

Is this correct?

[Kuchigakatai]

What can I do to make the construct state forms of nouns as irregular as Arabic broken plurals? I'm not sure how to go about it. Stuff like making nouns like sārb-e and ʾābz-a have construct state patterns like surūb and ʾubūz but nouns like semṭ-a and xaśr-a become something like semeṭ and xeśer

I know I can just do some simple vowel changes for biconsonantal roots' construct state forms, such as:
kap-a / kapi > kap-a / keb

1. Add morphophonological sound changes in inflections
2. Apply phonological decay in vowels (drop them, add them, harmonize them on occasion)
3. Analogize the results to everything that looks similar and their mother
4. PROFIT!

You could begin with, say, [ˈsa:rub], plural [ˈsa:rubun]. Then:
1. An inflection suffix with shape -VC(C) attracts stress and makes the stressed vowel long if it's in an open syllable. Now you have [ˈsa:rub] plural [sa:ˈru:bun].
2. Words ending in -VrVC simplify to -VrC, and this now makes the singular become [ˈsɑ:rb]. Meanwhile, pre-tonic open syllables with [a(:)] reduce to a schwa, and word-final [n] is lost, so the plural is now [səˈru:bu]. Then pre-tonic schwas tend to fortify by copying the next vowel ([suˈru:bu]), and word-final vowels are lost, so your plural ends up as [suˈru:b].
3. This new pattern of [ˈsɑ:rb] plural [suˈru:b] affects other inherited words with the shape CVrC, so original [kɛrtʃ] plural [ˈkɛrtʃun] is now [kɛrtʃ] plural [kuru:tʃ] (from pure sound changes you'd expect [kɛrtʃ] to have the identically-sounding plural [kɛrtʃ]).
4. PROFIT!

Just an example, you can also do it in a multitude of other ways.

Regarding [ˈxɑʃrɑ] plural [ˈxɛʃɛr], you could start off with collective [ˈxaʃarði] singulative [ˈxaʃaraŋ]. The [ i ] at the end of the collective modifies the preceding vowels through anticipatory metaphony, so [ˈxɛʃɛrði]. Meanwhile, "intertonic" vowels in an open syllable get lost: [ˈxaʃaraŋ] > [ˈxɑʃərɑŋ] > [ˈxɑʃrɑŋ]. [rð] then simplifies to [ri], so [ˈxɛʃɛri]. Then final vowels are lost and you have [ˈxɛʃɛr] singulative [ˈxɑʃrɑŋ]. Then word-final nasals are lost, and you end up with [ˈxɛʃɛr] singulative [ˈxɑʃrɑ]. PROFIT!

Contra Man in Space, I don't think your initial conditions matter much, although you may need a longer time scale in some situations. I can imagine accomplishing this kind of thing starting off Latin, modern English or modern Mandarin.

[Ahzoh]

So, to make sure I understand this, if the noun is tɑns-, and the gender ending is -ɛ for singular and -ɛt for plural, then the noun is declined as follows:

	Non-construct	Construct
Singular	tɑnsɛ	tɑns
Plural	tɑnsɛt	tɑns[ɑ]t

Is this correct?

Not quite.
A typical masculine gendered noun would be declined like this:
Nominative:

	Absolute	Construct
Singular	ɑːbzɛ	ɑːbɑs
Plural	ɑːbzɛn	ɑːb(ɑ)sɛː

Oblique:

	Absolute	Construct
Singular	ɑːbzi	ɑːbɑs
Plural	ɑːbzin	ɑːb(ɑ)siː

A typical feminine gendered noun would be declined like this:
Nominative:

	Absolute	Construct
Singular	sɛmtʼɑ	sɛmɛd
Plural	sɛmtʼɑn	sɛm(ɛ)dɑː

Oblique:

	Absolute	Construct
Singular	sɛmtʼu	sɛmɛd
Plural	sɛmtʼun	sɛm(ɛ)duː

[Ahzoh]

You could begin with, say, [ˈsa:rub], plural [ˈsa:rubun]. Then:

I quite like these, but I am not trying to create broken plurals but rather, broken construct states. The plurals are all regular, as indicated in the tables above.

Which does, of course, lend an issue as I can't historically derive broken construct states from collective forms of nouns, so I have to get them from somewhere else.

Ideas I have considered is that a noun's construct state stem alternations would originate from morphemes that indicate either alienable possession or inalienable possession. Another idea is a morpheme used for possessing person, like the agentive noun form C₁āC₂C₃- which has the construct state form of C₁aC₂wāC₃-

[Kuchigakatai]

I quite like these, but I am not trying to create broken plurals but rather, broken construct states. The plurals are all regular, as indicated in the tables above.

Ah, I see I definitely misread your post.

Which does, of course, lend an issue as I can't historically derive broken construct states from collective forms of nouns, so I have to get them from somewhere else.

Why would collective nouns present a problem?

Ideas I have considered is that a noun's construct state stem alternations would originate from morphemes that indicate either alienable possession or inalienable possession.

Yeah, that's a good option.

Alternatively you could start with a language that already has unpredictable construct morphemes (they just happen to be mere affixes, not changing the phonology of the noun they attach to at all). As time passes these unpredictable affixes do wreak phonological havoc.

Another choice would be to use an adposition that means "of" and reinterpret it as part of the possessed noun. As time passes, the adposition-now-affix wreaks phonological havoc on the word it attaches to.

Another choice would be to copy what Hebrew did, using the prosody of a phrase to alter words by how much stress they receive within a phrase. Since a possessor noun often has greater stress than its possessed noun, it can have a different phonological evolution. In Biblical Hebrew, construct nouns often present what looks synchronically like vowel reduction: [le:ˈβɔ:β] ~ [ləˈβaβ] 'heart', [bərɔ:ˈxo:θ] ~ [birˈxo:θ] 'blessings', [ʃɔ:ˈlo:m] ~ [ʃəˈlo:m] 'peace', although historically some "reductions" are conserved vowels because the absolute form underwent vowel lengthening. There is also French-like liaison in the [θ] of [ʃɔ:ˈnɔ:] ~ [ʃəˈnaθ] 'year'. This is simple vowel reduction, but this ablaut already looks pretty decent.

Biblical Hebrew also has "pausal" forms that do the opposite of reduction, lengthening vowels for extra emphasis. Naturally, most "pausal" forms involve absolute nouns or verbs (not construct nouns, because they didn't have much stress). E.g. [qaw] > [qɔ:w] 'measuring cord, measuring line', [jo:ʔˈxal] > [jo:ʔˈxe:l] 'he eats'.

Another idea is a morpheme used for possessing person, like the agentive noun form C₁āC₂C₃- which has the construct state form of C₁aC₂wāC₃-

I don't understand this bit. It sounds like the starting stage of your language already has a broken construct state? Or is that w in CaCwaaC some kind of regular infix that doesn't count as broken?

[Ahzoh]

Biblical Hebrew also has "pausal" forms that do the opposite of reduction, lengthening vowels for extra emphasis. Naturally, most "pausal" forms involve absolute nouns or verbs (not construct nouns, because they didn't have much stress). E.g. [qaw] > [qɔ:w] 'measuring cord, measuring line', [jo:ʔˈxal] > [jo:ʔˈxe:l] 'he eats'.

Vrkhazhian has no copular verb, so the predicate noun is put into the construct state, so I could definitely have a pausal construct state.

I don't understand this bit. It sounds like the starting stage of your language already has a broken construct state? Or is that w in CaCwaaC some kind of regular infix that doesn't count as broken?

The starting language had a possessive morpheme wā that became an infix but the pattern CaCwaaC is a suppleted form of the agentive noun pattern CaaCC- so yes it is kind of irregular for that specific noun type but would be regular for other human nouns.

----------------------------------
My vowels underwent these sound changes:
/æ æː ɑ ɑː e eː o oː i iː u uː/ > /ɑ ɑː ə əː i iː u uː/ > /ɑ ɑː ɛ ɛː i iː u uː/
The mid vowels merged into a schwa which eventually lowered to fill the front open space while the open vowels merged into the open back vowels.

Given that the old alienable and inalienable possession suffixes were -i and -u respectively, this makes for some nice irregularity:

originally /æ/:
a ā > e ē (alienable)
a ā > a ā (inalienable)

originally /ɑ/:
a ā > a ā (alienable)
a ā > e ē (inalienable)

originally /e/:
e ē > i ī (alienable)
e ē > a ā (inalienable)

originally /o/:
e ē > e ē (alienable)
e ē > u ū (inalienable)

/i/:
i ī > aye āye (alienable)
i ī > u ū (inalienable)

/u/:
u ū > i ī (alienable)
u ū > awe āwe (inalienable)

A noun like ḳīśa "head" can become either ḳūl ʾamāk "my own head" or ḳāyel ʾamāk "my head trophy"

But I want to do more, especially involving consonants affecting vowel quality or length. I have a couple consonant mergers such as:
/θ ð/ > /s z/
/ç ʝ/ > /ʃ ʒ/ > /ɬ ɮ/
..And generally /h/ assimilates with a following consonant or lengthens a vowel preceding it if it's word-final.
I don't know if sibilancy or lack thereof would affect vowel quality or length.

[Knit Tie]

lateral affricates

Oooh?

I've met native speakers of English with lateral affricates instead of postalveolars; given the Turkmen lisp and the Armenian and Polish l-dealveolarization, sure, why not

I'm actually thinking of lateralising a lot of affricates in my latest version of my current conlang - could I ask you to tell me more about all three of these things?

[Nortaneous]

Oooh?

I've met native speakers of English with lateral affricates instead of postalveolars; given the Turkmen lisp and the Armenian and Polish l-dealveolarization, sure, why not

I'm actually thinking of lateralising a lot of affricates in my latest version of my current conlang - could I ask you to tell me more about all three of these things?

Guy I used to work with had consistent lateral affricates, something like [t̠ɬ̠ʲ d̠ɮ̠ʲ], for /tʃ dʒ/. Turkmen has s z > θ ð, i.e. a lisp as a sound change; this is well-attested elsewhere (Zhuang, Burmese, probably *s > *θ as an intermediate for *s > t in Vietnamese), and there's also some Sinitic variety or other with s > ɬ. Armenian had ɫ > ʁ and Polish had ɫ > w, both of which are extant speech defects in English. (ɫ > ʁ is IME the most common 'speech defect'. Tom Brokaw has it sometimes. I can't find it now, but there's a Youtube video I've probably posted before where Tom Brokaw talks very clearly about how Trump has vioghated the ghaw.)

[Kuchigakatai]

While I've met my share of English speakers who have /s z/ [ɬ ɮ], I wonder why in all my years I've never met a Spanish speaker with /s/ [ɬ]. (I have met people with /s/ [θ] though.) There is probably something about the phonetics of the lamino-alveolar Spanish /s/ that makes it non-existent or much more uncommon than in English. Same goes for /l/ [w] in parallel with English [ɫ] > [w].

On the other hand, I've met Spanish speakers with a [ɾ] > zero defect, who would say the likes of quiera /ˈkjeɾa/ [ˈkjea], hora /ˈoɾa/ [ˈoa] even in careful speech. I used to be one of them myself in fact until I was the late age of 9 years old, but I replaced [ɾ] with /d/ [ð].* I can't find the post, but Linguoboy mentioned a short while ago that he's noticed that [ɾ] is often dropped in American English in a similar way in fast speech, but I've never met anyone who legitimately can't pronounce it at all.


* So, hora [ˈoða], corto [ˈkoðto], tener [teˈneð]. I was able to pronounce it in Cr position though: preguntar [pɾeɣunˈtaɾ]. This also meant I actively avoided the extremely common word pero 'but', as it'd sound like pedo 'fart', which never failed to make my little classmates laugh at me. I'd avoid the conjunction, or otherwise use adverbials meaning 'however' like aun así ("even then") or the rather formal sin embargo if I really felt I needed to...

[Knit Tie]

Thanks! The lang I'm currently working on has ɫ → ɣ before vowels and full vocalisation otherwise. It also has three coronal series - dental, alveolar and "retroflex," which is actually apical postalveolar. And that actually brings me to my next question - how realistic it is for a language to develop three coronal series like that, plus a palatal one? I know that proto-Dravidian and most Australian languages had this system, but the retroflexes in there were subapical palatal, not just postalveolar.

[Nortaneous]

Thanks! The lang I'm currently working on has ɫ → ɣ before vowels and full vocalisation otherwise. It also has three coronal series - dental, alveolar and "retroflex," which is actually apical postalveolar. And that actually brings me to my next question - how realistic it is for a language to develop three coronal series like that, plus a palatal one? I know that proto-Dravidian and most Australian languages had this system, but the retroflexes in there were subapical palatal, not just postalveolar.

Are Australian retroflexes typically subapical? If so, I don't know of any good examples of a dental/alveolar/postalveolar-but-not-subapical plosive contrast, but dental/alveolar/subapical-retroflex is well-attested.

[Knit Tie]

Thanks! The lang I'm currently working on has ɫ → ɣ before vowels and full vocalisation otherwise. It also has three coronal series - dental, alveolar and "retroflex," which is actually apical postalveolar. And that actually brings me to my next question - how realistic it is for a language to develop three coronal series like that, plus a palatal one? I know that proto-Dravidian and most Australian languages had this system, but the retroflexes in there were subapical palatal, not just postalveolar.

Are Australian retroflexes typically subapical? If so, I don't know of any good examples of a dental/alveolar/postalveolar-but-not-subapical plosive contrast, but dental/alveolar/subapical-retroflex is well-attested.

Actually, reading into it, Australian retroflexes that are just apical postalveolar are well-attested (e.g. Nyangumarta), so it seems my question is answered with some ANADEW.

Anyway, I have another question, actually: I'm thinking about a phonolody with no native liquids at all, which makes me wonder what can I do with the Am-English tapped r to get rid of it before it manages to phonemise. Turn it into /n/? /d/?

[Travis B.]

On the other hand, I've met Spanish speakers with a [ɾ] > zero defect, who would say the likes of quiera /ˈkjeɾa/ [ˈkjea], hora /ˈoɾa/ [ˈoa] even in careful speech. I used to be one of them myself in fact until I was the late age of 9 years old, but I replaced [ɾ] with /d/ [ð].* I can't find the post, but Linguoboy mentioned a short while ago that he's noticed that [ɾ] is often dropped in American English in a similar way in fast speech, but I've never met anyone who legitimately can't pronounce it at all.

/me waves.

Well, not really - I can pronounce [ɾ] - and nasal flaps as well - perfectly fine, but it is lost in my speech (and in the speech of others here, such as my daughter) extremely frequently from intervocalic positions after stressed vowels, along with in a few other cases (e.g. after /r/ preceded by a stressed vowel). I usually have to make a conscious decision to re-add it in these positions, usually when someone does not understand its elision, or if its elision just does not feel appropriately formal for the occasion. Note that it is preserved in other positions (e.g. it will almost never be dropped in two successive syllables) - it is not being dropped without rhyme or reason.