(Conlangs) Q&A Thread - Quick questions go here

[Ahzoh]

What are ways to develop word-initial geminates? I asked this question on the other board and was told it could arise from elision of first syllables like in Malayalam, vowel syncope, and this thing Japanese did to get "tte" from "to iu".
But what are other ways, especially involving hetero-organic clusters?

Also, is /s/ > /a/ ever possible like "ism(a)" > "iam(a)"?

[Porphyrogenitos]

Well, pretty much by any other way you get geminates, but just at the start of a syllable.

Deletion of a word-initial vowel before a geminate:

attama > ttama

Deletion of vowels (due to whatever reason) between two already identical consonants:

titama > ttama

Assimilation of a cluster and then deletion of an initial vowel:

aktama > attama > ttama

Assimilation of a word-initial cluster:

ktama > ttama
mbama > bbama
rtomo > rromo

Development of initial geminates in certain syntactic contexts due to word-initial mutations resulting from sandhi effects:

ad kasa > ak‿kasa
ad > a
ak‿kasa now reanalyzed as a kkasa

Then musu, coined centuries after a lost its final /d/, becomes mmusu after a because it's a regular morphosyntactic process now

And I think s > a would be plausible if /s/ was shifted to /h/, which is both possible and extremely common. From there coda /h/ could vocalize to /a/, especially if it first became [ɦ]

[Sumelic]

is /s/ > /a/ ever possible like "ism(a)" > "iam(a)"?

With enough sound changes, almost anything is possible. I can imagine something like /isma/ > /iʃma/ > /iʂma/ > /ixma/ > /iχma/ > /iʁma/ > /iɐma/ > /iama/.

[Ahzoh]

Assimilation of a word-initial cluster:

ktama > ttama
mbama > bbama
rtomo > rromo

Any C1 could assimilate to C2? Or are there certain ones that are more resistant? And I only want /p t d k s z/ to geminate.

Any how might I prevent this from being leveled with other words that share the root since the conlang is a tricon language? Have *həsólyā > ssəlya but *māhəsólā > mahsəla and *həsuólē > hsuli/husli.

[DesEsseintes]

is /s/ > /a/ ever possible like "ism(a)" > "iam(a)"?

With enough sound changes, almost anything is possible. I can imagine something like /isma/ > /iʃma/ > /iʂma/ > /ixma/ > /iχma/ > /iʁma/ > /iɐma/ > /iama/.

Or just /isma/ → /ihma/ → /iama/

[Porphyrogenitos]

Theoretically, yes, any consonant could undergo complete assimilation; essentially taking on all features of the consonant adjacent to it and retaining only its existence as a unit of length. E.g. amta > atta or abhu > ahhu.

But complete assimilation is more "forceful" or "extreme", so it occurs less often crosslinguistically (not that you couldn't have it occur - a lot, even!), and when it does occur it more often occurs between consonants that already share a lot of features - e.g. sz > zz only involves changing one feature, voiceless to voiced.

I could definitely see it getting blocked/reversed out of analogy if there are alternating forms of one word/root that put the two consonants in and out of the assimilation environment. Unless you mean you don't want these cases to be analogized out of existence? Like are you saying you want it to be ssəlya, mahsəla, hsuli/husli or do you want it to be hsəlya, mahsəla, hsuli/husli?

[DesEsseintes]

I’m mulling over devoicing processes in Núta. Two voiceless fricatives /s h/ occur in Núta, and I’ve decided that all unstressed vowels devoice before /h/, regardless of the quality of the previous consonant. Devoicing is indicated with a dot above:

/neyáwahte/ neyá·wȧhte

Furthermore, the /w/ in onset of the devoiced syllable may also devoice to [ʍ]. Similarly for /n j r/ → [n̥ ɾ̥ j̥].

Now, I’m thinking /s/ can also cause devoicing, but I’m thinking this may only occur if the onset is one of /t k ʔ/ (i.e. any stop). I don’t think the asymmetry is unnaturalistic, but it does bother me a bit. Thoughts? Any suggestions on tweaking this?

The phonology and a brief outline can be found here.

[Davush]

I’m mulling over devoicing processes in Núta. Two voiceless fricatives /s h/ occur in Núta, and I’ve decided that all unstressed vowels devoice before /h/, regardless of the quality of the previous consonant. Devoicing is indicated with a dot above:

/neyáwahte/ neyá·wȧhte

Furthermore, the /w/ in onset of the devoiced syllable may also devoice to [ʍ]. Similarly for /n j r/ → [n̥ ɾ̥ j̥].

Now, I’m thinking /s/ can also cause devoicing, but I’m thinking this may only occur if the onset is one of /t k ʔ/ (i.e. any stop). I don’t think the asymmetry is unnaturalistic, but it does bother me a bit. Thoughts? Any suggestions on tweaking this?

The phonology and a brief outline can be found here.

I like the devoicing especially since it's not that common in conlangs. I have a hard time devoicing /a/ though for some reason. I like the asymmetry too - I think we over think these things sometimes, and then examples from natlangs can be even weider...maybe you could go the Japanese way and only have devoicing of high vowels with /s/? I suppose one way to think of it is that the -voice feature of /h/ has more 'spreading' influence (maybe because it is more 'vowel-ish?) than /s/ so devoices vowels more regularly? Perhaps /s/ is on its way to devoicing all vowels, but not there yet? (Or these explanations may be totally implausible... )

@Esoanem, thanks for the example from Latin > French/Spanish. Incomplete deletion of /p t k/ seems interesting. I wonder if that was a case of a sound change starting, but not reaching all parts of the lexicon? Or was it more random?

A problem I am also having is deriving the Classical Qutrussan stop series. The problem is, atm a single unvoiced consonant between vowels must have come from an ejective, and I think that amount of ejectives is probably unlikely.

E.g.
/puttuku/ < /p'ottuko/ but the reflex of this should be /'puttuggu/ because /k/ usually > /gg/ intervocalically.

For this type, I can always say voicing harmony comes into play, but words with mixed voicing are a bit more difficult:
/qaubu/ < /k'a:upo/ - why didn't /p/ remain as /p/ because of voicing harmony? Or would it be reasonable to just say 'voicing harmony *usually* works' but in some cases it randomly didn't?

[Ahzoh]

Unless you mean you don't want these cases to be analogized out of existence? Like are you saying you want it to be ssəlya, mahsəla, hsuli/husli or do you want it to be hsəlya, mahsəla, hsuli/husli?

I am saying I want the former. Although maybe I should go for hsuli > ssuli.

[Creyeditor]

is /s/ > /a/ ever possible like "ism(a)" > "iam(a)"?

With enough sound changes, almost anything is possible. I can imagine something like /isma/ > /iʃma/ > /iʂma/ > /ixma/ > /iχma/ > /iʁma/ > /iɐma/ > /iama/.

Or just /isma/ → /ihma/ → /iama/

/isma/ →/isəma/ → /isama/ → /ihama/ → /iama/ might be the least weird solution. Vowel epenthesis, fricative lention and /ə/→/a/ are unproblematic.

[DesEsseintes]

is /s/ > /a/ ever possible like "ism(a)" > "iam(a)"?

With enough sound changes, almost anything is possible. I can imagine something like /isma/ > /iʃma/ > /iʂma/ > /ixma/ > /iχma/ > /iʁma/ > /iɐma/ > /iama/.

Or just /isma/ → /ihma/ → /iama/

/isma/ →/isəma/ → /isama/ → /ihama/ → /iama/ might be the least weird solution. Vowel epenthesis, fricative lention and /ə/→/a/ are unproblematic.

That’s only if you’re happy with schwa epenthesis occurring in virtually every cluster in the language. I’m not sure Ahzoh wants the ramifications of that.

[Pabappa]

In southern Spain vowels are centralized before a coda / s/.... I could see this developing into diphthongs. Finnish has nakra>naura, which i suspect is part of a long and polyconditional series of changes winnowing down various inherited clusters, such that every cluster would behave differently and also be affected by the vowels. But somewhere in between the unconditional shift of Spanish and the highly conditional shift of Finnish I'm sure you can find a way to get /a/ at least some of the time.

[Porphyrogenitos]

Debuccalization of coda /s/ to /h/ (or /x/) and vocalization of the coda /h/ to /a/ is the simplest option.

[DesEsseintes]

Debuccalization of coda /s/ to /h/ (or /x/) and vocalization of the coda /h/ to /a/ is the simplest option.

That’s what I said.

[Salmoneus]

Debuccalization of coda /s/ to /h/ (or /x/) and vocalization of the coda /h/ to /a/ is the simplest option.

Clearly. Though I might say /h/ > /@/ > /a/.

An alternative is coda /s/ > lost with compensatory lengthening of the vowel > diphthong... but obviously that might interfere with more things.

[Creyeditor]

is /s/ > /a/ ever possible like "ism(a)" > "iam(a)"?

With enough sound changes, almost anything is possible. I can imagine something like /isma/ > /iʃma/ > /iʂma/ > /ixma/ > /iχma/ > /iʁma/ > /iɐma/ > /iama/.

Or just /isma/ → /ihma/ → /iama/

/isma/ →/isəma/ → /isama/ → /ihama/ → /iama/ might be the least weird solution. Vowel epenthesis, fricative lention and /ə/→/a/ are unproblematic.

That’s only if you’re happy with schwa epenthesis occurring in virtually every cluster in the language. I’m not sure Ahzoh wants the ramifications of that.

Maybe only in sC clusters? It just came to my mind, when I saw someone claim glottal stop to /e/ for hebrew where most people assume insertion of /e/ and deletion of the glottal stop.

[Vlürch]

You could use -ablative as the suffix instead of -elative , and therefore prevent vowel elision .... and this preserves nearly the same meaning as the original.

True, but I don't want to because there are literally no results on Google for postablative case except some kind of spam site. I know that's a stupid reason, but well...

Grammar writers mostl come up with names for new cases on the spot, based on some (more or less) Pseudo-Latin. So no pressure for conlangers to do any better.

Oh, that's good to know. I guess calling it anteelative is good enough, then...

[DesEsseintes]

Suppose a language with extensive vowel sandhi has rules such as the following:

a + a → a, e + e → e, i + i → i, etc.
a + {e i} → ai
a + {o u} → au

The diphthongs ai au then simplify to e o. Now here’s the question: Would it be odd for the e o resulting from diphthongs to be long while the e o resulting from e + e and o + o are short?

e + e → e
a + {e i} → ai → ē

[Porphyrogenitos]

I don't think so. Diphthongs often tend to be longer than single vowels.

Although, if you mean that there was an original /ee/ that stayed long but the new /e/ + /e/ just became /e/, that might be a little odd. I guess if you conceived of the rule not as /e/ + /e/ = /e/ but rather as "a vowel following a vowel of identical quality is deleted" it wouldn't be implausible.

[DesEsseintes]

I don't think so. Diphthongs often tend to be longer than single vowels.

Although, if you mean that there was an original /ee/ that stayed long but the new /e/ + /e/ just became /e/, that might be a little odd. I guess if you conceived of the rule not as /e/ + /e/ = /e/ but rather as "a vowel following a vowel of identical quality is deleted" it wouldn't be implausible.

There was no original long ē. After I posted the question I actually thought of a solution: sequences of identical vowels simplified at an earlier stage than the evolution of diphthongs into long vowels. Then there’s no ‘clash’. Basically, it’s the same thing you suggested.

Thanks.