Conlang Random Thread

[Pabappa]

i dont think this is how triconsonantal roots evolved in Semitic .... its just an example path to follow for how to do it in a conlang. semitic may have taken more than 10000 years to get the way it is, involving many tiny steps that were unrelated to the evolution of triconsonantal roots, and maybe even going backwards in some places.

i dont know much. i do know that there are other non-Semitic languages in the Afro-Asiatic family that have similar grammars, e.g. Egyptian, and some, such as Somali, where the vowels dont seem to change much between different forms of a given word. and it is believed that Semitic, at least, passed through a phase with many biconsonantal roots, because the present triconsonantal roots often form families in which those sharing the same final two consonants have transparently related meanings.

[Ares Land]

(One thing I’ve long pondered about triconsonality, by the way, is why it’s so rare. In our world, it seems to be just one of those unique quirks which pop up from time to time, like Austronesian alignment or distributed exponence or IE-style fusion, and up to now I saw no reason why it should be attested in any conworld, or even any language outside Semitic. But from what you’re saying, it sounds like it’s a feature which is really easy to evolve — vowel mutation, syncope and analogy rules aren’t exactly hard to come by. Perhaps it’s just rare to have all of (a–e) happen to the same language in the right order, even if each by itself is common.)

That's a pretty good question.
I think AA had some non-concatenative morphoogy already, a tendancy towards CVC roots and CV prefix, which helps fit the triconsonantal pattern and a high tolerance of difficult clusters. The simple vowel system also helped.

For conlanging purposes: the roots don't have to be triconsonantal either. (I did something like that with the Tarandim family: the roots need two syllables, but that may mean one, two, or three consonants; the patterns will apply regardless.)

Unless I’m missing something, shouldn’t the syncope rule come before e→i here?

Also, how would this work for roots where the first proto-vowel isn’t one of /e o/? Though I’m assuming you could simply add more vowel shift rules.

These aren't terribly well worked out
Adding more vowel shift rules would work! But both PIE (with ablaut) and proto-Semitic started out with a very reduced number of vowel qualities, and that reduces the number of ways the vowel can change.
(It's hard to generalize a pattern with English strong verbs -- even then, it happens fairly often -- but if English had only three vowel qualities, ablaut would probably be extremely productive.)

[bradrn]

(One thing I’ve long pondered about triconsonality, by the way, is why it’s so rare. In our world, it seems to be just one of those unique quirks which pop up from time to time, like Austronesian alignment or distributed exponence or IE-style fusion, and up to now I saw no reason why it should be attested in any conworld, or even any language outside Semitic. But from what you’re saying, it sounds like it’s a feature which is really easy to evolve — vowel mutation, syncope and analogy rules aren’t exactly hard to come by. Perhaps it’s just rare to have all of (a–e) happen to the same language in the right order, even if each by itself is common.)

That's a pretty good question.
I think AA had some non-concatenative morphoogy already, a tendancy towards CVC roots and CV prefix, which helps fit the triconsonantal pattern and a high tolerance of difficult clusters. The simple vowel system also helped.

Yeah, it’s true that AA languages were generally predisposed to do this sort of thing, which helps a lot. (Though I must admit I haven’t seen much evidence of this myself: the only branch other than Semitic I’ve looked at in detail is Chadic, and that doesn’t have too much non-concatenativity.)

Incidentally, I forgot to mention Yokuts, the only family outside AA which I’ve seen claimed as having a triconsonantal system. It isn’t quite the same thing, but its morphology is certainly no less weird than Semitic. Not sure I entirely understand the system myself, but the gist is that Yokuts verbal roots are CVCVC with an extreme form of vowel harmony — both vowels must have the same quality. But then each vowel has something like ~20 mutation grades, several of which result in the vowel disappearing entirely, and then various verbal affixes impose various grades on each vowel.

For conlanging purposes: the roots don't have to be triconsonantal either. (I did something like that with the Tarandim family: the roots need two syllables, but that may mean one, two, or three consonants; the patterns will apply regardless.)

Not sure I remember this aspect of Tarandim (or anything of it, really); could you give a quick overview of the system please?

Adding more vowel shift rules would work! But both PIE (with ablaut) and proto-Semitic started out with a very reduced number of vowel qualities, and that reduces the number of ways the vowel can change.
(It's hard to generalize a pattern with English strong verbs -- even then, it happens fairly often -- but if English had only three vowel qualities, ablaut would probably be extremely productive.)

Huh, that’s true as well. I’m most familiar with Modern Hebrew, which has /a e i o u/, but now that I think of it, Proto-Semitic did indeed have a reduced number of vowels. As you say, that probably helped a lot with the analogy.

[Ares Land]

Not sure I remember this aspect of Tarandim (or anything of it, really); could you give a quick overview of the system please?

There's a bit on it here: https://verduria.org/viewtopic.php?p=31396#p31396
I use the CVCVC pattern to describe it, but in fact verbs can follow any of these patterns: VCV, VCVC, CVCV. (The most common one happens to be CVCV)

(Simbri is related and uses a similar system.)

[Ahzoh]

besam-o 'I build' > be.sa'mo > besmo
besam 'He builds' > be'sam > bsam

That's the other thing. These kinda guides all involve having some grammatical form that is affix-less, usually the affix-less form indicates the third person singular masculine. But my conlang has -CV shape person suffixes for al persons

I am also thinking that my proto-lang was most likely initial/left-leaning stress becuase initial stress is seen in words like mázi-0 "sea-CNS" and mázi-0-ku "sea-CNS-3fs.POSS"

I am thinking of having a proto-paradigm like this (first person and third person feminine respectively). Also yes, this language doesn't have basic tense or aspect distinctions, but it has basic modal distinctions instead.

Code: Select all

voice-stem-mood-modifier-person-logophor-negation
0-    prux-0   -0       -na    -aʡs     -pa
0-    prux-a   -0       -na    -aʡs     -pa
0-    prux-a   -ʔa      -na    -aʡs     -pa
0-    prux-a   -ki      -na    -aʡs     -pa

0-    prux-0   -0       -ku    -aʡs     -pa
0-    prux-a   -0       -tsu   -aʡs     -pa
0-    prux-a   -ʔa      -tsu   -aʡs     -pa
0-    prux-a   -ki      -tsu   -aʡs     -pa

And somehow it is supposed to become all this:

Code: Select all

pVruḫ-na   "speak\REAL-1s"
paraḫ-na   "speak\IRR-1s"
parḫ-an-na "speak\IRR-DIR-1s"
parḫ-Vk-na "speak\IRR-INT-1s"

pVruḫ-ku   "speak\REAL-3fs.REAL"
paraḫ-tu   "speak\IRR-3fs.IRR"
parḫ-at-tu "speak\IRR-DIR-3fs.IRR"
parḫ-Vk-tu "speak\IRR-DIR-3fs.IRR"

Code: Select all

pVruḫ-nässä   "speak\REAL-1s.REL"
paraḫ-nässä   "speak\IRR-1s.REL"
parḫ-an-nässä "speak\IRR-DIR-1s.REL"
parḫ-Vk-nässä "speak\IRR-INT-1s.REL"

pVruḫ-kässu   "speak\REAL-3fs.REAL.REL"
paraḫ-tässu   "speak\IRR-3fs.IRR.REL"
parḫ-at-tässu "speak\IRR-DIR-3fs.IRR.REL"
parḫ-Vk-tässu "speak\IRR-INT-3fs.IRR.REL"

But no matter what stress I have, there is no way to elide the modal vowel that would trigger the creation of an ablaut distinction.

[bradrn]

Assuming you have iambic [EDIT: sorry, trochaic] feet assigned left-to-right, you could use a rule which deletes all odd vowels (like in Nishnaabemwin, but deleting word-final vowels as well). If you add a vowel-coalescence rule before this, and do vowel mutation along with the syncope, that would give you something like pruxnʡspa, praxnaʡsp, praxʔanʡspa, etc.

More interestingly, you could assign iambic trochaic feet right-to-left allowing primary-stressed degenerate feet at the left edge, giving you stress patterns like:

'prux.naˌaʡs.pa
ˈpruxˌa.naˌaʡs.pa
ˈprux.aˌʔa.naˌaʡs.pa

Which gives you an interesting alternation between elision and non-elision of the modal vowel if you then apply some syncope and mutation: pruxnaʡsp, pruxanaʡsp, praxʔanʡsp.

[Ahzoh]

Quite interesting results. Given my language's opposition to onset or coda consonant clusters and how the laryngeal stops cause gemination, the result would probably be:
paruḫnespa or paruḫnessep
paraḫnespa or paraḫnessep
paraḫḫanespa or paraḫḫanessep
Which aren't very pretty.

Actually I think I'll discard the negation suffix, everything it combines with is ugly and overly long. The -ess- infix in combination with the person markers already pushes the limits to how long the words should get in that direction.

[Znex]

Quite interesting results. Given my language's opposition to onset or coda consonant clusters and how the laryngeal stops cause gemination, the result would probably be:
paruḫnespa or paruḫnessep
paraḫnespa or paraḫnessep
paraḫḫanespa or paraḫḫanessep
Which aren't very pretty.

That gives me strong Akkadian/Old Aramaic vibes and looks really fun to me honestly, but do what suits you.

[Ahzoh]

Quite interesting results. Given my language's opposition to onset or coda consonant clusters and how the laryngeal stops cause gemination, the result would probably be:
paruḫnespa or paruḫnessep
paraḫnespa or paraḫnessep
paraḫḫanespa or paraḫḫanessep
Which aren't very pretty.

That gives me strong Akkadian/Old Aramaic vibes and looks really fun to me honestly, but do what suits you.

I do like me some geminates, although I try avoid having word forms where there’s like three or four geminates in a row. That feels wrong to me to have so many.

I’ve also thought of a CaCaCCa pattern, if only it worked so well with my general scheme of eliding weak consonants altogether.

On an unrelated matter, I have something of a non-high-vowel harmony (like Akkadian) where /a e/ cannot really coexist in a word. But I’m not really sure on the extent. Is it only within roots or are affixes affected?

Or maybe a harmonizing change only occurs in one of either category but not the other? That is, a laryngeal-containing affix doesn’t trigger harmony in the root but only the affixes, and a laryngeal-containing root only triggers harmony within the root and does not extend to affixes. That way I can have affixes like -nesse while allowing roots like eber to not turn the 1st person suffix -na into -ne

[JT the Ninja]

Is it normal to want to re-work a verb system literally every time you look at it? I've seriously re-worked the Ahnakáh verb conjugation system at least 2 or 3 times, but it's still not satisfactory.

I almost want to scrap all the person endings entirely. The language really doesn't make use of affixes outside of verbs, so it would make sense to have it indicated by something else. Can't just use sentence position, though, because the word order changes for declarative vs. interrogative.

...wait...maybe I do what English does a lot and have nouns/verbs alternate stress patterns? The typical stress is on the first syllable...maybe have verbs put the stress on the final syllable? And this can lead to some vowel mutation, perhaps?

...I keep making more work for myself...[]

[bradrn]

Although I don’t remember you before (you must have been active before I joined), welcome back JT the Ninja! Pickles and Tea!

Is it normal to want to re-work a verb system literally every time you look at it? I've seriously re-worked the Ahnakáh verb conjugation system at least 2 or 3 times, but it's still not satisfactory.

Only 2 or 3? I’m impressed you can keep it that low. I personally rarely overhaul the whole system at once, but I regularly do this for individual subsystems of the grammar. (Do you want to know how many times I’ve gone over the aspectual system for my previous conlang?)

I almost want to scrap all the person endings entirely. The language really doesn't make use of affixes outside of verbs, so it would make sense to have it indicated by something else. Can't just use sentence position, though, because the word order changes for declarative vs. interrogative.

Agglutination — polysynthesis, even — with no person marking is well-attested: e.g. in Shoshone, Pirahã, Maidu etc. (Awtuw is an interesting case: the verb doesn’t cross-reference person, but does mark subject number.) See https://www.tandfonline.com/doi/pdf/10. ... ccess=true for details.

...wait...maybe I do what English does a lot and have nouns/verbs alternate stress patterns? The typical stress is on the first syllable...maybe have verbs put the stress on the final syllable? And this can lead to some vowel mutation, perhaps?

This is a pretty interesting idea, actually. I’d be really interested to see if you go anywhere with it. (Or perhaps I’ll just steal borrow it for my next conlang…)

[JT the Ninja]

When I say 2 or 3, I'm counting the major overhauls, where I just deleted everything and started from scratch. The minor iterations are much more numerous.

The language itself is...I guess you'd call it isolating? It marks most things with articles and auxiliaries instead of endings. Apart from the current verbal endings, Ahnakáh only has a couple affixes:

[*]Adjectival - leved "boy", levede/levedi "boyish" (singular/plural)
[*]Adverbia l- pelen "dryness", pelenyé "dryly"
[*]Participial - dag "light", dagese "lighted"
[*]Infinitival - káh "knowledge," el-káh "to know"

Everything else usually uses an auxiliary word, be it an article or an auxiliary verb:

Nouns:
[*]lon "ring" / "a ring"
[*]i lon "the ring"
[*]e lon "rings" / "the rings"
[*]ni lon "(to the) ring" (accusative)
[*]ke lon "(for the) rings" (dative)
[*]ti lon "(of the) ring" (genitive)
[*]Er al lon ye "It is the ring" (complementive)

Verbs:
[*]el-veron "to need"
[*]Ah verona "I need"
[*]Ah ve verona "I needed"
[*]Ah náh verona "I will need"
[*]Ah fi verona "I might need (now)" (suppositive)
[*]Ah fi ve verona "I might have needed" (suppositive past)
[*]A fi náh verona "I might need" (suppositive future)

There is a pretty nice system of suffixes I put together for different aspects, though:

[*]el-dar "to run"
[*]Ah dara "I run"
[*]Ah darano "I am running"
[*]Ah darala "I run habitually"
[*]Ah darare "I have finished running"

There also forms for each of those in the subjunctive and imperative. And you can further add the auxiliary to mark past/future/suppositive.

The mixed system of auxiliary/modal verbs for some things and suffixes for others does bug me, though, as does the prefix for infinitive (which I've been thinking I might nix in the future). I've also gone back and forth on the person endings. Mostly they look just like the pronouns were tacked at the end of the verb, and I don't like that. So I could just say drop the person endings entirely (or have it reduplicate the primary vowel if you need to add a suffix) and make the pronouns mandatory instead of optional.


Or something...[]

[Qwynegold]

Interesting challenge — I think I’ll have a go. Not sure how much time I’ll have to do this soon, so let me pre-register my choices:

• A2 Fixed stress location of a type that you seldom or never use in conlangs. When affixes are added to a word, the stress moves so that it will stay on the specificly numbered syllable.
• A4 Consonant mutation
• B6 Uses prefixes way more often than suffixes
• C3 A retroflex series, and not just on fricatives/affricates

Oh, and if we have to show a sketch in this thread, then could someone split this off as a separate ‘Conlang Challenge Thread’ or something?

Hmm, maybe create a thread just for presenting your conlang (sketch)? And send me a PM with the link! Btw, which fixed stress locations do you seldom if ever use?

[bradrn]

Interesting challenge — I think I’ll have a go. Not sure how much time I’ll have to do this soon, so let me pre-register my choices:

• A2 Fixed stress location of a type that you seldom or never use in conlangs. When affixes are added to a word, the stress moves so that it will stay on the specificly numbered syllable.
• A4 Consonant mutation
• B6 Uses prefixes way more often than suffixes
• C3 A retroflex series, and not just on fricatives/affricates

Oh, and if we have to show a sketch in this thread, then could someone split this off as a separate ‘Conlang Challenge Thread’ or something?

Hmm, maybe create a thread just for presenting your conlang (sketch)? And send me a PM with the link! Btw, which fixed stress locations do you seldom if ever use?

In that case, I think I’ll just put it in my scratchpad thread. (No need to create a new thread just for one post.)

As for fixed stress, I was actually a bit unsure about that at first, but I had been reading about stress systems and figured I’d end up finding something. I eventually went for a system where stress is fixed at the morphophonemic level, but becomes unpredictable at the surface (though always on one of the first two syllables).

[Man in Space]

Laghá ĝár, ör laghá genr.
[làʕhá ŋáɹ | ø̀ɹ làʕhá ʕèndɹ̩]

laghá

CAUSE.STATE

ĝár

vouch

ör

but.S

laghá

CAUSE.STATE

genr

observe

'Trust, but verify.'

[Raholeun]

I eventually went for a system where stress is fixed at the morphophonemic level, but becomes unpredictable at the surface (though always on one of the first two syllables).

I'd be interested to see how that works.

[JT the Ninja]

I eventually went for a system where stress is fixed at the morphophonemic level, but becomes unpredictable at the surface (though always on one of the first two syllables).

I'd be interested to see how that works.

Few things are as predictably unpredictable as language. []

[Qwynegold]

I eventually went for a system where stress is fixed at the morphophonemic level, but becomes unpredictable at the surface (though always on one of the first two syllables).

I'd be interested to see how that works.

Same.

[bradrn]

Oh, OK… I don’t have the whole grammar sketch finished yet (exams are getting in the way), but that part is reasonably presentable:

I divide phonological rules into segmental and post-prosodic rules. The segmental rules apply first, followed by foot assignment and then the post-prosodic rules.

Segmental rules:

1. Retroflexion: tr sr → ʈ ʂ
2. Consonant–Prenasal Deletion: C → ∅ / _ [+nasal][+stop]
3. Palatalisation: n ŋ t k s ʂ → ɲ ɲ c c ɕ ɕ / {i,j} _
4. Intervocalic Lenition: p t s k → β ɹ z ɰ / V_V

Following the segmental rules is syllabification and foot assignment. Feet are iambic and assigned left-to-right; primary stress is given to the leftmost foot, and degenerate feet are not tolerated. Heavy syllables [those having long vowel or coda] cannot occur in the weak part of the foot, and are assigned a separate foot if this would occur. Thus e.g. |əɰəwəzəekrəaːɰampəicəβuniːɨŋ| is footed as (əˈɰə)(wəˌzə)(eˌkrə)(aː)(ɰaˌmpə)(iˌcə)(βuˌniː)ɨŋ.

Post-prosodic rules can modify this prosodic structure, and are as follows:

1. Metathesis: If a foot has structure (CV.V(C)), the first syllable undergoes metathesis becoming (VC.V(C))
2. Unstressed Vowel Deletion: If a foot begins with an unstressed vowel, that vowel is deleted, with consequent resyllabification of any following consonant. If the previous foot ends in a vowel, that vowel may mutate: ə+V ɨ+V a+e a+i i+a a+o a+u u+a u+i i+u → V V e e e o o o ɨ ɨ
3. Stop Resolution: Illegal sequences of stops are resolved by changing the first stop to a nasal at the same PoA as the second stop
4. Stress Clash Resolution: If two stressed syllables are adjacent, the second is destressed (becoming a degenerate foot in the process)
5. Stressed Vowel Lengthening: /a e i o u/ → /aː eː iː oː uː/ in stressed syllables. /ə ɨ/ do not lengthen, but mutate to /a i/.

(Note that any of the above may be subject to change. Also, the quote doesn’t consider consonant mutation, which is treated separately.)

And a worked example:

|əɰəwəsəekrəaːkampəikəpuniːɨŋ|
→[P] |əɰəwəsəekrəaːkampəicəpuniːɨŋ|
→[IL] |əɰəwəzəekrəaːɰampəicəβuniːɨŋ|
→ (əˈɰə)(wəˌzə)(eˌkrə)(aː)(ɰaˌmpə)(iˌcə)(βuˌniː)ɨŋ
→[UVD] (ˈɰə)(wəˌze)(ˌkrə)(aː)(ɰaˌmpi)(ˌcə)(βuˌniː)ɨŋ
→[SCR] (ˈɰə)(wəˌze)krə(aː)(ɰaˌmpi)cə(βuˌniː)ɨŋ
→[SVL] (ˈɰa)(wəˌzeː)krə(aː)(ɰaˌmpiː)cə(βuˌniː)ɨŋ
/ˈɰawəˌzeːkrəˌaːɰaˌmpiːcəβuˌniːɨŋ/ chawəzēkrəāchampīcəꞵunīɨŋ ‘will I not repeatedly make you see him sitting?’

Compare e.g. |kəkəwəsəekrəaːkampəikəpuniːɨŋ| /kəˈɰawəˌzeːkrəˌaːɰaˌmpiːcəβuˌniːɨŋ/, in which the initial vowel is preceded by a consonant and thus remains unsyncopated giving second-syllable stress. Note that an initial heavy syllable will also receive primary stress. (In fact, thanks to Stressed Vowel Lengthening, the stressed syllable can usually be found from the surface form, though there are cases where this is impossible.)

[Raholeun]

I am working on a new language and attempting to translate some poetry in it. Unfortunately I am getting sort of stuck on a certain construction and could use some inspiration.

In any conlang of your choosing, how would you translate and gloss:

"I let you [VERB] first"