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

[Omzinesý]

Thank you both

If I understood corretly, this is the usual way of gaining glottalized resonants
nt -> nʔ -> n̰
Nearly all languages have resonant + stop clusters.

Salmoneus mentioned phonations like creaky voice. I think it could develop from a tone and tone can develop from stress.
We once discussed a German dialect that developed a creaky voice from word-final r.

"I did just want to add one thing that was implicit in my earlier post: phonation contrasts and related contrasts (like glottalisation) can develop from stress and word position. You could, for instance, have initial or final consonants regularly become glottalised (though this could also be accomplished through epenthetic initial or final glottal stops, as in germanic languages, which could then yield glottalisation through feature spreading / fusion)."
That is a good point too. Glottalization can be an emphatic (in the non-Semitic sense) feature in the beginning.


I was actually playing with an idea about a PIE lang with glottalized resonants. Glottalization there could well be caused by "laryngeals". Some theories also pose that the "plain voiced plosives" were actually glottalized, implosives maybe.

[Ahzoh]

Not sure what I want to do about superheavy syllables ([C]VVC). I do not really like having them but their existence is necessary to maintain a distinction between singular nouns and plural nouns (lumbum "person" vs. lumbūm "people") but I don't like having words like dāltin "they reveal" or tīlnēze "I who destroyed" where there is a superheavy syllable preceding a heavy syllable.

A solution I have is to add an epenthetic echo vowel after the offending syllables, but I do not like having such after my case endings. I also do not mind a superheavy syllable being followed by a light syllable as much as I hate superheavies being followed by heavies.

Dealing with superheavies is also important because odd-numbered syllable stems take geminated forms of person suffixes when they end in short syllables.
nāta-s "night-SG-NOM" (even-numbered stem) nāta-na "night-SG-1cs" and nātā-na "night-PL-1cs"
sammali-m (odd-numbered stem) "crocodile-SG-NOM" semmele-nna "crocodile-SG-1cs" but semmelē-na "crocodile-PL-1cs"

The old way:
natâlna "I shall destroy" natâlna-kkun "I shall destroy them" but natâlnēze "I who shall destroy" natâlnēze-kun "I who shall destroy them"
New way:
natâlana "I shall destroy" natâlana-kun "I shall destroy them" but natâlanēze "I who shall destroy" natâlanēze-kkun "I who shall destroy them"

So I'm just not sure if it's naturalistic to only have superheavy syllables word-finally

[Creyeditor]

Not sure what I want to do about superheavy syllables ([C]VVC). [...]
So I'm just not sure if it's naturalistic to only have superheavy syllables word-finally

Definitely naturalistic. But you can also have fun with prosodically conditioned echo vowel insertion or degemination. I guess it's up to you

[Ahzoh]

I wonder if it would be naturalistic if relativized verbs (verbs wherein the subject suffix contains the -ēz- infix that creates relative clauses) were not able to take object suffixes.

would allow for clean prosody (— = stressed, ∪ = unstressed):
tīl- "perish":
tīlna "I perished" (— ∪)
tīlinakku "I destroyed her" (— ∪ — ∪)
tīlinēze "I who perished" (— ∪ — ∪)

Doesn't work so well with relativized and object:
tīlinēzeku "I who destroyed her" (— ∪ — ∪ ∪) or tilnēzekku (∪ ∪ — ∪)

Though the tense-mood prefixes would turn it all into iambic:
natīlna "I shall perish" (∪ — ∪)
natīlinakku "I shall destroy her" (∪ — ∪ — ∪)
natīlinēze "I who shall perish" (∪ — ∪ — ∪)

And then I noticed that most plurals of nouns are also iambic in nature while singulars of nouns are trochaic.

[eldin raigmore]

I wonder if it would be naturalistic if relativized verbs (verbs wherein the subject suffix contains the -ēz- infix that creates relative clauses) were not able to take object suffixes

As long as they can take objects, I don’t see why it would matter that they can’t take object-suffixes.

[Creyeditor]

I wonder if it would be naturalistic if relativized verbs (verbs wherein the subject suffix contains the -ēz- infix that creates relative clauses) were not able to take object suffixes

Sounds like an original pattern of differential object marking (or differential object indexing)

[DesEsseintes]

I wonder if it would be naturalistic if relativized verbs (verbs wherein the subject suffix contains the -ēz- infix that creates relative clauses) were not able to take object suffixes.

Yes, this would be perfectly realistic. A linguist whose name escapes me has proposed a scale of “verb-likeness” where finite verbs are at one end of the scale and have the most verblike properties such as being able to be qualified by adverbs and take direct and indirect objects (and other arguments). Nouns are at the other end, as in most languages they are unable to do verblike things (and are therefore not classified as verbs).

In-between you can have all kinds of gradients. Nominalised verbs for instance can retain some verblike characteristics in languages like Thai which can incorporate adverbs and direct objects. However, they tend to lose at least some of their verblike characteristics. Your relativised verbs would be a perfect example of such a gradient on the scale. Participles, if your language has them, could be another gradient, with even less verblike characteristics than the relativised verbs you speak of.

[Ahzoh]

Interesting, They might fall somewhere halfway between a full-verb and a participle or gerund. They can take objects but unlike a main clause verb this is not mandatory. They cannot be used like an adverb or be substantivized, though they agree with (or indicate) the subject.

I call them relativized verbs (or verbs with relative pronoun suffixes) but prototypically they simply indicate an anaphoric reference and thus could be used to refer to the same subject over the entire discourse, though that is slowly dying off and they are often only used to indicate the same subject in a relative clause.

[Ahzoh]

I have two phonemes /z/ and /ɮ/, for /z/ it possesses two phones in complimentary distribution [z] and [r]. Likewise /ɮ/ has [ɮ] and [l]. However, they are phonemically distinguished when geminated, thus /zː/ vs. /rː/ and /ɮː/ vs. /lː/.

I need ideas for how a language arrives at such a situation and also what environments to find the phones of these phonemes. I know that [z] and [ɮ] can occur word-initially as word-initial fortition is common, likewise [r] and [l] can occur intervocalically as well. Other than that, I'm not sure what can occur where in other environments.

---

On the other hand, is there anything in the phonetic/articulatory nature of /ɮ/ that makes it unstable and more likely to become /l/? For comparison /ɢ/ is unstable due to uvulars causing retracted tongue root and voicing causing advanced tongue root as well as the interacting nature between voicing and air travel time in the mouth.

[Salmoneus]

I have two phonemes /z/ and /ɮ/, for /z/ it possesses two phones in complimentary distribution [z] and [r]. Likewise /ɮ/ has [ɮ] and [l]. However, they are phonemically distinguished when geminated, thus /zː/ vs. /rː/ and /ɮː/ vs. /lː/.

I need ideas for how a language arrives at such a situation and also what environments to find the phones of these phonemes.

One of two things is likely to have happened: either two phonemes have merged when not geminated; or, one phoneme has split into two when geminated. The former is too simple to have to explain; the latter can happen when gemination (or more generally fortition) can be applied to either allophone and the conditioning environment is then lost.

A third possiblity is that one of the geminated phonemes has nothing to do with the allophone of the other phoneme, and simply resembles it coincidentally.

In terms of conditioning environment for allophonic variation, presumably it's just the usual. Intervocalic, prevocalic, postvocalic, initial, final. Stressed, unstressed. Secondary articulations, quality of adjacent vowel (eg phonation, height, roundedness), presence of semivowel. Etc. I guess my immediate thought would be /r/ > [z] before /i/, /j/, which would be allophonic in most instances but would become phonemic if the vowel could sometimes be lost resulting in a geminate: /arira/ > /azza/, /arura/ > /arra/. But obviously there are countless alternatives.

On the other hand, is there anything in the phonetic/articulatory nature of /ɮ/ that makes it unstable and more likely to become /l/?

The former is a fricative requiring central occlusion, which is obviously going to be unstable as it requires fine control of the tongue edges independently of the centre and most people do not have super-agile tongue edges. Articulatorily it's going to be likely to either develop into a plain stop (moving the edges up a little) or a plain approximant (moving the edges down a little) - the fricative requires accurately positioning the edges within a narrow window of frication.
Acoustically, meanwhile, it doesn't sound that different from other fricatives while being harder to pronounce, so it's going to want to merge with (or simplify to) /Z/ or the like. [I mean, the IPA symbol is literally just an /l/ and a /Z/ stuck together...]

[Ahzoh]

The former is a fricative requiring central occlusion, which is obviously going to be unstable as it requires fine control of the tongue edges independently of the centre and most people do not have super-agile tongue edges.

Strange, this sounds like it could apply to voiceless latfrics as well, which are comparatively more stable.

Being harder to pronounce, so it's going to want to merge with (or simplify to) /Z/ or the like. [I mean, the IPA symbol is literally just an /l/ and a /Z/ stuck together...]

I merged /ʃ ʒ/ with /s z/ with the result being only /ɬ ɮ/.

I thought many times to only have /s z ʃ ʒ/ but my language is inspired by Mesopotamian which have those (at least /s ʃ/) out the wazoo, so I figured it would be unique and interesting to have /s z ɬ ɮ/ as the only coronal fricatives.

[Salmoneus]

In terms of conditioning environment for allophonic variation, presumably it's just the usual. Intervocalic, prevocalic, postvocalic, initial, final. Stressed, unstressed. Secondary articulations, quality of adjacent vowel (eg phonation, height, roundedness), presence of semivowel. Etc.

The former is a fricative requiring central occlusion, which is obviously going to be unstable as it requires fine control of the tongue edges independently of the centre and most people do not have super-agile tongue edges.

Strange, this sounds like it could apply to voiceless latfrics as well, which are comparatively more stable.

Everything is comparatively more stable than something else. Voiceless lateral fricatives are still very rare sounds. According to UPSID, ~90% of languages don't have them. [and from an eyeball-scan of the ones that do, it seems that that 10% is probably greatly exaggerated by their areal presence in North America]. That's certainly more than the 3% that have the voiced version, but it's vastly less than the 78% that have a lateral approximant.

So yes, there are also reasons why the voiceless lateral fricative is rare.

As for why the voiceless may be more common than the voiced lateral fricative, I'd suggest:
- the voiceless version is far more distinct from /l/, and hence far less likely to merge with it. Indeed, I suspect that there may be devoicing of the voiced fricative in some languages in order to maximise distinctness.
- languages in general are far, far more likely to lack voicing distinctions in ANY fricatives, with the single fricative series then conventionally notated as voiceless (even though allophonic voicing will be common)
- voiced fricatives may not be as acoustically distinct from one another as their voiceless counterparts, so that voiced fricatives are more likely to merge with similar fricatives than voiceless ones are. I don't have any statistics to back that up, but it makes sense to me acoustically and I think I've seen signs of that in language inventories (eg merging /z/ and /Z/ while keeping /S/ distinct from /s/).

Being harder to pronounce, so it's going to want to merge with (or simplify to) /Z/ or the like. [I mean, the IPA symbol is literally just an /l/ and a /Z/ stuck together...]

I merged /ʃ ʒ/ with /s z/ with the result being only /ɬ ɮ/.

I thought many times to only have /s z ʃ ʒ/ but my language is inspired by Mesopotamian which have those (at least /s ʃ/) out the wazoo, so I figured it would be unique and interesting to have /s z ɬ ɮ/ as the only coronal fricatives.

Did I say you couldn't do that? You specifically asked why it might be more unstable; don't come back to me with 'but I don't like it being unstable!'

Empirically, this is a very rare sound, and I've tried to suggest reasons why that might be. Either accept that your conlang will have a rare sound in it, or remove the rare sound from it.

[Ahzoh]

Did I say you couldn't do that?

I didn't say or imply that you did. You misunderstand, I only said what I said as more of a "funny that you mention that" tangent which also lead to a tangent about why I chose to have latfric over postalveolars.

In terms of conditioning environment for allophonic variation, presumably it's just the usual. Intervocalic, prevocalic, postvocalic, initial, final. Stressed, unstressed. Secondary articulations, quality of adjacent vowel (eg phonation, height, roundedness), presence of semivowel. Etc.

I was more wondering about say, whether /z/ is likely to be [r] before consonants or if it's likely to become [r] after consonants. I also wonder if it could just be as simple as "[z] in syllable onset, [r] in syllable coda"

Well, I'd prolly say the weaker version of /z/ is [ɹ] (or a flap) and not [r] which doesn't feel like a weaker version of /z/ to me

[Salmoneus]

I was more wondering about say, whether /z/ is likely to be [r] before consonants or if it's likely to become [r] after consonants. I also wonder if it could just be as simple as "[z] in syllable onset, [r] in syllable coda"

I covered that: "Intervocalic, prevocalic, postvocalic, initial, final". Yes, syllable position (even if I didn't explicitly say coda/medial/onset) can be significant.

Is it likely? No, of course not. But it's plausible, obviously.

[why would these things happen? Well, syllable structure and sonority hierarchy encourage sequences like CrV and VrC, and discourage sequences like CzV and VzC. There's also the question of how consonant-adjacent /z/ would cope with a rule requiring voicing assimilation (i.e. it might shift to something else to avoid the assimilation rule). But, on the other hand, feature assimilation would encourage CzV and VzC (partially anticipating or prolonging closure).]

Well, I'd prolly say the weaker version of /z/ is [ɹ] (or a flap) and not [r] which doesn't feel like a weaker version of /z/ to me

Famously, "rhotics" can freely vary between wildly different phones between dialects (including over time). While a direct /z/ > /r/ doesn't seem like "weakening", no, a weakened rhotic approximant could then easily be shifted to /r/ (or any other rhotic) over time. [or, vice versa, a trill could weaken to an approximant or tap, which could then strengthen to a voiced fricative]

[Nortaneous]

On the other hand, is there anything in the phonetic/articulatory nature of /ɮ/ that makes it unstable and more likely to become /l/?

I don't think so. I'm sure there are examples of ɮ losing frication, but I don't know of any - offhand the only developments I know of are ɮ > ð as an incipient shift in Khalkha Mongolian and ɮ > ld (sometimes with ɬ > lt) in some Sino-Tibetan languages.

I could believe that ɮ is less stable than l for the articulatory reasons Salmoneus mentioned, but this doesn't apply to ɬ because ɬ l̥ never contrast - that is, frication isn't an essential component of sounds that we would write as ɬ, but is for sounds we'd write as ɮ.

[Sequor]

Mongolian also seems to have ɮ > ɬ too, does it not? Judging from my ears, not descriptions.

[Creyeditor]

I have two phonemes /z/ and /ɮ/, for /z/ it possesses two phones in complimentary distribution [z] and [r]. Likewise /ɮ/ has [ɮ] and [l]. However, they are phonemically distinguished when geminated, thus /zː/ vs. /rː/ and /ɮː/ vs. /lː/.

I need ideas for how a language arrives at such a situation and also what environments to find the phones of these phonemes. I know that [z] and [ɮ] can occur word-initially as word-initial fortition is common, likewise [r] and [l] can occur intervocalically as well. Other than that, I'm not sure what can occur where in other environments.

---

On the other hand, is there anything in the phonetic/articulatory nature of /ɮ/ that makes it unstable and more likely to become /l/? For comparison /ɢ/ is unstable due to uvulars causing retracted tongue root and voicing causing advanced tongue root as well as the interacting nature between voicing and air travel time in the mouth.

Sorry, if I am a bit late to the party and sorry if I missed anything in the discussion. I hope I can still offer a new perspective. IIUC, you have two relevant phonemes (/z/ = [z]/[r] and /ɮ/= [ɮ]/[l]) for short consonants and four relevant phonemes /zː/ vs. /rː/ and /ɮː/ vs. /lː/ for long consonants. This is not unusual at all in natlangs. There seems to be a phenomenon where geminate or long segments or suprasegments are immune to neutralizing [synchronic/diachronic] phonological changes. So you could have word-initial fortition (r > z) and intervocalic lenition (z > r) apply to singleton instances of these consonants only. Similar for the lateral (l > ɮ/word-initial and ɮ>l/intervocallically). These can probably be extended to other strong positions (e.g. simple onsets (of stressed syllables) and weak positions (complex codas, complex onsets, onsets of unstressed syllables, codas). Word-final positions would probably prefer the sonorant allophones instead of the voiced fricatives, since voiced obstruents are less stable word-finally.

I can also tell you the exact articulatory reasons why [ɮ] is less stable (although many of it was already mentioned by Sal): It is a fricative, it is a voiced fricative, it is a lateral fricative.

• Fricatives are unstable, because you have to fine tune the aperture/narrowing of your articulators, in this case the sides of your tongue. It is easy to close them (non-laterals/stops) and it is easy to open them in a relaxed way (lateral approximants). If you want to do anything in between you have to carefully adjust it.
• Voiced fricatives are less stable than voiceless fricative and voiced sonorants (such as lateral approximants) because maintaining voicing is difficult in obstruents. In sonorants air can escape your oral/nasal cavity almost without any restrictions. This allows for so called spontaneous voicing. With obstruents, there are notable restrictions (in the case of lateral fricatives the centre of your mouth is closed and the sides of your tongue form a narrow constriction). This makes it easier to produce voiceless sounds.
• Lateral fricatives are less stable than central fricatives because it is more difficult to form a constriction with the sides of your tongue than with the tip of your tongue. The tip of your tongue is much more flexible.

[Nel Fie]

I've had an idea for an agreement system, but I'm not sure kind of category it would even fit in, if it's plausible at all. Here's how it works:

1) With intransitive verbs it works like a volitional active-stative system:
- The argument of the verb is marked as an agent if it's an intentional action,
- Or as a patient if it's not intentional, or the verb describes something that is not identified as an action (semantically or grammatically.)

E.g. if we use suffixes on nouns for marking, the phrase "John walks" would be rendered as 'John.A walk', and the phrase "John trips" would be 'John.P trip'

2) In transitive situations, the same marking is used - but only for one of the arguments. Whether it is the agent or the patient that is marked depends on other grammar design choices. For example, the choice could imply focus, or determinacy.
The "leftover" argument would receive a third agreement marking, which is the same regardless of the role of the argument.

E.g. using the "fits-all" oblique case for our third case, and in imitation of English active and passive constructions, the phrase "John walks the dog" would be rendered as 'John.A walk dog.OBL', whereas "The dog is walked by John" would be 'dog.P walk John.OBL'.

So, what would this be, if it's anything? It doesn't sound crazy to me, but it doesn't fit any descriptions of alignments I know about (though I only have short, theoretical definitions in my head, so that's probably the weak point.)

[Omzinesý]

I've had an idea for an agreement system, but I'm not sure kind of category it would even fit in, if it's plausible at all. Here's how it works:

1) With intransitive verbs it works like a volitional active-stative system:
- The argument of the verb is marked as an agent if it's an intentional action,
- Or as a patient if it's not intentional, or the verb describes something that is not identified as an action (semantically or grammatically.)

E.g. if we use suffixes on nouns for marking, the phrase "John walks" would be rendered as 'John.A walk', and the phrase "John trips" would be 'John.P trip'

2) In transitive situations, the same marking is used - but only for one of the arguments. Whether it is the agent or the patient that is marked depends on other grammar design choices. For example, the choice could imply focus, or determinacy.
The "leftover" argument would receive a third agreement marking, which is the same regardless of the role of the argument.

E.g. using the "fits-all" oblique case for our third case, and in imitation of English active and passive constructions, the phrase "John walks the dog" would be rendered as 'John.A walk dog.OBL', whereas "The dog is walked by John" would be 'dog.P walk John.OBL'.

So, what would this be, if it's anything? It doesn't sound crazy to me, but it doesn't fit any descriptions of alignments I know about (though I only have short, theoretical definitions in my head, so that's probably the weak point.)

Are you speaking about verb agreement or cases of nouns?

[Nel Fie]

I'm talking about verb agreement (or perhaps more accurately, morphosyntactic alignment?). Sorry if I confused that in the description.
I used cases on nouns for the example, but theoretically it could just as well be done via marking on the verb, word order, adpositions, etc... or whatever else languages can use to describe relationships between verb arguments.