Zompist Bboard Again

Ahzoh wrote: ↑Sat Apr 13, 2024 3:58 pm

bradrn wrote: ↑Sat Apr 13, 2024 3:46 pm I intended the comitative reading there, since that’s the applicative function which tends to be closest to causatives (again cf. sociatives).

That can't be correct.That doesn't make sense. The instrumental applicative should be what is closest to a causative since an instrument is semantically closer to a causee (the thing being caused to do something). And a causative is defined by a causer affecting a causee that results in the causee doing something.

No, there are many instances where they are very close, e.g.:

The nurse walked him / The nurse walked with him
Mother is playing with the child / Mother makes the child play

This kind of thing is essentially the ‘sociative’ category introduced by Shibatani & Pardeshi.

This is also why the papers people write on this topic talk about how languages with instrumental applicatives have those applicatives behave identically to, and be conflated with, causatives.

Which papers would those be, then?

My language does have split ergativity between animate nouns and inanimate nouns, so I suppose whether it's causative or applicative depends on the primary object's animacy.

I don’t really see why those two would have anything to do with each other.

The instrument semantic role would be assigned to the less animate thing while the causee semantic role would be assigned to the more animate thing.

That’s not what you said, though. Just because the alignment of the subject and object is split on animacy, that doesn’t necessarily mean the indirect object behaves the same way.

bradrn wrote: ↑Sat Apr 13, 2024 4:27 pm No, there are many instances where they are very close, e.g.:

The nurse walked him / The nurse walked with him
Mother is playing with the child / Mother makes the child play

This kind of thing is essentially the ‘sociative’ category introduced by Shibatani & Pardeshi.

The first example shows complete identicality, but the second example is ambiguous where "mother makes the child play" is not necessarily identical to "mother is playing with the child". Maybe the mother isn't playing with the child but just makes the child go outside and play with friends or something. And that is my point when I said causatives don't inherently imply co-action and co-actors. They certainly can, but they don't automatically do. And that is what makes comitative applicatives distinct from causatives: the obligatoriness of entailing co-actors.

And those examples are yet more distinct from:
I write a letter with the pen / I make the pen write a letter
I make Billy write the letter / I write the letter through Billy

Which papers would those be, then?

There is this one in particular: https://www.researchgate.net/publicatio ... syncretism

That’s not what you said, though. Just because the alignment of the subject and object is split on animacy, that doesn’t necessarily mean the indirect object behaves the same way.

The causee and the instrument both go into the primary/direct object slot in this case. The entire thing about causatives and instrumental applicatives is about how said primary/direct object is being made to perform an action itself and so animacy should be relevant.

Analysis of the secundative is weird. I hear one interpretation that secundative alignment simply makes no distinction between direct object and indirect object, only between primary object and secondary object, whereas I see another interpretation that says the primary is the direct object while the secondary is the indirect object and so the alignment has the arguments essentially "swap" places compared to the typical indirective/dative alignment.

Ahzoh wrote: ↑Sat Apr 13, 2024 4:48 pm

bradrn wrote: ↑Sat Apr 13, 2024 4:27 pm No, there are many instances where they are very close, e.g.:

The nurse walked him / The nurse walked with him
Mother is playing with the child / Mother makes the child play

This kind of thing is essentially the ‘sociative’ category introduced by Shibatani & Pardeshi.

The first example shows complete identicality, but the second example is ambiguous where "mother makes the child play" is not necessarily identical to "mother is playing with the child".

(Note that the case where they’re identical is precisely the sociative category.)

The point I’m making is that, contrary to what you said about the instrumental applicative being closest to a causative, there is another continuum which runs from the comitative through the sociative to the causative. This is all in that paper I linked earlier.

Which papers would those be, then?

There is this one in particular: https://www.researchgate.net/publicatio ... syncretism

Thanks! This is much more interesting than what I’ve been reading, and seems much more similar to what you have. (Though it seems that there are still some verbs where only the one interpretation or the other is valid.)

Incidentally, I see that they also identify this ‘in-between’ category as a sociative (citing Shibatani & Pardeshi as their source). It would seem that the sociative is just the point where the causative/comitative/instrumental/benefactive distinction completely disappears, and hence the key intermediate point for this kind of ‘universal transitiviser’.

All in all, this is extremely interesting. I may well steal it for my own conlang, which already has just the right diachronic prerequisites for this development. (Namely, a development of an instrumental SVC into a direct object marker, combined with widespread causativisation.)

That’s not what you said, though. Just because the alignment of the subject and object is split on animacy, that doesn’t necessarily mean the indirect object behaves the same way.

The causee and the instrument both go into the primary/direct object slot in this case. The entire thing about causatives and instrumental applicatives is about how said primary/direct object is being made to perform an action itself and so animacy should be relevant.

Well… remember that the causee is affected, as much as it affects something else. So we should already expect its agency to diverge from what we might ‘expect’ given its animacy. Besides, if the sociative is where ‘causee’ and ‘instrument’ become the same thing, then animacy isn’t necessarily the clue to semantic role that we’d like it to be.

Analysis of the secundative is weird. I hear one interpretation that secundative alignment simply makes no distinction between direct object and indirect object, only between primary object and secondary object, whereas I see another interpretation that says the primary is the direct object while the secondary is the indirect object and so the alignment has the arguments essentially "swap" places compared to the typical indirective/dative alignment.

Personally, I just see it as a matter of swapping the case-marking. Is there really much more to it than that?

jal wrote: ↑Sat Mar 30, 2024 4:23 pm Fed some conlangs to ChatGPT and asked it to analyse them. It was eerily correct.

I fed it a sample of something recent (a translation of The King and the God)... it did guess it was a conlang without me even asking, but its analysis of the grammar was quite wrong. I'm not sure if it misunderstood the glossing format, although it wasn't complicated.

I'm not sure if the conlang guess was driven by the choice of text, or something about the input data itself.

bradrn wrote: ↑Sat Apr 13, 2024 7:01 pm Personally, I just see it as a matter of swapping the case-marking. Is there really much more to it than that?

Yes there's a lot more to indirective alignment and secundative alignment than case-marking, just like there's often more to ergativity than simply case-marking (though it sometimes really is just that). Such as which argument gets promoted to subject in passive constructions, for one.

Ahzoh wrote: ↑Sun Apr 14, 2024 4:27 pm

bradrn wrote: ↑Sat Apr 13, 2024 7:01 pm Personally, I just see it as a matter of swapping the case-marking. Is there really much more to it than that?

Yes there's a lot more to indirective alignment and secundative alignment than case-marking, just like there's often more to ergativity than simply case-marking (though it sometimes really is just that). Such as which argument gets promoted to subject in passive constructions, for one.

Well… the thing about ergativity is that languages with true ergativity show its effects throughout the grammar. Things like preferring antipassives over passives, changing which arguments get relativised or coordinated, etc. Without those, it really is just a matter of swapping the case-marking: it’s interesting to investigate which arguments get which cases, but not much more than that.

By contrast, I don’t see the indirective vs secundative distinction having such notable effects on anything else in the grammar. It affects stuff immediately related to case-marking (e.g. passivisation), but as far as I’m aware, it doesn’t change much more than that.

Been using a word generator to generate roots for a proto-language and its family (mostly nouns). I am greatly satisfied with it satisfying my constraints. Idecided to use a word generator becuase I have a problem of overutulizing certain phonemes and under-utilizing others, which is not conducive to make a naturalistic conlang. Mostly it has to do with /f/ and /ŋ/ which are far less frequent than they should be.

The engine allows you to decide the percentage of frequency of syllable types and phonemes, so now I have new problem of deciding how frequent to make the phonemes.

My first approach is to base on the cross-linguistic commonness of phonemes, after all if certain phonemes are more common in language's inventories then maybe that translates to being more common in words/corpus? But maybe i.e. /t/ being the most common phoneme on earth doesn't mean that it's one of the top 5 most used in words in a particular language. And certain sound changes could make a phoneme more weighted to be more frequent

Idk, maybe there's a scientific way to approach this? Maybe some Zipf law insight?

This suggests phoneme frequency follow a Yule distribution, whatever that is:
https://www.reddit.com/r/conlangs/comme ... equencies/
The word gen I use (https://lingweenie.org/conlang/lexifer-app.html) seems to follow a slightly different variation of this distribution (Gusein-Zade)

I have decided to formulate a series of questions whose answers I think will give me insight into what I shall do:

1) Would sonorants/resonants be more frequent than obstruents?
2) Would coronals be more frequent than velars?
3) Would coronals be more frequent than labials?
4) Would labials be more frequent than velars?
5) Would lenis/voiced obstruents be more frequent than fortis/voiceless obstruents?
6) Would fortis/voiceless obstruents be more frequent than ejective/emphatic obstruents?

I think then I can determine a relative-frequency hierarchy I can adjust as I see fit.

Ahzoh wrote: ↑Thu Apr 18, 2024 7:09 pm I have decided to formulate a series of questions whose answers I think will give me insight into what I shall do:

1) Would sonorants/resonants be more frequent than obstruents?
2) Would coronals be more frequent than velars?
3) Would coronals be more frequent than labials?
4) Would labials be more frequent than velars?
5) Would lenis/voiced obstruents be more frequent than fortis/voiceless obstruents?
6) Would fortis/voiceless obstruents be more frequent than ejective/emphatic obstruents?

I think then I can determine a relative-frequency hierarchy I can adjust as I see fit.

1) Depends
2) Depends
3) Depends
...

A given language's consonant distribution is fairly random and has a lot to do with historical phonology. There's probably a vague tendency for less marked sounds to be more common, but counterexamples are very easy to find - in Tuu languages complex clicks are very common, while in Wichita the very unmarked /m/ is found in only two roots in the whole language.

Also word frequency has a lot to do with phoneme frequency - /ð/ is common in English text because of like three words, but pretty rare in roots.

Darren wrote: ↑Fri Apr 19, 2024 2:56 am 1) Depends
2) Depends
3) Depends
...

A given language's consonant distribution is fairly random and has a lot to do with historical phonology. There's probably a vague tendency for less marked sounds to be more common, but counterexamples are very easy to find - in Tuu languages complex clicks are very common, while in Wichita the very unmarked /m/ is found in only two roots in the whole language.

I thought hard and decided upon a series of relative order of phonemes, such as:

1) l > r > j=w
2) n > m > ŋ
3) t > k > tɬ=tʃ > p
4) kʼ > tʼ > tɬʼ=tʃʼ > pʼ
5) b > d > dɮ=dʒ > g
6) x > s=ɬ=ʃ > f
7) ʔ=h > ħ > ʡ

Ultimately that lead to:
l > t > n > k > m > b > r > kʼ > x > s=ɬ=ʃ > ʔ=h > j=w > ħ > tʼ > tɬ=tʃ > d > f > ŋ > dɮ=dʒ > g > p > tɬʼ=tʃʼ > pʼ > ʡ
Which I'm pretty happy with and think this is a naturalistic order of phonemes by relative frequency (using the Gusein-Zade distribution)

Also word frequency has a lot to do with phoneme frequency - /ð/ is common in English text because of like three words, but pretty rare in roots.

I'm more concerned over phoneme frequency in words than phoneme frequency in corpuses, for the purpose of wordgens.

---

The frequency of phonemes in Vrkhazhian will be even more skewed because of the deaffrication of /tɬ tɬ dɮ tʃ tʃʼ dʒ/ into /t tʼ d/ (when geminate) or /ɬ (ɬʼ) ɮ ʃ (ʃʼ) ʒ/ (when singleton) and subsequent mergers of the postalveolars with the central alveolars.

I think phoneme frequency is an underexamined aspect of conlanging, so I support this endeaver. Here's a fascinating paper about phoneme frequency in context in Korean.

With the new series of Doctor Who just a few weeks away, I had a thought - has anyone ever tried to make a Time Lord/Gallifreyan conlang? I don't think there's ever even been a nonsense Gallifreyan used in the programme, never mind a conlang. From memory, Gallifreyan names seem to be either nonsense or re-purposed Earthly names like Avalon(ia), Kasterbouros (the -borous sounding a lot like Latin boreus), Arcadia, and possibly some others. Google turned up nothing except a fan wiki of Gallifreyan as presented in the show at different times, but nothing of an actual conlang; nearly all the other results are for the "circular" writing system which dates from the 2005 revival.

Jonlang wrote: ↑Sat Apr 20, 2024 3:47 am re-purposed Earthly names like Avalon(ia), Kasterbouros (the -borous sounding a lot like Latin boreus),

I know nothing of Doctor Who, but that last name reminds me of Greek ouroboros more than anything Latinate.

Jonlang wrote: ↑Sat Apr 20, 2024 3:47 am With the new series of Doctor Who just a few weeks away, I had a thought - has anyone ever tried to make a Time Lord/Gallifreyan conlang? I don't think there's ever even been a nonsense Gallifreyan used in the programme, never mind a conlang. From memory, Gallifreyan names seem to be either nonsense or re-purposed Earthly names like Avalon(ia), Kasterbouros (the -borous sounding a lot like Latin boreus), Arcadia, and possibly some others. Google turned up nothing except a fan wiki of Gallifreyan as presented in the show at different times, but nothing of an actual conlang; nearly all the other results are for the "circular" writing system which dates from the 2005 revival.

I think Kiwikami gave it a try over at the CBB.

bradrn wrote: ↑Sat Apr 20, 2024 4:09 amI know nothing of Doctor Who, but that last name reminds me of Greek ouroboros more than anything Latinate.

Seconded.


JAL

Ahzoh wrote: ↑Fri Apr 19, 2024 3:18 am The frequency of phonemes in Vrkhazhian will be even more skewed because of the deaffrication of /tɬ tɬ dɮ tʃ tʃʼ dʒ/ into /t tʼ d/ (when geminate) or /ɬ (ɬʼ) ɮ ʃ (ʃʼ) ʒ/ (when singleton) and subsequent mergers of the postalveolars with the central alveolars.

I think you may need to evolve your words over a few thousand years of sound changes.

You may need to make some of your distributions dependent on the position in the word. While n > m > ŋ may be good for initial position, in a Tai language I noticed that /ŋ/ was the commonest final nasal.

Richard W wrote: ↑Sat Apr 20, 2024 12:03 pm

Ahzoh wrote: ↑Fri Apr 19, 2024 3:18 am The frequency of phonemes in Vrkhazhian will be even more skewed because of the deaffrication of /tɬ tɬ dɮ tʃ tʃʼ dʒ/ into /t tʼ d/ (when geminate) or /ɬ (ɬʼ) ɮ ʃ (ʃʼ) ʒ/ (when singleton) and subsequent mergers of the postalveolars with the central alveolars.

I think you may need to evolve your words over a few thousand years of sound changes.

You may need to make some of your distributions dependent on the position in the word. While n > m > ŋ may be good for initial position, in a Tai language I noticed that /ŋ/ was the commonest final nasal.

I can't think of thousands of years of sound changes. I also have to start from somewhere, so some things are lost to the vagaries of time.

Ahzoh wrote: ↑Sat Apr 20, 2024 12:24 pmI can't think of thousands of years of sound changes. I also have to start from somewhere, so some things are lost to the vagaries of time.

Also, thousands of years are enough to change about any sound into another. Given levelling of irragularities over time, I think you can get away fine with a regular proto-language, and develop irregularities from there.


JAL

Anyone ever wondered (in conlanging terms) what English might have looked like if the Norman conquest had exerted more influence on English phonology and grammar but less on vocabulary?

Otto Kretschmer wrote: ↑Wed May 01, 2024 7:15 am Anyone ever wondered (in conlanging terms) what English might have looked like if the Norman conquest had exerted more influence on English phonology and grammar but less on vocabulary?

No, because that's not how language contact works. Vocabulary is in most situations the easiest thing to change.

(Also, it did exert plenty of influence on phonology: that's how the voiced fricative series became phonemic, IIRC.)