How would this be distinguished from cu vois in normal speech (if at all)?
I like this! If you're going for more natlangy than loglangy, you'd be free to have the state be not just semantically but also phonologically null :D
Do you want to be able to attach subject-pronouns to verbs? If (that and you use an affix to nominalize) or (you use a particle to nominalize) then you could backform/reanalyze a resumptive pronoun this way! Seeing "changer of x-state" as a relative clause "someone such that RSM changes state x". I am told that AGNZ is the gloss for agentive nominalizer, so: "AGNZ cu vois" / "(AGNZ/RSM)-cu vois" / "cu-(AGNZ/RSM) vois" / "cu (AGNZ/RSM) vois", do you see what I mean?
That is beyond my planning at this point, but I think I'd count it as a point for the other order. I also considered a (potentially limited) ban on underfilling, which would mean cu would need to be supplied at least one argument (the patient), which would make it possible to use cuvois. Maybe visible from the examples: I am considering gemination as a marker for agglutination versus separate words.
Same! And I am wondering whether it couldn't be phonologically null in a loglang as well :)
What do you mean by "attach"? And by "subject"?
Relative clauses marked by a different word order is a nice option with this forced bracketing by the split verb. Maybe we do need the phonologically non-null state after all!
Being able to for example say cu ba vois as cu-ba vois or ba-cu vois (or some other affix, maybe cu-b vois or something, but you get the point).
Like generally I think that with a sufficiently advanced case system you don't really need any specific nominalizer? Just throw a resumptive pronoun in at the right location in a relative clause! An x-doer is a "one who (RSM does x)", a having-x-done-to-them is a "one who (someone does x to RSM)" etc., and you might be able to make relative clauses so easy/cheap/transparent to construct that [determiner] cu RSM vois can just be how you nominalize cu [...] vois.
EDIT: I guess I should have called it "agent" rather than subject
I'm only just starting to grasp the implications. If this is the only nominalizing option, then we probably want a shorter way to say "VERB-ho-MEANING", e.g. reducing the infix to just -h-, and also we'd probably want VERBs to be as short as possible, e.g. one syllable at most.
Yeah, basically exactly like that, except I didn't want to prescribe too many details about the grammar and/or morphology for it... Because I'd rather put the burden on you to, if you're going for a loglang, figure out how to then make sure relative clauses end properly unambiguously
I don't know if you want phonemic vowel length, but a reasonably common sound change that could happen in cases like this would be compensatory lengthening, where the -h- ends up disappearing entirely and and instead lengthening the preceeding vowel. cu nā-leu vois, naaleu, maybe even nah-leu or really whatever orthography you'd like (speaking of which, I find your usage of the long s delightful).
Yee the phono/morpho is still in very early stages, but I think long s is a given.
To be able to talk about grammar, let me introduce/reuse some placeholder words that are hopefully reasonably intuitive. I'll deal with SSM and all that jazz later.
Also, as far as I can tell, Latejami's distinction between state verbs and action verbs is a bit nebulous. A reasonable analysis might be something like
stative state: some parameter describing the patient's state is a constant function of time
dynamic state: the parameter changes from one constant value to another over a salient/contextual time interval
stative action: the rate of change of the parameter is constant over a given time (???)
dynamic action: the rate of change is not constant (Maybe a Dirac pulse? I need to think about this more.)
Since I'm still unsure about these action verbs, I will be ignoring them for now.
PAd: Agient, Patient, dynamic - agent causes patient's change of state (ex.: "informs", "opens", "repairs")
A=Ps: Agent=Patient, stative - agent maintains their own steady state (ex.: "survives", "keeps busy", "stays awake")
A=Pd: Agent=Patient, dynamic - agent changes their own state (ex.: "escapes", "learns", "lies down")
Patient, agent, and oblique
[Placeholder. I hope the pattern is clear enough by now but I may add the above cases plus oblique explicitly at a later time.]
Meaning carriers: MC (I desperately need better nomenclature...)
For now, let me only define one MC, namely NULL. It will serve as a placeholder and combining it with a VERB yields the most general possible version, e.g. PAs-NULL ≈ "patient's state is maintained by agent".
Pronouns
I'll simply use the usual 1s, 1p, 2s, 2p, 3s, 3p, (maybe 4s, 4p?), RSM for now.
Determiners
e: \exists
a: \forall
n: \not\exists
[Details on scope and negation to follow, not opening that can of worms yet.]
Examples
These will basically be glosses, so we won't need actual phono and morpho for now. Will use brackets to mark relative clauses, but just as a visual aid. I don't think they are necessary?
PAOd 2s 3s 4s NULL ≈ "Your state in relation to the-other is changed by them."
POs e[Pd RSM NULL] 3s NULL ≈ "Something, [whose state changes in some way] experiences a steady state in relation to them."
Compiling all possible VERBS with at most the roles patient (P), agent (A), oblique (O) — in both stative (s) and dynamic (d) versions — we get this basic table with 8 different combinations:
P?
A?
O?
s/d
#args
P
-
-
s
1
P
-
-
d
1
P
-
O
s
2
P
-
O
d
2
P
A
-
s
2
P
A
-
d
2
P
A
O
s
3
P
A
O
d
3
And adding onto that, VERBS where two or all three roles are identical entities, we have another 12 combinations:
P?
A?
O?
s/d
#args
P
-
P
s
1
P
-
P
d
1
P
P
-
s
1
P
P
-
d
1
P
P
P
s
1
P
P
P
d
1
P
A
P
s
2
P
A
P
d
2
P
P
O
s
2
P
P
O
d
2
P
A
A
s
2
P
A
A
d
2
If this is the extent of our possible VERBS, that'd mean that 20 would be all it takes.
I am aware that this doesn't cover 4-ary verbs, which would probably be dealt with differently.
Are there any other important verbal concepts that these don't cover?
Would it be nice to split up knowledge of facts and knowledge about topics?
If so, most of the shared-role declinations become somewhat meaningless, because an entity can't fact-know itself.
Maybe the shared-role declinations are special cases that shouldn't get their own VERBS?
On the other hand, even if just as syntactic sugar, they seem practical.
Do the Ps and Pd versions even make sense? Can I be in a state of knowledge without an oblique thing that I know?
This should probably be just a grammar thing, not a metaphysical one.
Maybe this is simply how the grammar deals with underfilling? Is there a qualitative difference between the different declinations or is it just a matter of what roles I explicitly mention?
Is that along the lines of what you meant?