So I'm trying to make a verb system that has the following goals:

1) it should, like the rest of look very Georgian; it should nail the Georgian aesthetic, and

2) it should replicate the chaos of Georgian verbs by

2a) distinguishing present, future, aorist (perfective past), imperfect (imperfective past), and perfect,

2b) whose encoding is distributed across the combination of multiple morphemes rather than a single dedicated TAM slot,

2c) which morphemes, synchronically, have no apparent inherent meaning, by being made to show up in multiple and contradictory TAM encodings, and

3) lexical mood would be cool too I guess.

2b/2c have consistently given me the most trouble. I have tried searching for a description of the development of Proto-Kartvelian verbs to use as inspiration but have not been able to find one, except some off-hand lines from Tuite(? If I remember correctly?) that Proto-Kartvelian probably originally only had aspect, not tense (like PIE).

So let's say I start from a Proto-Kartvelian-esque/PIE-esque starting place of just aspect - some verbs are inherently perfective, others inherently imperfective, and some sort of morphology to swap aspects. Looking through the EOG:TAMLW and WLG, they strongly imply tense comes from aspect and aspect comes from grammaticalizing auxiliaries, so I guess that's what I'm doing, because I have no other leads.

The problem I've been running into over and over and over is "the present and imperfect are both imperfective. Which one does the imperfective aspect turn into, and once it does, *how do you get the other one?*"

One way is evolve the imperfective two separate times, perhaps once from "go" and once from a copula, e.g.


    t'q'   -a
    t'q'   3.SG.M
    "he t'q'ed" (aorist)

    t'q'   -v   -a
    t'q'   go   3.SG.M
    "he was t'q'ing" (imperfect)
    < lit. "he goes t'q'ing"

    t'q'   -ul    -i     -a
    t'q'   PTCP   STAT   3.SG.M
    "he t'q'es" (present)
    < lit. "he is a t'q'er" or "he is a t'q'ing one"

The issue with this is it seems to fail goals 2b and 2c. Aside from the root and person marker, there isn't any morphology shared across multiple and contradictory conjugations so as to make the meanings of the individual morphemes inscrutable. We pretty plainly just have a slot that is realized -∅- in the aorist, *-v-* in the imperfect and *-uli-* in the present, and I don't want TAM encoding that straightforward.

Alternatively I could evolve the present/imperfective distinction the way I think some IE languages did, which is to start slapping the aorist's person markers onto the imperfective stem.

But... um, as you can see, there aren't currently separate person markers for the perfective and imperfective; they're all just using *-a* for 3.SG.M at the moment. 3 persons (1st, 2nd, 3rd) x 2 numbers (sg, pl) x 2 genders (masc, fem) = 12 seperate markers I need if gender is distinguished in all persons, which was the plan. Add onto that I had been planning for verbs to be polypersonal so I need 12 additional *object* markers (which may or may not be the same as 12 genitive person markers for verb constructions that originate from a genitive construction). Having to come up with another 12 on top of that... it's just... a lot. And it doesn't really feel motivated to have separate markers for the perfective vs. the imperfective without an explanation for *why* they would be different in the perfective vs. imperfective.

Perhaps as long as we're using auxiliaries, the different person markers originate from different conjugations of the auxiliaries, e.g. "is" vs. "was" or "goes" vs. "went". But I feel this just moves the problem. Now I have to come up with 12 suppletive forms of "go" or "be" (possibly 24 if they themselves need to contrast perfective vs. imperfective) with no explanation for why they would have the forms they have, other than just "it goes back to the proto, they were suppletive in the proto too ¯\\\_(ツ)_/¯". It's a lot of random forms to come up with and no very satisfying explanation for why they are the way they are.