Please help.

[masako]

I don't have much advice here, but:

Are prepositions even a thing in a head-final SOV lang?

There are a handful of SOV languages with prepositions and at least two languages that have prepositions and are probably head-final (relative clauses and adjectives before nouns). At least one (Tigré) has all these properties.

This is excellent, thank you.

[Jonlang]

I don't have much advice here, but:

Are prepositions even a thing in a head-final SOV lang?

There are a handful of SOV languages with prepositions and at least two languages that have prepositions and are probably head-final (relative clauses and adjectives before nouns). At least one (Tigré) has all these properties.

This is excellent, thank you.

Wasn't Latin pretty much SOV with prepositions, too?

[masako]

Wasn't Latin pretty much SOV with prepositions, too?

Colloquial Latin, yes. But SVO was predominant.

[masako]

So, update, I am moving ahead with using prepositions and SOV. I am basing this on Persian's syntax, and the subordination of Quechua.

I would very much like thank everyone who contributed to this thread. I sincerely appreciate you.

[Kuchigakatai]

Wasn't Latin pretty much SOV with prepositions, too?

Colloquial Latin, yes. But SVO was predominant.

I think you read and said that the other way around.

The change can be observed across time actually. Early on, Plautus and Cato the Elder (3rd-2nd centuries BC) use a lot of SOV. In the 1st c. BC, Cicero uses it less often in comparison, especially so in his personal letters as opposed to his speeches and books, showing a difference in register where SVO is more colloquial. By the 2nd century AD SVO is apparently slightly more common in the likes of Gellius and Apuleius, and then, after the 3rd century crisis, 4th century Latin then appears as more SVO than SOV, especially strongly so in less formal writings like Augustine's Sermons and Egeria's Pilgrimage.

I once came across a paper that also noticed that in 7th-8th century Latin, legal documents from France are SOV more often than Italian ones. This difference is still noticeably present in Old French and Old Italian as late as the 12th century, where the retention of the distinction of nominative vs. accusative in the former may have facilitated some retention of SOV as a secondary word order... It is a bit unclear to what extent the French were simply more interested in imitating the ancient SOV though, as there's also a long history of written Romance adopting SOV from Latin (13th century Old Spanish was mostly SVO and VSO, but by the 15th century it would be innundated by the SOV order, much like how it was by borrowings from Latin at the time...).

[Richard W]

Wasn't Latin pretty much SOV with prepositions, too?

Yes, but adjectives and relative clauses mostly follow rather than precede, much as in Persian, where prepositions are predominant.

[Moose-tache]

If you're going to be be using prepositions (as opposed to postpositions) that are transparently derived from verbs, how will the word order shift be made apparent to the listener?

What I mean is, yempa-n tahe follows head-final order. But once the phrase is used as a locative clause within a large sentence, it becomes tahe yempa. If you kept the head-final order throughout, it would look like any other SOV language with postpositions from verbs, like Japanese or Muskoki.

na [yempan-tahe] [ueson-ina] maloha mitan-anya. This is the most dead-simple way to have objects clearly connected to their verbs. But if we derive prepositions from the verbs, then we get something like: na [??-tahe] [yempan-???] [???-ina] [ueson-???] maloha mitan-anya. The object of anya is clearly mita, but it looks like the object of ina is yempa or something. The first time I read "Na mitan tahe yempa ina anya," it looked like "I see the table eating under a dog," because that's what matched the previously established rules of Kala grammar, i.e. unmarked subjects and head-final clauses. You would either need a way to mark the verbs as targeting objects after the verb instead of before it, or you would need a rule saying main verbs follow head-final syntax, while all other verbs follow head-initial syntax.

If your goal is to make it unambiguous how each clause is nested within the other, I wouldn't sweat it too much. These kinds of things are frequently ambiguous in natural language. In the sentence "The dog eats the bone under the table," it is unclear whether the bone or the eating is being described as under the table, but it hardly matters. Sometimes natural languages rectify this by differentiating relative clauses from adverbial ones.

tl;dr: your entire dilemma can be solved by just keeping predicates intact using their default morphosyntax, instead of separating and rearranging them like in your original examples.

[Richard W]

If you're going to be be using prepositions (as opposed to postpositions) that are transparently derived from verbs, how will the word order shift be made apparent to the listener?

The verb takes the accusative, the preposition the nominative.

Indeed, nominative (yempa) and then preposition (tahe) might suggest that what the table was under something! Furthermore, might Na mitan yempan tahe anya mean "I see the dog while I'm under the table"?

[masako]

tl;dr: your entire dilemma can be solved by just keeping predicates intact using their default morphosyntax, instead of separating and rearranging them like in your original examples.

Even your "tl;dr" is technical. But I get it, thank you.

Indeed, nominative (yempa) and then preposition (tahe) might suggest that what the table was under something! Furthermore, might Na mitan yempan tahe anya mean "I see the dog while I'm under the table"?

I mean, maybe...I guess. The idea of ambiguity being inherent in lengthy constructions seems to be common across language typologies. I think it's something I'll address in the grammar rewrite, but it doesn't require any additional rules, necessarily. However, a few examples will likely be included to make sure there is clarity in the explanation.

[Moose-tache]

Here's an example of what I tried (perhaps less than coherently) to explain above.

In Japanese, you could have a sentence like: musubu tabete iru. "He/she is eating musubu." Modifiers generally go to the left, so: oishii musubu tabete iru "He/she is eating delicious musubu." But a clause can modify a noun too. To do this, you simply grab the entire clause like a bear and drop it without guile or ceremony to the left of the head noun: musubu tabete iru kani "The crab that is eating musubu." In some languages this might require a relative marker. For example, in Korean we would say musubu meogeun ge, but really this is optional from a conlanging perspective. I'm not even including object markers, because the head-final order makes that obvious, too. It's the crab that's eating the musubu, because musubu comes immediately before eat, which comes before crab. If you start with a strongly head-final language, there is virtually no special syntax needed to make these nested clauses work.

This can lead to verbs being used as postpositions. Compare these sentences in Choctaw (yes, I'm talking about Choctaw. Everybody drink).
oka mahli pila iyaalitok. "I went south."
towa iit pila! "Throw the ball this way!"

Again, no special syntax or morphology needed. But what happens if we put "tabete iru" before "musubu?" Or if we put "pila" before "oka mahli?" Basic syntactic rules no longer help us, and we need to make new rules to explain where to find the objects of these verbs.
*tabete iru musubu kani "eating something, some musubu, a crab."
This is meaningless in Japanese, so we would need to mark the verb in some way to tell the listener that the object of the verb follows a different pattern: *tabete iru-derp musubu kani. Leaving the verb unmarked simply doesn't work, because there is no way to know when the listener is expected to switch from normal OV syntax to special nested clause VO syntax.

So for Kala, you already have all the tools to make postpositions from verbs with no extra work: you just put the clause to the left of the thing it modifies, maybe with adverbial or relativizing markers if you want some redundancy.

*yempan tahe musubun ina. "He/she eats musubu under the table."

Rearranging the internal structure of the nested clause just makes things harder. That's fine if you want that. Maybe you want lots of weird details in your syntax, like natural languages. But since you approached this problem from a perspective of how to most effectively parse a complex sentence given basic Kala grammar, it would be strange to then abandon default head-final grammar for no reason.

[masako]

*yempan tahe musubun ina. "He/she eats musubu under the table."

So;

mita yempan tahe ueson ina
The dog eats the bone under the table.

??

[Moose-tache]

That's exactly what I would expect.

There is as I'm sure you've noticed some ambiguity inherent in doing things this way, or any other way. The clause yempan tahe could modify the eating, or the bone. The subject mita could also be the subject of nested clauses, or not. Natural languages routinely ignore these ambiguities (is a "happy lion hunter" a hunter of happy lions?), but you can always make things clearer if you wish. Verbs can be marked as relative clauses, or adverbial clauses, or there may be some morphological indication that subjects have changed. The sky's the limit.