Finite clause types at last

In the winter of 2012, while on a dog walk in Washington Manor, out of the blue I had an idea about Koa clause marking. This idea felt crazy and revolutionary, but also beautiful, and possibly even ingenious. I dismissed it as too revolutionary and let it go for almost a decade, but never forgot it; and in the past few years of feverish Koa development I haven't been able to stop thinking about it.

The idea concerns the i that precedes the VP when it lacks a pronominal subject. Prior to 2012 I assumed the rule was simply that verbs have a pronoun slot that must be filled, so if an overt pronoun is absent, i is required as a placeholder. This is the case (more or less) in Toki Pona, which originally inspired this marker.

The revolutionary idea was that perhaps I had misanalyzed my own syntax. What if, instead of being a  pronoun, i is marking the clause type? And if that's the case, what if the ko I had previously regarded as a specifier that nominalizes clauses, or a complementizer, is actually just marking another type...and the u I thought was a relativizer was marking yet another? And what if there could be other markers too?

Borrowing from the original post in January 2012 in which I introduced and then immediately discarded this idea, this would give us the following marking for finite clauses:

i - verbal
ko - nominal
u - adjectival

...and the following template for the VP:

(SUBJECT) TYPE (PRONOUN) (TAM) VERB (OBJECT)

There's an important bit of subtlety here that I need to make sure to touch on before continuing. When languages nominalize clauses they usually become non-finite, sort of by definition; but as I said above, clauses with these markers are all finite. What one has to remember is that every Koa predicate can be verbal, nominal, or adjectival depending on its syntactic position; in this system, clauses are no exception. These clause type markers identify what the clause is doing in syntax -- i.e. whether the clause is acting like a verbal, nominal, or adjectival predicate -- but the clauses are still finite regardless.

Clause nominalization, the nonfinite variety, is an entirely different topic...though it does also exist in Koa, and would also be able to used for similar purposes! This is all very easy to confuse, and I think I've been consistently inconsistent or imprecise in my language here in the past: in Koa there's an extremely important distinction between a nominal clause and a nominalized clause. I should possibly ideally find a different term for one of these!

Anyway, those who have been following the plot closely may have noticed that I've been nonchalantly slipping structures like these into example sentences and translations for more than a year. After I discovered last year that Macedonian does "nominal clauses" this way too, e.g.

не сакам [ Јуле да знае ]
NEG want.1SG [ Julie NOM.CL know.3SG ]
"I don't want Julie to know"

...for which the precisely parallel Koa translation would be...

ni-na-halu [ le Iúliki ko ilo ]
1SG-NEG-want [ NAME Julie NOM.CL know ]

...and also discovered that Basque does nominal (and conditional, and adverbial) clauses this way:

[ bere aita Californian dago-ela ] esan_du Mikelek
[ his father California.LOC be.3S-NOM.CL ] say.3SG Michael.ERG
"Michael says his father is in California"

le_Mikele i sano [ taémaka ko ne le_Kalipónia ]
Michael VB.CL say [ his.father NOM.CL LOC California ]

...and that Swahili does relative clauses this way:

kitabu [ a-li-cho-ki-soma mtoto ]
book_β [ 3SG.SUBJ_α-PAST-REL_β-3SG.OBJ_β-read child_α ]
"the book that the child read"

ka tusi [ ka toto u luke ]
DEF book [ DEF child REL.CL read ]

...I was convinced that this isn't a completely outlandish, typologically ridiculous strategy at all. It's certainly not augmenting Koa's IAL-worthiness, but as I keep pointing out, in some cases the art is starting to feel more important than the charter, and these structures feel right for Koa. They're simple and elegant, and Koa syntax as a whole actually makes more internal sense when they're handled this way.

And so, by the power vested in me, I hereby admit these structures into the Koa canon. As ever I have the right to change my mind, but I'm going to give them their fair shot and see how Koa feels in practice when it makes full use of them.

I alluded above to the possibility that there might even be more than the three clause types referenced so far, and indeed many more have surfaced. Here are the categories and subcategories of clause types as identified so far:

i - verbal: neutral unmarked
vo - verbal: presentative
oe - verbal: obligative
ea - verbal: hortative
ko - nominal
u - adjectival
ve - adverbial
ha - conditional

Notes about distribution:

1) Verbs with pronominal subjects have some special properties, which affect the use of these clause markers. i is not used (or is extremely marked, TBD); ko is optional; and u is also optional. The remaining markers are required in all circumstances regardless of clause type.

2) A clause carries one and only one clause marker; BUT

3) Serial verbs to the right of the main verb are marked by i regardless of the clause type of the main verb (see example below).

In any event, this is really just an introduction and opportunity for me to welcome this new(ish) syntax out loud. Many of these types will need a post of their own to demonstrate usage; first up and currently in progress is a discussion of the newly discovered and rather exciting conditional structures, which have been patiently waiting for someone to notice them for 25 years.

Oh, and about those actually legitimately nominalized clauses I mentioned earlier...it turns out I got those right in that same 2012 post as well, actually. They're formed with specifier version of ko, and all clause markers -- as markers of finiteness -- are omitted. One example for now, pending a full treatment in a later post:

le Mia i sano ko [ ka moa Ø ma-lalu i poli ]
NAME Mia VB.CL say NOM [ DEF chicken NONFIN IMPF-sing VB.CL much ]
"Mia said the chickers were singing a lot"