On Tue, Sep 18, 2012 at 5:13 PM, selpa'i
<seladwa@hidden.email> wrote:
Will xorban ever have something like sumtcita?
It already has them IMO. A "tcita" would be the adverbial operator "ju" (in Lojban the thing that makes modals out of selbri, I forget what it is) in conjunction with a formula such that it forms an adverb. Compare:
sa xrma bjra = Some horse runs.
ju stra = ("A" being) fast (adverb)
sa xrma (ju stra) bjra.= Some horse runs fast.
Literally: "Some horse, in the same/composed situation, is fast and runs."
In effect, "ju stra" is a modifier of the head predication "bjra". Now that we have that, a case tag is just an adverb that has an object (in Lojban sumti). We can add objects to adverbs easily. An object is just a binder followed by a formula, in which position that formula is called a "restriction". Here is the typical way:
ju plnaki = ("A") using something "I"
ju li tpli plnaki = using legs
sa xrma (ju li tpli plnaki) (ju stra) bjra = Some horse runs fast using legs.
(Advanced usage note, ignore for now: As shown, it's very often better to stick the object in front of the formula
instead in front of the "ju" so we can reuse the variable immediately. I
prefer "i" for these short-scope bindings. But if you want the
variable of the object to be reused later, then stick the object before
the "ju".)
What about relative clauses?
That's just a modification of a restriction. Instead of "ju", often we can use the ordinary logical-and operator "je", because restrictions are usually stative and composition semantics feel weird for them, but if you think you need "ju" feel free:
je pnsika = such that "I" owns "A"
je li nnli pnsika = such that the man owns A
sa (je li nnli pnsika) xrma (ju li tpli plnaki) (ju stra) bjra = The horse such that the man owns it runs fast using legs.
I assume that you can make do without either, but they are probably a
useful shortcut.
We already have them. Any shortcuts would be syntactic sugar as far as Classical Xorban is concerned, but yeah, non-classical Future Xorban may get some by popular demand.
If, say, "mi citka lo plise ca lo nu lo gerku mi batci", you can just
turn it into a bridi based on cabna:
la fa le plse ctka'ake li fi lo grko btcoka'a cbnaki
At this stage of CX development, what you have there seems flawless, but I am not sure we yet have consensus or a complete design wrt all the semantics involving worlds, tense, situations and the like, so things could change later.
Obviously, this requires quite a bit of forethought.
(Can you reset the l- variables when entering an abstraction? I guess if
not, you'd run out of V's pretty quick if the sentence gets long and has
lots of nesting)
You can reset a variable as soon as it's out of scope of its binding. IMO it's dangerous to do this during a current binding (nested bindings of the same variable Considered Harmful). See what I did with "i" in the final horse example, which seems safe. One of the things I am trying to contribute to Xorban is a style with patterns and practices that make it relatively easy to handle variables with confidence.
Would this one be correct as well (hoping that the other one even
worked, my first try at producing anything non-trivial)?:
la fa la plsa ctka'aka le fe la grka btcaka'a
[... cbnaki]. Both nested "la" bindings work but might confuse the listener. I would especially avoid the "la plsa" usage.
Anyway, since this is probably not always convenient, I'm wondering if
maybe a subordinate clause system would be useful. I think you only
really need one word to get both sumtcita and relative clauses. (And it
doesn't seem parentheticals are really about that, am I right?)
Jorge just added parentheticals. They are one means of handling nonrestrictive clauses, though I have not explored that usage yet.
I have to step away from the keyboard but I will send this first reply for now and reply to the rest of your email later.