[YG Conlang Archives] > [engelang group] > messages [Date Index] [Thread Index] >

[Date Prev] [Date Next] [Thread Prev] [Thread Next]

Re: [engelang] Xorban: Termsets



On Thu, Sep 27, 2012 at 10:14 PM, Jorge Llambías <jjllambias@hidden.email> wrote:
 

On Mon, Sep 24, 2012 at 12:36 PM, Mike S. <maikxlx@gmail.com> wrote:
> On Mon, Sep 24, 2012 at 8:57 AM, Jorge Llambías <jjllambias@hidden.email>
> wrote:
>>
>> In selma'o notation, this would become:
>>
>> sentence := formula | CA formula | WA
>> formula := CCA | term formula | foretree formula | formula aftertree
>> aftertree := HI | term aftertree | JA aftertree aftertree
>> foretree := HE | term foretree | JA foretree foretree
>> term := NA | LA formula | JA formula
>
> Eight parts of speech so far isn't too bad. As a side note, I'll quickly
> mention here three things about trees I've discovered so far that the
> speaker will have to remain aware of:

I'm not too happy with all that, since it is now not apparent from the
production "term formula" whether the term has scope over the full
formula or not. I think he/hi need more work.

co ma'a xrxe

The simplest solution that I can see: (1) terms simply have to combine with whatever can consume them first, and then the trees combine with the "core formula" secondarily, as reflected in the order of the rules (bottom first).  Thus there is a two-level precedence hierarchy:  when in doubt, terms combine first.  This does necessitate a speaker move "f-" term from the trees into the core predicate in order to maintain logical form, which is nothing complicated compared to what happens in English wrt to clefting and wh-fronting.  It seems like the natural thing to do anyway.

There is another solution possible though:  (2) Certain terms in an prior tree (namely bindings of variables not otherwise bound and "f-" terms) that are not specifically overridden in the subsequent tree are assumed to be in the inner tree as well (it generally won't make sense to apply "f-" to a tree anyway, because trees aren't formulas and therefore can't have event arguments;  and it doesn't harm anything to add more variable bindings).   Aftertrees would be considered more prior than foretrees, the better to reflect their afterthought nature.   This could not apply to "na", though.  Here's an example previously shown:

lo ( fo la ma djna le nnle li crdi he )( le nxle li ckti he ) dndakike
"(O) is the situation in which John gives to the boy a radio and to the girl a book."

Note that under solution (2) the second foretree will inherit the fronted terms "fo, la ma djna" from the first foretree which would naturally consume them and hide them.  It wouldn't inherit "le nnle, li crdi" because "le nxle li ckti" would block/override those bindings.

It would work similarly either way.  There are going to be some syntactic restraints either way or any other way you can think of, and as I mentioned I am in no rush to make these particles official before they're fully tested.

--
co ma'a mke

Xorban blog: Xorban.wordpress.com
My LL blog: Loglang.wordpress.com