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


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

Re: [engelang] Xorban Development



Jorge Llamb�as, On 05/09/2012 00:37:
On Tue, Sep 4, 2012 at 8:49 AM, And Rosta<and.rosta@hidden.email>
wrote:
Jorge Llamb�as, On 04/09/2012 02:56:
On Sun, Sep 2, 2012 at 8:27 PM, And Rosta<and.rosta@hidden.email>
wrote:

The rules are:

Every word tries first to bond syntactically with words
preceding it; if it fails, then it bonds syntactically with
words following it.

That would make it very difficult to ever end a sentence, since
every sentence would always remain available for some future word
to grab it. Doesn't that impose an even more unbearable burden on
the listener than having to remember the restriction on some
variables? Or maybe you could have a full stop.

Rather, let every word try to be final in a macrosyntagm, and only
 if that is not possible should it set forth links to following
words. I'm assuming that binding is syntactic.

How can you tell when a macrosyntagm is finalized?

It's a valid question, and one could choose various alternative answers. I would say that the parser tries to create a macrosyntagm first. If it succeeds, then that macrosyntagm is no longer accessible to what follows.

For example, if I think I have a finalized macrosyntagm, and the
negation operator "na" comes along, does it try to be final and does
 that mean that what I thought was a macrosyntagm wasn't one after
all, or is the old macrosyntagm part of the new one?

I'd opt for the finalized macrosyntagm no longer being accessible to the parser, so na would be sentence-initial.

I'm not completely sure what a macrosyntagm is. Is it any part of a
sentence that can itself be a sentence? But then unless you have a
full stop (or something like my illocutionary operators that "close"
a sentence) I don't see how you can ever be sure that the putative
macrosyntagm is not part of a bigger one that encompasses it.

A macrosyntagm is a phrase that's not contained within a larger phrase.
Anyway, so that we don't get too sidetracked, let me recapitulate how we got onto this. All variables must be bound, so when we come to an unbound variable, the options are:
(1) Ungrammaticality
(2a) the variable is implicitly bound and is unrestricted
(2b) the variable is implicitly bound and preserves the restriction it had the last time it was bound
(3) the variable is bound by something that follows it

(1) pains my sensibility. (2a) had originally been my preference, but one short variable (which we're currently calling o'e but which I would give the form /ai/) would suffice, so it doesn't make sense as the default for variables in general. (2b) is okay for a subset of variables, e.g. V'u, but is (for reasons I've gone over before) objectionable as the default. So (3) is merely a solution of last resort.

I'd be fine with declaring that the default rule for unbound variables varies by Xorban dialect and is outside the common core (the 'kernel'?). Then we could get on with developing the kernel in which all variables are bound.

Unless you allow sentences like "for every x, if snow is white
then it is now raining", which I don't really see a reason to
disallow.

I don't see a need to introduce a rule specifically to disallow
them, but in the syntactic system I outlined, I think it would make
sense to say that in "bcda fgha li", li's dependents must follow it
rather than being "bcda fgha". And likewise in "li bcda fgha", at
least "fgha", and prob "bcda fgha" wd have to be dependents of a
following quantifier.

If you do allow things like "for every x snow is white",

In the suggested variable-precedence system, it wouldn't be generable.

then I don't understand how you can tell that "li" in "bcda fgha li"
is not supposed to take "bcda fgha" as its dependents. Why couldn't
it then be followed by "jkla la" so that all the 'a's end up
properly bound?

Right. So ever variable must be bound; every binder must bind.  That excludes"bcda fgha" as dependents of "li".

So it seems that getting an operator to look to the right for
its dependants would tend to be quite difficult.

How so? The job of the parser is to establish predicate--argument
and quantifier--variable relations, so at any given word, the
parser knows whether or not it contains an unbound variable.

Even if that was to be part of the syntactic stage of parsing, I
still don't see when you would allow "for every x snow is white" as a
valid sentence and when you wouldn't. If "for every x" follows "snow
is white", does it take it as a dependent?

No."for every x snow is white" would never be a valid sentence.

Would you use vV for something like "when John visits his
grandmother, he usually brings her chocolates" too?

You could, yes, though in this instance you could have "John -A,
John's grandma -I, occasions when A visit's I -U, occasions when A
 brings I chocolates -E, majority of U are E". But if the sentence
 were "When John makes a cup of tea, he usually forgets to drink
it", the vV system wd be best I think.

Is that because John and his grandma are typically pictured as
individuals, while "cups of tea" is typically seen as "dividual"?
That's really the only thing that distinguishes the two cases.

Good point. The answer to your question is Yes.

So, whenever dividuality matters, quantifier scope matters, and vV is needed. And when dividuality doesn't matter, quantifer can be lV and vV isn't needed.

I think it's good to have both systems, the xorloan one and the classical one augmented by vV.

--And.