Re: [lojban] semantic parser - tersmu-0.1rc1

[Jorge Llambías <jjllambias@gmail.com>]

On Sat, Nov 26, 2011 at 1:07 AM, Martin Bays <mbays@sdf.org> wrote:
>
> It seems related to donkey anaphora -  consider
> {ro te cange poi ponse su'o xasli noi ke'a xi re darxi cu broda}.

Right, it's the same thing. The grammar allows both ke'a and ri to see
as its antecedent something that in principle it should have no
business seeing as its antecedent, and so we have to cope somehow.

> Also, note that the scope-respecting rules sometimes are the most
> intuitive - e.g. in {ro da prami de noi se prami da}. I'm not sure what
> the scope-breaking rules would make of that; I guess something
> nonsensical.

In principle as nonsensical as "no da prami de noi se prami da".

But what you call scope-respecting rules are the rules for "poi", not "noi".


>> I think it's "broda be na ku" that must force tight scope.
>
> I don't see why.

Mainly because that's what the sytntax suggests. It's when we go
against the syntax that we end up with problem cases.


> But in {broda da gi'e brode da}, the first {da} is to the left of the
> {gi'e}. So why shouldn't the {gi'e} be in the scope of the {[su'o] da}?

Because "gi'e" connects two bridi-tails, and so each bridi-tail is
subordinate to it. It's for the same reason that in "na broda gi'e na
brode" the first "na" is in the scope of "gi'e" and not the other way
around.

> So do you also have {da broda .i je da brode} -> {ge da broda gi da
> brode}? If so: that's clearly contrary to CLL (and much usage).

Yes, as I said, that's what I would have. I find the CLL rule somewhat strange.

>> >>   ko'a .e ko'e broda su'o da
>>
>> How does that work for ".e"? It seems that if you do what I do, you
>> end up exporting "su'o da" to newly created preneces.
>
> In a sense, yes; e.g.
>
> da ko'a .e ko'e de broda
> Prop: EX x1. (EX x2. broda(x1,ko'a,x2) /\ EX x2. broda(x1,ko'e,x2))
>
> ; but I think of this as exporting the connective to the prenex, and
> then the inner existential to that same prenex within the scope of the
> connective.

Same with "gi'e".

Presumably in "su'o da .e su'o de" you have the first "su'o" within
the scope of ".e", yes?

> Basically: when parsing, we always have a notion of 'current prenex',
> which corresponds to the closest instance of 'statement' or
> 'subsentence' we're below. All sumti connectives, termset connectives,
> quantifiers and tagged terms (currently just {naku}) export to that
> prenex, in order.

Yes, but I'm just parsing gi'e, na and selbri tags before tail terms
rather than leaving them for last. That's what the syntax suggests,
and I just don't see the advantage of doing otherwise.

mu'o mi'e xorxes

-- 
You received this message because you are subscribed to the Google Groups "lojban" group.
To post to this group, send email to lojban@googlegroups.com.
To unsubscribe from this group, send email to lojban+unsubscribe@googlegroups.com.
For more options, visit this group at http://groups.google.com/group/lojban?hl=en.