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

[John E Clifford <kali9putra@yahoo.com>]

The reason for wanting to have a real grammar of Lojban hardly seems to need 
explaining; why would we want less?  Especially given our claims about it all 
being spoken logic (of some sort, at least).  For that "logic" we might need 
more quantifiers than Lojban has at present, but the ones it does have are a 
good start.  And, of course, there is the shift to an intensional logic which is 
absolutely required (and ought, I suppose, to already be part of your back 
translating program).  Eventually, of course, pragmatics has to enter in -- in 
elisions and anaphora if nowhere else -- but that is also a part of any useful 
reconstruction, so not a new idea.  And, of course, there is the notion 
(untested) that the real grammar of Lojban will be much simpler than that of any 
natural language.



----- Original Message ----
From: Martin Bays <mbays@sdf.org>
To: lojban@googlegroups.com
Sent: Mon, December 12, 2011 12:53:45 PM
Subject: Re: [lojban] semantic parser - tersmu-0.1rc1

* Monday, 2011-12-12 at 10:26 -0800 - John E Clifford <kali9putra@yahoo.com>:

> Well, once again, it is trivial to get some Lojban sentence for each FOL 
> sentence; the trick is to get every legitimate expression of that FOL 
>sentence.  
>

Ah! Yes, that's harder. Unclear why you'd want to do it, although
finding the *shortest* lojban sentence expressing a given FOL sentence
is a reasonable problem. If we talk literally just about FOL, that
shouldn't be too difficult, as the fragment of lojban which corresponds
to FOL is quite small and easy enough to understand (modulo minor
controversy regarding some edge cases, as discussed in this thread).

> Or, in other words, a grammar that explains every Lojban sentence in terms of 
> FOL (well, I am ready to concede this may be SOL and that, of course, very 
> extended).  In particular, it has been noted, non-standard quantifiers (like 
> "most" and "few", say) and restricted quantifiers present problems at the 
>lowest 
>
> level -- what is a rational way to derive them from logic (or how must we 
>expand 
>
> our notion of logic to make a derivation, if not trivial, then at least 
> rational). 

Extending FOL with generalised quantifiers corresponding exactly to
those in lojban seems the obviously correct thing to do.

Giving semantics and proof theory for the resulting logic is a separate
issue, and I don't think we should expect anything very clean or
complete (because pragmatics is crucial to understanding e.g. "most").

Martin

> ----- Original Message ----
> From: Martin Bays <mbays@sdf.org>
> To: lojban@googlegroups.com
> Sent: Mon, December 12, 2011 10:24:00 AM
> Subject: Re: [lojban] semantic parser - tersmu-0.1rc1
> 
> * Monday, 2011-12-12 at 09:16 -0600 - John E. Clifford <kali9putra@yahoo.com>:
> 
> > Well, point one goes against orthodoxy for languages in general and
> > certainly for Lojban in particular.  Point two is generally true, but
> > much harder is not impossible.  In general, the move from FOL to
> > Lojban ought to seem easier than the reverse, but even that may be
> > illusory.
> 
> I'm not sure what you're getting at here. FOL to Lojban is entirely
> trivial. Assuming a relational language and that every relation is
> the interpretation of some selbri: use {su'o da zo'u} and {ro da zo'u}
> for the quantifiers, and use geks for the boolean operations. There are
> no issues at all.
> 
> Martin
> 
> > On Dec 11, 2011, at 9:37 PM, vitci'i <celestialcognition@gmail.com> wrote:
> > 
> > > On 12/11/2011 09:31 PM, John E Clifford wrote:
> > >> Wait.  If we have a means of converting FOL into Lojban, it has -- to be 
> > >> acceptable -- to provide a rule for a means of converting some logical 
> > >> expression into each Lojban sentence.  So, in particular, it has to 
>provide 
>
> >a 
> >
> > >> systematic explanation for even those sentences which are problematic to 
>the 
>
> >
> > >> back conversion.
> > > 
> > > That doesn't follow. There may be some well-formed Lojban expressions
> > > that are not produced by converting any FOL expression. Alternatively,
> > > there exist some relations which are much harder to compute in one
> > > direction than the other.
> > > 
> > > -- 

-- 
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.