On Friday 28 September 2001 14:33, Karl Ramberg wrote:
> "Ohshima, Yoshiki" wrote:
> >   I believe that the good ability of
> > reading/writing/speaking the mother language is more helpful
> > for kids to acquire more things in long-run.  If I would
> > have kids, I hope they would learn *correct* mother language
> > before learning programming.  I wouldn't want to teach them
> > some compromized (wrong) sentenses.

I agree. But sythesis is far easier than analysis, so what I was 
proposing was a division between automation and manual labor that would 
actually reduce the amount of bad sentences in the system.

Your example might be coded as something like:

( '[S]
  agt(have(equ>own),' , {'I'. 'you'. 'she'} atRandom , '
  obj(have(equ>own),' , {'dog(icl>pet)'. 'fox(icl>pet'} atRandom , '
  qua(pet,' , {1. 2} atRandom , '
  [/S]' ) asUNLString printString

This isn't really correct (see 
http://www.dfki.de/~becker/VorlesungSS99/unl/index.htm or, for a more 
readable version but without the graphics, 
http://www.dfki.de/~becker/VorlesungSS99/unl/tsld001.htm) and is way 
too awkward, but the idea is that something like this code would 
generate a correct sentence in any of the supported languages.

With a fully manual system coverage won't be good. Then children will 
face a few correct sentences in their native language and many others 
in languages they don't understand.

> >   I'm sorry I can't suggest any good ideas, but natural
> > language issue is certainly time-consuming.  I'd say I want
> > SqC to put higher priority on "cutting edge" or "break
> > through" type of works.

The code to be written isn't too complex. The major time sink is 
collecting the needed database. Unfortunately the UNL license didn't 
give me the impression that it would be possible to use their work in 
an open source project.

> I find icons and pictures a interesting way to deal with
> this. Numbers are language independent. So is music notation.
>
> Anybody have any pointers to research/work in this direction ?

Weren't some scientists using musical notes and colors to communicate 
with extra terrestrials? Oh, wait! That was a movie... "Close 
Encounters" something :-)

But seriously, we are already using numbers to communicate: ASCII or 
something reasonably similar. If you use an encoding, then there must 
be a context and you have nearly all of the complications of natural 
languages.

For communicating with pictures, I found Chis Crawford's 
"Trust&Betrayal" game very interesting. Strangely enough, there are no 
pictures in this essay: 
http://www.erasmatazz.com/library/JCGD_Volume_1/Siboot.html

-- Jecel