Re: Universal grammar

In article <1161934779.658332.265010@xxxxxxxxxxxxxxxxxxxxxxxxxxx>, "Franz
Gnaedinger" <frgn@xxxxxxxxxxx> wrote:

Back to language. What do you think of Rob Freeman's
approach of a grammr based on examples instead of rules?
I find it promising. That is the way I proceed in English.
I don't consult grammar books, I rely on songlines and
sentences I remember.

When I learned language, one focused much on grammar and spelling, but
nowadays, one has realized that it is important to practice, practice and
practice. The same as always has been known in say music performance.

Prepositions are a problem of mine:
in the airplane? on the airplane? on an island? in an island?

I think linguists say that the treatment of prepositions is quite
arbitrary in European languages.

Dictionaries offer some examples, but never enough for me,
so I began to google the groups for the correct use of
prepositions, for correct grammar, also for slang words
I find nowhere else. A billion of messages are my grammer
book and dictionary. Examples instead of rules.

I think this works in humans, because with sufficiently experience, they
can build rules intuitively. This will not work with computers, or at
least not as easily, even with AI, because they have very different
experience.

One picture that comes to my mind is AI and chess, where the strongest
computer programs on the fastest computers are nearly abreast with the
strongest humans, the grandmasters. One knows pretty well what makes a
strong player: the ability to scan forward. If a "ply" is a half-move, a
chess grandmaster has the ability scans some 10-15 plys ahead, I think.
The human needs only to scan a couple of hundred positions in order to
achieve that. The computer needs to scan perhaps a million times more
positions in order to achieve that. How is it possible for a human to scan
just a fraction of positions to achieve the same or better result than a
computer?

So I think this is because the human will deliberately choose moves where
most positions are bad, thus cutting down radically one the needed
possibilities. This can be achieved via an opening theory. The computer
does not have that ability of intent. One might also note that
when computer programs get stronger, humans can play against them in order
to learn the playing style. But a computer chess program cannot become
much stronger in by merely running on a faster computer, because the
search is exponential. Computer programs need humans in order to become
significantly stronger.

So there is something similar going on the parsing of languages.

I wished there was something similar for mathematical
purposes, an easily searchable electronic library of numbers
and number sequences, together with the algorithms that
produce them. Euler's memory for everybody, so to say.

It is difficult because of how the human brain works. It inputs data
fairly exactly, but it is immediately filtered through cognitive processes
before put into memory. This is why, one believes, some with a certain
types of brain damage where only this cognitive filter has been damaged to
be disabled, can develop seemingly amazing abilities, like absolute pitch,
being able to instantly perform a piece that has been heard only once, and
such.

Tools don't replace the human hand, they prolong the arm.
Computers don't replace the human brain, the enlarge the mind.

One uses to say that computers do it differently. They are good at higher
speed exact computations, but poor at understanding structure, that is,
cognition. This is also true about the proofs my program can do: it easily
zips through low level, axiomatic proofs, which humans find very hard to
do, but it can't understand anything structurally more advanced; proof
searches are exponential. Therefore, I am opting for a system where the
human can enter structure, where the program can support by
scanning through all possibilities and check those gory details humans
rather do not want to bother about.

--
Hans Aberg
.

Relevant Pages

  • Hydra v Adams
    ... HYDRA beat Michael Adams 5.5-0.5. ... Perhaps Hydra is simply much stronger than ... None of the games were decided in prepared miracle-rook-sacrifices. ... if the top humans continue to refuse to play ...
    (rec.games.chess.analysis)
  • Re: BSG thoughts(spoilers for the most recent episode)
    ... We already saw that they are stronger and have better stamina ... Note too, that even though the Hulons are stronger, etc, than Humans, ... >> Their fighters are not piloted by Cylons, ...
    (rec.arts.sf.tv.babylon5.moderated)
  • Re: AB squad stats
    ... > Greig Blanchett wrote: ... first generation of humans ever to have more food and rest than they knew ... therefore had even healthier, stronger kids. ... > how England had picked these giant backs to try to run through the Aus ...
    (rec.sport.rugby.union)
  • Re: Universal grammar
    ... strongest humans, the grandmasters. ... whencomputer programs get stronger, humans can playagainst them in order ... cognition. ... Going from examples to general principles can be done, ...
    (sci.lang)
  • Re: AB squad stats
    ... >> first generation of humans ever to have more food and rest than they knew ... stronger kids. ... Then full time professional rugby players, ...
    (rec.sport.rugby.union)