Re: Reductive Paraphrase again
John and John,
If your conclusion is strictly true, then a program that "experiences"
the meaning of words might be able to handle linguistic dialogs based
on its experiences. But how does that program start? If its lexicon
is randomly invented, nobody else will be able to communicate with it.
So starting with a nucleus of words that have widespread meaning
(e.g., WordNet) could at least base the program in communicable
reality.
Here's an example of a European research project that uses some of
Luc Steele's methods:
http://www.eet.com/news/design/technology/showArticle.jhtml?articleID=164902452
Care to opine about whether this has some chance of adding new
knowledge about linguistics?
Rich
John Sowa wrote:
> Rich and John B.,
>
> Children (and adults) learn words by associating them with
> images, feelings, actions, and other experiences. There are
> very few basic words that can be adequately defined by
> reductive paraphrases -- and most of those are found in science
> and engineering.
>
> Mathematics is the primary source of examples of reductive
> paraphrases that are precise and reliable -- but only because
> the people who invented those ideas deliberately designed them
> (or chose them) to be reducible by formal methods.
>
> Unlike some people who try to set up artificial barriers
> between common sense and logical and/or mathematical reasoning,
> I believe there is a continuity. Logic and mathematics are
> based on perfectly natural human abilities, but they have been
> developed to a high degree of skill. I would compare them to
> gymnastics and other atheletic skills, which require a very
> high degree of professionalism. Another example would be the
> musical skills of professional musicians.
>
> All those skills -- mathematics, gymnastics, and music -- are
> "natural" in the sense that everybody can do them to some
> extent, but doing them at a professional level requires an
> enormous amount of dedication and practice.
>
> Summary: I would say that reductive paraphrase is possible for
> a small percentage of the words that people use -- primarily
> words in science and engineering that have been deliberately
> designed for reducibility. It may be possible to develop
> approximate definitions for others, but those approximations
> would always be highly specialized for some specific purpose.
>
> For more detail on these and related issues, I would recommend
> my article on the Challenge of Knowledge Soup:
>
> http://www.jfsowa.com/pubs/challenge.pdf
>
> John
>
>