Re: nature -> "human brain" -> "language terms" ==>> knowledge ?
On Friday 18 March 2005 02:07, Alexander Povolotsky wrote:
>
> Rob Freeman <lists@CHAOTICLANGUAGE.COM> wrote:
> >There is nothing wrong with your method Alexander.
> >
> >If it is of your own devising you are to be congratulated, because it
> >seems to be basically "distributional analysis", which is a key
> >Machine Learning technique (and one which revolutionized linguistics
> >in the... 30's?)
>
> Well ..., I wonder was there Object Oriented analysis at that time ? ;-)
Perhaps, but the performance overheads of well structured OO code were
probably unacceptable on slide-rules and note-pads :-)
> Also please note that the method of the processing I am proposing creates
> graphical hierarchies of nodes, thus lending itself to further
> "post-processing" using methods from the Theory of Graphs - was that
> covered in the 30's (or later) ?
There is a survey which is a bit more modern. I don't know if it is the best,
it is just one I know:
Powers, IJCL, 1997, "Unsupervised learning of linguistic
structure" (http://www.infoeng.flinders.edu.au/papers/19970003.pdf)
> If this classification was implemented (statically) - could you point me to
> the reference to (or even better to the instance of) such complete statical
> database of such grouping for any given language ? (such reference will be
> indeed much appreciated on my behalf !)
Just look up American Structuralism, Alex.
From the point of view of the discussion here I think the most interesting
thing to note is the way American Structuralism was deposed as the dominant
paradigm for linguistic study after 20 years or so in the top job.
It was Noam Chomsky who knocked it off its perch
The way Chomsky displaced Structuralism was to make the observation that
distributional analysis resulted in representations (phonemic in point of
fact) which were "incoherent and inconsistent".
He used this observation to discredit distributional analysis and supplant the
whole method and theory of structuralism with his own theory that language
structure must be innate.
That was the great revolution in linguistics of the '50s, and one we are still
largely with to this day.
It is a sharp observation, and one Chomsky is to be credited on.
But there is another way to interpret the experimental observation of
"incoherent and inconsistent" linguistic representations. It wasn't
palatable, or even it seems conceivable to Chomsky and others, but it is a
perfectly reasonable interpretation to assume that linguistic representations
_really are_, fundamentally, "incoherent and inconsistent."
-Rob