language?

btanksley@hifn.com btanksley@hifn.com
Fri, 7 Jan 2000 09:55:37 -0800


> From: David Vennik [mailto:soma@dynamite.com.au]

> What we need is a program which processes 'natural' language really.I
> have now decided that the only thing preventing a computer from
> understanding human language is the lack of a sufficient 
> model (an exact
> and detailed specification) of how it works. We can all 

This is already solved -- human languages are parsed by a "unification
machine", which is a step above a context-dependant PDA (I'm not sure how it
relates to a Turing machine).  There are several parsers and lexers at
http://www.sil.org/htbin/search-softcat.pl?kw=parser, and other
language-related software elsewhere on the same site.  Interestingly but
unsuprisingly, UGs can also parse the lesser languages, so (DFAs << DPDAs <<
NPDAs << unification machines).  Unification grammars can detect type errors
and so on, though.  They also turn out to be _very_ good at error detection;
they can usually parse the meaning even when there are grammetical errors
(it turns out that most language errors humans make are actually not
formally errors, but rather shifted meanings!).

-Billy