Re: Icon

Keith J. Miller (
Wed, 5 Jul 1995 22:00:55 -0400 (EDT)

I was surprised and pleased to find a query about Icon on the corpora
list. We use it here at LAS for in-house purposes (processing of large
text files), and I have been wondering why more people weren't using
it. We were introduced to Icon when we hired a new employee who used to
teach in the U of A computer science dept, with Ralph Griswold, the
founder of the Icon language. He, too included it in a survey of comp
lgs. course he used to teach there. I am told that Icon is an offspring
of, if not similar to, SNOBOL (which enjoyed a wider popularity).

I find Icon rather simple, and even perhaps fun, to use. There is a relatively
good book concerning its use, titled (I believe) *The Icon Programming
Language*, by Ralph Griswold and his wife (whose name escapes me at the
moment). As I said, I found it powerful, and relatively easy to use,
although there are some concepts that are likely to be new at first.
Comp ling / AI students will likely be drawn to it because of its
suitability to these tasks, and some features that are slightly
reminiscent of PROLOG (e.g. generators in Icon are similar to PROLOG's
automatic backtracking, and Icon programs are also driven by the notion of
success and failure). I would encourage you to at least look at the
language and experiment with it a bit. Another nice feature is that it
is available via anonymous ftp, although the creators do ask you to let
them know that you are using it. In return, they do readily provide
answers to any questions you might have regarding the language, in
addition to the discussions in the usenet newsgroup.

Good luck, and I'd be interested to see what others have to say, and what
you eventually decide.

----- Keith Miller


Keith J. Miller Language Analysis Systems
Georgetown University The Center for Innovative Technology
Department of Linguistics Herndon, VA 22070
Computational Linguistics 703-834-6200