[Corpora-List] Suggested Track for Studying Computational Linguistics

[Mark P. Line]

John F. Sowa wrote:
> I partially agree with Mark Line:
>
> ML> I think statistical NLP is not linguistics --
>
> But I would qualify the following point:
>
> ML> it's what computer scientists do instead of
>  > linguistics.
>
> I would say that it's what *anybody* would do when
> the problem is not sufficiently well understood to
> suggest a more explanatory model.

Well, that may be very nearly true if "do" refers specifically to the NLP
enterprise, as opposed to empirical and theoretical research on human
language itself. But most linguists don't do NLP.

In any event, I don't think there's any dearth of useful explanatory
models for human language. In the rush to get products to market using
statistical approaches, most of those models have been ignored because
they are believed to take too much time to implement -- or because
non-linguists have difficulty understanding how to use them.

That means that even in the NLP arena, it's the patient, long-term, deeply
linguistic R&D programs that will win out in the end.


> I also agree with Christopher Brewster:
>
> CB> I think computing involves a lot of hours of studying
>  > and acquiring a set of skills which dealing with the
>  > difficulties of the linguistic aspects of NLP do not.
>
> But I would add that linguistics requires skills that
> are independent of the skills of a computer scientist.
> Linguists who only know one natural language may be very
> good at what they do, but their intuitions into the nature
> of language may leave something to be desired.

This is a good example of what I was driving at. It's possible that it's
more a matter of cognitive style than learned skill as such. That might
mean that somebody considering a choice between linguistics and computer
science as an academic home might do well to assess his or her cognitive
style before making a final decision. (As a matter of personal preference,
I could do with fewer math-heads in linguistics.)


> For any student interested in linguistics (computational
> or otherwise), I would recommend the "reflections" by
> Barbara Partee, who majored in math as an undergraduate
> with a minor in philosophy and Russian.  Then she went to
> MIT to earn a PhD under Chomsky and then taught linguistics
> at UCLA, where she learned formal semantics from Montague:
>
>     http://people.umass.edu/partee/docs/BHP_Essay_Feb05.pdf
>     Reflections of a Formal Semanticist as of Feb 2005
>
> That background is hard to duplicate, but some similarly
> varied set of skills combined with a great deal of native
> talent would be conducive to innovation.

Well, Barbara Partee's approach to language is almost an outlier in its
mathematical mindset (mathematical, not computational).

I would hold up people like Tom Givon, Dick Hudson, Syd Lamb and Ken Pike
(just to name those that have had the most influence on me personally) as
good examples of what can be accomplished with a deep understanding of
what human language is like.

But again, it may still be largely a question of cognitive style.


-- Mark

Mark P. Line
Polymathix
San Antonio, TX