[Corpora-List] Quantitive Corpus Linguistics

[Dom Widdows]

Hi Serge,

Thanks for the summary about Husserl - I will raise his urgency on my
ever-growing and guilt-ridden reading list :) Do you know of one major
work in which he sums up his main positions - or do you need to follow
them over time (like Wittgenstein)?

>  The usual problem I found with putting philosophical ideas into our current research concerns their abstract nature.  Some sort of dynamic interpretation of source texts is quite common in computational linguistics, see Yorick's work on preference semantics.  Some sort of time-consciousness is relevant to linear parsing and memoization techniques.  The crucial question is whether we can get from philoshophical ideas anything useful apart from metaphors.  For instance, it follows from Husserl that we acquire our life-world via interaction with the real world and society.  For linguistics, this means that a corpus is the ultimate source for acquiring ontologies and lexicons. So what?  We didn't know this?

Ahhhhm - it's embarrassing to point out, but we appear to have behaved
as if we didn't know this for several decades of linguistic research,
and used loaded terms like "statistical" and "anecdotal" and "shallow"
to dismiss methods that tried to bring this insight. Now the
pendulum's swung the other way, but is in danger of getting stuck
there too!

> And the mechanisms we use to get a semantic lexicon from this ultimate source (e.g., distributional semantics) are  very different from the ideas put forward by Husserl.
>
>  My puzzled 2p about the power of philoshophy,

You've made me think about this for a few days, which is good. Here
are two main reasons why I think looking at philosophy is useful to
engineering.

Firstly (as William Durant said in the foreward to "The Story of
Philosophy"), we need to remember that once something becomes
concretely implemented, people don't call it philosophy any more -
natural philosophy contributes to natural science, moral philosophy
contributes to law, political philosophy contributes to politics and
government. I have a good friend in robotics who thinks that one of
the main reasons people think of robotics as often a failure is
because once it's successfully mass-produced, it's not called a robot
any more, it's just (for example) a plane that can fly itself. Given
where we're at in computational linguistics (early), if we want to
gain insight from the way other fields developed, we're likely to find
a lot that was called philosophy before it was called science, let
alone technology.

Secondly (and I think your point about Husserl and ontology learning
exemplifies this), the things that don't work out practically often
remain the domain of philosophy. If you want to look for negative
examples, or to find good reasons why a technique won't work, the
first place to look is in the domain you're working in - in this case,
perhaps computational linguistics. Good reviewers will be quick to
point out if you've missed something important here. But the second
place to look is often philosophy (or at least, might be found in a
book with a philosophy shelfmark). In retrospect, it often looks clear
why some research programs were destined to succeed and other to fail
- but this kind of 20/20 hindsight is never fully available at the
time, but often philosophy is the nearest guide. So we might not read
Husserl to tell us exactly how to create ontologies, but we should at
least learn how not to create them.

OK, those are both of my cents at the moment!
Best wishes,
Dominic

_______________________________________________
Corpora mailing list
Corpora at uib.no
http://mailman.uib.no/listinfo/corpora