[Lexicog] Interesting lexical discoveries

Mike Maxwell maxwell at LDC.UPENN.EDU
Tue Feb 3 14:08:46 UTC 2004


Patrick Hanks wrote:
> 3. Collocational patterns seem obvious when pointed out, but are not
> easily called to mind. From this I conclude that social salience
> (what we actually do with language) and cognitive salience (what we
> think and say that we do) are independent variables.  If this is
> right, introspection is a flawed research technique -- downright
> misleading, in fact.

I would think that introspection can be used in collocation research in two
ways: thinking up examples, and figuring out the relations among given
examples.  As you say, the former is subject to omission.  But I suspect
that the latter is downright necessary.

More generally, I suspect any lexicographic technique, used by itself, is
flawed and misleading, corpus searches included.

There's an interesting comment on page 185 of Manning and Schutze's
Foundations of Statistical NLP:

    It is instructive to look at the types of collocations
    that a purely linguistic analysis of text will discover
    if plenty of time and person power is available so
    that the limitations of statistical analysis and computer
    technology need be of no concern.

They then show some collocation results from the BBI Combinatory Dictionary
of English, and further comment:

    Naturally, the quality of the collocations is also
    higher than computer-generated lists -- as we
    would expect from a manually produced
    compilation. [!]

On pg. 187, they mention work by Church et al on "a program of computational
lexicography that combines corpus evidence, computational methods and human
judgement to build more comprehensive dictionaries that better reflect
actual language use."

In sum, I suspect the key to in-depth lexicography lies in some such
combination of human and computer effort, exploiting the strengths of each
(including, if I may say so, human intuition and introspection).  J. C. R.
Licklider would have loved it.

    Mike Maxwell
    LDC
    maxwell at ldc.upenn.edu




Yahoo! Groups Links

To visit your group on the web, go to:
 http://groups.yahoo.com/group/lexicographylist/

To unsubscribe from this group, send an email to:
 lexicographylist-unsubscribe at yahoogroups.com

Your use of Yahoo! Groups is subject to:
 http://docs.yahoo.com/info/terms/



More information about the Lexicography mailing list