[Corpora-List] help needed: automatic extraction of syntactic alternations

[D.L. Theijssen]

Dear Corpora List members,

I'm currenctly studying the dative alternation in (British) English:
'I give him a book' vs. 'I give a book to him'. Since I want to do
this computationally (in a way similar to Bresnan et al. 2007), I need
a large set of instances. I have created a set of 915 instances by
employing the syntactic information in the ICE-GB corpus and manually
filtering the candidates found. This was a time-consuming process and
unfortunately, the data set is still too small for my purpose. Since
there is no time left for the manual collection of more data, I am now
trying to extract the instances automatically.

Around the year 2000, a number of researchers have attempted the same,
but for the purpose of automatic semantic lexicon learning (e.g.
Schulte im Walde 1998, Lapata 1999, McCarthy 2001). They mostly
evaluated the lexicon learned, not the extraction of the individual
cases. For me, both the recall (not missing too many instances) and
the precision (not getting too much noise) of extracting the
individual cases are very important.

Are you aware of research focussing on the automatic extraction of
individual instances of the dative alternation (or other syntactic
alternations)? Or linguistic research on individual instances obtained
automatically, for example by following the approach in Lapata (1999)?

Any suggestions are most welcome.

Daphne Theijssen


------
Daphne Theijssen MA, PhD student
Department of Linguistics
Radboud University Nijmegen
Tel: +31 24 3611134
Email: d.theijssen at let.ru.nl
Homepage: http://lands.let.ru.nl/~daphne

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