Corpora: grammatical relations

[eneko agirre]

Dear colleagues,

We are trying to extract grammatical relations from english corpora, and would
like to know whether someone has publicly available software for doing this. The
software could be fully independent of be applied to the output of a syntactic
parser.

We had thought of using a publicly available parser (Collins, 97) and extracting
the relations from there. Unfortunately the output of such a parser is just a
syntax tree, and some post-processing is needed to find subjects,
direct/indirect objects, differentiate complements from adjuncts and so on. Has
someone done this work before?

Thanks for your help.

Eneko Agirre
IxA NLP group
Basque Country University