I am one of the authors of BART, a modular framework for coreference resolution. BART came to be in the JHU Summer Workshop project "Exploiting Lexical and Encyclopedic Resources for Entity Disambiguation" as a joint effort, and it is being actively used and developed by multiple research groups.

The code for the parser from (Versley and Rehbein, 2009) is available in source code from bitbucket. This includes parts that are based on code from Helmut Schmid's BitPar (included with kind permission), and is available under similar terms as BitPar, i.e. for non-commercial/research purposes only.

The Open Corpus Workbench (CWB) allows you to efficiently store and query large (>100M words) corpora. cwb-python is a Python interface (similar to the existing Perl one) that allows you to quickly retrieve, e.g., a certain sentence, or the occurrences of a certain word.