Building a Large Annotated Corpus of English: The Penn Treebank

5.2 Future directions

1. Name early uses of the Penn Treebank.
   Training stochastic parsers and POS taggers, inside–outside re-estimation from partially bracketed text, using prosody to disambiguate parses, bootstrapping lexicons, and finding examples for linguistic theory with tgrep, a fast tree-pattern search tool.
   • Marcus et al., 1993, Building a Large Annotated Corpus of English (Computational Linguistics), §5.2
2. What limitations of the 1993 annotation scheme did the authors acknowledge?
   Its essentially context-free representation leaves many argument/adjunct relations unmarked (e.g. complements after a sentential adverb get “trapped”), and some category schemes are inconsistent. Users wanted richer, less skeletal annotation and explicit predicate–argument structure (logical subject and object), planned for the next phase.
   • Marcus et al., 1993, Building a Large Annotated Corpus of English (Computational Linguistics), §5.2