Comparing linguistic interpretation schemes for English corpora

Project AMALGAM explored a range of Partof-Speech tagsets and phrase structure parsing schemes used in modern English corpus-based research. The PoS-tagging schemes and parsing schemes include some which have been used for hand annotation of corpora or manual postediting of automatic taggers or parsers; and others which are unedited output of a parsing program. Project deliverables include: a detailed description of each PoS-tagging scheme, and multi-tagged corpus; a “Corpus-neutral ” tokenization scheme; a family of PoS-taggers, for 8 PoS-tagsets; a method for “PoS-tagset conversion”, a sample of texts parsed according to a range of parsing schemes: a MultiTreebank; an Internet service allowing researchers worldwide free access to the above resources, including a simple email-based method for PoS-tagging any English text with any or all PoS-tagset(s). We conclude that the range of tagging and parsing schemes in use is too varied to allow agreement on a standard; and that parserevaluation based on ‘bracket-matching ’ is unfair to more sophisticated parsers.