Automatic Mapping Among Lexico-Grammatical Annotation Models (AMALGAM)

THE MULTI-TAGGED CORPUS

Using the AMALGAM multi-tagger a corpus of 180 sentences was created. The "multi-tagged" corpus consists of the following texts:

• The Industrial Parsing of Software Manuals (IPSM) text (60 sentences)
  These sentences were selected for parsing during the IPSM'95 Workshop (International Workshop on Industrial Parsing of Software Manuals, May 1995, University of Limerick, Ireland).
• The Lancaster/IBM Spoken English Corpus (SEC) text (60 sentences)
  These sentences were extracted from the beginning of the SEC.
• The Corpus of London Teenager (COLT) English text (60 sentences)
  Sentences selected from the COLT corpus.

The texts were tagged (raw output from the AMALGAM tagger, including errors) using the Brown, ICE, LLC, LOB, UNIX Parts, POW, SEC and UPenn tagging schemes.

The tagged texts were also proofread and edited by human experts in order to remove any errors made by the AMALGAM tagger.