The Second Release of the American National Corpus contains over 22,000,000 words of written and spoken American English, annotated for lemma, part of speech, noun chunks, and verb chunks. Part of speech tags using the Penn tagset are included for all data in the Second Release, and many documents are also PoS-tagged using the Biber tagset.
The ANC Second Release is available for research and education for a nominal licensing fee from the Linguistic Data Consortium. Commercial users can obtain the corpus and gain rights to use it in commercial products by joining the ANC Consortium. Please consult the LDC Catalog entry for the ANC Second Release.
The First and Second Releases of the ANC include materials which have been acquired to date, and therefore the current release of the ANC is not balanced. There has been no hand-validation of the XML tagging or the annotation. Headers are typically minimal, although most contain complete information concerning domain, subdomain, subject, audience, and medium. Check the list of known bugs and caveats for a description of the limitations we are currently aware of.
One of the aims of the Second Release is to get feedback from the community about its structure and annotation, so that modifications can be made, if necessary, for the final release of the full 100 million words. We therefore invite comments and bug reports from the community of ANC users. Please contact anc@cs.vassar.edu.