copyright and text corpora

Dear Copyrightistas:

The following question has come up on the Corpora list, which is dedicated to folks who use large collections of text - corpora - for various kinds of research in linguistics.

  Suppose that I duplicate large amounts of copyrighted texts on a server, but only release extremely small contiguous samples (say, < 100 characters):

• for free,
• in response to research inquiries,
• as anonymous usage examples; ie. without reference to source or author.

  The canonical query is 'show me all +/- 7-word contexts that include the following word...' Obviously, this is very similar to the functionality of a Web search engine, which also shows small samples of (usually) copyrighted text.

  My assumption would be that if each individual citation is arguably "fair use" by virtue of size, research application, and absence of monkey business (eg. you can't reassemble a readable text), then the system as a whole is "fair use."

  Is there any case law, US or otherwise, that comments on whether the simple fact that the underlying texts are stored and searched electronically obviates what would otherwise be obviously "fair use" citations?

  Thanks in advance for any assistance,
  Doug Cooper Received on Wed Jun 18 2003 - 20:02:12 GMT