Development of Creative Commons licenses

[Nancy Ide <ide AT cs.vassar.edu>]

Creative Commons licenses are causing us a big problem.

We are creating a 100 million word corpus of American English texts  
produced since 1990 (the American National Corpus -- http:// 
AmericanNationalCorpus.org), intended to be comparable to the British  
National Corpus of British English. We go to the web to find texts  
that can be included in the corpus--especially fiction, but also  
blogs and anything else we can find--often looking for Creative  
Commons licensed material. The problem is, the "default" CC license  
most authors choose includes "share-alike" and  "non-commercial use".  
These two restrictions make it impossible for us to include the data  
in the ANC, since (a) the ANC is used by publishers of dictionaries  
and English-as-a-second-language texts, who use the corpus to search  
for linguistic patterns and word usages, which is therefore a  
commercial use; and (b) the viral "share-alike" restriction means we  
have to let anyone else use the materials under the same license, and  
this would violate our strict usage requirements for researchers and  
commercial enterprises. Attribution, the other part of the default CC  
license, is not a problem for us, since information accompanying each  
text identifies the author and source.

Our experience tells us that many authors would not only be willing  
to contribute their texts to the ANC, but would be happy to know that  
their prose is being used to develop the next generation of American  
English dictionaries and other reference materials. Furthermore, by  
far the largest body of users consists of linguists and computational  
linguists, the latter of which use the ANC to gather statistics about  
linguistic patterns. No one actually reads the corpus (most uses  
involve concordances showing sentences in which a given word, phrase,  
or linguistic construction appear), so there is nothing to lose in  
any case. The worst that could happen is that a sentence from a text  
might be used in a future dictionary as an example usage.

Creative Commons licenses clearly don't cover a situation where the  
text may be used as a *research tool*, even if the goal is a  
commercial product. In fact, the structure of your information  
implicitly encourages the "attribution/share-alike/non-commercial"  
license.  Could there be a license or option that would provide for  
use as a linguistic resource?

In any case, if anyone has thoughts on how to reach authors who might  
contribute material to the ANC, even if they do use the default CC  
license, we'd appreciate hearing them. We have an on-line mechanism  
for text contributions, with a license, etc. that makes contribution  
easy.

Nancy Ide
Technical Director, American National Corpus Project