What are the conventions for corpus usage?

Question

Suppose I'm working with a language that doesn't have much data available; the largest corpus around is a translation of the Bible, which is free to read online, but is presumably still copyrighted by the translator. Or suppose I've found a corpus of books to scan in, but those books were recently published and thus still under copyright.

What can I do with that corpus, by law and by academic convention? For example, if it's freely available for reading (or if I've purchased a copy), does that also mean "freely available for running a statistical analysis on"? Can the results of that statistical analysis be redistributed, even if the original text can't?

I'm not sure if there's much legal precedent in this area, but I'm sure there's quite a lot of academic precedent in linguistics—corpora have to come from somewhere, after all.

EDIT: I'm asking this question from America, but information from anywhere would be useful!

Answer 1

I can only speak for Germany, and IANAL (I am not a lawyer). The situation is basically as follows:

• You can collect material from accessible sources (from the web, from radio broadcasts, from TV) and do analyses on that material
• You can do so within a closed collaboration with some collaborators including students and guests visiting your institution (they just have to sign that they don't take the corpus with them)
• You cannot legally redistribute the collected material without permission of the copyright holders (sigh)
• You can redistribute some materials that do not allow the reconstruction of the original (e.g., word frequency lists)

Some legally largely un-tested tricks I have encountered to circumgo the restrictions on redistribution:

• Tear the material in parts short enough to be not copyrightable and scramble the order of these parts (typical unit of the parts: sentences). You loose some research questions (e.g., cohesion) but you can do still a lot with this.
• Thin the material by creating voids in the material (typical: Alternating chunks of 100 words visible and deleted)
• Parallel and coordinated crawling: Communicate with your partners at other places which websites to crawl and make sure that you crawl the same content by exchanging checksums of the crawled texts.

For more information you can check Juristische Handreichung für die Geisteswissenschaften (DARIAH-DE Working Papers 12, 2015) (in German)