EDIT: Part of the community has decided on a less catchy and more representative title than "Socially acceptable plagiarism (with regards to translation)". Let's run with that for a while. GRP.10.27 END EDIT.
The catchy part of the title refers to reusing words or ideas without permission, but without 'rocking the boat'. In mathematics, this is often done by proper attribution of the source, by naming the author/speaker, or providing other clues that the phrasing one has just used is not original.
If this were a discussion forum, I would invite answers that lists various forms of socially acceptable plagiarism; instead I prefer brief mention of such to the comments. This post is about the propriety and etiquette involving translations.
My situation is that I want to devote time to producing an English translation of some papers in German, and make the results available. Assume (although I may end up doing something different) that I place my efforts in a PDF and put it on some web page for public Internet access.
Primary question: What trouble do I get into by doing that?
Answers to this may depend on what things I did wrong, especially if I use certain words, phrases, images, parts of images, without crediting the source or asking for permission. So let's add to the context that the papers are from before 1980, the authors are likely unresponsive, and the source journals are likely discontinued, although online access to the sources are not freely available. I will list the original sources in the PDF, but the following question arises:
What trouble may happen if I do not contact some representative and ask for permission to use old material?
For material published in 1980 and afterwards, there is less of an excuse not to seek such permission or rights, and I would accept anecdotes that might lend guidance and are not obvious applications of common sense, but I am interested in the amount of effort someone in academics expends in order to reuse published material, especially in translated form. I know of a few examples in book publishing where more effort is made in getting such permission, however that appears more expensive than I feel the current scenario merits. This leads to:
What is "due diligence" in producing such a translation? Does it matter if the translation is provided gratis or for a fee?
I don't expect to sell access to the translation; it is my intent to make it freely available to any individual researcher. If someone else wants to put it in a book and sell that book, however, perhaps I could grant them such rights in exchange for a small monetary (or caffeine-ary) consideration.
Finally, let's assume I provide my own translation, except that for some small sections (possibly in a different language) I use someone else's translation of the same or related source. Let's say that I am concerned especially about a fragment that is (roughly) three paragraphs or about 200 words long. The answer to the following questions may be length dependent: if so, consider that I also have a 20 word fragment that is of concern.
How do I attribute this fragment? Do I use a footnote, or mention it in a preface? What is due diligence for making sure I can use this fragment?
This question is barely suitable for MathOverflow; I ask it here because the papers and output are mathematical, and the conventions in mathematics and mathematical publishing may not be addressed were I to ask these questions elsewhere. If they are addressed elsewhere, please provide a pointer to such material.
Also of interest, although I do not need the answers here, is if I have the same situation as above, except that I provide an interpretation (which is laced with my own perspective) rather than a translation (which attempts semantic fidelity and objectivity with respect to the source paper). The answers may stay the same, but it feels like a different situation to me. (A similar situation is mentioned in Mathematical etiquette: Rephrasing / restructuring a work, limited release (with attribution) acceptable? , which has some useful advice, although it does not involve natural language translation.)
Gerhard "Yes, It's About Jacobsthal's Function" Paseman, 2011.10.26