I realized that at Requests for new languages [1] we have a number of proposals for projects in moribund languages [2]. In brief, when roughly less than 1000 dominantly older persons speak one language, this language will be dead when those speakers die. Even some larger languages [than mentioned ones], like Lower Sorbian [3] is (with ~15.000 of speakers) are deeply endangered and it is almost predictable that this language won't be alive in the next century. But, cases like Lower Sorbian one is -- are border cases -- and I don't see a problem with creating such project inside of the standard procedure.
However, we have some number of cases where project is requested for a language with less than 100 older speakers.
My proposal is to do the next in the cases of moribund languages: * Reject proposal for project creation. * Suggesting them to put their language corpus at [multilingual] Wikisource. * Allowing them to work on Incubator if they really want to spend some efforts on language revival. * If a project at Incubator shows possibilities to be a live one, they may ask for project again, when they will have to pass all necessary steps (localization of MediaWiki and so on).
This is a kind of a "political issue", so I prefer to see discussion here before discussion at Language subcommittee.
[1] - http://meta.wikimedia.org/wiki/Requests_for_new_languages [2] - http://en.wikipedia.org/wiki/Language_death [3] - http://en.wikipedia.org/wiki/Lower_Sorbian