nohat has listed lots of appropriate references. I'm a fan of Language Log and OED myself. I also think Google is a great tool, at least for comparison purposes.
I don't think a reference to one of these is necessarily important, though it lends credibility. More than anything, I think an appeal to one of these sources should provide the ability to find more information on the topic.
I also think we will find at least some situations haven't yet been addressed by these credible authorities. In these cases, do you reference academic papers if possible? How far should you attempt to answer a question on your own—can tweaking and reviewing a home-grown answer be a community responsibility? I would say yes, especially so in the case of grammaticality judgements, where native speaker intuition is infallible.
Subjectivity in areas of debate will be unavoidable, and permissible to some extent. E.g., "How do I pronounce x" would be fine, but "is this better than x" is obviously something to be closed. I think StackOverflow's rules would work fine: "Avoid asking questions that are subjective, argumentative, or require extended discussion."