First, "self-plagiarism" is probably a misleading term. I realize that it's often used humorously, but I think some people might take it seriously, and think that "self-plagiarism" really is kin to "plagiarism." It's not: The harms of plagiarism are that (1) it deceives the reader into believing that your work is your own invention, when it really isn't, and (2) it wrongly denies credit to the people whose work you’re copying. Neither of these applies to copying from yourself. (In this I agree with InstaPundit.)
Second, there are problems with copying from yourself, but it's also easy to deal with those problems. Setting aside the purely copyright law questions that arise if you've given away or sold the copyright, they are:
Copying from your past published work may make a reviewer -- for instance, a law review editor deciding whether to publish the work, or a professor grading your paper -- think that your work is novel when in fact you've already said it before.
Copying from your past published work may annoy readers who have read your old work, are now reading your new work, and are slowly finding that they've wasted their time rereading what they'd already read before.
The way to deal with these problems, I think, is simply to drop a footnote at the start of the discussion -- or, if you prefer, have a sentence in the text saying something like "As an earlier article of mine discussed . . ." -- that makes clear that the following paragraph, section, or whatever else is adapted from an earlier work. This signals to the reviewer that this material shouldn't count as original, and it signals to readers that they can skip this section if they read the earlier piece.
There's no inherent need to paraphrase your original words (if you edited them well the first time, they're probably the best words for the job), or to put them in quotes (which would just be needlessly distracting). You might still want to reword the material if the context is different enough to justify such alterations, but there's no ethical obligation to do so.
Of course, if the whole article is just a version of your earlier piece, then the introductory warning may signal to the law review editor and to readers that the piece isn't really novel -- which is only fair, and which discharges your obligation to them. (The editors might not mind this, for instance if you're adapting a piece from a law review into a communications studies journal, or if you're publishing a trimmed down version of a long piece; the important thing is that they realize that this is going on, and get to decide whether they want to publish the article nonetheless.) You should also signal this on your c.v., so that people who are counting your publications for various professional reasons realize that they shouldn't double count. On the other hand, if only a small part of the article -- say, the background section -- is borrowed from an earlier piece, then there's no need to stress this on the c.v., and law review editors probably won't be much annoyed by the duplication.
Third, all this refers to scholarship being reused as scholarship. If you write an op-ed (or a blog post) based on your scholarly work, I don't think you even need to point this out (though you might want to, since it might make your op-ed seem more credible). Editors expect op-eds not to have been published before in newspapers, but I don't think they expect that the material is novel in the way that scholarly work should be novel; I suspect they actually like it when the op-ed is based on published scholarly articles. And virtually no readers are likely to have read the original scholarly piece, so you don't need to worry about wasting their time by giving them something they've already read.