This probably depends on field, but in Computer Science it is common for people to publish similar material in workshops, then conferences, then journals. Each time it is acceptable to reuse some material from one level to the next as long as there is some new material (journals sometimes will publish verbatim but that's anohter matter).
So, is it acceptable to borrow significant text from one conference paper to use in a second conference paper? In my experience the answer is no, but I'm curious if there are any guidelines, and if so, whether they vary across disciplines.
BTW, after posting I came across this: Attitudes towards self-plagiarism ... certainly related to this. I commented there, and should have mentioned here, that a complicating factor with regard to reuse of text arises from the use of blind submission: if you can't say who you are, you can't admit to copying your own text.