Answer to the question: no it is not plagiarism. Plagiarism is defined as passing off someone else's text, idea, representation as your own. As the author in question is actually the author of the old paper, it is not plagiarized.
If the author signed over the exclusive copyright, it is a copyright violation.
If your conference requires contributions to be novel (but see below), then it fails on that account - regardless whether the abstract is copy&paste or rephrased (and even in that case you may report that the citations are not appropriate, and previous work is missing).
In addition, as reviewer you may decide that it is not interesting enough (e.g. because it is known already) even if there is no formal novelty requirement.
Personally I find self-plagiarism a misleading and unfortunate term (but yes I do know what it means, and I totally agree that we need to fight the underlying problems). But IMHO is not a straightforward extension of plagiarism, and in my experience, there are better terms to unambiguously name the problems.
In addition, the term has the clear connotation of an offense - but whether and how much one should cite oneself depends a lot on circumstances. In particular, not saying that this your idea is 5 years old is offensive (only) if there is a requirement or expectation of novelty. On the other hand, one would expect a lifetime-achievement award lecture to be full of widely known ideas.
But even in a paper I'd not cite all my previous papers that are somehow relevant but only the one (or maybe two) that is of most use to the reader. Excessive self-citation isn't better than not citing yourself - it is just a different problem.
I thus find it much more practical to require a contribution to be novel (no duplicate publication) and substantial (no salami publication - though for a conference presentation less is often more). There are further requirements (outside the self-plagiarism questions), e.g. the contribution must be the authors' own (no plagiarism).
With students if necessary for the purpose I clarify that novelty includes written from scratch. In addition, even if that is independent of self-plagiarism, I make people aware of the fact that they can violate the copyright of somehing they hold the (in my legislation inalienable) authorship rights to.
(I do think though, that academia would be better off if the novelty requirement would be somewhat dropped in favor of replication studies)
I'd like to throw in a slightly different view from a field with presumably very different conference culture, which I'd formulte as
When reviewing conference abstracts, I ask myself:
- Overall quality (as far as one can tell from 200 words)?
- How well does it fit with the interests of the audience?
- I rather disregard "global" novelty compared to novelty to the audience at that conference.
Our conferences do not have novelty requirements and neither do they ask for an exclusive copyright transfer (unlike our journals). Ultimately, if the topic is too old and well-known, it will fail at the "interesting?" question.
Long story:
we submit 200 words or maybe 1 page abstracts - so 1.5 copied pages are plain impossible. More importantly, our conferences don't count as peer-reviewed publication, for that we publish in a proper journal (we often don't even have conference proceedings but instead have a special issue of one of the reputable peer-reviewed journals). In particular with my medical colleagues I see a trend that they are concerned someone could steal their idea - so they will present only results that have already been published in a peer-reviewed journal (pretty much the opposite of "novelty").
For me the important point of a conference presentation is to tell the audience something interesting for them. I freely admit that I hate presentations that don't convey any useful message to anyone in the audience beyond "I, the author, am a hero" or "This [totally useless crap] is sooo new".
If there is already an accepted paper, I include the reference into the conference abstract, for several reasons: the reference serves as "topic has passed peer-review" tag and of course it is an additional advertisement for the paper. The propriety* of citing oneself is only a minor point as we don't have strict novelty requirements; and the lower the number of allowed words, the less usual it is.
Due to our few 100 word limits, I'd also never look for much originality in a conference abstract: once you've found a formulation that saves 5 words, you're not gonna give it up easily ;-)
As I work at the interface between some disciplines doing statistical data analysis for chemical data and medical diagnostic problems, I attend conferences with rather disjunct attendees.
Consequently,
- I've been presenting the same software (poster) at three conferences (within 2 months) with a total of probably close to 2000 attendees but an overlap that I estimate to be < 20 people. I submitted it to those conferences because I thought it interesting for their respective audiences and I did not expect many people to attend more than one of them.
- I'm invited to speak in more detail about a topic that I first presented 3 years ago at a chemometrics conference. (The paper was published 2 years ago and of course I refer to it: after all I want people to read it - but not to the previous conference) Again, I am the overlap in attendance between the statistics-heavy first conference and the application-centered conference now.
And, by the way, I was taught about scientific writing to avoid synonyms and always try to stick to the same terms (once good terms are found) - even if that is repetitive throughout the paper, and use easy language in order to be unambiguous and as understandable as possible to other non-native English readers. This is another source of not so very original formulations.