The best defense against plagiarism is having a distinctive scholarly voice and point of view. By this mean that nobody else's prose, nobody else's ideas, should satisfy you. You will rarely spend whole pages summarizing other people's ideas, or presenting raw data. When you do, you will make it clear that that material comes from other sources, because you want to make it clear that that it is not yours. You will never confuse a sentence someone else wrote with one of your own.
Plagiarism arises out of a position of weakness. The student is trying to get up to a certain level of professional or trying to fill space in the paper. The senior scholar uses plagiarized verbiage as filler, without taking pride in every paragraph. Maybe he is overcommitted and has to churn things out fast and can't be bothered.
When I cite something I think is very brilliant, I make sure I am extra careful to give full credit. I feel self-confident enough to lavish praise on other scholars when appropriate, especially when their work helped mine along. When I quote anything, I am conscious of the style gap between quoted material and my own prose. I don't mean that I write better than anyone else, but that I have my own ethos of prose that will be distinctively different from that of other writers, good or bad. It also helps that I rarely agree with anybody else. (Just kidding.)
I think I have improved as a writer, but I look at some things I wrote 20 years ago and they still are more or less fine with me. I would change sentences, but I would do the same with sentences I wrote a year or month ago.
Plagiarism is somewhat more likely for more routine information, stuff that doesn't seem to belong to any particular person in the field. If I find a good statement that I think expresses a consensus view, I will quote that verbatim: it saves me some time, and I don't have to state my own views as though they were the consensus (they rarely are; I overestimate my agreement with other people many times.)
On the mechanical level, I make sure quotes go around a foreign phrase the second it goes into a word-processing document. It isn't even allowed to be there one second unattributed, without the protective cocoon of quotation marks. I never take notes paraphrasing something that I might confuse at a later date with my own notes from my own brain. I will paraphrase something in a word processing document with the source in front of me and provide the source in the act.