I disagree with the two existing answers (gerrit and Patric).
In mathematica writing, it is not necessary to put quotation marks around very short fragments of descriptive text where that text is the obvious and natural way to express the idea. For example, if Smith has written a paper whose main result is
Theorem. Every even number is divisible by two.
then it is perfectly acceptable to write
Smith [cite] shows that every even number is divisible by two.
without quotation marks and without clumsy rephrasings such as
Smith [cite] shows that all even numbers have two as a factor.
The significant intellectual contribution of the work you're citing is the theorem itself, not the obvious wording that they used to express it. As you say, anybody who understood the concept would probably choose to phrase it in that way, even if they'd never seen the paper you're citing. Mathematical writing would be completely unreadable if every phrase that had ever appeared before was put in quotation marks. After all, Smith was hardly the first author to talk about even numbers – are we going to accuse him of plagiarism for not acknowledging that the phrase "even number" is a quotation from somebody else?