i agree with both to an extent.
i agree with brendan that labelling is vital to a language analysis. it makes no sense to discuss the effects of persuasive techniques without informing the reader what technique you are actually referring to, and can only result in confusion and a lower score.
however, i also don't believe that you should limit yourself to simply naming it them as puns, rhetorical questions and whatnot, as it does get extremely monotonous for markers. i suggest that you inject some variation into the piece by labelling them in different ways. for example, rather than saying that "the author has stressed the negative effects of this scheme with hyperbole", you can say "the author has stressed the negative effects of this scheme by exaggerating the truth", or labelling it more subtly through quotes as neophyte has mentioned. this will give a more sophisticated response.