I personally believe that while it is certainly true that CPAN is one of the major points of strengh for Perl and similarly important are the communities, you're stressing too much the accent on tools as opposed to the basic language itself, which is the starting point for the latter ones, and everything else.
For example, when I once gave a presentation myself about Perl I took a very personal approach, that perhaps was best suited to the environment I was talking to: namely a universitary one as opposed to the $job one you're most likely to refer to... However I started writing the proof a combinatorial identity in two ways: i.e. respectively with managing fussily all indices on sums and omitting (most of) them on assuming null values for certain quantities outside of certain ranges, which made the latter just as sound and as unambiguous as the former, but much more readable at the same time. I did so to illustrate an analogy for dwimmeries. Then one should necessarily talk about contexts. Talking about natural language principles in Perl, especially in terms of statement modifiers, or at least hinting at them would be very important to either get one intrigued about the language itself or letting her decide that it's "not for her." Also, I think that it would be wrong both to omit OO programming in Perl altogether and to refer in detail to any model in particular, be it the one extreme of the DIY "standard" which may scare some or the opposite one of, say, Moose. OTOH one should mention that both exist, and hint to the full spectrum of intermediate possibilities, but above all... now yes: the huge amount of CPAN modules, be they "procedural" (please let me pass the term for lack of fantasy finding a better suited one!) or OO or... whatever!