The way I look at it, the rules state that visible enemies within the specified distance close, and then the warrior wallops them.
The rules don't specify why those enemies close. The flavour text suggests a narration, but the rulebook already tells us that flavour text is just suggestive.
So then we can either insist - in a way that the game rules don't require us to - that the flavour of the power must be the same every time, even if that produces incoherent fiction.
Or we can do what the game rules permit and even encourage us to do, which is to allow the narration/flavour to vary with the situation, maintaining as constant only what the game requires, namely, that when the player uses this power, the PC's visible enemies within the specified distance close such that they can be walloped.
To me it seems obvious that the first is a silly approach to the game, and the latter sensible. Among other things, as I posted, the second helps produce a sense of dynamism, disrupting the tendency of turn-by-turn resolution to imply a stop-motion world.