Generative tradition, as far as I (an amateur) understand, revolves around the 'poverty of the stimulus' argument.
So I can understand why I clearly get the meaning 'she shelved her books' or at least the grammatical soundness of it. But I see no reason to believe the structure in (2) of , pg. 2 (image uploaded in  for quick view) where the NP 'her books' is assigned(i know assigned is not the correct term) 'on the shelf' in the structure.
Assuming the verb 'shelve' is derived from the noun shelf with its idiosyncratic meaning of 'putting an item on the shelf'. It might as well had been something absolutely arbitrary, like, 'taking an item off the shelf', or 'creating a shelf'. I guess this is a neat argument. In the article, it's taken that the sentence has an internal syntax that somehow assigns the structure (2) to it. Now yes, we have a very specific usage of the verb shelve in English and so on, but let's just suppose so I can come to a point.
In the same vein, in this same article , another sentence "it cowed a calf" goes through the same operation. "the cow had a calf" it says. I just wonder WHY? I understand cowed is some 'verb in past tense' with the 'colorless green...' argument, but why should cowed = gave birth to a calf? and why should the syntax have anything to do with it? Isn't it in the realm of semantics/meaning which is just arbitrary(like that of shelve)? UG alone is just not enough to account for such bizarre etymologies, not at least without context.
I hope the above description places my doubt clearly, even if the doubt itself is wrong(due to ignorance). I request some readings if possible. The way I've found the literature, it's a big fat mess and(without any formal linguistics training) nothing even came close to a neat sourcebook-style treatment of this tradition. The book itself of which  is a review is unavailable on the web, so there are other things I've been reading (like chomsky's mildly inaccessible papers and Adger 2003, Carnie 2012). At least trying.