What would it look like if Shakespeare had written Huckleberry Finn, or H.P. Lovecraft had written Harry Potter? I'm going to see if I can find out! There's a recent algorithm that can re-draw pictures in the style of a given artist ( http://arxiv.org/abs/1508.06576 ), and I want to see if the insights it gives us about images can transfer to understanding text.
you might want to look at https://github.com/ryankiros/neural-storyteller , in particular he 'style-shifting' operation that allows their model to transfer standard image captions to the style of stories from novels.
#72 might also be relevant. That uses word2vec rather than the recurrent neural nets or deep convolutional neural nets, but you might be able to borrow some ideas. There's also this RNN implementation: https://github.com/karpathy/char-rnn (as decribed in The Unreasonable Effectiveness of Recurrent Neural Networks) that you might find interesting.