I've been coding in perl for more than 10 years, and recently picked up python to do natural language processing. So I am new to both python and NLTK. I started off with the popular "Natural Language Processing with Python" by Bird, Klein, and Loper. I found it hard to follow because the dots were not connected. In that book, I'd read an example, and I *thought* I understood it, but coding it was a different story because there was no transition between the different sections. So I had to go to github to find random sample code using the tools I was interested in.
Then I picked up NLTK 2.0 Cookbook. It doesn't spend much time explaining concepts, but I find I can type the code in pretty much word-for-word (and replace it with my customized needs), and the code works. I find it much more useful, and faster to code using this book as a template. It's limited in the recipes it gives, but provides the main NLTK recipes, so I'm quite happy with it.
Since I had already picked up the Bird book, I knew exactly which recipes I needed. If you don't know which chapters are pertinent to your needs, I could see it being difficult to know which recipes to use because there's not much explanation and the terminology is different if you're new to natural language processing. The terminology and concepts are better explained in the Bird book.
A good book. Worth checking out some sample pages on Amazon to see if the recipes will be useful to you, and if you're finding difficulty implementing what's in the Bird book, this one provides better instructions.