I've just started cataloguing my library at LibraryThing, something I'd describe as last.fm for books and without a Winamp plugin. It's a slow process: I'm away from my regular library, and the books I've taken on holiday all seem to be fairly strange. My copy of Something Happened (thanks Trav!) either has the ISBN printed wrongly, or the Library of Congress have the wrong ISBN, and I should really use the check digit to see which. I'm proud that it's the oldest hardcover edition listed though. I had to enter the details for my copy of Steppenwolf manually - it wasn't listed at all.
They have a feature (recently written about in Language Log) that lists 'unsuggestions'; books, according to the collected data, that are antithetical to those you are interested in. There are some truly, truly dreadful books out there - the challenge is to find one, a 'seed', that generates a fantastic library. Unfortunately, too few people own Who Moved My Cheese? For Kids for it to generate unsuggestions, but you get the feeling it would work quite well. I found At Knit's End: Meditations for Women Who Knit Too Much which generates a very interesting library, even if it's a little uneven. GEB:EGB, a bunch of Asimov, Clarke, Wells, Pynchon, Vonnegut, Burroughs (Junky), Nietzsche, Phillip K. Dick, Goethe, Camus, Robert A. Heinlein - even Niven's Ringworld. Of course, there's a few Tom Clancys (I want to write Clancies) in there, and C. S. Lewis really doesn't do it for me, but altogether a pretty good mix.
Let us know if you find something worse and thereby better.
Well, it's more of a one way distinction. If you see an author consistently use 'an', it's a pretty clear indicator they use the initialism. But, I think that the other case really isn't such of an indicator: too many people don't know that 'an' isn't just for written vowels.
The ANSI standard is the initialism. The Google count is 1,170,000 for the acronym - the first page is full of Microsoft Knowledge Base articles in line with their official policy for their SQL Server product. There are 969,000 hits for the initialism. Disappointingly, it looks like I might be in the minority.
Speaking of 'an', apparently a newt was originally an ewt; the story is that the 'n' shifted at some stage. No time to fact check it - it seems pretty far fetched? It would bring the name closer to eft and the like. OED lists newte going back to the 1400s, although ewte is mentioned in the same attribution, so who knows.
By the way, why is it pronunciation, but pronounce? Lord knows. It seems that's just the standard these days. OED attests it either way for both of them. There's even pronountiation. Pronounce, pronouncing, pronunciation. There's also pronunciating, but who knows where that falls. The Latin had just the u, but it's filtered through the French who, at various times, seemed to have just-u, just-o and both. I subconsciously conflate them all - I even seem to pronounce pronunciation pro'noun-ci-a-tion. Ugh.
dominic@doublemu:~$ uptime 15:23:04 up 128 days, 10:46, 1 user, load average: 0.00, 0.00, 0.002 ^ 7. w00t.