This is what the reaction word cloud looks like for the Rachel Maddow Show online community:
It turns out, it was easier to copy whole pages of responses and then delete out the names and time stamps than it was to copy each word one at a time. So I expanded the whole Facebook thread and copied the pages of comments from the original blog post and pasted into a Notepad document. I then used ctrl H to replace repeated unwanted text with blank spaces. Even using as much bulk cleaning of the text as I could it was still pretty labor intensive. I stopped when my eyeballs started to make sandpaper noises in my skull - but still, we have a bigger sample than the Washington Post/Pew poll: 1,796 words.
If you're doing the math on the difference between the number of replies and the number of words, a bunch were cut for being curses, ridiculous made up words, or full comments.
I assume the folks at Pew took only the most common responses to make their cloud. In the early goings I wondered if we'd have any pattern at all in our words as it seemed like everyone was giving a different response. As it is, some words, including my favorite of the night, Legerdemain, aren't even visible.
I didn't realize Wordle differentiates based on capitalization. "Sucks" would probably be bigger if SUCKS, Sucks and sucks were all counted together. Likewise for "crap." Meanwhile, the font I chose for this display doesn't make a distinction between capital letters anyway. Humph.
In the mid-to-upper right is the word "unerican." That's supposed to be un-American but the "am" was deleted when I was cleaning out the time stamps. As far as I can tell, that's the only significant casualty of that process.