Yeah, you can do a Negative Binomial turned to Normal for this one. The keys used are the alphabet, plus !?.,;:' and space, meaning 26 + 8 = 34 keys. Assuming the probability of a key is the same as for any other key, the probability of a key being the right letter is 1/34. Based on the Negative Binomial distribution the mean number of keys before getting all the letters is: letters * (1 - 1/34) / (1/34) = letters * (34 - 1) = letters * 33.
Each letter is about 0.007%, so there are 14285 letters. I think the max speed is about 450 ppm, so about 7.5 keys per second, per monkey.
If we just use the mean, the expected time is 33 keys/letter * 14285 letters / (10 monkeys * 7.5 kps/monkey) = 6285 s = 104,75 min = 1 hour 44 minutes 45.4 seconds.
A lot of keys are needed to get the full text, so aproximating using Normal distribution is appropiate, we can use it to try getting a 90% probability. The mean of keys is: letters * 33. It's variance is: letters * 33/(1/34) = letters * 33 * 34 = letters * 1122. The 90th percentile is letters * 75.94 ~= letters * 76.
And so, the most time for 90 % of people will be 14 475.5 s = 241.25 min = 4 hours 1 minute 15.5 seconds.
That's a lot of waiting...