This is very interesting, I'd be interested to see what this analysis would look like if you only looked at posts over a certain reward amount, and filtered out auxiliary verbs like "can, will, would, should."