Here is an exceptionally interesting chart via Hot Air:
If I am reading this right, the three recessions that appear to be of the longest duration are the last three: the recession of 1990-1991 that arguably led to George Herbert Walker Bush being a one-term President (even though it had ended by the time of the election); the recession of 2000-2002 (including the bursting of the tech bubble in 2000 and 9/11/01); and the current recession, which began in late 2007 and accelerated with the crashing of the mortgage finance market in the fall of 2008. Why?
I would offer two potential theories, neither of which I am entirely comfortable with, but both of which I think have at least some plausibility.
First, our economy has increasingly moved from a manufacturing-based economy to a knowledge-based economy (or, less charitably, from an economy that makes things to an economy that pushes paper). But, as anyone who has worked in a large, "knowledge-based" company could tell you, there are a lot of "knowledge workers" who don't work very hard, who don't produce very much, and who are, therefore expendable. When a recession comes, then, it's easy to justify laying people off from knowledge-based companies, because everyone knows they are expendable anyway, and it becomes harder to justify hiring them back, because you don't need them to actually make anything.
Second, I would hazard the guess that unemployment benefits have been extended far longer in the past twenty years than in prior recessions. Thus, people who are out of work don't have to look for new jobs (or take jobs in a different field or with less benefits or at less pay), because they don't have to. Ironically, then, the government's remedy for unemployment actually creates more unemployment, or at least lengthens the period of unemployment for those it tries to help.
A third theory occurs to me: it may very well be that the length of recessions is contingent on the degree to which the government has distorted the marketplace. If relatively low (as in the 1950s), recessions will be relatively short, because the market will rebound. If government distortion (intervention) is relatively high (as it is now), recessions will be relatively longer, because the market will react slowly because of uncertainty about what the government might do next.