For the last three recessions, median income started dropping prior to the official recession and continued to drop long after the recession officially ended. In the 1990 recession, median incomes dropped for four years. In the 2001 recession, they dropped for five. Around 2009, they dropped for six.
The fascinating thing, to me anyway, is that median incomes dropped prior to the recessions, that from the perspective of the median family, the recession started years before being recognized by official statistics. Subject to the normal caveats about correlation, causality, and small sample sizes, you might even say that median family income drops are reliable indicators of future recessions.
It's also interesting that while many are wondering why it's taken so long for incomes to recover, the time till income recovery doesn't look all that different from the previous recessions. The biggest difference isn't the length of time till income recovery started, but the amount of income lost. In addition, the severity of the initial income drop was associated with a shorter time gap between the income peak and the recession itself. By the time we hit a recession in 2001, incomes had already been falling for two years.