I come from a python background. I typically write long unit test descriptions that link the test to a user story. The test is as much about justifying why the code exists and how it fits into the larger system as it is about testing for correctness. As such my tests often have about as much docstring text as test code.
it syntax. Something like Describe makeFoo - it should return a foo. I don't see many long descriptions. This works fine for TDD but I get less sure on where to document why the code exists in the first place. Why does makeFoo return foo seems more important me than if it does in fact return a foo.