If you have not shot the person yet I recommend recording them in front of a green screen. Then it is very easy to place behind them.
If the footage is already shot; You can create a mask around the part of the person where the text is and then key frame the control points so they move with the person. OR you can let Motion do it for you. This is covered in my Creative COW Apple Motion Training DVD. I have a sample of what I'm talking about at the end of my kinetic typography video tutorial.