Converting the Background of a single layer image (say, a new scan, or an image imported from a camera card) to "Layer 0" or "Layer 1" or "whatever-name-you-choose-for-the-layer" will assure that any area in that layer which is erased or masked will display transparency underneath (usually indicated by the checkerboard pattern).
CTRL + J simply copies the Background into a new layer. Erase or mask anything in this duplicate layer of the Background, and the original will show through. Also in this case, it will allow you to toggle the visibility of the original Background, so you can view the checkerboard transparency region(s) if they’re present, and if you want to. In a file that consists only of the image-as-Background you can’t make the image invisible. (This is in reference to PS CS2…dunno if it’s changed with CS3/4).
Besides all that, duplicating the Background is just a prudent way to go. Futz around all you want with the duplicate layer…if you mess up you know you can always go back to the original. Alternately, you could just choose to copy the image file at the OS-level and work on that, leaving the original untouched.
It’s just a good habit to get intoNEVER do anything to your original image data that you can’t recover from.