In the old days, you could scrape text off of emulators (like the VT100). They were text based.
You can scrape text off of web browsers because they are text based (HTML).
But a typical windows form is a rendered graphical image. You would have to resort to OCR. That's a pretty big can of worms you don't to open.
You might be able to use the old Windows Get Text message. You would need to get the main window handle and then iterate the child windows.
The windows message is WM_GETTEXT.
http://msdn.microsoft.com/en-us/library/ms632627(v=vs.85).aspx[
^]
I would google WM_GETTEXT and whatever language you are using.