Sure, you can definitely build a debug version by setting the target’s scheme to build/run debug, but when you build it, the target is a framework which is compiled object code that your app then links to, so no breakpoints since there is nothing in the workspace connecting the built framework and your project changes. I think that stepping through OpenEars in a debugger is probably not going to be a good approach for learning how Pocketsphinx works, since OpenEars is really designed for painless continuous-listening app and vocabulary building and that means it ships a drag-and-drop framework to keep things simple.
I think an elegant way to use your approach for learning Pocketsphinx would be to step into the Pocketsphinx executables for Linux, which are easy to build on (for instance) Ubuntu. I always keep an Ubuntu virtual machine (or several) around for when I need to get quality time with Pocketsphinx directly.