When using the "Correct Recognized Text" feature, the UI is cumbersome and requires either hitting Tab repeatedly or constantly switching between the mouse and the keyboard.

The user should be able to cycle through multiple corrections using only the keyboard:
1. Press "Correct recognized text"
2. In the resulting dialog, the text entry cursor should automatically be placed in the "Recognized as" field.
3. When the user hits Enter/Return, it should be as if the user pressed the "Accept" button on the UI.
4. The next potential correction should appear, and the cursor should once again be placed in the "Recognized as" field. (GOTO 3)
5. If the user hits Escape, it should be as if the user pressed the "Cancel" button on the UI.

Okay, I can't help but editorialize just a bit: This is really basic UI design. Shouldn't this be pretty intuitive to whomever designed this?