It looks like working with text and listening to speech requires the same language-processing brain circuitry, and it's not optimized for parallel processing.