I wrote, and my company shipped, a free icon/text configurable speech generation system for the Amiga that does essentially what the iPad app in question here does at least two years before the date of the patent in question (the complaint dates the patents 1995 and 1997 -- Talkboard hails from 1993 and before, though I can only document it to 1993 -- that's the copyright date in the archive.) The application is called "Talkboard" and is still available from our company's historical archive.
Talkboard presented a layered interface pretty much just like the one in the iPad app, It used a synthesized voice and provided for unusual phonetic construction so as to obtain the best clarity (the Amiga's text-to-speech could be.... quirky.) You could load and save phrase banks, and one phrase bank could partially replace another or completely replace another. Single words or short phrases or entire complex sentences could be stored for 1-click or multiple click retrieval. The phrase could be represented by any shorthand that was convenient. It came with presets, but was really intended to be customized by the end user - what a kid has to say and an adult has to say tend to not be the same things in most cases. It could also be driven from ARexx, a system-wide scripting facility, and could dynamically change definitions based upon whatever criteria you needed it to.
As far as I can tell, there's nothing unique, new or even interesting in the two patent claims.
Hopefully that's of use to the EFF or the defending party.