And you make the classic mistake of faulty analogizing.
I totally acknowledge that it takes domain knowledge to come up with a good solution. However, I specifically ask this question because it looks likely to me that that doesn't apply in this particular case. An app that uses a hierarchical display of symbols to allow input of language seems to me like a natural, intuitive approach to allowing a user to generate speech. I believe if a few people were to iterate on an app for this purpose, they'd probably have come across this approach pretty quickly. In other words, I'm saying this is more an archetypal case of a broken software patent, rather than of a specialized domain invention that just happens to use computer hardware.
I could not have come up with the equation e = mc^2, but I believe I could have come up with the basic idea of a hierarchical icon display used to input language without years/decades of research. It would have been even easier for a speech pathologist to achieve that in this era of software and programmer abundance, even if PRC had never developed their system and gotten their patents.
(Of course, if I've misunderstood the scope of the patents involved, I apologize.)