I'm sure Chocky has given his consideration to porting some decent toolkits to RISC OS, if he or someone else does it, fantastic, but I think to port QT would be a colossal task.
You're right about the languages thing, you always get people banging on about how great BASIC is, and don't mention the enormous advantages of langs like Java.
I think RISC OS could attract coders a while ago, say Sibelius, for the size of userbase, RISC OS has had some good software written for it. But these products have either been discontinued or fallen behind comparable products on other platforms. I used to write for RISC OS, but now I have savoured the delights of other platforms, I doubt I'll ever code for it again, RISC OS is nice to use, horrid to program.
Like you say, as the userbase diminishes, there is less money to invest in development, less money == less good products, and products have to get more expensive, further reducing the user base, and it's attraction to outsiders. It's a vicious circle which will not be escaped with the current activities of major RO companies.
If you say that all serious users are on RISC OS 4 or above, then I think puts the userbase at under 5000, which can't sustain any serious software development unless you're extremely keen to support the platform regardless of earnings.
It's sad, I had personally hoped that RON may provide new users, but it looks like that is not to be.