most of your ideas seem to center around learning, especially accelerated learning. there's also a meta post on hacking chinese. Why not combine these 2, and go for an idea that helps accelerate foreign language learning? There's this dude that owns the blog alljapaneseallthetime.com. He learned to speak Japanese fluently after 1.5 years with some very smart and saavy techniques. They include things like listening to any Japanese 24/7, even while sleeping...
Some of it is specific only to Japanese, but a lot of it is just incredibly smart. AntiMoon.com is also another good resource. The core message is to concentrate on input and understanding first...
The big businesses in this field: Rosetta Stone, Pimsleur, etc... they own this field.. but in reality, they all suck (well Pimsleur is ok, I guess). Their techniques is NOT how you learn languages if you want to be fluent.
And so many startups focus on connecting language learners through Skype or a community. Sexy.. cool.. but NOT how you should learn languages if you want to be fluent.
The way to learn is to be like a baby... just listen, listen, listen. Get used to the flow. Get used to the accent. Get used to the nuances... it's ok if you just do this for a month, 2 months or even 3... No stress to start speaking.. no fear of mistakes.... listen, and your brain will start making those synapses and connections. People say babies learn languages faster than adults.. that's total nonsense. It's because babies are learning 24/7 for 5 straight years before they start speaking a single word. If adults had the time to do that, they'll freaking learn how to speak Japanese or Chinese as well. Babies are NOT special.
How does this translate to a startup idea? No clue... ChinesePod.com does this very well for Chinese, so you basically should start with a single language.