100% agree. If `pure input' people, like Stephen Krashen and Steve Kaufmann, are to be believed then output (like speaking and writing) would eventually come after lots and lots and lots and lots of reading and listening, but I'm a bit skeptical.
Idk who they are, but have they ever learned a second language? It's probably just a gimmicky claim that got them a bit of fame (or infamy), and they stuck by it because of their career. I doubt they can present evidence.
And I don't mean: here's someone who was exposed to a large amount of English, and can now say a bunch of three word sentences. That's cheating. No, I want to see a large, representative group of people that haven't had any feedback on their speaking and writing during their learning reach almost error free, near-native fluency for a totally unfamiliar language (i.e., not an Englishman being exposed to French or a Mandarin speaker to Cantonese).
oh yeah that was another thing with Rosetta Stone that drove me crazy - their speech recognition software was not good. I eventually started skipping the speech lessons altogether because it simply would not accept my input. Instead I practiced repeating after the example speaker during the other lessons, and speaking my own sentences out loud when I was "composing" them when working from grammar books. But, again, what I really needed was a teacher and a class.
On a different note, for the one language that I did learn and used to be relatively fluent in after 6 years of taking classes in school (Spanish), I can still somewhat converse in via writing, but I can't speak it at all anymore really. They're totally different skills.