It makes so much sense to base output on outright copying of input, especially input you have read, to verify you heard correctly. This way there is no anxiety/wondering if I am saying the correct thing, or at least much less of it. If the input comes from a reputable source, which is most native speaker material, I can be confident that I am saying, broadly, the right thing. Of course, there are nuances of connotation, register, colloquialisms, etc. But this is a good method to ensure I am building a good foundation. Thank you AntiMoon and Khatz, as usual.