I don’t personally like SRS as an algorithm approach but repetition certainly exists in natural learning and my version is a simplistic recording and occasional revisiting (and re-re-visiting of resources). I don’t like the thought of a program doing this for me because I think my desire for repetition of resource is driven by its usefulness to me (how does the software know if a word is dud one and not actually worth learning in the context of the real language I am encountering?) Of course you address this Edwin by emphasizing the importance of how you feed the system.

What I am trying to do is use artificial tools to recreate the benefits of natural learning.

To be fair in your TV method you are doing the same Keith, the massive input may be natural but Video isn’t. This is probably what leads to the “attention” problem you have identified recently, if you see people preparing and cooking a meal for example in real life and you don’t understand the language you have a vested attentiveness interest if you know you will be eating the end result. Seeing the same thing on a Video won’t spark the same interest.

I think with the right tools and a reliable way to feed input I could use SRS and still feel that I was a natural language learner.

If there are only 2 categories, “structural” and “natural”, and nothing in between, I think SRS falls into the natural side.

I think this is like trying to categorize the ‘natural’ products sold on the street. They are extracted from natural sources (or at least claimed to be), but they are processed. Are they really ‘natural’?