Regarding learning onyomi in isolation:
[edit - I think the chance of correctly guessing the meaning of compounds in newspapers is closer to 50%, even less.]
Regarding "The Movie Method" TM :-): I'd like throw the idea of after effects into your discussion of efficiency and effectiveness.
As we read more, we process complete words not individual kanji. But before that occurs, when we see individual kanji, I believe we have an unconscious "sense" of the kanji - the sum total of all previous associations we've made. Its a fuzzy mix of all the onyomi, the kun yomi, the Heisig keyword, the vocab we've learned and their meanings, etc. And that "sense" evolves with input. So I think it makes sense to keep this batch of input reasonably pure.
Adding lots of arbitrary (and sometimes elaborate) layers of associations such as English words, movies, etc. might end up being mental flotsam and jetsam in the future. Sure, that clutter will eventually go away (more time), but consider why one would want to add it in the first place ....unless it's truly necessary.
Add: I realize this caution applies to RTK as well. For me, the efficiency benefits of RTK outweigh the temporary problem of having distracting English keywords and weird stories associated with the kanji. At least the keywords, unlike movies, reflect the meaning of the kanji (for the most part) which can be useful. (Note: non-beginners can try using RTK with Japanese prompts to avoid the English keyword problem.)
Katsuo Wrote:This suggests that someone who has learned the on'yomi should be able to predict them correctly in unknown words about 90% of the time.Stats is a distant memory for me, but wouldn't this mean that in a 2 kanji word the chance of getting it correct would then be about 80%. That sounds high, but if you look at in terms of misreading 2 words every 10 words... (There was an epic battle on this topic last year which went nastily sideways. No need to go there, I just wanted to share a comment made (I think) by JimmySeal which made an impression on me).
[edit - I think the chance of correctly guessing the meaning of compounds in newspapers is closer to 50%, even less.]
Regarding "The Movie Method" TM :-): I'd like throw the idea of after effects into your discussion of efficiency and effectiveness.
As we read more, we process complete words not individual kanji. But before that occurs, when we see individual kanji, I believe we have an unconscious "sense" of the kanji - the sum total of all previous associations we've made. Its a fuzzy mix of all the onyomi, the kun yomi, the Heisig keyword, the vocab we've learned and their meanings, etc. And that "sense" evolves with input. So I think it makes sense to keep this batch of input reasonably pure.
Adding lots of arbitrary (and sometimes elaborate) layers of associations such as English words, movies, etc. might end up being mental flotsam and jetsam in the future. Sure, that clutter will eventually go away (more time), but consider why one would want to add it in the first place ....unless it's truly necessary.
Add: I realize this caution applies to RTK as well. For me, the efficiency benefits of RTK outweigh the temporary problem of having distracting English keywords and weird stories associated with the kanji. At least the keywords, unlike movies, reflect the meaning of the kanji (for the most part) which can be useful. (Note: non-beginners can try using RTK with Japanese prompts to avoid the English keyword problem.)
Edited: 2009-05-19, 4:01 pm

] The fellow who created the movie method, Alex, asks for feedback and has already made some changes and added clarification to parts.