Yo, folks. This post is for a few answers from commentaries.
The link for the yahoo dictionary is http://dic.yahoo.co.jp/
Beware. It was 4 search options. The one you are looking for when you search for sentences is 和英, that is, japanese -> english.
About the number of sentences, counting only the sentences I kept in the SRS, I have 232 sentences. The tae kim pack had about 800 sentences and I already had gone trought 60% of them.
About on how to learn the readings, I believe there are now 2 choices.
1) Learn it from the sentences, and doing rote memorization for the readings.
2) Learn it from the sentences, but use some kind of mnemonic for the readings.
I'm tending to pick the number 2 because the success of RTK.
The method I plan to use is the DrMovie method. Or better said, a variation of it,
because I already did RTK and because I know a lot of japanese already from previous studies.
I'm already adding places for the readings I already know. From that little I've noticed that:
1) Adding a very familiar place for a familiar reading makes that place very strong.
2) Recognizing a learnt word from a sentence while watching anime/doramas makes that word really hard to forget.
3) For familiar words relearnt in the sentences is very easy to put the readings in their respective places.
So, operationaly speaking, these 3 things make mining sentences from anime scripts a nice option.
Now, alyks pointed me this post:
http://blog.feedmejapanese.com/2008/07/08/collecting-sentences-or-learning-japanese/
The sentences from yahoo are awesome, they are very rich, but they are not familiar, and they are not fun. They have no context.
There is one more thing I noticed while watching Macross Frontier. The series is great but there is an special point about it. It has many songs. And fansubbers are getting crazy lately and putting karaoke subtitles in everyone of them.
So, my next try on mining sentences will be on these karaoke subtitles.