How much care do you give to the study of the exceptions?

I'm not sure if it's worth moving them to the back of the cards...

Memorize their readings with separate cards.

Fair enough.

I have a question about the exceptions notation used in the deck.

For example, does "-匋:tou" indicate that every kanji with the 匋 component is read "tou" as well?
Whereas "-谷" simply points out that the 谷 kanji has different (or non unique) readings...

Have I got it right?

The reasons I used for not including a reading are as follow:
1. It's a component already included in another card (you study the reading separately).
2. It's an extremely irregular component with many readings (only useful to mark an exception to the general rule), or
3. It's a single-kanji exception (no advantage in learning the reading this way).

Technically, for the exceptions that include a reading, you could insert a new card in the deck to study it separately.

Feedback is welcome. :-)

Thank you for the clear explanation!

The deck is extremely useful... It's nice to see so much care put into something.
I'm studying the cards with no exceptions first, as they provide maximum benefit through minimum effort.

I'm a little bit surprised that there aren't many people interested in this approach... I think it's really efficient. I love the feeling when I meet a new word in Core and I'm able to guess its reading!

I've updated the deck to remove some rare and some fake second readings of some radicals (due to mistakes in kanji dictionaries). For example, according to http://jisho.org/ 亥 is never KAI, 菐 is never HOKU, and 道 is only read TOU in the word 神道 SHINTOU.

I've also added a few more groups to the list that I had skipped last time, as they seem more common and regular than what I initially thought.

DrJones Wrote:I've updated the deck to remove some rare and some fake second readings of some radicals (due to mistakes in kanji dictionaries). For example, according to http://jisho.org/ 亥 is never KAI, 菐 is never HOKU, and 道 is only read TOU in the word 神道 SHINTOU.

I forgot to upload the version that got rid of those? Oops.

Sorry to bump this up.

Did you guys make any further progress on this?
I read somewhere (maybe from Vempele or DrJones?) that the deck that was uploaded some time ago is now outdated, and also that there is a new one which relies on visual similarity between characters.

Any chance you could share those new resources? I would really appreciate.
Thanks!

Bumping it too. This seems a pretty interesting study. Good work in finding and filtering all those phonetic components.

As for the anki deck, it would be good to have extra fields with examples where it works (not only exceptions), purity and number of characters covered. The ordering of the deck seems suboptimal, but w/o those last two fields I can't fix it.

I'm starting to add examples manually... but it is quite laborious... I also moved the exceptions to the back of the card.

EDIT: now that I saw that some cards have examples in their mnemonic field. But only a few of them, usually when the component has more than one reading.