I've just started to import this in to anki and review, so here's some issues I've come across so far.
First, thanks nukemarine and others for organizing it and to everyone who entered data!
As for anki, how is everyone arranging their cards? I'm doing english to kanji and vice versa, but I think nukemarine at lease isn't using the english? I imagine you have to go from english to kanji if you want to learn how to use the grammar rather than just recognizing it?
When generating recognition and production cards in anki, is it possible to ensure that they are ordered as recognition then production for each fact?
I noticed and fixed a typo in the 9th sentence in part 1. I know phauna has checked parts 3 and 4 but has anyone checked any other parts? Are people reviewing and finding errors but not fixing them in the speadsheet or did everyone just skip the first part?
I've found the best way to find errors in text entered from a book is to enter it twice independently, mechanically find the differences, then compare each difference to the text. That'd be nice if people were interested in entering it all again (!) but I guess most errors should come up during review, as long as the fixes make it back to the spreadsheet...
When fixes are made in the spreadsheet, how to merge them back in to anki? Anki doesn't have a merge function, but it should be possible to overwrite fields by using sqlite directly on the deck db. The UBJG# would make this easy as it's a unique key.
The revison control features in google docs are pretty lame -- there's no way to record a comment for the change and it's difficult to see the changes between revisions. I guess another part could be added to record overall change comments for each revision. For each individual change there is a "Notes" and/or "Comments" column in some parts being used for this. It'd be nice if every part had both and were used consistently with Notes for grammar notes (that you might want to see when reviewing) and Comments for comments on changes (and stuff that you only want to see when editing the spreadsheet).
I think it'd be useful if the grammar point was included with each sentence so you can check what you were supposed to be aware of when reviewing that sentence.
Nukemarine Wrote:I've got many of the hiragana only words converted to
kanji (hopefully the right kanji).
Kanji Text - Convert all to kanji if possible, leave as kana if
unsure. Numbers can be Roman Numerals. Include punctuations
Kana Text - Convert numbers to kana equivalent, include punctuations.
English Text - Generally what's done in the book.
(I guess he meant arabic numerals

)
I don't think all possible kana to kanji is a good idea, nor converting numbers to kanji. Converting kana to kanji for the missing (outside of JLPT 3 and 4) kanji is helpful for those who've finished RTK if not for those who haven't but just want to study for JLPT3. However converting numbers and usually kana words to kanji is harmful if you want to learn to write in the standard fashion. If there were 2 columns, one usually kana as kana and the other everything as kanji then one could be used for recognition and the other production.
Then there's a less common but more serious problem where the kana and kanji version of a word have different uses. This has happened at least early on in part one where うち has been kanjified.
Nukemarine mentions this, but some of the english sentences need adapting to make them stand alone. I think some people are doing this, but just early on there's:
006-01 私は朝はパンを食べます。I eat bread in the morning.
006-03 今日の朝も、パンを食べました。 I ate bread this morning.
The last sentence should instead be something like "I ate bread this morning too." for it to be possible to go from english to kanji.