Reading A Whole Book in Anki - 三体

[markhavemann]

I had an idea a while back to use Anki to read books, but I couldn't find anything about it on the internet so I might be the first person who do this.

 

I've just finished read a full book (Jordan Peterson's 12 Rules for Life 《人生十二个法则》) only in Anki to see if it would work. It's actually been a much better learning tool that I imagined it would be, so I thought I would share it as I start my next book, 《三体》. Actually I've already read about a third already, but it was some time back and I thought I would start over.

 

The first thing I should say is that the reading part is linear and doesn't actually used the SRS part of Anki. There is no repitition, even though it's in Anki.

 

I've included the deck package for 《三体》 if anybody wants to try (is this ok? not sure if I'm breaking site rules here). I haven't used it yet or tested importing it so I'm not sure if there are any big issues, but I'll be happy to try fix them if anybody else finds this useful. 

 

 

三体.apkg活着.apkgankiImportFile三体II·黑暗森林.tsv

 

 

Why?

 

So here's my rationale for using Anki for something so far away from what it was meant for:

 

1. Bite sized bits of language - each card contains only a sentence or a phrase
2. Read on the go, and sync reading progress between devices easily
3. Looking up words is easy (copy and paste on PC, and Pleco is directly accessible from inside Anki on mobile)
4. Marking things you don't know, or want to check later is really easy, and also syncs between devices
5. Easy flashcards - keeping a sentence with a new word (or that you just want to review later) is so easy, and can just be moved into one of your other anki decks that you actually SRS with.
6. It's easy to set a daily reading goal, and those goals are nice and measurable which is motivating (I make it 20 cards a day, and usually do more than that)

 

How?

 

That's the why, this is the how:

 

1. Use a script to automatically split a whole book into flashcards (12 Rules for Life had 5500, Santi has 7000+)
2. Set the interval so high that you will never see a card twice 
3. At the end of every week or month, check all the cards that you have marked (starred), look up words, move to other decks, convert to another card type, whatever you want.
4. Delete all cards that you've already reveiwed (and haven't repurposed) so you never see them again
5. Keep reading

 

How it Looks

 

Each card has:

1. previous line for context (if you decide to keep the sentence, but it also helps to stop your eyes from getting confused when changing to the next card)
2. the current line
3. next line for context
4. Unique ID for the card, (book name)
5. progress through the book (this is card 2 out of 7533)

 

I think there is some other functionality put in here but I can't remember.

 

Anki Settings

 

These are the settings I use to make sure I don't have reviews the next day. I can't remember which ones are important. 

 

 

 

ankiImportFile三体II·黑暗森林.tsv

[markhavemann]

I noticed a problem with paragraph breaks not being shown (they should be indicated by a thick black line).

 

I've reuploaded with this fixed, as well as a deck for 活着 since I think it's easier to approach.

 

I'd love to include vocabulary cards which are shown just before you encounter them in each card or chapter. Not sure what that would involve though.

[alantin]

On 4/18/2022 at 3:13 AM, markhavemann said:

I'd love to include vocabulary cards which are shown just before you encounter them in each card or chapter. Not sure what that would involve though.

 

Now this is getting interesting. I read your original post a couple of days before and was a little skeptical about the idea, because I didn't think it would bring anything new versus reading the book on Pleco for example, but I think this idea of reviewing the vocabulary before each sentence or a chapter is very interesting. I would probably review the vocabulary before each sentence or at most a paragraph because a chapter just gives you too much time before exposing you to the word in a sentence.

 

Creating these vocabulary cards should be quite easy to pull off with a CTA script. Here is an idea on how to do it off the top of my head. You have the book separated into individual sentences in the anki deck already, so first export them into a csv and separate them into individual text files. I think there is a sample script already in CTA that reads a bunch of text files in a directory and does stuff on them. Modify that so that to read each text file and place each vocabulary item along with its definition into a csv file next to the file containing the full sentence. Then cat them all into a text file so that the vocabulary definition end up before each sentence and modify the lines so that they can be imported as a csv into a new anki deck (I personally use vim macros and excel for this sort of editing). Set the options in Anki import so that each unique line is imported only the first time and any duplicates are ignored.

 

What you should end up with is an anki deck with vocabulary cards before each sentence card where each respective word is seen the first time.

 

You could also add tags after sentences and words to later recognize them in Anki and change the types of the cards based on them etc.

 

A real coder would probably create a more elegant work flow to create these cards, but this is the kind of quick and dirty stuff that I do to bludgeon stuff I want into Anki decks. ? I'll probably try this out myself at some point if you don't pick up the idea, but for now I'm out of energy to do that. Please let me know if you find the suggestion helpful! ?

[markhavemann]

On 4/19/2022 at 2:53 AM, alantin said:

I would probably review the vocabulary before each sentence or at most a paragraph because a chapter just gives you too much time before exposing you to the word in a sentence.

good point

 

On 4/19/2022 at 2:53 AM, alantin said:

Modify that so that to read each text file and place each vocabulary item along with its definition into a csv file next to the file containing the full sentence.

I'd go a step further and also show many many times the vocabulary item appears in the book. That way, anything that only appears once can be mostly ignored, and just used to understand this time, but a little more effort could be made for items with say 5 or 10 occurences, possibly moving them to a separate deck with proper SRS review options.

 

I don't use CTA but I think it keeps track of the vocab that you know, is that right?

 

I guess that would be the basis of creating a cards that aren't massively full of unecessary items. If you're at a point where you are reading a novel, you probably know a couple thousand words at least. If there is such a list and someone wants to post theirs here, I'll play around with excluding those words from the vocab items if I get around to doing this before @alantin does.

 

[alantin]

On 4/19/2022 at 3:17 AM, markhavemann said:

I don't use CTA but I think it keeps track of the vocab that you know, is that right?

 

I guess that would be the basis of creating a cards that aren't massively full of unecessary items.

 

It does if you tell it which words you know, but for this it's dictionary, word parsing and LUA scripts are the useful part. Now that you mention it, I do have something like 10k words marked known in CTA so I could use it to filter out those words. B

 

But Even if you don't have any filter for it, the ten or so chapters in a book should already have between 80-90% of the unique words in a book. You could mark frequent words with labels like "appears-over-1000-times", "appears-over-100-times" or something like that and then just get rid of them. Or you could use a list of HSK1,2,3,4 etc words to exclude those words.

[ebj]

Is there any possibility you've done the same for the next two books in the series?

[markhavemann]

On 1/17/2023 at 1:57 AM, ebj said:

Is there any possibility you've done the same for the next two books in the series?

I have done it for the second book, I guess I'll only do the third one when I get to it in a month or two (reading on anki is just a sort of side then when I'm travelling and stuff and lately I haven't had much of that).

 

I've uploaded the second book in the main post at the top. The format is TSV so you can just create a new deck and import it, make sure to choose the same card type as the other cards (maybe called, "book", I can't quite remember).

 

For importing, the order of the fields should be: uniqueID, leading line, line, trailing line, positionInBook, tags

 

Let me know if you have problems getting it working, I can make a video showing how or just put it into an Anki deck file for you.