With subdecks for each card type, how exactly does Anki gather cards of the same note?

Anki Scheduling
1

Sorry if this answered elsewhere, I haven’t been able to find it or figure it out myself.

I’m wanting to separate my language learning cards and try learning the listening cards (card 1) and speaking cards (card 2) separately. It seems like the best way to do this is by subdeck. Ideally I would like to study in this order:

  • Listening - new cards
  • Listening - learning and review cards mixed together
  • Speaking - new cards from the same notes as the listening cards
  • Speaking - learning and review cards mixed together

I created a Listening subdeck and a Speaking subdeck (and set the deck override in the card settings, nice feature). If I study from the main deck it would gather cards from notes in the way I want, but I can’t use the study order above IIUC. So I’d expect to study subdeck-by-subdeck, but I’m hung up on whether/how the Speaking cards will be gathered from the same notes as the Listening cards (bold text above). It’s not clear to me from the display order settings as described in the Anki Manual. What will Anki do?

When I create notes, I tend to add ~80 new cards, while the new card limit is 40. So Anki might select Speaking cards from notes that haven’t been studied in Listening yet, which I disprefer (I guess I could try it, but I’d rather learn both cards from each note on the same day).

Up until now I’ve just had all cards together in one deck.

For reference, current display order settings:

image
image923×313 16.3 KB

Feel free to suggest alternative approaches

2

You could suspend the cards that you add until you’re ready to learn them.

You could also use reposition for this.

1 Like
3

If you want your Listening study session and your Speaking study session to remain completely separate – yes, you’ll have to keep the cards in separate decks and study them one-at-a-time.

You have your New card gather set to “Deck, then random notes.” That all but guarantees you’ll get different Speaking cards than Listening cards, which is not what you want – Deck Options - Anki Manual .

Make sure you’re setting these Display Order options on the deck you click to study – that’s what controls that learning session. The sibling cards should already share the same New-queue number. In your case, since you’re clicking the Listening and Speaking decks separately, either “Deck” or “Ascending” will give you what you want.

1 Like
4

Thanks, that makes sense. I’d overlooked the new-queue number. I’m often adding notes with similar information sequentially - so having insertion order set to “random” should be enough to mix those up, right?

And setting new card sort order to “card type, then random” will show new Listening and Speaking cards in different order IIUC.

A slightly different question - I set review sort order to “difficult cards first” on the assumption that I’m more likely to hit ‘again’ on more difficult cards. If they show towards the end of the study session then there’s less time before the next showing. Don’t know if my assumption is correct though.

Now using:

image
image923×872 35 KB

1 Like
5

It should.

The sort order is the order the cards are shown to you in after they are gathered. In each of your two separate study sessions, you’ll be seeing a different card type, right? In that case, any of the Random options will work for you.

I don’t think that’s necessarily a useful metric for you. That “difficulty” isn’t how difficult you think the cards are, it’s using the “D” component of FSRS – based on your past results with a card, but also on your parameters, how many times you’ve studied the card, etc. I guess I’d say – all things being equal – you’re not doing anything wrong by picking that, but you shouldn’t see it as a guarantee that you won’t have lapses later in your session.

6

Thanks for the extra responses

Since I’ve split the decks and my review behaviour varies, I thought I’d try splitting the deck options for each and optimising the FSRS. The parameters for speaking look like an improvement (I think I wasn’t seeing many review cards soon enough), but the parameters for listening look a little extreme:

Speaking
0.5111, 2.3730, 6.3781, 14.8073, 7.2623, 0.6353, 1.4688, 0.0010, 1.5027, 0.1112, 0.9878, 1.9726, 0.0949, 0.3632, 2.3083, 0.0762, 3.0164, 0.3572, 0.4878

Listening
0.9123, 3.9633, 35.1183, 100.0000, 7.0656, 0.6099, 1.6274, 0.0010, 1.6840, 0.0000, 1.1482, 1.9395, 0.0884, 0.3196, 2.2749, 0.2138, 2.8818, 0.5367, 0.6856

If I hit [good, good] on a new listening card because I made the card in the past few days and it’s fresh in my memory, I won’t see it again for a whole month, which seems excessive to me. I feel like I’ve done something wrong somewhere. The 100 for w3 feels like a red flag to me.

Prior to this the FSRS parameters were optimised for a combined deck containing all these cards

7

The downside of splitting your cards into different presets is that it also splits up your review history, and FSRS has fewer reviews to consider. Click Evaluate in each of those presets – what’s your RMSE and how many reviews is it counting?

That is the max for that value, which can also be caused by there being insufficient data. If you look at Stats > Answer Buttons for that deck or preset, have you ever used the Easy button?