Is it worth it to make (some) Native -> TL flashcards? Or to hide context?
So getting the preface out of the way to avoid 30 comments saying the same thing:
- this is not as a main source of flashcard learning (which are and will continue to be predominantly TL->N);
- this is not a replacement for output with actual human beings;
- this is not a blanket "should I, for every flashcard I have, make a reverse one"
So I've been using readlang to read articles and books for a while. This has made me a huge stack of words and phrases, and recently I picked up anki again and made flashcards out of them. It's going great. I can read articles from Canon van Nederland which are about 600-700 words long and translate 10-20 words/phrases per article. The site claims the versions of the articles I use are at a C1 level (bonus: if you're learning dutch it's a great site - they have settings where the same content is available in levels from A2 up to C1)
However, I then reach the common problem that my input (mostly reading) is both far stronger than my output (mostly writing) and also that I'm suspecting I am picking up context more than actual words.
For instance, I have a card:
> Hij keek weer naar de Hokkie-robot, die onverzoenlijk stil en doelgericht voor hem stond, met zijn slagstick in de aanslag geheven.
I know that onverzoenlijk here means unyielding, but I will look at onverzoenlijk and go "uhhh well it's a negative (on) and it's an adjective..oh it says Hokkie-robot, so because I've read hitchhiker's guide 500 times it's going to be unyielding silence".
I probably would not be able to tell you without the context. I think if I saw it in another context I don't know if I'd pick it up. I definitely could not tell you that the Dutch for "unyielding" is onverzoenlijk.
So I'm thinking of making some flashcards that prompt for output. I know this is usually a bad idea (you don't want to reinforce that TL->N->TL translation loop), and it certainly would not work for cards with longer phrases (e.g. emerged from the smoke curtain - my card for this uses dook uit but I could also produce it with kwam..uit I think).
I don't think I can make a pictorial representation of a lot of these sorts of cards (you're not looking at a picture and thinking immediately that it is representing unyielding unless you know you've used that specific image..).
So, two questions:
- Is it a bad idea or a harmful idea to make cards for words or phrases without context? Or should I be replacing all my contextful cards with hidden context until I show the reverse side? (and then being more lenient about words with many uses)
- Is it a bad idea to make output cards for singular words and short phrases where there is no ambiguity? Should I trust the process that output follows naturally from having a confident level of input, and my problem is that I just don't do enough outputting?
EDIT/UPDATE: Well, I tried going through my cards to pick out potential words for making output cards, made about 50, then encountered 6 different translations for "approach" and realised it was not going to be a good idea.
However, wrapping the sentence context in an Anki hint field so I need to click it to see the context seems like it has legs!
UPDATE2: oh god hiding the sentence by default is absolutely going to kill my stats in the short term