Do you Use ANKI for Grammar? If yes, How?

[nfera]

@davideroccato

That’s why I prefer to review even only 1 card per day, but that card needs to help me integrate a grammar concept.

To get over the example I made of pattern recognition of ‘awake’ = ‘would’ve come’, the answer is not necessarily less cards. In fact, having more Anki cards actually reduces pattern recognition, because then you may have a few other cards which have ‘awake’ on the front. But it’s also got to do with exact design of what to put on the front. Some other solutions when you recognise pattern recognition include:

• Write cards that are very similiar to each other, so you need to consciously read the card first
• Edit the card to mildly change the sentence, like change ‘awake’ to ‘not sick’ or something in the example
• Just delete the card. You’ve gotten in a few reviews. Perhaps that has sufficed for your goals

Anki card writing is a real skill, and is not easy, if you are serious about it.

As I mentioned earlier, you can use the Anki spacing algorithm to schedule external, more dynamic exercises. Eg. Go to ChatGPT and paste in the following: Give me 5 cloze sentences to test my understanding of X grammar concept in English. Then open up a new tab with ChatGPT, paste in the prompt, and you’ll get 5 completely new sentences. It’s slower, but you have zero chance of pattern recognition.

But you don’t necessarily need to use Anki for spacing, retrieval, and repetition. You can just apply the concepts in, say, having one grammar session per week or one per month where you go over grammar exercises you found particularly challenging. Or for your weekly grammar study, tell ChatGPT you are an advanced learner of English and they should give 20 cloze exercises testing your grammar. ChatGPT will occasionally repeat similar grammatical concepts over the weeks and months, no worries. Or if you’ve already been through an advanced grammar book, just buy another one or two grammar books and go through them. Just jump around. Same concepts, different exercises.

[davideroccato]

Very interesting discussion, lots of ideas here, thanks.

I need to dedicate a bit of time to figure out what’s the best scenario for the time I have available for perfecting my grammar skills, outputting them, and remembering them in the long term. It’s really a question of figuring out what works for us, that is not too taxing on our mind, but at the same time gives results, and can be tweaked to increase the mental effort step by step.

[LeifGoodwin]

@nfera

@ericb100

You can probably do some google searches like “optimimum fsrs settings for language learning” or something along those lines and find some suggests

You just need to turn FSRS on. The default parameters are where you need to start and they are decent. Then every month or two, go in and click ‘optimise’ to update the paramters based on your data.

I was unaware of FSRS, and I have now enabled it. I’m not someone who likes fiddling with these things, so it’s good to hear that the defaults are fine.

@nfera

To get over the example I made of pattern recognition of ‘awake’ = ‘would’ve come’, the answer is not necessarily less cards. In fact, having more Anki cards actually reduces pattern recognition, because then you may have a few other cards which have ‘awake’ on the front. But it’s also got to do with exact design of what to put on the front. Some other solutions when you recognise pattern recognition include:

• Write cards that are very similiar to each other, so you need to consciously read the card first
• Edit the card to mildly change the sentence, like change ‘awake’ to ‘not sick’ or something in the example
• Just delete the card. You’ve gotten in a few reviews. Perhaps that has sufficed for your goals

This is a problem for me, as I mentioned in an earlier post. If I am not careful, my brain learns the two sides of the card as abstract patterns rather than semantic concepts. To complicate matters, many words will have multiple meanings, such as a literal meaning (raise a car), and one or more metaphorical meanings (raise one’s spirits).

My solution is to avoid simple terms such as la montée, le montage and dépister and use phrases and sentences such as la montée de la marée, le montage du podcast and Les forces de l’ordre utilisent un test salivaire pour dépister la consommation de drogues. The increased complexity of the text forces my mind to think, rather than pattern match, and the context aids recall in real situations. I believe that research has shown that learning in context, as opposed to rote learning, is more effective. Obviously the complexity of the sentence relates to the students level in the language, and I don’t normally use long sentences, unless they provide a benefit i.e. explaining usage in context. I wonder how that integrates with FSRS?

[nfera]

@LeifGoodwin

I wonder how that integrates with FSRS?

FSRS by default uses parameters which were calculated across a large dataset of many users. However, this is only the default. The real power of FSRS comes when you optimise it for your own data. You do this simply by going into the settings and clicking ‘optimize’.

Ideally, you should have separate decks (and assign a separate preset for each) for cards you expect to have significantly different retention rates. For instance, if you are learning two languages, each should be a separate deck (and preset), where you optimise your FSRS parameters for each. If you want to mix all cards together during your review session (which is recommended, as contextual interference is beneficial for learning), just make both decks as sub-decks and a third deck and study that.