jeff_lindqvist wrote:luke wrote:There is also the assumption that someone can learn 500 words a month and keep that pace going for almost 3 years. Those with Anki vocab experience know this projection may be optimistic and unrealistic.
It would be an interesting challenge, though.
500 words per month means ~16.66 words per day. Perhaps there is a decent deck out there, with exactly 16000 words... Assuming they are all introduced at the right time, learned equally easily (to the level where you'd select "Easy" on the second run-through on the first day you see them, and every consecutive time you see them), you use the default interval settings, and most importantly - never miss a single day - how many words would you have to review on the final day? (It's probably dead easy to calculate this, but I'm too tired at the moment...)
I suspect it's not quite as easy to simulate as it seems... But running it through the (somewhat unpredictable!) Anki simulator, I get the following figures. I've based it on the settings and retention rates for the 2000-odd Latin words I've learned in the last few months, and rounded it to 1000 days. Note that this isn't modelled with the simplifications you suggest above - it's modelled more accurately using my own learning steps, interval modifier, and retention rate at every step, as it's been for those 2000 words.
- 16 cards a day, recognition only:
- Maximum reps 513 (not on the final day... Anki has more randomness in it than that!)
- Average reps 363
- 32 cards a day, both directions:
- Maximum reps 975
- average reps 723
Interestingly, the recognition-only average reps is slightly less than my own average reps for the last three months, which has not been a period of intensive learning. I don't think that this level is particularly unsustainable... Using averages worked out from my own word cards, this would take about 15 minutes a day, which isn't bad at all. The issue would be also maintaining enough input and output to make sure those words stuck in a usable way, especially the later, rarer words.
The two-direction one is harder to model, without a deck that does this directly. Some of the cards would be a lot easier, because you get them twice, and some are harder because they are active. Would it also get disproportionately harder as you go on, since later infrequent words are less easy to turn into active vocab?
Realistically, even with one direction, your retention rate is going to be very different at different stages - low for the first thousand, then increasing rapidly as you get used to the language, and then at some point hitting much rarer words and so dropping again. I've less experience of this far end of the bell curve, but have hit it in Greek, so that individual word cards are much less effective than at earlier levels. If I read more and more widely this would be less of an issue, of course.
The other issue is that if you were enough of an Anki addict to keep going at that level for 1000 days, you are probably not the kind of person who can resist sneaking in some extra Anki drills - grammar, sentences, whatever - that would really slow you down. Which is of course exactly what I have done and why I've averaged 10 words a day for the last 6 months rather than 16.
All this presumes a reasonably familiar language. I don't have the stats directly for my own Farsi experience, but it took me, as I recall, basically double the time to learn an equivalent number of words. Latin might not be a category 1/2 language, if it had a category, but the vocabulary is as easy to an English speaker as a romance language. Just Anki-ing your way through a category 4 or 5 language would be seriously tough. There's a reason all the AJATT learners of Japanese use vast volumes of sentence cards - the words are just much harder to pick up in isolation than a related language would be.