Ratio of new words vs reviews

General
1

@Steve How does readlang decide when to add new words to start learning vs reviewing old words. I suspect that the ratio may be off for me as my number of words in the learning set has been growing pretty consistently over time. Is there any way to control this or force reviewing old words only for a while?

2

Currently, Readlang doesn’t have separate controls for new words vs reviews. When you practice, the system shows you words that are either:

  1. Scheduled for review
  2. New words you’ve never practiced

Both types go into the same “ready to test” pool, sorted by favorites first, then by word frequency.

I take your feedback and maybe it could be an option in future to restrict practice to only reviewing old words. But personally, I think you shouldn’t worry too much about the number of words in the learning set growing. This is likely to happen if you read more than you do you practice sessions.

3

So, is there no specific expected behavior when there are neither favorites nor applicable word frequency data? I’ve never used favoriting, so I feel like this situation is likely typical for my reviews. But somehow I don’t have the sense that new words are drawn completely at random. Am I wrong?

edit: specifically I think it’s choosing the most recently created new words. That suits me fine, if that’s how it works, since I’d normally want to see those ones I had just been reading.

4

I suspect this approach has broken down a bit. This is my current status in Slovak which is a heavily inflected language:

  • I have > 13k “learned words” and > 40k ready to learn words
  • Of the 13k learned words only like 6k are actually unique/different lemma. I have lots of the same word but with different endings due to conjugation/cases.
  • The new words that get added are pretty heavily biased towards different forms of words that I really already know as the frequency shows them high.
  • I’ve done some manual review of words that are lower frequency, and I suspect those are just not getting reviewed because they are “pushed out” by higher frequency words that are just different forms of words I already know.

Thinking about this more, I think frequency guided is great for beginner and lower intermediate, but is actually counter productive for higher levels of vocab in inflected languages. Am I thinking about this the right way?