Progress Report - Day 8

I just finished my 50 words for the day and I thought it would be a good time to share some statistics. I've been keeping track of the number of repetitions required to remember a word for the first time. I also keep track of the number of leeches and will report some basic numbers here as well.

Here's a look at how I recorded my new card repetitions on day 3.

1/23
7:24 59
7:36 51
7:47 33
7:53 22
7:58 15
8:02 12
8:04 7

The left column is time and the right column is the number of failed cards left for review. When I add new cards I set Anki to "show failed cards at end." I then go through my new set of 50 words to get an overview. As you can see there are more than 50 cards starting. This is because I had some failed cards leftover from a previous session. In the future I'll try to have a clear queue when I start a new session.

By the 5th review I have successfully answered 75% of the cards. When I took a look at the first week data this number was pretty indicative of the week as a whole. 76% of new cards are remembered (answered correctly the first time) by the 5th review. Six reviews and up appear to be represent the point of diminishing returns.

Here's an overview of what I've discovered so far.

• 76% of new words are remembered after 5 reviews or less
• 15-18% of new words will not be remembered (i.e. they will eventually fail enough times to be flagged as leeches)
  
• Day 7 - 15% of all new cards were suspended (cumulatively)
• Day 8 - 18% of all new cards were suspended (cumulatively)

Today was the most difficult list of new words I've encountered. I think that explains the large increase in number of suspended cards. I define most difficult quantitatively (see the numbers above) and qualitatively. It felt like there were many unknown Kanji readings during this session.

I plan on augmenting my strategy as of today. When learning new cards I will manually suspend any cards that have not been remembered by the 6th review. The reason is that these words have a high probability of becoming leeches and will most likely be a waste of time. It will be easier to learn these words after the 120 day challenge. This is inline with my original intention of remembering a minimum of 33% of the word list. If I maintain this pace and leech rate I should be able to claim around 4920 words.

I also hypothesize that the suspension rate will decrease over time. There are two reasons for this. As I progress I continue to learn new Kanji readings and bolster known or partially-known readings. This will improve the retention rate of new words. Secondly, I've been going through the word list backwards. That is to say that I've been learning the JLPT1 word list first which statistically contains more of the less frequently used Kanji than the JLPT2 list (I've already finished levels 3-4).

That's all for now. I plan on updating these numbers at least once per week. If you can think of any interesting statistics you'd like to see please leave a comment.