Tired of Anki? Try Goldlist

Discussion in 'Learning Techniques and Advice' started by Bob, Jul 13, 2014.

  1. Bob

    Bob Active Member VIP member

    Joined:
    Apr 18, 2014
    Messages:
    339
    Native Language:
    English
    Intermediate Languages:
    Cebuano, Greek_ancient, Spanish
    Basic Languages:
    Filipino, French, Hebrew_clasical, Mandarin, Portuguese, Russian
    The Anki vs Goldlists vs Iversen stylewordlists has gotten me thinking about expanding my thoughts on Goldlist.

    Goldlist is an SRS with no upper limit on how long you should wait. Reviews don't make a huge bubble in the middle of the process, instead dropping off quite quickly.

    I haven't really talked about it much here, but I have a log on HTLAL Adventures in Cebuano where I tried this out for myself, starting around page 3.

    The original idea by David J. James is here with special attention given to Japanese .

    In short, this is the only vocabulary aid that I use for foreign languages, and I feel I can chalk up my high level of Cebuano reading to this. I probably would have gotten there anyway, I just think this is faster in the long run, and less stressful. It has some other possibilities too (I used it to remember what 2 birds looked like). It has the advantage of being very systematic, so you will always know exactly where you are, and how much further you have to go.

    Here's the basic idea: write down a list of words that you would like to know, with their definitions, and any other oddities, then come back after at least 2 weeks and rewrite 2/3 of the list that is hardest for you. Do this 7 times.

    At least 2 weeks? That's right, you don't have to rush to make sure you get things done by a deadline.

    Why does this work? I really don't know. But the idea is that after 2 weeks, you are working with long term memory and not short term. This is a reversal from the normal model that says you have to keep something is short term memory until it goes over into long term. We're going into long term directly. Also, there seems to be a breakpoint after 7 times, and you don't really need to do anymore.

    After 2 weeks of doing basically nothing but writing this stuff down (and ideally pronouncing them), there is always about 1/3 that seems easier than the rest. Could you pass a vocabulary test at this point? no, but it won't take long now to get these words to be usable. In my experience if I read (or hear) something twice, even if it's the same material, and I'm able to ask what it means, then that word is pretty much mine now.

    The "standard" Goldlist will start with 25 words. Then the list will get smaller.

    25 -> 17 -> 12 -> 7-9

    This completes the "bronze" level.

    After this, grab enough of these lists of 7-9 to have about 36 words, and from that make a new list of 25. Do it again

    25 -> 17 -> 12 -> 7-9

    This completes the "Silver" level. I stop here.

    So how do you know which words are in this ideal state of limbo? David simply says to go through and mark which words on the list that you know better than the others. Don't hide the translations! Look at everything and see which ones feel stronger. I think this is the most confusing part of the system. Some people have tried this and say "hey no way, I don't know 8 of these 25 words! I barely know 2!, this doesn't work." And by this they mean that they don't automatically know these yet. It's not yet usable. But remember, you're not going to be at that point yet, but you will be shortly.

    If you want to test it out, perhaps you could try to do it like I do. Find 25 words that you could use (so that you can try it when you're done). Don't pick words where you say to yourself "oh yeah, I remember what that is", because that is the end result of this method. Two weeks later come back and do this:

    1) Go through the whole thing once, top to bottom, looking at the words and definitions.
    You are not memorizing! only exposing. You don't want to cram, or find some cutesy way to remember the word.
    2) cover the translations and see what, if anything, you can easily recall. Mark them.
    The first time around I can normally pick 4 (If it's not a brand new language)
    3) Look at them all again bottom to top. Try to recall the ones you marked while doing this.
    4) Cover the translations again and see what you know now.
    5) repeat steps 1-4 until you have chosen about 1/3 of the words.
    5) Adjust your choices as necessary.

    With 25 words, you will usually have to stop after 8 words have been chosen. These are the 8 words that you were able to memorize without too much effort.

    Using this process I can learn to read a language. It takes a bit to ramp up a completely new language, because you don't know enough yet to get context clues. I'm doing this with Jonah in Hebrew, It's taken about 2 weeks at a leisurely pace (I'm quite busy)
    Last edited: Jul 14, 2014
  2. Peregrinus

    Peregrinus Active Member

    Joined:
    May 27, 2014
    Messages:
    613
    Native Language:
    English
    Intermediate Languages:
    German
    Basic Languages:
    Spanish
    More info on the GL method:

    HTLAL: Has anyone tried the Gold List method?
    HTLAL: Trying Goldlist in a classroom setting (by ElComadreja - our Bob)
    Language Geek Blog: The Gold List vocabulary method
    Victor's Blog: The Goldlist Method (with pics showing pages of lists)
    Uncle Davey's youtube channel (creator of GL method - scroll down to GL playlist)

    To summarize:

    1) Take 25 headwords and write them with definitions and other info like gender, constructions, etc., at the top of a new page.
    2) Wait at least 2 weeks but not more than 2 months, and distill the first list by choosing 70% of the words that one knows least well - write those at the top of the opposite page (1st distillation).
    3) Wait another 2 weeks to 2 months and distill again - write the 70% below the first distillation (this is the 2nd distillation).
    4) Same again, after next period make a 3rd distillation -writing those words below the original list on the left page (note inventor uses different color ink for each list and checkmarks for the distillations).
    5) After you have 3 distillations for several headlists, you transfer the 3rd distillation from each to a new list and start over.
    Last edited: Jul 14, 2014
  3. Bjorn

    Bjorn Active Member VIP member

    Joined:
    Apr 17, 2014
    Messages:
    165
    Native Language:
    Norwegian
    Intermediate Languages:
    English
    Basic Languages:
    French, German
    Interesting, I have read about it before but I dropped it since I didn't quite understood how it works and also because pen and paper is not my favorite tools.
    Thanks to Bob and this link from Peregrinus http://en.vvb.no/2012/03/the-goldlist-method/ I think I got it now.

    The question is how important is the handwritten aspect in this method versus typing it once?
    In a spreadsheet you just have to type the words once and then just moving around the words into different "books" after each distillation.
  4. Peregrinus

    Peregrinus Active Member

    Joined:
    May 27, 2014
    Messages:
    613
    Native Language:
    English
    Intermediate Languages:
    German
    Basic Languages:
    Spanish
    I suspect that if the method can be successful for a lot learners (which I have not seen yet, i.e. a lot of success stories), then the handwriting aspect may be crucial. And doubly so for ideographic languages like Chinese. One is already not spending as much time long term with a word as with other methods, so handwriting may be what insures a meaningful amount of time spent per word.

    Obviously with any method there are lots of possible permutations of the details. As with typing vs handwriting, normal handwriting vs. a slower scriptorium like writing, doing 1 list per day or 5, etc.

    BTW re multiple Goldlists per day, one instruction I did not give above was that Uncle Davey says that there needs to be some time interval being doing lists.
  5. Bob

    Bob Active Member VIP member

    Joined:
    Apr 18, 2014
    Messages:
    339
    Native Language:
    English
    Intermediate Languages:
    Cebuano, Greek_ancient, Spanish
    Basic Languages:
    Filipino, French, Hebrew_clasical, Mandarin, Portuguese, Russian
    Something else to be said is where do you get your words? David describes a scenario where he has a book in 2 languages. He reads along and whenever he sees a word he doesn't understand, he uses the other book to get the meaning. When the words have gone through the process he reads the book again.

    I usually just use a dictionary. Later on, If I'm already B1/2 reading, I'll underline words I can't figure out and look them up after I'm done with a section. Then I read it again.

    On a smaller scale I'll list the glossary of a language course (or the words from the dialogs) and then do the course.

    For Hebrew, after listing the high frequency words, I also listed the additional vocab for Jonah. I finaly just got through reading the thing for the first time, but it's still shaky. I think this is because I'm trying to absorb every stinking word at once.
  6. Big_Dog

    Big_Dog Administrator Staff Member

    Joined:
    Jan 11, 2014
    Messages:
    1,039
    Native Language:
    English
    Advanced Languages:
    Spanish
    Intermediate Languages:
    French, Japanese, Mandarin, Russian, Swahili, Thai
    Basic Languages:
    Korean
    The method doesn't appeal to me because, unless I'm misunderstanding, it takes out the easy words. I find most kinds of exercises to be more efficient when there is a decent amount of easy material in the mix. I believe this has actually been proven to some extent, but can't find a link to the experiments. Somewhat related, from FSI Lessons learned from fifty years of theory and practice in government language teaching:
  7. Bob

    Bob Active Member VIP member

    Joined:
    Apr 18, 2014
    Messages:
    339
    Native Language:
    English
    Intermediate Languages:
    Cebuano, Greek_ancient, Spanish
    Basic Languages:
    Filipino, French, Hebrew_clasical, Mandarin, Portuguese, Russian
    Well that's the thing, you take out easy words, next time you take out the words that are easy now, eventually it's all "easy". Then being able to make sense out of everything is quite a boost. After I do this, I ideally find texts that have some of these new words in it, so that I'm not having to check every word. Again, this is impossible to do right at the beginning so it's a good idea to start with something like a Teach Yourself or a text that repeats itself allot.
    Last edited: Jul 14, 2014
    Big_Dog likes this.
  8. Peregrinus

    Peregrinus Active Member

    Joined:
    May 27, 2014
    Messages:
    613
    Native Language:
    English
    Intermediate Languages:
    German
    Basic Languages:
    Spanish
    As Bob indicated, you do have easy words today, which you take out or send further to the future. Then the hard words of today eventually become the easy words of tomorrow. Only if you already had a fairly good vocabulary, and were now only adding lower frequency words with no connections to other known words (i.e. in same word families), would you be confronted with all hard words. Even then, unless your vocabulary learning activity via whatever method were the sole activity, you can counterbalance with other different activities, like appropriately graded reading and listening.

    In this respect I don't think GL differs from Anki, unless the algorithm used by GL (2 weeks plus instead of the SRS algorithm), is inefficient in teaching vocabulary (which it might be).
  9. Iversen

    Iversen Member VIP member

    Joined:
    May 1, 2014
    Messages:
    91
    The paradoxical thing is that the 'easy' words may be the hardest ones to learn, because they often are irregular and used in lots of expressions, where their contribution to the total meaning isn't obvious. I sometimes include them in my own wordlists (for instance by quoting forms based on irregular stems), but they basically should be learned by other methods. I suppose the same argument applies to goldlists and Anki.

    On the other hand, many political, economical, technical and scientific terms are easy to learn because they are loanwords or derived from loanwords.
  10. Bob

    Bob Active Member VIP member

    Joined:
    Apr 18, 2014
    Messages:
    339
    Native Language:
    English
    Intermediate Languages:
    Cebuano, Greek_ancient, Spanish
    Basic Languages:
    Filipino, French, Hebrew_clasical, Mandarin, Portuguese, Russian
    Quite true. I still put a word in if it's got something weird going on, and put the oddity there with it. Sometimes I'll put a weird form of the word as it's own item. Assuming "go" was easy, something like "went" is a good candidate to make its own item. You could also make a line like:

    go, went (past) Move somewhere

    and choose the item when the whole thing is easy for you. Or what I do is just take out what I happen to recognize at the time.

    Also, idiomatic expressions can be their own item. A Cebuano example:

    layo sa tinae it's not that big a deal (lit. far from the intestine)

    edit: can't seem to get tabs to work on this forum :p
    Last edited: Jul 15, 2014
  11. Big_Dog

    Big_Dog Administrator Staff Member

    Joined:
    Jan 11, 2014
    Messages:
    1,039
    Native Language:
    English
    Advanced Languages:
    Spanish
    Intermediate Languages:
    French, Japanese, Mandarin, Russian, Swahili, Thai
    Basic Languages:
    Korean
    Nicely stated.

    Actually, I think it's quite a bit different from Anki. As you pointed out in another thread, Anki doesn't take out the easy cards, it just bats them into the future. Also, many people have found that it's no fun working with leaches. Personally, I delete cards that take more than 3 or 4 attempts. If I see that word or phrase at a future date, and think it's worth memorizing again, I'll put it in again as a new card. Often that solves the problem, but sometimes it gets deleted again.

    Finally, I think one of us doesn't understand the Anki analog correctly. It's probably me. But I thought the analog was pretty complicated, meaning not really easy to think about, because it actually tries to create a session where you get 90% correct (assuming you have your forgetting index, or whatever he's calling it now, set to 10%). That means that you aren't just going to get your hardest cards. There is some complicated switching out going on to try to make the session somewhat easy for you. I'm not the only one who's heard that you should have easy cards in the mix - Supermemo used this fact too. But maybe I'm wrong about all this.

    I'm not one to talk, because I haven't tried the method, and don't fully understand how you determine words to be eliminated. But working with material which is currently difficult, regardless of whether or not it will be difficult in the future, doesn't seem to be following the "mix in some easy material" principal. That being said, I respect the fact that the method works well for some people.
  12. Peregrinus

    Peregrinus Active Member

    Joined:
    May 27, 2014
    Messages:
    613
    Native Language:
    English
    Intermediate Languages:
    German
    Basic Languages:
    Spanish
    I don't really read that kind of stuff about Anki, i.e. their rationale for their specific implementation of the algorithm, although of course I have read excerpts as they come up in discussions in their help list. The creator has very rigid views as I said as to his implementation of of the SRS algorithm. But though he is an excerpt programmer, his understanding of learning is not much different than ours, i.e. he's just another dude on the internet with an opinion. He is intent on forcing users to do Anki a certain way with as little leeway as possible, for the legitimate reason that otherwise they will cry it does not work, when it was an individual's implementation of Anki that did not work.

    It is true that by via tweaking various settings regarding easy/hard, one can determine a certain pass/fail rate for individual sessions. I am very conservative and my numbers of reviews and review times are correspondingly higher. Similarly with Gold List, one has to make a judgment about how well knows a word, and that series of judgments impacts results. Probably only forcing a user to type/write an exact definition, which wouldn't work because they are somewhat nebulous, would one be able to get a more accurate pass/fail rate (assuming no partial credit).

    Re a mix of easy vs hard, again I believe that absolutely, and have read that FSI paper in the past more than once. But I don't think there has to be a balance in every individual learning activity, just overall. In fact learners who have progressed to an intermediate or advanced stage (however one defines that), probably underestimate what they can learn by going over a basic course again quickly, especially regarding usage and fine points of grammar.
  13. Big_Dog

    Big_Dog Administrator Staff Member

    Joined:
    Jan 11, 2014
    Messages:
    1,039
    Native Language:
    English
    Advanced Languages:
    Spanish
    Intermediate Languages:
    French, Japanese, Mandarin, Russian, Swahili, Thai
    Basic Languages:
    Korean
    My opinion on that is if you are able to control it, make sure you have some easy material inter-leafed in. That's because there will be many instances in language learning where you'll have to do hard stuff, and you won't be able to control it. When you have a chance to be efficient, be efficient.
  14. Bob

    Bob Active Member VIP member

    Joined:
    Apr 18, 2014
    Messages:
    339
    Native Language:
    English
    Intermediate Languages:
    Cebuano, Greek_ancient, Spanish
    Basic Languages:
    Filipino, French, Hebrew_clasical, Mandarin, Portuguese, Russian
    For the record, I just read straight through Jonah and only had to check about 6 words.
  15. Cainntear

    Cainntear Active Member VIP member

    Joined:
    Apr 29, 2014
    Messages:
    343
    Native Language:
    English
    Advanced Languages:
    Catalan, French, Italian, Scottish_Gaelic, Spanish
    Intermediate Languages:
    Corsican
    Basic Languages:
    Dutch, German, Irish, Polish, Russian, Welsh, Sicilian
    Indeed. It's quite refreshing to have a piece of software designed that way when a large part of the competition prides itself on flexibility, but in the end just gives the user enough rope to hang themselves... from the moon.
  16. Bob

    Bob Active Member VIP member

    Joined:
    Apr 18, 2014
    Messages:
    339
    Native Language:
    English
    Intermediate Languages:
    Cebuano, Greek_ancient, Spanish
    Basic Languages:
    Filipino, French, Hebrew_clasical, Mandarin, Portuguese, Russian
    As stated by James down in the comments somewhere, the 2 months is not strictly necessary. It's more of a scheduling tool to keep things on track. I had some Cebuano lists up on the Gold level that were far more than 2 months that I tried out. It definitely took longer than usual to find my words, but it all still worked.

    And since this info is hard to find all at once, I'll add that even though James says Going through the Silver level is enough for most people, he does a "Gold" level after the Silver level (hence the name of the system). After that you do one final distillation, and you'll be left with maybe 150 words from a large project. These words you then put on "shirts and coffee mugs" etc.
    Last edited: Jul 26, 2014
  17. Peregrinus

    Peregrinus Active Member

    Joined:
    May 27, 2014
    Messages:
    613
    Native Language:
    English
    Intermediate Languages:
    German
    Basic Languages:
    Spanish
    Bob, thanks for the further instructions on using GL. "shirts and coffee mugs" is hilarious. Probably the only thing for the most recalcitrant words using GL.

    You (and I guess your students), are still the only ones I see reporting sticking with GL for a long period, let alone having success with it as you have had. As I said in another post, I suspect the reason may be trying to cram too many lists in a day, or at least not slowing ramping up to find a sustainable number of lists per day.

    What still bothers me about GL is there not being either a scientific basis as with Anki, or at least untested yet scientifically plausible bases for the constituent parts as with Iversen's method. But what works works, even if one does not understand why. However perhaps the repetition for less well known words is science based, just not optimized perhaps as the underlying algorithm of Anki tries to do. The flexibility, which Iversen's method shares, of not being tied down absolutely every day is very attractive.

    What have you found to be a sustainable rate of new words per day or month?
  18. Bob

    Bob Active Member VIP member

    Joined:
    Apr 18, 2014
    Messages:
    339
    Native Language:
    English
    Intermediate Languages:
    Cebuano, Greek_ancient, Spanish
    Basic Languages:
    Filipino, French, Hebrew_clasical, Mandarin, Portuguese, Russian
    The suggestion is no more than 10 new lists a day. Of course this would mean roughly 5 hours (including breaks) so not everyone can do this. This seems about right if I have no backlog (and nothing important to do at home). Otherwise 6 new lists feel like my limit. For our students we set it up to be an average of 2 new lists a day. (50 minutes max) Whenever I stop adding new words it quickly feels like my workload is negligeable. And, usually, if I had time 2 weeks ago to write down 25 words, then today I have time to distill.

    Here is his article on making a schedule keeping in mind that you would like to stay in the 2 month range. I feel it's a bit complicated.
  19. Peregrinus

    Peregrinus Active Member

    Joined:
    May 27, 2014
    Messages:
    613
    Native Language:
    English
    Intermediate Languages:
    German
    Basic Languages:
    Spanish
    Yeah that's awfully complicated, and would require an awful lot of time. I used the word "awful" twice intentionally :). Obviously the lists pile up over time and one has to budget time for the long run accordingly. Perhaps if he limited himself to advocating just a couple or three lists per day max he would entice more learners to give GL a shot and who would then have a better chance of sticking with it.

    Looking at the total time for the project he gave as an example, i.e. 128 hours for 3000 words /lines of head list, I am pretty sure that works out to over twice what Iversen would take to learn a comparable number of words, even allowing for the fact Iversen has approximately a 15% attrition rate of lost words (128 hours *60 minutes /3000 vs. 20-30 minutes for 30-40 words). Although I can't make a good estimate for Anki, my strong suspicion is that 128 hours would be far more than required for a similar amount of words with Anki, the learning plus subsequent review time for which would start out slow and then peak and decline once no new words were being added, even allowing for card creation time.

    Uncle Davey says that one can take between 2 weeks and 2-3 months between distillations, but it would be nice to see an experiment where learners tracked different batches of lists on different (but internally consistent) time frames, to see what produced the best results as to distillation intervals. Perhaps using shorter intervals would result in better retention and fewer needed distillations.
  20. Bob

    Bob Active Member VIP member

    Joined:
    Apr 18, 2014
    Messages:
    339
    Native Language:
    English
    Intermediate Languages:
    Cebuano, Greek_ancient, Spanish
    Basic Languages:
    Filipino, French, Hebrew_clasical, Mandarin, Portuguese, Russian
    My scheduling is much simpler:

    1) Do I have 20 minutes?
    2) Do a list
    3) Go do something else
    4) Go back go step one when I get a craving

    It would be good to get the hard numbers for those playing at home, but at the moment I still prefer Goldlist because I'm not trying to memorize anything. It's a nice lesuirely walk that I can take. It also keeps me in the mind frame of "think, don't translate". With anki after a few weeks I find myself beating my head against the wall with large amounts of leeches and super easy cards. I also feel forced to do reviews "today". There is also a definate end, and it's not when I'm tired of seeing a card.

    On another note, (and maybe this should go back to the comparison thread) after going through Iverson's method, do you "know" the words? could you pass a vocab test just doing that?

    Also, I've found I can do this at any time of day when, for example, I don't have the brain power to read an intermediate foregin language.

    James may have these numbers on himself as the Goldlist started out much more complicated.
    Last edited: Jul 26, 2014

Share This Page