Anki vs. Gold Lists vs. Iversen-style Wordlists

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

  1. Peregrinus

    Peregrinus Active Member

    Joined:
    May 27, 2014
    Messages:
    613
    Native Language:
    English
    Intermediate Languages:
    German
    Basic Languages:
    Spanish
    @Iversen,
    Thanks for the reply. My intent in asking the question was the thought that perhaps your long-term attrition rate is not as high as you think, in that a previously forgotten word from a list may show up again in a subsequent list and then stick, though that would be semi-dependent on their frequency. Taking your words from various sources as you do, you have to come across words in some type of frequency, either overall for the language or in a topical domain. So perhaps the more frequent ones that are initially forgotten get listed again without an explicit plan for that to happen, while extremely infrequent ones do not. Thus perhaps your system "recaptures" a good part of words that are lost via attrition on the first go round. But this would be impacted by your subjective judgment in applying "guessable" that you discuss to whether to list a word or not.
  2. Iversen

    Iversen Member VIP member

    Joined:
    May 1, 2014
    Messages:
    91
    That is true, but as you probably remember I made word counts in Serbian for both pages in the part of the dictionaries I just had worked on with my wordlists AND pages in the part I hadn't reached yet, and the number of known words in the latter amounted to around a third of the words on these pages. So I definitely didn't start the vocabulary campaign from scratch. I have in fact also warned against doing dictionary based wordlists before you already have some experience with the target language, among other things because it makes it much easier to remember words when you can form associations to other words in this language. So it is almost inevitable that I have seen many of the words on my lists in the months before the test.

    When I choose words for the lists I don't try to avoid guessable words. For instance a large part of the Serbian words on -ија er loanwords that have -y in English and -i in Danish (like "копија" ='copy') - but you can't always count on this, so it is still worth taking notice of the cases where the rule applies. Of course the percentages of retained words get higher when not all words are totally opaque and indigenous, but the goal is not to make the exercise as difficult as possible, but to learn a representative selection of the words in the language. And with a dictionary of 12.000 headwords (and an even smaller one with 8.000 headwords) the risk that I memorize extremely rare words fairly limited. I don't think I list the same words again and again very often in wordlist after wordlist, but it might happen with words from texts. However these words were apparently not memorized well enough the first time, so I don't see this as a problem. Actually I'm fairly sure that some of the words I have used in a dictionary based list has been on a previous 'free' list, and of course they will be easier to remember if I have used them before. But again: the goal is not to make the lists more difficult than necessary. On the contrary: when you work with worklists you should try continually to adjust the difficulty up and down through choice block size and difficulty of the words in such a way that it is a challenge to remember each block - but not so high that it becomes a strain or so low that it becomes trivial.

    Apart from that: I made some wordlists for О, П, Р and С in July (before my Québec tour in the first two weeks of August) and managed to do one repetition round for them all - but not the one two days later (as in the original experiment setup). For Р and С I could do one a few days later because I had brought the 'latest' letters along in my luggage. And then I made a late repetition for all four letters today, so now I have figures for three repetition rounds for Р and С, but only two for О and П. The funny thing is that the missing repetition round doesn't seem to have had much impact on the results of the late repetition here 3-4 weeks later. The 'letter size' seems to be more important: due to lack of time I did a less thorough memorizing session for letters with many words (П and С) than with the 'shorter' letters before leaving home, and now I have had to face the consequences: a higher percentage of 'lost' words and also higher overlap, because I simply didn't do the job properly the last days before my departure. And now Novi Sad is only 1½ month away, so I from now on I suppose I have to emphasize the activation of the vocabulary I already have got rather than on learning even more words.

    The Goldlist repetitions and the final test will however take place as scheduled.
  3. Iversen

    Iversen Member VIP member

    Joined:
    May 1, 2014
    Messages:
    91
    In my Serbian vocabulary project I have reached a milestone: I have reached the end of the alphabet and done at least two repetitions of all the words on my original wordlists. Not all words have been through 3 rounds: 'П' and 'Р' got their second and last review after 4 weeks, and the letters from 'Y' onwards (Latin 'u') were reviewed after one week so here I assume that the 2. round just was delayed - although the control situation was more like a third round without round 2. And then I discovered to my dismay that I had skipped a column in 'T' during the 1. and 2. round (round 2 was affected because I just did that round by annotating the checklist from round 1, whereas I wrote a new copy of the words for the 3. round - and then I discovered the missing section)

    Results:

    1. round (always after 1 day):
    4780 words, missed 958 (20%)

    2. round (mostly after 2 days, at the end of the alphabet after 1 week):
    4810 words, missed 805 (17%), incl. 278 (6%) overlap with round 1

    3. round (after 2-4 weeks)
    2669 words, missed 431 (16%), incl. 239 (9%) overlap with round 1,2

    The most striking thing here is that the percentages for missed words are almost the same in the three rounds - but the words I couldn't recall were to alarge extent new ones every time, which suggest either 1) that I relearn the words I miss in a given round, or 2) that there always is a certain amount of memory malfunction which in principle can hit any word. That 1) could be true is a matter of course, but there are also cases where 2) is the more likely explanation. Such as missing "шеик", which according to my dictionary is a loanword which means 'milkshake' - but when I saw it in round two I immediately translated it as 'sheikh' in my mind, and then it counts as missed.

    The slightly better results in round 2 and 3 suggest that relearning plays an important role, but if my relearning functioned really well the percentages for missed words should fall more. I can't say for certain whether it is the time pressure during control sessions or the general 'noise' that should be blamed for this situation. Before this experiment I did my 1. (and mostly only) round like a relearning situation, whereas this time I did more rounds, but I did them as control rounds where I didn't review the words before testing my memory - which is the situation Anki users always find themselves in.

    From now on I think I'll revert to my earlier review method where I first look the words through, then copy the words 'blindly' (with covered translations) adding translations wherever something is dubious or inaccessible. But I think this experiment has shown that at least one extra round is beneficial. And then I think I'll add one element: if you run through the translations in the original wordlist immediately after doing a repetition round most of the original can be recalled - but the decay rate is fairly high: I am much less successful just a day later. My hunch is that doing such a round based on the translations might help to keep make the relation from translation to original word stick, which should be relevant for the later activation phase.
    Peregrinus and Bob like this.
  4. Peregrinus

    Peregrinus Active Member

    Joined:
    May 27, 2014
    Messages:
    613
    Native Language:
    English
    Intermediate Languages:
    German
    Basic Languages:
    Spanish

    In my hybrid experiment, and with Anki alone, more often than not, the miss rates between rounds tend to cluster close together and not go down dramatically, the same as with this report. Anki has the ability to filter for missed words for a time period, and I used to add a round at the end of each day where I reviewed the words I had missed earlier. While it made some difference the next day when those missed words showed up again, it was not enough for me to keep doing it. And as has been noted before, some words randomly seem to be missed in various rounds in addition to some words always being missed. I suppose that more time spent on the first couple days in additional review rounds could help, but I don't think it is worth it time-wise as long as Anki is making sure I have close to a zero long term attrition rate.

    Although I have not done it since I do use Anki long-term while using Iversen lists now for initial learning, a benefit of doing Iversen style lists digitally with a spreadsheet is that one could easily continue the process at any time in the future. One could add a column of logical fields, and then quickly review older lists and note misses, and then sort the lists to come up with just the (current) misses. And then of course treat them as one or more new lists. In a way it would be like using a spreadsheet Anki style but not to a fixed review schedule.
  5. Peregrinus

    Peregrinus Active Member

    Joined:
    May 27, 2014
    Messages:
    613
    Native Language:
    English
    Intermediate Languages:
    German
    Basic Languages:
    Spanish
    Some further observations on my now standard hybrid Iversen-Anki word list method. While I still have not improved much in mnemonic associations with the exception of Iversen's excellent suggestion to kind of blend a L2 word with part of the L1 definition, and thus am not as effective as someone who has ability with such associations, I find that the initial use of a digital Iversen list in a spreadsheet prior to continuing with them in Anki, still is producing better results than Anki alone. New words seem to fall into one of 4 categories: easy, somewhat easy, somewhat hard and hard. The effect of using the Iversen list at the beginning has helped to move many/most somewhat easy words to easy, a lot of somewhat hard words to somewhat easy, and a few hard words to somewhat hard. Thus I end up with fewer reviews and time spent on same for the overall list, while still spending the most time long term on hard words.

    From an average of 30 new words per day, taking a day off occasionally when the reviews started to pile up, I am now averaging 45 words a day, doing about 60 alternate days and 30 or so on the other days. This has had the effect of increasing reviews, and I sometimes pass a word with a low next interval like 3/4 days when I did not really know it, just to keep reviews down, since I can always fail it later. But as I have said before, the same time, i.e. 40-60 minutes that previously maintained a far smaller vocabulary of learned and partially learned words, now maintains a much larger one, currently around 12,500. Yesterday I had 580 words to review, which ended up being 710 review with failings and re-reviewing, and spent 48 minutes doing so (broken into chunks over 3 or 4 hours). Since I log this every day in a spreadsheet, I can look back a year ago and see this is only about 5-6 minutes more for about 25% more reviews covering a far smaller total vocabulary (which I don't log every day but probably then was 3000 or less). The effect of gradually pushing words that become well-known into the future really adds up and works to allow the same time being sufficient to maintain an ever growing vocabulary. Obviously which words become well-known and which don't varies and depends on how often I encounter them in extensive reading and listening, thus whether they get reinforced often or not.
  6. Iversen

    Iversen Member VIP member

    Joined:
    May 1, 2014
    Messages:
    91
    Using three-column wordlists as an introduction to the words which you then follow up through Anki seems to be a very sensible model - the only drawback I can see is that you might be tempted to use a keyboard to make the transfer to the digital format in Anki easier, and I personally think that writing things by hand as more efficient. But that may be different for different persons. I have always been less concerned with the way words on a list are reviewed as long as they are reviewed at least once. So using Anki for the purpose doesn't go against any of my principles - except that I don't like to be checked by a machine, and that I prefer seeing a lot of words at the same time instead of one at time.

    It seems likely that my good results with just wordlists (no Anki) are due to many years of association training - if I don't spend enough time and effort initially to build good association 'bundles' for each word I also forget them. But with those 'bundles' I retain the words fairly well.

    I would like to quote a few passages from HTLAL relating to the end of my Serbian dictionary perousal: "
    In my Serbian vocabulary project I have reached a milestone: I have reached the end of the alphabet and done at least two repetitions of all the words on my original wordlists. Not all words have been through 3 rounds: 'П' and 'Р' got their second and last review after 4 weeks, and the letters from 'Y' onwards (Latin 'u') were reviewed after one week so here I assume that the 2. round just was delayed - although the control situation was more like a third round without round 2. And then I discovered to my dismay that I had skipped a column in 'T' during the 1. and 2. round (round 2 was affected because I just did that round by annotating the checklist from round 1, whereas I wrote a new copy of the words for the 3. round - and then I discovered the missing section)

    Results:

    1. round (always after 1 day):
    4780 words, missed 958 (20%)

    2. round (mostly after 2 days, at the end of the alphabet after 1 week):
    4810 words, missed 805 (17%), incl. 278 (6%) overlap with round 1

    3. round (after 2-4 weeks)
    2669 words, missed 431 (16%), incl. 239 (9%) overlap with round 1,2"


    So my 'word loss' has actually been minimal during the test period, which might seem surprising. However it should be considered that I have been listening assiduously to Serbian and Croatian TV, and I have also studied a fair amount of articles - mostly about popular science.

    One more quote from HTLAL:

    Since Sept 1 I have done a quite significant follow-up study to my protracted Serbian vocabulary experiment: a word count using two dictionaries: my Serbian<->English dictionary which I used for the experiment and an old Serbocroatian<->French dictionary (Obod Cetinje Medicinska Knjiga) with approximately 26.000 headwords in the Serbian part according to my estimates. (...) This dictionary uses Latin letters which I have tried to avoid, but it has nevertheless been useful in many cases where I couldn't find a word in neither the English one, not the Italian one.

    I divided the words on 6 pages from the Serbian-English dictionary into known, 'borderline cases' and not known, and I got 252 - 39 -104 words in those categories, i.e. 64% - 10% - 26%. Which clearly shows that my vocabulary marathon with its almost 5000 wordlist words has had a lasting and durable effect on my Serbian vocabulary - and not just an effect limited to the words on the lists. The Serbian<->English dictionary has roughly 12.000 words, so even a third of those is a fairly solid background for further work in the language. But I wanted to check my level on another dictionary - and preferably a bigger one - so I grabbed the old French one from the shelf and went through the left column on three pages, which yielded 166 words: 81 known (49%), 17 borderline (10%) and 68 not known (41%). Those 49% is undeniably less than the 64% from the other dictionary, but quite satisfying if you take into account that this was a bigger dictionary and that it used Latin letters. Actually I only wrote the words from the first page in Latin letters - after that I decided to quote the words in Cyrillic letters, but I don't feel that this had any effect on my results.

    And now I can presumably speak Serbian? Nope. I have not had time to train active skills too, so even though I can put sentences together in my head now I don't feel ready to speak. Knowing the right words is a good start, but the ability to apply the morphological and syntactical rules on the fly is quite another thing, and whether I can get that put into order in the month up to Novi Sad is still an open question. But I can already now see the effect on my reading capabilities. I have been reading articles from Wikipedia for some time now, including a long article about the Triassic period, and I feel less and less need to consult the dictionary or the translation of a bilingual printout (...)"


    As you can see I have more or less settled for the term "borderline cases" instead of the misleading "guessable words" - I have scored many loan words as known simply because I assumed they would look in a certain way .. and they did. And these are the real 'guessable' words, not those which have a form or meaning which was guessable, but nevertheless surprising in some way.

    One consequence of my wordcounts is that the test of Russian 3-column wordlists against Russian goldlists has become compromised - my background vocabulary in Serbian has risen so much due to my dictionary exercises that that I may experience some spill-over into Russian, which could drown any difference in efficiency between the two methods - maybe it had been better to choose a non-Slavid language for the purpose. I'll make the fourth and last destillation of the goldlists soon, and then maybe a week later I'll do the comparative tests. But the most interesting part of the test could be my score on the 100 "hands-off" word list which I haven't seen since I made it.

    Right now I'm watching Croatian TV (HRT1), and lo and behold, they are interviewing a Danish expert on renewable energy - in English, but with subtitles. And there are a number of words there which I haven't seen in my dictionaries because this is a fairly new knowledge field - like vjetroturbini (wind turbines). But I would prefer listening to more Croatian (or Serbian rather than English) - I need listening training more than reading exercises.
    Last edited: Sep 8, 2014
  7. Peregrinus

    Peregrinus Active Member

    Joined:
    May 27, 2014
    Messages:
    613
    Native Language:
    English
    Intermediate Languages:
    German
    Basic Languages:
    Spanish
    Actually I transfer words from Anki, where I have already entered them and their definitions, to the spreadsheet. But I do type them into the spreadsheet. I am a fairly fast touch typist in English and almost so in German since the keyboard for same which I enable is not that different.

    I doubt that I will ever make much progress with mnemonic associations, but there is another aspect to my hybrid method that occurred to me the other day. We are all familiar with the phenomenon of waking up to find various words and phrases in a L2 that we have read or listened to bubbling up to our consciousness. For me this is especially with audio sources to which I have listened a number of times, but also vocabulary studied through Anki. I notice that I especially seem to have this happen with the words I have listed and studied in the spreadsheet, and with a little effort can recall many of them from a session. This effect is less pronounced from Anki, and I wonder if Anki does not interfere with the effect for the wordlists. That is, if I only used Iversen style word lists I might find my attrition rate to be lower. I am not willing to stop Anki and test the hypothesis though, at least with German. But I am considering as I have mentioned, restarting Spanish with word lists only. I would still put them in Anki to serve as my personal glossary, but I would not review them there.


    If you are willing to do so, perhaps you could test Goldlists again on a different language that is currently on your backburner in order to avoid any such problems. I guess the effect of some of the GL lists being compromised, is that they have kind of turned into another repetition round of your normal lists.
  8. Iversen

    Iversen Member VIP member

    Joined:
    May 1, 2014
    Messages:
    91
    Actually the gold- and other lists involved in this experiment are in Russian, but the overlap in vocabulary with Serbian/Croatian is so big that I expect a spill-over effect. I'll edit the previous mesage to make this clear.
    Last edited: Sep 8, 2014
  9. 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
    I've found Goldlist to go crazy fast for Portuguese and Italian, even after I try to filter out words that would be too obvious. I would not at all be surprised if you got unexpected results with something else close like that. I'm routinely taking out 40 or 50% instead of the normal 33%. And it doesn't take as long to find them.

    on another note, it would be useful if there was a Portuguese dictionary out there that only lists words that are different from Spanish
    Last edited: Sep 8, 2014
  10. Peregrinus

    Peregrinus Active Member

    Joined:
    May 27, 2014
    Messages:
    613
    Native Language:
    English
    Intermediate Languages:
    German
    Basic Languages:
    Spanish
    FSI has a Spanish to Portuguese course, and there are probably others. What I would do is find a list of false friends and a list of common/frequent words that differ quite a bit, and learn that well. Then I would study morphological similarities and differences, usually endings. With the latter knowledge I would adjust Spanish words if possible and otherwise just insert Spanish. After that I would just try to do a ton of extensive reading.
  11. Iversen

    Iversen Member VIP member

    Joined:
    May 1, 2014
    Messages:
    91
    Well, you can adjust the difficulty of the words on your lists (whether golden or not) simply by choosing different words. I include a number of guessable, but nevertheless relevant words when I feel that the other words in a 5-7 word block are hard, simply because I have found out that I learn more if the average level in a group is just at the level where there is some resistance, but not enough to block me. You can do the same thing when you choose the words for a goldlist.
  12. Iversen

    Iversen Member VIP member

    Joined:
    May 1, 2014
    Messages:
    91
    I have now checked my resultats after the Goldlist - 3 column - 'lazy' list experiment. and the results were more extreme than I had expected. First I took my original 4 goldlists with 100 words, covered the translations up and checked how many I knew by writing my best guesses at the menings - but it turned out that I mostly didn't even have a plausible proposal for a meaning. I happened to see the translation of one word, so I had to eliminate this word - the remaining 99 words could be divided into 18 known, 8 'something in that direction' and 73 totally wrong or no proposals. It was a fairly gloomy experience.

    Then I checked my 'do-nothing' list, where I had chosen 100 unknown (or only vaguely known) words about one month ago and not done anything after that. Here I overlloked one word, and the remaining 99 words could be divided into 17 known, 10 'not-totally-wrong' and 72 total failures. That is, practically the same dismal level as with the goldlists.

    Suitably disheartened I proceeded to check the result of my own three colum wordlists, which I had reviewed twice since I made them (plus one time where I just looked them through), and that was like leaving a black cave: out of 100 words I know 64, 10 were not too far off the mark and only 26 were totally lost or wrong. These results lie reasonably close to those I found in my Serbian experiments, and the difference could be due simply to the fact that I have spent much more time on Serbian than on Russian in the test period..

    By the way - I should probably give some examples of the middle category. For instance I had translated "заграница" as frontier area, but it means foreign countries in general ('udland' in Danish). And I had written 'tålmod' (=patience) for "терпимост", but apparently it means 'tolerance' . However larger discrepances than these were counted as fullblown misses.

    Now these resultats were fairly unexpected - I had a fairly good feeling about the goldlists and genuinely believe that I retained something from them, but it is pretty clear that I didn't. And I think I know why: I had done exactly as I was told: read the words aloud, but DON'T make associations. And association making is alpha and omega for my memorizing. When I saw the translations to the words in the goldlists again I could see all the golden opportunities I deliberately had missed, like identifications of loanwords ("шпалеры" = 'espalier'), derivations ("лодочник" 'boatman' from "лодка" boat"), sound associations ("срам" - 'shame') and other types of associations ("фокусник" 'magician' - i.e. a person who cleverly manipulates the focus of the spectators) . When I did my own wordlists I exploited these possibilities to their limits, and that's in my opinion the crucial factor behind my good results with my own layout and utter failure with the goldlists. Maybe others have a mind that functions on other premises, and maybe I ought to do a second experiment where I used associations to my heart's delight, but otherwise followed the rules. But for the moment I prefer spending my time on my own kind of wordlists which suits me like a glove.
    Last edited: Sep 16, 2014
  13. Peregrinus

    Peregrinus Active Member

    Joined:
    May 27, 2014
    Messages:
    613
    Native Language:
    English
    Intermediate Languages:
    German
    Basic Languages:
    Spanish
    Iversen, thanks for the further report. Re GL though, isn't a failure rate in the range of 70% for each distillation what is called for? GL calls for multiple distillations over a period of time to supposedly achieve the same effect of your own lists. So I am not surprised at the results from any one distillation.
  14. Iversen

    Iversen Member VIP member

    Joined:
    May 1, 2014
    Messages:
    91
    More destillations wouldn't help. I don't remember the information about the 70%, but my argument is a simple one: I put 4 x 25 words into a machine, and I got 17 (or 27) back at the other end - practically the same fraction as I got without doing anything apart from copying words to a piece of paper. Destilling the last 4 x 6 words more times can't change that. And as I see it the culprit is the warning against association making, which amounts to telling me that I should refrain from remembering those words. A word without associations is for me a non-word, a thing that I have turned my back on. And the question is why a successful language learner and teacher like David James/uncle Davey/Huliganov would issue such a warning - which reminds me of warning against putting gasoline into a car. Maybe he has grown disillusioned with the kind of associations used by memory artists (who typically have to remember known things in a specific order, where language learners have to remember more or less unknown items in no special order), but there are other kinds of associations which are more in line with the task at hand.

    I can't see that I have committed 'sins' that could explain the results. I have taking time to write the words by hand, I have recited the words as I was told, OK, I have added less grammatical information than I could have done, but at least marked perfect/imperfect aspect and gender of substantives where it wasn't clear etc, I have destilled roughly a third of the words away each time as suggested, and even though the time interval has been down to 10 days (because I wanted to finish the exercise well before Novi Sad) I have not been below that ... and my use of yellow A4 paper (one sheet per list) has permitted me enough space to keep the four sections well clear of each other. On the other hand, if you take a cynical stance I'm certainly the wrong person to test the goldlist concept, given that I have an alternative which I have been using since 2007 with good results. But even with this background I had expected to get results above the 'zero level' indicated by the do-nothing list, and I'm fairly frustrated that I have been looking at least 4 x 6 words words 4 times (= the last remaining words) without being able to remember them. But again, this must be due to the instruction that tells me not to use associations - an instruction which in my mental circuitry is translated into an order not to remember anything.
    Last edited: Sep 16, 2014
  15. Peregrinus

    Peregrinus Active Member

    Joined:
    May 27, 2014
    Messages:
    613
    Native Language:
    English
    Intermediate Languages:
    German
    Basic Languages:
    Spanish
    I think the reason why Huliganov warned against associations is because the GL method is supposed to be done in a stress free manner. As with any novel system, it is hard to separate out the components that are useful and necessary from those that aren't (and which might work against those that are). Perhaps indeed GL is more suited for those who don't have much ability with associations, although I like the the hybrid method I am using now despite not being good at same.
  16. Peregrinus

    Peregrinus Active Member

    Joined:
    May 27, 2014
    Messages:
    613
    Native Language:
    English
    Intermediate Languages:
    German
    Basic Languages:
    Spanish
    A couple more comments on the hybrid Iversen-Anki method I have been using almost every day to learn 30-60 words per day.

    -Since the words are already in Anki with their definitions before I type them in the spreadsheet, they are also thus alphabetized by Anki. Since I always put the definite article before the substantive, this has the unfortunate effect of making it too easy to guess the gender on the German columns of the Iversen list. So for the review the following day I generate a random number column for the list and resort and then redivide into groups to mix it up.

    -I have found it very easy to remember groups of 5-7 while doing the Iversen list part of the method and now have switched to 8 word groups, and occasionally 9 to not end up with too small of groups. This has worked out well, but probably only because I already know so many German words and their bases. If beginning with this method for a new language I would revert to 5-7.

    -I have also occasionally experimented with only doing the first day of Iversen review and relying on Anki afterwards, instead of day1:Iversen, first Anki review->day2:Iversen->day3: Anki again. This seems to work OK, but not enough to forgo the 2nd day Iversen list review, and again probably only works because of the reason I gave above of already knowing so many words and their bases.
  17. Iversen

    Iversen Member VIP member

    Joined:
    May 1, 2014
    Messages:
    91
    I would also have expected the sequence with a regular 2. repetition round and then Anki to function better than a sequence with an Anki round between the two rounds with wordlists. But there are several ways to do the first repetition round with wordlists. When I did the Serbian experiment I covered the translation columns in my original list and went through the words, and when something was awry with a word I uncovered the translation and copied it to the test sheet. In this way I didn't get more information beforehand than you do with Anki. But I don't think that this is the most efficient method: even with the same layout I think it is better first to look through the list and do some rememorizing whenever you feel there might be a problem. Ideally all words should then come out as known - which obviously isn't the idea behind a vocabulary test, but close to the situation you want to be in later. OK, the test wasn't a waste of time as I did some rememorizing during and after the test rounds, but for my part I see it as a wasted opportunity to recall a word if you simply can't remember it. You learn more from being able to remember it - with a bit of trouble. The people behind Anki clearly saw this in another way, but that's not my problem.

    Earlier this week I took around 100 words from an Indonesian dictionary and put them into a wordlist, and yesterday I did the first and so far only repetition. However it was clear that I remembered fewer words than I did in Serbian, so there I went back to an earlier layout, where I copy 5-7 translations blockwise to my repetition sheet and then reconstruct column with the original words - like you do in the first round, but without the first column (which I have on my original wordlist if I need it). This can be seen as an invitation to rememorize the words, using simple repetition associations of various kinds. After that I did a partial test round using the simple setup with just one coloumn plus comments at the problem cases. And this triple procedure seemed to solve the problem, but only at the expense of slower progress and more work. However I reckon that I can switch to the simple setup later.
  18. Peregrinus

    Peregrinus Active Member

    Joined:
    May 27, 2014
    Messages:
    613
    Native Language:
    English
    Intermediate Languages:
    German
    Basic Languages:
    Spanish
    One other thing I forgot to mention is that if I find the majority of words easy to remember on the second day's repetition of the Iversen list, I go through the entire list for a column and seek to know them all instead of in smaller groups before I write that column. Thus if I had 6 groups of 8 and just treat it as one big list and if I peak on word #20 then I finish going down the list and again from the top and only write (type) after I know them all. This gives me more time to "almost forget" which is one of the principles of Iversen's method.

Share This Page