From lojbab@lojban.org Sat Feb 5 08:58:53 2000 X-Digest-Num: 355 Message-ID: <44114.355.1925.959273825@eGroups.com> Date: Sat, 05 Feb 2000 11:58:53 -0500 From: "Bob LeChevalier (lojbab)" Subject: Re: LogFlash cheating At 11:31 AM 02/04/2000 -0800, Joirin Couwe wrote: >Originally this was going to lojbab only but then I >thought there might be some ideas to discuss in >general. The context is my LogFlash training and >participation to the research (as a subject). > >Am I allowed to "cheat" a little bit with LogFlash? Certainly. Though we want to know what techniques worked for you. >I'd like to drop almost all words after new word >review. I find that most words I did "remember" >correctly after review were only in short term memory >and don't succeed to advance to higher levels but >crash back to drop pile in the 1x query anyway. That is to be expected. The ones you know will stay up and not need to be practiced, and the ones you do not know drop back, and get extra practice as often as daily (if you keep up the regimen). Our experience is that usually if you get a word right on recog 1, you will 80-90% get it right on recog 2. Your percentage on recall 1 will be very low, similar to what you get on new words, and recall 2 is very high again (usually involving 1 letter errors in the Lojban). But dropping words back intentionally from recall doesn't help much because the drop back practice is in the easier recog direction. >There is still few words that I can remember at once >and correctly like sumti, big surprise :). So should I >take 'drop all first just to practise' as a consistent >behavior or just for the words I know or feel that I >really need practise. The program already makes you practice the ones that you don't know. If you somehow manage to guess a word, then you must know it well enough to get by. I would let those move up (it also gives you a small feeling of progress even if only 2 or 3 words a day make it up the gauntlet to "under control"). Thus, my experience with LogFlash on a typical session was: New words 40% (after review), drop back 50%, Recog 1 60%, Recog 2 95%, Recall 1 50%, recall 2 80%. When I did it there were no under control lessons until you got all new words done, and as a result I was down to 60-70% when I started my second pass through the words - people report that they have better success with the words being popped up for review each day in the under control lessons that start a few days after you get the words into that pile. In short I guess I am telling you that you don't need to intentionally get words wrong. If you don't know it, you'll get them wrong without trying, and if you get them right, you know them well enough for now. >Can I drop words with by my will at next levels too. >Or am I skewing results in some harmful way? I mean >you can't see whether I answered wrongly with >intention or didn't I just know the word. Nor can we see whether and how much you are trying to practice writing Lojban while using LogFlash. It'd be nice to know, but we are not at the stage where we need to control that variable. Consider that every time you practice a word, it is the equivalent of looking the word up in a word list and writing it. If you look a word up in a dictionary often enough and then use it, you learn that word. By "cheating" all you are doing is looking the words up extra times when you have successfully guessed them. Seems like unnecessary work, but it hurts no one. >The hardest part is to get all 40 words into memory. You won't. In my case, I didn't spend more than 10 seconds reviewing them, preferring to let the error practices teach me. Nora on the other hand spent 30 seconds or more on each word trying to develop a good memory hook. She got 60% of new words right, I got 40%. But over time it evened out, and indeed I got the words mastered before she did (but this was because on my 3rd and later passes through the list, I went up to 100 and then 250 words a lesson, so that by the end I was typing half the word list each day). The practice you are doing on the first pass through is just familiarization. Don't expect all the words to stick. If you get the percentages I did, then of 40 New words, 16 make it to recog 1, 10 to recog 2, 9 to recall 1, 5 to recall 2, and only 4 to under control. So 36 dropped back without my trying. Clearly you get to practice all the rest at least one time. >The secondly hardest part is to learn the words which >try to slip from memory all the time. But I'm not sure >where should I put one part: recognizing words in >opposite mapping, from english to lojban. I find that >*very* hard too. But it's not comparable to >tediousness of memorizing anything in the beginning. Nora found that memory hooks were valuable to her, even if very creative. "jganu" looks like "jagged", so she though of jagged mountains, which you draw with sharp angles, hence "angle". Her memory hook for "manci" was even stranger, but relies on American cultural lore: she likes bread, and eats it for a snack, and a common brand name for bread here is "Wonder bread". So "manci" to "munchies" to "wonder bread" to "wonder". On the other hand, I found that the tedious typing of the words was more help to me (and greatly enhanced my typing speed by the time I finished - I do not touch type, but hunt and peck at up to 35 words a minute) >Should I try various sizes of the new word bunch? Like >60 or 80? (How's that option changed? I haven't seen >that option since I generated my profile file.) You can go into the options screen at any time. There is an option to regroup new words, which I think is distinct from the option to drop all words back and start over. The time you spend per day will be roughly proportional to your new word lesson size. Double the new word lesson size and you double your total time. We found that the ideal size is one where the number of error words that you are practicing is between 10 and 25 on as many lessons as possible. Fewer than 10 and the multiple drills of the error words end up being more of a typing practice rather than memory drill. More than 25 or 30 and the error lessons get VERY tedious (I once had over 100 error words, after I stopped doing LogFlash for 3 days with a full drop back pile - that was a painful 2 1/2 hours, and I missed half the words again the next day). >Does the 'x1 some English/other English word/still >more x2' notation mean that some english, other >english, still more are different interpretations for >the concept? Probably. No. We are approximating the meaning of the Lojban word by giving multiple near-synonyms. The Lojban words do not necessarily mean exactly what any of the English words mean in all contexts. Hence "x1 is sad/depressed/dejected/[unhappy/feels sorrow/grief] about x2" means that sad, depressed, and dejected all approximate the meaning of badri, and that the others (in brackets) might be used in some contexts for badri. The ones in brackets may have a better word to translate them directly other than badri. > Maybe I should try to express >my question better: is the mapping from lojban l to >english e like the following (e1 = some English, e2 = >other English,...): > > +--> e1 >l --| > +--> e2 > | > +--> e3 > >which means l is some concept and mapping to English >could be any of the forllowing English concepts e1, >e2, e3 (conceptual difference between (e1|e2|e3) and l >is relatively small). > >Or more like: > >l --+--> e1 > | > +--> e2 > | > +--> e3 > >L means mostly e1 with e2 and e3 appropriate >translations sometimes. That is the distinction between using brackets or not. >There might be other possible versions for l to e >mapping but I'm a little bit too tired to think more >about those, now. > >Few other notes (maybe something I have thought to >include to my program): >- I'd like get some noise words into erroneous >practising. As you said, typing some word 6 times >might become just typing exercise. When you're asked >some other words in the meantime you just have to get >a good practise to write the erroneous word right. That one is not an option now. But the lessons where you have very few error words are also ones where you know the words pretty well anyway, so the practice is less important. Besides, the error words go back to dropback where you will practice them more tomorrow and the next day. >- I don't like to drop to the base level when I can't >recall english to lojban mapping. I have to type the >english name for the word many times before I have >another chance to exercise lojban. Typing in English >enhances English recognition and at the same time it >may harm lojban. At least it doesn't get it better. Yes and no. Dropping the word back means that you didn't know it perfectly, so practicing it more does not hurt (it also makes the lower lessons a little more pleasant because some of the words are ones you already know in the recog direction). I believe that you have an option on the options screen to not have recall words drop all the way down. We don't recommend this, because you will get very large recall 1 lessons and have to focus on too many words at once, and because when you get them right on recall 1, you only have one more chance to miss them before they are gone into under control, where you won't see them for a while. A word that you miss on recall should be practiced more than once before you file it away. In the long run, it doesn't matter what direction, as long as you get some recall practice, because the two way connection of the words to their keywords gets burnt into your mind. >- Lojban writing is checked too little. You either get it wrong or you get it right. That is all that is needed. If you mean that the 2 recall practices are too few, just remember that the under control lessons start a few days later (after you have had a few days to forget the words). And when you get the entire list don, you will be doing intensive recall direction practice on the words. At that point, you will only practice recognition when you miss the words, and the reverse of what you now say will be true. >- Some English words are really hard for foreigner. I >might have heard 'immerse' or 'submerged' before but >bath is much more common form. With this comment I >don't have a intention to say that this is necessarily >bad. My English vocabulary gets better without >noticing :). There are a lot of subtle benefits to LogFlash. %^) >- Some English cluewords are good idea. The problem is >many times to remember the proper synonym for some >clueword you recall. Yes. Technically you can edit a keyword in the review mode outside of the lesson practices, if one is too hard, but this is risky because if you get a duplicate keyword, you have no way to tell which Lojban word applies, and there is no "restore defaults" to straighten out a mess. In theory, you could replace English keywords by Finnish ones (or Japanese for that matter) that way. >- For some words I really feel like I've learner them. >For some I recall them immediately on the screen but >not in the real world, real situation. So memorizing >the word might hook itself to some specific >environment. > >- I don't feel I really know the words I have already >got to the top level. So maybe some real use is >needed. You won't know the words until after around 3 passes through all the words. But after the first pass, you will know enough that writing Lojban will not require painstakingly looking up every single word. >- I feel most of the words are easier to learn in a >form of some example sentences. The best thing would >be of course to have few "clue sentences" and then >generate questioning sentences when needed. Still, it >might be too hard, although I have seen somewhere >mentioning about program which generates random >sentences. This is a good way to review the words - make up sentences using them. lojbab ---- lojbab lojbab@lojban.org Bob LeChevalier, President, The Logical Language Group, Inc. 2904 Beau Lane, Fairfax VA 22031-1303 USA 703-385-0273 Artificial language Loglan/Lojban: http://www.lojban.org