Date: Mon, 17 Nov 1997 16:12:20 -0500 (EST) Message-Id: <199711172112.QAA01775@locke.ccil.org> Reply-To: David Moore Sender: Lojban list From: David Moore Subject: Re: logflash software X-To: lojban@cuvmb.cc.columbia.edu To: John Cowan In-Reply-To: Rick Nylander's message of "Mon, 17 Nov 1997 08:40:45 -0800" X-Mozilla-Status: 0011 Content-Length: 6584 X-From-Space-Date: Mon Nov 17 16:12:22 1997 X-From-Space-Address: LOJBAN@CUVMB.CC.COLUMBIA.EDU Rick Nylander writes: > I have found that using the default of 20 new words per session to be > _way_ too much for my little brain to handle. And I'm only in the "new > word review" mode. This means I must frequently turn off new word > introduction just so I can clear out some of the old stuff. I (perhaps > foolishly) thought that there is a good reason that the default is set > to 20. > > It would seem that, upon entering the "gaining control" mode, you'd be > juggling at least 100 words between the various piles (on a good day). Yes, but most of the words won't fall at that point, so it'll go fairly quickly. Most of the time is spent when words fall, because you have to go through them 6+ times at that point. Also, since you are less comfortable with the fallen ones, you're slower at finding them. So probably having 10 cards fall is about the same time as 80-100 cards not falling. You might want to tune the number of new cards based on both the number you think you can handle in a session and the amount of total time you can concentrate on it. > I've struggled through about 340 words this way, but I'm considering > starting over with a smaller number of new words per session. > > Or am I expecting to progress too fast? I think getting through 340 words is pretty good, especially without a class environment to reinforce the words in other ways. And you got all of that in under 3 weeks. > My wife recalls reading an article that said you should generally > introduce about 5 new words per vocabulary lesson of a new language. I > think I might be able to handle 7 to 10 with no problem. I used the ladder system while taking german, and was able to get up to rates of 60 words in the new word pile a day. But this involved 2 hour vocab sessions at the peak. I also modified the system slightly by splitting each dropback into more reasonable piles. If say 35 words fell out of the new words, instead of doing all 35 until I got 6 sets in a row correct (or 10 total sets), I split it into 4 piles of ~9 each where I did that. Also, if I went a day or two w/o doing cards, I'd disable adding new words as too many would fall out of the recall1 and recog1/2 stacks. If I hadn't done the splitting of large falling piles, I don't think I would have learned it nearly as well. Since I saw questions about stats for lojban logflash, perhaps someone is interested in this. Fere are some ending stats I had for a german vocab set. Note that `Day' is really session number. I didn't think to track real days in my stats until it was too late. I did have the UnderControl pile retesting words as well, but I just realized the stat program ignores that. : sdnp5 ; ./scripts/deck_history Cards.old vocab1 Number of cards in each stack. DAY # 44 45 46 47 48 49 50 51 UnderControl 1615 1622 1632 1635 1639 1641 1646 1652 Idle3 8 10 4 4 2 5 6 1 Idle2 10 4 4 2 5 6 1 6 Idle1 4 4 2 5 6 1 6 3 Recall2 4 2 5 6 1 6 3 1 Recall1 2 5 6 1 6 3 1 2 Recog3 5 7 2 7 3 1 2 1 Recog2 7 2 7 3 1 2 1 0 Recog1 2 7 3 1 2 1 0 0 DropBack 9 3 1 2 1 0 0 0 total 1666 1666 1666 1666 1666 1666 1666 1666 Percentage dropping back from each stack. DAY # 45 46 47 48 49 50 51 TOT Idle3 12.50 - 25.00 - - - - 9.71 Idle2 - - - - - - - - Idle1 - - - - - - - - Recall2 - - - - - - - - Recall1 - - - - - - - 6.07 Recog3 - 14.28 50.00 14.28 - - - 16.15 Recog2 - - - - - - - 1.48 Recog1 - - - - - - - 4.31 DropBack 22.22 - - - - - - 7.62 NewWords - - - - - - - 12.43 total 5.88 2.27 5.88 3.22 - - - 5.46 I also have some information on how many times a particular card got dropped, which I used occasionally to help identify words that were giving me particular problems, to try to find some other strategy for them. : sdnp5 ; ./scripts/drop_histogram Cards.old vocab1 Total drops: Times Dropped Count 18 1 9 1 8 2 7 4 6 9 5 9 4 56 3 96 2 234 1 489 0 765 Drops after most recently reaching UnderControl: Times Dropped Count 5 3 4 4 3 4 2 1 0 1652 -1 2 What tends to surprise me here is that the distribution is such that more cards were dropped 0 times than 1, and more dropped 1 than 0. My initial expectations were to have some sort of bell curve around 2. Drops after reaching undercontrol are a bit confusing since it's quite likely that a lot of those weren't tested again since the last time they made it. The 2 cards at -1 mean that 2 words never made it at all. I believe that I looked into a time indexed drop histogram and it was the case that words that fell out of undercontrol were more likely to fall again in lower levels, so for some reasons those cards were more difficult. -- David Moore | Computer Systems Lab __o UCSD Dept. Computer Science - 0114 | Work: (619) 534-8604 _ \<,_ La Jolla, CA 92093-0114 | Fax: (619) 534-1445 (_)/ (_) | In a cloud bones of steel.