Using Pleco Effectively

[jiacheng]

After talking with some other people who use pleco, it has occured to me that the array of options can be intimidating for the novice user, but extremely valuable for someone who has been using it for a while. While the Instruction manual does a good job of detailing all the functionality, we're pretty much left to set up our own tests, vocabulary sets, score files, etc. I was thinking that it might be useful to other users to discuss the types of tests and methods we use to incorporate pleco in our study routine. Perhaps the most popular use scenarios could be incorporated into the manual as kind of a "best practices" reference.

I'll go ahead and start the discussion by what kinds of tests I use. Hopefully others can follow up with the details the study methods and test types that you find the most useful:

Recognition:
Test type: self scored, show head initially, then reveal pron & def.
card selection: Repetition spaced, automatic scoring with its own scorefile:
General Usage: I have this test scheme set up to include all the vocabulary from the books and lessions that I have studied in the past. When I want to study a new lesson, I add in the new category into the larger pool. This way i get to acquire new vocabulary while also maintaining words that I've learned long in the past.

Writing:
Test type: self scored, show pron & def, then reveal head.
card selection: Repetition spaced, automatic scoring with its own scorefile:
Usage: I use the same pooling method here as I do with the recognition set. This ensures that I can write every word that I have studied. I will typically lag a bit on this test relative to the recognition, i.e. If i'm on lesson 5 in recognition, I might be on lesson 2 with writing. I don't use the character recognition since I want to ensure that I am using proper stroke order(this can be verified by tapping the character within the flashcard session).

Expression:
Test type: self scored, show def, then reveal pron & headword.
card selection: Repetition spaced, automatic scoring with its own scorefile:
Usage: Same pooling method. The category set may lag behind the recognition test somewhat, but is generally ahead of the writing. Often times a word may have several synonyms, so in these cases allow myself to come up with multiple possible answers in certain cases. If the revealed word is one of the ones I thought of, then I mark it correct. This is more of a verbal test, so I don't try to think of the specific characters, only the proper pinyin with tones.

Review Tests:
Test type: same as the corresponding reading, writing, expression.
card selection: random, scoring none. uses score file from it's parent test.
There are 3 tests that are basically for reviewing previously missed cards. They correspond to the reading, writing and expression tests. They use the same scorefiles as the main test, but do not record any statistcs. I use a filter to require last answer = incorrect. I find that these tests are good for quick short term acquisition of new vocabulary as well. As with all the tests, I use the "Review incorrect cards at end" and the "loop" option.

Tricks:
Change the order in which the simplified and traditional headword is revealed, showing first the set you are less comfortable with. This forces you to get familiar with both sets, even if you mainly concentrate on one.

I pretty much stick with the repetition spaced algorithm most of the time because what I care about is getting the word in my head and keeping it there forever, and this is the most effective algorithm for committing information to long term memory.

 

[ben_gb]

Hi,

Thanks for your posting. It is interest to read how others are using Pleco for learning chinese.

I am mainly focussed on learning to speak/converse and to read.

At the moment, I concentrate on testing myself using the english definition as the question (like your 'Expression' test), as I feel that this should help with learning the words for talking. I started out with using my own definitions, from text books etc, and I am gradually switching cards to use definitions from one of the dictionaries, if the entries are suitable.

One of the downsides of the dictionary entries is that they often include multiple meanings, but many of those are not commonly used, so it can be difficult to know which ones to learn (not all dictionaries seem to put the most common definitions first).

Also, I find that trying to learn from a complete dictionary entry can make it slower for you to learn. Most of the articles I've read about learning/memory say that it's important to keep the 'units' of information you are learning as simple as possible, just a few keywords. This means I don't show the Xrefs or examples on dictionary entries during flashcards. But I have sometimes found that having a few example sentences can be useful, to help memorise the usage, so in this case I use the OX dictionary definition, which includes examples that are not turned off by the settings.

Learning from the definition at first seems more difficult than learning from the headwords, but I have found it actually helps me learn to recognise the characters too...so much so, that I currently don't bother reviewing the 'headword'-first flashcards. The trick for me is to try for form a picture in your mind of the characters before they are shown. Then when you move away from that card, again try to picture the characters in your mind while pronouncing them.

I'm using the "frequency adjusted" scoring, and bumped up the "aggressiveness" factor to 5, with automatic scoring (but I enter 1 to 6 based on how well I remember), and if I get it right 3 times, then it's considered 'learned'.

I'm not sure this is quite ideal, as Im finding a lot of cards get repeated too quickly, but I'm a bit wary of messing around with the scoring controls too much as I don't want to ruin the scores I've built up so far.

I think the scoring stuff is definitely an area which needs to be improved, at least in terms of better descriptions about what all the controls do and how they interact. It may also useful to have a range of "preset" scoring schemes which clearly describe what they do and are maybe based on 'best practice' of flashcard memorisation (or maybe it is already?). The problem I feel is that I don't think my scoring setup is quite right, but I'm not really sure what can be done fix that.

One of the frustrations I've found (which probably applies to any flashcard system), is that knowing a flashcard well and always getting it right when it's displayed during tests, doesn't mean it will pop into my head when actually I need to use the word during a conversation. Even though when I'm testing/reviewing I usually say the definition and chinese words out loud, as well as picture in my mind what the word is (if it's a physical object), it still doesn't seem to be available for 'instant access' when I'm talking.

Not sure if there are any ways to get around this?

On the other hand, after I've used a word or phrase a few times in a conversation, then that seems to reinforce my memory of the word when I'm doing a flashcard test. So if I've used a word in conversation that I haven't yet seen the flashcard for, when the flashcard does come up, I have it remembered and up to the 6 score far more quickly.

By the way, I completely skip any of the writing/stroke order tests. I barely handwrite anything in English, and would write even less in Chinese, so putting lots of effort into learning to write is not worthwhile for me. I have previously tried learning to write characters and found that if I stopped writing for a week or two, they rapidly disappeared from my memory. The flashcard learning that I described above does mean that I can type pinyin fine, so I am able to send emails in chinese to friends etc (though grammar is another matter!!).

Anyway, hopefully other people will contribute to this thread and share their techniques!

Cheers,
Ben

 

[jiacheng]

Hi Ben,

I've also noticed the problems with the dictionary containing large numbers of meanings for the expression test. This seems to be especially cumbersome for single character words. I think your suggestion of using a custom card could be very effective for these cases. You can always highlight the word and tap it with a flashcard session and browse through the various dictionary entries as well. So far, for these cases, I don't try to memorize every definition, just the most important ones, or the ones relevant to the lesson I am currently studying, and I tend to be somewhat lenient as to what I count as a correct answer for those. If I come across another meaning in a lesson for that card, then I can just reset the score and be a little more strict as to coming up with the right definition the next time around.

I haven't purchased the Oxford Dictionary yet, so I pretty much just stick with ABC. I'll take a closer look at the demo entries and see if it could be useful.

It's very interesting that you don't bother with the headwords. You might try creating the headword test, but set the scoring very aggressively since you would already have gained a significant ability to recognize the characters from learning from the definitions. It could be a good way to solidify your ability to recognize the characters, but not add a whole lot more work.

While having several different tests seems cumbersome, it is somewhat mitigated by the fact that you have already acquired some knowledge of a card from other test types, and that having different test types basically solidifies your "directional" knowledge. In my case, the percentage correct tends to be much higher on the expression and writing tests than they are on the recognition (headword) test.
In fact, many times on these tests, I will go ahead and set the score to 800 new cards that I have never reviewed since having previously learned the components of a word, I can often write and express a new word after only having reviewed it in the headword test. And if I don't know it, then the score just gets set back to 100 the next day. This eliminates a decent amount of unnecessary work.

As far as not being able to get a flashcard right, but not having it pop in your head when you want to express it, I think doing translation exercises from lessons helps with this, but in the end, there's probably no substitute for real world usage and live interaction. It definitely solidifies your knowledge of a word that you have learned in flashcards. On the other hand, adding a word that you have picked up in conversation into your flashcard pool also further solidifies your knowledge of it and makes sure that you won't forget it later on.

Another thought is that the frequency adjusted algorithm could be leaving gaps in your long term memory. While i think frequency adjusted is great for short term acquisition of new vocabulary for next week's test, I believe that the repetition spaced algorithm is far superior for keeping words in your long term memory. With frequency adjusted, you could study a word several times in a day, and it would have the same score as if you studied that word several times over a number of weeks.

Frequency adjusted gives you no baseline goal for how often you should be reviewing your cards. I'm certainly not accusing anyone of being lazy , however, with frequency adjusted, there is no specific target, so it's kind of a best guess as to whether you think you've reviewed enough flashcards to maintain everything you've learned in the past.

Repetition spaced gives you a baseline target for how much you should review, and in my experience, it's a very good target. You could always tweak it by adjusting the aggressiveness as well. It is also more thorough than frequency adjusted in that if you are doing all your assigned cards, you are guaranteed to review a card within the next X days provided that you are testing all the cards assigned to you and not accumulating a large backlog. With repetition spaced, it's only statistically likely that you will study a card in the next X days, and that likelihood is reduced when you review fewer cards. Over the long term, it seems like a guessing game at best to know how much reviewing is enough with frequency adjusted.

As for me, I also skip the stroke order tests. It's probably very useful for a beginner, but once you get a good feel for stroke order, you can guess the correct order of new characters 99% of the time. If you're ever in question about any character, you can just look it up in the character info when you are doing other tests.

I do agree that writing is very tedious, however, I find that being able to write a character forces you to pay attention to the components, and in turn makes it easier to recognize it, especially when you see a character in a word context different than the one in which you originally learned it. You may find that if you add in a writing test, that it doesn't add too much to your workload since your miss rate will be low, having learned it previously from other tests. It will also likely reduce your miss rate of other tests, since you will be reviwing any given card more frequently.

 

[jiacheng]

Recently I've noticed that there may be a weakness in my scheme in that I don't review cards exclusively from the audio or pinyin. It could be that this type of test could significantly improve listening ability. I'm experimenting with some adjustments in my writing test to reveal the English definition after the pinyin & audio and separately from the headword. I suppose with these adjustments, calling it a dictation test may be more accurate.

So far, I haven't been counting it as a miss if I'm not able to come up with the proper word in my mind from just the pinyin as it could be quite difficult to do out of context considering the large number of homonyms in Chinese. I may start counting it as a miss in the future and just be more lenient if I happen to know a homonym of a particular word.

 

[ben_gb]

The reason I used the frequency-adjusted scoring is because it seems a bit more flexible as to how I can fit the flashcard reviews/tests into my day.

I find that I can usually spend 20-30 minutes at a time, maybe 3 or 4 times a day. This seems to translate to about 40-60 cards at each sitting, so in total maybe around 200 cards a day (including new ones and reviewing old ones). However, sometimes, I may have to skip a day or two if I'm not in a position to do the tests, but this doesn't mean I then have time to do 600 cards on the 3rd day.

My current full card list is around 3300 words, and of this I have "learned" about 700 (ie got right at least 3 times), and I also use 'limit unlearned' set to 20, so I don't get overloaded on new cards.

At the moment, I do have a goal which is to try and get through these 3300 words by September 14th, because I am starting a new chinese course then. That means I need to be "learning" about 50-60 a day. However, I do already know many of the words which are in the 3300 list, so actually the number of completely new words is a lot less.

I just tried switching to spaced-rep, but it only offers me 5 cards for the first session. Though I notice if I bump up the 'limit unlearned' to 40, it increases the batch to 25.

Maybe I'll try this for a day or two and see how it goes.

But this is a good example of where I think the scoring area needs greater explanation, as I kind of know what I want to achieve and the time I've got to spare, but I am not clear on what settings to change in order to get there in the most efficient way.

Regarding the testing on the pinyin. I am gradually getting to the stage where I can ignore more and more pinyin, so I wouldn't do any testing just on that, as I feel it is sort of wasting my time learning to read pinyin words which I will never use. I feel it's quite important that when you hear a word spoken, that you try to picture the chinese characters in your mind and not the pinyin, so I think that anything which reinforces pinyin in your mind would best be avoided! Also as you say, there are so many homonyms it's probably not worth the trouble. Much of the time, the context tells you what the word should be.

Using the audio is probably better than reading the pinyin, but it's still not ideal as again there is no context, and also the audio can't take into account all the possible tone changes which might occur when the word is used in a real conversation.

For improving listening, there's really nothing better than watching chinese TV programs. The key is to pick shows which a. you are interested in, b. are quite visual, and c. are also quite focussed on a particular subject area. For example, I've been regularly watching food and travel shows (from Taiwan). Each show is different (so not boring), but often the same words and phrases are used from show to show, and because the camera is showing what they talk about, you can often guess what the word means. I've found that, after spending many hours watching these sorts of shows, I am picking up more and more words and phrases, and can sometimes follow the conversations even if I don't understand every single word. I often watch with Pleco at my side, so I can quickly check any important or frequently used words, and then add them into the flashcards.

Ben

 

[jiacheng]

Personally, I don't use the "limit unlearned cards" option. The reason is that the vast majority of cards in my master pool are learned. Another reason is that i think most people tend to underestimate their own memory capacity. I have often surprised myself when pulling in a new list of 60-80 cards into my master pool. I find that after reviewing them to the point of being able to get each one right at least once (using the review incorrect cards at end + loop option), that the probability of remembering one of those new cards on the second day is usually over 50%. Once I've remembered it correctly after a night's sleep, and boosted the memory with a subsequent review, the probabilty goes up even further that i will remember the card when it tests me 4 days later.

It is definitely annoying to get a huge backlog that seems overwhelming, but the reason reason that these cards are in your queue to be tested is that according to the algorithm, you are statistically likely to forget these cards. I have built up a backlog on several occasions and I have a few tricks for dealing with it.

1. Use a score filter to test the lowest cards first. These are the most fragile memories since they are newly acquired and require more frequent reinforcement than cards you've been studying for a year. Assuming it takes a fixed number of seconds for each card, this is really the most effective use of your time since getting a card right one time effectively reduces the testing frequency from once a day to once every 4 days. Once you've knocked out the lower scored cards, you can unset the filter, or raise the filter threshold and try to do as many as you can of the cards with higher scores. And if you still have a backlog, at least they are all higher scored cards, which you are less likely to forget anyway.

2. Don't pull in any new cards into your master pool until you can get the backlog back down to 0.

3. Tweak the scores of cards to more accurately represent how well you know them. If you bring in some new cards, set the score to 800. You'll still be tested on them right away, so if you happen to have picked up that word in passing, there may be little need to study it. This will reduce your workload in the coming days. If you are just recently starting to use pleco and are building your pool based on material that you acquired years ago, then you can simply bump up the scores for those since you would not likely need to review those very often.

4. (I haven't personally had to use this one yet) Try temporarily removing some older material from your master pool. Keep in mind, however, that cards with high scores don't really contribute that much to your daily workload, so you might not gain that much from pulling them out.

In reality, having a backlog is not a terrible thing. As long as you're able to shrink the backlog somewhat day after day, then at least you're moving in the right direction.

If the backlog continues to grow and be unmanageable, then I would come to one of two conclusions.

A. you are studying with too much repetition and you need to increase the aggressiveness setting.
B. your goals are too aggressive for the amount of time you can allocate to learning new material. Assuming you have found your optimal aggressiveness setting doesn't need adjusting, then by building up a backlog, something has gotta give. Either you can't learn the new material, or you're gonna start forgetting the old material.


Some observations:
By limiting the unlearned cards, it may pull the unlearned cards too slowly and it may not pull in the cards that are most in need to review. Since most of your cards are unlearned, you may pull in 5 new cards in a particular session. I could be mistaken here, but even if you happen to get them right one time, it will probably not pull those same 5 cards in the next day since they are still considered to be unlearned. This is problematic because new memories are fragile. You need to reinforce those memories over the next few days to ensure that your initial effort to acquire them doesn't go to waste.

Assuming that I correctly understand the 'limit unlearnd cards' option, it seems to be counterproductive. The reason being that if you forget a card, then it again becomes unlearned. Yet by using this setting, you limit how many of these you can study. and these are the words we need to spend the most time on to be able to commit them to long term memory.

 

[ben_gb]

Well, I suppose I can try turning off the 'limit unlearned' function and see what happens.

My understanding is that it limits the number of cards that you've reviewed at least once but have not learned yet (by whatever definition of "learned" you use). So, if it's set to 20, then it will only introduce new, unseen cards once the number of seen-but-unlearned cards drops below 20.

However, what I'm not sure about is if you don't use the "limit unlearned", does the system just keep throwing new cards at you, until you've seen them all at least once. And if so, how are already seen cards prioritised.

[...time passes...]

OK, I just tested this. Turning off "limited unlearned" presents me with over 2500 cards in the queue!

So it does seem that "limited unlearned" can be used as a sort of tap, to let you limit the new cards being introduced to a trickle, rather than a flood.

The other thing I've noticed is that so far, since switching to spaced-rep, it hasn't presented me with any of my older cards for review. So far it's all been new ones.

Poking around a bit, I also notice that the scores on some cards are now in the 102401-204800 range, which seems to imply that I won't see them again for 3-5 years??? This is assuming the 100 points per day (but maybe I've misunderstood what this means). Seems a bit excessive, as I'm sure I don't know those cards THAT well.

 

[jiacheng]

After reading your post and then checking the manual, it seems that I misunderstood how 'limit unlearned' works. It is certainly a lot more useful than I orginally thought. I would still be careful with it and don't set it too conservatively, otherwise you might be placing an artificial limit on how quickly you are able to acquire new material.

It would seem that the scores you have accumulated by using frequency adjust are not quite as useful in the repetition spaced context. You might try making a duplicate of the scorefile and then cutting them down by 1/2 or 1/4 using the batch function. This could be tricky since it doesn't support adjusting them by a percentage. You would probably have to select a range, say 20000-40000 and then set all the cards to 6000. You would want to start out with the lowest ranges first, otherwise you will end up lowering the scores multiple times for cards with very high scores.

I'm not sure if the score progression with frequency adjust is the same as it is with repetition spaced, but if it is, having scores as high as yours could mean that you're way over studying material that you already know, thus leaving less time to acquire new material. In a repetition spaced context, it would take years to build up scores as high as yours. and the time between reviews roughly doubles every time you get it right, so if the time to the next test is 2 years, you've probably been studying that card for around 2 years. Of course the details are a little more complex and score progression gets adjusted based on the difficulty rating, but it's a useful way to think about it.

To get to a score over 100000, you would have to get around 18 correct views in a row, a process that might take place over 3 years with repetition spaced. If you got to this score over say, 6 months with frequency adjust, then you may be reviewing 6 times the number of cards necessary to remember them. If this is in fact the case, you could be devoting a whole lot more time to learning new material.

 

[ben_gb]

Hm...this seems like an opportune moment for Mike Love to chip in and explain how all these score mechanisms interact with each other!!

It does seem that the scores aren't compatible between spaced-rep and frequency adjusted, especially when you have so many other variables such as difficulty factor and aggressiveness to take into account too.

I'm going to try starting a new profile/score file just using spaced-rep with a section of my cards and see how that works over the next few days.

Thanks for your hints and suggestions, though maybe this has taken your thread a bit off the original topic!

Cheers,
Ben

 

[jiacheng]

After I adjusted my writing test to show the pinyin and auto play the audio first, I think that I'm gonna stick with this method for a while. My main goal is to associate the sounds with the characters as well as the meaning. Showing the pinyin is just to clear up any ambiguities in case I didn't hear the audio clearly. I'm really hopeful that by improving my ability to identify words by the sound out of context, that this will improve my listening ability. I'm thinking that it could be troublesome if I start to accumulate a number of homonyms in my deck, but so far with around 5000 cards in my pool, it hasn't been an issue.

ben, i wasn't sure what you meant exactly by getting away from pinyin. Are you referring to how some elementary texts that place the pinyin under the characters? If that's the case, then I would agree that it's a crutch to try to pull away from. But in general, I think pinyin is an excellent tool to use, just as someone might use the pronunciation codes in an english dictionary. Granted, it's rare that someone would need to read an entire text in pinyin, but I do think that to be able to conjure up the meaning and write the characters based on the pinyin would probably be fairly equivalent (or at least have a strong correlation) to being able to understanding a word when spoken.

I notice that non-native english speakers often ignore the pronunciation code of a word and end up trying to pronounce it the way it's spelled, which is obviously a mistake since spelling and pronunciation are often inconsistent in english. So I think it would also be a mistake to ignore the pinyin since if you are unable to come up with the pinyin with tones for a word, then it's likely that you are unable to pronounce it correctly, provided that you know pinyin well.

 

[ben_gb]

I do agree pinyin is definitely useful when you are initially learning the correct pronuciation of a word...there really isn't any other way to learn that, unless you use an alternative, like zhuyin, or use the audio (perhaps the ideal way).

For people who've initially learned alphabetic languages and been brought up in an "alphabetic world", there does seem to be a bias where your eye almost subconciously gets drawn to roman text. So, for example, if you read a textbook which has chinese on the top line, with pinyin underneath, unless you physically block off the pinyin, your brain will pull in the roman text out of the corner of your eye, even if you are trying to concentrate just on the chinese characters.

You can see this too if a page is flashed up in front of you containing a mix of chinese and alphabetic words. Your eye automatically gets drawn to the alphabetic parts, and the chinese parts tend to be a bit of a blur unless you force yourself to concentrate on them.

A lot of this seems to be due to familiarity - you've got decades of learning in English to counteract - and as you become used to reading chinese, and learn more characters, this effect hopefully will lessen.

There was a discussion about this some time back on another forum, and the general feeling was that to help counter this, you should try and wean yourself off pinyin as much as possible, especially for reading longer texts (but of course you already need to know a reasonable number of chinese characters, or you'll be overwhelmed). Also, for testing listening, it is better to use audio and the character itself, rather than showing the pinyin.

I suppose what you want to achieve is when you hear the sound, your brain immediately associates it with the object or concept itself, rather than going through the mental translation of sound -> pinyin -> character/object. When you say something, you want your brain to go directly from concept/object -> sound. And, when you read something you want your brain to go from characters -> object/concept and maybe the sound too.

Using your writing test as an example, I'd probably have it just playing the audio, then I have to write the word, and only see the pinyin afterwards. If I saw the pinyin with the sound, then I wouldn't be sure that I was not being given a nudge by seeing the pinyin.

Some people do put lots of effort in trying to read something like this:

Zhè shì yī ge lǎo gùshi.

Shìqing fāshēng zài 1946 nián xiàtiān. Nà nián wǒ jiāngjìn shí suì, zhù zài Sìchuān Chéngdū jiāoqū d Bǎihuā Qiáo. Zhōngguó Kōngjūn Tōngxìn Xuéxiào d jīdì zài nàli. Wǒ bàba shì nà ge xuéxiào d jūnguān.

But, I feel it would be better to use the chinese character version and read from that. I already know most of the characters for this text (and I can understand what it's saying from the pinyin), and those that I don't I can check individually. It will be slower initially, you will speed up as you get more used to it. Chinese also doesn't have spaces separating words, so if you learn from this kind of text, it's another area you need to get used to with real chinese text.

In summary, I think the issue is that once you get beyond the beginner level, everytime you use pinyin, you are spending less time with characters, so you are slowing down your becoming familiar with them. To help this, it's good to find ways to minimise the use of pinyin. But of course, everyone is wired up differently, so what works for me, might not work for you!

For me, remembering the tones is one of most difficult things. But using Pleco's 'coloured tone' feature on the characters has helped a lot with this. Now, with many characters, I don't need to look at the pinyin at all.

Ben

 

[mikelove]

We've actually been thinking about letting people make multiple flashcards for different "sub-definitions" of a single entry; pick out a few #s and just base the flashcard on those. This would likely only work in the ABC and ACE dictionaries, but could be a nice way to limit the amount of text to review. We could also just open things up and let people edit the text of definitions in cards, but I'm a bit uncomfortable with that since it makes it harder for us to manipulate them later on (e.g. for duplicate-checking).

The scoring options definitely need to be explained better, yes, though I think also the system needs another redesign - 2.0 is too close to 1.0 in that it treats repetition-spaced and frequency-adjusted study as two different things, what we really need is something that offers the guaranteed progression of repetition spacing (i.e,. you'll see a card at least often enough so that you're unlikely to forget it) but with the flexibility of frequency-adjusted (so if you feel like cramming / studying more, you can). The overall architecture of the 2.0 flashcard system I'm very happy with - it's an insanely powerful / flexible system in general - but scoring / card selection is kind of a weak point.

The "batch" function actually does support percentage-based score adjustments - the popup list for that is after the field where you enter the score, normally says "pts" but can be switched to "pct." The scoring calculations for frequency-adjusted and rep-spaced actually are the same, but there's a bit of a distortion effect in frequency-adjusted sessions because the software's trying to make sure it doesn't show you the same card twice in a row (or too many times in one session) - made worse if you've enabled "don't repeat cards in session": even a card which according to its frequency should be coming up only 1/1000 as often as a brand new one will end up coming up a lot more often because you don't have any other cards left to review, and that'll mean its score increases much faster than it would in rep-spaced.

Oh, and one thing hardly anyone seems to use: the Category options in "Card Filter" can be very useful for temporarily blocking out some cards you don't want to learn - just add all of those cards to a special "Buried" or "Low-Priority" category and filter them out of your sessions for a while.

 

[jiacheng]

Mike:
One way to do this might be to just allow the different selections of a definition to be highlighted, expanded or collapsed to an icon by clicking on that section and have the state be saved with the card. Maybe another button could expand/collapse all of the definitions for the entry.

ben:

So far I've been looking away from the screen and letting the audio play, and try to come up with the word in my mind. I then check the pinyin to ensure that I heard the audio correctly. The next step is that I reveal the meaning. If it was in fact the meaning that I remembered, then I will try to write the word. So far, homonyms haven't been much of a problem. Obviously they'll be more of an issue with single character words, so I just try to come up with as many words that match a particular sound before I reveal the definition.

One problem with the audio is that it often doesn't sound quite accurate in some cases, such as 2 consecutive 3rd tones, or words that use the neutral tone, so it's good to at least be able to refer to the pinyin in those cases.

I have noticed the tendency for my eyes to be drawn to the pinyin in the example sentences, or on some childrens books that use it and it is definitely a distraction. In the case of a book, I cover it up with something. Zhuyin would probably be a good solution for this and not terribly difficult to learn if you already have a good command of pinyin. Maybe there could be a flashcard option to force zhuyin, similar to the option we have to force simplified & traditional currently. I just wonder, once you start becoming more familiar with zhuyin, would your eyes start being drawn towards it? Or maybe that would never happen since most folks would likely have learned zhuyin at the same time or well after they started learning chinese characters.

 

[ben_gb]

Mike:

I've been using rep-spaced for some days, after previously using freq-adjusted.
What I've found is that while on the one hand, I felt freq-adjusted was in many cases showing cards that I already knew quite well too frequently, but some new-ish cards were not shown for ages.
On the other, rep-spaced has so far been more consistent/predictable, however, I feel that for new cards, it's not showing them frequently enough. I've reduced the aggressiveness (and I use the scoring buttons), but haven't quite achieved what I am aiming for.

I think what would work is if newer cards could be shown 2-3 times within a day, say if you give them a score of 4 or lower. Or another idea is if a card is "unlearned" (based on whatever definition the user has chosen), then the difficulty factor is automatically adjusted so the card gets shown much more frequently. After it goes over the "learned" threshold, then it switches to the regularly difficulty/scoring method.

At the other end of the scale, I wonder if it might be possible to have a maximum score setting? Seeing that scores could go so high so that cards are not shown for a year or more does make me think I'll likely have forgotten it by then. I would generally like to make sure I see any card at least every 4-6 months or so.

jiacheng:

Yes...I agree on the audio problem. The whole tone change stuff is a bit of a nightmare! I've ended up just using the audio occasionally to help me improve my pronounciation, especially on some of the harder sounds. I am fortunate, though, in that I currently have access to a lot of spoken chinese resources (TV, friends etc), so I tend to use those to improve my listening.

Regarding Zhuyin, I guess this could help getting around the problem of one's eyes being biased towards alphabetic characters in the flashcards, maybe not longer term though, as it's usually much simpler than the character it's representing, and perhaps your eye will be drawn to the less complex shapes?
One of the annoyances if you want to use zhuyin elsewhere, like in textbooks, is that it is normally displayed on the right-hand-side of every character, so as you become familiar with it, your eye tends to cheat a bit (especially to get a hint on the tone, which is quite obvious even if you don't know the rest of the zhuyin characters), and it's not really possible to block it out.

I have found that after I've become VERY familiar with a character, then my eye will not be so easily diverted to see the pinyin or zhuyin, and I can even mentally 'block them out' when reading some texts.

Ben

 

[mikelove]

jiacheng - that's a great idea, though I still think we'd need to do something more for it on the duplicate-handling side as well - might just be a matter of improving the duplicate card prompt, though (as we're already planning to do), letting you browse through the cards that the system thinks are matches and see them rendered with their current hidden / visible section settings before deciding whether or not to create another card. Along with perhaps also adding an optional dialog that comes up after creating a card to let you configure which sections are initially visible (and maybe changing some other settings, adding it to additional categories or altering the initial score / difficulty).

Once we added the option to customize the pronunciation display mode in flashcards, you wouldn't even need to use Zhuyin since you could just hide the pronunciation altogether with the "None" option - there'd have to be a way to adjust it separately for the pronunciation and definition sections of each card, though, since you'd presumably still want it visible in the former.

ben_gb - good idea about a maximum score setting. With rep-spaced, if you set the points per day to something lower than 100, say 50, and use "Hours" instead of "Calendar" as your "day type," that would cause cards with the lowest score to come up more than once a day. Certainly another argument for a new scoring / card selection algorithm, though.