How can I scale my vocab learning?

[SteadyCamel]

I use Pleco for my flashcards. I now have about 2,600 flashcards and I'm facing a common problem in that just staying current with these using SRS is generally overwhelming. I've built the deck slowly so I know it pretty well, but with an 85% hit rate, I still have ~110 cards / day to review. I stopped adding new cards six months ago and my daily card load has hardly gone down.

 

My intuition is I should be adding more cards faster (scaling on prior vocab) but I'm doing exactly the opposite.

 

How have other people dealt with this?
 

[Xiao Kui]

I assume you're reading words in context (books, articles, tv shows, etc.) and that's where you're getting the words for your cards?  Even if you're already doing it, I would focus more on reading words in context than making flashcards.  You're just going to retain them better.  My own process is intensive/active reading, extensive/passive listening, and extensive or intensive/active for watching TV.  I tried Anki for awhile to keep up with all the new words I was learning in Thai, where my level is only intermediate, but it was hard finding the time for the cards and I think I'm growing my vocabulary better listening to podcasts, because the most important words I need to learn inevitably get repeated, sort of a natural SRS.  I learned Chinese in my 20s, Spanish in my 30s, and now Thai in my 40s. My memory not being as good as it was when I was younger made me realize I had to learn things in some kind of context in order for my brain to make the proper connections necessary for retaining them - context assigns importance to the new words.

 

That is assuming you are at least an intermediate or advanced level.  Now if you are a beginner drilling flash cards in order to build up sufficient vocab for reading, I suggest stopping with the flashcards and using a graded reader until you are ready for kids' books. Graded readers have repetition built in - if you still need flashcards for the graded reader words, make a smaller deck.

Regardless of your current level I think you have TOO MANY dang cards in your deck and need to focus on learning words in context, painful as it may be at first.  Now, if you are using the Refold method and sentence mining for your words, I don't know much about that, but I am skeptical of the Refold method in general, and especially for Chinese.  Sorry for the lame lecture, but seriously context is king when it comes to acquiring new vocab!

 

I still add new words in Pleco, just out of habit, but I don't have time to review my decks, yet I'm still learning new Chinese words that stick all the time.  The super low frequency ones have a low ROI and can usually be defined through context, anyway.

 

[SteadyCamel]

Xiao Kui,
Thanks for the feedback. Maybe I can set aside my large deck and create a new one that is partly mde up of new words and partly made up of words I need to re-learn. That would enable me to focus on some new vocab and free up more time for reading. 

 

I have a question on the reading. I'm currently reading Chariman's Bao articles at the HSK5 level. I can read them, but it's a bit slow going partly because there is always quite a bit of unfamiliar vocab and partly because about 20% of the grammar requires some thinking on my part. This level feels right to me, but it means I read less–perhaps two short articles a week. Any thoughts?

 

As for listening, it's a similar story, though my listening skills are lagging my reading skills. There are a couple of sources with video segments I work through on YouTube. If I spend half an hour a day, I get through about 30 seconds. It's slow because I try to figure it out without looking at the subtitles. I'm curious what your reaction is here?

Thanks!
 

[markhavemann]

1. Delete cards that keep coming up and you can't remember. If you haven't seen the word in context in six months then it probably Isn't important. You can always re-add later.

 

2. Don't use Anki's default settings: you might find the same retention rate with much bigger intervals, mine is set to hardly ever letting me see cards if I get them right (like maybe 4-5 times a year at most in the first year) and my retention is 95% (I just checked it now for the first time in ages and realise it's actually too high)

 

3. Don't be too strict on yourself: if you got a card wrong because you are tired, but would normally get it right, just pass it anyway. You can fail it next time if you really forget. 

 

4.1. Make better cards: only one (maybe two) definitions per word. For example 尊重 has the dictionary definition of: 

• to esteem
• to respect
• to honor
• to value
• eminent
• serious
• proper

but I would only put "(to) respect" on the back of a card, no need for the other stuff, it just makes it harder to remember and easier to fail, without adding much.

 

4.2. Make better cards: don't test yourself on pronunciation AND meaning on one card. Make separate cards for each.

 

Edit: @imron advocates deleting decks and starting fresh. I agree with this but I've also realised If your intervals are set right then there's almost no need because the cards you would want to delete will almost never come up to bother you. On the odd occasion that you see a card with a five year interval you could reasonably just delete it manually if you can still remember it.

[Insectosaurus]

I agree with some points already made, especially just adding one or two definitions. I would stick to one card per word though, and have never have and problem of a conflict between them. You can use mnemonic devices to help you remember that meaning. Mnemonics can be whatever you want them to be, as long as they help you remember the word.

 

Increasing the quality of the cards is generally a good thing. Abstract words can use images just as much as concrete ones. They don't have to be perfect. If you don't find a good image searching for the word in Chinese, then search for the English keyword. Or, if you have a good dictionary, check in what context the word is generally used and a picture of that. The important thing is to link the front of the card with the back, and images work wonders for that. My cards look like this.

 

Also, if you experience the load is still too high, just nuke the deck and start over. This time, put a limit on how many words you learn each day, perhaps ten. If you have a reasonable limit of new words per day, and don't fail a majority of them each day, you could have a deck with thousands and thousands of cards without reviewing taking up much of your day.

 

One important thing is the leeches interval. Don't have 0 percent. That might make sense for med school students and the like, but for a language learner it's torture. I have mine on 80 percent, but I can see arguments for everything between 10 and 90...

 

On 4/14/2023 at 3:31 AM, markhavemann said:

Don't use Anki's default settings: you might find the same retention rate with much bigger intervals

 

Would you mind sharing your intervals?

[markhavemann]

On 4/14/2023 at 4:26 PM, Insectosaurus said:

Would you mind sharing your intervals?

 

I haven't really messed with these in a while and I can't remember exactly why I made each value what it is, but I think the most important part is starting ease, and I also like the long intervals for new cards.

 

Edit: "Leech action" is missing in the screenshots above. I have it set to "tag only".

 

Since these settings filter out easy content really quickly, I'm not afraid to fail a difficult card a lot just see it every day for a week, because I know that once I can remember it, it will disappear into interval oblivion pretty quickly, maybe to reappear in a few months and restart the cycle if I've forgotten it.

 

On 4/14/2023 at 4:26 PM, Insectosaurus said:

I would stick to one card per word though, and have never have and problem of a conflict between them.

I can understand why most people feel that way but what's your logic for this? I think including pronunciation on meaning cards is a good thing (but not testing it), but completely isolating pronunciation is also part of what made my Anki time so much more enjoyable.

 

One of the main principles of Anki is testing for the absolute minimum amount of information on each card, and testing for pronunciation and meaning at the same time violates that. It seems inefficient to have twice as many cards but the easy stuff gets filtered out quickly enough for me.

 

Here are my retention stats. I haven't added many new cards in a while so my reviews are really low at the moment.

 

 

[Insectosaurus]

On 4/14/2023 at 12:36 PM, markhavemann said:

I can understand why most people feel that way but what's your logic for this? I think including pronunciation on meaning cards is a good thing (but not testing it), but completely isolating pronunciation is also part of what made my Anki time so much more enjoyable.

 

I add ten new cards per day. That means 10 new words per day. If I add two cards per word, that means 5 new words per day. If I had trouble remembering both meaning and reading I would go for two cards per word, but since I don't I just have never seen the point for me personally.

 

Our intervals are certainly vastly different. I have 2m 14m, then 2 days graduation. I'm not mathy enough to figure how vastly different they are in practice though.

 

Edit: I asked our friend ChatGPT, and it seems like with your settings a card will appear 7 times in a year, compared to 10 times for mine, providing you answer correctly. It's not a huge difference.

[markhavemann]

On 4/14/2023 at 6:42 PM, Insectosaurus said:

I add ten new cards per day. That means 10 new words per day. If I add two cards per word, that means 5 new words per day. If I had trouble remembering both meaning and reading I would go for two cards per word, but since I don't I just have never seen the point for me personally.

There is some point to this, I suppose after getting past that initial learning of the most common characters you stop really meeting words with characters that you don't already know, and the new ones are few enough that they aren't hard to remember.

 

Somehow having the pronunciation separate helps the meaning to stick more easily for me too though, and I do like the pronunciation card as simple pronunciation/shadowing practice since it's rare for me to not have audio with a card.

 

On 4/14/2023 at 6:42 PM, Insectosaurus said:

Edit: I asked our friend ChatGPT, and it seems like with your settings a card will appear 7 times in a year, compared to 10 times for mine, providing you answer correctly. It's not a huge difference.

I would have expected that to be much more, very interesting. 

 

Edit: accidentally posted this before finishing.

 

Ok maths time. If you do 10 new words/cards a day you will have 3650 cards in a year. Let's say it takes on average 3 seconds per card, that gives you 10 950 seconds, which is just over 3 hours of reviewing. So (assuming a correct answer every time), my settings are 21 hours in a year vs yours of 30 hours.   

[骏马的丕沿?]

Read books.

[Ledu]

@SteadyCamel

 

Use flashcards or not? It is kind of a difficult topic. Many experts, polyglots and successful language learners don't actively promote the use of flashcarding. Some used flashcarding and later seemed to regret it.

 

Last week I did about 8 hours of reading and 7 of flashcarding. I will continue doing flashcards for 1 hour or so a day. I work with one textbook at a time and make cards for each book. Seeing these words and sentences is another form of Srs. I use Pleco and do 4 tests set to random. It is basically Edsko de Vries setup on the Pleco forums except I set mine all to random not Srs.

 

I see flashcarding as reviewing. You could "brute force" your way and do 2 hours or more of flashcards a day. I don't care about forgetting them. I don't use Anki. I use a premade Hsk 4 deck and decks I make for each textbook I read. Currently I am reading a series called Ledu. It teaches you vocabulary based on the components etc. It is on Blcup. Maybe you could try that book or a vocabulary building book + flashcarding. 

[SteadyCamel]

Thanks,

骏马的丕沿, do you have a book suggestion? I presume you mean a normal book as opposed to a reader. I had considered this previously and then had a hard time getting one and decided I wasn't quite ready, but maybe now is the time to make the plunge.

 

Ledu. I'll take a look at that. 

 

 

[markhavemann]

On 4/18/2023 at 5:31 PM, SteadyCamel said:

do you have a book suggestion?

The classic first book suggestion is 《活着》, but you can also check out this thread: https://www.chinese-forums.com/forums/topic/58156-what-are-the-first-10-books-you-read-what-are-the-first-10-you-would-read-now-that-you-know-better/

 

[SteadyCamel]

Do you have any tips on where to buy an ebook version? I know this can be a tricky topic... I live in the U.S. and need simplified Chinese.

[markhavemann]

Chinese books are usually pretty easy to find. Just go to Google and type "《book name》 txt" and you can find a txt file and usually .epub and .mobi on the same sites.

 

The first search result on Google seems to have all three file types. I only checked the .txt file and it looks like the right thing.

https://www.qinkan.net/book/1200.html 

[SteadyCamel]

Wow, thanks! The "txt" makes all the difference. I was going down crazy rabbit holes trying to find this. You can't imagine.

[Ledu]

@markhavemann Thanks. I thought it would be easier. The epub looks great on my ereader. The text file comes out jumbled though, no matter which site I download from. At least the epub works fine. Now I know a few ways to get books in Chinese. You are an inspiration!

[markhavemann]

On 4/25/2023 at 8:31 AM, Ledu said:

The text file comes out jumbled though, no matter which site I download from.

That's probably because of encoding. Often these files are encoded in GB2312, but lots of readers will just assume UTF-8. You can download Notepad++, then use the bottom part of the encoding menu to convert to UTF-8 and save the text file, that should fix the problem if you do want to use the txt file rather than mobi or epub. 

 

I downloaded the text document from the link I posted above and you can see it's using Chinese encoding:

[Ledu]

@markhavemann  I appreciate this advice. Thanks again. I use Mac's though. Took me a little while but I was able to use TextEdit on Mac to open the file and change the encoding. I used the GB encoding option. Now it is on my e-reader and looks great and ready for reading with the Pleco app. I put this link just in case anyone is using a Mac.  

 

https://support.apple.com/guide/textedit/if-characters-arent-displayed-correctly-txted1028/mac.

[SteadyCamel]

The foregoing has been incredibly useful. I have started reading 《活着》.

 

However, I need to return to my unresolved issue of vocab, which is my single biggest hurdle both in terms of progressing in my studies and in terms of time management. Simplistically, I currently use SRS on 2,600 cards. I can't really maintain that and grow the deck. I've devised a system I will likely migrate to where I largely set aside the 2,600 and focus more on a new deck. (I'll maintain a separate study deck for re-learning old cards.) 

 

The thing I can't get over conceptually is how will this get me ahead? If I eventually need to know 10,000 words (pick your #), how is it that deserting my current deck for a new one will get me to 10K faster than simply adding to my existing deck slowly but progressively? If I set aside the 2,600, they will steadily fade… Keep in mind that the vast majority of time I spend on my current 2,600 is on the 20% that are in some sort of gray area that requires review to become learned. The other 80% don't really require any time to maintain. (I can cut them but doing so won't really solve much.) Am I going to eventually have to return to my prior set of words? Is this just self-deception?

 

One argument is that I will actually remember the first 2,600 better than I expect–even the 20% that aren't yet committed to memory–and that I'll simply learn more by studying new words. New words, the argument goes, will add familiarity with more characters, lead me to make fresh word connections, etc. In fact, perhaps it doesn't matter too much which words I study as long as they're not too obscure and as long as I use repetition to lodge them in my memory to some extent? Just spend a little time each day on vocab because you can't SRS 10K words?
 

[markhavemann]

Remember that SRS is just a tool. The final goal is to make these words part of you, and to eventually have your reading/watching etc. serve as the "SRS" for most of your vocab. 

 

On 4/30/2023 at 8:14 AM, SteadyCamel said:

If I set aside the 2,600, they will steadily fade…

It doesn't matter. If you don't see them in your reading etc. then maybe they are beyond your current level or outside of the scope of your life right now, and actually not important to know for the moment.

 

Imagine this scenario: You delete a card from your deck (or move it to an inactive deck). A few weeks or months later, you see that word in a text. You can't remember the meaning but you know it's that card that you deleted (or no longer study anymore), and feelings of regret and remorse well up from the deepest part of you for ever having doubted the power of Anki. With tears in your eyes, you add the word back into Anki, knowing you'll never be foolish enough to delete another card ever again. Every time you review that re-added word you are filled with feelings of shame. You look away whenever it comes up for review, too embarrassed to look directly at the screen and be reminded of that time your faith in the algorithm faltered. But at the same time, something strange has happened, that word is forever cemented in your memory. You get it right every time. In fact, you wonder if it's even necessary to even have the card in Anki at all anymore....

 

Jokes aside, deleting and re-adding a word because you see it later makes it really easy to remember in the future.

 

On 4/30/2023 at 8:14 AM, SteadyCamel said:

In fact, perhaps it doesn't matter too much which words I study as long as they're not too obscure and as long as I use repetition to lodge them in my memory to some extent? Just spend a little time each day on vocab because you can't SRS 10K words?

Yup! Also, the more words you learn, the easier it becomes to learn new words. Just keep going and things get easier. Efficiency is important but don't get too caught up in the details, just have fun with it.