Flashcards/Anki - Splitting up pronunciation and meaning

[markhavemann]

I'm not really sure where to post this. Hopefully this is the right place.

 

So for all of my Chinese language learning "career" I've always made flashcards (in Anki) in what seems to be the standard way, ie. One side with characters, one side with Pinyin and one side with meaning. As most people seem to do, I started out with two cards for each note, character->meaning/pinyin and meaning->character/pinyin. As things grew I realised that meaning->character was very time consuming and ambiguous and I deleted those (as many people also seem to do later) and now just do character->meaning/pronunciation. 

 

Recently I've also been adding audio to my cards which I think is a really nice supplement. 

 

Now, when reviewing it's easy to run into a problem of when to fail a card. If I forget the meaning it seems like an obvious fail (even if I remember the pinyin). If I remember the meaning but forget the pinyin altogether that also seems like an obvious fail, but what if I remember the pinyin but forget the tones? It's easy to push those through or choose "difficult" so the card appears sooner. The problem with that is that I still don't know the tones and will probably forget them again, but failing the whole card seems like I will be wasting my time. 

 

In the past few months I've been more interested in internalising tones and pronunciation rather than just remembering which tone a word should be, so Anki started to become less important for tones anyway. Recently though I have been thinking that it seems a little silly to put that much information on a card, and pronunciation should be something that could be drilled separately, without caring about meaning (in the same way as phonics is taught to kids, separately from word meanings). That's not to say that meaning isn't important, just that things could be more efficient if they are focused on separately. 

 

So what I have begun to do is to create two cards for each note, one has the character on the front, preceded by "pronunciation:", and on the back ONLY the audio file of the sound. So the goal is the hear the word in my head and say it, not to link it to pinyin or anything like that. These cards automatically get added to a separate pronunciation deck. 

 

Then the second card has character and audio on the front with meaning and pinyin (probably unnecessary) on the back , but the goal is ONLY to remember the meaning for these, and the card will not be failed if any part of the pronunciation is not remembered. 

 

I feel like with this kind of approach, characters could be added ONLY for their sound too. This might sound a little silly but there are lots of characters that I see in names of peoples and places, and would like to pronounce, but they are not so common that I've come across them in any meaningful way yet, so choosing a random meaning for them from the dictionary simply to make a flashcard for their pronunciation doesn't seem really useful. But learning just their pronunciation may be useful, and also make it easier to simply add a meaning in a related two character word later on. 

 

Has anybody done anything similar to this before? I'd love to hear any thoughts and ideas. 

 

 

[roddy]

22 minutes ago, markhavemann said:

if I remember the pinyin but forget the tones?

Double fail for thinking tones aren't a part of pinyin. Back of the class. 

[Wurstmann]

39 minutes ago, markhavemann said:

This might sound a little silly but there are lots of characters that I see in names of peoples and places, and would like to pronounce, but they are not so common that I've come across them in any meaningful way yet, so choosing a random meaning for them from the dictionary simply to make a flashcard for their pronunciation doesn't seem really useful.

Just wait until you come across a word containing the character naturally. You can also learn the names of people. Picture on the front, the name in Hanzi on the back.

 

I don't like having multiple cards for one word. My cards look like this:

For writing I have an extra deck:

[imron]

1 hour ago, markhavemann said:

but what if I remember the pinyin but forget the tones?

Then you didn't remember the pinyin, because pinyin includes tones and the sound shouldn't be separated into parts.

 

1 hour ago, markhavemann said:

The problem with that is that I still don't know the tones and will probably forget them again, but failing the whole card seems like I will be wasting my time. 

Whenever I encounter a card where I got something wrong (no matter how small, and regardless of whether it was meaning, pronunciation or character) I always stop and spend a couple of minutes going over in my mind what when wrong, and thinking about why I couldn't remember it, noting the difference between the correct and incorrect version and creating a false future memory where I come across the word and am unsure of it but then I remember the correct part.  I spend the time to make sure that next time I encounter this word I will not make the same mistake.

 

If it was only a small mistake I might still then pass the card, but if it was a significant mistake I would fail it.

 

This dramatically increases review times.  Across all reviews on a given day I tend to average maybe 30 seconds per card (this has been consistent over years).  Most cards are instant, it's just the mistaken few (and subsequent cementing the corrected information in my mind) that then blow out the average time.

 

However it's necessary to do this - my aim isn't to amass the largest possible deck of cards to review, it's to make sure I am learning everything to the point where I can recognize/use it instantly (or near instantly) in the real world.  In order to avoid spending all my time reviewing flashcards, I also tend to delete my deck when it starts taking up more than about 30 mins a day to review.  This allows me to focus most of my attention on words I have recently learnt, and I then use reading and consumption of other content to 'review' every other word - and if the words don't come up in the content that I'm reading or consuming, then it's not a useful word for me to know at the moment anyway so I don't let it bother me.

 

What you describe about setting up front and back sounds quite similar to my Pleco setup I wrote about here, except I don't use audio.  The only thing I would caution is not to use single characters - always try to create cards with multiple characters to avoid ambiguity.

 

1 hour ago, markhavemann said:

so choosing a random meaning for them from the dictionary simply to make a flashcard for their pronunciation doesn't seem really useful.

Seems useful to me - plus you learn another word - be careful though because a lot of characters have a different pronunciation when used in a name (especially a surname) so make sure you choose the one with the pronunciation from the name.  For words like this, you can also internally decide you don't care so much about the meaning and rather associate it with the name of the person/place from whatever you were reading.  You could also just use the full name on the card, rather than splitting it in to separate cards with separate characters.

 

 

[Wurstmann]

16 minutes ago, imron said:

However it's necessary to do this - my aim isn't to amass the largest possible deck of cards to review, it's to make sure I am learning everything to the point where I can recognize/use it instantly (or near instantly) in the real world.

We have a slightly different approach then. The Anki cards are there to remind me that this is a word that exists and it is pronounced like that and it roughly means this. I get a more nuanced understanding of the word over time from encountering it numerous times in the real world.

[imron]

I used to see things like that too - and to worry that I'd forget words if I didn't keep following the algorithm - but then I realised it ok to forget words, especially if they're not coming up in the content that I'm consuming.  I don't need reminding that a word I'm never seeing outside of a flashcard exists.  There are still plenty of other new words that are coming up in the content I'm consuming and I'd rather spend my effort on them.

[Flickserve]

Oh. I prefer a different approach for my cards (when I get off my backside).

 

Each note has an extra card.

 

This card is a card for listening skills. 

 

It plays the audio of a sentence without seeing any words or pinyin. Visually it is blank. 

 

it plays the sentence audio a few times (mine is five) so I get a few chances to listen.

 

I try to work out the whole sentence, each word and the whole meaning of the sentence. I would often shadow in my head the sentence. 

 

Then I reveal the answer. It has the chinese characters, the pinyin and the voice again. I look at the chinese characters first and then the pinyin. Sometimes, if I was interested in a certain word, I would put the word in another field below the sentence. 

 

The voice repeats another ten times. This one I shadow.

 

 I pick the sentence initially to repeat five times on the front of the card - probably if I get better , I can reduce it to three times. That gives me a chance to listen again to words I have missed.

 

I pick ten times for repeating and shadowing after seeing the written words in the sentence on the back - you can set it for more repetitions to shadow more times according to your preference. Or just press play audio again. 

 

I don’t ‘learn’ the pinyin. The pinyin is there to aid my shadowing and just check I have got the right pronunciation. How about tone? Am I shadowing the correct tone? Because the repetitions are quite quick and multiple, I can zone in on a part where I feel something wrong. If I think there is an inconsistency between what’s I hear and the pinyin, I check that word in Pleco for a more correct pinyin and it’s usually the tone that’s a problem. (My original pinyin is made from the Chinese toolkit in anki which might have an error). 

 

If there is a discrepancy, then I repeat the audio and shadowing again trying to correct the tone for that particular word. 

 

I found the lower intermediate sentences from chinesepod quite useful for this as the dialogue is said out loud with some emotion. 

[Flickserve]

I forgot to write my rationale.

 

For me, I tried the ‘show chinese word and try to remember the meaning and pinyin with tone’ strategy. Quite unsuccessful with that.

 

Realised someway into learning mandarin that my listening skills were quite poor so I did the setup as above.

 

Initially listen five times to the sentence audio - that’s to help give enough repetition to listen to every syallable possible and also make an attempt at understanding.

 

Show answer with hanzi, pinyin and translation. At the same time, the sentence audio is playing another ten times. Verbally, I try to read out loud the sentence for which the audio which is going on at the same time. 

 

Reading out loud may be subject to dissatisfaction. Not a problem as now you can start to shadow the sentence by using the repeat audio function in anki. You get another ten attempts (or more until you are fairly satisfied). I set it to ten as that makes quite a nice set of the sentence in a loop without having  to keep pressing the repeat audio button. Bear in mind that in “pronunciation best practice technique “, you might have to repeat a sentence over two hundred times for accent reduction. There’s no point in pressing repeat audio two hundred times, hence the setting of ten sentences (your preference of number) and making things reasonably automatic. 

 

pros:

a) flexible practice which you can semi-control the number of loops of the sentence. 

B) listen first multiple times to catch each sentence and the rhythm plus pace before seeing the hanzi and pinyin. And then that is reinforced.

C) checks your listening skills in pinyin. Does the pinyin and tone match to what you thought you heard? Sometimes it doesn’t - that could be a regional accent or connected speech. For myself, I don’t actively try to remember the tone as it is written - I try to remember the tone within the sentence maybe drawing out with the finger just like native Chinese do. 

D) was there a word you didn’t catch at all - Maybe the speaker said a word really quickly or that it was very softly spoken. Very quickly , you will be able to check because the sentence audio is repeating again

E) reading the hanzi with the audio- the aim is to start associating the hanzi with the right tone in the rhythm of a sentence. Pattern recognition 

F) get enough sentences for your own pronunciation practice if so inclined

 

cons:

A) need to create your own cards - time!

B) as a relative lower intermediate, I found chinese pod short dialogues made into decks more convenient. You can remember the general storyline and use that to help jog your understanding by context.

C) not sure how good it is for advanced learners with bigger vocabulary. 

D) it doesn’t practice your hanzi recognition straight off in the traditional sense.

E) you still need some exposure to the hanzi outside. For example, there is a podcast on nutrition and I had a look at a nutrition label to see if there were any hanzi that looked vaguely familiar. In fact, when writing this point, it suddenly occurred to me I could note those difficult words and maybe type those into a search engine for graphic images. Nouns would be better for that. If you want to practice more with those words within, you could do the technique by @tysond . Have a huge bank of anki cards with audio created from many different sources, search for sentences with the character, copy that card into a new deck and review accordingly to your preference. 

 

[Wurstmann]

@Flickserve

For making audio flashcards fast I'd recommend using subs2srs.

[Flickserve]

@Wurstmann

 

It's making the .srt which is the rate limiting step.

[Wurstmann]

@Flickserve Right. The fastest way to get subtitle files is downloading them from Netflix.

[realmayo]

I've got a vague concept of top-down versus bottom-up when it comes to discriminating and remembering sounds and tones as a learner of a language where the sounds and/or tones are fairly alien.

 

 

Top-down would be where you have a higher level concept that there are, say, four tones.

Self-testing, then, involves hearing a sound with a view to assigning it into one of those four buckets.

The brain's main focus is: which tone is this going to be?

 

Bottom-up would be where you hear a sound and describe what you hear.

The brain's main focus is: what am I hearing?

 

What I'm trying to say is that: certain audio flashcard or drill-down methods for me risk a side effect where I'm listening-to-assign.

I think that listening-to-assign is a great drill when you're first learning alien tones and sounds in a new language, because it forces you to concentrate on just one element of what you're hearing.

 

But I suspect that it's actually a barrier when your aim is good memory of a word's full pronunciation. Full, in terms of tone, stress, relation to neigbouring sounds, and initial & final or consonant & vowel components etc.

 

For Chinese: I think the target should be dictation. A simple flashcardy approach might be Q: Audio + English translation A: Write it out in pinyin (and characters too if you like) -- both for words and short sentences. (Not unlike Flickserve's approach, maybe)

 

This is flashcards as a way of learning and then practising, as much as it is prevent-forgetting.
 

 

Similarly, I wonder if the traditional method of seeing a character and testing yourself on the pronunciation is a flawed method of learning.

Because -- it's fake. I mean, when you read a book in Chinese, I don't think (maybe I'm completely wrong here) that you're sounding out the tone (or indeed the other sounds) of each character.

 

So: testing yourself on seeing a character and coming up with its pronunciation: you are not honing a skill that is fundamental to reading Chinese or to speaking Chinese.
 

[Flickserve]

2 hours ago, Wurstmann said:

Right. The fastest way to get subtitle files is downloading them from Netflix.

 

I don’t have Netflix.

 

if you do, we have a thread for anki cards from films.

 

https://www.chinese-forums.com/forums/topic/52517-subs2srs-anki-deck-index/?tab=comments#comment-404358

[Wurstmann]

Oh, didn't know about that. I will begin posting some decks then.