On flash cards, spaced repetition and sentence mining

Most language learners are familiar with flash cards, fewer learners have heard of SRS or spaced repetition and even fewer learners know of sentence mining. Googling these terms will clear things up for most people. What these activities have in common is that they are deliberate learning tools or tasks, that take us away from listening and reading or using the language.

I believe that a certain amount of such activity is useful in helping us to notice the language. The amount that we should do will depend on our interest or tolerance for this kind of activity. My tolerance level is low, and here is why.

I find it easiest to just listen to a language, since I can do it while doing other tasks. I enjoy reading, which I do on its own, or on the computer in order to easily look up and save words from texts that I am listening to.

I do invest some time in the deliberate study of words that I have saved. I look at them in lists, and I use flash cards. I also invest time in upgrading these entries in my database, by tagging them or editing the phrases that are captured in LingQ.  The deliberate review of vocabulary represents less than 10% of my language learning time. Since I have, on average 50 minutes a day to spend on language learning, this means 5 minutes a day on vocabulary review.

An SRS system pushes words at the learner based on an algorithm, which is designed to optimize retention in one's memory. I do not like having a steady flow of words pushed at me. I prefer to review my vocabulary when I want, and in the way that I want, and that can vary. Sometimes I review chronologically by the date when I first met them, sometimes alphabetically which enables me to see relationships between words which start the same way. I review my tagged lists sometimes, and I can search my saved words by root or component. Whatever I do, I can only review a small portion of the words that I have saved. I have saved over 20,000 words. Reviewing the words is more of a brain massage to make my brain more attentive, rather than an exhaustive effort to learn the words.

I believe that most of my learning of words takes place when I listen and read. When I read in LingQ I can see the previously saved words highlighted in yellow. Every time I import an item from Exho Moskvi in Russian, of about 30 minutes in length, I see 200 old friends. I often click on them while reading, just to refresh my understanding of them, especially in the light of the new context where I now find them.

I know and respect greatly the language learning activities of Khatzumoto, but I cannot bring myself to learn sentences, or as it is sometimes called, to do sentence mining. I prefer phrases. In LingQ  I review the automatically captured phrases when I review my saved words. Since I cannot review all my words, this is a bit random, but it all helps. I especially focus on phrases that embody grammar points, like cases in Russian. I know that all of this review is very much a matter of random sampling.

However, even though I spend most of my time reading and listening rather then reviewing words or sentences, there is no question that my vocabulary has grown immensely. I read with far grater ease now than I used to.

Comments

[Edward]

hi Steve,
Khatzumoto's site is so great! I want to import his article to LingQ. Do you think he would like to record the sound for me? Dose he know about you? Maybe I should just coax him to join LingQ. He could be a Japanese and English tutor, and provide his context with sound files in LingQ Library. Hehe. I'm so proud of my own design: the SRS Vocabulary Cards. I will email pics to you.

[igordesu]

Hmm. Yes. But remember that the purpose of an SRS is not to "exhaustively learn" anything. The purpose of an SRS is to retain that knowledge after you have already learned it. The only real reason that it is useful for language study is that it helps you retain words and phrases that you don't come across often enough when reading and listening. In a way, it helps simulate the "natural SRS" that native speakers have in their experience with being surrounded by a language for years.

[Edwin]

Khatzumoto did mention Steve's name in one of his Youtube videos. He didn't make the videos himself, but his friend did.

[Mitch]

Khatzumoto's site is great, and he has good things to say about your blog, too!

Have you ever contacted him directly? I'm sure the discussion would be most interesting, and would make good LingQ material. (As you did with Paul Nation.)

[Ana]

Well, I do like studying flashcards (really!), and I do feel they are worth studying, more than you would recommend. Of course, listening and reading are easier and more enjoyable, mainly in more advanced stages. But I think if one really wants to speed up learning, a good amount of flashcards every day are absolutely needed.
But I don't like complicated schemes like SRS. For me, it is a waste of time thinking about "how well" you remember every word. It is a somewhat silly attempt at controlling things which are not at all controllable, at least not at that level of detail.

[Mairo Vergara]

I am a big fan of SRS! The problem with sentence mining and SRS is that nobody can really do it the way you have to do it. It takes time and effort, and the results don't show up early. SRS is a long-term tool, it only shows its true power after some months of reviewing. Two months ago I created a Anki deck for my English studies and I just reached the 1000 sentences mark. I am starting to notice to results.... I know thing that "I shouldn't know", my vocabulary is really growing and those structures and many patterns are just cementing into my mind like magic.

SRS isn't just about memorization. It's also about reading. Although it isn't like reading a book or a magazine, it still is reading. Everyday you have to revise your sentences, read it aloud, etc. If if you keep Srsing, you keep reading.

I believe though that SRS isn't a method for everyone. It's too systematic, it requires discipline and diligence, and sometimes you just don't want to revise the damn thing. AS Katzumoto use to say, it is a hardcore method, probably not suited for the average learner.

Sentence mining is all about reading and listening. Actually, you mine the sentences from your extensive reading and listening. Lingq and Katzumoto method are not too different. In fact I find the methods very similar. In both methods you have to read and listen a lot and save words and phrases for revision. I just think that LingQ vocabulary reviewing system isn't so good as a good and organized SRS deck, specially if you are using Anki, which is such a amazing SRS.

[Steve Kaufmann]

I am willing to try anything. I will start saving long phrases on LingQ and see what happens. We are, at present, limited to 100 characters in terms of the lenght of phrase or sentence we can save.

We may could, in the future, change the nature of our SRS/flash card system but it is not a priority now. We do plan to allow people to export lists and import lists, so that our users could, fairly soon, be able to integrate their LingQs with their favourite SRS system.

In the same vein, we are working right now on allowing users to customize their default dictionary, rather than having to rely on Babylon which may not be ideal for their combination of target language and native language. This was identified as the most desirable improvement in our recent survey and should be available soon.

[Ana]

These are good news! I think a truly working dictionary is essential to make LingQing less painful and bring more people into it. (although I like to study lingQs, I don't like too much to create them...)
When I was studying French, I almost always had to go to an external dictionary, which is considerably more time consuming. And take into account that I could deal well either with Portuguese or English hints.
This resulted that I frequently found my French study time far less effective than my English study time.

[Chris]

I fully agree with what Mairo has said above. Another thing SRS are great for is learning Chinese characters, especially if you want to remember how to handwrite every character you can recognise. If I hadn't stumbled upon a system like Anki, I'm not sure I'd have had the confidence to learn to write them all otherwise. And the real beauty of them is, as said, the longterm effect it has. Also, no card is considered "remembered for good", so every vocab item (or character) you enter in will always be reviewed again sometime in the future. That's extremely handy.

[Katie]

I use SuperMemo flashcards along with LingQ (and other sources) for content. It scares me how well it works. I also used it to study for an exam in a class (nothing to do with languages), and scored 100%, with no anxiety. I had never experienced anything like that before. I don't know if the algorithm used is perfect, but I enjoy the discipline of daily review. It only takes about five minutes (per language). It's not the focal point. Reading and listening is a huge part!

[Steve Kaufmann]

Katie,

I believe you. However, if I look at my profile at LingQ I see that I have saved 220 LingQs the last 7 days, average about 1000 LingQs per month and have 23,000 saved LingQs. I prefer to go at them based on my criteria, and I wonder how an SRS system would handle this volume of cards.

[Katie]

Hi Steve,

In SuperMemo, if you know a word well upon its first review, you might see it again two or three days later. If you know it well the second time, you'll see it maybe a week later. And the periods between review will increase exponentially.

If you review a word, and you don't know it well, it'll go back into your cue to review again that day, and then it'll come back in one or two days, and you start over. For example, there was an expression in Russian that I had completely forgotten. It's now back in my cue, and I'm learning it all over again.

I divide my words into folders based on from where I got the content. There are some "topics" or folders that I am completely tired of, so I just delete them. It is possible to have to review hundreds of words a day, which is a complete pain. I also sometimes cheat, if I see it's in a folder for a topic I'm not interested in anymore. I don't worry about it so much.

This is far from perfect, but it works well enough for me enough to keep at it daily. I'm sure for quite a few, it wouldn't be useful at all. One problem I have is that your daily que doubles or triples and so on, based on how many days you may have missed. I have a work around for this, too, but it can be frustrating.

I enjoy reading about how you learn languages, because it seems to me that you have faith in your own intuition, which I think is important. Maybe it's perhaps your relaxed state that allows you to learn in a natural way. So sometimes I wonder if I'm perhaps too attached to SuperMemo. Maybe I should lighten up a bit.

Except it's too fun for me, in my geeky way. In a day or two, I'll try to write an explanation for how I'm using it to learn Italian through LingQ.

One immediate criticism might be that after as much time (in days) that I've spent learning Italian, I really cannot speak it (tho' I can understand more and more, especially reading), but perhaps it's a question of focus, as I'm learning four languages at once. I wonder if I'd be farther along if I'd focus only on Italian instead of Russian, Czech, and Spanish.

It's been a personal experiment I've been conducting, comparing from where I get the material, and how quickly it translates to some level of fluency, which is likely a separate topic. So far, in my completely unscientific study, Colloquial Czech is winning out over LingQ Italian, but LingQ Italian is more enjoyable, because I'm listening to "real" conversations, versus canned, hypothetical dry dialogues. Then again, perhaps over time, I'll be "fluent" in Italian, whereas with Czech, I'll only have memorized the textbook.

[Igor Efremov]

There is a guy, Rob Waring, ESL teacher in Japan, who studied spaced repetition systems, his study you might find useful http://www.robwaring.org/vocab/principles/systematic_learning.htm but everything else on his website is biased rubbish.

[Chris]

Anyone wanting a good, in depth explanation of how sentence mining can help, I'd recommend this article which I just came across:

http://www.glowingfaceman.com/2008/12/sentence-mining.html

He basically explains at length and in detail how sentence mining works and how to "mine" most efficiently for sentences, particularly once you get to the more advanced levels.

[Steve Kaufmann]

Igor,

I read the link quickly. I was not impressed. When I saw the following sentence I stopped reading.

6. We continue like this until all the words are learned.

That is simply not how we learn. We can look at some words once and learn them. We can look at other words 10 times and not learn them. One day these words may click. It is not something that any system can change. The brain is in charge.

[Steve Kaufmann]

Chris,

I had a quick read. I do not necessarily agree. Mostly I mine content for words, then short phrases that embody usage and grammar, and only occasionally for phrases that are actual sentences.

But more than that I listen and read.

The mining activity helps me notice what is going on when I listen and read. I learn most of my words by seeing them highlighted in yellow in link, old friends in new surroundings, that finally start to stick.

[Sarah]

Government documents now available on AudioBooks. Keep up with Obama's fast moving government! To download the latest government documents on audio, visit www.ativioaudio.com.