Beli Tsar wrote: ryanheise wrote:
Beli Tsar wrote:Kauffman talks about truly enormous levels of repetition with his mini-stories to get those high-frequency words (among other things) in during the early stages. We don't all have the patience for that; cutting the repetitions by 3/4 or more and doing a couple of minutes of SRS a day would get similar results.
Some people may be able to breeze through their daily SRS queue in 2 minutes, while other people may struggle to recall each card without the added context you get when just engaging directly with the story.
That's totally fair. While I do wonder if many problems with SRS are more to do with the difficulty of learning to use Anki itself well than with the words, there's no question that it can be harder for some people. Indeed, just changing the language you are learning can make it much harder too.
I should have put my point in a more balanced way - Kaufmann is too dismissive of SRS Frequency lists: he gives the impression they do not give added utility. They do, even if it's not always for everyone or easy to use. Done right, for a lot of learners doing a lot of languages, they make the initial period of language learning easier. I've learned a lot from Kaufmann in the past, but this feels unhelpful.
One thing to grant Kaufmann is that he does preface his views on SRS by saying things like "I personally find that ..." or "For me, I ...", and he says that if you enjoy SRS, then it could be worthwhile for you. I think the premise of Kaufmann's YouTube channel is that if you're interested to learn languages in the same way that he likes to learn languages, you can keep listening to his perspectives. He built the LingQ platform around the way he personally likes to learn languages, but he does at least acknowledge that other people should do SRS if they are so inclined.
Beli Tsar wrote:
ryanheise wrote:It is also almost certainly the case that you do not actually need flashcards for the most frequent words. If you consume 10,000 words of input in a day (e.g. listening to a 1 hour audio book, or reading 3 chapters of a novel, or perhaps a 20 minute audio / 1 chapter of a novel but 3 times over again), you will hear the top 100 most frequent words at least 10 times each day, and you'll hear the top 2000 most frequent words at least once each day. There must be a point on that scale above which you ARE getting enough repetitions of a word naturally, and you can save SRS for the words below that point. Graded content would boost those numbers further.
Again, point taken, and certainly you don't need
flashcards for any
words, let alone these ones. That said, words vary in difficulty, and the first 100 by frequency tend to be brutal - they are all the annoying little connectives and prepositions, words which give little clue as to their meaning, and often flexible in ways that don't map onto other languages. As has often been observed here on the forum, those are the hardest words to learn, whereas plenty of rare words are actually concrete nouns that are pretty easy to remember. There's a reason textbooks don't introduce the first 100 words by frequency all at once.
In other words, I personally find all the help I can get useful with these words, and that means SRS. And in the earliest stages of language learning, when SRSing very common words makes sense, one hour of listening or 3 chapters of reading is beyond what most people can handle anyway. The point of it is to make that day come sooner.
Words such as connectives and prepositions are precisely the kinds of words that I find SRS to be the least helpful with compared to reading/listening in context. When I first started learning Japanese, I used SRS cards based on a word frequency deck, and I found these flashcards in particular to be ineffective:
の - of, in, at, for, by, ... (The #1 most frequent word in Japanese)
に - at, on, in, to, for, ... (The #2 most frequent word in Japanese)
と - and, or, with, if, ... (The #9 most frequent word in Japanese)
で - in, at, from, by, ... (The #11 most frequent word in Japanese)
And to do a good job of creating effective flashcards for these words would take a lot of time. The frequency dictionaries unfortunately don't tell you the individual frequency of each meaning of the word, so to make a good flashcard, you need to spend some time to decide which meanings you are going to focus on, and then perhaps create multiple cards so that you don't overload a single card with too much meaning.
The good thing about these words, however, is that they are so common. As soon as I started reading basic dialogues, with an online dictionary for each word (which is basically what LingQ does, or web browser plugins like Rikaichamp/Rikaichan), I very quickly got a handle on these words. I really did not need them in my SRS at all because they occur multiple times even within the same sentence, they're that common.
If you think about it, software like LingQ, LWT and Rikaichamp does part of what SRS does, at least for L2->L1 vocab flashcards. With a flashcard, you turn over the card to see what it means, but you try to remember the meaning on your own before you look. With LingQ, LWT and Rikaichamp, you hover your mouse over a word to see what it means, but you try to remember the meaning on your own before you look (and you don't need to look if you're already confident). The difference is that reading tools like LingQ give you rich context for free but don't follow the optimal mathematical repetition intervals, while SRS gives you the optimal mathematical repetition intervals but without the free context. However, my own experience is that the optimal mathematical repetition intervals do not help me at all when it comes to learning words that extremely common like this anyway, and what matters much more is having multiple examples to get used to how these words work.
As for whether 1 hour of listening per day is too much for the average learner, all of the stats I gave scale linearly. So you can just halve that and listen to 30 minutes a day, and you're going to hear the top 100 most frequent words at least 5 times each, and the top 10 words at least 100 times each, and the top 1000 words at least once each. You can keep halving that, and you just need to halve all the other stats. But whatever amount you arrive at, my point is still the same, which is that there will be some point on that scale above which you are going to get sufficient repetition of those words through your reading or listening material, and you can save SRS for the words below that which occur in nature less frequently than the SRS would deem optimal.
What I'm interested to explore is how we can use computers to help us select input material to fit the forgetting curve more closely. For more about this idea, you can read my post SRS vs natural repetition
ryanheise wrote:Finally, there is a very simple answer to the question "Why slow down that process?":
SPEED is not the only goal, or even the main goal, for everyone. For example, I enjoy taking the scenic route to the grocery store, even though it may take me longer to get to the destination. People should make an individual choice that will help them fulfill their own goals, and one of those goals may be to enjoy the journey.
Yes indeed, and well said. And there's no question that finding SRS enjoyable is... unusual; so many may not wish to make it part of their hobby. Thank you for the reminder.
Nonetheless, I think there is a danger with Kaufmann's approach that it actually makes language learning harder. SRS is a tool that can make language learning faster, and (done right!) easier. That in turn makes it more accessible
. In one sense, I don't want to rush the journey myself - but I'm just far too busy not to take advantages of tools like this. I couldn't learn languages without them, simply due to lack of time.
And indeed, it was SRS that opened up the very possibility of language learning for me - it took away the burden of not knowing vocab, of forgetting it again and again, so that I could begin to read, and so make progress. That's what I should have been trying to say in my original post - that Kaufmann's dismissive attitude (and behind it, Krashen's) potentially shuts people off from something that, for some at least, makes language learning far more accessible, more successful, more rewarding, and ultimately more enjoyable.
Yes, but there is a certain unfair asymmetry to what you're saying: i.e. The danger of Kaufmann's approach is that it makes language learning harder, while SRS is a tool that can make language learning easier. Well yes this is true, but the reverse is also true and it depends on the person.
Now I think it's perfectly OK to be an advocate for your own preferred method, and I think it's also perfectly OK to listen to a different point of view, consider the reasons given, and then decide that it does not resonate with your own experiences, and dismiss it as "unhelpful". But if you hold up a mirror, this is what both you and Kaufmann are doing. We all do it. For Kaufmann, he also also plugging his product, so that's different, although I don't think there is any principle that says that when marketing one's product, one should not try to advocate for it.
The concern would be whether he is falsely advertising his product, but I don't think he has gone to that extent, and rather, I think he has generally been one to acknowledge that there are different learning approaches that may suit different people.