Will a Spaced Repetition System (SRS) Help You Learn a Language Faster?

spaced repetition systems, SRS, Anki. Most language learning systems, including

LingQ, do incorporate some form of spaced repetition into their program of learning.

It's a tool to enable people to remember words primarily.

Although SRSs can be used in science, can be used in other fields.

Even in language, it can be used to try to remember phrases and words.

The whole idea is that with a system of spaced repetition, and specifically an

expanding space, in other words, there is an algorithm which someone or different

people have different algorithms, which say that if you space the repetition out

initially, let's say two days, 10 days, 30 days, 50 days, whatever it might be,

that this is going to improve your ability to retain the items that you're learning.

So, I have not been a great practitioner of spaced repetition systems or a user

of these spaced repetition systems.

I did, however, use a form of space repetition when I was

learning Chinese characters a long, long time ago, 50 years ago.

I

had a box here of these lovely hand drawn Chinese characters, and

there's a thousand of them here, and I went through all one thousand characters

with these squared Chinese textbooks.

And I would write one character, let's say 10, 15 times,

and then a couple of columns over.

I would put either the sound, the pronunciation, or the meaning.

Then I'd pick up another character or another flashcard, and I would

continue doing that, write it 10 times.

Eventually I ran into the first card and I'd have to remember

it before I had forgotten it, and I would keep on doing that.

I started out with 10 a day.

I got it up to 30 a day.

Of course,

forgetting a lot, relearning.

And I did that for a thousand characters. Thereafter, to learn the remaining

3,000 characters that I had to learn,

I just relied on coming across them in my reading, writing them out a few times.

So that's my experience with space repetition.

Of course, that was not an algorithm.

It was not an expanding space, spaced repetition system.

It was just some form of spacing.

Now, doing some research on the subject, I discovered that while there are

studies that show that some kind of expanding spaced repetition is

beneficial to learning, there are other studies that show that it is the

repetition that's beneficial to learning.

That even if you have constant spacing, there is no significant difference.

And I will leave some links here in the, uh, description box so that you can go

and check these different people out.

So it's not obvious that all of the research confirms that what we

consider to be space repetition, this expanding spacing is all

that beneficial for learning.

Now the next thing is applying this to language learning. Because

it's one thing if we're learning,

you know, facts, scientific facts, even historical data. Or even in the case of my

Chinese characters, while these characters were being learned sort of away from any

context, at least, there is a significant connection between the characters

because there's a lot of repetition of elements in these characters.

So that I was learning and getting a greater and a better sense of how

these different elements combined in these characters that I was

learning when we're learning words.

There still are some elements or components.

But it's not as clear as was the case when I was learning Chinese characters.

So when we learn a language, of course we're concerned about input, we're

concerned about comprehension, and we are concerned about producing the language.

And so we need, obviously, we need words in order to understand the

language and eventually to produce it.

And as you know, I'm a big proponent of inputb-based learning,

getting the language in you.

Getting words in you so that you understand the language

and eventually can use it.

So in studying how the brain learns, there is a thing called "Event-Related Potential,"

which measures how the brain deals with sentences that it hears or reads.

So the brain tends to predict what's going to happen in this sentence or

phrase or whatever. And it basically has

certain predictions in terms of the meanings, sort of the semantics if

you want, and in terms of the syntax.

In other words, how the structure should work based on its familiarity with the

language, its exposure to the language.

It makes certain assumptions.

And in this way, researchers are able to tell what is important for the brain

to be able to understand that sentence.

Well, it turns out that while frequency of exposure is extremely important,

lexical frequency, in other words, how often they've seen the word.

However, as the learner has more and more exposure to a variety of contexts,

the lexical frequency, in other words, frequency as a factor becomes much

less important than previous exposure to a variety of different contexts.

And in the description, they talk about the lexical frequency being

sort of a bottoms up, bits and pieces

approach to how the learner reacts to words and phrases in a sentence.

Whereas over time, with enough exposure to the language, it's more

of a top down form of reaction that dominates our ability to understand

and eventually to predict what's going to happen in the sense or to react if

what happened in the sentence didn't correspond to what our brain expected.

But it's the buildup of that familiarity with the context, what to expect,

which is more important than the frequency of exposure to the words.

So that is important.

Bear in mind that, as I said when I was talking about Spitzer's

book, learning takes place when we have frequency of exposure.

When there is similarity. Obviously, if I'm going to study, say on LingQ,

Finish, as opposed to Dutch,

I'm going to have an easier time with Dutch because it's more similar to what I know.

Also in that similarity, I think there's also included in that

some sense of association with

similar things, related things.

All of this contributes to whether or not we're going to get the neurons firing,

whether the synapse is going to take place.

But the other thing, the third thing, was relevance or importance, which

is also an important consideration in determining whether the brain is

going to choose to learn this thing.

If

it considers it to be, you know, important.

It's important to bear these things in mind when we consider how a

space repetition system works.

Obviously, if we take the position that exposure to a lot of content and a lot

of context reduces the importance of the frequency of exposure to the word itself,

that's a consideration.

And if our approach to comprehension is very much tied up with context

and relating words to other words in context that we have seen before,

structures that we have seen before, it reduces the specific importance

of just the frequency of exposure to the word, which is kind of what spaced

repetition system tends to stress.

That's insofar as

comprehension is concerned.

Now let's look at production.

The interesting thing about production is that planning and memory are

located in the same parts of our brain.

Because to plan something, you have to be able to remember something that

you want to plan to do in the future.

And when we produce a language, apparently, and I, again, I'll leave

a link to this in the description box, there's a lot of planning involved.

What are we going to say?

How are we going to structure it?

Which words are we going to use?

And the planning involves a process of,

first of all, using the easily retrieved words first.

Because that gets you going.

Reusing patterns that have worked for you before and suppressing those things that

might seem similar but aren't relevant.

A word that sounds almost the same or is spelled almost the same. Or

interference from another language.

We have to suppress those.

So the brain has sort of a three-part process, finding the easily

retrieved words and phrases to use, reusing things that worked before.

And suppressing those things that they don't want.

But all of this requires a significant memory.

And if that memory is best achieved through massive exposure to the

language rather than isolated review of words, maybe that's

where we should orient ourselves.

If we look at spaced repetition systems per se, it seems that in the case

of language, language learners might have, depending on where they are in

the language, 500 words in their deck.

5,000 words in their deck. A large number of words potentially in their deck.

Now, language learning is a function of attitude, of time, and of the ability

to focus on what it is you're doing.

So obviously, if you like doing spaced repetition, if you like doing

flashcards, it's something you should do.

The research that I did (I went to ChatGPT) suggested that the average

person spends 30 seconds on a flashcard.

I can't imagine spending 30 seconds on a flashcard.

I do, and I'll explain a little later how I use flashcards, but

in my case, it's a few seconds.

It's not 30 seconds.

It's just exposure.

But I guess people who are trying to remember that flashcard might

spend 30 seconds on a flashcard.

At least that's what ChatGPT told me.

So if we look at word frequency, as I've said before, word frequency

declines very, very quickly.

So in the most frequent words, obviously just in any content,

any context, you are going to get exposed to the very frequent words.

The bigger issue is if we look at word frequency, and I'll show you a list

here and you'll see that words that are in the 50th spot in terms of word

Frequency, or 500th spot or 5000th spot, may only appear once or twice in a book.

In other words, these statistics were based on Google's corpora of

content, but some of these words, even fairly common words, like, you

know, "road" or I don't know, whatever, they may not appear that often.

How are you going to learn them?

Obviously, you can guide the process.

If you're reading "Harry Potter" the word "Harry" is going to appear a lot of times.

I've never read "Harry Potter" but words having to do with magicians and so

forth are going to be there more often.

If you're talking about economics or history, there are certain relatively

rare words that will appear more often. But nevertheless, you are

chasing a very small number or a low frequency of occurrence.

Of these words that you nevertheless require, because on any given page,

there's going to be 10 of these low-frequency words. And not understanding

them is going to limit your ability to understand what's on the page.

So then perhaps that's a place to use spaced repetition systems,

because you can then go after words

that won't appear naturally often enough for you to learn them.

But with spaced repetition, you can then zoom in on them.

The problem is, and this is the dilemma, if there are potentially 10-, 20,000 words

that you need to learn, if you actually do them in spaced repetition systems,

you will end up spending almost all your time on these flashcard reviews.

Obviously, if you have seven or eight hours a day to spend learning the

language, then that might be where you choose to spend your time. And let me

stress, it's entirely a matter of personal choice. So I would not choose to do that.

In fact, I don't choose to do that.

But people who enjoy doing that, then

that's what they should do.

I find that when I'm reviewing flashcards, if it's divorced from any

context, then I quickly lose interest.

So I find it hard to focus.

So once you're not focusing, once you no longer consider it relevant,

important, then I can't continue.

So I occasionally do go through lists at LingQ, especially of

my Status 3 words.

Trying to move them to known because I have a goal.

I want to increase my Known Words total, but still I can only do it for so long.

10, 15 minutes max.

It's all about sort of "bits and pieces learning" versus "top down learning."

When we're dealing with flashcards, when we're reviewing grammar rules, we

are dealing with the bits and pieces.

When we are just reading and listening, we're in the sort of

"top down learning mode," and I prefer to spend most of my time

in a "top down learning mode." But there is sort of a compromise,

which I now really enjoy doing.

In the case of my Persian, given where I am, call it a [CEFR] B1 level in Persian,

still struggling with a writing system that's still not very familiar to me.

I enjoy going into Sentence Review.

The content that I'm using is timestamped.

I can listen to it.

I can go through it, look at the words that are still not known to me.

I'm still dealing with content that I've read before, so there are no more sort

of unknown words, but there's a lot of yellow words that I still don't yet know.

If I want to, I can go into any one of these sentences and go through the

matching pairs flashcards exercise.

I can go through the exercise of reconstituting the sentence.

That's really sort of drilling down into "bits and pieces learning,"

which, of course, slows me down.

So I don't do a lot of it.

I do it on a sampling basis because I think it helps give me a better sense

of the structure of the language.

It just increases, at some level, my familiarity with the language, even

though it's slows down my taking in as much of the language as possible.

And I think the lesson here, or the message that I would like to end

with, is in language learning, of course, our attitude is important.

The amount of time we spend, and remaining focused so that we pay attention to

what's happening in the language.

So therefore, we need to do what we like doing.

We need to do what we're going to happily spend time doing, and we

need to do things that we think are important so we remain focused.

And so I don't think any one system, you know, you should

do this or you should do that.

You should use SRS, should not use SRS.

I don't think that's good advice.

We never really know when we're engaged in some language learning

activity, how effective that activity is compared to some other activity.

We know that as long as we are engaged with the language in some way, we are

learning. Now, are we learning less well than if we were doing something else?

Are we learning faster than if we were doing something else?

It's very hard to tell if we have a sense that we're improving.

If we have a sense that we are enjoying the process, then we should

continue doing what we're doing.