A thousand words in 6 weeks

Those of my readers who are also on Olly Richards mailing list – and there may well be a sizeable crossover due to common interests – will have noted that he has a guest on this week, one “memory scientist” Anthony Metivier telling people that according to his method it is possible to learn 1000 words in six weeks.

Now just for the record let me say that learning words in isolation isn’t optial, better to learn short phrases showing the word in use with its collocations and recalling a range of grammar, but you certainly can learn words if you want to.

I just wanted to compare the results of people using the GoldList Method and certainly my own experience using the GoldList Method with this run rate of 1000 words in six weeks.

I prefer to use the term lines, and how many new words equate to 100 new lines depends entirely on the material. If it’s a dictionary it can be 100% or near. Probably the average is around half of that, and in some cases even less. Certainly I take several lines for each new word in Japanese, while in Czech I have 24,000 lines of Headlist and I know that there are around 18,000 words in there. Imiagine that we wanted to focus on words, we’d prepare material in order that a line was a word. So for this thought experiment I will take the idea 1 line of GLM = 1 word per Mr Metivier’s Method. GLM is very flexible so it will work around that.

To learn 1000 lines in GLM means to entirely distil them away. This cannot actually be done in six weeks as you can do a maximum of two distillations in that time. So instead you have to apply the long-term run rate which is 3 line repetitions on average per line of Headlist, because a 1000 line Headlist will distil out at somthing like this:

H = 1000,
D1 = 680,
D2 = 460,
D3 = 310
Bronze total 2,450
D4 = 175,
D5 = 125,
D6 = 90
D7 = 60
Silver Total 450
D8 = 40
D9 = 25
D10 = 15
D11 = 10
Gold Total 100
Grand total 3,000

3000/1000 = 3

It will vary from maybe 2.6 to 3.4 but in the main it will be around 3.

So to learn 1000 lines to the long term memory you need to do 3,000 lines in those 3 weeks.

That will be the equivalent of learning 1000 words, but you won’t necessarily know which of them they are. It won’t be a question of guaranteeing that all the words in a list of 1000 are in long-term memory, instead it is a question of following a long-term run rate.

So, how 3000 lines in 6 weeks is 500 lines a week.

That’s the same as the 5,000 level target on the 70 day challenges.

So effectively what Metivier is doing and what we are doing is a very similar result.

In our case, it should be possible using an average of 1.5 hours per day.

What is more interesting is to see which method gives the best passive recall two years after the six weeks in question are over.



Answer to Question comparing Goldlist and Mnemosyne Methods.

Ebbinghaus forgetting curve
Not hard to see where the two weeks of short-term memory fits in with Ebbinghaus' ideas!

I have been lucky enough this week to receive questions from two people on YouTube about aspects of the Goldlist Method, along with their permission to respond here so that I don’t have to fiddle about with the 500 character cut-off or however many it is over there.

Let’s kick off with this one from YouTube channel WellConditionedChimp

I’m wondering whether you are familiar with Mnemosyne, an open source computer program that is reminiscent of your method – it makes digital flashcards that come up for review after a variable interval of time. The interval is determined by how quickly you remembered the material the last time, if at all. In what ways is your method superior to this one?

I assume that you are referring to the Mnemosyne Project in which case I was not familiar with it, although it seems to be building on Piotr Wozniak, who in turn builds on other researchers going back to Ebbinghaus. In my case I only learned about Wozniak’s work on memory after my own system was complete, but as you will see if you read the Polyglot Project (available via syzygycc channel on YT as an e-book for free, or in paper printed and bound on Amazon.com for $16.95)  you will know that my inspiration came from reading second hand about Ebbinghaus, plus my own experience as a linguist, plus the fact that getting back into numbers in order to become an accountant started to make me think along the lines of a numerically controlled learning system for languages. Continue reading “Answer to Question comparing Goldlist and Mnemosyne Methods.”

Do elephants need the Goldlist method?

The eye of an asian elephant at Elephant Natur...
The eye of one who doesn't forget in a hurry...

I’ve put down a goldlist for years and picked it back up and continued. The long term memory is the long term memory. Humans and elephants have it in spades. The difference is that we can turn ours off by switching on the short-term memory in the process of conscious cramming or deliberate rote learning. Elephants probably cannot do that – an elephant never forgets. Their sample rate is higher as their brain is 7% Hippocampus and not just 5% like ours. They have a language which we have a lot of difficulty in understanding as it is in infrasound, travels 10 km and they use they feet and trunks as well as their huge ears to pick up the auditory signals. We need machines to hear any of this, and then we don’t really experience it but see it as vibrations on a screen. They on the other hand can eavesdrop on human speech and they take a particular interest when their keepers describe what plans they have to do with them.

If elephants were formal linguists and polyglots then they probably wouldn’t need something like an SRS or a goldlist method, as they are very natural in their use of their facilities. But since we humans do very silly things with our minds in aid of learning, under the influence of schoolteachers utterly uneducated in how the brain actually works and using a “one size fits all” method for learning, we do need something that can get our minds working more optimally again while approaching the learning of other languages. And that’s what this method and some other methods try to offer.