Ed Cooke: Memrise and the Membus project

This is the actual fluency podcast, helping you become a better language learner by providing motivating weekly inspiration, as well as top tips and tricks from expert guests. And now, your host, Chris Broholm.

Hello everyone, and welcome back to another episode of the actual fluency podcast. And as I say that, I realize that's how I start almost every episode. Hmm. Anyway, uh today, I'm happy to talk to Mr. Ed Cook from Memrise.

And uh you might be wondering, Chris, how come we're getting three episodes in one week? And I'm like, yeah, you know, that's a bit unusual, I'll admit. Um, but first of all, the two of the episodes were kind of connected. That was with Mr. Michael Levi Harris. And I figured, you know, you shouldn't have to wait a whole week to listen to the continuation of that episode. And on the other uh point is that I've had a few delays lately where an episode would be, you know, two, three days delayed from the posting date. And I just figured, you know, why not just get them all out now uh to compensate for that.

So, you're getting a lot of episodes this week. I hope you can catch up or keep uh stay updated on your podcast listening devices, whether that's on the phone or computer or whatever.

But uh yeah, so I'm really excited about today's talk. I've been trying to get to talk to Mr. Ed Cook for a while, and we finally had a chance to catch up a little bit. And so what we're talking about is just in general, the Memrise project, and also their newest project, the Membus project, which I I briefly talked about in the previous episode as well, in the introduction.

And you can still get involved with the Membus project. You can go to uh the Kickstarter and search for Membus, and you'll be taken to the donations page. And basically, you can contribute any any amount you uh you feel like, and any amount you can afford. And actually, you can get to go on the bus, which is one of the most epic multilingual road trips ever attempted. You know, you can stay uh five or seven days on the bus and go around Europe recording native speakers uh speaking sentences and words.

So, that's extremely epic, and I would definitely have gone have done that, paid for that if I had the money. But uh some of you probably know, I'm not the best uh place financially. Uh looking to improve on that, but hopefully, uh I'll get another chance to go on. What I've said is, or actually, what they said from Memrise was that if I could find two people who got their spot on the bus using the pledges on Kickstarter, they would invite me to come along as well.

So, if any got any of you guys listening out there are like, I want to take a a week's paid vacation, um, you know, the going on the bus isn't cheap. I'll be honest with you. You know, the minimum payment, I believe, uh, is uh, about 700 pounds, which is a lot of money for five days. But you also get everything included, and it's one of the most epic road trips ever.

And actually, we'll go down in history, I'm sure, because we're at the sort of renaissance for language learning technology. So, you know, it's a kind of an epic, probably once in a lifetime or, I mean, if it comes up again, that would be surprising. But yeah, it's definitely an epic project. And, you know, if you can get involved, I highly suggest you you do.

And, you know, let me know about it. Uh, be sure to tell them I sent you. So, I can come on the bus. But, um, that's it for now. Uh, enjoy the episode with Ed Cook. And we also get into some learning uh, techniques, how to use Memrise the best, how to improve your memory, how to, yeah, basically learn the best.

So, I really hope you enjoy this week's very practical episode, and I hope to see you in the next one. Take care.

All right, we're back with another episode of the actual fluency podcast. And today, I've got the CEO of Memrise, who's also a master of memory. And we're here to talk about the Membus project, Memrise, and the memory, of course. I mean, how could we not?

Um, Ed Cook, how are you doing today?

I'm doing extremely well. Thank you very much for having me on the show.

Fantastic. And, you know, this Membus project has obviously been the talk of the town the last few weeks. Um, can you just give the listeners who haven't heard about it yet a brief summary of what the project is?

So, um, well, the Membus, to begin with, is a, it's a double-decker bus, uh, which, uh, we have got our hands on and painted into the marvelous colors and, and so on. And, and, and the Memrise Membus project, um, it relates to a learning experience we've been developing, um, which, um, is based on very short videos of native speakers, uh, using language in context.

So, um, you know, obviously with, uh, with language learning, the reason you want to be able to speak languages is because you wish to be able to, you know, communicate and encounter people and learn about the culture and, uh, basically open up the horizons of your world. That's what we kind of think of as the essential quality of, of language learning.

And, um, and the Membus, and so, yes, and so the, the video, um, learning experience is basically to say, rather than learning from phrases or from written, written-down text as, you know, most language apps and books obviously, um, operate, why not, um, actually have, as an essential part of the language learning, uh, an encounter with native speakers using language in context.

Um, and this has many charms. Um, uh, some are practical, and some are aesthetic. Um, you know, when you're learning from short video native speakers, you're, you're actually kind of meeting people who speak the language. You're seeing their clothes, their humor, their physiognomy, uh, how they behave, you know, and some something of their humor. You're, uh, and actually encountering real people in some sense.

Um, and then, you know, practically, also kind of encountering how the language is actually spoken, how, um, different accents work, how, um, how, well, you're, you're actually just encountering real language. And so, it's got these kind of two other marvelous benefits. This kind of mode of learning that you're learning real language, and you're also in direct contact with the reasons you'd wish to speak a language in the first place.

So, that's the that that's the kind of the video experience. So, we, you know, we we're kind of in a situation where like, okay, well, this is super cool. We've done it for English to some extent. Um, but how can we go out and, um, and collect, you know, what's required for kind of language, tens of thousands of videos of words and phrases being used in context?

Um, and so, it's quite a long story. But, you know, we we experimented around, running around with film cameras and talking to people and so on. And we found that there's this rather wonderful methodology where you get a kind of a bunch of people, uh, five or so filmmakers. You hit a town, and you just accost pedestrians. You say, excuse me, like, we're making this video dictionary. Would you mind contributing a phrase? And, um, it turns out you can actually record an enormous number of these phrases, but also to a really high quality. You you can shoot in beautifully. You can have super great audio, um, and the rest of it.

And so, um, and so you on the basis of this methodology, we sort of thinking, okay, we need to go around Europe and like collect loads and loads of languages. Well, what is the most, uh, effective and also, let's face it, fun way of doing that? And so we decided that the way of doing that is to go on a kind of on a road trip. Um, and so yeah, we picked up this double-decker bus, painted it with Membus on the side and all these cool Memrise graphics. Um, next summer we're going to go on a kind of epic, um, and I'll come back to why epic is the word for it, but an epic road trip around Europe, um, to basically hit all the towns and countries of Europe and collect tens of thousands of videos of native speakers in context, which will then pipe into the Memrise app to create a kind of a learning experience which we hope will be basically of unparalleled joyfulness. Um,

Fantastic. And obviously, people can get involved as well. We should probably get that in

Yes, um, so, yes. Yes, I mean, this is, um, so, uh, Yes, so, um, we are, um, currently running a Kickstarter, um, which is part of, um, you know, it's both to kind of partly fund the project, but also to, um, gather a community, um, behind behind it.

And so, you know, if you just search Google for Membus, um, it will come up. The project will come up. And as well, the page on on on memrise.com. And, um, and there, you know, we're looking for, we're looking for volunteer filmmakers, we're looking for people who wish to, um, meet us along the way, introduce us to their friends, just to, um, have some fun and meet us, uh, and, uh, and yeah, so we're looking for it for for anybody who's kind of inspired by this project to kind of to help.

And, you know, and I will say that through the Kickstarter, we've had some wonderful, you know, wonderful incoming interest. You know, people who are like, well, you know, I'm, I'm currently, you know, uh, recording video of, um, video testimony of people in Africa and Northern Iraq. Uh, this has some great project. I'd love to get involved and, and so on and so forth. So, um, all all volunteers welcome. But anyone who wants to meet us along the way, maybe record some of their own language with us or or or otherwise help out would be fantastic to meet.

Sounds good. And there's, uh, just a few days left to to, uh, to do to contribute to that Kickstarter. So, please, if you're interested, you know, get going and get on that Kickstarter. There's lots of different levels you can contribute all the way from, you know, uh, getting, uh, some Mem Memrise swag or you can go all the way up and sponsor a wheel, which is, uh, very exciting. Uh, and, uh, you know, depending on what people, uh, what kind of budgets people have, there are lots of different levels. So, you know, go check it out if, uh, you haven't done so already.

All right. So, uh, on the basis of that, I I just had a few questions about your the methodology you're talking about. Um, has there been any kind of link between or have you or have any kind of benefit to memory of watching somebody speak a phrase instead of just listening to someone speak a phrase? Does it help us memorize it better?

Well, I suppose, um, um, I think, I think it does. Um, I I don't know if that is the principal benefit.

Um, and I think that, um, that it's important. I mean, one of the things about memory is that it's quite a complex cognitive function. So, it's not, it's not just a kind of storehouse in the center of your mind, um, but rather it, um, it interacts with all of your other kind of faculties.

So, um, for so, for instance, um, perception, you know, you learn to perceive things, you learn to understand an accent or something. And, and that learning is itself memory. Um, but we don't often call it that.

Um, and so, so there are huge kind of, um, the valuable things for the language learner which takes place when you learn from video which wouldn't necessarily fall into the category of, oh, I can remember this word better, but you remember the word more like it actually is. Um, and you remember more, um, more of the subtle fringe features of what someone's actually saying and speaking.

Um, and the second thing is that, you know, the the context of of language usage, well, I mean, just as generally, as a feature of memory, the context of recall, um, is most effective when it's, um, oh, sorry, the, um, the the our facility at recalling something is closely related to the connection between the situation we're in when we're trying to recall and that in which we learn something.

So, there are kind of lots of famous studies where, you know, if you smell marmalade while you're studying maths and then you smell marmalade when you're doing a maths exam, you'll do do better at that. And it can, it can link to anything. You know, if you're if the space, people would perform better in exams if they, um, do the exam in the same space they studied.

One rather comic experiment involved scuba divers learning to solve, uh, maths problems and remember things. And it found that if you learn something while scuba diving, you recall it better when scuba diving than when you're back on dry land, and vice-versa.

So, so these kind of context effects are enormously important. And obviously, the vast majority of the time when you're actually speaking a language, you, um, you're, you know, you're face-to-face with somebody. You're actually seeing, uh, someone, including their face and their intonation and the fact that they perhaps speak quite fast and have facial expressions and so on.

So, so that I think is another advantage. You know, it may be that, um, it may be that that, um, there's no massive improvement in how fast you remember the word, but that you can recall it much easily when you actually need to. And that you remember more richness and so on and so forth.

Right. So, okay. So, and, and also, I suppose, it's a general improvement of the learning process also because it's more enjoyable to actually see people and, you know, from, from the demos I've seen, you know, people are quite smiling and they're quite happy to share their knowledge.

Yeah, I mean, I think for us, we've always wanted to develop a learning experience which, um, which is authentically joyful. So, where, you know, you can imagine even if you're really tired after a long day and you normally sort of, you know, watch Netflix or whatever, where there's just a kind of a flow to it and a pleasure and a kind of encounter with things which are beautiful and vibrant and fun.

Um, we've always wanted to create something like that, which is also like brutally effective for learning. So, where you can like learn an incredible pace. And we think there's a really interesting path, um, with this video because it's just so enjoyable to see people's faces.

Uh, and, you know, it's I I I'd like to to reflect on the fact that if you look at different nationalities, and so far, we've just done a little trial project on Danish people and French people and so on. Um, one of the things that strikes you is like, oh my God, the Danish people are so Danish, uh, and the French people are so French. Uh, and equally, like, English people are so English.

Um, and it's that kind of, um, encounter with like the reality of the people, which is, I think, uh, part of the joy of the of the learning experience and something which we really want to develop and turn into a a kind of authentically rich form of learning experience.

Right. I talked to Maria a little bit about the project as well. She's heading heading the project. And, um, one of the questions I had was about the accents, because initially, when I thought about recording something in English, I was thinking, well, does it help somebody in let's say, Brazil to learn English from somebody in Glasgow?

And I initially saw it as kind of a weakness in, in, in the, in that you had to have kind of a high language standard in, in the sense of like, high German or high English. But Maria was kind of saying that it's it's probably, you know, beneficial as well to show the diversity.

What do you think about, you know, having a a wide range of of accents? Would it be helpful for learners of English or let's say, all languages? Or do you have to kind of take some steps to, you know, make it a little bit more like the high standard version.

Accessible. Yeah. Yeah. Well, I mean, I think it's a little bit of both. So, I think that, um, that in terms of like, the robustness of your comprehension and the depth of your language skill, having that facility of understanding lots of different accents.

And also, it's also different voices, um, and actual human voices, you know, quite often on on online learning programs, you get a kind of robot voice which says everything, you know, the kind of, uh, a slightly kind of tinny version of, um, a perfect form of the language. So, I think generally speaking, it's very healthy.

Um, you know, I think that the the the dream of the video project in the long term is that one will be able to kind of select one's accent. Just like, okay, I'm going to be I want I want to learn like Texan English or English from Oxfordshire or English from Scotland or whatever. And, you know, it could be amazingly specific. I want to be able to speak like the Beatles. I'm going to learn English from Liverpool, um, or what have you.

But, um, that's probably a little bit down the line. Um, and I think, you know, we we because we have wonderful data on the degree to which, um, people are correctly understanding the videos, uh, we're able very quickly to say, okay, well, this person's accent's a bit thick. We're going to whatever. Um,

That's pretty funny. You can you can just see it based on the test results.

Yeah. It's it's it's It's it's it's astonishing, actually. It's so cool what's possible with, um, kind of information technology. You know, like, you know, because on Memrise, people might learn the word for spoon in French, you know, 500,000 times a year or something. And so, you get incredibly precise data about, you know, what the best way to learn a word for spoon is, et cetera, et cetera. So,

And and you use that data to improve the the sort of the side, I suppose, the service.

Yeah, so that's the, um, that's the idea. I mean, it of course, these exist slightly more complex in reality than in, um, than in one's imagination. But that's like, um, emphatically something we're going to be able to do with the video stuff.

Um, and, um, I mean, it's interesting. I think, um, you know, we have some really exciting product ideas around, um, around optimizing the, um, about using the data for kind of at a very fine-grained level where, you know, the dream version would be like, okay, so, um, Gerald, um, has lots of experience in Spanish and a bit of experience in Portuguese but has never learned the Cyrillic alphabet. And he's a sort of 42-year-old farmer. And on the basis of all that data, we know that he's going to find the following joke really funny and never forget the Russian word for bus stop. Or Stavka, what if you know. Yeah, that that's dream, you know. And I think, you know, um, it's, uh, something to be aspired to. something to be doing.

I guess you could even like AB test individual speakers. Uh, let's say if you and I recorded the same phrase. You could send 50,000 people to your recording and 50,000 people to mine.

Yeah, yeah, totally, totally.

And if people learned the phrase better from you, you could just say, well, we probably shouldn't have this guy then. So, we should have Ed say it instead of Chris.

Yes, well, I think that's exactly what we will do. Although, you know, um, I don't think we'll be too slave to the bell curve, so to speak. You know, we can, um, we can, um, you know, if if we find someone extremely amusing, um, but completely incomprehensible, that's might be a good experience for language learners to have. You know. Anyway.

Yeah, you just mentioned Liverpool before. Yeah, it goes under that.

Um, so what what other, um, uh, memory, uh, well, I I suppose I keep talking about memory, but of course, that's because you're you're the memory guy. You're the, you know, grandmaster of memory. And, uh, and, you know, Memrise sort of builds on on the memory aspect.

But but what other other sort of behind-the-scenes improvements towards retention rates and learning rates have you been doing in the last one, two years that people might not have seen because it happens behind the scene?

Well, um, I mean, there's, there's lots of, I mean, there's a hundred tiny little details really. Um, you know, a tweak to this, uh, algorithm, um, um, a subtle change to how we handle this interaction of a particular test type, um, a way we adapt to the difficulty of a word or to a person's particular strengths and weaknesses.

Uh, you know, I think it's it's one of those things which has, um, you know, we still have a long way to go. I think there's, um, there's the absolutely perfect form of this would be the and, you know, and this is what we're kind of trying to achieve, and I think we will, we we will get there, we'll get very close.

But the kind of the dream version of this is something where, um, we're kind of exactly anticipate, you know, we're like the kind of perfect dance partner of your brain. Um, where we're exactly anticipating, you know, each of your your your mind's moves, where where it's the micro confusions, tiny little interactions between distant memories, moments of inattention and so on. And we're just doing everything to make sure that the the information's going into your brain in as as sort of delightful a fashion as possible.

Um, and so that's that's the uh that's the dream. And, you know, and I think, um, you know, we've made a lot of progress and we've got some quite exciting stuff coming up in the next year. Uh, which I think will move the needle quite a bit. Um, Great.

Yeah. Yeah, one one of the metrics we have is kind of speed of learning. Um, so, um, you know, roughly speaking, measured in words per hour. Um, and it's a complex metric because you have to say, well, you learn words per hour to remember for what period of time.

But, um, we've been doing this this really cool competition with a bunch of cognitive science departments around the world. Um, and it's called the Memprize. And we've actually got a URL for this, Memprize.com.

And, um, basically, we are, um, we got, you know, we approached with University College London, some professors there, the guy who Professor David Shanks, who runs the memory and languages lab, UCL. We approached twenty of the top cognitive science labs in the world. Uh, and we were like, well, you've been studying memory for thirty years, design your ultimate learning methodology. And so there's so so many of them did.

So, we got about twenty experiments which got done where people were like, well, I've been studying this element of memory. So, I'm going to create a thing like this. And it was kind of to learn eighty Lithuanian words or something in an hour. And so, they all ran experiments. And then, um, provided us with their kind of empirical results.

And, you know, some of the some of the experiments worked, some of them didn't. And some of the best ones, um, you know, improved the amount someone learned a week later from, you know, against a pure flash cards use case. So, they improved the rate from sort of twenty words per hour to like forty-four or fifty words per hour. Um, and by the way, we internally, we think Memrise is at about forty-four. Okay. That's what I think.

Uh, any-ho, and so, we've taken these five top experiments and then we've coded them ourselves. And we're about to run this massive experiment to, um, to really gain some, um, to really gain some, um, some like massive scale empirical data about these different approaches.

Um, and it's quite cool because what's interesting is that in cognitive science, although an incredible amount is known, and there are just so many unbelievably brilliant people coming up with super cool theories about lots of things. It's not really an empirical question, like, what's the fastest way to learn? It's more like, okay, if I add this mnemonic, what's the influence on long-term retention or whatever? I I I kind of compare it to like scientists have identified like lots of ingredients which help learning but have never thought to cook up the ultimate recipe or something.

Um, anyway, so so we're going to about to run this thing. And then we'll we'll see what happens. We'll see which entry wins. And then we hope to do this experiment every year for many, many years. And then because these results are all going to be open source, the people from the previous next year could be able to say, okay, these guys won this year, but that bit's stupid. We're going to add this thing and the rest of it. And hopefully hopefully, we'll create a kind of feedback effect. And we'll be able to like seriously advance the science of learning, uh, to a point which just doesn't exist at the moment. And that would be beneficial for us, of course. But it'll also be beneficial for every other learning startup and every school and da da da. So, that'd be pretty cool.

And real learner. Yeah. Yeah. Yes. Amazing.

And, um, also, you speaking of new features, uh, not too long ago, you released, or at least I got access to. I know sometimes you do staggered releases. Uh, but I got access to the new functions of the app, uh, which had the, uh, speed review, uh, which I wanted to just, uh, mention, uh, because in, in sort of, um, if I if I just had to ignore everything else and look at the speed review as a as an independent unit, I would say that's kind of a, I'm too lazy to do the full extended review.

Is that is there some form of truth to that or is it What is the Why did you come up with the speed review first of all? Let's let's talk about that.

So, um, I think the the thing is with with learning tech is that one's always balancing, um, motivation and, uh, engagement. Uh, so, so motivation and learning effectiveness. Um, and then they're not, they don't always completely coincide.

So, um, so, um, for instance, um, there's lots of evidence, um, in and this is kind of in the kind of meta-psychology or meta-cognition kind of area of psychology. Um, where there's lots of evidence that, um, that people radically misunderstand what's most effective for their learning. So, so if you kind of get someone to use one of two techniques, quite often, the one which works like twice as well as the other one, they'll say, no, that one didn't work very well, whatever. Anyway.

So, so one and so there's always this kind of challenge, um, in, um, in designing learning tech to make sure that it's as effective as possible. But also to kind of communicate that to the learner. Um, anyway, that's just a kind of a general proviso to to answer that.

But, um, the kind of cognitive reason for having speed review is just that like speedy recall, um, means that, you know, so memories have kind of, um, have actually have have speeds, kind of thing. And you kind of notice this when you try and eat slowly. Um, so if you have like, you get a fork full of food, and rather than going at your normal pace to your mouth, you actually slow it down by a factor of two. Um, it's actually really painful and the rest of it and the rest of it. But you do that for a bit, and you can actually reset the pace at which that memory, that mem that action memory works, kind of thing.

And quite often, you can get into a kind of pace of recall as well. So, you have a certain speed at which you produce an answer, um, which sometimes may not be quite fast enough for conversation because you when you're when you're talking, you obviously have to have a capacity to find words exceptionally fast in order for you to be able to do. Anyway. So, like it's it's obviously a desirable facet of a memory that you can find it really fast. So, that that's kind of one thing.

At the same time, like people often uh or Memrise have, um, um, thousands of items they've learned. Right. Where, you know, where actually they, you know, you fundamentally know those memories, um, and, you know, we could give you a typing test, and you could get it right. But, um, but that actually takes a bit of time and effort and the rest of it. And you can actually get pretty much the same benefit from just speedily recalling it.

And so, that's kind of one of the motivations is to aid people who have a, um, have a lot of memories to review. And I think the final reason is just a bit of fun. It's just like gameification. Gameification. Yeah, fun fun is fun, you know, and, um, fun fun my my 70-year-old father, um, who, um, who rather hilariously, was not, I mean, I've been doing Memrise for like six years or whatever. Um, he wasn't using the Memrise app until we were made editor's choice on Google Play. And and he was like, oh, Ed, you're your apps. I'm going to download it. So, he downloaded it. He got into doing some Italian.

And I saw him, uh, recently, and, uh, he was like, okay, so, Ed, I can't talk. I've got to do my I've got to I've got to do my speed review. And I I just watched and it was crazy. He's 70 years old. And he could and he he was putting together strings of like 65. I mean, you know, viewers trick listeners who, uh, if you don't know the system, but I can't get to 65 myself. Yeah. So, it was it was quite interesting to see see my old man just You know, whatever. And just making these kind of speedy recall things. And, you know, uh, and it's quite a bit of fun. So.

Yeah, it it definitely is. And, uh, I guess, uh, one of the things you also just mentioned that that is a big thing is the most efficient method to learn something might not be the most enjoyable, or enjoyable at all. And then that defeats the entire purpose of effectiveness.

Yes, it's it's um, yeah, it's it's it's a balance. And, um, you know, I think actually at the heart of it, there is a place you can find where you kind of combine perfect effectiveness with perfect enjoyment. Um, um, but the kind of the thing you've really got to optimize there is not like how fast are the memories being laid down over a two-minute or five-minute period, but just like, how much will someone wind up learning after a year? Kind of thing. And that's the thing, you know, you want to maximize, and that's the thing you want to, um, I mean, for for me, like, the the moments in my life where I've had my most, um, my most, um, kind of impressive and, um, enjoyable learning experiences, have always been like 50% heart, 50% head.

I've always felt like in my brain, I'd say, oh my God, I'm a genius. I'm so fast, it's so cool. And then in my heart, I'd been like, oh my God, it's so cool what we're learning. I've learned, you know, like, uh, I'm discovering new things, I love it. I want to learn more, I want to learn more. So, that's kind of that's the loop which we're kind of aiming to create. And, um, it's not a job I imagine we're ever going to finish, but it's quite an enjoyable, uh, task to try and, uh, put it together.

Yeah, exactly. I mean, I I think a lot of people are jealous of, uh, of your job of of, uh, just playing around with the memory and, uh, things.

Um, one thing I I was, uh, well, maybe not surprised, but, um, when I've seen some of your other stuff on YouTube and, uh, interviews you've done, you always talk a lot about the, um, um, I'm not sure what it is a memory palace, I suppose. You know, the journey method where you kind of go through, you mentioned it when we were, uh, when you were teaching people how to memorize a deck of cards.

It always surprised me that that method didn't have any sort of, uh, presence, you know, uh, in Memrise. Yeah, it just it just doesn't seem to be like a a journey or, yeah, you have the mems, of course, uh, being the, uh, associations. But you don't have the sort of spatial relationship. Is that something that you've tried to implement or thought about implementing or

Yeah, we certainly thought about implementing. I mean, I've spent, um, years of my life daydreaming about the ideal way to do that. It's tricky.

It's, um, I mean, I think, I mean, actually we can come back to the Membus on this topic because, um, the courses we're going to create on that journey, I think we're going to, um, integrate the actual sort of story and the journey into the courses. So that, um, so that there is a kind of to begin with, a narrative element. And secondary, a kind of spatial element to kind of categories and so on.

Um, but the the thing which memory palaces are most effective for are learning sequences. Okay. So, so like if you need to learn, um, so, let's say that in order to speak a language, you needed to learn a thousand words in order. We'd definitely do it, uh, in a memory palace. Um, and, um, and then you'd be able to recall, you'd literally be able to begin with the first word. And then, you know, it wouldn't take that long, you know. You'd be able to study on it for like ten hours. And then you'd be able to begin with the first word and get all the way to the last word.

But that's not really the required task actually very much in life. I mean, I'm there's there's surprisingly little one actually has to remember in sequence. There's a small amount of biology. There's some arbitrary tasks like, oh yeah, and there's poetry, of course, is a big thing you have to learn in sequence. But, but actually, there's not that much in language which needs to be learned in sequence. So, that's kind of part of the reason we haven't emphasized it too much.

I do actually think that there are some really interesting possibilities. And, um, I cannot wait for virtual reality tech. Oh yeah. To really hit the mains. I mean, this is a bit of a long one, long-term one. But I think in about three years time, when we've kind of completely nailed mobile stuff, maybe. Um, that will be like the next frontier. And there there are just incredible opportunities for structuring knowledge around, um, around memory palaces.

And, and the thing which is super interesting is that memory palaces are actually quite flawed in terms of their perfection because you're always using like an arbitrary space. So, I mean, for those, uh, viewers or whatever, who don't understand how this works, you basically take you take a space like this room, for instance, and you'd have like a sequence of say, twenty objects. And what you do is, rather than trying to remember them just as a string of, um, of verbal sounds, you actually attach them to the space in your imagination.

So, you say, okay, we'll begin with the postman here. Just imagine a postman here. And then over there, we're going to have a label. And then, you know, and you go around. And it turns out that, um, this is an exceptionally strong way of of, uh, installing memories in your mind. No doubt it has some kind of fundamental evolutionary story behind it. So, you know, when you're, you know, we're good at space, uh, animal, mammal brains have been doing this for whatever, 200 million years, you know. Uh, the brain's good at space.

But the thing is, is that the the actual space doesn't really encode any information other than the sequence because it's just an arbitrary space. You're just saying, okay, the next thing along, and then, okay, But there's a possibility with virtual reality that we could actually create spaces which themselves encode information.

So, so the the position of the door and the shape of the room and what happens when you walk out of the room and the next rooms along and so on, each of those things could contain structured information. So, rather than imagining things in the space, all you're doing is learning the space.

And I've been thinking about this for years. And, you know, and and so I used to do a thing called memory walks where you take a bunch of people, and you walk them around, and you'd learn say, all the prime ministers in order or the presidents or whatever. Um, but of course, if you actually had power over the space, you could actually make the space, um, you know, reflect, um, not just like, um, the sequence of the presidents, but you actually have the president there. And you could also have, um, um, their causal connections to the next president. And, you know, let's say there was a plague at that time, you could have an object which kind of signifies a plague and so on. And just literally by walking around the room, you could absorb the history.

And, you know, I in particular, I've used to build a school like this, an actual architectural space where, you know, if you were studying 13th-century history, you'd go to that point in the school, and you'd do your lesson there. So, all the memories would be attached to already meaningfully defined timeline.

But I suppose you could just have one room because the virtual reality can just generate the room. So, if you just have the space, there's a big like gym room or something, the the virtual reality system could actually just generate every Yeah. every reality based on what the subject.

Yeah. Um, yeah, definitely. And, and there's a I mean, there's a particular thing which I always think is very interesting, which is, um, making connections between abstract concepts. Um, sometimes also with history. Um, and this is like inherently a very challenging thing to remember.

You know, when you're studying philosophy or, um, some arcane area of cultural theory or whatever it is, um, just holding all the connections in your mind because it sometimes requires months of study to just sort of be, okay, I'm thinking about this concept, and of course, that relates to this way of framing and understanding, you know, this. And this is kind of what you're doing when you're a student and you're getting getting used to a new area of study.

But there's a possibility that you might be able to use memory palaces which are kind of like three-dimensional diagrams, you think about it. Yeah. Diagrams which you're walking around inside, and you can kind of look around at the rest of it, might be actually teach those kinds of complex concepts. Anyway, um, for the future.

So, but yeah, I mean, I think, um, just going back to Memrise, there's, um, there's two two things which which I think we we should hopefully make, uh, more of. I think that the video, um, the video provides a certain sense of space, you know. You you kind of remember not just the foreground but the background. You have a sense of the space there. Um, and this will help kind of contextualize and connect memories.

Um, and then hopefully through the road trip, um, you know, which, um, I should talk about more in the Membus. Um, hopefully through the road trip, we should be able to kind of create narratives within the courses. Um,

Right, so so it makes sense that you don't just have a sentence that says, John is going to the beach. You actually have you you knew the pre-story to that, or based on the memorized story, so you can if you're unsure of a word, you can always think back to the previous action, or and then go from there.

Yes. Yeah. And at the very least, think back to the context. So, we can sort of for for, um, I don't know, have have you ever seen a a Snapchat story?

A Snapchat story. Well, I've been Snapchatted messages but I haven't seen a story per se. Because I personally find Snapchat very boring as a messaging platform.

But, um, but they have this amazing thing where people can make their messages or the or the little snaps they do public. And they get kind of curatorially linked into scene. And then I, um, I saw one of these from Mecca. And it was just, you know, it was just a collection of whatever, five-second clips from next to each other. But it was kind of looking at it from the inside from every angle. You saw them drinking water. You saw them going out to pray. You saw their hotel accommodation, how they got there. You saw them chatting and eating and the rest of it.

And it was just kind of like, um, it was just like a sort of intimate documentary style that kind of well over all the stuff which actually happens. If you get me. Rather than just the normal iconic photo, or whatever, which tells you nothing. Uh, or gives you a very misleading global impression of it.

And, um, and I'd I'd I'd like the kind of Memrise courses in the future to have this kind of quality where, you know, you know, going to the beach in France is quite different from going to the beach in Germany. Or beach in Peru, you know, whatever. And to be able to capture the texture of these experiences with all the contextual phrases and all of the, um, all of the kind of quality would be pretty pretty beneficial.

And and that would then mean that when you actually were in a beach in Peru or what have you, you kind of had the lingo. You kind of knew what goes on. You maybe you're too surprised. You you're whatever. And in fact, it might it might tempt you to go there in the first place, which would be even better.

That's nice, too. Yeah.

Uh, one of the big discussions, uh, that has been going on for a while. And I haven't seen any new sort of studies released on this is, uh, should you learn from, uh, a translation, or should you have the original target language only?

And obviously, Memrise works a lot on going from a language to another one. Um, is it English obviously the the big one? Um, have you what's the first of all, what's the benefits of doing one or the other? And also, now that you're starting to support video, will you will the model at some point have videos that don't have any text under them, or will

That's a very interesting idea. That's a very interesting idea. I mean, so you just be in sort of pure video mode, and you just have to work it out. And maybe the action you'd have to It's always good to have people forced to interact because that mean actually then forces you to understand things rather than just let them fly by or what have you.

Um, um, well, I mean, I I I think that the obviously in some sense, when you're learning a phrase or a word or what have you, there's there's some kind of connection going on. So, you know, you learn the word for spoon, and you're kind of, um, you know, you've got to successfully connect the word to the thing in the world or, you know, which might in fact be a pattern of behavior and so on. So, so in some sense, you're always doing some kind of mapping.

Um, and so the kind of question I guess you could say is like, do you map via a verbal connection to the concept you already have, or simply via some other source of connection to the concept, which might be like an image or or whatever. Um, and, um, yeah, I mean, I think, um, you know, I think, um, I think it is obviously possible to learn merely from images and stuff. I mean, mainly, um, I mean, it's of course, the child is supposed to be the the the compelling us to this. Um, and, um, yeah, and of course, children are amazingly stupid and take ages to learn even the basics of a language. And they also have incredibly rich contextual information.

Uh, so, I mean, I think I I think it is a viable thing, and it obviously, um, in some sense, it has some some sort of obvious advantages. Like, um, there's no recourse to something you're comfortable with. And you're kind of thrown in a deep end. You know, whatever. You know, the rest of it. You're just thrown in to just doing.

Um, but, uh, I think, you know, certainly with kind of technology, I would be very very excited about designing a system which went, uh, just just straight to the language. But, uh, but I think, you know, if you can give people the experience of sufficiently rich contextual situations with sufficiently rich interactions, then then it could be compelling. Like, like if you go to a foreign country where nobody speaks your language, and you're just forced to pick it up, make sounds, and hope people respond to you like, you know, the spoon, spoon.

Right. Yeah, I was just thinking if you let's say, let's say you had to learn the word for spoon, there'd just be a picture of a spoon and no, and there would be maybe the word in the foreign language would be there, but it wouldn't say the English equivalent. You'd have to map the translation to the image.

Yeah, I mean, but even the image is not the it's not your internal conception. It's a it's an external visual symbol of that thing. Um, so, um, Yeah, I mean, and obviously like, obviously, nouns are the easy thing to go through.

But, you know, if you're sort of thinking like, um, if you're trying to understand a phrase like, she walked elegantly with a quizzical expression on her face. Right. You know. And it's like, okay, well, maybe you can make a visual representation of she walked elegantly.

But when it comes to like, she would have bought him, um, um, a nicer umbrella, um, but she decided that since he had no idea how to look after himself, uh, she would instead just give him a birthday card. Uh, you know, with that sort of thing, just like that's kind of elaborate thing to form a visual image of.

Very long sentences on the Memrise course. Not not long sentences, but things which refer to different modes and times and, um, and elements of reality which aren't directly visually represented in that thing.

So, I mean, I I I at one point I got quite into, um, cognitive linguistics, which, um, is a kind of area of, um, the cognitive science of language, um, acquisition and usage, which is, uh, it's just that way, it's full of geniuses, this discipline. You know, there's just amazing minds creating extraordinary diagrams to kind of often just to illustrate like, um, why a preposition is used a certain way in French and not in English.

So, you know, it's just like, you know, in French, you might say like, I drink in the cup, rather than I drink from the cup. Anyway, there'll be enormous diagrams of subtle fields of meaning interacting and all these different schemas and stuff to explain that simple thing. And it's thrilling from an intellectual perspective.

But, um, but didn't provide any immediate opportunities for like, oh, that's a great way to teach that because, um, even if it's getting right to the core of what the meaning actually is, sometimes, uh, sometimes it kind of for a kind of short-term, I just pure translation is just a bit easier.

Yeah, I suppose that's the thing.

Um, I was I have to say I mean, I have to say, I mean, I I I should add that, um, the, um, that, um, yeah, I'm by no means, uh, I'm by no means an expert on what the possibilities are, you know. Like, we have we have an approach of thing which we're kind of working on. But I certainly wouldn't rule out that, um, some budding young thing could, you know, invent an experience which makes just a pure like, learn from the image really compelling.

Right. Well, the whole idea is flawed anyway, I suppose, as a, you know, learning as a child. People often, you know, parents often say the best thing. But I always think it's very flawed, um, argument presentation. And as you say, it's it takes them years to even get to say a basic wrong sentence. Yes. So.

I mean, it's also the case that they are simultaneously building up their entire conception of the world. You know, from a level which is just right down to the basic sensory motor space, you know. Yeah. How do I stand up? And they have like one or two full-time peers all the time. I mean,

Exactly. If you had a French student who just walk around you for 24 hours a day, every day, I'm sure you'd learn French very quickly.

Yeah, I mean, in order to try and achieve that, I several times proposed to elegant Parisian ladies, but unfortunately, we haven't had any positive results so far.

Didn't work. That's one experiment you still have to do. And I think even if it works out successfully, they probably wouldn't want to walk around in circles chatting in French anyhow.

All right, well, um, uh, I was talking about earlier if we could give the people some very practical and actionable tips and tricks to how to improve their Memrise experience.

And, um, why don't we assume that I want to start a new language on Memrise, and I find the best course, which is probably created by you or you can see from the community popularity usually. So, I start my course.

And what are the things that when I before I even click the first word or sentence, what do I need to know to maximize my results and, uh,

Yeah. Well, I mean, I think the first thing is definitely to to click the button. Um, and I'm not entirely joking there. Um, you know, the difficult bit is always the beginning. Right. As in that, you know, and this is not just the first time you use it. But, as a Memrise user, every day, you know, you want to be using it as often as possible.

Um, and there's a kind of a neat little trick you can use, um, you know, let's say you're you find exercise a bit unpleasant. Well, it's a very good idea just to sort of do one press up. And then having got down and do one press up, you kind of go, okay, well, I'll do I'll do five or or whatever. Right. You know, I'll do twenty. It's a similar thing with learning.

So, just like getting, um, really like choosing an amount to learn each day, which, you know, which in its basic form doesn't add up to very much, is is a is a really good idea. So, um, on Memrise, you can set your yourself a daily goal. So, you can say, okay, I'm going to make, you know, whatever, learn a thousand points a day, you know, sort of five words a day. And getting into that rhythm is a really good idea.

I think the second tip, um, is to do with the use of mnemonics. So, we have these things mems on the site, which are user-generated, um, memory tips, which basically help you make the connection between a foreign word and a and your your native language in your mind. Um, and these are incredibly powerful if you use them.

Um, and sometimes they're not necessary, but it's a really good idea if you ever get a word wrong to kind of hold yourself up for five seconds, and just to really think, okay, you know, whether using a mem created by the community or in your own mind, just to explicitly make the connection because, um, you know, memories are connections in your mind. And the vast majority of the time, when you don't learn fast, it's because it's not because you forgot anything, it's because you just didn't really make the connection in the first place. And having that kind of explicit moment where you consciously connect one thing to another in your mind is, uh, is really powerful for them. So, that's, uh, that's that's another tip.

And then I I do recommend like adding a bunch of friends to your personal leaderboard. Um, because there's something about banter around, uh, competitive learning, which is deeply, um, deeply effective. Um, and then what else?

Um, you know, I think, um, you know, I think, um, you know, as I say, setting the goal, and making sure that it becomes a daily habit. Because you know, if you think about it, you learn ten words a day. It takes like fifteen minutes to learn ten words max, you know, or whatever. Yeah. You do that for a day once a day for a year, then you learn 3,650 words. Um, and that's pretty comfortably enough to be able to understand an news in a newspaper or or or whatever else. Um, and it's of course fairly addictive just to be just to be learning but, you know,

Right. And try to make sure to keep the reviews down every time. And and just I guess follow the the whatever it tells you to do, that's I mean, that's why you designed it that way, right? So, people don't have to think about when to review stuff themselves.

It'll come up at a sort of a what what does that when does it come up actually? What what is that defined by the the time it takes for it to request a new review?

So, um, one of the features of Memrise is a system based on the concept of spaced repetition, which basically says, you know, you've got a lot of memories in your brain, in fact, all of them are fading. Um, as that's a new take on life, I'm afraid to say. But in actually quite a good way because, you know, millions of things go into your mind every day, and most of them are complete nonsense. So, you kind of want everything to be fading unless it's important.

And the way, and the kind of hack the brain has for for realizing when something's important is just to say, well, if it repeats, then it's actually a valuable pattern in the world. If it repeats, or if it's exceptionally interesting and dangerous or beautiful or sexy or what have you. Uh, any-ho, so, all memories are fading. And they fade at different rates.

And one of the key things which determines how quickly a memory fades is basically how often you've seen it before because each time you get a repetition, it fades slightly slower the next time. And so, if I were to do that's if I can do this in a sort of gestural fashion. But if you know, memory goes into your head, it's immediately going, whoa, and about five minutes, you've almost forgotten it. But if you remind yourself then, um, it will be it'll go back up, and this time it'll fall slightly slower. So, after about an hour, you'll be about to get it. And doing it again, then it'll be it'll be the last a day this time. Actually, it'll be be much more solid. And if you remind yourself in a day, well, actually, this time it's going to take a month. And and so on and so forth.

Um, and it and it of course, it depends on the memory, it depends on who you are. And we try and track all these things. Um, and so the Memrise system is basically sort of is like a guardian for your memory. So, it tells you like, okay, you need to remember this one hand now, and this one now, and these five now, whatever. And and it takes looks across your whole past learning history.

Even if you've been learning for years, and you've learned like, funny enough, a guy came into the office today. I going to chat with Roberto, who's the first overlord I've ever met on Memrise. So, he had a 100 100 million points, he's learned more than 100,000 items. Anyway, he popped in, he's from Turin, in Italy, and he's like, hey, I'm an overlord. And so we hang out with him.

Um, and so he has like, um, you know, he has 100,000 things he's learned on Memrise. And and, you know, every day, we make a suggestion of the ones he should most review. And he goes through them. And that's how he can kind of keep that many items fresh in his memory.

He must have many reviews though every day.

Yeah. Yeah, he's I think he's he's quite a devoted student of Memrise. Yeah. Okay. I've talked about four or five hours a day learning.

That's impressive. Yeah. That's one of the problems I had when I was uh sort of really getting my head in was that suddenly the reviews would just stack up, and I would miss a day. And I'd have like 1,000 to review even before adding new words. And I I just crumbled. I don't know if there's any tricks or tips in that situation.

Um, well, I think in a way, that's bad product design. And we have, um, some, you know, because, um, because, you know, what's important is to maximize your chances of remembering as many things as possible.

And if if we start showing you numbers like, okay, you've got a thousand to review, that may be true. But probably if we showed you 500 or or or 200, um, and then just make sure you did it, then you'd end up learning more. So, again, it's that example of effectiveness and motivation.

Um, and good product design should probably have been a bit more encouraging. So, I do apologize about that. And we do have some on that. We're doing quite we're doing quite a lot of data work on our spacing algorithm at the moment.

Because, of course, you know, we've got millions of people who've learned tens of millions of words. So, so we have a lot of data. And we've got some quite interesting ideas about how we can improve that in the future so that you don't find yourself up against a wall of reviews. So,

Yeah. Yeah, that's the problem. And then then I suppose it's just, um, you know, keep using it every day, or as regularly as possible. And just keep the reviews down. And add new words. And I yeah, I guess the balance of adding new words is just to make sure not to add like a hundred words in one day and then no words for you know, you want to spread it out, right?

You do. That's actually been actually that that would be another tip for Memrise users is like, little and often. And this has been a um established principle of effective memorization since, uh, you know, early antiquity. So, um, you know, in in in in the early classical Greek texts.

Um, oh, by the way, I I must come back to I had this thought today because Marie has been talking about the Membus tour being epic. Oh yeah, that's true. Yeah. Yeah. And then I suddenly remember today that the word epic come from the Greek word epos, which means word. Um, and so so I suddenly struck me what an appropriate thing to call our Membus road trip epic because we will be, in fact, epically collecting hundreds of thousands of words.

Yeah. That's that's crazy for eleven languages, if uh the Kickstarter is successful. And just a reminder to everyone, you know, you can still get involved for a couple of days on the Kickstarter page. Just go search Membus, you'll find the Kickstarter page. You can get involved at various levels. And you can even join, you can get on the bus. You can actually be a part of the bus. I mean, how cool is that?

And, um, Ben, the uh, CEO, CEO of Memrise, has promised me a seat if, uh, I can find two others. So, uh, so, two will pay for for my seat if if that works out. Oh, great. Excellent. Yeah.

So, uh, so, if anyone's listening out there and they're like, I want to get on the bus, you know, remember to say Chris sent you. And, uh, we'll have a the most epic, uh, road trip under.

Um, I also did sort of, uh, on a whim say that if you make the 100k a goal in your Kickstarter, and somebody gets me on the bus, I'll get a a decently sized tattoo of Memrise somewhere.

That's very good. I I'd recommend That's all in, all in. Yeah, that's very good. I appreciate your devotion to the to the brand. And, um, we'll we'll we'll do our best to make sure it happens.

We still got about 40k to go, so you didn't need to uh, worry too much right now. Yeah. Right. Exactly.

But, uh, thank you so much for coming on, Ed. And, uh, sharing your love for memory and Memrise. And, uh, we can't wait to see the new features. I've I've seen some demos of the new video function on Memrise. And it looks fantastic. I hope my data plan doesn't get swallowed up on the first day. But,

You know what it won't actually. Our incredibly clever engineers have somehow managed to compress and serve the video in super small things. So, it's actually hardly more than a picture. So, you can be like, browse like Instagram, and it'll be fine.

Yeah. Amazing. All right. Well, thank you so much. And, I'm really looking forward to the, uh, Membus tour and, uh,

We can't wait to see you on the bus.

Yeah. I hope so. Otherwise, I'll just stoke the bus through Europe and run behind it or something. Yeah. Just come in the way. Okay. Have a great day. Thank you so much. Bye.

Thank you for listening to the actual fluency podcast.