Anki vs Memrise: In which platform can I learn words the fastest?

For people who compared them and took stats, what are your findings?
How many words per hour can you learn on each platform for similar type of material?
Where do you get better retention? Who has the better SRS algorithm? Who is the most effective of the two?

I sometimes wonder if the six repetitions per word are absolutely necessary, or maybe I can cut there a bit to save on time.

I’ve seen a guy who said he can learn at Anki 400 words in 3 hours.
I’m skeptical I can reach that at Memrise.

There is absolutely NO clear-cut answer to your question!

How many words of a new language you might be able to retain depends on sooooo many variables!

I have been using memrise to learn Swedish and have found it very helpful. I have learnt - according to memrise - over 11,000 words in three years, most of these Swedish.

BUT, for me, with a background in languages and a passion for learning and teaching them, I already had a head start, PLUS, I am fluent in German and my native language is English, which means that I am familiar with - via cognates - at least 50% of Swedish vocabulary the minute I meet it. So, with these factors present, I have been able to learn Swedish words very fast.

If I had chosen a language with a different alphabet and from a completely different language family, the situation would look completely different.

“learn 400 words in 3 hours” - this guy is talking UTTER RUBBISH!!!

What on earth does he even mean by that???

Equating quantity with quality is never a good way to judge anything, I’d say.

6 Likes

It is good to be skeptical because it is impossible to learn that quickly. If you consider that a native speaker knows about 20,000 words. By this logic , it would take 150 hours of studying to know every single word a native speaker knows. Absolutely false, ridiculous, etc etc. I would bet that after those 400 are “learned”, the 400 new words that person learns the next day involve 395 of the ones from yesterday and 5 new ones.

For reference - I have actually learned around 6,000 words in about <1.5 years on Memrise. I learn Chinese and my native language is English . Those are numbers you can expect if you work hard I think.

3 Likes

according to most contemporary handbooks on the psy of learning, the average passive vocabulary of an English native is around 150,000. Collegiate dictionaries have around 200,000 entries, or similar. It is said (not “it”, in fact David Crystal) that English has a vocabulary of around 500,000 (of course, with archaisms and specialised languages). I do bet that German, French etc have similar numbers (ok, let us please leave aside the fantafab claims of some arrogant English natives that English would be the richest language on earth…)

1 Like

I don’t think that is indeed the case. I consider the source to be very reliable.

To give you an example, in early 2011, I conducted a small and very unscientific experiment to see how well this method works for learning vocabulary. Naturally, I’ve been using similar methods almost from the start, but I decided to test the limits. Going through the vocabulary list to an advanced level proficiency test, I found there were around 2000 words I didn’t know. It took me little more than three hours a day for five days to go through that list, averaging about 400 words per day, 135 words per hour or just over two words per minute. Using spaced repetition software, I was able to retain about 95% of these words, spending another ten hours over the following weeks. This doesn’t mean that I could use all these words properly, though, and learning words are of course only one part of language learning.

See also his comment to laurenth down below in the comment section of the article.

I will suggest to read this too: http://www.hackingchinese.com/when-perfectionism-becomes-an-obstacle-to-progress/

Rough summary:

When perfectionism becomes an obstacle
I’m prepared to say that perfectionism is always bad if taken to extremes, with the possible exception of pronunciation. I really think that learning tones and sounds properly in the beginning is well worth the effort in a way that writing characters beautifully or perfectly is not. So why is perfectionism so bad? Because it is inefficient. Spending too much time on something might mean that you spend less time on widening your horizons and learning more, which would in the end lead to better results.

Smart, not lazy
I think those of you who frequent this website know that I don’t advocate this approach because I’m lazy. The point here is that if the idea is to get really good at something (like Chinese) , focusing on knowing everything you learn until you know it perfectly is a waste of time. That time could be spent practising pronunciation, broadening vocabulary or reinforcing grammar. I’m sure this gives a lot more in return for the time invested. I don’t mean to say that a solid foundation isn’t important (see my comment about pronunciation above), but I’m saying that you won’t get very far if you spend years just laying the foundations. I’d much rather learn 100 words and remember 90%, than learn 50 words in the same time and remember every single one. I’d much rather be able to write 100 characters other people can understand than 10 people think are beautiful.

I’m going to surpass that in 3 months of study. I could have easily condensed it into 2 even.
My max achievable pace for me at Memrise is probably close to 76 words/hour.

According to Memrise, the mean is about 60 words/hour, based on their time completion estimation for an individual course.

His speed is basically about double that.
If I see each word 3 times, or half, before moving on to the next, that looks to be reasonably within reach.

See response above and also this article: How To: Learn 100,000 Words by 8th Grade

There is a difference between recognizing words and knowing actually how to use them.
Your estimation for the knowledge level of a native speaker is also quite low in my opinion.
I will also add that this didn’t account for reviewing, but rather just the pace of studying new materials.

1 Like

I use both anki and memrise for one of my own courses. I like the flexibility of anki but honestly memrise keeps me more motivated. As in number of words learned: I think that’s up to what your brain can handle, not what the platform offers …

There is a script for that (memrise auto learn) by cooljingle - lets you skip the repetitions if you think you don’t need them (for individual words).

Going from first principles, of course it’s going to be Anki. Different review intervals if the word is easy or difficult, options to set your own review intervals, and no review interval cap. And, I think, easier to make courses, with more options and stuff you can add.

How much of a the difference that makes in your learning I think depends on how difficult to remember what you’re learning is for you.

The thing Memrise has going for it is its community-created courses, community, gamification, and ease of use. I couldn’t get into Anki, I very quickly got into Memrise.

1 Like

Have you tried http://www.languagecourse.net/vocabulary-trainer.php vocabulary trainer VT, which is available e.g for Android?

They even have a short FAQ on the website, once you step into the VT url with some interesting questions and answers.

IMHO it is much faster than Memrise “to learn” (let’s say to be presented) new vocabulary, as they cut out the 6 planting steps.
They use use a Leitner 6 index card box system for review, just like Digital Publishing software does it.

However, I did find some bugs on their VT website for a test quiz and also storing my words and index card boxes progress on their server was a bit (too) buggy for me.
I lost many words (stored/learned words list can only be seen on the Android app) when I tried to do offline practice on Android@Bluestacks because of missing WLAN.
So I have no clue what the server did and why sync errors on Android can result in loosing words stored on the server.
I am not sure if there even is a real offline / cache mode for learning / review words without Internet (on Android) and syncing the state back to the online server, once you are at home.

My tries where on Bluestacks Android emulator software, so it may be not relevant, as I also have some severe problems with the offline/cache mode (Non-Pro) of the Android DuoLingo app, where the app crashed, did not restart and I lost all multiple lessons…
I have no idea if the are related or not.

But you generally wanted to hear about different concepts, so either VT or DP are different concepts than Memrise, AnkiSRS, DuoLingo, etc.

Have you already tested “Flashcards Deluxe” (there was a longer answer about this software in another Memrise community thread from someone) or “SuperMemo”?


So currently I am not using VT, but my plan was to eventually contact their technical support team, once I maybe have a free minute and the nerve to play/fight with them about my findings.

Usually I am alwaysbored of 1st/2nd level support as most of the time these people are the wrong ones or even give out wrong informations.
So actually I would have to directly speak to their 3rd level, the app programmers themself, or write the app specialists a longer e-mail, which is not that easy too accomplish too.
Chat / Skype or something would probably be the best to track issues together or find out interactively how the app / server sync / offline usage should be working as designed.
I need to personally hear that from a senior programmer by himself…
And the company is not located in Germany, so I am not keen on it to give them a phone call either :wink:

Sometimes because I have been learning Chinese I forget that words and word families are different due to verb conjugations and adapting root words for different parts of speech. The number of word families an educated native speaker knows on average is 20,000.

Considering you seem to be learning the language I am learning, I will let you know that learning to recognize characters is very easy because of how visual and component based they are. Additionally 80% of the characters are semanto-phonetic (I think that blog you linked has an article on that).

If your goal is to be able to simply recognize words, then yes you could sustain 76 words/hour. But at my peak of learning 50 words per day for a period of 60 days… I reached a point where I was doing 1000 flashcard reviews every 24 hours. You will reach a saturation point where your entire day will be soaked up doing review. Do the math yourself. Unless you intentionally forget or stop reviewing these words, Memrise will not be the platform for you, and you will burn out and quit.

I fundamentally question your statements about “76 words/hour” , it doesn’t matter how long it takes to click… what matters is an arbitrary X number of days from now what percentage of those words do you know unconsciously without thinking about it. That is the level needed to begin reading comfortably, speaking, listening, or any practical language skill.

Within a week you will understand the leaky nature of your memory and you will re-adjust your claim. Anyone who has successfully learned a language can attest to the fact that learning that many non-cognate words in an hour is simply impossible.

5 Likes

I will answer people in stages as to not make this post too long.

Not really. There can be potentially a lot of embedded inefficiencies in the code.

Even the moving time between screens when answering questions and the waiting time before starting a new session can waste in accumulative a lot of time.

I actually found today something to elevate just that with a the Memrise+ script.

Here is another cool feature of Anki to increase efficiency which doesn’t really exist here: http://www.hackingchinese.com/killing-leeches/
It similar to the idea of the difficult words section here but not quite.

Also consider that Anki is self-evaluation, so even if you accidentally mistype something you can still mark it as good and continue as if nothing had happened.

There are a lot of settings in Anki so you can really custom fit it to your needs.
It might worth the extra time it takes to learn how to use it, as the savings and the increased efficiency can be great since we do it every day and for many hours.

Here it is more a one solution fits all kind of thing, so ,understandably, it will not be optimized, if someone is really looking to boost efficiency the best he can.

Yes, I’m familiar with it, but I’ve never considered to use it in that manner.
It can be hard to know when is the correct time. You might think you know it well enough, but only to be mistaken shortly after. It might require some experimentation to see if you can lower it first to 5 and then 4, without an adverse effect on retention. it’s an interesting suggestion.
On the other hand, it might also be a bad idea to trick the algorithm like that…
Since no one really tested it it’s hard to know if it’s really working or not.

How does the introductory phase work at Anki exactly? I’ve never really used it yet.

I too didn’t really like it too much from first glance (which is why I’m here and not there), but maybe it’s just a matter of getting used to it? I’ve spent here more than 2 months at Memrise and I’m still learning new things every day, so one must really spend a lot of time with an app to really know it and get used to it.
The community is excellent. Much better than what is going on with duolingo, I would say.
But does the points really that much matter to you? We’re here to learn not to collect points. If it’s really that much more effiecnt why not move to it? You might shave off years at the end…

Anki is not community created courses? How is it different? There are shared decks too.

Why won’t you guys actually test it with 2 different courses on the 2 platforms, and see what gives you better results after a 1 month period? That is what I’m considering doing in Oktober.

Anki is a big name in the industry, which a lot of people use and like very much, so giving it a real shot is a given for me. That is what I believe.

For the accidental user, who just want to learn a language or two, I guess it does not make that much of a difference at the end, but I’m shooting for an overload position and that is a lot of words and work ahead of me. I’m not sure Memrise is really suited to handle such big quantities.

the same goes for Memrise…

one could use the same amount of time to read sth serious about learning, instead of various easy-not really accurate-not even true stuff that learning sites offer for the potential client

I didn’t try any of them yet but heard about some of them. Getting to know each app closely really takes a long time, so every few months I’m just testing one and see if it’s any better. At the moment, I’m focused only on the big names, Anki/Memrise, which I heard about from many different people and sources.
I don’t think I need so many different flashcard apps, but I’ll give them a try anyway, so thanks.

I’ve installed it today. Looks quite good. There is one thing that bugs me though. There is an annoying sound every time I tap. Do you know how to disable it by any chance?

I’ve been doing that for 2 months now at a higher pace than that, exactly to test if it is sustainable over a long period rather than just in bursts. The first month I average 133 words per day, and with many days going over 200. The second month, which I’m now closing, I took first a break for 20 days, without reviews, to use ReadLang instead, and with the last 10 days of the month, I’m going to end up with about an additional 2000 words, while managing to reduce my backlog down to zero. I don’t have yet a steady pace, as it’s difficult to keep up with, but I’m woking on it. The reviews are sustainable since I’m able to review more than 1200 words in 2 hours using exclusively speed reviews. If I could disable typing tests I will only use the blue review instead. Is that possible? That will help greatly.

I’m trying to find an equilibrium point between review and study that will not exceed 5 hours a day. That is something that I aim to reach a stable sustainable point, if I have the time of course. Whatever it will end up to be the final pace, I want to be able to reach 0 reviews per day, if I wanted. I actually not too worried if I accumulate some backlog and review things late, as the adverse effects don’t impact me greatly. I’m still searching for a balance point as every day I’m learning new ways to improve inefficiencies, such as the catch-up review script, which I didn’t use until today, but can really help. I’ve tested it and it works great! I’ve used it on a course which teaches the Arabic script, which I didn’t review for 2 months since completing it, and it pushed the words for 96 days or something. A real time saver. I knew I don’t need so many reviews there, so it was the perfect application for it.

I’m preparing now for the final month. The goal is about 5000 words for this month. I’ve found an incredible plugin for Memrise called Memrise on Steroids, which will greatly help me keep my goals and be on top of things. I’ve set goals for all courses I’m going to study and looking forward to trying it. The goals are the numbers in yellow.

I’ve set the goals in such a way that it will spread out over quite a few courses (about 8 languages) but no more than a total of 100,000 points a day, which I will spend just on learning new materials. I expect that should take about 3 hours or a little bit more than that. If I exceed the allocated time of 3 hours I will lower my goal to compensate. An additional 2 hours I will spend on reviews, so I expect to end up with about 150,000 points daily, more or less. If 5 hours is exceeded, again, I will continually try to adjust it. That is the plan at least.

Yesterday backlog was 0. Today it climbed up to 1380 (indicated by the number in blue. Green are words that I’ve yet to study).

In the meantime, I’m still doing a lot of homework to try to learn the functionally of all the different scripts, so I can make an overall a better use of the site with better judgement.

3 Likes

The problem is that with Memrise it feels more like a lot of patches, which is cumbersome. I hate to think what will happen if they overhaul their site one day, and all the scripts, which I began to heavily rely upon will suddenly become broken and obsolete, with no one to fix them.
There is also core functionality which is out of our reach to modify.
There is also the added risk they’ll decide to become a paid only service one day.

With the help of a few scripts, I’m happy to report I was able to improve my pace to 93 words per hour.
The gap is closing. Might be no need to check Anki after all. :slight_smile:
Still, I need to do it a little bit faster.

A lesson in Swedish 6.

Edit: A lesson in Spanish 5. 120 pace.
That is already basically almost as the pace he reported.

1 Like

Now it’s getting weird…

I’ve shattered it. :joy::hugs:
A pace of 234.
Too bad I can’t take a 1-hour lesson like this.

A lesson in Spanish 5.

@Casper_duo

In the User-Created Global Leaderboard thread, you said:

If all questions are 150, you can even get to 160K per hour. The problem is that I don’t get enough to review… And I’m doing more than 200 words per day. My review count is 0 now almost always as I’m doing them immediately. Very slowly racking up which makes it difficult to gather a lot of points.

At some point, I did some calculations estimating how many words per day I should plant if I wanted to earn 100,000 points per day using normal review (no speed review). Roughly speaking, if you were starting with a new account, and you planted 100 words per day, after about three weeks, you’d have enough words to water each day for close to 100,000 points.

Note: My math assumed that you always get everything correct…

To reach Overlord in one year, you’d need ~270,000 points per day, which means planting 250-300 words per day (depending on your accuracy and how many words you already have planted and coming up for review.)

I’m not doing speed reviews anymore as there is really no point.
I can get the same experience from the blue review with the scripts. It’s more beneficial points wise and also the lack of the timer provides better results.

The minimum you need is 666 words for review to reach 100,000.

Learning 300 words per day is about 100,000 points.
That will leave 170,000 points that you need to get only from reviews (or 1132 words).

How many potential review points can you get at such a pace, day in day out, I’m not sure.
The math is more complicated.

Planting 300 words a day has become less of an issue as I can now reach above 100 per hour.
The overall speed has increased greatly.

It looks like I wasn’t clear enough in my explanation. If you plant 100 words every day, you’ll have enough words coming due for review (after three weeks) so that between planting and reviewing, you can earn 100,000 points per day.

Can you show me your math?

At which streak do you receive 150 points? 8/8?

That will suggest that I will need to wait until I reach 96 days, before I get 600 words per day, at a 100 per day pace.
And I will need more than 600 to reach 100,000 points. This also takes into account a perfect score.

I’ve not been learning yet for such a continuous prolong time without breaks, but I understand it will sort itself out in the more distant future if I’ll be consistent. The pace seems to be sufficient enough to support it.

1 Like