by [anonymous]2 min read20th May 201274 comments


Personal Blog

The following protocol is very dumb, and relies on a lot of mental brute force, but I find that it works very well indeed.

First, learn the alphabet. the most basic survival phrases, the phonetics, the basic grammar. This may be the steepest part of the learning curve. I recommend that you get at least an introductory book to help you through this phase, the A1 level.

Now comes the fun part.

The key element to accelerated and efficient language learning is wanting, needing, craving to understand and to be understood. Choose a text, any text, which you know will trigger all your "I want to know what it says!" instincts. I myself find that highly dramatic works, with lots of suspense and high emotional torque, are ideal. Take a dictionary and just look up every single word that you don't know. You write each one of them down, with, first, the exact pronunciation, then, their definition in the original language, the translation to your own language, and, optionally (and I do recommend taking that option), a couple of examples of its use, better if you come up with them yourself.

It's very intense, and one can get deeply immersed in the flow, so pay attention to the clock: there's a very high risk of Tetris Effect/Just... One... More... Word... effect taking place. Like going to the gym, you need to pace yourself: if you go everyday for two weeks and then give up for three months, we won't be achieving much. Keep it at six hours weekly maximum if you're doing this on the side, three hours is a reasonable rate.

The next step would be, once you're fairly confident you won't make a fool of yourself, to join a forum where topics you care a lot about are discussed very seriously, and then trying to contribute to the discussion. This will force you to write a lot, very quickly, and your interlocutors will be very unforgiving of mistakes, so you'll be very motivated to check and double check. Giving a teacher a sloppy piece is simply laziness, giving it to a discussion board is an affront.

I focus this much on the written media because that's where you'll get the most information bandwidth, so to speak, and because it's both easier to pick apart and to put together than the oral language, kind of like the difference between a turn-based game and a real-time strategy game. Subtitled movies can be a fun, low intensity tool, but the difficulty when learning languages other than English is to find versions with good subtitles (you DO NOT WANT the bad subtitles).

But you may well have to learn to speak the language properly, and one of the fast ways of improving one's ability to learn languages is, well, singing. Usually there's opportunities to join a choir for free (especially if you're okay with church-y environments), and they're a wonderful learning opportunity for beginners. The more you master voice and rhythm, tone and timbre, the easier you will find it to attune yourself to any particular language's set of sounds and paces.

And know this: the more languages you learn, the easier it becomes to learn more. Additionally, the same way learning programming languages helps you clarify the way you think and pattern your thoughts more logically, learning new human languages will give you a firmer grasp of the metaphors on which it is all built, and a better understanding of both universal human psychology, and your own language (like Monsieur Jourdain, who found out to his great amazement that, his whole life, he had been speaking in prose... but, hopefully, what you find out will be more meaningful in all sorts of ways).



New Comment
74 comments, sorted by Click to highlight new comments since: Today at 1:37 AM
Some comments are truncated due to high volume. (⌘F to expand all)Change truncation settings

Evidence that this works better than other methods being...

Seriously, with such a huge number of people trying to learn a second language (like 90% of all humans) we should have some proper studies by now.


That's essentially how I learned English. You need to add an important caveat: it takes about 10-20 years.

(ETA: on second thought, I'm maybe overstating it - what I wanted to point out is that it may take a long time to learn a new language without immersion - but that it's doable at all is good news.)

You need to add an important caveat: it takes about 10-20 years.

Learning languages is expensive; I'm not sure it's a good idea at all for ordinary English speakers, and I see a distinct lack of calculation in this post or other comments.

If you plan to never, ever live in a non-English-speaking place, yeah, learning languages other than English is not terribly useful. Anyway, the utility function is not up for grabs, so if the reason I'm learning Irish is because I want to considerations of usefulness are not too relevant.

If you plan to never, ever live in a non-English-speaking place, yeah, learning languages other than English is not terribly useful.

Living in the same country you were born in is the lot of something like 90% of humanity and usually has been, and English-speaking may well embrace more territory than you might guess.

For example, when I visited Belgium and the Netherlands in 2005, I was shocked at how many natives spoke English and how well.

And just today, while finishing a Wikipedia experiment, I was surprised to note that the smaller German Wikipedia was sending something like half as many visits to my (English) DNB FAQ as the larger English Wikipedia. (Although I just checked, and contrary to what I thought, the German entry wasn't specifying that my page is in English, which is likely inflating click-throughs; I've added an 'Englisch' warning, so we'll see how things change after 100 days or so. Good to know this for context in my experiment, too.)

As a Dutch person with a German girlfriend, I'm in both countries quite often. It's common knowledge in both countries that the Dutch are good at English, and it's common knowledge in Germany that the Germans are not very good at English. Apart from that, fully English courses, or just English lecture slides, are common in our exact sciences university. In Germany apparently not so much, although I don't have first hand experience. Looking up actual numbers, this seems to be somewhat true. The English Language in Europe [] wikipedia page has a nice bar graph [] and map [], created from data from an EU survey [] In the Netherlands, 87% indicate that they speak English. In Germany it's 51% and in Belgium it's 52%. Across all of Europe, it's 51%. Oh, and if you're ever back in the Netherlands, you're welcome to drop by :)
Probably, but I doubt this figure will stay this high much longer, especially for people born in countries living economic hard times such as Ireland. (Also, people who stay in their home country tend to get lower-status jobs than those who emigrate.) But it might be further away from where you are. Most people emigrating from Ireland and Britain end up in North America or Australia just because they don't have decent-enough German or French to try mainland Europe. Northern Europe is unusual in that respect. Staying a few days in Paris without decent French was awful, so I can't even imagine how it would be to stay there for a longer period without learning French. (And while I never had any kind of language problems in the Netherlands, I still guess that if I looked for a job there it'd be a helluva lot harder to find one than if I could speak Dutch.) Well, the kind of people who are likely to be interested in stuff like DNB are likely to already be able to read English, wherever they grew up (if anything, I'm surprised that lots of people read the article on GNB rather than the one in the first place); but this needn't generalize to most two-digit-IQ people, and if you live in a country you'll have to interact with them too.
Statistics []. My guess is that the percentage for Belgium would be a lot higher if you excluded the part that speaks French.
So to followup the experiment: I checked Google Analytics [] now and was surprised to see that after I added the English warning, late May - early June 2012 saw a big spike in traffic from the German Wikipedia to the DNB FAQ, many times the daily average. Presumably this must have been from some article in the German media on dual n-back. Unfortunately, the data ends at 10 June 2012... because that was when I moved to Amazon S3 to save on hosting costs, breaking all my existing redirects (Amazon S3 doesn't do .htaccess or redirects) - and the German article was pointing to a redirect. D'oh! Anyway, there's no obvious sudden drop on 20 May 2012, but the big spike contaminates all the relevant days. I guess I'll wait another ~100 days and see what happens now that everything should be working properly...
Update: traffic has still not recovered: [] I infer this means the English warning is indeed deterring a lot of traffic.
You seem to imply that if you want to some day live in a non-English-speaking place, you should learn languages now. No, only if you know where you'll eventually want to live should you learn that language now (and maybe you should wait for immersion, which is easier).
Learning a second language is likely to make it easier to learn a third language afterwards, and so on.
Yes, but I think the expected value of correctly guessing where you want to live is higher than the value from this effect. But I think you're just rationalizing. Compare this comment of yours to the earlier one. Are they really talking about the same thing? Are either of them responsive to Gwern?
I meant something like “if you think you might live in an A-speaking country you should learn A, but even if you think you might live abroad but you're not sure where, it's better to learn some language A than no foreign language at all, because even if you end up living in a B-speaking country, learning B as an adult who has never learnt an additional language before will be harder than if you had already learnt a foreign language A when younger.” Gwern's point is that learning a language other than English is expensive and useless, and I'm pointing out cases when it's not useless (and, in the great-grandparent, that I want to do something expensive just for the hell of it it's my own business -- going to the cinema once a week would also be expensive and useless, for example). Or am I missing something?
Gwern didn't use the word "useless." If it's useless, the magnitude of expense is irrelevant. The important thing is to compare costs to benefits. Listing benefits without quantifying them doesn't contribute to this. Incidentally, the particular benefit you mentioned was in the original post.
Not necessarily: there are such things as terminal values. The utility function is not up for grabs.
Aren't there lateral benefits to learning something as complex as a new language? The level of mental focus and commitment required must have cognitive rewards and I would think any level of cognitive improvement would be of great value. In order to learn any language, it requires a certain level of immersion in cultural concepts/perspectives outside of your own. Broadening cultural awareness and gaining new perspectives certainly contributes to an individual's ability to see the world with increased clarity. It seems to me that measuring the worth of learning anything in terms of how directly one might make use of it cannot measure its total value.
Indeed. “It's expensive to learn and it has no obvious immediate practical benefits” applies to lots []¹ of skills/knowledge, so it seems silly to me to single any of them out. 1. Well, cooking is a bad example, but you'll be able of thinking of better ones.
I can't say I will never live in a non-English speaking place, but since I have no particular plans to at the moment, I have no reason to learn any particular language out of hundreds, either.
As I said elsewhere, learning a language is likely to make it easier to learn another one later on; also, some languages are more influential [] than others, so if you choose to learn German or French (if you're in Europe) or Spanish (in America) you have a non-negligible probability of finding the very language you chose useful in the future.
So, the plan is: invest 1000 hours now, so that when you actually need to learn a language you only have to invest 800 hours then? Anyway, I hear you saying that you have intrinsic motivation here, so these kind of calculations aren't really relevant to you, which I understand.
Out of curiosity, and related to what you said upthread: are you learning Irish?
Yes. (I took an introductory course in UCD when I was on an exchange there last year, the course on the BBC website, and now I'm taking Gaeilge gan Stró! - Lower Intermediate Level from
Cool! I don't know that website; is it any good? Má bhéadh aon suim agat chun cleactadh Gaeilge a dheanamh, chuir PM dom! Ní féidir liom é a labhairt go líofa, ach tá mé ceart go leor.
Deep immersion courses are freaking awesome. It's almost like sink-or-swim, but you've got handlebars and floaters and lifeguards. The big trap is to start speaking English with everyone rather than whatever language you're learning, but that's inevitable at lower levels. Also, you meet lots of awesome people: you see, it takes a special kind of folk to go spend a prolonged period of time by themselves in a foreign country learning a foreign language. They tend to display a fairly large subset of the rationalist virtues [].
I have seen courses using so-called direct method, where speaking any language except the one you learn is strictly forbidden. (It helps if the participants come from different countries.) How is it even possible to start? Pictures. A lot of pictures. It is important to have a huge database of unambiguous pictures. Sometimes you use body movements to ilustrate a concept (just like people do naturally when they don't know the word). Gradually it becomes possible to have a trivial conversation. Also the lessons take a few hours, to repeat and repeat the given amount of material again and again. I have seen this method used for teaching Esperanto, and one week (4-5 hours a day) was enough to make one able to speak simply about everyday topics. (Here is the textbook [], but I suppose it would not have the same effect without the real teacher and classmates.)
This is basically what Rosetta Stone does. It can be pretty effective, even without social reinforcement.
If your goal is to speak the language with no substantial disadvantage compared to a native speaker, yeah, but learning to have basic conversations and to get the gist of written texts will often take one year or even less, especially if your language is not too different from the one you're learning (e.g. both are Standard Average European [] languages), and especially if you've learned additional languages before.

For a website by somebody with a successful track record of language learning, have a look at . As well as a lot of motivation talk (which overlaps LessWrong's akrasia discussion), it focuses on two very important concepts:

  1. Learning in real context Rather than isolated words, you should look at whole sentences or paragraphs. These should be from real texts or spoken language aimed at native speakers (start with children's books if you want lower difficulty). Definitions are useful, and Rational_Brony's advice to look them up in the original language is good, but a definition only tells you what a words means, not the subtleties of when to use it. The best way to learn that is exposure to huge quantities of real context.

  2. Spaced repetition software This dramatically increases your memorization eficacy, and it works very well for language learning. gwern wrote an excellent article about it which you should read if you're not already familiar with it:

How many languages are you fluent in?

Which ones, out of curiosity.

Very nice. I disagree with those who said it should have gone in discussion. Too much bolding, way, way too much bolding. Consider including links to, or discussion of Fluent In 3 Months or All Japanese All The Time

Speaking as someone with okay German and poor French who's currently learning Mandarin, gwern is correct, the returns for learning a language other than English that you don't need to use now are pretty low. It's really, really hard to get good at any language, and almost all multi-national workplace, English will be either the working language... (read more)


I would say it helps a great deal with pronunciation to just go around making funny noises. Experiment with what sounds you can make orally, it really helps. Then learn some IPA. I have done so, and I have yet to encounter a language with a sound I cannot pronounce.

Well, making weird sounds on command is one thing, making them in the right place in the middle of a word in a normal sentence speaking at a normal rate is another, at least for me. (The extreme example is tongue-twisters, which can be very hard even if each single sound individually isn't.)
I concur. For instance, as a speaker of Australian English I not only don't pronounce the 'r' at the ends of syllables/words in English, I also speak a non-rhotic version of all my second languages. Living in the US and being constantly mocked and made aware of this difference for a full year has only made me marginally better at producing them on demand.
It so happens to be that I can do most tongue twisters with little effort, and that normal sentences are rarely tongue twisters.
I'm relatively good with tongue-twisters too, but I still mix up sounds which I didn't grow up using regularly if I'm not very careful, e.g. sometimes when speaking English quickly I accidentally use the Italian-style r (which sounds more or less like the English flapped t) instead of the English-style one, leading to misunderstandings like “what is” when I mean “where is” and the like.

I'm gradually learning German through song lyrics. I just open a page with lyrics and Google Translate, put on the song and try to make sense of it, memorizing words and expressions. When a real English translation is available, I check that afterwards to see how I've done.

There's a wonderful site called Leo [], which has an absolutely awesome German-to-Other-Languages dictionary. There's a wonderful Leo Dictionary Add-On available on both Firefox and Chrome which allows you to look a word up in there just by selecting and clicking on the option in the contextual menu.
Leo, in my line of work, is one of the most useful resources on the web. I don't think there are comparable resources for any other languages besides German, which is really too bad. (If anyone knows different, please link!)
There's [], which is very nice. Not quite as super special awesome as Leo, but nice.
Thanks for the pointer, I'll check it out.

@Reasons for not learning a language:

The more languages you learn, the more access you have to humanity's written riches. That includes newspapers, research papers, literature, scientific and otherwise. Relying on translators to bring that material to your language is a dicey proposal, especially if you're looking for something very specialized. It's also a lot more comfortable to read stuff in good Native Language than in horribly butchered English, which is all too often the case with science papers (not that scientists as a whole tend to show much writi... (read more)

Also, this [] was posted a few weeks ago, though it's somewhat disputed.

I'm surprised that nobody seems to have brought up any mental benefits of speaking more than one language. I'm not sure how strong the evidence is, but there has definitely been research that claims to point in that direction.

As goes for most of the arguments for learning second languages for reasons other than actually speaking said language: what makes you think that this unspecified advantage applies more to learning languages than to the other competing possibilities of intellectually challenging activities like learning math/statistics/programming/art/etc etc.?
Of course (I think I should have pointed that out in my first post), but physics/math/etc also take a long time to learn properly, so the time required becomes much less relevant. I do not mean that one should necessarily learn a new language instead of learning math - although I might say that if you already know a lot of math (enough to get significant benefit to your thought processes), it might be useful to spend some time learning something that trains different aspects of mental processing (like learning a language). If I had to speculate on what any specific benefits might be, I would suggest that it comes from having more than one independent lens through which you interpret the world/more than one basis for your thought processes, and the benefit to mental flexibility you can get from switching between them (I'm not sure if I am properly communicating what I mean by that though).

I used this technique to learn to read German in about 300 hours (1 hour / day for a year).

However I could not find a way to use it to learn Chinese or Japanese, where reading is very hard and dictionary lookup is very slow due to the non-phonetic characters. I would love to find a way around this.

Lookup will accelerate over time. For example, while the characters are non-phonetic, there are many characters with the same pinyin (minus the tones). Just by looking at a character and noticing that it is similar in shape to one you already know the pinyin for, you can often guess that the pinyin will be the same. For example, "据 剧 居" are all "ju" (with different tones) in Chinese, so by knowing one, you can easily look up the other two. There are probably even more, but these are the 3 I'm familiar with at the moment. Another option would be to watch Chinese or Japanese TV programs, either on DVD or computer so that you can pause and rewind. Almost all Chinese TV programs are subtitled, so you can hear the pronunciation and use that to look up the character if it is unfamiliar.
If you use Firefox Rikaichan [] is an option (Note, I'm not sure if it's compatible with the latest version; haven't upgraded in forever). Chrome has a version named rikaikun [] is also an option. Also, install an IME so that you can type in the language, I would personally recommend google IME because it's much more comprehensive than the default microsoft one. You can use this to check words if you think you know how they're pronounced, typing it out in a text field and rikaiing it if you want a super quick reference on the computer. For getting faster at dictionary lookup, memorize the basic radicals and understand how the strokes are done. You can use Heisig to do this (Be warned, while Heisig advertises itself as a kanji learning tool, it doesn't actually teach you anything you would want to know, such as proper usage, pronunciation, nuance and cultural context) or you can work through Kanjidamage [] (WARNING: PERHAPS, MAYBE, CONTAINS MANY PENIS JOKES USE WITH CAUTION ). I have a recommendation for works you could read and which are compatible with all the tools I have outlined so far but have a preference against revealing it. PM me if you want to know. Other helpful resources: A dictionary [] Phrase lookups [] Both are completely in Japanese, although the dictionary has both E-J and J-J options (英和 and 和英 radio buttons). Unfortunately I have no answer to the Chinese question, or rather, the answer is "grow up in China and then let your reading and writing atrophy because you like English better." There's a dictionary [] but otherwise I'm at a loss at what else would help.

Ach! I'm getting downvoted... is it because I emphasize everything? It can't be because of lack of structure, and it's definitely informative, and the paragraphs are even... The sentences are too long?

The frequent bolding did make it hard to read. Also, changes in fonts like this are frequently a sign of crankiness so they likely bring up negative associations. People may also be downvoting for the post being in Main rather than Discussion. It is on the borderline of which of those one might reasonably put it- I suspect that the lack of sources/citations may push it more into the Discussion section.

We have a "No Original Research" rule?

We have a rule that discounts anecdotal evidence appropriately.

Most posts in the sequences don't have any citations. A lot of the sequences look like original research to me, but I don't know all the research that has been done, so I might be wrong. [] []
37 ways that words can be wrong is a sort of meta-post that is meant to tie together other posts, so I wouldn't use it as an example unless you're implying that all of them go uncited or completely unresearched.
Yes, I daresay that the expected quality of mainspace posts is substantially higher than some of the posts in the Sequences. And that's ok.
I disagree; I think our standards are too high for main. If we lowered our standards maybe main would actually see activity. When people see posts like this getting downvoted they learn that Less Wrong probably doesn't want to read what they write. Even if the stuff they want to write is pretty good.
I didn't downvote, but 1) I think this belongs in Discussion not Main, and 2) I would have expected an article with such a title to be about reasons to learn a new language -- you should have titled it “How to Learn a New Language” instead
I'd also recommend an introductory paragraph, where you explain what the post is going to be about, your basis for believing your information is correct, etc. Something like "this is a post describing a specific strategy for learning a new language. I've used it to learn Mandarin, French, Urdu, and Hindi." First because the opening is rather abrupt, and second because (as you can see) without citations everyone assumes you're working only from anecdotal evidence. If you aren't, you should definitely give your sources. And if you are, you should explicitly make that disclaimer, because otherwise it feels (at least, to me) like you're trying to make a stronger claim than just "hey, here's something that works for me."
I think it's because you're wearing your rationalist shoes.
... You mean these []?
The analogy is that if someone made a Main post saying "hey guys, you should wear Vibrams because they're better for you", that could get downvoted because it's not a suitable topic for Main, even though a disproportionate number of people here wear Vibrams. Part of that is how the post is motivated- if it's cast as an example of "here's how ergonomics can make your life better, and making a one-time investment can pay large dividends," then it might get upvoted whereas it might not if it's a standalone endorsement for shoes.
My suggestion would be to add an introduction. There are many more things to be read than time to read. It's incumbent on you as a writer to convince people that what you have to say is worth the time investment. And you need to make that case clearly, convincingly, and concisely right at the beginning. For this particular article, you need to establish two things: * Why the reader should care about learning a foreign language. You take this as given, but I submit that it's not as obvious as you might think. It sometimes seems like everyone else in the world is trying to learn English - why shouldn't I let them do all the work? * Why the reader should listen to your advice. As far as we know, you're just some random person on the internet. Even if I am interested in learning a foreign language, why should I trust your suggestions? A paragraph or two addressing those two points would go a long way towards convincing your potential readers that your article is worth their time to read.
I didn't downvote or upvote you, but here's my feedback: -I disagree that learning (non-english) languages has a high value/cost ratio. Thus, I'd rather not see the enthusiastic exhortation in the title (though I would be much less opposed to it if I thought that you were exhorting people to do something which actually did have high value/cost ratio.) -I am fine with/appreciate the bolding, italics, and caps -- things that help me grok someone's emotions as they're writing something like this are appreciated, because it lets me get inside their head better. -Citations would probably be appreciated on most things. -This is unlikely to be generalizable to most of your audience, but I really don't like things that require brute mental force -- they don't hold my attention.
I can scarcely imagine myself being fired up enough about something to use that much emphasis, so it seems unlikely your "get super motivated!!!" strategy will be very useful for me.