The magic of words


In Ursula K. Le Guin’s Earthsea series of novels, magic power can be gained over something by discovering it’s true name in the ancient language of the dragons. Young wizards who are training in the magical arts have to spend significant time memorizing many of these ancient words, and it can be hard to keep going. To help them along, their teacher gives a little speech about the ancient language.

In his speech, he explains to them that because their language is related to the ancient language, they can find clues within their words. Some of their words are made of bits of the older words. By investigating the older language, they can learn new things about their own. Also, the more words they learn, the more things they can precisely describe and thus have power over. This motivates them to keep going in their long task.

So, as an illustration of how to learn from parallel texts, I’d like you to take a look at the parallels between the different language versions of some of this speech. I’ll lay out 3 sentences along with their corresponding sentences in the other languages. First is English, then German, then Dutch, and finally Polish to give some contrast. These are from the official translations, not google translate.

  • But magic, true magic, is worked only by those beings who speak the Hardic tongue of Earthsea, or the Old Speech from which it grew.
  • Aber Magie, wahre Magie, wird nur von denen ausgeübt, die das Hardisch der Erdsee sprechen oder die Ursprache, aus der es stammt.
  • En magie, ware magie wordt alleen gewrocht door hen die de Hardische taal van Aardzee spreken, of de Oude Spraak waaruit deze is voortgekomen.
  • Ale magia, prawdziwa magia, jest dziełem tylko tych istot, które mówią hardyckim narzeczem Światomorza albo też Dawną Mową, z której to narzecze wyrosło.

  • That is the language dragons speak, and the language Segoy spoke who made the islands of the world, and the language of our lays and songs, spells, enchantments, and invocations.
  • Das sind die Sprache der Drachen und die Sprache Segoys, der die Inseln dieser Welt schuf, und es ist auch die Sprache unserer Lieder und Epen und unserer Zauber- und Bannsprüche.
  • Dat is de taaldie de draken spreken, en de taal die Segoy sprak toen hij de eilanden van de wereld schiep, en de taal van onzwijzen en zangen, onze spreuken, oproepingen en bezweringen.
  • Dawna Mowa to język, którym mówią smoki, język, którym mówił Segoy, ten, co stworzył wyspy świata, język naszych ballad i pieśni, zaklęć, czarów i wezwań.

  • Its words lie hidden and changed among our Hardic words.
  • Die Worte dieser Sprache sind versteckt in unserem Hardisch.
  • Haar woorden liggen nauwelijks herkenbaar verscholen tussen de woorden van ons Hardisch.
  • Jego słowa spoczywają, ukryte i zmienione, pomiędzy naszymi hardyckimi słowami.

For those of you who already speak some German or Dutch, you’ll notice right away that the translations are not exactly literal. There are some words that have been removed or added. Also, even for those who don’t know any of these languages, you might have noticed that there are some changes in word order.

If I were starting these languages completely from scratch (which I sort of am with Dutch and Polish, although I have a background in germanic languages to help me with Dutch), then the first thing I’d look for is some “anchor” words. Typically these are proper nouns for people and places, and they tend to stay roughly the same between translations. This will help you even with unrelated languages like Chinese, where the foreign names are usually spelled out somewhat, using rarer characters as phonetic approximations.

In this case, the words that are going to transfer across all translations are “Hardic” and “Segoy”. Due to their connections as indo-european languages, you’ll also see Magic / Magie / magie / magia corresponding. And then among the germanic ones you’ll see more words corresponding like dragon / Drachen / draken (which amusingly seems to be “smoki” in Polish, as far as I can tell). I also guess “ballad” as the Polish word for song. If words were chosen differently by the translators, the different versions could be made to correspond even more closely.

The next thing you can try, is finding a passage that repeats itself with only a slight change, and then see what changed. A good candidate for this is “the language dragons speak, and the language Segoy spoke”. Looking at the polish, you’ll see “język, którym mówią smoki” and “język, którym mówił Segoy”. Without looking at any dictionaries, I would guess that mówi- is a stem for “to speak”, with -ą added for present and -ł added for past. I would also guess that język is the word for language, which is somewhat confirmed by looking at the sentence after that too. I could be wrong, but I would be aided by reading along further in the story.

This is basically the first time I’ve ever looked at Polish in this detail, yet I can still find patterns and start working things out. This is only with 3 sentences. By finding slight hints at patterns, and then seeing those patterns represented in hundreds or thousands of different sentences, you can learn a lot of the language without ever looking at a dictionary. It would take ridiculously long to look up every single unknown word in the unknown sentences.

So, this is why I suggest that learning with books can be really productive, even from the very start. It’s made much easier by having a translated version of the text to make it comprehensible, and you should probably limit your dictionary lookups to those words that you’ve already seen many times (the “high frequency” words). By seeing many many somewhat-comprehensible examples, you can learn a lot very quickly.

Another thing that might help make things more comprehensible, is to do a quick browse through a bunch of grammar examples, just to see what’s possible. No need to try and memorize any tables or do any “exercises” from textbooks, because you’ll pick up their workings naturally as you read through your novel. Looking at a bunch of clear examples is helpful though, because it lets you see what’s possible in the language…to see what’s out there for you to discover. This helps you notice it when it comes up for real in your novel.

Just remember that Exposure comes before Knowledge, not the other way around. Don’t wait until you’re “ready” to expose yourself to the language, because then you never will be. Also, go out and buy some real books. As Khatsumoto has said, you have to own before you pwn. If you have no books on your shelf, you will have limited access to the language.

Have fun reading!


10 days of intense Swedish: day 9


In an attempt to bring up my Swedish comprehension level, I’ve been trying a little experiment. I’m trying to see how much time I can put in over a 10 day period, focusing on Listening-Reading. Using English text and Swedish audio, I’m studying at least 5 hours per day, and trying for more. While not for the faint of heart, this method suits me right now because I have a block of spare time, so I can throw all of my effort into this project. So far, after 8 days I’ve done 44.5 hours of studying in this fashion. Today is day 9.

I’ve discovered two difficulties. One is obviously to do with time. If I waste 15 minutes every hour chatting with people on the internet and surfing web pages, then it’s actually quite a significant fraction of my daily time, even though it seems insignificant in the moment. On the day where I tried the hardest, it took me over 15 hours to get 10 hours of productivity, but on most days it was far less than that. That extreme day meant that I only got up to cook/eat/shower, and I didn’t even leave my apartment the entire day. Not very sustainable, but it was interesting.

Incidentally, I also thought it would be easy to get 8 hours of productivity in, because it’d be similar to an office work-day. I found out, however, that in any 8 hour period of a “work-day”, not all of it is productive time. Trying quite hard, I’d be lucky to get 6 hours of productivity over that 8 hour period, which I suspect is actually far more than I ever did at a real job. It’s really hard to stay on-task for that length of time.

The next problem is attention. While raw time is important, its value will vary depending on how much attention you can pay to the various language features you encounter. While Listening-Reading, I noticed that sometimes I would get absorbed in the English text and block out the Swedish audio, and other times I was only half-listening to the Swedish audio, but not trying super hard to pick out every single word and come up with the meanings.

Overall I found this method quite satisfying. I fully understood everything since I was reading in English, but I was also picking up a lot of new Swedish words from the audio. So far I’ve read John Boyne’s The boy in the striped pyjamas (“Pojken i randig pyjamas”), Paulo Coelho’s The Alchemist (“Alkemisten”), Roald Dahl’s The Witches, and most of Stieg Larsson’s The girl with the dragon tattoo (“Män som hatar kvinnor”).

One thing I’d prefer in the future is to use parallel texts, where each sentence or paragraph of English is matched up with the transcript of the target-language audio. This would be vital for starting a language from scratch, I think, but I managed without it this time because of my previous experience with Swedish using other study methods. I haven’t created parallel texts yet for my Swedish books because I’ve had a lot of problems finding PDFs of the Swedish text online, so I’ve had to settle for my regular Swedish paperbacks. I have some tentative plans in the near future to try this method from scratch for other languages, and for those I’ll definitely be using line-by-line parallel texts.

I’m continuing today, after I get a few things done in the real world. I’m enjoying my new apartment here in Berlin, but I still have to buy some vital things like a fridge. It’s fun to ignore the world and study Swedish all day, but responsibilities sometimes intervene.

I should be able to easily reach 50 hours total by the end of my 10-day study period, but I also hope that I can raise that to 100 hours by the end of the month. Hopefully that should put me on pretty firm ground in Swedish, and I’ll be able to continue my studying a little bit more casually by just reading a lot of books whenever I have the chance.

how much input do you need?


I’d like to draw everyone’s attention to this excellent article from Antimoon: “How much input do you need to speak English fluently?”

It first caught my attention because it mentions the little facts that you have to know about a language, but which are not classified as “grammar”. For example:

You can give an opinion, but not an advice; buy a cake, but not a bread; move a table, but not a furniture; share a fact, but not an information.

Then I realized that I really liked his comparison of language learning to facial recognition. When you remember someone’s face, you don’t remember a bunch of specific and logical facts about various features. You don’t remember that the distance between the eyebrows is 0.754 times as big as the length of each eyebrow, or something like that. You just know, because your brain has a special part for recognizing faces without use of the conscious part of your mind. Much of language acquisition is similar. You don’t need a bunch of explicit rules, because as you acquire the language naturally through exposure you’ll just automatically start to develop some magical part of your brain that absorbs the needed rules and principles. This is all done below the level of conscious thought and calculation. Perhaps later you can try to develop your explicit ability to describe such rules, but that’s mostly a boring job for professional linguists with Phds, not for those of us who just want to speak and understand a language.

Finally, there’s the question of how much input you actually need in order to speak and understand a language fluently. The Antimoon article does a nice job of going through the various factors in this, but basically comes up with a nice ballpark number of 1000000 sentences of exposure. The author then considers how much work it would require to be exposed to this much of a language within 3 years, which is what he says it took him for English. Remember that this is for a very high level of proficiency, where he speaks and writes almost perfect English. I personally believe that you can gain a really good level in a language related to your own in less than a year, and that there’s always room for improvement.

His summary makes this enormous task seem quite doable. He breaks down the 1000000 sentences in 3 years into 6400 sentences per week. This then gets broken down to 1600 written sentences per week (~60 pages) + 4500 spoken sentences (~6 hours of audio) per week. This is a very doable amount for almost anyone. Just find whatever content interests you, and try to expose yourself for about an hour a day, and then you can expect to be highly fluent in 3 years according to this estimate. For those of us who are even keener, you may be able to up this amount to multiple hours of exposure per day and thereby rapidly increase your learning rate.

To get these hours of exposure in, just try to fill up your “spare” time (like riding on the bus, or sitting at your desk at work if possible) with audio content in your target language. When you get home, instead of watching TV in your native language, have a selection of target-language books sitting on your coffee table so that you can just pick one up and start reading. Or better yet, have those same books along with an MP3 player containing the audiobook versions so that you can listen and read at the same time. If you really want to watch TV, order some dubbed versions of your favourite shows if they are available. If not, order some original shows in your target language.

If you use whatever method you can to understand more vocabulary, you will soon be enjoying real adult materials in your target language. I couldn’t understand even 1 word of swedish in december 2009, but right now I’m listening to an audiobook in swedish and I’m able to follow along with the story most of the time, and I catch a lot of the vocab. I’ve been slack recently too, so I probably just barely averaged 1 hour per day, if that. I just very efficiently got myself up to speed on basic vocabulary, and put a lot of basic sentence material into Anki flashcards so that I could review them regularly, and did as much listening and reading as I could. Now that I have this basic level, it’ll be easier to ramp up my intake, and hopefully soon I’ll be racking up dozens of hours of content and thousands of sentences worth of exposure, on my way to fluency.

You can do it too, just start reading and listening in any way that you can 🙂

the writing of Claude Piron


Lately I’ve become really impressed by various articles about Esperanto that I’ve read by Claude Piron. Piron was a psychotherapist and taught from 1973 to 1994 in the psychology department at the University of Geneva in Switzerland. He apparently spoke Esperanto since he was a small child, and is a notable author in it.

I first encountered him as I was reading his instructional novella “Gerda malaperis” (“Gerda disappeared“). It’s a book that gradually introduces Esperanto vocabulary, moving from the basics quickly up to an intermediate/advanced level by the end of the book. It seems very well thought out as instructional material. I’m reading it online at, where it’s available with audio and vocabulary lists for each chapter.

After working through some of this instructional book, I somehow stumbled upon Claude Piron’s articles in English. In them, he demonstrates a powerful ability for critical thinking and analysis of arguments. I particularly enjoyed his Psychological reactions to Esperanto.

Also illuminating was his shorter article, The language of power, wherein English is analyzed. He asks if English is actually an “international language”, and argues strongly that it isn’t. Very few people in the world actually speak English, and because of its great irregularity and mishmash of romance and germanic roots, it takes thousands and thousands of hours of work in order to master it. Only about 3% of people in India can speak it, despite the elite there being quite good at it, and even people in France who rate themselves as “quite good” at English were unable to figure out 3 short English paragraphs in one test.

As a consequence, most speakers of English as a second language who don’t already come from a germanic language background, are from a higher economic class where they can afford to go to fancy schools and spend significant time living or working in an english-speaking country. Not many people have the opportunity to spend 2000 – 10000 hours learning English, and those of us fortunate enough to be born in an English-speaking country have received a free-ride in that department. In the words of one Korean that Piron quotes, he could have achieved several PhDs in the time it took him to learn fluent English. In contrast, Esperanto can be learned by most people in around 150 – 200 hours, so on a scale of months rather than years or decades.

Beyond the realm of language politics, Piron had some interesting articles about the evolution of Esperanto itself. Quite an interesting read from a linguistics perspective, and even more interesting for me as a learner of Esperanto.

In many of his articles, such as Linguistic Communication – A comparative field study, Piron stresses the hierarchy of power that develops in circumstances where some people are native speakers of a language, and have to communicate with those that aren’t. Inherent in this situation is the fact that those native speakers will always be authoritative, and the others will be in an inferior position. This could be remedied somewhat if everyone opted to speak a language other than their native language, to level the playing field, but of course in the realm of power relations this is rarely an option. In such situations, it makes plenty of sense to take as a working language one that takes an order of magnitude less time to acquire, one in which everyone is on equal footing.

It seems clear to me now that this is a role that Esperanto could and should play. Not as a “replacement” for any other languages, but as a tool of international communication that levels the current language hierarchies. Everyone can and should speak their own language or dialect in their “home” situations where everyone else around them can also do so, but in those situations that require communication with outsiders that don’t speak that language, Esperanto is the logical and efficient tool for the job. Whether it is up to the task is an empirical question rather than philosophical, and I think this has been proven by the 100+ year tradition that it has enjoyed as a language for poetry, novels, theatre, children’s play, and international communication.