Wednesday, October 28, 2009

Quantifying language learning, diminishing returns, and wanderlust

There's an interesting post over on Language Fixation talking about just how much is involved in learning a language, which was inspired by an interesting post on GlobalMaverick talking about the mess between being a beginner and being fluent. There's three points I'd like to focus on from these posts: quantifying language learning, diminishing returns in language learning, and wanderlust (to new languages).

Read more... Let's start with quantifying language learning. From GlobalMaverick:
I don’t think it’s possible to say with any sort of accuracy that you’re at 73.6% of a native speaker’s abilities…
And Language Fixation concurs:
[T]he distance between “beginner” and “native” is huge and unknown…
I disagree with these statements, but I would be able to agree with them if they added "with the currently existing language-learning tools". A language is a limited body of knowledge, consisting of discrete units: vocab, grammatical rules, intonation patterns, etc. Once you've got a way to count these units (and all of them certainly can be counted), then it's simply a matter of figuring out what the average native speaker knows to be able to determine whether you're at 73.6% of a native speaker's abilities. An SRS system with each of these items broken down into a discrete units could do just that. I'd be surprised if a tool like this doesn't pop up soon enough, but until then, GlobalMaverick and Language Fixation are right; there's not a real accurate way to get the whole picture of where you are in a language.

Now let's get to diminishing returns. From Language Fixation:
For me there’s also a tremendous thrill at the beginning because everything is so new, and because every time you sit down to study, you’re actually gaining a rather large percentage increase in your knowledge. After day 2, you know TWICE AS MUCH as after day 1.
This gets to a great point of debate in language learning. As you progress farther and farther into the language, finding those discrete units that you don't already know takes more and more time. Thus, going from 90% to 99% proficiency will take a heckuva lot longer than going from 0% to 10%, or probably even 0% to 50%. The point of debate this brings up is whether it's better to get a bunch of languages up to 90% proficiency, or one or two up to 99% proficiency.

And that gets us to linguistic wanderlust, i.e., the desire to work on getting a new language up to 90% before the previous one is up to 99%. As Language Fixation describes, the speed at which you can do this always make it attractive to the serial language learner.

I've had to restrain myself from indulging my own wanderlust (despite frequent flirtations with Russian, Korean, and Arabic) and focus on the seven languages in which I've already got some degree of proficiency in, and all of which need a lot more work, whether on the way to 90% or 99%. I suppose this puts me in the aim-for-99% camp, but we'll see if I can maintain my self-control (I'm pretty sure the next time I see a good Korean grammar book in Japanese, I'm going to buy it).

Link: the size of a language [Language Fixation]
The messy in-between [GlobalMaverick]

Wednesday, October 14, 2009

Japan's foreign-language teaching industry valued at a cool $8.7 billion

There's an interesting article in The New York Times saying how Japan is gaga over learning English from Obama's speeches. They apparently even go gaga when they can't really get what he's saying; he's just that moving.

While that's fun and all, the most interesting part of the article was this little tidbit:
The publishers [of learning resources using Obama's speeches] are trying to tap into a foreign-language teaching industry [in Japan] that the Yano Search Institute said was valued at ¥767 billion, or $8.7 billion, in 2008. The figure includes the cost of books, CDs, dictionaries, e-learning programs, standardized English tests, and the cost of private language lessons. The institute, in Tokyo, says the majority of the spending is aimed at learning English.
Just to see it with all the zeroes, that's $8,700,000,000. To put that number into perspective, that's more than the GDP of 46 countries as listed in the CIA World Factbook. And that's in 2008—hardly a stellar year for the world economy.

So if you're in the language-learning industry and Japan's not a major focus for you, it's probably time to make a Japan plan.

Labels: , , ,

Monday, October 12, 2009

75-year-old Iranian guy crowdsourcing a Farsi-English translation in Geneva airport

Here's an interesting anecdote about a 75-year-old Iranian guy crowsourcing a Farsi-English translation in the Geneva airport. Sounds like he's doing a pretty good job of getting it done that way, but perhaps someone should tell him about how to get his foreign-language writing corrected online for free.

Labels: , ,

Saturday, October 10, 2009

The best free online Portuguese-English dictionaries

Continuing my series on free online dictionaries (previously covering Japanese, Chinese, and Spanish), after the jump you'll find my favorite, free, online Portuguese-English dictionaries.

Read more...
  1. WordReference.com: Like with Spanish, WordReference.com comes in first on my list of Portuguese-English dictionaries. However, its Portuguese dictionary is not as good as its Spanish dictionary primarily because there seems to be a big gap between the two in terms of word count. For Portuguese too there are forums available in which you can get answers to specific questions, but the forums are much less developed than in Spanish. You should still be able to post any question about how to say something in Portuguese that you might care to, but the body of already-answered questions is much smaller than in Spanish. What's more, the Portuguese dictionary seems to be very Portugal-centric, so for those of us that consider Brazil our home in the Portuguese-speaking world, that's a big minus.

  2. Reverso: Reverso's another solid choice for a dictionary, but Wordreference.com seems to have it beat in terms of breadth, and certainly in terms of the forums. The one thing I do like about Reverso is that, when there are multiple Portuguese definitions of the English term, it gives you a parenthetical that helps you figure out which is the appropriate term (take the entry for "stock", for instance). It's sometimes clearer than Wordreference.com because of this.

  3. LookWAYup: Look way up doesn't have the word coverage that the above two do, and its definitions are pretty basic, but on occasion I've found it has filled a blank the other two couldn't.
Just as with Spanish, I'll also note Ultralingua here. It's a pretty solid dictionary, comparable to Reverso, but they limit you in the number of searches you can do per day, hoping you'll pay to have no such limit, so it's not really free. That said, it's worth checking when you're struggling to find a definition in one of the dictionaries above.

All in all, the currently available dictionaries for Portuguese aren't quite as good as those for Japanese, Chinese, or Spanish. The most fundamental problem is their limited word coverage, but Portuguese dictionaries also don't have as many example sentences or options to hear the pronunciation of the words online. However, with Brazil on its way up as one of the BRIC countries, I'm hopeful that this situation will improve over time.

As always, let's hear what you've got to say in the comments below—especially if you can point me to an even better dictionary that I missed!

Labels: , , ,

Learning by Mixing target language words into native language texts?

I'm pretty sure this is not a great idea:
Waikato University PhD student Michael Walmsley is working on a project which will help language learners build their foreign vocabulary by reading texts online where some of the words have been replaced with words in their target language.
Read more... My big problem with this is that it'll keep learners away from actual exposure to the target language; instead of reading a pidgin text, why not just go to the native language? If the learner refuses to read native language texts, I think that learner's got a bigger problem than can be solved by throwing in some target language words to a native-language text.

Walmsley relates this to his own experience with Japanese:
I can always find time to read the newspaper to keep up with current affairs, but I can't do that in Japanese. It's just too time consuming. This way, busy people can squeeze in language study by combining it with their general reading.
I read that as saying that his priority is reading the news rather than learning his target language. I'd recommend he find short newspaper articles (such as those found here in Yahoo's headline news in Japanese) and read fewer articles while taking the time to understand them. If he's serious about learning Japanese, I'd imagine that he can get by without the crutch he's trying to create.

Link: New tool ahead for language learning [Waikato Times]

Labels: ,

Thursday, October 8, 2009

The best free online Spanish-English dictionaries

My series on free online dictionaries (previously covering Japanese and Chinese) continues. Today I've got my favorite, free, online Spanish-English dictionaries for you, after the jump.

Read more...
  1. WordReference.com: WordReference is my first choice in Spanish, and it's often the only choice I ever need. Not only does it provide you free access to a few different traditional (read: dead-tree) dictionaries whenever you look up a word, but it also has its own content.

    If none of those get you the definition, then it's time to turn to WordReference.com's forums. With hundreds of thousands of entries in the forums, I've almost always found that whatever I'm looking for that's not already in the dictionary is already in the forums. And, when that fails, posting something in the forum will get you a quick response. For example, I put up this post when I was trying to figure out how to say "I'll spare you the details" in Spanish (f.y.i., it's "Te ahorraré los detalles"). I got the answers I needed in a minute. That's right; one, single minute.

  2. Reverso: Reverso doesn't have quite the depth of WordReference.com's content, but it is a solid dictionary and once in a while I manage to find a phrase there that I couldn't find in WordReference.com. Given the breadth of WordReference.com and its powerful forums, I rarely use this, but it does provide a good back-up.

  3. SpanishDict.com: SpanishDict.com takes third place here because it doesn't seem to have quite the breadth of the first two. It does have concise definitions, so it might be a good thing to use if you're just starting out and don't need the level of detail found in WordReference.com. Like WordReference.com, it also has forums used in a similar manner, but they're no where near as developed as WordReference.com's.
There are also two other online dictionaries that bear mentioning but not inclusion in the above list: Diccionarios.com and Ultralingua. Because you can only perform a handful of searches in these dictionaries per day (they limit you in the hope that you'll buy an unlimited version), they're not really completely free, but they can serve as a good fall-back when none of the other dictionaries are getting what you need. If you use them only in that way, you're pretty unlikely to ever hit their daily-use limit.

You might note that I didn't mention example sentences or the ability to listen to words being pronounced at all above. That's because, unlike the selections in Japanese and Chinese, there are no dictionaries that I'm aware of that implement these features in Spanish. Although you can find pretty much any word you like in the above dictionaries, you're often on your own to find a way to hear the pronunciations (here's one way to do that) and to find example sentences. There are of course some example sentences in the dictionaries above, but there's definitely no systematic sentence collection from net sources like that found in the Japanese-English dictionary ALC.

As always, let's hear what you've got to say in the comments below—especially if you can point me to an even better dictionary that I missed!

Labels: , , ,

Wednesday, October 7, 2009

Get your foreign-language audio recordings corrected online for free

I've already given you the low-down on how to get your foreign-language writings corrected online for free. Now I'd like to turn to how to get your audio recordings corrected for free.

Unfortunately, your options here are still pretty limited. As far as I can tell, there are only two places where you can submit recordings and get them corrected by native speakers, neither of which are close to making the feature ideal: Livemocha and Lang-8.

Read more... Lang-8
  • Overview. Lang-8, based in Tokyo, is a two-person project by Yangyang Xi, CEO, and Kazuki Matsumoto, CTO, that focuses letting language learners get their texts corrected. However, with this little tip, which Lang-8 supports by making adding audio easy, you can get your audio recordings corrected as well.

  • Content. Lang-8 is set up as a journal or a blog, but you're free to post whatever you feel like posting.

  • Making corrections. Correctors can leave comments for you, explaining what you did wrong. There's no feature for them to record a message for you directly. Although they could leave a recording in the comments in the same way it can be posted in the entry's body, no one has done so yet for me.

  • Speed of corrections. Just as with text, the corrections come very rapidly. Waiting a day for corrections would be a long time to wait.

  • Correction presentation. It is up to individual correctors to apply formats: bold, strike-thru, red, and blue text. Your results will vary.

  • Languages. You can post in any language you want, and native speakers of all major languages are well represented on the site. I'd wager that it'd take longer to get corrections for less frequently studied languages, but I've not tested that hypothesis.

  • Interface. Lang-8's interface is alright; it's nothing to rave about, but it gets the job done.

  • Bottom line. I love that I can record whatever I feel like recording to Lang-8, but I don't like it takes a bunch of steps to post audio recordings and that there's no easy way to post audio recordings in the comments.
Livemocha
  • Overview. Livemocha's main product is it's Rosetta Stone-like language-learning courses, but the coolest thing it does is connect you with tons of native speakers, including through corrections of your audio recordings (see my complete review of Livemocha here).

  • Content. For audio recordings, you're supposed to read outloud a text related to your lesson; there's no discretion involved in what you're supposed to record. Learners can and sometimes do add their own audio at the beginning or the end of the recordings, but they generally follow the script. Of course, you don't have to follow the script and you can surely find flexible human users who'll correct your audio recording for you regardless of what it contains.

  • Making corrections. Correctors can easily record their own recordings in reply to your audio recording, which is the major benefit of submitting audio recordings for correction on Livemocha. Correctors also get a comment field in which they can make comments and variously format the comment text.

  • Speed of corrections. Livemocha has a very large user base, so corrections come back very quickly, certainly comparable with Lang-8.

  • Correction presentation. If there's an audio recording attached to a comment, it's readily available for you at the click of a button. Like Lang-8, it is up to individual correctors to format their textual comments. Again, your results will vary.

  • Languages. Arabic, Bulgarian, Chinese, Czech, Dutch, English, Estonian, Farsi, Finnish, French, German, Greek, Hindi, Hungarian, Icelandic, Italian, Japanese, Korean, Polish, Portuguese (Brazil), Portuguese (Portugal), Romanian, Russian, Spanish, Turkish, Ukrainian, Urdu.

  • Interface. As far as getting audio recordings corrected goes, I've got no major complaints. The interface allows you to get the job done.

  • Bottom line. While I love that correctors can easily supply their own recordings in response to yours, I don't like that you're nominally limited to Livemocha's specified scripts.
Despite the inability of my correctors to easily supply audio recordings in their comments, I've tended to use Lang-8 more for getting my audio recordings corrected, largely due to its content flexibility. Nevertheless, there is a lot of room for improvement—whether on one of these sites or on the site of a new provider of this feature.

Labels: , , ,

Tuesday, October 6, 2009

The best free online Chinese-English dictionaries

Continuing my series on free online dictionaries (for Japanese, see here), today I present you with my favorite, free, online Chinese-English dictionaries, after the jump.

Read more...
  1. nciku: Nciku is my first stop when looking up a Chinese word. I've found that they simply cover more of the words I need than the other dictionaries on this list. They often supply numerous example sentences, which also puts them ahead of most of the other dictionaries here.

    They supply the pinyin for the words you look up, and pinyin is supplied for example sentences via a pop-up when you hold your mouse over them. The latter is kind of a pain because you can't copy and paste it and you can't see it without doing something, but you can always copy and paste the Chinese word itself into the search field to get copyable pinyin.

    Example sentences can be pronounced via a pretty natural text-to-speeh program which goes beyond the simple character-by-character pronunciation of MDBG, but there's no pronunciation button available for the definitions.

    Another very convenient feature is that you can write characters by hand to look them up—a great feature when you can't recall the reading.

    If you're curious about what "nciku" means, see here and here.

  2. Dict.cn: My second stop is usually Dict.cn. More aimed at the Chinese learner of English, Dict.cn's strong point is their example sentences sourced from the net, similar to ALC for Japanese. There is no pinyin and no pronunciation of Chinese words.

  3. MDBG: If you dig around the internet for Chinese-English dictionaries, you'll surely run across some of the numerous online dictionaries based on the CEDICT project. But the site serving as the caretaker of the CEDICT project and one of my favorites Chinese-English dictionaries is MDBG.

    MDBG has a large breadth of vocabulary and always provides the pinyin. MDBG can even color code characters and pinyin syllables by tone. Audio pronunciations are provided, but they pronounce each character as if it were alone, i.e., they don't take into account systematic tone changes based on the previous tone. Like nciku, you can also write characters in with your mouse to look them up.

  4. iCIBA: A lot like Dict.cn: aimed at Chinese learners of English, strong point is example sentences (which also seem to be sourced from around the net), no pinyin, and no pronunciation of Chinese words. They tend to have more, but simpler, example sentences than Dict.cn, and also will serve up related encyclopedia entries.

  5. Reverso: Reverso has pinyin but no pronunciations. The word coverage generally seems to be less than the others above, but they are an additional option that I turn to occasionally.
Know of a dictionary that belongs in this list? Drop a line in the comments below!

This post was updated on October 6, 2009, to clarify the name of nciku and to note that nciku does in fact have pinyin (thanks to doviende of Language Fixation for the clarifications).

Labels: , , , ,

Google's getting into the language-learning game

Google Ventures, Google's venture capital arm, has invested an "undisclosed amount" of its $100 million in EnglishCentral, Inc., an English-language learning website where learners can watch popular videos (such as a clip from Forest Gump or a Red Bull ad) and then get graded on how well they pronounce the words spoken in the videos via EnglishCentral's "unique speech recognition platform".

This investment represents nothing more than Google dipping its toe in the water of the language-learning world. Let them get in up to their ankle or knees, and we'll all think back to the quaint days when we thought Rosetta Stone was a big player in the language-learning world.

Links:
EnglishCentral
Google Ventures, Atlas back language startup EnglishCentral [Mass High Tech]
Google Ventures Invests In English Language Learning Startup EnglishCentral [paidContent.org]

Labels: , , , , , ,

Exposure is a great way to learn languages... for computers too, it seems

Ben Taskar at Penn State has a computer learning language by watching TV, listening to audio, and reading texts.

Sounds like a better language-learning program than most people get exposed to.

Link: Machine Learning by Watching and Listening [PhysOrg.com]

Labels:

Is the FTC about to give Rocket Languages a call?

As I've pointed out before, Rocket Languages has been conducting a pretty blatant (yet successful) astroturfing campaign. I thought that they might have been getting a bit worried when the New York attorney general started looking into the practice in other industries.

But now they'd better be worried; the U.S. Federal Trade Commission will start cracking down on astroturfing as of December 1 of this year.

Read more... From the Associated Press (via Yahoo News):
The FTC will require that writers on the Web clearly disclose any freebies or payments they get from companies for reviewing their products.
From the Financial Times:
The new focus on transparency in online advertising comes after several companies have been exposed for “astroturfing,” or creating fake grassroots campaigns, and as advertisers race to spread word of their products through new services such as Facebook and Twitter.
Astroturfing? Hello, Rocket Languages!

The Financial Times continues:
The main target of the new guidelines appears to be the widespread practice of viral marketing in which companies recruit non-employees to talk up products in exchange for samples or promotions.
Rocket Languages' M.O. is not enticing website owners through samples and promotions but rather through commissions on the sales by customers directed to Rocket Languages' products from the third-party sites.

However, it doesn't look like that difference is going to put Rocket Languages in the clear. Here's the relevant part from the actual regulation:
When there exists a connection between the endorser and the seller of the advertised product that might materially affect the weight or credibility of the endorsement (i.e., the connection is not reasonably expected by the audience), such connection must be fully disclosed.
So... ya think getting a commission for each click-thru sale is a "material connection" that "might materially affect the weight or credibility of the endorsement"? By, say, making these "associates" write overly positive reviews to increase the number of sales and, hence, the commissions they get? Hmm... ya think?

Watch out, Rocket Languages. It looks like the FTC might have you in their sights.

Links:
Guides Concerning the Use of Endorsements and Testimonials in Advertising [Code of Federal Regulations]
FTC Publishes Final Guides Governing Endorsements, Testimonials [Federal Trade Commission]
Celebrities face endorsement crackdown [Financial Times]
FTC: Bloggers, testimonials need better disclosure [Yahoo News]

Labels:

The danger-zone skill level in language learning

When you're learning a language, there's a level of skill that I like to refer to as "the danger zone".

To show you where it comes into play, I would say that your ability in a language roughly progresses as per the following graph.

You start with the obvious beginner, intermediate, and advanced levels. After that, you get to proficient, where you can use the language to get done whatever it is you need to get done, but you're still obviously not fluent. At this level, and those below it, your lack of fluency lets you make mistakes that would otherwise be considered rude without you being considered rude; fluent speakers will generally assume you just don't know enough of the language to speak with sufficient social grace.

Read more... To skip to the top of the skill chart, you've got the native level—which very few language learners ever obtain—and fluency. At the level of fluency, they may still be able to tell that you're not a native speaker, but you're language skills are so good that they assume that if you say something rude, you're actually being rude—as opposed to it just being a linguistic screw-up. And, since you're fluent, they're probably right.

Between proficiency and fluency, you've got what I've cornily designated "the danger zone". At this level, you can easily communicate just about whatever you want to communicate, and some fluent speakers will start to think that you are fluent as well. However, in reality, you're not, and some of the mistakes you're making—which may be considered to be rude—are merely just you mangling the language.

This is just where I find myself in certain areas of Japanese. In casual conversation, I might have punched through to fluency. However, in the formal business Japanese that I constantly need to use at work, I find myself smack in the danger zone; I'm probably just about good enough to fool some fluent speakers into thinking that I'm also a fluent speaker in this regard, but in fact I'm still learning to smoothly use the various polite forms that are necessary in these kinds of communications. I probably get some leeway simply because I'm clearly not Japanese, but I definitely need to be careful because once in a while I screw something up that would obviously be considered rude if said by a native speaker.

So that's the diagnosis, but what's the cure? Well, it's really no different than learning anything else in a language: exposure, exposure, exposure. I'm constantly hearing this language around me, so I'm pretty confident that I'll get it sooner or later, but for now I'll just have to struggle through and try not to offend anyone too horribly.