Hi everyone,

I recently bought Rosetta Stone in Japanese for my son, and had a few issues with the company in terms of bugs and poor service, so I decided to write a freeware competitor. (See the Rosetta site if you are not familiar with the idea. Basically, it is a picture-and-audio program that attempts to teach language monolingually and immersively... http://www.therosettastone.com.au/ )

At the moment, my program is similar to older versions of Rosetta stone - you hear a phrase in the language you are learning and click the matching picture - but I plan to expand it in future. It currently lacks content, but I hope to generate content through an online community effort.

Unlike Rosetta Stone, the program comes with a lesson editor - which was by far the most complex bit of the program to write. I intend to get native English speakers to contribute English lessons, native Japanese speakers to contribute Japanese lessons, and so on, to create a huge shared resource that is absolutely free. Once we have a few lessons, the payoff for contributors will be excellent: contribute one lesson, and get a lifetime supply of free lessons in any language.

I've made a brief Youtube video to illustrate the idea for non-English speakers:

I have an account at Lang-8 and a couple of Japanese speakers have already expressed interest in contributing. Obviously, the more English lessons available, the more attractive it will be for Japanese contributors.

One of the main hurdles right now is translating the interface into Japanese. Here is my first attempt at the main menu items. The arrows ( => ) indicate changes suggested by Lang-8 users but I don't have enough Japanese and they don't have enough English for me to clarify the corrections:


Japanese Interface 日本語インターフェイス
English Interface 英語インターフェイス
Lessons レッスン
Options オプション
Import Lesson インポートレッスン
Download Lesson レッスンをダウンロード
Arrange Lessons レッスンの手配 => レッスンの予約
Library View ライブラリビュー
Review Weakest 最弱を確認 => 弱点を確認
Users ユーザー
Add User ユーザーの追加
Remove User ユーザーの削除
Study Language (as in, Language of Study) 調査の言葉 => 言語を調査
English 英語
Japanese 日本語
Editor エディタ
Lesson Editor レッスンエディタ => 文章を書く機能
New Lesson 新しいレッスン
Find Lesson レッスンを探す
Edit Current Lesson 現在のレッスンを編集


'Review weakest' is a shortened form of 'Re-do the lesson that has the worst score.'

Cheers,

Torrential.

While all of those translations are technically correct, the heavy use of loan words makes it seem kind of unnatural. Many computer/technology terms are written like this, but from my experiences with Japanese software, kanji and hiragana menus usually dominate. Although I have indeed seen programs that use 「オプション」, most tend to use 設定 i my experiences. For example 「ライブラリビュー」　would probably be ライブラリを見る or something in a native program. But hey, like I said, those translations are all correct, but I think it might help to get them to sound a bit more "authentic."　

Thanks,

That's just the sort of feedback I am after. To be honest, I am surprised they are even "technically correct", as I used Google Translate.

What do you think of the suggested changes marked with => ?

For instance, I am unsure what "文章を書く機能" actually means. The lesson editor is the part of the program that lets users compose lessons by choosing pictures, writing captions, selecting click zones and so on.

Cheers,

Torrential.

dude I'm sorry but this program already exists and it's called anki. using the core6 deck you can learn thousands of words/phrases with native audio and pictures. it's freeware and has an online database of that and thousands of other kinds of decks contributed by native speakers in almost every language.

and it also has the thing rosetta stone and your program lack, which make the relatively ineffective... a spaced repetition system to time flashcards to pop-up when you're likely to forget them.

ankisrs.net

sorry to be a downer

it also has an interface in english and japanese... as well as at least 30 other languages.

dtcamero Wrote:dude I'm sorry but this program already exists and it's called anki. using the core6 deck you can learn thousands of words/phrases with native audio and pictures.

Yeah, Core6k is the only way to use Anki (rolls eyes)

Also, Core6k is not the same as what the guy above is talking about, though Rosetta Stone's approach is a little bit inefficient anyway.

[Zgarbas]

The company I worked for translated all its games into Japanese and I was appalled at the lack of actual Japanese in it. Everything was just the English word transcribed into katakana. The games sell pretty well in Japan, so I assumed that the people somehow don't mind the fact that unless you know English they're unintelligible.

Rosetta Stone looks nice and makes you feel good. No need to hate on the approach just because there's better ideas out there. The price, though... yeeuck.

dtcamero Wrote:dude I'm sorry but this program already exists and it's called anki. using the core6 deck you can learn thousands of words/phrases with native audio and pictures.

Anki is great, don't get me wrong. But it's possible to make learning software that can do things that Anki simply can't. Look at brainrush, it tries to gamify the learning experience a bit. It's not perfect either, but it's an example of things you can do differently. Also Torrential mentioned Rosetta Stone for his son, Anki quite frankly is not kid friendly.

Now rosetta stone methods aren't perfect. I think there are much better ways you could go about it, but it also isn't the worst.

I think the main trick would be coming up with proper lesson plans for people to use. Obviously rhinospike is perfect for getting audio samples. However, I'm no linguist and am always experimenting with how I learn.

Hi there,

I'm familiar with Anki. As far as I know, it misses the mark for what I am proposing. For instance, it relies on users self-marking and then self-rating. Personally, I think all that meta-cognition is a distraction. My program would involve listening in Japanese, then clicking the relevant picture or part of a picture, and immediately listening to more Japanese. The program would know when you were wrong, and you would not have to stop to rate yourself. The overall feel would not be trying to recall stuff, which is the atmosphere Anki creates, but solving a problem while in a monolingual framework.

For instance, the audio prompt could be giving directions from my house to a friend's house. The user has to listen to the directions and click on the right house in a town map. I can't see Anki doing that.

I'm also familiar with spaced repetition, and plan to add it. There are more sophisticated approaches than raw SRS, though. Instead of repeating an item exactly the same way for each SRS cycle, which is the Anki approach, I feel it would embed the item better to move on to a more complex item in which the knowledge to be reviewed is an assumed component. That is, the program could create learning dependencies between items and use an SRS model in the background to guide the process.

I had planned to use Rhinospike to get the process started. If there are already databases of monolingal picture-audio combinations, then these would be ideal for immediate importing into my program.

The other potential benefit I saw was that intermediate students of, say, Japanese, could write lessons for beginners, and advanced students could write lessons for intermediate students. Native speakers could check the lessons and provide audio (or Rhinospike could provide audio). This could actually be the part that lifted language skills the most. I know a couple of Japanese teachers. They could have a class of students write questions (like the map exercise mentioned above) while a sister-class of Japanese kids writes questions in English, and then the two classes could provide the audio for each other. The result is a valuable learning experience for all of them, and a set of demo lessons for next year's class.

Anki is of value, but it is not the only possible approach, and the different programs could be used together. (The simplest way to combine them would be to export all failed items to Anki.) If the lessons are designed well, though, there would not be many failures. The repetition of Rosetta-style programs is organic and the complexity increases incrementally, so simpler items do not really need review.

I have always thought a weakness of Anki was the fact that the program does not realise that 'Turn right at the Post Office' is actually a review of 'The Post Office,' and reviewing the more complex item negates the need to review the simpler item. Maybe Anki has fixed this since I last looked?

Cheers,

Torrential.

BTW, I had a quick look at the brainrush site. For learning animals in Spanish, in one of their demo lessons, I found the format quite reasonable.

Torrential Wrote:I'm familiar with Anki. As far as I know, it misses the mark for what I am proposing. For instance, it relies on users self-marking and then self-rating. Personally, I think all that meta-cognition is a distraction.

I'm not an Anki diehard fan or anything, but what evidence do you have to support this? For example, how many languages do you already speak well? (And I mean more or less fluently, not fake-polyglot-Benny達 style). You are not building anything new or more useful than what's already available. On the other hand though, you could gather a lot of naive customers that ignore all this…

Torrential Wrote:I have always thought a weakness of Anki was the fact that the program does not realise that 'Turn right at the Post Office' is actually a review of 'The Post Office,' and reviewing the more complex item negates the need to review the simpler item. Maybe Anki has fixed this since I last looked?

This is an interesting idea, but it shouldn't be the default functionality. Write a plugin.

Torrential Wrote:I'm also familiar with spaced repetition, and plan to add it. There are more sophisticated approaches than raw SRS, though. Instead of repeating an item exactly the same way for each SRS cycle, which is the Anki approach, I feel it would embed the item better to move on to a more complex item in which the knowledge to be reviewed is an assumed component. That is, the program could create learning dependencies between items and use an SRS model in the background to guide the process.

Another interesting idea, write a plugin.

I've downloaded the latest version of Anki, and tried out the deck at this link:

https://ankiweb.net/shared/info/4221416043

The images are not available at the provided link... Any pointers as to where I could find them?

Also, is there a more suitable deck for beginners, particularly one designed for monolingual learning?

delta Wrote:

Torrential Wrote:I have always thought a weakness of Anki was the fact that the program does not realise that 'Turn right at the Post Office' is actually a review of 'The Post Office,' and reviewing the more complex item negates the need to review the simpler item. Maybe Anki has fixed this since I last looked?

This is an interesting idea, but it shouldn't be the default functionality. Write a plugin.

This sounds true, but I'm wondering if it is actually true.

Before you write a new SRS-type system, I would go and do some research to make sure your ideas are backed by science. Just because you can click on animals, it doesn't mean you understand Spanish (although it might), or it's even an efficient way to study (although it might be).

Couldn't some of you guys said things more properly?

Couldn't a "Make sure that your software isn't already out there. There's this and that..." suffice?

If some people like Rosetta Stone approach, then a free software can be a gem.

Keep up the good work, Torrential!

partner55083777 Wrote:Just because you can click on animals, it doesn't mean you understand Spanish (although it might), or it's even an efficient way to study (although it might be).

This cracked me up.

[RawToast]

undead_saif Wrote:If some people like Rosetta Stone approach, then a free software can be a gem.

Keep up the good work, Torrential!

+1

Personally I didn't like Rosetta Stone, but many people have used it and benefitted. These people would love a free and community driven alternative. Anki isn't the 'be-all and end-all' software for language learning and alternatives should be welcomed (like the Kanji learning on this very site!)

I personally would like to see one person who has actually used Rosetta Stone as a primary study tool and learned a language. I think it's a sham that makes you feel like you're learning something, and while giving you the endorphins of playing a video game, takes the user only nominally (if that) in a positive direction. You feel great, enjoying yourself pushing the llama button over and over again, but it's the interface that creates that feeling, not the learning.