Back

学 | Manabi: iPhone flashcards app for power-learners of Japanese (in beta!)

#1
[Image: iHLmg6m.png]

For the past 3 years, I've been building an iPhone app for studying Japanese via flashcards. Today, (with Fabrice's gracious permission) I'd like to share it with the Kanji Koohii community! I'm opening it up as a private beta before the public launch so that I can start taking your feedback into account while I iron out the remaining details. This forum has always been one of my top resources over the years, so I'm excited to have something of my own to give back.  Smile

Why build a new flashcards app?
Valid question! Big Grin I built this out of my own frustrations with existing software out there. Besides what I hope to be an overall superior user experience, it comes with extra Japanese-specific features.
  • Flashcards are best for retaining, not necessarily for learning new material. Despite that, it's a popular use-case; so, Manabi lets you scroll down on a card to learn more about it: it'll automatically pull in example sentences from Twitter and Weblio, and let you easily look up the word in a real dictionary (not just EDICT, as wonderful as it is). This helps you build new associations to native materials if you decide to use Manabi to learn new expressions that you didn't encounter "in the wild". (It's also just fun to use Twitter for studying!)
  • Features a novel method of grading cards that optimizes for the common case, while still letting you mark them as "Too Hard" to "Too Easy" if you need to.
  • The flashcard editor understands kanji, so it automatically puts furigana above them without making you use a special notation.
  • Kanji-writing cards automatically include stroke animations.
Overall, Manabi provides a "fuss-free" Japanese flashcards experience catering to intermediate/advanced-level learners. That means no need to micro-manage custom deck configurations; everyone's decks, including the shared deck library, have the same great features built-in. And since Manabi's spaced repetition algorithm is smarter about prioritizing the cards with the highest impact to review, that means less panic over falling behind, less of the dread and guilt that come with an accumulated backlog of reviews.

Of course, Manabi has the standard features we've come to expect from flashcards in the year of 2016: a refined "SM-2" spaced-repetition algorithm; decks can be shared with other users (and any non-destructive changes are automatically synced to subscribers of a deck); self-graded rather than multiple choice, to allow for active recall; and so on. (There's no offline support yet, but I'm working on it—I just wanted to get this into people's hands first for those who have connectivity most of the time, but I totally understand the need for studying on the subway, on vacation, etc. so I can assure you it's coming.)

So, let's get started!
Sign-up for the private beta (or just send me a PM here). I've already launched the beta at the start of the year, and I'm improving it on a regular basis now that it's live. To be up-front on pricing: Manabi is free while in beta, and it will be a commercial product once it launches. I'm not looking to quit my day job with Manabi—pricing is not yet finalized, but will be used to fund its continued development, and will include a discount for students. As thanks for early feedback from the community, I'll extend free service to beta testers for a generous period after launch.

Screenshots
[Image: LYf4ipD.png][Image: RHoaWns.png][Image: PqvYpnQ.png][Image: mg85UHA.png][Image: QxOdHui.png][Image: Xh5gVgz.png]

http://beta.manabi.io/
Edited: 2017-08-08, 7:44 am
Reply
#2
[Reserving this reply for future use.]
Reply
#3
Looks interesting. Signed up! Examples from Twitter sounds fun, looking forward to see it in action. Will I be able to import from Anki? CSV?

To be honest, it would take a lot for me to switch from Anki to another app, but it's always interesting to try other options.
Reply
Thanksgiving Sale: 30% OFF Basic, Premium & Premium PLUS Subscriptions! (Nov 13 - 22)
JapanesePod101
#4
Will there be an Android version?
Reply
#5
(2016-10-24, 7:59 am)sremvik Wrote: Looks interesting. Signed up! Examples from Twitter sounds fun, looking forward to see it in action. Will I be able to import from Anki? CSV?

To be honest, it would take a lot for me to switch from Anki to another app, but it's always interesting to try other options.

Thanks! No automatic import tool for Anki, since Anki decks support so many different configurations. It would be nice to build a CSV importer though, hmm. (I do have a CSV exporter on the way, for the record.)

If you have a deck that you authored, I can help you get it into Manabi! Please send me a DM.
Reply
#6
(2016-10-24, 9:36 am)Wurstmann Wrote: Will there be an Android version?

Sorry, iPhone exclusive for now. I do plan on a companion iPad app, but no solid plans yet for Android.
Edited: 2016-10-25, 9:05 pm
Reply
#7
wahnfrieden Wrote:Features a novel method of grading cards(...)

Interesting, are you adjusting review intervals based on some "difficulty" score associated with each character? ie. certain characters are more difficult to remember than others, on average, across all flashcard users.
Reply
#8
(2016-11-01, 10:32 am)ファブリス Wrote:
wahnfrieden Wrote:Features a novel method of grading cards(...)

Interesting, are you adjusting review intervals based on some "difficulty" score associated with each character? ie. certain characters are more difficult to remember than others, on average, across all flashcard users.

I should have provided more details - I was referring specifically to the user interaction for grading cards. It's best illustrated with an animation, but I'll do my best with words for now: instead of presenting the user with 4-5 buttons where most of the time only 1 or 2 of the buttons should really be used, it shows Wrong and Right buttons only. Then when pressing Right, the user is prompted with "How hard was it to recall?" and the Right button changes to "Good" and two smaller buttons slide up out of it for "Too Hard" & "Too Easy" (as users of Anki, etc. would be familiar).

This sounds complicated on paper but since the "Good" button instantly replaces the "Right" button, it's as simple as double-tapping "Right" once you get in the habit. Then if you really need the "Too Hard", "Too Easy" buttons, they're easily available, just not promoted front-and-center.

Cards can also just be swiped left/right to grade them Wrong/Good.

This UI makes it much clearer to SRS newcomers what the intentions of the buttons are. When other applications present 1-4 or 1-5 buttons at once, it's not obvious that most of the time you should just be using the "Good" button by default if you were able to recall. And since you can just swipe the cards, there's no extra friction for the usual case.

But to address your question directly - Manabi's SM-2 algorithm does have an "ease factor" for cards per user, currently. As Manabi grows over time, I do plan on building out smarter behaviors that take advantage of the data available from its centralized backend model, like what you mention - e.g. seeing that people tend to have difficulty with a particular expression or kanji, and adjusting the default "ease factor" for new users of those cards based on that.

I think there're a lot of exciting opportunities from having this data! Mnemosyne has collected review data for years but I haven't seen anything done with it materially (would love to be corrected if wrong here) - I think that project would have a hard time taking advantage of it in a content-specific way since then it'd have to ship substantial databases out to users, being desktop/offline-only software. With Manabi, since everyone shares the same backend servers, cards could be easily cross-referenced automatically for difficulty adjustments etc.

**Quick update**: Sorry for the radio silence the past couple weeks; I'm currently building out a web frontend to Manabi to let you create and edit cards, so you don't have to only create content on the iPhone. I'm getting close to having this working well enough, then I'll ship the beta to everyone who's signed up for it, since I think it's pretty important to be able to more quickly create and share content from your computer, not just on the iPhone.
Edited: 2016-11-14, 10:40 am
Reply
#9
Quick update:

I've created a basic companion web app to make flashcard creation & editing easier, so you're not limited to the iPhone. Eventually I'd like to add more and more features to the web version of Manabi, but for now it'll be a supplement to the iPhone app for content management.

I'm excited to say I managed to port the iPhone's card editing functionality to the web, something I haven't seen anywhere before (please correct me if I'm wrong, I'm super curious): live-updating, editable furigana inside the text field! Rather than having separate expression and reading fields, there's one field for the Japanese expression with automatically-generated furigana that you can edit inline. Check out this screenshot - the furigana is updated as you type, and you can click to edit it.

[Image: QoIxqng.png]

Now that I have the companion web app ready, I'm going to focus on getting the first version of the iPhone beta out in the coming week or two, into the hands of everyone who's signed up.  Smile If you still need an invite, please drop your email at the bottom of http://beta.manabi.io/ so that I can add your device to the forthcoming beta.
Edited: 2016-11-27, 10:57 pm
Reply
#10
I've created a Slack channel to discuss the beta and get feedback, if anyone's interested: https://slack.manabi.io/

As another quick update, here's a screenshot of the WIP companion web app:
[Image: grMhiDU.png]
Reply
#11
It's been a while since I gave a progress update - now that the beta has launched, besides ironing out some early issues and polishing UI, I've been working on a couple features inspired by the usage so far.

I've built a great set of tools for creating new flashcards: you can make them on your iOS device or on the web, and both have automatic furigana generation inline with the text editor in a seamless way I haven't seen elsewhere before. However I've realized that even though flashcards work best when you make your own based on material you've picked up elsewhere, it's not what people typically want to do when they're first trying out a new tool.

So with that in mind, I've been building out a more substantial set of material that you can jump into right away. I've edited the "Core 5000" word frequency list to cull items that aren't suited for vocab-oriented study, and improved many definitions and some of the ordering for cohesion, to come up with an optimized collection of decks called Core 4000.

Here's a screenshot of some WIP. I have a bit more optimization to complete, then will be sending out a new build very soon. Smile

EDIT: The update is live!

[Image: k33LfaT.png]
Edited: 2017-02-13, 10:27 pm
Reply
#12
Good luck with it. Is it too late to participate to the beta testing?
Reply
#13
Looks nice! What's the URL for the web app?
Reply
#14
Looks lovely, although I've become an Anki wizzkid and this can't provide what I expect now. Definitely good for beginners who don't wanna learn Anki's ins-and-outs and want to drill vocabulary in isolation. Really love tool, congratulations.
Reply
#15
(2017-02-18, 7:42 pm)deignouze Wrote: Good luck with it. Is it too late to participate to the beta testing?

It's not too late! I'm aiming to launch publicly within a couple months, but it's never too late for feedback. Add your email at the bottom of https://beta.manabi.io and I'll add you to the next build. Smile

(2017-02-18, 9:56 pm)Bokusenou Wrote: Looks nice! What's the URL for the web app?

https://manabi.io - you can login after creating an account in the app. It's only supplemental to the app for now - the web version makes it easier to manage your content and create cards.

(2017-02-19, 6:50 am)NinKenDo Wrote: Looks lovely, although I've become an Anki wizzkid and this can't provide what I expect now. Definitely good for beginners who don't wanna learn Anki's ins-and-outs and want to drill vocabulary in isolation. Really love tool, congratulations.

Thanks for the kind words. I saw your video about how you use Anki - you have a nice setup; I'd certainly like to get more into things like cloze deletion in later versions. What would you say would be the feature you'd miss most if you switched away from your Anki setup?

One of the reasons I build Manabi was to have a standard set of powerful Japanese features for all users. While you can set up some pretty sophisticated decks in Anki, it's frustrating to lose out on functionality when you go to use someone else's deck.
Edited: 2017-02-21, 9:15 pm
Reply
#16
(2017-02-21, 9:06 pm)wahnfrieden Wrote:
(2017-02-18, 7:42 pm)deignouze Wrote: Good luck with it. Is it too late to participate to the beta testing?

It's not too late! I'm aiming to launch publicly within a couple months, but it's never too late for feedback. Add your email at the bottom of https://beta.manabi.io and I'll add you to the next build. Smile

(2017-02-18, 9:56 pm)Bokusenou Wrote: Looks nice! What's the URL for the web app?

https://manabi.io - you can login after creating an account in the app. It's only supplemental to the app for now - the web version makes it easier to manage your content and create cards.

Oh well, I don't have an iPhone or iPad so I guess I won't be able to check it out unless either the web app gets more complete functionality or an Android version of the app shows up... Anyways, thanks for answering my question!
Reply
#17
Update:

I'm readying Manabi for a Fall launch. Due to my current immigration status (I moved to Canada a few years ago), I have to wait a few more months before I'm able to put a commercial iPhone app in the App Store. So it's looking like I'll be able to release around October.

In the meantime, I've built a free app that complements it: Manabi Reader. Besides studying flashcards, I've found one of the best and most enjoyable ways of learning Japanese is by reading, and making a habit out of reading short articles every day. This app helps you do that, and will tightly integrate with Manabi for flashcards later.

Manabi Reader is essentially a Rikaikun clone for iOS (tap words to look them up instantly) with some nice additional features. It lets you open any of your own content, but also comes with a curated selection of reading material that I'll be updating continuously. This includes news sites, blogs, and folktales so far, for beginner through advanced-level reading abilities. It also cleans up the articles to make them easier to consume on your phone.

I've sent it out to anyone who signed up for the Manabi beta, and I'll be putting it into the App Store very soon as a free app. More info/screenshots soon!
Reply
#18
I'm trying to find the easiest way to make good monolingual flashcards - does this app have any built in support for that?

This may sound weird but I've recently decided I don't want to review mature cards automatically at least. I'd rather just assume they're learned and if move on with my life. Maybe crazy from a pure SRS perspective.  Is this possible to setup in your app?
Reply
#19
(2017-06-22, 9:08 am)kraemder Wrote: I'm trying to find the easiest way to make good monolingual flashcards - does this app have any built in support for that?

This may sound weird but I've recently decided I don't want to review mature cards automatically at least. I'd rather just assume they're learned and if move on with my life. Maybe crazy from a pure SRS perspective.  Is this possible to setup in your app?

I'm definitely interested in improving the monolingual support, but there's nothing specifically catering to it yet in the app. That said, when you create your own flashcards you can just as easily use Japanese text for the "meaning" field. Later, it'd be great if I could have pre-made monolingual decks available, or monolingual dictionaries tied in.

I hear you on mature review fatigue. Currently Manabi just lets you mark a card "Too Easy" - my plan is to add a feature soon that detects that you've rated a card "Too Easy" either very early on or multiple times in a row, and asks if you want to archive it / consider it permanently "learned". I feel like this could be a good fit both generally and for your specific use case - generally I want to avoid making users spend time configuring Manabi, and instead have the app intuit your intentions more naturally. How does this sound to you?
Reply
#20
(2017-07-02, 12:44 pm)wahnfrieden Wrote:
(2017-06-22, 9:08 am)kraemder Wrote: I'm trying to find the easiest way to make good monolingual flashcards - does this app have any built in support for that?

This may sound weird but I've recently decided I don't want to review mature cards automatically at least. I'd rather just assume they're learned and if move on with my life. Maybe crazy from a pure SRS perspective.  Is this possible to setup in your app?

I'm definitely interested in improving the monolingual support, but there's nothing specifically catering to it yet in the app. That said, when you create your own flashcards you can just as easily use Japanese text for the "meaning" field. Later, it'd be great if I could have pre-made monolingual decks available, or monolingual dictionaries tied in.

I hear you on mature review fatigue. Currently Manabi just lets you mark a card "Too Easy" - my plan is to add a feature soon that detects that you've rated a card "Too Easy" either very early on or multiple times in a row, and asks if you want to archive it / consider it permanently "learned". I feel like this could be a good fit both generally and for your specific use case - generally I want to avoid making users spend time configuring Manabi, and instead have the app intuit your intentions more naturally. How does this sound to you?

Yeah that sounds nice. I'll know better when I try it of course but it sounds good. I downloaded Manabi Reader but I don't have the flag card app - the reader is looking very nice. I like it more than Wakaru which is my current go to app for reading books on my phone. I'm really looking forward to your app coming out on the App Store.
Reply