1. Forum
  2. >
  3. Topic: Japanese
  4. >
  5. All too common: kanji not acc…


All too common: kanji not accepted

I find the Japanese course's lack of kanji acceptance baffling. See the example below.


I realize the course is still in beta, but the progress in accepting what is the default Japanese script is slow-going. It makes progressing through the course rather tedious. My advice to the course contributors would be to spend a few hours going through reports and correcting these omissions.

January 27, 2018



I don't think any of the course contributors read the forums. I've not seen a single post from them here. Hopefully I'm wrong though.

This course only has four contributors (if that). For a language like this, where the complex script means there are a multitude of possible ways of writing the answers correctly, you'd think they'd want a lot more contributors.

I've submitted hundreds and hundreds of sentence reports since I started the course in April. I've received 62 feedback emails. 23 of these emails were sent in the space of two hours, all from mhagiwara. If that can happen in two hours all by one contributor, it really makes me wonder how frequently they check the sentence reports and go through them. The last feedback email I received was a month and a half ago...

I've got sentences I reported 8 months ago that haven't been fixed, even though, for some of them, other sentences with the exact same problem were fixed over half a year ago...

Like other people, I worry quite a bit about what's actually going on with our course. I try to think optimistically that there's a good reason they don't want to have more contributors (such as some major technical problem with how the course works that they're trying to fix first) and a good reason why we don't see much sign of life from the current four contributors (such as they're ignoring the discussions, the incubator info page, and the sentence reports, because maybe they are working feverishly on new sentences / tree extension / some other major behind the scenes thing). But it's a struggle to keep from getting frustrated when thinking about these things, thanks to the complete silence we endlessly hear... ^^;


Oh... I didn't even know you're supposed to receive feedback for reports! I only got one, from the Chinese course, but I thought that was a single occurence, because I sent like 20, if not more, reports for Japanese with no emails back.

I started Japanese when I first got on Duo - 81 days ago. So I think it's safe to assume that during this time corrections haven't been made. I'd really like to hope something big is coming up, like the beta phase exit that keeps the contributors busy...

"Hope is the last thing ever lost"


62 feedback emails from EN→JA course
↑ Screenshot of all 62 sentence report feedback emails I've received for this course (click thumbnail to view full size image). ^^

I'd really like to hope something big is coming up, like the beta phase exit that keeps the contributors busy...

For a course to exit beta, I believe usually it almost solely depends on the number of sentence reports first dropping below a certain target level. At least I remember it was like this when the reverse tree was in beta, where the target was to have fewer than 3 reports per 100 users for the month and this continuing for a few consecutive months:


If this course has the same kind of requirement, then I think it's unlikely the course will be leaving beta any time soon. (Although somehow the Chinese course is listed under phase 3 in the incubator now, so maybe this isn't the case...?)


I just looked at the incubator and you're right, Chinese is already out of the beta. What also stands out that the Chinese course has 8 contributors on 761.000 users and Japanese has 4 contributors on 4,89 million users.

So if so many users do Japanese, why not putting more effort in it? It's the course with the most users in the beta, even surpassing a lot of courses already released. I bet there are a lot of people wanting to contribute who speak both English and Japanese.


Wow, testmoogle, that's really a lot of feedback!

rizzeau: I read in the comments that some people actually applied to contribute to the course. They fulfilled all the requirements, but they were ignored or refused.

I can't comprehend why, surely the current contributors could use some help, seeing as the course seems to be stuck?

Edit: checked and currently on Duo: Hideki and Ayakita are active (have a streak), Mhagiwara and Jkanero - inactive. But that's only as far as using Duo...

At comment below: Mhagiwara is surely busy, because he also contributes to Chinese and Korean courses. He is listed as software engineer, just as Hideki (senior software engineer). Maybe because of the course's complexicity, and Mhagiwara being busy, it's too much work for just one person. Most people who apply to contribute are not into programming, they "only" speak both languages fluently. If a Jap-En software engineer/ programmer applied, I think it's more likely he would be accepted. Duo could make an announcement to fill that position in the job offers, or state the reason for the delays at least. I agree that the lack of communication is a problem and should be fixed as soon as possible.

I am doing the reverse tree too, and the structure is a bit different from the En->Jap version. Much more casual vocabulary, multiple grammar structures are sometimes possible in answers, sentence endings like です/だ can often be ommited without tripping "wrong answer". Some phrases can't be accurately translated, so they are approximated and sound a bit unnatural in one or the other language. Necessary compromises... That must have been hell to program it.

PS. This is not a "hate" post by any means (just to make it clear, because I don't know how it will be perceived), just trying to reason and figure out the possibilities. ;)


If they have time to hold a streak, you should have time to read the forum once in a while, but that's my two cents.

Even since I found out the forum, I'm finding a lot of negative posts about the course about things I absolutely agree with. Also on Reddit I've found that most topics on this course is about how it should be improved and people are actively saying that they should another app or way to learn Japanese (including me).

What's most striking is just the lack of communication and I think that the organisation of Duolingo has to step in, either with accepting new contributors who fulfil the requirements, or just give a statement with what's going on. Even if it's that they are focussing on other stuff.

With a huge user base you can expect some improvements to be made or at least some communication. Like I said before, the user base exceeds even a lot of languages that are out of the beta.

What I also found out is that the English course for Japanese is a lot bigger! I made a post about this earlier. I understand that you can't use some kanji from that course to this course (like N1 kanji), but vocabulary is (like "apple", which is a freaking N5 word). The knowledge and vocabulary is there, improve the beta!


People are disappointed, Duo should communicate or appoint new contributors, please improve the beta and use our input.

End of rant


I edited my post above to reply, because as more "reply-to-the-reply" posts appear, the more squeezed they get and take up more space vertically.


The quick Chinese beta exit is a surprise, but I think it might have to do with the course set-up. The vast majority of it winds up being character<->pinyin matching about which there is no possibility to report anything. Hence, the level of reports coming in is mechanically much lower. Chinese also has only one writing system and inflexible word order, so overall it's probably pretty manageable translation-wise, at least relative to other completely-not-related language pairs. And the number of sentences seems quite small relative to what other trees with a similar number of skills have.

Another odd case is English for Thai: it took like three years to build and then got out of beta in maybe two months. Most likely scenario: Thai-speaking users, necessarily new to Duolingo, just didn't even know about submitting reports. (Sure I'd like to think they just did a really good job with all that time they took, but...)


That's the exact reason I keep using the mobile version. I read opinions that the web one is better, as there is more typing and less word-picking.

But when I type, my answers can go unaccepted, because inputting kanji is more convenient (not to mention a great learning method) than typing a string of simple kana.

So for now, I'll stick to the app. When Japanese comes out of beta, I'll do a rerun through the tree on the web version solely, for testing and correcting purposes. I really wish that accepting kanji worked well right now, though. I feel like I'm missing out on a lot of practical writing experience... さあ, would that I could :(

Does anybody even know if the course is still actively being developed or was it just shelved and is not on the way of coming out of beta antytime soon? The WIU reports give away nothing. I looked for any kind of blog about the course's development and found null.


It is possible on the desktop web interface to choose the word bank rather than typed input, at least for inputting Japanese. But it remains very sad that I resort to this, because I'm not actually practicing my ability to generate Japanese words; it's more like solving a little puzzle-game with limited variables.


Same here. I use the iOS app almost exclusively - a lot less frustration this way.


Luckily all of my kanji is still accepted. I'm also doing the reverse tree and I'm finding words and kanji there that I haven't seen at all, like the kanji for woman: 女の人 (おんなのひと) is something everybody knows (I assume), but there I found 女性 (じょせい).

I think everybody wants some clarity what the contributors/mods are working on. Because a lot of people think that this course is being neglected.

Edit: I did a course higher up the tree and OP was right, a lot of my kanji was not accepted, the kanji for "grandfather", "warm", "hot", "cool" were all denied. I reported it.


i just recently finished the japanese tree for the trophy and it doesnt even get to the level of 2nd intermediate class i took in university, there is a lot of room for improvement and I hope that they are working on it to make it easier for learners


If you mean that it covered two beginner semesters and one intermediate semester, then that itself would surprise me (in a good way). It's the shortest tree for a natural language on Duolingo, and Duolingo tends to claim it gives you one or two college semesters' worth of content.


Duolingo isn't meant to be a full language course. It's basically just something for reviewing vocabulary.


This is one of the reasons why I'm using Lingodeer and other apps to learn Japanese. Once I finish Lingodeer, I plan on testing out of the Japanese Duolingo course just for the trophy. I'll probably use it for the clubs aspect and try to get to level 25, but at the state it's in now, I don't think its as extensive as the other Duolingo courses. I hope this is fixed when it comes out of beta, mostly because I know there will be people who think you can become fluent in Japanese just with Duolingo. I think the course has less than 100 kanji. That's definitely not going to happen lol. I hope they add more kanji in the future.


I don't know if fluent, but you should be able to assume that you can at least speak/read/listen to the language on N4 level (I'd be happier with at least N3), but in its current state it's not even on N5.

I saw yesterday that a little bit is explained in the web version, but it's still not the case in the app where the most users are (according to Duo's own statistics 80% use the app).

At this point everybody just want some answers, I think people even settle with "I'm sorry guys, but we're swamped with other stuff". It's the constant radio silence that is killing the motivation of a lot of people.


I agree. Usually, Duolingo courses get you to around an A2. If you have really great memory, or you keep refreshing your weak skills and you read and study all the grammar notes, you might end up as a B1. It all depends on how much work you put in, in my opinion. It's great that the community can volunteer to create courses, but at the end of the day some courses become more developed than others because of this.


I am relieved to find this discussion. As someone who once studied extensively and then let it slide for 20 years, I've chosen to warm up with early lessons, and I prefer typing in the web app (The "word bank" approach is too easy, like a cheap multiple-choice test. But whenever I type, there's a 50/50 chance of running into really arbitrary limitations (can't use kanji for "kanji", can't write "still not" rather than "not yet" etc.), and then there are lessons that are super-repetitive (one of the classroom lessons alternates between exactly 2 sentences). I would feel encouraged if I heard that someone is putting effort into improving the Japanese unit in some way. Heck, perhaps Duolingo could allow a crowd-source solution whereby people with sufficiently verified Japanese skills can actually weigh in and if enough of us hit "report" with the SAME exact complaint, the system starts accepting it pending review by Duolingo staff. Of course, generating additional content requires more centralized planning. But they could start with the low-hanging fruit...


Given how long it took English for Japanese to come out of beta, I assume a few hours every day for months would probably do the trick.

Japanese<->almost any other language is just a difficult pairing for the Duolingo set-up. You probably don't run up against the 3,000 translation max in Norwegian very often, but apparently it's routine in Japanese.


@piguy3: Yes, it seems to me that Japanese-English needs not a mere LIST of acceptable translation, but a primary list PLUS a bunch of rules about how any of those sentences can be subjected to substitutions/syntactic re-orderings, implicit subject assumptions, variations based on formality, and of course kanji-vs-kana for direct Japanese input. There's a manageable list of variation-directions with respect to any one of these issues, but the list of whole sentences is pretty unmanageable.

I noticed once that duolingo flagged a typo in my English, but still accepted my answer. I think it would be great if the Japanese module erred on the side of accepting a VERY wide range (including all those translations that a threshold number of well-qualified users report as "should be accepted"), even while nudging us toward the preferred translation. Then people who actually have significant Japanese knowledge can shrug off some of the arbitrariness and get on with learning.


Some kind of "provisionally accepted" category based on frequently-reported translations would certainly be welcomed by users. How technically feasible it would be (people commonly substantially underestimate the technical difficulty of seemingly small changes; for instance it is reported it took half of Duolingo's staff six months to make the changes to allow there to even be a Japanese course) and how the details would work in practice I have no idea.


It's been a year since this was posted and I still get almost every listening exercise wrong the first several times as I try to surmise which words they want in kanji, which in kana, and which in a combo of the two. Is the only option to use the word bank? I don't consider the word banks to be useful, because I don't learn from them like I do from typing in the answer.


yeah that happened to me!

Learn Japanese in just 5 minutes a day. For free.