1. Forum
  2. >
  3. Topic: Japanese
  4. >
  5. Translation from English into…


Translation from English into Japanese: A problem in the word selection questions

First: I am already able to read and speak Japanese, so this issue is not really a problem for me. But I think, it is one for beginners. The examples are out of my memory, so they may differ a bit in the real questions.

The issue occurs in the questions, where one have to translate an English sentence into Japanese by selecting boxes with words. I can understand, that the okurigana are divided from the stems, but a separation in the middle of a word?

This type of question is normal in Duolingo, but the way, it is implemented in the Japanese from English course is not optimal, I think.

The words are devided on unusual places. It is difficult to explain, so a picture shows more than 1000 words ;-)

Why are words like 男の子、女の子、女の人、男の人and some others partly or even always divided into there single characters? I. e. (男)(の)(人). Some other words like がっこう, まど and ゆうめい are always (?) devided into two boxes (ゆう)(めい), (ま)(ど) and (がっ)(こう) and one have to connect them by selecting them together in the correct order (of course) to answer the question successfully.

In English, it would be like (Scho)(ol) or (fam)(ous):

  • "These women are famous."
  • men e these ol house ous ar fam wo girl.

This is ridiculous, isn´t it?

And that´s not all: Sometimes, not only the words are separated, but even the sentences at any place:

  • There are a lot of men. (男の人がおおぜいいます。)
  • が ます ど 人 いい うるさい の おおぜ 男

男の人 and especially おおぜい are each one word and shouldn´t be divided. The stem of います is indeed い so it could and maybe should be divided from the stem.

So the selection boxes would be better something like

  • が ます ど 男の人 い ま おおぜい うるさい

I think, this makes the learning progress unnecessary difficult and should be changed.

June 27, 2017



As a complete beginner, yes, I certainly agree. I also wish the words in the boxes would be pronounced when you click on them — like they are in every other TTS tree on Duolingo.


I couldn't agree more. I think DuoLingo could benefit from these suggestions, as I too had found it funny when they split the words or the verbs in places where it didn't need it.


Hmm - I've noticed this too but it hasn't bothered me that much. Words with a long vowel like いもうと or おとうと are often broken up, so I am guessing that the goal is for people to remember to add the extra う.

I read yesterday (wikipedia, I think) that the average total length of stay on Duolingo is two hours. Not much! And I wonder if it's even shorter for Japanese - at least in my group, there is almost zilch activity. So from that perspective, maybe you have a point, and leaving words whole might increase people's stick-to-it-ness.


This is interesting: May I ask you, if you are a beginner in learning Japanese? Or did you have some Japanese knowlegde, before you used the Duolingo Japanese from English course?

And yes, いもうと is also one of these examples. But as I wrote: I think this makes thinks unnecessarily more complicated.


Yes, it makes it more complicated though I have to admit I enjoy the puzzle aspect - just now I was searching for 学校 when I finally noticed there was がっ and こう available. Figuring that out was kind of fun!

I've had two years of Japanese continuing ed which goes quite slowly - so I'm not an absolute beginner but by no means intermediate. (To give you an idea: NHK's "Easy" News is really, really hard for me.)


To give you an idea: NHK's "Easy" News is really, really hard for me.

Hmm, there was a website many years ago, where you could watch recent interesting news with Japanese subtitles (which are common in Japanese TV anyway), a written transcript in Japanese, a translation into English and some more useful things.

I tried to find it again now, but maybe the website isn´t available anymore? But if I find it, I will post it here.


That would be nice. Right now, I take NHK's written transcript and plug it, sentence by sentence, into Google Translate.

imiwa? also helps - there is a feature where the content of your clipboard is broken down and analyzed. So between those two, I can get an idea of what's going on - but it is hard work.


Ok, this is an interesting fact. Maybe I am wrong. From this sight, it may have an additional learning effect.


I've seen this too and it's frustrating. Even more so though, are when the kanji choices are ones I haven't been taught yet. Then when I look at the tool tip for the original sentence none of the words in there match the choices I have. I actually can't finish Time 3 because of this problem so I can't move forward with my learning.


I basically let those one sit until the end of the lesson when hopefully there's only one or two and I can get the answer into short term memory.

Hopefully, the contributors can work out what the problem is even when the only report option is "there's (some other) problem with this sentence."


I do that too and I find it very confusing to "learn" this way, especially because there are no grammar notes (I normally use the website).

I'm a complete beginner (apart from knowing Hiragana and some Katakana when I started) so it is all very challenging and I am progressing very slowly. Waiting for the web version...


Japanese would obviously be more than hard enough even with a perfectly polished course, but the kanji-only hints when all the answer choices are in hiragana and other beta-version roughness, yeah, they make it extremely tough going.


Thank you for sharing your experience. That's what I thought. Because I am already able to read Japanese and know all the words and grammar in this course, it is a bit difficult to imagine, how this issue may affect an absolute beginner.

I really think, this type of question is not solved optimally.


I don't mind if stems and particles/suffixes are separated actually. I think it doesn't hurt to be forced to piece together the correct words from such pieces. And I don't think I mind having multi-kanji words divided into kanji, but I'm not certain how I feel about that yet... However, I agree that completely unnatural splits like "おおぜいいます" in your example, are pretty unreasonable.


I don't know. I kinda like the 'puzzle'​ thing. If the sentence has 'not go' You have to find the root iki, the polite mase and not​ forget 'n' to make it negative. I'm a beginner, but not complete beginner that I think it would be too easy otherwise.

Beginner as in I know hiragana​ and katakana, but not kanji, can say sentences. Not working much on reading, since my priority is speaking. So, Duolingo is a good way to make sure I don't forget the kana.


This is interesting. But I mentioned, that I understand, why the stem must be separetad from the okurigana. But single words? So you would also accept this also in other languages, as the example, I provided for English?

  • "These women are famous."
  • men e these ol house ous ar fam wo girl.


In English, not exactly like that because that's not how English works. I mean splitting iki and masen is ok to me because it will remind the learner that masen negates. Splitting mase and n is ok to make learners notice the difference between -masu and -masen. For example, if you're looking for masen and there's no other option that starts with ma-, you can just​ click on it without actually reading the entire thing. But if you split mase and n, you have to pay attention to be sure your characters are complete.

In English, that would be equal to splitting donotgo to do not go (which is already how it's like in English).

Not sure how it would work in English. Maybe in English, it would be like:

sleep __

choices: sound, ly, ness, ing

-splitting makes the learner pay attention to how -ly, -ness, and -ing changes the meaning of the root word and choose accordingly.

I'm not saying they should do this in English (after all, reading characters isn't something to practice in English if you're not a kid). Just trying to illustrate why I think it works/can be helpful in Japanese.

But I don't know. Maybe I just haven't encountered any awkward splitting yet. It would be awkward maybe if they split ocha into ochi ya or natto to na tsuto. These are words with those small characters that mark some changes in pronunciation and they are just simple nouns. There's no point in splitting them.


When you read my initial post, you will see, that I´ve already mentioned, that I understand splitting okurigana from the stem. And I understand your point of view, but I´ve seen some very arkward splittings (and provided one of them here), that may encumber the progress.

For me it is no problem to find out the words, because I already knew them very well, before I tried out the Duolingo course. But a beginner? This may be confusing.

Regarding the ん I have to think about it. Actually I think, splitting the ん from the negative formal okurigana don´t make any sense, because as far as I know there are no different endings, you could choose from to add to ませ. The positive form is ます (masu) and you can´t connect ん to this, because you would have to change ます to ませ first, but in my opinion it is more logical to learn the negative and postive forms as they are: ますand ません. And for the informal expressions, you don´t need ますor ません. So splitting ん from ません makes things simply more difficult, I think.

Same for splitting somewhere in the middle of words. I´ve seen it in no other Duolingo course and if I remember correctly, also not in the English from Japanese tree, but I will check this later.

The only reason for splitting ん from the okurigana may be from the programmers view, because there are some expressions in colloquial speech, where you don´t need the ん after ませ, i. e. ・・・くださいませ。

So no, I am unpersuaded about the advantages of splitting Japanese words elsewhere.

But I may not be the target of such a course, because I am already able to read Japanese and my view is from this perspective. I can´t imagine the 'puzzle' effect for me it is not funny. Perhaps, because I am reading too fast and it is ridiculous for me to search for single syllables instead for whole words, especially in case of longer sentences.


A lot of the "issues" in course, so far I can attribute to difficulty in "programming." For example the audio for the kanji naka (as in Tanaka) is always "naka" even when it should be chuu (as in the kanji for China). I just assumed they link that picture with the audio. To accommodate changes in the kanji pronunciation, they'll have to make several files and make sure the correct combination of picture and audio will show up on a certain question. This difficulty also shows in when you click words and the audio will read it per character and not the entire word. To me it's like they decided "ok this picture goes with this audio." It's easier for them that way, I guess. I think this same-character-several-pronunciation thing is also the reason why Google Translate is still having a hard time. So, somehow I can understand the challenges they need to hurdle.

The splitting of words may have been intentional or it could be the only easy way to design stuff at the moment. Either way, I'm not really bothered with it (I'm actually more concerned with the audio). We do agree that splitting iki + masen, for example is alright. I do, however, further that it's ok to even split words into iki + mase + n, and other words like this. Here's why:

My impression of this course is that it puts a prime on getting to know the characters. The point isn't just to teach you a word so you know what it means. It's to let you know a word enough that you know what kana or what 2 or 3 kanji to put together to make that word. The equivalent of this activity isn't splitting an English sentence into:

"These women are famous." men e these ol house ous ar fam wo girl.

Instead, it's like solving a jigsaw puzzle, where the pictures aren't clearly cut to show the eyes, the nose, and such. Sometimes, a piece just shows part of the eyes and you have to find the rest. Japanese writing is basically pictures. You put pictures together to express a message.

This works for me because I'm using DuoLingo for reading practice and intro to kanji. If users here think reading isn't their priority at the moment, then this isn't the right material for them. You don't need to learn the characters to know how to speak. Reading and speaking are two different skills.

BUT, there's an advantage if you at least know the hiragana and katakana, mainly because materials online usually use kana. Another benefit is you'll only need to deal with one more script (kanji) if you eventually decide to start reading. If these aren't enough reasons and you really want to start speaking Japanese, then start speaking the language and set aside learning the characters for now.

Given that, I guess, I do agree this isn't for complete beginners. Because I think complete beginners should start with speaking and not so much with reading. Unless, of course, they don't really mind the order in which they learn the four skills (speaking, listening, reading, writing).


Ok, thank you for sharing your thoughts and I think, I now understand you better. From this point of view, it may be helpful to split words in single kana for some students (surely not for all).

It is perhaps a question, how someone learns. As I described, I rushed in the past to learn all kana in two steps within two weeks, because I wanted to be able to read words and sentences and didn´t want to think about single characters. I learn and read words as they are like in other languages and that´s why I don´t like, when words are intentionally splitted at random places.

But I understand now, that other could see it differently.


No, prob. To me it's just that 'This material is imperfect, but can I make it work for me?' With all it's imperfection, this material is big on reading. I like that I can learn a few kanji early on. So far these are words I already know how to say and use in a sentence, so I can just focus on getting to know the kanji.

So it really depends on the learner's goal. If they think the material isn't align with their goals, they should leave right away and spend their time with materials and activities that are.

That's why I stopped taking the German course. I didn't know how to make it work for me.


Obviously everyone makes an individual determination as to whether what's provided currently merits further attention, but the logic of Duolingo, both explicitly stated and as evidenced by ever so many decisions they've made, is to be an introduction to a language for complete beginners. Hard not to see this broken-up-word business as a departure from, and detriment to, that mission.

It's not hard for me to think about the position of a complete beginner, for in Japanese I am one. Japanese doesn't separate words in writing. I assume there are intonation distinctions that demarcate words, yet I at least certainly can't hear them yet (in marked contrast to many other languages I don't speak at all but where I at least think I can hear where words start and end). In short, I'm left not even knowing what's a word and what's a part of a word. It's easy enough to figure out what's going on when it's a case of okurigana, but splitting things like random nouns as I see mentioned here makes no sense at all, given the purported objectives of the course.


Yes, that's exactly what I thought. Thank you sharing your experience with that issue.

Additionally I haven't seen this issue in any Japanese textbook for Japanese children.

I still think, the real reason of all these splits is only to ease the programming. It is simply more easy to code (with the disadvantages we encounter in this course).

You may see that in the fact, that many splitted parts does not have a pronunciation or at least a wrong one and also there are wrong pronunciations in some sentences.

As an example could be the kanji 今: This character stands for the idea 'now' and the Japanese word for now is 'ima'. But when it is connected to other kanji like in 今夜, the chinese reading 'kon' is often used. So the compound is spelled 'konya' not 'imayoru'. So they have to program at least for the kanji many boxes with a flag for different readings and have to provide the correct one to choose from. And that's not all: Certain compounds can have a special spelling depending on the sentence, they are used in: So 今日 is spelled generally as 'kyō' and not 'konnichi' (which is generally written in hiragana). But sometimes you can see even in Japan 今日は for konnichi wa and it depends on the sentence, if it is spelled as 'konnichi ha' or 'kyō wa'. An example sentence: 今日は良い天気ですね! Here 今日は is spelled as 'kyō wa' and not as 'konnichi wa'.

So you see: This is really complicated, because some kanji can have really many different Japanese (kun yomi) and Chinese (on yomi) readings and there are many exceptions from the rules.

And for the okurigana (flexions): splitting them from the stem reduces the code, because they are redundant. I. e. ます and ません and also others like なかった or たくなかった can be added to any stem (Japanese is really strict here), so that the programmers only need the different stems and the different okurigana once (just an example to show, it is of course a bit more complicated). It is an example for using same patterns in different words. They can do it, but actually it is far from solved optimally.

But in my opinion there is no real underlying advantage for learners.

And although a very enthusiastic and very motivated learner could fight through it, the course would be more effective without these splits, I think.

I worked now about 3/4 through the reverse tree (English from Japanese) an there are no bad splits at all.


@piguy I get your point. And I agree that this course can be improved.

I'm not sure​ though if DuoLingo ever posed itself as for "complete beginners." Free and 'personalized' course as far as I know is what they tout. It has a broad user base, I'm not sure if it's a good idea to design content specifically for complete beginners. This is designed like a game. So they can't make it too easy (complete beginner level) nor too difficult (intermediate and up)--just my two cents, I don't really know what they factor into they're decisions.

From my experience, Duolingo offers an ok introductory course.

But I think an "introductory course" doesn't necessarily mean for "complete beginners." And by complete beginners, I mean pre-A1 in the CEFR scale. An introductory grammar course, for example, isn't a good idea even for A1 beginners.

I'm a beginner, too. I'd put myself on the A1 level. I don't study grammar yet. I don't even know the term 'okurigana' even though I'm sure I've heard/read it before. The thing is if I think 'I don't need to know this right now,' then I just set it aside. For example, I don't need to know the term okurigana (or study grammar) to say:

Ringo o tabemasu.

Atode ringo o tabemasu.

Asa ringo o tabemashita.

Mainichi ringo o tabeteimasu.

Or to: - introduce myself

  • tell the time

  • ask simple questions like what is _ or where is __

If you already know some words and expressions, I don't think that classifies as complete beginner.

Anyway... Back to Duolingo.

I've actually only encountered the words for 'desk' and 'chair' here. So far, I haven't seen them split into parts. I can always tell them from the choices even without my internal voice actually reading them. It's good if you want to read Japanese sentences (It's like looking at an English word or sentence and knowing what it says without your mind 'saying' it). But it's probably not that good if you still need to learn the word as in my case. I've managed to hold myself from clicking 'isu' right away and actually read/say it aloud first. I'm​ starting to do the same for desk, which I can't remember now. I know it's three hiragana characters with e and maybe tsu. If I see it, I can tell it right away, like it really stands out from the choices for me.

If all the words are complete, it's easier to fall for this trap.

Another kind of dependency is hachicspr's example (sorry for mentioning​ you, I don't mean to put you on the spot). She knew the word shimate and her mind automatically looked for something that looks like it. Now, our brain is good in scanning stuff. If that picture of [ しめて ] was there she would've clicked it without second thoughts. But, it wasn't, throwing the mind off because what it's expecting wasn't there. Now, it has to pause and look at each choices and sort them out. The brain learns better (stores something for longer term) if it's​ doing some hard work, as far as I know.

And yeah yeah we can all just be done with it if Duolingo just presents each word as a single unit. I'm just saying there's also a benefit on how it currently is (at least when it comes to the issue of splitting words). And if it's too difficult for you right now, maybe this isn't the right activity/way of studying for your current level.

As I've mentioned, the strength of this material is in reading/character recognition. If you need to build vocab or become familiar with common sentence structures, look for other material/activities that will expose you to those kind of stuff. There are other things in this course that I think aren't helpful for beginners:


  • using 'wa' in sentences where a 'ga' or 'o' would have been ok. I think it's better letting a student get used to situations that uses 'ga' or 'o' before showing when it can be 'wa,' which can change the meaning or show contrast. Instead, you see wa sometimes, ga/o sometimes, and none when someone​ might expect a 'wa' based on how other sentences were.

I guess to me this are bigger issues since they don't seem to add any value other than maybe you'd be compelled to research on it (which I don't really think beginners should be spending so much time on).

I'm at that point where I might not be able to help someone by explaining​ grammar details and why it's this particle and not another, but somehow I can understand the meaning and the subtle changes in it when the particle changes. So I don't really mind if just for myself, but I'm curious if it's actually helpful to others.

With that, I only recommend this and other character-heavy or reading-based materials for those whose priority is not speaking or as a supplementary material.

@InuzukaShino I think we can all agree this material at its current state isn't at its best (I even wonder if Duolingo, in general, is good for beginners). I also agree and mentioned how this splits can all just be in the programming.

But I guess what I'm driving at is that the content of a learning material is one thing, playing its strengths and weaknesses to your advantage is another.

For learners who self-study it's an important skill to practice.


i completely agree with this. when i first encounter something like this, the question asking for "please turn off the electricity" and i'm searching for しめて but can't find it. after i got wrong i just realized they put it like 電気 - の - はし - めて - ください


Certainly! I am a beginner myself and I thought this was a bit weird. Also, I noticed that they made the answers blatant/obvious...or they put the answer and an answer to another question/definition when you select a box answer.

Ex: What is "Good Afternoon"?

(こんいちわ) (おちゃ) 

Will this change soon? I am still a beginner so I was thinking perhaps when it gets harder this would change as well.

~ありがとう ございました


I think the same. Who works on Japanese course can divided words like in Japanese books for kids: ex. in Japanese books for children, they use only kana and make space word to word. 男の人
が おおぜい います。 It will be nice if they introduce new words first in kana with meaning in English and later in kanji, and continue with it.


It seems reasonable to split up words entirely mae out of kanji to me as we have to work out how to put it together but splitting up a hiragana word like まど into ま and ど which it seems to do a lot is, just as you say, ridiculous. Like splitting window into win and dow.

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