My dictionary is too slow and has too many useless words!

After using jisho all these years, I decided to try a different dictionary, so I downloaded midori.
I wanted to find a Japanese-English dictionary that’s easy to use offline while playing games, or for help in translating in the wild by taking photos of the text.
It started fine, I was playing a game and scribbling the sentences on my iPad into midori kanji and all which works pretty well (first time trying the scribble option with Japanese) as long as you scribble the strokes fast enough.
Since it’s the first time I was using it, I just kept on adding sentences, to test the scribble recognition, 5 vs ら、 3 vs ろ 、n vs ん 、and then this happened:

Now, I’ll be the first to admit my Japanese is super basic, almost えんぴつです, but even I know that ほしかったんだ shouldn’t be a major event for a dictionary ね even when it’s being spelled 欲しかったんだ.

Which led me back to square one - how can you tell that a (digital) dictionary is reliable?
Did I know that Wanikani (not a dictionary) is a reliable source when I started? Nope. I took a leap of faith ‘cause you gotta start somewhere, but at this stage I really want to find one reliable source to work with dictionary wise, otherwise it’s exhausting.

So tell me, how do you use your dictionaries? Can you tell when it’s wrong? Can you tell when it’s correct? How can you tell when you can’t tell because you don’t know - that’s the conundrum.

3 Likes

Every dictionary has at least one “mistake”. :face_with_open_eyes_and_hand_over_mouth:

2 Likes

As you don’t separate words by spaces in Japanese your dictionary tries to parse your entry and finds ほりし and かったん as one of the possibilities, another being ほりしかった, a conjugated form of ほりす and んだ. The dictionary can’t know without context what is ‘useless’ for you. It did give you the translation you were looking for (欲す)plus some others - you can’t blame it for that.

It would be the same in English without spaces. If you entered ‘hisword’ the dictionary could understand ‘his word’ as well as ‘hi sword’ (I’m sure there are better examples).

4 Likes

Presumably the OP doesn’t want archaic words like 欲す, which also doesn’t work grammatically with the かった there, and only the correct interpretation of しい in past tense.

3 Likes

As I saw the furigana ほりしかった I thought that that was the input.

I tried the same input こういう欲しかったんだ in takoboto, and the result is better than I expected:

1 Like

I assumed the app was automatically applying furigana to the entered words based on what was parsed. If the ほり was applied by the OP, then that would make the mistake easier to understand.

1 Like

Here it is, as others have said, not the dictionary itself that is wrong, but the app’s attempt at parsing the words (which has tripped it up also in its attempt to apply furigana). Parsing is hard because of the lack of word boundary markings, and different apps will do better or worse at it. Generally I would recommend not relying on an app for that (or for adding in furigana, which is also a hard problem) and instead picking the words you want to look up yourself.

The dictionary information itself is clearly the standard EDICT data file that basically every free J-E dictionary will use (and which is the same data jisho uses). The one thing I would kind of critique the app for there is that EDICT marks the ほりす entry as “archaic”, but the app isn’t showing you that information.

13 Likes

Hmmm one “mistake”, that, that means I found it! I found it and I can keep on using midori forever! Woohoo ٩(^‿^)۶

Just to clarify, the title of this thread was a tongue in cheek paraphrasing the “wanikani is too slow and teaches useless words” made me laugh thinking what if people reacted to a dictionary that way.

@ Leebo assumed correctly, the furigana was added automatically only after you click translate, depending on what the dictionary chooses, unless you opt out.

I wasn’t asking for a particular anything at any point, this was simply a first run to learn and develop familiarity with the UI. The sentences where totally random since I was simply copying the game dialogue (A Short Hike if anyone is wondering), and that’s why it was so puzzling.
It didn’t make sense, so I wrote ほしかったんだ hiragana only to see how it would deal with it without the kanji.
And that’s not all, this is what you get when you click the words on the left column:

So the short answer was wrong, but even in the case it was correct, why would it prioritize the archaic option first?

Does this mean it’s better to use the dictionary form only?
If so, the UI is a bit misleading.

1 Like

I saw the EDICT in the app description (and probably saw it gazillion times when using Jisho online and forgot), but it was kinda meaningless.
There are so many options and I don’t really have the knowledge to be able to choose the right dictionary. That was part of why I opened this thread, I needed information that is more up to date and informed than what I was able to find, and maybe some reality check about what to expect from a Japanese digital dictionary.
It’s a change of pace moving from curated material to using a dictionary on your own with random material. So I’m a bit overwhelmed, and there’s a learning curve here. I’ve been reading about dictionary form but I never really used the “formula” outside of grammar drilling apps/books.
There’s also the possibility that I still need my training wheels.

Well, you can see in the second screenshot that it does, but in a different section. I still think it’s a weird mistake, though.

It’s android, right? I’m an applehead…
It does deal better with that sentence.

1 Like

Yes. It seems that there is no iOS version.
I didn’t even know that you can enter whole sentences and that is so good at parsing them. Up to now I only entered single words. So thank you for your post!

1 Like

You’re welcome (╹◡╹)I’m glad it helped you discover this feature (and a little bit jealous - it looks good and useful).
The technology exist so it made sense to me that the next step is to merge a dictionary with a translator, what makes the difference is probably the material the ai behind the sentence translation learned.

The thing is, technology changes open source starts and stops and there are forks that remain and what not, and new apps emerge and old apps being updated. Old thread data become absolute. So after contemplating a bit I decided it might be beneficial not only for me but for everyone to be up to date with the current trends in dictionary world, and maybe introduce beginners to the why and how of it.

1 Like

Ihelpedmyunclejackoffahorse

2 Likes

For what it’s worth, Midori has been my go to iPhone dictionary app for years and I think it’s great. Granted I don’t use the translate section because I don’t need an app to parse sentences for me. I also know that the developer is pretty responsive to feedback. If you submit this as a bug, there’s a chance he’ll fix it. Though there’s also a chance he’s using some third party parser and can’t really do anything about it.

3 Likes

Same as @seanblue, I’ve also used Midori for years. I also have never used the translate section. It is pretty good at kanji recognition when drawing them, even without correct stroke order and such. But the more complex a kanji is (or the more uncommon/unusual/obscure it is), the less likely it’ll find the right kanji. Usually radical search works in those cases.

I found early on in my journey that my overall comprehension started leaping forward when I stopped trying to fit everything perfectly to English grammar (like a good translation does, or even a mediocre or bad one does). Literally translation can be useful if it helps you wrap your head around how Japanese grammar actually works. So I never use translation features unless I’m completely and utterly stuck. And if I want that, I go to one that specializes in that, not a dictionary app that have probably thrown it on as a value add. If it isn’t the main feature, it ain’t getting the same attention as it would need to be great.

4 Likes

First thing first : @seanblue and @MissDagger - thank you for your input as midori users.

Yeah, that’s how I used Jisho, just kanji and kanji through radical search, but this translating feature caught my eye, and I really enjoyed writing those sentences, instead of going kanji by kanji, it’s more of a personal preference, I guess, a change of rhythm.
I haven’t tried it yet, but one of the translating features is text from photos recognition, and I thought it might be useful when I’m in Japan, next week. I didn’t use any of this things the previous times I’ve been to Japan, and managed fine with what I needed, using Japanese most of the times, but I’m a curious person and it seems like this feature could be helpful at times.

It really depends on what I need at a particular time. For my pace at the moment, parsers are quite useful training wheels when I don’t feel like putting all the effort, I want the exposure with a little less work a little more fun, and the way takoboto does it (I do need to check midori on my iPhone as well) looks like a great transition from grammar apps to dictionary only. It tells my brain it’s the same concepts it’s seen before just different environment.
I think in that manner, a parser is better then a translator, since you still gotta do some grammar work on the sentence as a whole, while with translation, you don’t really need to work on the relation between all the parts, unless you choose to.

wow, midori, nostalgia hit me now,

one of the first apps I bought for the first ipad released

Nowadays I use takoboto on android

2 Likes

BTW, I assume 頑の中 was meant to be 頭の中. Just a mistake in kanji recognition?

1 Like

Yeah, good call, thanks for catching it, should have been 頭の中, she’s telling him she was caught up in her head all morning (I knew she was telling him about how she was, and I knew she was feeling down, so the idea of feeling stubborn in the inside did seem plausible, yet a bit odd, but not far enough to dwell about it).
It’s probably me not recognizing the kanji in the game as 頭 seeing 頑 vocabulary in my reviews every day, and trying to distinguish pixelated kanjis on a small screen, and my lack of knowledge - but that’s the whole point of what I’m doing, learning and expending my knowledge, so it’s a really good mistake.
With the game itself, It’s hard to tell sometimes, the first sentence of the game talks about the 景色 but the 景 had 口 radical at the top instead of 日. It’s a good practice, though, since recognizing kanji in the wild is like that, and that’s why I want to start using the dictionary more, than relying on a closed educational environment the feeds you everything, or I’ll remain in the beginners level forever.

Your call was even more important than my personal learning experience - I checked the visually similar lists of both 頭 and 頑 - they have 願 for both of them, and both of them for 願 but none of them for each other. I’ll write them in one of the threads that deal with this kinds of suggestions.