I can't believe how difficult it is to get it to work properly.

I finally managed to create a vertical text, with nice furigana from an "innocent" txt file using this method: http://ebookyaro.blogspot.jp/2012/10/kin...epub3.html but with KindleGen instead of Kindle Previewer (because I'm on linux). I used the argument -locale jp.

Anyway it's looking perfect, except for one thing: it won't select words, but entire sentences. I don't get it, it's properly tagged as Japanese. So I thought: "Well maybe there's something else missing!" so I compared the xhtml files to those of こころ (a free ebook on amazon.co.jp, which is working fine with the dictionary). They are - as far as I could tell - the same.

Has anyone been able to find a solution to this? It's not my first time looking around and damn would it be useful! Surely it's a little tag missing, because the bulk of the text is formatted just the same way as こころ. Any ideas?


Edit: I tried with Kindle Previewer as well. I also tried following exactly the steps from http://timagawa.hateblo.jp/entry/20130113/1358088551 and no matter what, I get a nice looking text that just won't select anything but entire sentence (actually from one punctuation mark to the other, such as commas...) Also I know the dictionary is working, because the few one-word sentences pop-up correctly.

I had a similar problem, using amazon's "send to kindle" service solved it. I guess they have more powerful converters on their servers.

I wonder why people keep buying Kindles, when you can buy a Nook instead, root it and install regular Android apps.

I think apps like Ankidroid, epwing dictionary readers, Jade Reader or Rikaichan for Android should work, which would make the Nook the ultimate Japanese reading tool ever.

Japanese and E-book readers: A guide

BTW, I noticed their conversion service on the "send to kindle" became very smart (recently?), it took a RPG book with page numbers and those repeating chapter titles in the top of every page, and stripped those perfectly, and I just tested with one of those academic journal PDFs that are formatted in 2 columns (also page numbers) and it converted it pretty well too.

@visualsense Really? Makes no difference for me Do you by chance have details like, format/orientation you sent?

When the text is horizontal, the dictionary parses correctly, or at least it lets me delimit words. It's only when the text is vertical that it won't work. And then again, Amazon free ebooks are both vertical and working flawlessly, so there has to be something I'm missing.

GREAT NEWS!

It's actually working. I sent a vertical mobi file and received a vertical azw3 file parsing correctly (note: I tried simply converting to azw3 on Calibre before without success). So thanks visualsense!!

Not so great news: I investigated a bit by editing and comparing the working and the non-working azw3 files. It seems that the only thing that will make the magic happen are those 4 "unknown" files, which are binary files. When I remove those from the working edition, it still kind of works, but I have to use "Go To" and it doesn't parse the text correctly. I could be wrong though, maybe I've missed something I wish there was another way than relying on Amazon's magical service, but at least it works!

it was horizontal, a mobi I bought outside Amazon, the reason I sent it (a file that was already a mobi) was it had some crap formatting that made the characters super small, I was surprised when it fixed the dictionary thing too

oh, great it worked

I guess that parsing Nihongo without space between words is too hard to do it realtime on a kindle, they just code it on those files.