I'm looking for some tool that can help me but I don't know if this exists. What I would like to do is to be able to convert the output of for example Google translate to a sentence that contains the hiragana for Kanji written in [] after the kanji.


This PDF describes some of the difficulties in ascertaining how to read the two-kanji name 淳子. This might provide some hints as to why such a tool is ultimately an impossibility.

Even if we ignore proper names (since these are notoriously variable in Japanese), we find strings like 金山, which could be read variously as かなやま, きんざん, or こんせん, depending on context.

Ultimately, I don't think such a tool exists. If it does, its capabilities are probably restricted to those sets of kanji-spelled words that have single unambiguous readings.

  • If the OP really wants to use Google translate (putting aside the horror of such a proposition) then it provides romaji output as well as Japanese. It should be possible to unambiguously convert this to hiragana. I assume the romaji has the correct readings of the kanji; I've never checked. May 7 '19 at 21:39
  • @user3856370, I doubt that Google Translate could have the correct readings for all kanji-spelled words, simply due to the polymorphic nature of Japanese words. For instance, names are sometimes impossible even for native speakers to guess at with any accuracy. Some kanji strings are also hard to figure: 食べ物 is rather unambiguously read as たべもの, but what about if we remove the okurigana and spell it as 食物? I count eight "official" readings for this in my dictionary, each with slightly different connotations. Google could take a statistics-based approach, but that won't always be right. May 7 '19 at 22:37
  • My point is that if Google is translating from the English then it already ought to know which meaning of a kanji it is intending to use so there should be no ambiguity. May 8 '19 at 6:57
  • @user3856370: "ought to" can be a dangerous assumption. :) See, for instance, the oddities described here regarding Google Translate's inability to correctly identify a kanji term. While English input and Japanese output might be more likely to produce a correct reading, I'm not sure I share your optimism at present. May 8 '19 at 16:08
  • 1
    My thinking was theoretical rather than from any optimism in Google Translate's abilities. I'm quite sure your right that it will still manage to screw things up :( May 8 '19 at 20:35

Unfortunately getting kanji readings right is not an exact science, but you can get some approximation by using the text glossing feature in WWWJDIC (and few other similar sites/apps). You'll likely quickly see missing or wrong readings for many kanji combinations.

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .