Sometimes I’m kind of dissappointed with the quality of the word definitions that come up for target language Japanese, host language English. For example:
イケメン (ikemen) → twink
Where are you getting this definition?
The EDict dictionary file I use in my own software has “(n) (uk) good-looking guy/handsome man/hunk/Adonis/cool guy/”
Google Translate has “handsome”.
It’s nice that you can recognize some conjugations/inflections and context, but when the word is in dictionary form, I think you can do a lot better. I think just using Google translate would be better than the current mechanism.
I consume a lot of Japanese language content via the combination of Netflix and Language Reactor, and I share your concern about the quality of the built-in hover dictionary. They have just returned the ability to turn it off.
For better quality Japanese->English hover dictionary I have happily used Yomichan extension, and now 10ten Japanese reader extension, in addition to the Language Reactor extension.
But agreed, it would be nice if LR could directly make use of EDICT, KANJIDIC, etc.
On the hand LR is not specifically just for learning Japanese
Yes, I agree Korean also has issues with would definition quality. I had to turn off my hover dictionary for Korean as well.
It’s on their roadmap—but it hasn’t been updated in a while:
I know with any languages that are more context-based/have many variations, endings, compound words, etc., the LR team is still figuring out how to improve them.
I think English and other non-Asian language TL learners have fewer issues with dictionary definitions than learners who are learning Korean, Japanese, etc.