If I click on one of the English words used to define a kanji character (usually by accident), renshuu tries to look that word up in the dictionary. This used to work, but today it converted the word to romaji first, breaking up the result into nonsense. Is there a hidden setting that controls this behavior? I know that automatic conversion is useful for beginners, but I don’t want to use romaji input ever.
Edit: this apparently only happens for words not in the dictionary. I’d still like the option to disable romaji input, though.
There is no available setting. It will, of course, search for it in English first, but if a)there are no results for the English (which is primary), it does try to read it as romaji and see what it can do with it.
There is probably a level at which it can "grade" the results and say "this probably isn't what they want", but that's a not-so-simple thing to fix. As long as there is at least one English result, though, you'll never reach this page.