True. The current Kobo search mechanism used for Japanese dictionary is very poor.
I could not find 調 in my books to check. However I think I can explain why the results are as you have observed:
- if Kobo can find exact match to the character, e.g. 調, it will get its spelling from an index file (ちょう【調】) and using this information is able to find a proper word definition from a definiton file (ちょ.html)
- if Kobo cannot find exact match to the character (e.g. 来) but can find similar word beginning with the character we are looking for eg. 来意 (the spelling is different: らいい however) so it uses らいい【来意】 to display a word definition (using らい.html file)
- if Kobo cannot find exact match nor any other word beginning with it, it will display something odd or no definition at all. I did not figure it out yet. Eg.
- 思う -> おもう[思う] is ok but others are wrong: 思 -> し, 思い -> おもい[重い], 思って -> nothing is found
- 来る -> くる[来る] is ok but: 来 -> らいい[来意], 来た -> きた[北] are wrong.
So Kobo is using kanji AND its spelling to find a word definition. It works fine if the spelling is known and there is exact match but in other cases we can get odd results.
The big issue is with conjugated verbs IMHO, eg. Kobo can find 思う but cannot find 思います, can find 来る but cannot find 来ます, etc. Workaround would be to add all conjugations verbs to the dictionary