Faster ways to generate a table than comparing to the text or using relative search for Japanese? Plenty, indeed most are.
If you have some text already then why not change it? Know what one bit of text says. Leave the start and end of the stuff you know and change the middle to the start of your list of things to check. Rinse and repeat for the rest of the list of possibilities/things you want to know. This might also be the only way you learn some characters, or some types of punctuation if they are in the game but not used by the final script. You can improve this further by doing runs and patterns rather than limiting yourself to a known section and figuring out what you changed.
If you are lucky you might even be able to change something in RAM and have it be reflected on screen.
I also like name entry/high score entry here -- they are not always the same but sometimes they are and that means you can literally choose characters in real time and see what they correspond to. Most SNES era games won't do Japanese characters for such things (or will be limited to Kana if they do) but you might be lucky.
The tile stuff varies a bit by system but if it is going to be one of those games where things get loaded in at will then the order of the characters in the pictorial representation usually corresponds to the order in the encoding/table. For Japanese then you are more likely to get a few gaps where certain values are just not used but still worth looking into.
Probably not for the SNES but on filesystem based systems you can get text encodings noted in the font file. The DS NTFR font format being a good example of this.
After this you probably start to head into either probability based setups, mathematical analysis and linguistic analysis. For Japanese then does it match the order of certain existing encodings, does it match the order of a given printing of a dictionary, does it match the order of the various levels of kanji tests for that/some year... - programmers are lazy by design and are not likely to hand encode a decent selection of kanji.
After this then time for assembly type analysis. If you know where some text is located in the game then you can follow its journey from the ROM into a character on screen. That will necessarily take you through how it decodes it and thus give you the keys to the kingdom. I seldom see hackers go in for this method for text (if I am figuring out custom compression, how a damage calculation or greater part of a battle engine works, or how any number of other things work or are handled by the game then I will go for some fun here with the disassembler/debugger but for text I do usually find the other methods get it done soon enough).
I have a few others detailed in http://www.romhacking.net/forum/index.php?topic=14708
but the ones above are the main methods I would use.