Although it's handy for other purposes, having your computer's language set to japanese won't affect the displayed language in the char field of a hex editor.
The way these display characters is by using each byte as an index in a map of characters. Displaying multibyte encodings is much more involved, which is why many hex editors don't try or only permit it as a special exception. The difficulty arises because plain bytes offer no reference as to when a character may start or end, or if it's even a character. In fact, you could have a valid shiftJIS string that would appear as nonsense just because it was preceeded by random data that happened to make it
look like it was the start of a char.
The big problem is that text and binary are probably muddled together. There's the blind luck method of poking around different files hoping you see something resembling text, or tracing loaded files. You could also use something like
xsearch to search for the hex version of a string you know is present in all the files in a directory. Unless they did something
fun, it would be either shift-jis or one of the unicodes. 3-4 searches.