Alright, so for data_msg_id.dat, I sure hope those first 4 bytes are a signature. If true then nothing there needs to be changed.
And then for data_msg_txt.dat: They used UTF-8 coding. Wow, how nice of them! A little bit hard to explain:
-From personal experience you need about 3 English characters per 1 Japanese character.
-UTF-8 encodes Japanese characters with 3 bytes per character
-UTF-8 encodes English characters with 1 byte per character
So length limits should not be run into. Also they left you 8 LBAs extra for this so even if it's a little longer that's not a problem.
I made this for you if you want to use it.
https://docs.google.com/spreadsheets/d/1AxatmFfONWI7i_H_49NKcOw3qo-E56rT_VFQ0dGMZos/edit#gid=0
The basic idea is to use a human-friendly tool, then turn that back into a file the game can read when it's all done. The advantage is you don't have to type common codes, like \n or ,x. Also if you care about balancing lines, it's easier.