No, I'm not a code breaker, but I have done localization for quite some time and have seen this in the past. 表示 turned into 沸.

You might run into more "odd", out-of context characters, and so my advice is to contact your client immediately and alert them that the some Japanese characters are garbled.

I said it "in the past" because this was the case with some compilers, but it should have been fixed for some time. I cannot tell you when this happened, but if you are looking at Trados Workbench or Multiterm, the chance is that this corruption happened during the alignment stage. Or, the Japaese texts were handled in a non-Japanese environment.

Here is a bit of technical background, if you are interested:

It only happens with certain Kanji whose trail byte (Kanji is double-byte, consisting of a lead byte and a trail byte) is 5C (in hex). 5c is Backslash in ASCII, and the parser thinks it is a backslash , which is an escape character in some environments, and skips reading the very next byte (the lead byte of the next character). So, 沸 is the result of

lead byte of 表 + trail byte of 示.

Some of the Kanji that cause the same problem are:

ソ、能、表、予、十、貼、構

And how they become depends on what follows these kanji.

Good Luck!

--------------------------------------------------
Note added at 6 hrs (2008-03-13 15:18:24 GMT) Post-grading
--------------------------------------------------

Hi Corinne,

Both Kathy's and your comments let me further -- This problem still happens easily with web pages or html/xml based help files. If possible, check with your client if the encoding specification is set properly. good luck!