All of the character lines are in a file named, e.g. Hatena.xml in Apricot470_x86/Characters/Hatena .
The format is pretty self-explanatory.
You would just need to translate the text in each <message>text goes here</message> block.
(The text editor needs to be able to grok UTF-8 encoding and Japanese text.)
Tick sequences are run at a particular time of day, month, year, etc.
The Activate sequences are the ones where the character is just chatting. If there's more than one, the script chooses one randomly when it generates an Activate event. At first, I would cut this portion of the file down to just one sequence and make it a no-op, ie
<sequence name="Activate"> </sequence>
which will eliminate the chat. Then, as you had time, you could either translate the lines or make up new ones.
There is also a words.xml file which contains words tagged with attributes - how to edit this is also self-explanatory. These are the words which are mined out of the feeds defined in Subscriptions.opml.
Comment too long. Click here
to view the full text.