The dictionary can be produced using aspell for Unix/Mac OS X or under Cygwin. You need to have aspell and appropriate dictionary installed.
Here is the example for Russian dictionary I've used:
aspell --lang ru-yeyo dump master | aspell --lang ru expand | tr ' ' '\n' > russian.dic
For German it would be:
aspell --lang de dump master | aspell --lang de expand | tr ' ' '\n' > de.dic
At the moment IDEA reads the dictionary in the system default encoding, so you may have issues if it differs from the locale of the dictionary, however next IDEA update will read the dictionary in UTF-8. When it happens, you can use iconv for conversion.
If aspell produces the directory in UTF-8 on your system, you will need to convert it into cp1252 for the current IDEA version:
iconv -f utf-8 -t cp1252 de.dic > de-cp1252.dic
However, if you build it on the same system where you run IDEA, the encoding should match (as aspell will produce it in system default encoding and IDEA will also read it in system default encoding). I used Cygwin on Windows with cp1251 encoding and it worked fine for Russian, but would be hard for German on the same system until IDEA reads it in UTF-8.