Sometimes it is useful to get the "guts" of a vocabulary while being willing to give up other information.
The following procedure has proven to be effective in a few cases. There are no guarantees!
1. have a good user established.
2. find the directory for the vocabulary to be recovered. It will generally be:
(7.0) C:\Program Files\ScanSoft\NaturallySpeaking\Users\<your name>\current\xxxx
(7.0) C:\Documents and Settings\All Users\Application Data\ScanSoft\NaturallySpeaking\Users\<your name>\current\xxxx
where xxxx is the folder with the base language model -- something like General_ or Legal or Medical or Cardiol_. The directory will have 4 files in it -- general.voc, dd10voc1.voc, dd10voc2.voc, dd10voc3.voc.
3. Make a copy of this directory elsewhere on your system -- the desktop is fine, as these files will be very temporary.
4. Navigate to the copy of the directory using Explorer. Then RENAME all 4 files as follows:
general.voc->general.top
dd10voc1.voc->general.to1
dd10voc2.voc->general.to2
dd10voc3.voc->general.to3
Key is that all 4 files have the same name but different extensions.
5. Open the "good" DNS user.
6. NaturallySpeaking, Manage Vocabularies, IMPORT... and select the general.top file from step 4.
It will ask you what vocabulary to merge into -- most of the time, you will be best off specifying a NEW vocabulary which you can name as you please.
7. Open the new vocabulary -- you should see your custom words.
top