autocorrection does not work with a specific dataset

[gavagai]

I just worked on a .doc dataset - say A.doc - and typed in an abbreviation for an autocorrect word. Nothing happened.
So I first thought, maybe its because I started with a .doc from somebody else. So I saved the dataset in NeoOffice-format as A.swx and started again.
The same: my shortcut "ak" for "akzeptanz" doesn't work and so not any other shortcut.
Next, I created a new document and there everything is fine.
So it has to be some attribute of my A.swx but I don't see which one.
When I am working with A.swx I can look up the autocorrection dataset via the "Extras"-menu and there are all the shortcuts for often used words.
The path in NeoOffice properties is fine too. (The autocorrections works with other documents very well).
Where can I look for the missing or wrong attribution (or relation between this dataset and the autocorrection file) of that A.swx dataset?

[narf]

Please try the following:

- select the menu Tools :: Autocorrect
- under the Word Completion tab, is Enable Word Completion checked?

thanks --fran

[gavagai]

Hi fran,
I checked it and it was not enabled. I enabled it but it didn't help. I even closed the file and reopened it: no effect (but it is still enabled).
I checked with a new textfile. The word-completion is enabled by default. So I think not being enabled was a heritage from the .doc-file I have started with.
Stimulated by your proposal I looked in the same menu under "Options", where there is a list of options to en/disable for immediate change (E) and change after saving (N). There is an option "Substitution table". It is enabled for both columns (N and E in German).