As far as I know the only way to do this is to change the connection method
that Word uses to get the data to ODBC, and set up the field delimiter in a
SCHEMA.INI file in the same folder as your data file. However, the ODBC
driver has some limitations compared to the built-in text converter (e.g.
you can't have multiline text fields, and ODBC will probably limit you to
255 fields).

If you want to try that, create or edit the SCHEMA.INI file in the folder
containing your file (let's say it is called myfile.txt). You need a section
in the SCHEMA.INI file like this:

[myfile.txt]
ColNameHeader=True
Format=Delimited(|)
MaxScanRows=25
CharacterSet=ANSI

or possibly

[myfile.txt]
ColNameHeader=True
Format=Delimited(|)
MaxScanRows=25
CharacterSet=OEM

(or use the locale ID of the character set you need).

Then check Word Tools|Options|General|"Confirm conversions at open", go
through te conneciton process again, and select the (Delimited?) Text files
via ODBC option when it is offered.

Your only other option is to preprocess the text file in some way, e.g. read
it into a .mdb or link a .mdb table to it in Access. But if the ODBC
approach does not work, that will probably not be easy either.

Peter Jamieson