The simple answer to your question is that it would look something like
ActiveDocument.MailMerge.OpenDataSource _
Name:="T:\PRT\ELGEXC14.CSV", _
Connection:="", _
SQLStatement:="SELECT * FROM T:\PRT\ELGEXC14.CSV WHERE EGEXCODE='14'"
as long as Word recognised EGEXCODE as a text column. And you probably don't
really need the file name, e.g. you could probably get away with
ActiveDocument.MailMerge.OpenDataSource _
Name:="T:\PRT\ELGEXC14.CSV", _
Connection:="", _
SQLStatement:="SELECT * FROM X WHERE EGEXCODE='14'"
Going a bit beyond that, there are three main considerations here:
a. which version of Word?
b. do you want to do anything in the SQL statement other than specify the
field list, a fairly simple WHERE clause, and a fairly simple ORDER BY
clause?
c. do you need to use a particular connection method?
Word can open a .csv file using either
e. its internal text converter
f. ODBC
g. OLEDB (Word 2002 and later only)
When Word is using (e), the SQL you can issue is limited. With (f) and (g),
Word uses the Jet driver/provider to read the text file and can, broadly
speaking, use any of the features of Jet's dialect of SQL, including string
functions, joins, GROUP BY and HAVING clauses and so on. You may not be able
to force Word to use (e) even if you want to. And you may find it difficult
to get Word to recognise the correct delimiters with either (f) or (g) - in
the case of (f), you can specify the field delimiter in a SCHEMA.INI file.