I can't recall if V6 has the Address Block feature - if it has I would use that or look at the alternaitves for automated address processing.
Alternatively if the address line is always consistently used in the original input, you can use the SPLIT functions splitting on the comma. You may also want to use the some form of TRIM function as part of the calculation for improved control of left and right aligment in the data fields.
I can;t improve on the Help information so see what you can glean from that and if it leaves questions come back and we can see what do to resilve them.
AFAIK there's an offer on to upgrade to v10 for users with old versions - so first of all I'd recommend you do that so you have not only address blocks, but the floating trap, the ability to parse HTML, PDF and XPS files, support for Excel 2007 and tons of other useful stuff!
State= rsplit(field;3;", ";1)
Town= rsplit(field;3;", ";2)
Street= rsplit(field;3;", ";3)
would do the job on your sample, but your data may vary...