If i understadn you correctly you want to bypass just some of the words in your current dictionary and not all of them? 

That is correct.

Admission. A colleague here + myself had been trying a test policy with test filter as follows
Condition
body-contains("blacklisted word")
attachment-contains("blacklisted word")
Action
deliver()
The policy executes this filter instead of the Profanity and also has other Content Filters checked on. The policy was always higher than Default.
When I posted initially I believed we had this test filter higher than the default profanity filter in the new policy but not so, or possibly we hadn't the 2nd condition "Committed". Profanity had been unchecked. On moving it above Profanity it does now allow mail out with the desired word whilst blocking the rest of the word list and applying the other required filters.

Despite all of the above whilst it is working in the above form would the new dictionary applied to the group be a better method (less invasive on resources,execution time etc) for tackling this.

Once again many thanks

I think it is one of those things that comes down to prefferance and ease of management.

We have a number of our content filters that we do not want applied to emails going to specific groups of people.

To do this, we have create a number of new Mail Policies that use the Default Anti-Spam, Anti-virus and VOF settings. we have then excluded the content filters e do not want applied to these messages.

Membership of this Mail Policies is managed by an LDAP group. So instead of having to make any changes to the boxes (and go through our Change Process), we can just add and removed people from the various LDAP groups.

Thanks for your advise in this