VMan = the Acapela Voice MANager, which lets you create aliases of the voices, so you can essentially copy an Acapela voice, say Heather, and change that voice's properties so you have Heather and Heather-2, which may have didn't speed and pitch properties.
As it turned out, I implemented Dmitri's idea, and then came up with an even simpler idea . . . If I parse a three character abbreviation into it's component letters, and then had it pronounced as $1 $2 $3, the NeoSpeech voices pronounced the "A" correctlyas a long "A", but Heather did not. The simpler solution was to use the Acapela dictionary editor to change Heather's pronounciation of a capital "A" to make the long A sound. Problem solved!