Trouble with Vista Voice Recognition and MFC-based app.

I have an MFC-based application that i'm trying to use with Vista's built in voice recognition feature. The platform that the application has been built with is Visual Studio 2005, using Windows XP.

I have a simple textbox, of MFC type CEdit, that i'm using to extract the output text provided by Vista Voice Recognition. However, the results are not so good. After speaking a sentence using VR(for example, "this is a test"), the results i receive from the event handler of my CEdit textbox is "thi" and then "s". In other words, for a word of length x i get the first x-1 letters of the very first word, and then the last letter of that word, and then all other words get omitted. I do not run into this problem when using other kinds of voice recognition software, such as Dragon.

Should i be using a different type of MFC structure? Is there a special SDK i'm supposed to be using to get this to work correctly for Vista?

I could really use some guidance for this. If anyone can offer some help i would really appreciate it.

I would ask on MSDN since it's MS sw development using VS tools.

Well, maybe you could answer this...how exactly does Vista VR output the text? One character at a time, or in words? I've seen it work fine using notepad, but i would certainly be curious as to how the output is being done exactly.