Some folks over at Pronexus recently asked the question: Will people use speech in Vista? [I think so ... but I'm probably a bit biased... ;-)]
They go onto ask for which things will people use Windows Speech Recognition in Windows Vista. For example, will people use WSR primarily to control the UI, or to perform tasks in a "Command and Control" type paradigm. Or will people use it primarily for entering large quantities of text (also known as dictation).
The foks at Pronexus aren't sure, but they hint at more of a command and control use pattern over a text input pattern.