Sounds a bit too straighforward to me. I think I've experienced more variance in the degree of stuttering regardless of whether or not a sentence was "set" or "changeable". To me, at least, context is a _factor_, but it's never a _consistent_ factor, if you know what I mean. Hypotheses that seek to unify and "streamline" what stuttering is supposed to be, seems to me to run the risk of not actually spanning our total experience with this variance.