Well, even assuming the Big 5 approach is at all close to the mark, dichotomies could still exist. Each data point still falls to one side of the bell curve or the other.On the other hand, it has also been demonstrated that the dichotomies don't exist; traits fall on a bell curve.
This is interpreted to mean that the dichotomous types cannot exist.
I do believe a normal distribution to be accurate for this sort of thing.... for many people to be in the middle and not at the extremes of the preferences. Even cognitive type itself is often not that expressed, for the majority of people just a slight flavoring. But it's still a distinct flavoring.