Hi Bernd,
This is very confusing, I agree!
Let me try to explain (at least as far as I imagine what happens). It has to do with the Bouinding Boxes.
First, if you don’t use the Pivot Point, the Bounding Box of all the letters is considered and if you set the Alignment to, for insatance Top and Right, that is what is used to place the Letters:

But when you use the Pivot Point, the Bounding Boxes of the individual letters is considered. From the bottom left (the (0, 0) of the Bounding Box (just looking at the 2D version here) to wherever you set the Pivot Point. In this example Bottom Center for the A and Top Right for the C:

That Vector is than used as the Position for the Placement of the Bounding Box of the letter.. Scaling by minus 1 then becomes clear (I hope).
The strange thing is, that when you’d use a Mesh, instead of a String, Converted to Mesh, it would use the Position for the Vertices of that Mesh, so you’d have to use the Offset, instead of the Position.