That's a possibility.

I must admit that I don't entirely understand it enough to implement it (the OP said "I don't believe that the algorithm for this is a
whole lot more complicated"), and I don't have a need for it myself, but if someone would like to provide some code for it, even if it's in the form of a function written in Python:

    def at_default_word_boundary(text, pos):
        ...

then I'll see what I can do! :-)