Linguists know how to talk

Ben Zimmer and John McWhorter have done a diavlog hosted by Bloggingheads. If you know the names, you already know if you're interested. If you don't know the names, they're real linguists who will undoubtedly replace some of your mistaken beliefs and superstitions about language with observations that will prove to be much more interesting.

Zimmer has previously said of the word diavlog:

Diavlog is a second-order blend, by the way: it blends dialog and vlog, with the latter element representing a blend of video and blog. Or make that third-order, since blog blends Web and log.

My question has long been this: Do we distinguish, with a proper surface representation, a diavlog [dia(log)+[v(ideo)+[((we)b)+log]]] from a diavlog [dia(log)+[v(ideo)+log]] that isn't designed for the web?

And how do we know that [v] isn't just an infix, excised from video and inserted into dialog?