You've got to feel sorry for the bloke: not only has he spent his life being misquoted, but nobody bothers to credit him for the research they haven't read!
Oh, I'm sure
somebody has read it!
In fairness, I suspect a number of people have performed similar such studies and research. From my quick scan of your link, Mehrabian's numbers related to a particular context (where people already knew each other?), so that leaves a wealth of other situations and contexts to look at. And I'm sure people from different disciplines have looked at it too, each from their own stand point.
Let us not stray too far from the point though. That being that an amount of understanding comes from the other than the words. It's probably daft (bordering on meaningless!) to attempt to put a percentage on that amount since it probably varies so greatly from context to context. I'd imagine, for instance, that in any legal context that the word is more concrete. In general conversation, less so.
It's actually quite an interesting exercise to record a conversation and listen back to it
very carefully. You might be surprised at the number of 'errors' people make, either in word choice or grammar. As a listener in context, you kind of just auto-correct. Sometimes filling in with what you
think they said rather than what they
wanted to say. Fiendishly tricky, this conversation stuff. It's a wonder we understand each other at all!
VERY colloquial!! 
