While I think that many of these questions are interesting, I believe that their sample is flawed. For example, there are many terms that are commonly mispronounced except by the cognoscenti (read ``superior jerks'') that are overwhelmingly shown as pronounced correctly in the survey.
There are many others as well. For example, aunt (ont vs. ant) is probably pronounced ont by about 20% of the population around here -- for whatever reason, the black folks. But it doesn't really show that in their graph.
I think that their sampling is flawed in a very similar way to the ``Dewey beats Truman'' polls were flawed. They're only sampling folks that are interested in this in the first place, which is liable to attract people that are hyper-correct. Not exactly a random poll.
Previous Topic
Index
