Wolfgang - I can understand that. And if the article had said what you have, I wouldn't have reacted the way I did.

But to have them go through the list of "changes" and then find that they are not even audible without computer analysis makes me wonder what standards are being applied here and how many samples were taken and what kind of statistical analysis was used? Are the sound samples from the '50s adaquate for the kind of discrimination they are attempting? I do truly wonder whether or not these are real changes or whether they massaged the numbers in order to produce the results they wanted.

And I agree that a change does not have to be perceptible in order to be real; however, under ordinary circumstances I doubt there are many people who would say a vowel shift has occurred unless it CAN be percieved audibly. [Voice recognition software and hardware is a special case scenario.]