January 27, 2007

Scrambling in internet folklore

Back at the dawn of modern time, when LLP still had that new plaza smell, Mark Liberman examined a widely circulated item about what was said to be research at Cambridge University on the comprehensibility of text in which letters inside words had been scrambled, leaving the first and last letters in place.  The claim was that such scrambled text was astonishingly comprehensible: TIHS IS AZANMIG!

Apparently, nothing dies on the internet.  Things propagate and then retreat, but are always ready to revive in force.  We might be entering a resurgent phase of the Cambridge Scrambling Tale; my e-mail suggests that after three or four years in hibernation it's awake and abroad again.

So if you've recently gotten one of the versions of this internet folktale, go back and look at Mark's 2003 posting and at the piece by Matt Davis (which Mark cited), especially at its "update 2" section, where Davis looks at relevant psycholinguistic research and notes that the material in the mailings seems to have been carefully chosen to be comprehensible.

I have little to add to Davis's discussion, except to note that (as I wrote to a friend at the time):

... lots of the versions have typos in them!  For example, the second word in your version [which began: "The phaomnnehil pweor of the hmuan mnid"] is "phaomnnehil", which lacks one "e" and has an extra "h".  It looks like someone was doing the letter transpositions by hand, rather than using a random-transposition scheme, which is what any actual researcher would do.

The second sentence had "rscheearch" (with an extra "ch") and "iprmoetnt" (with "e" instead of "a").  That's three erroneous words in the first 28 words, and most of those 28 words were little ones.  At that point I gave up checking the text.

zwicky at-sign csli period stanford period edu

Posted by Arnold Zwicky at January 27, 2007 02:08 PM