Inspired by Geoff Pullum, Håkan Kjellerstrand at hakank.blogg has written an icon program to generate plausible variants of "Henning Mankell" and compare them with the list in Geoff's original post. Kjellerstand is also the author of the perl module MakeRegex, which "composes a regular expression from a list of words", based on common prefixes. I've been hoping, though, that someone will follow up on David Beaver's post by writing a program to help with (various approaches to) estimating the statistical density of what David called "the Henning Mankell morpheme space".
This is a serious issue in psycholinguistics, as should be clear from reading what Geoff and David wrote. More on it later, maybe.
Posted by Mark Liberman at April 22, 2004 10:10 AM