Ray Girvan emailed to mention that the Henning Mankell (or was it Hanning Menkell?) thread reminds him of Sellar and Yeatman's "1066 and All That",
"where the authors deliberately play on the names of the pretenders during the reign of Henry VII, Lambert Simnel and Perkin Warbeck. (They refer to them variously as Warmnel, Perbeck, Wimneck, Warmneck, Lamkin, Lamnel, Simkin, Permnel, etc)."
Note that this is exponentially worse than the Manning Henkel problem, since there are not two but four dissyllables to conjure with.
The outlines of a Henning Mankell experimental paradigm are beginning to emerge -- present one or more reference names, and then (after a delay and perhaps some distraction) ask subjects whether each of a set of probe names was in the reference set. The aim would be to predict error rates and reaction times, based on a model of the structure of the morphophonemic subspaces involved.
In fact, there's probably already a relevant literature on this... In any case, I'll bet that such phenomena turn out to exhibit the same lack of sequential independence that was demonstrated repeatedly here (by google counting methods) for spelling variation in words like "emperor", "jennifer", and "attila".
Posted by Mark Liberman at April 22, 2004 03:52 PM