> And is there not strong evidence (cited by Maynard Smith
> and Szathmary) that there is a gene that affects it?
No, apparently not. Elman, et al., discuss this on pp. 372-8. Smith/Szathermary is based on work by Gopnik and Crago. This was eventually picked up and bandied about by Pinker as evidence for the existence of a language gene. Another study of the same family (Vargha-Khadem, F. & Passingham, R. 1990. Speech and Language Defects. Nature, 346, 226) reveals that the impairments are not specific to verbal behavior or to particular "morphological categories".
"The disparity between the Vargha-Khadem et al., and the Gopnik and Crago findings is explained by the fact that the former are based on a much larger set of experimental morphological contrasts than the original four past-tense items used by Gopnik and her colleagues: Two regulars (kissed and walked) and two irregualars (went and was). Yet these subsequent clarifications about the broad array of deficits actually present have gone largely unnoticed in public discussions of the KE family." - Elman, et al. p 376
-- bill