>Brosnahan, who noted that certain linguistic phenomena like
> aspiration of initial voiceless occlusives, affrication androunded
> front vowels showed a high correlation with blood groups.
Since blood groups are not even spread around the world, but are clustered
and regional, and since linguistic features are also not evenly spread, but
clustered and regional, it is therefore to be expected that there will a
"high correlation" between at least some blood groups, and some linguistic
features. Any such "high correlation" is therefore statistically
insignificant.
This seems at first sight to be another example of the kind of wrong
thinking that says (for example) the chances of the car in front of you
having exactly that number plate, and no other, are millions to one;
therefore it is statistically significant that that particular car is there.
This is nonsense.
Peter