I would not be surprised if math kids were more easily ID'd in that particular study, as it used multiple-choice standardized testing data, which, although it had measures of reading comprehension, didn't include measures of exceptional written language gifts at all (outside of mechanics and grammar). Also, it's much more difficult to objectively assess expressive language in general. So learners with exceptional verbal comprehension were somewhat sampled, but not those with exceptional and original language creation gifts.
Also, I think the pools weren't the same size (i.e., the reach of the SMPY math talent search, vs the CTY verbal talent search).
But there is other research that seems to suggest that high math girls are more likely to be also high verbal, while high math boys are not, which one could imagine might mean that high math is more common than high verbal (granted these are not quite as high overall, but still a relatively selected pool):
-about why women with math ability may choose non-STEM careers (short version--more choices). Includes the observation that high math women are often also high verbal, while high math men often aren't.
http://www.news.pitt.edu/women_STEMhttp://www.cds.web.unc.edu/files/2014/10/not_lack_of_ability.pdf
