Haim's quip about ten irregular names is a one sentence strawman. I'm not going to recite at length from Galdwell's book, or other related material. I'll just repeat one standard "example" that's been bandied about and seems to get at least some people's attention. This example certainly does not summarize all the various musings about language differences in arithmetic.
Example: suppose instead of twenty-three we learn to say two-tens-three, and instead of thirty-five we say three-tens-two. Now, what is three-tens-five plus two-tens-three?
Some people feel that this allow the language processing brain to line up better, (an EE type might say with a better "impedance" match), to the math at hand.
(All the obvious criticisms about additions with carries etc. are here ignored.)
The question this example raises for me is, does this kind of language shift allow a child to "get it" faster? Does it facilitate in an entirely natural way some of the conceptual work that is often addressed more laboriously by those pictures you (R.H.) disdain so much?
Some people think the answer is an obvious "yes" or "no". I think its worth an investigation.