A tribe in Brazil doesn't have words for numbers above 3, according to recent news reports. (I've heard the results dicussed on NPR's "Science Friday" and seen commentary on Language Log) Neither does the tribe engage in counting activity. It would seem, though, that these two facts are simply two perspectives on the same basic fact: they don't count. If they counted, they would have the words to do so. If they had the words, it would mean that they counted. It doesn't tell us anything very significant about language per se. It is not that the absence of numbers determines the poor mathematical skills of this people.

Imagine a tribe that didn't cook, ate all its food raw (now that we're reviving "primitive" stereotypes.) They wouldn't have a word for "medium-rare" either. And, astoundingly, they make remarkably poor "cooks," but not because of any purely linguistic deficiency. That's what I think Mark Liberman is getting at in his post on this topic: it's not that the absence of words prevent these people from counting, but that they simply don't count. From which I draw the conclusion that all the resources that exist in a language must enter into use. Imagine the contrary case: "Yes, Mr. Anthropologist, we have a quite elaborate number system, but we never actually use it." That would not be plausible, I submit.