It is a fair representation... the article states: Spaepen's experiments did not reveal which component of language is crucial to developing an accurate number

It is a fair representation... the article states: Spaepen's experiments did not reveal which component of language is crucial to developing an accurate number sense. However, she suspects that it is the "count list" – the familiar sequence of numbers that every speaking child learns early on. Children learn this count list well before they actually understand that "four" refers to four objects rather than three or six, says Michael Frank at Stanford University in California. So, being taught to count helps with counting more accurately (duh). I'm not arguing against the thesis of this study, mind you. I believe that language is of fundamental importance to thinking, and certainly, if you're never taught the word-concepts for numbers and counting, you are unlikely to be good at counting things. However, to prove conclusively that accurate counting emerges out of language (rather than out of being taught how to count), someone would have to devise a method of teaching counting that does not involve language and test that against the normal way. The experiment described in this article proves little. Of course, such an experiment is effectively impossible today, but just because we can't do it (yet) doesn't mean we should accept poor substitutes...