I'm sure I've seen a different article about it, though, because I've heard of a variation on the experiment where the subjects were people who had learned a non-English language as children, but then moved to America and had spoken English for 10+ years. Even though the entire questionnaire was given in English, people associated a physical object with attributes of its gender in the first language they learned as children.
For the remark about word genders, all I could find today was http://www.nytimes.com/2010/08/29/magazine/29language-t.html... which doesn't include sources.
I'm sure I've seen a different article about it, though, because I've heard of a variation on the experiment where the subjects were people who had learned a non-English language as children, but then moved to America and had spoken English for 10+ years. Even though the entire questionnaire was given in English, people associated a physical object with attributes of its gender in the first language they learned as children.