Many journalists seem to have the impression that the co-occurrence of certain word pairs is a good indicator of the way the culture is going, and so it may sometimes be. But one of the things the people posting on Language Log have tried to stress is that it is important to see that such claims are empirical: they could be wrong, they could be right, and you've got to do some work to determine which by looking at the way things actually are. What we are trying to fight against is this sort of thing:
It's difficult to find a piece of writing in the mainstream press which mentions the word 'bisexual' without finding that it is immediately followed by the word 'chic'.
This appeared in an article in a newsletter for bisexuals in Victoria, Australia, and was quoted by its author here.
Now, the claim is in an empirical one of a particularly clear sort. So, using Google News, we check the figures: for "bisexual", 984 hits in currently indexed news articles in English. For "bisexual chic": zero. Turning to the entire web (not just news sources), for "bisexual" we get 2,390,000 hits; and for "bisexual chic", a mere 870.
Could the author have meant to refer only to Australia? Google can check that too. In Australia we have: for "bisexual", 36,600 hits; and for "bisexual chic", 3. Only the latter figure is an overcount by 50%, because one of the hits is for the above quote itself. There are, therefore, exactly 2 hits for "bisexual chic". For a number like that, I am prepared to do an exhaustive reading of the text for all hits. One appears to be on the personal website of a creative writing student at Macquarie University. The other is a discussion of (mostly American) TV programs at Queerplanet, a gay website in Australia. Hence the number of hits at Australian mainstream press sites is zero.
The claim quoted is thus not just false but, staggeringly, overwhelmingly false. It is perhaps the falsest claim ever discussed on Language Log (though of course this is debatable; the BBC's science reporting constantly struggles to stay ahead in wild falsehoods). It is a very good example of the sort of claim we think people should stop making about language use. Difficult to find a case of "bisexual" that does not have "chic" after it is what he said. And that is utterly, outrageously untrue. These things can be checked, often in under half a second of Google time. Check them! I have spoken.
[Update 9/8/2004: Semantic Restructuring commented on this post, as an instance of "the growing reliance on google for statistical fact checks". I've added this link, at his request, because Trackback was not enabled at the time. --- MYL]
Posted by Geoffrey K. Pullum at September 6, 2004 04:59 PM