October 17, 2007

The Islamic language family

One small point went unnoticed in Sally Thomason's very sharp two-part critique (here and here) of Tecumseh Fitch's recent short article in Nature. It was spotted by my sharp-eyed Edinburgh colleague Bob Ladd. The artwork accompanying Fitch's article depicts the tree of Indo-European language relatednesses. And the branch leading toward such languages as Russian is labeled ISLAMIC.

It is possible that Fitch never saw this at the proof stage, and will learn about it right here on Language Log. In my experience of writing for Nature, proofs of commissioned commentaries and the like are sent by fax or PDF at a stage where the accompanying decorative artwork isn't necessarily in final form when the author checks the text. If I recall correctly, when Barbara Scholz and I read the proofs of our short piece ‘Language: More than words’ for publication (in Nature 413, issue no. 6854, 27 September 2001, page 367), we hadn't seen the picture that was to appear with it.

Bob and I both felt that the most likely explanation for the slip here (there is of course no such language family as "Islamic", anywhere in the world) does not lie in sheer ignorance. (We don't think anybody believes Russian is in a family of Islamic languages, except for those who have come to believe it in the last few days because they read it in a diagram in Nature, and those who have made a totally unjustifiable leap of inference from President Putin's current visit to the Islamic Republic of Iran.) Our guess would be that (a) somebody's handwriting was misread by an artist who was just copying lettering and needed a second cup of coffee and was not thinking about plausible historical relationships between languages, and crucially, (b) the direction of the slip was reinforced by the very high frequency of the word "Islamic" relative to what was actually intended, namely "Slavic".

Frequency commonly affects the direction that error takes; you may recall my speculation about the case I discussed in "Hammer, jammer, slammer, stammer, grammar". To check the guess about frequency in this case, I looked at the number of pages found by Google News (UK) for searches on the two relevant words. Slavic: 247. Islamic: 52,561. The defence rests. The slip was probably a frequency-reinforced error in handwriting recognition.

Perhaps, though, it would still be a good idea for Nature to print a correction.

Posted by Geoffrey K. Pullum at October 17, 2007 07:11 AM