My friend András Kornai, a fine mathematical and computational linguist who is just finishing up a very interesting advanced text in mathematical linguistics, has informed me of something that I did not know: that his Erdős number is 2. András co-authored a paper with the Hungarian mathematician Zsolt Tuza (the paper is "Narrowness, pathwidth, and their application in natural language processing", Discrete Applied Mathematics 36 (1992) 87-92; downloadable from here), and Tuza had several joint papers with Erdős (for example, Paul Erdos and Zsolt Tuza, "Rainbow Hamiltonian paths and canonically colored subgraphs in infinite complete graphs", Mathematica Pannonica 1 (1990) 5-13). This changes things with regard to what I previously said here. First, we now know there is a linguist with an Erdős number lower than Mark Liberman's. Second, this means there is a linguist with an irreducible Erdős number (András can never collaborate on a paper with Paul Erdős, who is now dead, so 2 is as low as he can go). Third, since András and I have published a joint paper ("The X-bar theory of phrase structure", Language 66 (1990) 24-50), I now know I have an Erdős number not greater than 3 (and a properly respectable one established by refereed research papers all the way through — in fact by two-author papers all the way through). And fourth, all of the roughly fifty people I have co-authored with therefore have an Erdős number not greater than 4. Who knew.
Posted by Geoffrey K. Pullum at April 15, 2007 01:35 AM