2014/07/26 by Matilde Marcolli, Marcolli, Matilde
Computer Science · Mathematics · #68P30 #91F20 #Computation and Language (cs.CL) #FOS: Computer and information sciences #Information Theory (cs.IT) #cs.CL #cs.IT #math.IT #msc:68P30 #msc:91F20
paper · pdf · doi:10.48550/arxiv.1407.7169
11 pages, LaTeX
arxiv created 2014/07/26 · arxiv updated 2014/07/29
We propose an approach to Longobardi's parametric comparison method (PCM) via the theory of error-correcting codes. One associates to a collection of languages to be analyzed with the PCM a binary (or ternary) code with one code words for each language in the family and each word consisting of the binary values of the syntactic parameters of the language, with the ternary case allowing for an additional parameter state that takes into account phenomena of entailment of parameters. The code parameters of the resulting code can be compared with some classical bounds in coding theory: the asymptotic bound, the Gilbert-Varshamov bound, etc. The position of the code parameters with respect to some of these bounds provides quantitative information on the variability of syntactic parameters within and across historical-linguistic families. While computations carried out for languages belonging to the same family yield codes below the GV curve, comparisons across different historical families can give examples of isolated codes lying above the asymptotic bound.