The goal of phonostatistics is to measure the phonological divergence between languages. Simons (1977a) reviews twelve different methods of phonostatistic analysis. In a later work (Simons 1984) he suggests a thirteenth method as best for use in the computational analysis of word lists. It is this last approach that is used here.
The phonostatistic method used in WORDSURV is based on a quantity called the degrees of difference. The degrees of difference (or DD) between sounds is the number of minimal steps that would be required to change one to the other. Consider a hypothetical language with the consonant inventory below:
The lines in the diagram show the minimal steps of change between sounds in the system. The difference between two phones is defined as the least number of steps between those two phones on a chart like the one above. For example, from the above chart one would calculate the following DD values for these pairs of phones:
WORDSURV allows the user to assign DD values to corresponding pairs in the SETTINGS module. By default, identical pairs have a DD value of 0 (zero), and all other pairs are initialized with a DD value of 1. This default strategy has the advantage of requiring minimal work in entering DD values into the computer.
Several of the phonostatistic methods reviewed by Simons (1977a) calculate DD values automatically, using such strategies as comparing generative distinctive features or articulatory features. Because WORDSURV allows the user to determine DD values, it is possible to account for sound changes that, while cross-linguistically complex, might be quite simple in the particular language family being studied.
Wimbish, John S. 1989. WORDSURV: A Program for Analyzing Language Survey Word Lists, pages 60-62. Occasional Publications in Academic Computing, number 13. Dallas, TX: Summer Institute of Linguistics.