High-Speed Computation of Lexicostatistical Indices

The purpose of this paper is to bring to the attention of interested persons the existence and availability of a high-speed computer program designed to aid in the determination of lexicostatistical indices, of the type developed by Swadesh, for assessing the degree to which any two given languages are cognate. The program was written by Carroll for the IBM Electronic Data Processing Machine Type 704 (with a 32,768-word magnetic-core storage) at the suggestion of Dyen, who wished to compute lexicostatistical indices among several hundred Malayopolynesian languages. The need for a high-speed computer program is suggested by the fact that the number of possible different pairwise comparisons among n languages increases geometrically with the number of languages; specifically, it is n ( n - l)/2. If the comparisons are based on, let us say, the words in each language corresponding to the 200 ‘meanings’ of Swadesh's standard list, the analysis of the data for 100 languages involves a total of nearly a million separate acts of comparison. The program which is now available enables the performance of this job in approximately one hour and a quarter—provided, of course, that the data have all been properly assembled and coded for input to the machine. The individual acts of comparison (that is, the comparison of the forms for a given meaning in two languages) are performed at the rate of about 300 per second for typical kinds of data

Algorithm · Computation · Computer Science · Lexicography and Language Studies · Linguistics, Language Diversity, and Identity · Second Language Acquisition and Learning

• An Indoeuropean Classification

  Isidore Dyen, Joseph B Kruskal et al.•Transactions of the American…•1992

  Add to list

• A Control Case for Glottochronology1

  Open Access•A Richard Diebold•American Anthropologist•1964

  Add to list