Características cuantitativas de la flexión verbal del Chuj

Abstract

This paper presents the application of two word-segmentation methods  -measurement of information content or entropy (Shannon and Weaver 1949) and that of the economy of signs (de Kockand Bossaert 1974, 1978)- to a small corpus of Chuj; a Mayan language spoken in Chiapas and Guatemala. The motive for this is to determine whether or not it is possible to discover automatically some set of inflexional affixes, in spite of the small size of the corpus. Results show that information content is adequate for the discovering of about 83% (precision measure) of verbal inflexional affixes, whereas application of the economy principle requires a greater sized corpus.