Automatic Discovery of Affixes by means of a Corpus: A Catalog of Spanish Affixes

Abstract

This paper describes one way in which an as complete as possible catalog of Spanish affixes can be built from a corpus. In order to accomplish this, the calculation and usefulness of different indexes associated to possible segmentations within a word (number of squares, de Kock quotients, entropy, bigram statistics and probabilities of affix occurrence) are explored. Furthermore, a general affixality index for word-segments presumed to be affixes is examined.