Automatic extraction of verb patterns from "Hauta-lanerako Euskal Hiztegia"

This paper presents some of the results obtained by means of the method we developed for the study of verb usage examples, emphasizing as we do so that the primary aim was the development of a method rather than the results per se, and dwelling on the importante of shallow syntactic patterns in obtaining the patterns of the verbs studied. We are concerned with the extraction of verb patterns from the verb entries examples of an ordinary dictionary in machine readable version. The corpus of verb usage examples that we have analysed is composed of 13.089 examples. A shallow analysis allowed us to detect the verb chains and phrasal units that appear with the verb under study. The use of an SGML (Standard Generalized Mark-up Language) data structure to represent the analysed verb entry examples facilitates the extraction of the information contained in this data structure. We present an evaluation of the basic subcategorization patterns found and the principal problems encountered in the automatic extraction of them.