Abstract
This study aimed to analyze the extended application of the LX-Parser, a syntatic parser, in a corpus composed of the initial passage from Peregrinação (published in 1614) written by Fernão Mendes Pinto (ca. 1510-1583). Manual and automatic extraction of noun phrases from the first ten chapters of the work were carried out. The hypothesis that the specificities of old texts limit the accuracy of the results generated by the considered parser was tested. The hypothesis was confirmed, since the results of this extended application did not prove to be productive due to the high frequency of problems in the produced analysis. It was identified that the main problems related to old texts are related to the issue of sentece extension, spelling and linguistic variation and change. In addition, there were also problems that are not specific to old texts, but still limited performance: the issues of structural ambiguity and linguistic categories.
Keywords:
Technology; Computational Linguistics; Historical Linguistics; Syntax