Academia.eduAcademia.edu

Aelius Falado 1.0

2014, Aelius Falado 1.0

Abstract

This project is included in the area of Corpus Linguistics and morphological analysis of Brazilian Portuguese, which is about automatic morphological annotation corpus of spoken language. The work aims to contribute to the improvement of the automatic morphosyntactic tagger Aelius, originally developed by prof. Dr. Leonel Alencar, from Universidade Federal do Ceará, coordinator of the project CompLin - Natural Language and Computation. The tags used in Aelius are the same tags of Tycho-Brahe historical Portuguese corpus (electronic corpus already tagged, composed of texts in Portuguese written by authors born between 1380 and 1845). The texts that were used for tagging come from the bank of Varsul project, which studies the linguistic variation in southern Brazil. We worked with automatic tag of 20 texts, containing 154.530 words. This project presents the tool for automatic tagging, the automatic morphosyntactic analysis process executed by the tagger and our auto-tagging manual review work, besides some suggestions of improvement to deal specifically with aspects of orality. We will verify the Aelius’ performance in automatic morphosyntactic tagging of a corpus of spoken language, an unprecedented challenge to this tool. From the errors of the tagger, we seek to infer certain patterns of annotation to overcome limitations presented by the program, proposing some suggestions of implementations so that Aelius can tag even more satisfactorily corpus of spoken language. We will deal especially with cases of interjections, apheresis, onomatopoeia and conversational markers.