UNIVERSIDADE ESTADUAL PAULISTA
"JÚLIO DE MESQUITA FILHO"

Faculdade de Ciências e Letras - Campus de Araraquara
  Agenda Pós-Graduação - Lingüística e Língua Portuguesa

Aluno(a) Michel Ferreira dos Reis
Titulo TEMAS E FORMAS EM HÉRACLES: um estudo baseado em keywords e treebank
Orientador(a) Profa. Dra. ANISE DE ABREU GONÇALVES D ORANGE FERREIRA
Data 23/01/2017
Resumo RESUMO
Este trabalho teve por finalidade realizar um levantamento das palavras-chaves na obra
Héracles, do tragediógrafo Eurípides, e demonstrar o uso de treebank para a anotação
morfossintática das sentenças. Para tal, a pesquisa fora baseada nos pressupostos teóricos e
metodológicos da Linguística de Corpus em relação à extração de keywords, partindo do
ponto de vista de que um texto quando comparado a outro ou a um conjunto de texto
apresenta palavras estatisticamente mais recorrentes e mais significativas dentro do texto
estudado, revelando temas e ou estilo textuais. Nesta pesquisa, para extração de keywords, o
corpus de estudo foi Héracles e o de referência foi constituído das outras dezoito obras de
Eurípides, todas em domínio público no site da Biblioteca Digital Perseu. Utilizando o
programa AntConc, configurado com o método de cálculo log-likelihood e com o limiar
crítico de 10,84, foi obtida uma lista com sessenta e quatro palavras não lematizadas, as quais
trouxeram temas importantes da peça como os personagens, a violência e o ambiente familiar
e de amizade que permeiam as ações e a loucura que domina o protagonista. Já com a
anotação sintática em árvore foi possível trazer pontos da gramática da língua grega, bem
como questões das orientações de anotação do Manual do Treebank de Dependência do Grego Antigo.
Palavras-chave: Linguística de Corpus, keywords, treebank, ensino e aprendizagem de
línguas, língua grega antiga.
ABSTRACT
This research had the purpose of performing a survey of the keywords in Héracles, by
Euripides, and to demonstrate the use of treebank for the morphosyntactic annotation of the
sentences. For this, the research was based on the theoretical and methodological assumptions
of Corpus Linguistics in relation to the extraction of keywords, starting from the point of view
that a text when compared to another one or to a set of text presents words statistically more
recurrent and more significant Within the studied text, revealing textual themes and or style.
In this research, for the extraction of keywords, the corpus of study was Heracles and the
corpus of reference was made up of the other eighteen works of Euripides, all in the public
domain on the website of the Perseus Digital Library. Using the AntConc program,
configured with the log-likelihood calculation method and the critical threshold of 10.84, a
list was obtained with sixty-four non-lemmatized words, which brought important themes of
the piece such as the characters, violence and the familiar environment and friendship that
permeate the actions and madness that dominates the protagonist. Already with the syntactic
tree annotation it was possible to bring points of the grammar of the Greek language, as wellas
questions of the Guidelines for the Syntactic Annotation of the Ancient Greek Dependency
Treebank.
Keywords: Corpus Linguistics, keywords, treebank, language learning and teaching, ancient
greek language.
Tipo Defesa-Mestrado
Texto Completo

APG 2.0
Copyright 2014 (c) UNESP - Faculdade de Ciências e Letras do Campus de Araraquara