Relatório de pesquisa 08/08
Correlates of Rhythm in Written Texts of Brazilian and European Portuguese, Antonio Galves, Charlotte Galves, Nancy Garcia and Claudia Peixoto,
submitted April 08.
Abstract
We address the question of detecting fingerprints of rhythm
in written texts. We study texts from 20th century
Brazilian and Portuguese authors. In these texts we codify
the syllables according to whether they carry main stress or not,
and whether they are at the beginning of a phonological
word or not. Additionally, periods are also marked. Modeling the
sequences of symbols obtained with this codification as Variable
Length Markov Chains, we estimate the patterns for each
text. This probabilistic model discriminates European
Portuguese from Brazilian Portuguese. Moreover, the model obtained for
each language has a clear linguistic interpretation, as it
captures structural features which have long been conjectured in
the literature.address the question of detecting fingerprints of rhythm
in written texts. We study texts from 20th century
Brazilian and Portuguese authors. In these texts we codify
the syllables according to whether they carry main stress or not, and
whether they are at the beginning of a phonological word or not.
Additionally, periods are also marked. Modeling the sequences of
symbols obtained with this codification as Variable Length Markov
Chains, we estimate the patterns for each text. This
probabilistic model discriminates European Portuguese from
Brazilian Portuguese. Moreover, the model obtained for each
language has a clear linguistic interpretation, as it captures
structural features which have long been conjectured in the
literature.
Mathematics Subject Classifications
(2000):
Keywords:
Copy
of the file:
rp08-08.pdf
(PDF)
rp08-08.pdf.gz (gzipped PDF)
April 23, 2008
Volta ao
indíce de Relatórios de Pesquisa