ESSV Konferenz Elektronische Sprachsignalverarbeitung

Title: How to identify elliptical poems within a digital corpus of auditory poetry

Authors: Hussein Hussein, Burkhard Meyer-Sickendiek, Timo Baumann


Ellipses denote the omission of one or more grammatically necessaryphrases. In this paper, we will demonstrate how to identify such ellipses as arhythmical pattern in modern and postmodern free verse poetry by using data fromlyriklinewhich contain the corresponding audio recording of each poem as spokenby the original author. We present a feature engineering approach based on literaryanalysis as well as a neural networks based approach for the identification of ellipseswithin the lines of a poem. A contrast class to the ellipsis is defined from poemsconsisting of complete and correct sentences. The feature-based approach usedfeatures derived from a parser such as verb, comma, and sentence ending punctuation.The classifier of neural networks is trained on the line level to integrate the textualinformation, the spoken recitation, and the pause information between lines, and tointegrate information across the lines within the poem. A statistic analysis of poet'sgender showed that65%of all elliptical poems were written by female poets. Thebest results, calculated by the weighted F-measure, for the classification of ellipsiswith the contrast class is0.94with the neural networks based approach. The bestresults for classification of elliptical lines is 0.62 with the feature-based approach.

Year: 2019
In session: Poster und Demonstrationen
Pages: 247 to 254