Benefits of a Combined RNN and Dictionary-based Lemmatization: The Three Armenian Editions of the Apocalypse of John as a Test Case

Van Elverdinghe, Emmanuel
(2023) Digital Caucasiology – a Change of Paradigm? — Location: Frankfurt am Main, Goethe-Universität Frankfurt (4.October.2023)

Files

VortragFrankfurt_231005.pdf
  • Closed Access
  • Adobe PDF
  • 3.49 MB
Request a copy

Details

Authors
Abstract
The methodology of lemmatization in the GREgORI Project combines two procedures: a traditional dictionary-based approach, relying on manually verified lemmatized and POS-tagged forms from previous corpora, and an analysis by a recurrent neural network model, which produces predictions based on the same dataset. This paper illustrates the differences between these two approaches and demonstrates the complementarity of their outcomes, using as a test case the Armenian text of the Apocalypse of John, accessible under three recensions in three (more or less unsatisfactory) editions. Once lemmatized, these texts can be compared with each other and with the Greek originals more easily and accurately, paving the way for a study of their textual history and translation technique.
Affiliations

Citations

  • APA
  • Chicago
  • FWB

Van Elverdinghe, E. (2023). Benefits of a Combined RNN and Dictionary-based Lemmatization: The Three Armenian Editions of the Apocalypse of John as a Test Case. Digital Caucasiology – a Change of Paradigm?, Frankfurt am Main, Goethe-Universität Frankfurt. https://hdl.handle.net/2078.5/101363