Analysis of metaphors

Computer-aided analysis of metaphors in theory-language educational text corpora

Institute(s): TU Darmstadt

Partner: Institute for Linguistics and Literature Institute for Linguistics and Literature and Institute for Philosophy, TU Darmstadt

Financed with funds from the project executing agency: Dec. 2014 – Nov. 2015 as pilot project

Project collaborators

Prof. Dr. Iryna Gurevych, TU Darmstadt
Prof. Dr. Petra Gehring, TU Darmstadt
Prof. Dr. Andrea Rapp, TU Darmstadt


Text-based sciences do not yet possess any digital tools for the recognition of original metaphorics. So far, metaphor research in the history of philosophy and science, if it wants to work with large corpora, is dependent on detours for automated metaphor recognition (e.g. comparison with previously created lexica, evaluation of word neighborhoods). Context-oriented metaphor definitions are not feasible for larger, comparative metaphor studies. Manual extraction or annotation procedures are difficult to comprehend. Large corpora cannot be sensibly analyzed in this way. The aim of the project is therefore to solve this problem in interdisciplinary cooperation. The aim is to develop (a) innovative computer science methods for context-oriented recognition, (b) methods for computer philological processing and presentation, and (c) methodologically reflected methods for the philosophical-content classification of metaphor occurrences in theory-language corpora. Thus, for the first time, corpus-based (and in this sense empirically validated) statements on theoretical metaphorics are to be introduced into the research discussion in philosophy and the history of science. To this end, an exemplary corpus of four interdisciplinary relevant classics from three centuries (Kant, Hegel, Nietzsche, Weber) will be processed, evaluated and made available to research. With the implementation of the project, the new qualities to be achieved by eHumanities in basic research and application become clear.


Associated Software Products

DKPro Core

Relevant publications

  • Do Dinh, Erik-Lân; de Castilho, Richard Eckart and Gurevych, Iryna: In-tool Learning for Selective Manual Annotation in Large Corpora. In Proceedings of ACL-IJCNLP 2015 System Demonstrations, p. 13–18, 2015
  • Gehring, Petra and Gurevych, Iryna: Suchen als Methode? Zu einigen Problemen digitaler Metapherndetektion. Journal Phänomenologie – Schwerpunktthema: Metaphern als strenge Wissenschaft, vol. 41, p. 99–109, 2014.
  • Do Dinh, Erik-Lân: Automatic Identification of Novel Metaphoric Expressions. Diploma Thesis. Technische Universität Darmstadt. 2013.

Further information

Within the framework of this pilot project, the development of the annotation software CSniper was promoted. CSniper enables the selective annotation of rarely occurring linguistic phenomena, in this case metaphors. The manual annotation of metaphors was carried out in direct collaboration with Prof. Petra Gehring and Prof. Andrea Rapp. The related project “Nature and State” investigated the use of metaphors in scientific texts.