22 March – Kristina Despot

In the talk, the methodology, goals, and results of the Croatian Metaphor Repository project will be presented. The project involves three levels:

1. Theoretical research of conceptual metaphor, metonymy, image schemas and semantic frames in the Croatian language;

2. Computational linguistic research on metaphor in natural language processing and artificial intelligence involving developing tools for semi-automatic metaphor detection and semi-automatic extraction of linguistic metaphors;

3. Psychological experiments to further explain the nature of the links between concept meaning and perception as manifested through primary metaphors.

The applied goal of the project is the creation of a metaphorical language database ‒ MetaNet.HR ‒  which is coherent with the UC Berkeley MetaNet: A Multilingual Metaphor Repository project. The database is available on the Internet (http://ihjj.hr/metafore/). It lists conceptual metaphors and metonymies, which are decomposed into source-target relations among cognitive primitives, image schemas, and semantic frames. Metaphors are divided by type (primary, complex and entailed metaphors), by level (general, specific), and by family. Their examples from language and other modalities are also provided.

The theoretical background for the linguistic analysis of conceptual metaphors is the Neural Theory of Language and Thought and the Neural Theory of Metaphor (Feldman 2006, Lakoff 2008). Formal annotation tools are taken from the embodied construction grammar (Feldman, Dodge and Bryant 2009). Linguistic metaphors are detected using MIPVU procedure (Steen G.J., Dorst, A.G., Herrmann J.B., Kaal, A., Krennmayr T., Pasma T. 2010). The project has significant potential for further international development.