Metaphors are versatile figures of speech and thought, referring to something in terms that usually refer to something else. Through embodied and culturally situated cross-domain mappings, metaphorical expressions are prime tools for creating clarity for complex and abstract phenomena, as well as for expressing emotion and perspectives. They may also be used to hide motives and even manipulate readers.
In an increasingly digitized world, text-based “frames” of reality often tap into metaphor’s world-shaping potential – as cultural artefacts have done for millennia. The bulk of metaphorical language consists in highly conventional structures that may be revitalized through contextual cues.
To capture the cognitive and communicative potential of metaphor, it is important to analyze its rich situated context as well as large scale patterns of figurative language. In qualitative analysis, reliable identification is traditionally a challenge – but a widely accepted solution is the empirically validated Metaphor Identification Procedure VU Amsterdam (MIPVU). In quantitative analysis, vector-space models and machine learning are promising venues, but much further research is needed. My lecture provides an overview of the current state of the field in quantitative and qualitative computational metaphor identification by means of a mixed-methods case study.
Lecture Slides will appear here soon.