Fri, 10:00–11:30, JOTA series: P19 (Faculty of Computer Science)
(slides)
How do we trace the paths of meaning as it travels across languages? A storm of protest is vivid in English — but does the same metaphorical force of a storm carry over into Japanese or Zulu? Patterns of metaphor and metonymy may be widespread, yet their specific mappings reveal the distinct cognitive and cultural contours of each language. To study these paths, we need more than isolated examples; we need multilingual maps. Wordnets, with their networks of synsets and semantic relations, offer a kind of semantic cartography, allowing us to explore how meanings shift, radiate, and connect.
In this talk, I introduce recent work on modelling metaphorical and metonymic extensions in lexical networks, focusing on ChainNet, and show how these patterns can be examined both within and across languages. This multilingual perspective is enabled by the Open Multilingual Wordnet, which brings together lexical data from many languages within a shared semantic framework and allows comparison of meaning structures at scale. I discuss both its strengths — such as the ability to capture fine-grained distinctions — and its limitations, including uneven coverage and structural inconsistencies across languages.
Beyond the analysis of sense relations, wordnets also provide a practical framework for organising and linking lexical knowledge. Resources such as the Open Multilingual Wordnet support applications ranging from cross-linguistic comparison to the development of digital dictionaries and learning materials. While they form only part of a broader ecosystem, such networks can contribute to language documentation, education, and efforts to make lexical resources more accessible to both researchers and learners. By modelling how metonymy travels and metaphor wanders, we begin to map not just words, but the deeper connections between language, culture, and how we make sense of the world.
Note: This is work done with Rowan Hall Maudslay.
Talk in the JOTA series of occasional language technology lectures at the University of Ljubljana, 10:00–11:30.