19 April – Marianna Bolognesi & Laura Aina
In this talk we will provide an overview of the current status of the COGVIM project (n° 629076 – Project Acronym: COGVIM; Call identifier FP7-PEOPLE-2013-IEF) and we will present the results obtained in WP2 and WP3 ( www.cogvim.org ).
The rationale between the COGVIM project is that the similarity between source and target domains in metaphors can be operationalized in terms of distributional similarity.
In particular, Laura Aina, MA student in Logic and Computation, will present the results of the distributional analysis that she performed on a dataset of verbal and visual metaphors expressed in A-is-B form (the COGVIM materials). The analysis has been conducted using Distributional memory (Baroni, Lenci 2010), which is a corpus-based distributional method that allows to retrieve semantic similarity between two words, based on their co-occurrences across corpora of text. The results show that Distributional memory seems to capture better the similarity between source and target domains of verbally expressed metaphors.
Dr. Bolognesi, COGVIM project coordinator, will present the results of a distributional analysis based on the same materials (i.e. the same A-is-B verbal and visual metaphors) but performed with FDT (Bolognesi 2014; 2016), which is a distributional method based on Flickr annotated images. The results show that FDT seems to capture better the similarity between source and target domains of visually expressed metaphors.
The results of the two analyses will be discussed.
Potgieterszaal, University Library
1012 WP Amsterdam