This study aims at modeling the semantic similarity between metaphor terms by means ofa distributional method based on a Big Data stream: Flickr tags. As explained in the article,this distributional model, Flickr Distributional Tagspace (FDT), captures primarily relationalsimilarity between concept pairs, that is, between tags that appear in similar tagsets(and therefore in similar pictures). A long established view in metaphor theory claims thatmetaphors pertain to the conceptual dimension of meaning, but while different modelsaim at explaining how language constructs and represents metaphorical conceptualstructures, we still know very little about how other modalities (for example, images)achieve metaphor construction and expression. A comprehensive theory, which argues infavor of the conceptual nature of metaphor, cannot afford to be biased toward the analysisand modeling of one specific modality of expression, thus neglecting potential modalityspecificdifferences. The present study, conducted through FDT, found that visual andlinguistic metaphors behave differently, in that the similarity between two aligned conceptsin a visual metaphor appears to be significantly higher than the similarity between twoconcepts aligned in a linguistic metaphor (which, in turn, does not differ substantiallyfrom the similarity between two randomly paired concepts). These findings suggest thatthe relational similarity between two metaphor terms (captured and modeled throughFDT) is crucial for visual metaphors but not for linguistic metaphors. An additional contentanalysis, also reported here, shows that the type of semantic information encoded in therelated tags (i.e., the contexts on which the contingency matrices of this distributionalmethod are built) differs, in relation to the modality of the metaphor: while situationrelatedand entity-related features are typically associated with concepts aligned in visualmetaphors, introspections, and taxonomic features are typically associated with conceptsaligned in linguistic metaphors.
展开▼