Theoretical and methodological issues in the identification of metaphorical language

Abstract The study provides a critical overview of metaphor identification methodologies and suggests further improvements to one of the most common procedures known as MIPVU. It presents the theoretical background to two major tasks of metaphor identification: the segmentation of text into linguistic units and determining their metaphoricity. The latter includes the demarcation of senses and the specification of source and target domain meanings. Difficulties involved in annotation include the establishment of boundaries to delimit metaphorical meaning-carrying elements, metaphoricity associated with word sequences rather than individual lexis, the treatment of compounds, particles and nominal inflections, as well as conflating or overlapping sense descriptions in the dictionary. Many of these issues are illustrated with a detailed analysis of Hungarian examples. A flexible but principled application of the procedure is recommended depending on the specific research agenda. The study advocates the utilization of the notion of decomposability in determining the metaphorical status of idiom-internal words and suggests a modification in MIPVU’s treatment of compounds. Decisions taken early in the process of metaphor analysis have repercussions in later stages when source domains and mappings are identified or quantitative analysis is conducted.