Abstract

This article describes the Metaphorical Source Domain Identification Procedure (MSDIP), which can be used to code source domains in metaphor identification. In the first part of the article, we describe the complexity of source-domain coding in corpus analysis. We argue that, in many cases, discourse is underspecified and multiple source-domain candidates may be relevant for a specific metaphorical expression. For instance, if a word like “fight” or “target” is used metaphorically, it could refer to either the source domain of war or sports. To make these issues explicit for analysts, we developed MSDIP, which builds on and extends the Metaphor Identification Procedure Vrije Universiteit (MIPVU). In the second part of the article, we explain the coding steps of MSDIP and subsequently report on a reliability analysis, demonstrating the reproducibility of the procedure. We end with a number of detailed sample analyses, demonstrating the role of co-text and context in selecting the likeliest source-domain candidate through MSDIP. These analyses show that MSDIP is both reliable and flexible in dealing with the complexities of real-life discourse during source-domain coding.