Why we mix metaphors (and mix them well): Discourse coherence, conceptual metaphor, and beyond

The paper explores the phenomenon of metaphors that occur in close textual adjacency, i.e. as metaphor clusters, but do not share a similar cognitive basis. Clusters frequently mix ontologies and are thus devoid of coherence that can be explained as emerging from a single conceptual metaphor. Evidence to that effect comes from a British corpus (Sun and Guardian) of 675 newspaper commentaries covering the 2004/05 EU referenda (in all, 2574 metaphors). First, it turns out that journalists combine metaphors into complex, yet well-formed arguments on a regular basis, with 39% and 62% of all metaphors respectively occurring in clusters. Even more strikingly, the data reveals that ontologically mixed metaphors account for 76% of all clusters and that almost all of these are straightforwardly comprehensible. This challenges the view of mixed metaphor as awkward language usage. I argue that mixing works because metaphors are typically embedded in separate clauses situated at different temporal, causal, speaker, or belief-related conceptual planes. By consequence, no strong joint processing pressure arises that could result in a perceived clash of metaphorical imagery. Thus, felicitous mixing is a natural by-product of the shifting logic of clauses in complex argumentation. In addition, I present a qualitative typology of how clustering metaphors interact in argumentation. It calls into question the view that conceptual metaphors are the coherence-maintaining device par excellence. While conceptual metaphors may create “internal binding” in ontologically coherent clusters, complementary “external binding” models are needed to explain the mixed clusters (and ultimately for a full explanation of all kinds of metaphor-based argumentation).