Context always makes clear that metaphorical mappings go from source to target -- not vice versa
What is it about?
In a metaphor, the projection (or "mapping ) of features is always from source to target, never from target to source. Some authors do not accept this, and argue that the mapping process goes both ways (that is, it is "bidirecttional") -- and thus conclude that the target and source in a metaphor can (sometimes) be reversed. Conceptual Metaphor Theory (CMT) fully endorses unidirectionality, but for instance Lakoff and Turner (1989) have incorrectly rejected Max Black's (1979) "interaction theory" of metaphor because it supposedly accepts bidirectionality/reversal of target and source. The present paper discusses three related aspects of this issue. The paper (1) argues that while Black points out that in metaphor target and source interact, resulting in some fine-tuning of the target, he firmly postulates unidirectionality as a key criterion for calling something a metaphor; (2) shows what is wrong with some experimental research that claims to have found that participants' responses reveal they reverse target and source in a metaphor. The main problem with the experiments is that they fatefully ignore pertinent pragmatic contexts of metaphor. Sperber and Wilson's "relevance theory" helps to explain this misguided thinking; (3) discusses some apparent exceptions to the rule of unidirectionality
Why is it important?
This is a serious issue, as accepting "birectionality" would enormously, and confusingly, stretch the notion of metaphor, which thereby would lose almost all of its power as a theoretical concept.