This paper discusses problems associated with translation of metaphors in literary texts from a relevance theory (RT) perspective. It begins with a general account of how metaphors are treated in translation theories (most notably by dynamic equivalence theory) and expounds the merits an RT-based translation theory has over others in accounting for various aspects of metaphor translation by drawing on such notions as efforts vs. rewards, poetic effects, and interpretive resemblance. Specifically, it is pointed out that the biggest reward (or contextual effect) obtainable from processing a metaphor is the image associated with it that can give rise to multiple implicatures with varying degrees of strength. This leads to the observation that an approach heavily skewed toward dynamic equivalence and thus tolerant to form modification (metaphor deletion, in this case) in favor of message communication may risk stripping RL readers of access to the richness of meaning inherent in metaphors. Based on these theoretical considerations, five specific methods of metaphor translation (M→M, M1→M2, M→S, M→P and M→S+P) are discussed with an analysis of examples taken from published translations of literary works. The analysis is geared toward elucidating how in each method processing efforts are reconciled with rewards and how this dynamic relationship relates to the goal of achieving equivalent relevance between SL and RL.
