Metaphor and Metonymy: Jakobson Reconsidered

LEON SURETTE Metaphor and Metonymy: Jakobson Reconsidered1 Roman Jakobson's '954 article 'Two Aspects of Language and Two Aspects of Aphasic Disturbance' has fundamentally altered the way literary critics use the terms 'metaphor' and 'metonymy.' Before that article metonymy was a little-regarded rhetorical trope distinguishable from synecdoche only with difficulty. Some indication of its prestige can be gathered from the fact that 'metonymy' is not indexed in either l.A. Richards's Practical Criticism or Wellek and Warren's Theory of Literature, although both works index metaphor.2 Since Jakobson, metonymy has been elevated to a status equal to metaphor, and is perceived as its contrary. In the process both 'metaphor' and 'metonymy' have ceased to designate only rhetorical tropes and have become the labels for cognitive relationships. 'Metaphor' is the label for relationsrups of 'similarity: and 'metonymy' for those of 'contiguity.' The following discussion is a critical analysis of Jakobson's article with a view to locating some of the confusions arising out of the displacement of tropological concepts by cognitive ones while retaining the same labels for both sets. Before turning to the article which established the current popularity of 'metonymy' as a rhetorical term, I should make clear what I mean by the distinction between cognitive and tropological concepts. Since much of the controversy endemic to rhetorical discourse is traceable to the placement of this distinction, one can expect any placement of it to be hotly disputed. Classically, a trope is a deviation, a 'turning' - some alteration of standard linguistiC practice so as to draw attention to the expression so denominated. Perhaps the simplest model of the trope is the pun. In this humble trope an acoustically similar word is substituted for the 'correct' one: 'Bare with me.' Here the 'turn' is a mis-selection of a word or lexeme - 'bare' meaning to expose, instead of 'bear' meaning to carry or, in this case, 'put up with.' The explanation or explication here offered is, in my terms, a cognitive one. I call it 'cognitive' because the explanation appeals to the sense, meaning, denotation, or reference of the lexemes in question , and not just to their linguistic roles.3 Even the pun cannot be construed within the horizon of language, but requires an appeal beyond language to a cognitive realm where expressions have content and not just form. UNIVERSITY OF TORONTO QUARTERLY, VOLUME 56, NUMBER 4, SUMMER 1987 558 LEON SURETTE Since Gottlob Frege's work, logicians and linguisticians alike have built their systems on the axiom that meaning is a purely formal matter involving definitions and operational rules. Within this Fregean universe there is no room for 'dirty' cognitive elements involving the content of the expressions - that is to say, information about the properties, characteristics , locations, relationships, etc of those entities and events the expressions might be supposed to 'pick out' or designate.' Cognitivism - varied and disputed as it is - involves the assumption that expressions have contents and that their contents have a role to play in their interpretation. The main stream of twentieth-century thinking has resolutely denied the pertinence of content in interpretation. It has done so in pursuit of the will-o'-the-wisp of 'objectivity.' Within linguistics, structuralism follows this tendency in its insistence on language's independence of any cognitive content it may be supposed to carry - on its independence ofmeaning and reference. Richard Boyd sums up one current view of the pertinence of content in interpretation of linguistic expressions as follows: Scientific metaphors raise truly fundamental issues about language and linguistic competence, and the theory of reference required to understand them has several quite startling consequences, which are important both to an understanding of metaphorical language, and to an understanding of language in genera1. We shall discover, for example, that there is, in an important sense, no such thing as linguistic precision; there are rational strategies for avoiding referential ambiguity, but they are not a reflection of rules oflinguistic usage (as the empiricist theory suggests).' Semiological theories of language such as Saussurean structuralism or that of Jakobson's Prague school are grounded on the assumption that linguistic precision is attainable - indeed, that such formal precision is virtually inescapable within linguistic...