TEXT 2. The translation of metaphor

As I see it, the main and one serious purpose of metaphor is to describe an entity, event or quality more comprehensively and concisely and in a more complex way than is possible by using literal language. The process is initially emotive, since, by referring to one object in terms of another ('a wooden face', 'starry-eyed'), one appears to be telling a lie; original metaphors are often dramatic and shocking in effect, and, since they establish points of similarity between one object and another without explicitly stating what these resemblances are ('he leads a dog's life', but elle a du chien), they appear to be imprecise, if not inaccurate, since they have indeterminate and undeterminable frontiers. However, there is no question that good writers use metaphors to help the reader to gain a more accurate insight, both physical and emotional, into, say, a character or a situation. Further, it is not difficult to show that a one-word metaphor, once it is accepted as a technical term, so becoming a metonym (e. g. 'dog', chien, cane), as a 'truck', 'tub' or a 'mine car', and becomes a more or less dead metaphor, may be added to the technical terminology of a semantic field and therefore contributes to greater accuracy in the use of language.

I have never seen this purpose of metaphor stated in any textbook, dictionary or encyclopaedia. The issue is clouded by the idea of metaphor as an ornament, as a figure of speech, or trope, as the process of implying a resemblance between one object and another, as a poetic device. Further, linguists assume that scientific or technological texts will contain mainly literal language, illustrated by an occasional simile (a more cautious form of metaphor), whilst the purpose of metaphor is merely to liven up other types of texts, to make them more colourful, dramatic and witty, notoriously in journalism. All emotive expression depends on metaphor, being mainly figurative language tempered by psychological terms. If metaphor is used for the purpose of colouring language (rather than sharpening it in order to describe the life of the world or the mind more accurately), it cannot be taken all that seriously.

Words are not things, but symbols of things. On Martinet's model we may regard words as the first articulation of meaning, and since all symbols are metaphors or metonyms replacing their objects, all words are therefore metaphorical. However, as translators we know that words in context are neither things nor usually the same symbols as individual words, but components of a larger symbol which spans a collocation, a clause or a sentence, and is a different symbol than that of an isolated word. This is the second articulation of meaning and to this extent language itself is a metaphorical web. Lastly, as Gombrich has pointed out (1978), metaphor is literally translation, and dead metaphors, i. e. literal language, are the staple of accurate translation.

Metaphor is in fact based on a scientific observable procedure: the perception of a resemblance between two phenomena, i. e. objects or processes. Sometimes the image may be physical (e. g. a 'battery' of cameras), but often it is chosen for its connotations rather than its physical characteristics (e. g. in 'she is a cat'). Violence is exercised on reality when the objects or processes are identified with each other, which in the first instance produces a strong emotive effect. Gradually, when the metaphor is repeated in various contexts, the emotive effect subsides, and a new term that describes reality more closely has been created, e. g. étonné which in a seventeenth-century text might be translated as 'thunder-struck', but is now translated as 'astonished'.

(Newmark P. Approaches to Translation. Cambridge, 1988. P. 84-85)