Book chapters will be unavailable on Saturday 24th August between 8am-12pm BST. This is for essential maintenance which will provide improved performance going forwards. Please accept our apologies for any inconvenience caused.
One of the twentieth century's most memorable and amusing figures for the writing of poetry is Robert Frost's ‘Like a piece of ice on a hot stove the poem must ride on its own melting.’ Typically, it was Frost who undermined his own simile, insisting, in a 1931 talk called ‘Education by Poetry’, that ‘all metaphor breaks down somewhere’. By viewing the whole concept of figurative language as inconclusive, he freed it from carrying too much responsibility for consistency. ‘It is touch and go with the metaphor,’ he wrote. ‘That's the beauty of it. Until you have lived with it long enough you don't know … how much you can get out of it and when it will cease to yield.’ Likening the act of writing a poem to ice melting on a stove proved Frost's point. The simile breaks down right away: a piece of ice on a stove invariably evaporates and disappears, whereas the test of a successful poem is that, having melted itself more or less from delight to wisdom, it remains in the canon as a work of literature, a work of art.
Whatever the limitations of the melting ice metaphor, it is certain that figures of speech are not only the stuff of poetry; they are built right into our language. They are part of the mechanism through which we work with words to make words work for us. By language I do not mean the noises people make in circumstances of extreme distress or pleasure. Babies cry, lovers moan, a man pushed off a cliff hollers Aaaaah. I can't think of such outcries as language because verbal communications must be to some extent creations of the mind – or more accurately, of many minds coming to (changeable) agreements or adopting habits over hundreds of years about what, approximately, certain repeatable noises or combinations of sounds and symbols mean. Nothing is ever fixed about language.
Without wanting to get entangled in theoretical barbed wire, I would like tentatively to suggest that although language relates to meaning in many ways, these can be sorted into three general categories.