The myth of the four tropes

I’d like to take a moment to dismiss one of the most persistent schemes – one which has appeared in various forms over the decades, and which we might refer to as ‘the myth of the four tropes’. Theorists from Giambattista Vico in the eighteenth century to the twentieth-century rhetoricians Kenneth Burke and Hayden White (and even as recently as Jonathan Culler) have claimed that metaphor, metonymy, synecdoche and irony are the four ‘master’ tropes, and constitute the most meaningful categorical distinctions. They are, in Vico’s wonderful phrase, ‘the corollaries of a poetic logic’.1 The idea is that these four tropes describe the basic types of relations that can exist between two terms. Metaphor relates through similarity; metonymy relates through direct link; synecdoche relates via the part to the whole; and irony relates through inversion.

I’m afraid, as the scientists say, the idea is so bad it isn’t even wrong. Synecdoche is a useful term, especially when extended to species-for-genus and vice versa; but it’s merely a subset of metonymy. A thing’s belonging to a larger or smaller categorical or hierarchical class is just another attribute in the set of its aspects and connotations; in the mind, ‘parts’ are just attributes of wholes, and set-membership is an attribute of a set-member.2 (The entire mess, incidentally, has its origin in a failure to distinguish between the myriad expressions of trope and the deep conceptual mapping that often reveals them as very similar mental operations. This is the error that leads people to conceive, say, of ‘metaphor’ and ‘simile’ as conceptual distinctions, though in this regard they’re virtually identical, and their difference, while not insignificant, is largely presentational.)

Irony presents a stranger problem. The key phrase here is ‘relates through inversion’, a nonsense that should have the alarm bells ringing. Yes, irony appears to subvert or invert the relation between a word or statement and its referent, between the signifier and its signified. But the inversion doesn’t create the connection: something else does first. Irony serves only to pervert a pre-existing relation, not forge a primary one. Indeed, most of the stock examples of ironic trope – e.g. ‘Gorgeous weather today!’ when it’s raining, and ‘Working hard, are we?’ when you’re clearly not – reveal irony as little more than an presentational mode (and so does any study of its more sophisticated manifestations in literature, film or drama, where it may take structural, gestural and generic as well as expressive forms). The inversion takes place at a linguistic, structural or interpretative level, but not at a conceptual one. Absolutely no one is thinking, deep down, that ‘YOU ARE NOT WORKING = YOU ARE WORKING’. Trope-types, if they’re going to form a useful taxonomy, can really only refer to relations being made within or between concepts. The whole subject of irony is therefore far more appropriately discussed under the linguistic heading of pragmatics or stylistics, or in our case, poetic address.3 Metonymy and metaphor, as we’ll see, are certainly significant and central fixtures in our neural circuitry; but I doubt much evidence will ever be found for an important ‘irony centre’ in the brain where concepts are inverted in sense before they seek a linguistic representation, however strongly certain politicians and US states give the impression of a bypass operation being widely available.

The idea that there are only two trope-classes, metaphor and metonymy, has been around for some time now. It was given its most famous defence by Roman Jakobson in the 1950s, when he argued that they represent the two fundamental ‘axes’ of human communication: selection and combination.4 There is a horizontal dimension of syntagm (here, something like ‘a rule of syntactic series’), based on rule, combination and contiguity; and a vertical dimension of paradigm, based on selection, substitution and similarity. If you take a dinner menu, the syntagm is the part of the menu that goes appetiser – main course – dessert, and the paradigm is what allows you to select between cheesecake, gateaux and ice-cream, the last three sharing the property of ‘sweetness’ we consider appropriate for entry to the conceptual domain of ‘dessert’. These days linguists tend to talk more about things like grammaticality and lexicality, but it’s the same broad principle.

Metonymy and metaphor reflect the two most fundamental ways our brains make useful inferences about the world and connect the information they extract from it. I’ll refer to the first type, the metonymic type, as intra-domain, and the second, the metaphoric, as inter-domain. Another way of thinking about it is ‘classical category’ and ‘family resemblance’. Classical categories allow us to apply the same rule to many subjects in any arbitrarily law-governed context, while family resemblances allow us to see similarities between disparate subjects, and rely on our memory to compare and verify those similarities. The first works by relation, the second by correspondence. In the first case the rule is prime; in the second, the item; in the first, grammatical procedure; in the second, memorised lexis. (A wonderful demonstration of this has been given by Pinker and his team in their work on regular and irregular verbs.)5

Tropes of relation, like metonymy, allow us to ‘decline the world regularly’ and apply broad and often arbitrary rules in a given context, and eat the five courses of our meal in the culturally prescribed order. Tropes of correspondence, like metaphor, allow us to find ad hoc rules based on common resemblances between memorised items. Though the menu may only offer bream, turbot and rollmop herring for the fish course, this allows us to ask ‘do you have any shrimp?’ without sounding like a lunatic, whereas the request ‘do you do chicken livers?’ will convey the impression that you have a broken paradigm-axis.

The question of whether we might consider ‘symbol’ – where concrete detail stands for an abstract quality – a distinct third category is moot, and I will pursue the matter at length in another chapter. Symbol has often been regarded as a mere subset of metaphor where the tenor is text-absent. I believe this is a mistake, based on the failure to see that the symbol combines elements of both metonymy and metaphor, and is also a formal expression of the unique tropic properties of the abstract realm. My strong feeling is that symbol is a third class of trope, though its uniqueness lies less in its being a composite sign (which it is) than in its being a sign of pure semantic projection; symbol is a tropic means of assigning meaning to the world. It seems plain to me that, while metaphor and metonymy are really just human reflections of the basic universal functions of paradigm and syntagm, ‘symbol’ is an entirely human contribution, and describes the unique process whereby we confer the quality of ‘meaning’ on things which do not intrinsically possess it (i.e. ‘everything’). As all this rather flies in the face of accepted practice, the proof will be miserably long; I offer my apologies in advance for the abstruseness of the argument, discussed in the chapter ‘Concrete, Abstract and Symbol’.

The later chapter on ‘Four Semes, Four Tropes’ points the way to an alternative scheme. Theorists have long made the distinction between the ‘pure symbol’ and the ‘literary symbol’; ‘symbol’ here is a mere homonym. The pure symbol is the result of an arbitrarily forged connection, one which can arise through a cultural, social or wholly subjective process. It ‘means’ nothing but that which is arbitrarily ascribed to it. The so-called ‘literary’ symbol is almost precisely its opposite; it’s formed through an entirely different conceptual procedure, and is less ‘literary’ than ‘the trope which generates what we call meaning’. In my own scheme there are still four tropes, but they are rather different from our traditional candidates. I refer to them as metaphor, metonymy, symbol and asymbol (which connects to ‘aseme’, the anti-seme of arbitrary link; the word ‘asymbolic’ has some minor currency). They correspond to inter-domain, intra-domain, supra-domain and what we might call extra-domain processes. (One might visualise them respectively as a circle; two intersecting circles; one circle enclosed within another; and two circles whose edges touch.) They work, respectively, through the attributes one domain encloses, the attributes shared by two domains, the enclosure of one domain within another, and the arbitrary contiguity of two domains.

Peristasis and metonymy

Peristasis and metonymy are ‘tropes of relation’, and both concerned with registering contextual saliences, that is, ‘relevant detail’. They are tropes which show we are aware of the rules of our immediate frame of reference, or which body those rules forth. They are both intra-domain tropes which essentially differ only in their formal expression: peristasis is additive, while metonymy is subtractive.

Let’s begin with peristasis (peri – around; stasis – stand), a term which will be new to most readers: I’ve stolen it from a little-used rhetorical figure that refers to the ‘description of attendant circumstances.’6 It’s the difference, if you like, between saying ‘the woman’ and ‘the old woman’; between ‘the boy’ and ‘the boy with the red setter’; ‘the office’ and ‘the office we have in Reform St’; ‘the takeaway’ ‘and ‘the greasy takeaway’, ‘the sum’ and ‘the difficult sum’. Peristasis adds to the subject by declaring or implying some significant secondary attribute, that’s to say one either relevant to the specific domain in which it passively finds itself, or (as is more usually the case in a poem) propositional of a specific domain in which it actively places itself.

Now you might think a mere casual expansion in the noun phrase is such a natural and basic feature of language it’s barely worth identifying as a trope at all. Usually we’d just subsume such things under ‘descriptive language’ and the wretched category of ‘image’. What a peristasis does, however, is declare an additional detail, usually in the form of qualification or predicate, that for contextual, specific-domain reasons we find relevant, important and worthy of our attention. As we’ll see, it’s really a kind of pre-metonymy (cognitively, I think it functions specifically as a potentiated metonym), and shares its deep form. ‘Peristasis’ bears roughly the same relation to ‘metonym’ as ‘metaphor’ does to ‘enigma’ (my word for ‘riddle’): metonym omits the ‘subject’, enigma the ‘tenor’. I would also argue that this ‘additional detail’ may be unspoken, i.e. text-absent, and present only as an ‘active connotation’ drawn forth or implied by the context – or what I’ll soon call the ‘thematic domain’ – in which it appears. Similarly, what looks like a simple subject is often a contextually identifiable salience, telling us we have a metonym: for example ‘the woman’ may indeed be metonymic if the person concerned is being pointed out as the only female member of a gang of male football hooligans; the full sense here is really ‘that human in the group who is female’, or ‘the woman one’. Formally, this is arguably identical to pointing out someone as ‘the spots’ in the context of a series of otherwise immaculate mugshots; a peristatic expression of the same detail would be ‘the spotty one’. Peristasis is a contextually relevant expansion; metonymy is an understood contraction of a peristasis. Their presence in any text is a matter of sensitive interpretation.

Since peristasis encloses the idea of the so-called ‘poetic image’, thinking of it as such will at least give us a conceptual handle on the subject. It’s important to note, though, that calling things ‘images’ has generally been a disaster for sensible poetic analysis. It’s a vague and outmoded category, which falsely privileges ‘the visual’ where it should receive no special treatment. In more enlightened uses of the word, ‘the image’ will be broadened to include ‘the sensual’ – though this still tends to exclude qualities like ‘emotional tone’ and ‘abstract quality’, which are just as important to literary description as any others. That’s not to say the category of ‘the visual’ is unimportant; it’s just one of several concerns in the context of any linguistic analysis.

(To air a further irritation: ‘the image’ has allowed people to talk about ‘ekphrastic’ poetry as if it constituted a meaningful literary genre. It’s a cool-sounding word all right, and ekphrasis did indeed have a nobler role in classical verse – but in our own age, it’s almost as arbitrary a genre as ‘poems about animals’, much as I love poems about animals. Poetics concerns itself most usefully with word-stuff, because poems are machines made of words and concepts; ‘imagetalk’ is often a way of avoiding any more rigorous analysis. Since ‘peristasis’ is unlikely to take off – I propose that we might also adopt the neutral word ‘detail’ as a serviceable, catch-all alternative to ‘poetic image’.7 The word ‘image’ is fine, provided its use is confined to ‘visual representation made of words’.)

In the generic space of our standard ‘resting state’ definitions, ‘an old woman’ is just ‘an old woman’. However, additional knowledge of the specific domain in which the detail occurs can assist us in seeing why the speaker chose to name that attendant detail, and not another. This turns the detail into a peristasis. Indeed, the attribute will often be unusual, striking or odd enough to alert us to the fact that we actually have a significant peristasis – and that its full meaning must be derived from the rules of its specific domain, the very presence of which it may have alerted us to.

To take the examples I gave earlier: the woman I described as ‘old’ is also Irish, I happen to know – but since I was pointing her out at the back of the bus, her Irishness is pretty useless in her quick identification. The boy with the red setter has a blue jumper, but in the final of Crufts, he is most relevantly identified by his dog. The office in Reform St is very spacious, handles billing, and is painted yellow – but this firm has two offices like this, and I needed to direct you to this one, because I work there. You get the picture. Peristasis is at the heart of poetic description, where poets select their qualifiers and descriptive language with obsessive care: this sharpens and directs the way in which a particular detail is interpreted, and implies which rules may be governing the whole local domain of the poem. Simplistically: if a yew tree is described as ‘a thousand-year-old yew’, it is likely that the poem in which it appears is concerned in some way with ‘eternity’, ‘history’ or ‘time’. (And even more so if the peristasis has been made even more salient by metonymy, and shrunk to its theme-indicating aspect, i.e. ‘I sat below the thousand-year-old yew.’) No detail can be arbitrary in a poem, which (unless it declares itself otherwise, through surrealism or some other explicit strategy) is almost invariably read as ‘consciously composed’. As we’ll see, the reader is even entitled to insist that within the context of the poem-reading contract, a detail can never be merely descriptive. Even if its deeper meaning isn’t yet understood, they will proceed with their rereading on the assumption that it has one, and its symbolic or peristatic importance will gradually manifest.

The concept of peristasis is, I think, invaluable in understanding poetic metonymy. A metonym is just a peristasis with the subject removed. The metonym narrows the subject by referring to it only via a declared aspect, part, direct connotation, causal relation, ‘feel’ or quality. To put it more succinctly, a metonym ‘caricatures something in context’. This saves time, which is why it’s often called a ‘trope of contraction’ – and poets love metonymy, because in poetry, time is money. Compositionally, it’s a two-part process: the first stage is the identification of a contextually relevant and salient feature of a subject; the second is the elision of the subject, leaving the related feature to stand for it, and point to it.

The text-present part of the metonym, the connotation/aspect/feature/part/consequence of the thing, is functionally identical to an ‘index’ in semiotics. A beeping smoke alarm indicates ‘danger’ in the generic domain. However, we also interpret indices contextually, and in the specific domain of my kitchen, it usually indicates ‘toast’.8 Often these indices are complex, and composed of a succession of relational links – one thing stands for another thing, which stands for another thing – each of which might be constrained by generic or specific frames. The noise of a smoke alarm may say ‘danger’; but it does so through first indicating ‘the presence of smoke’, which generically indicates ‘fire’, which, in the context of ‘the home’, generally indicates ‘get the hell out of here’. A smoke alarm would go off near an outdoor barbecue too, but could safely be ignored. (I’ll refer to the relational aspect and the subject of the metonym as ‘index’ and ‘subject’.)

In literary theory, a compound metonym or a chain of substitution is sometimes referred to as a ‘metalepsis’. My alcoholic great-great-grandfather was known as ‘Drink-the-Bible’ from the way he disposed of the money he earned as a door-to-door book salesman. He drank neither bibles nor the money he earned from their sale, of course, but whisky: ‘Bible’ here is metaleptic. (Working backwards: he drank ‘whisky’; this is short for ‘whisky bought with money’; this is short for ‘whisky bought with money from bibles’; this is short for ‘whisky bought with money from bibles which were representative of the books he sold’.) The fact that several terms in the metaleptic chain might be skipped means that it can play only with such conventional and idiomatic connections as can be easily elided – e.g. one might casually say my forebear ‘drank all the proceeds of his bible-selling’, without even being aware of the metonymic substitution ‘proceeds’ for ‘booze he bought with the proceeds’. In Sonnet 10, Shakespeare refers to the aristocratic family of the Fair Youth as ‘that beauteous roof’. The qualifier ‘beauteous’ directs the vector of sense back to the lovely boy himself, while ‘roof’ lies at the end of the metaleptic chain. ‘Roof’ is a synecdochic metonym for ‘house’, and ‘house’ is a metonym for the aristocratic family of which the Fair Youth is a member. If we take Shakespeare’s performance to be rather disingenuous, and suspect him of being really more concerned with the youth’s desirable beauty than with his ability to replicate that beauty through procreation, we might diagnose that beauteous is itself a metonym, and not only refers to the fine features of the Fair Youth’s noble line, but also points back to the ‘real subject’ of Shakespeare’s undeclared love. (It very much depends, as they say, on ‘what you think he meant’, but to refuse to speculate on his intentions is a failure of intelligent analysis. Compassion prevents me from making a further assault on the already whimpering ‘intentional fallacy’.)

Many metonyms in the generic domain will be very familiar, and are based on standard formulae, themselves straightforwardly indicative of what we find important on this planet, and in this human life. These might be place-for-person-or-institution (‘the White House’ for the US presidency; ‘Scotland’ for the Scottish football team); part-for-whole (local-type synecdoche, ‘wheels’ for ‘car’; ‘ass’ for ‘whole body’); whole-for-part (‘society’ for ‘the rich’); substance-for-whole (‘lead’ for ‘bullet’); place or time-for-event (Waterloo; 9/11); symbol – in the Peircian sense of arbitrary signifier, what I call an ‘asymbol’ – for signified (‘flag’ for ‘nationality’; ‘rose’ for ‘love’). All these formulae are the intra-domain equivalent of the LOVE IS WAR-type mapping formulae employed in inter-domain conceptual metaphor; that’s to say they indicate the broad conceptual operation behind the specific verbal expression. With intra-domain tropes these formulae are syntagmatic and rule-based; with inter-domain, paradigmatic and item-based. As we’ll see, they are both infinite sets.

Metonymies of the effect-for-cause and cause-for-effect type are standard literary (and especially poetic) substitutions; a hoe-handle described as ‘sweat-seasoned’ likely means ‘by work’ not merely ‘by sweatiness’, and the reader is asked by the very salience of the image to ‘read in’ and look down the causal chain towards the full sense. In George Mackay Brown’s ‘Hamnavoe Market’,9 we find ‘Johnston stood beside the barrel. / All day he stood there. / He woke in a ditch, his mouth full of ashes’ – and infer the hideous, drunken end to Johnston’s day through the evidence of the morning after. Elsewhere in the same poem, ‘A gypsy saw in the hand of Halcro / Great strolling herds, harvests, a proud woman. / He wintered in the poorhouse’ – and again, we interpolate the patent uselessness of the gypsy’s predictions at the causal lacuna of the second and third lines. (An example of ‘enthymeme’, if anyone still cares.)10 [For a short handbook of rhetorical figures still in use, see endnote 10.] In another stanza, Garson receives ‘an eye loaded with thunder’, after a fight in a boxing ring – but we don’t need to see the thrown punch, only its effect.

Many metonymic contractions are so familiar you don’t notice them. They’re so culturally convenient that they have become unconscious, and index has become wholly conflated with subject. When a kid uses a container-for-contained metonymy and asks you to ‘zip me up’, they mean ‘zip up the coat-of-me’, of course. ‘Name-for-person’ is even weirder, but metonymy it is, somewhere; ‘Don Paterson’ is a linguistic sign arbitrarily attached to me, but it can also be seen as shorthand for ‘he-who-goes-by-the-name-of-Don Paterson’, and this convention is again born of the generic domain, where we regard ‘names given by our parents’ as important and useful designators. [This statement is hardly uncontroversial. See endnote 11 for a defence.]

My name is not ‘Stevenage’ – but it was recently, in a conversation between two ticket inspectors about how many taxis to order to replace a cancelled train. In A&E, the same person might be better described by the index ‘the broken arm’ – or in a café as ‘the ham sandwich’, as in the famous textbook metonym ‘the ham sandwich wants his bill’. All these specific-domain contexts – café, train, A&E, the bookie’s, church, Facebook, kitchen, confessional, bedroom, academic conference – have their predominant concerns, which then produce the rules by which their contents are usefully organised and prioritised. In other words, while we might repeat the textbook definition that metonymy consists in ‘the substitution of a related term for the thing itself’, in non-generic contexts, it’s the rules of the frame or context that promote certain kinds of relation, and make the substitution both possible and meaningful. Most poetic peristases and metonymies are products of subject, index and frame; indeed without a clear frame, the indexicality of the attribute is weakened drastically. We can consider indexicality itself a product of an enclosing, rulebased domain; the effective power of an index reflects the strength of the local rule which governs it.

It’s been particularly difficult to nail what kinds of relation are up for grabs; one would require a full description of all possible attributes of a thing, and various theorists have produced some clines and systems and taxonomies that make my own look like models of sober continence. I’ll discuss these in due course, but as I’ve said, any aspect, connotation or relation can form either the shared ground between two domains, or the substituted term in an intra-domain metonymy. We tend to derive our metaphor mappings from category-type (e.g. ‘argument is war’) and metonymic formulae from rule-type (e.g. ‘container-for-contained’), but these are both descriptive of instances, not categories. The two sets are symmetrically infinite. The first is obviously so; however some theorists have tried to establish a finite number of metonym-rules, mainly because they find the prospect of an infinite number depressing. But they are forgetting that the ‘rule’ is generated not between the index and subject, but via the local by-laws of the specific domains and ‘given situations’ which enclose them – of which there are, of course, an infinite number.

Incidentally, if one were to attempt a true categorical list of ‘kinds of connotation’, this would inevitably be derived from something like the ontological categories Ray Jackendoff devised in conceptual semantics, a finite list of nine possible schemata which each have infinite generative extensions.11 These primitives are very clean and Aristotelian, and seem comprehensive: Thing, Event, State, Action, Place, Path, Property, Manner and Amount. They would generate broad categories for all possible attributes, simply because they cover everything that humans could possibly be interested in. (If you’re susceptible to this sort of thing, I defy anyone to look at Jackendoff’s conceptual-semantic parsing of a sentence and not weep at the beauty of it – and be astonished that no one had thought of doing this before.) Of course, it’s important to remember that articulate bats – not to mention those silicon-based dogfish in the Beehive Cluster, and that intelligent gas-cloud floating off Betelgeuse – will doubtless have other categories too. Our own may well turn out to be universal, but they are unlikely to be any more comprehensive than are our perceptual faculties.

All it takes to form a specific domain is for it to recur – or merely occur, if its exigencies are clear and compelling enough to propose a set of rules. In life, we go … ‘I’ve been here before, and I am now cognisant of what should be prioritised in this situation’. This is a dentist’s surgery, where my dinner plans and choice of comedy underpants are largely irrelevant; if I stagger towards the reclining chair yelling ‘I can’t sleep!’ I can assume my dentist will not ask me if my bed is lumpy, or if I’m worrying about my relationship. Or if your dog goes missing, the salient rules of ‘I’ve-lost-my-dog-domain’ establish themselves immediately: when you scream at your partner, ‘We have to call everyone!’ – you don’t mean the priest, the bin-man and the Arts Minister. You mean ‘We have to call the dog pound, the police, the vet’s and the local radio station!’ i.e. ‘everyone related to dog-losing’. However, any recurrence of a situation will produce rules. Take Lautreamont’s famous scenario, ‘the chance meeting of a sewing machine and an umbrella on a dissecting table’: it might be pure surrealism to begin with, but stare at this little tableau every day for a month. If, two months later, you happen to chance on an old Singer, resting on an operating table or otherwise … Just try not thinking of an umbrella. Within the domain of Singer sewing machines, umbrellas have now assumed a particular, if arbitrary significance.

The act of reading and rereading a poem forms and then confirms a specific domain, ideally one close to that which the poet had devised. (The poem is too unstable a sign for it ever to be identical; however, many poets are in permanent denial of this fact. Poetry can never be a mode of clear communication – though if it doesn’t aspire to be, the results are doubly confusing.) While an initial reading will begin to conjure up this space, rereading will confirm and solidify its governing laws and priorities. The way we describe things in poems – whether in the form of peristases or metonyms, metaphors or symbols – will depend on the nature of the poem’s specific domain and how it has formed in the act of rewriting, a heuristic process reflected in that of the reader’s rereading. I’ll refer to the poem’s domain as the ‘thematic domain’. This is simply the domain of ‘what the poem is about’, and its ‘aboutness’ remains in flux until the poem is fully composed or fully understood. These states are never attained but essential to posit, so they can guide the consistent direction in which our composition or interpretation should be heading.

To give a rather facile example: suppose you are writing a poem about a beloved aunt’s funeral. In it, you describe our old poetic standby, a tree. You will immediately, and probably unconsciously, start to look for aspectual details that support the thematic domain. (Indeed this specific domain radically reconfigures the core and secondary connotations of ‘the tree’ as it would normally appear in the generic domain of the parole, essentially operating as if it constituted a micro-language.) You might describe the tree’s many rings, as a way of introducing the themes of age and the passing of time; or its place by the cemetery wall, or its turning leaves. You may reduce it to an attribute or connotation alone, and show the reader just the tree’s shadow, or the ghostly whispering in its foliage. You may draw out an unspoken connotation by context (‘peristasis by inference’) – if you chuck a ‘yew tree’ into a funereal poem, it will be hard for many readers to stifle the mythical associations the yew has with death and rebirth; or to take a less ‘symbolic’ example, a ‘willow’ or a ‘blackthorn’ will have more ‘tearful’ and ‘darker’ qualities drawn from them in our funeral poem than they would if they appear innocuously in a poem about a children’s picnic. And so on. At the very least, the peristasis you use is unlikely to include many positive and optimistic words, since they would contradict the core funereal attribute of ‘sad’. All that said, the reality of poetic composition is that the coherence of the result rarely reflects a coherent process: because the poet usually brings little more than a hunch to the empty page, a vague intuition that a couple of things, events, feelings, ideas might somehow be connected, it could easily be that a free-floating attribute suggested the inclusion of the tree in the first place: we need a shadow in this poem, so what might cast it? Indeed it’s just as likely that the lone peristasis that proposed the funereal theme itself; you had begun the poem as merely a note about a tree in shadow, but had identified some hidden proposition, slippage, play or give in the image that proposed the larger theme – one you had perhaps been unconsciously ducking.12 (I am convinced that an ignorance of the messiness of the poetic process must account for about 50 per cent of errors in critical analysis, where the poet is credited with the deliberate creation and brilliant timing of an effect that was achieved through a mixture of luck, intuition, accident, error or unconscious gesture. Poets will rarely contradict a positive critic, though, no matter how misguided their praise.)

Once the specific domain of the theme has been established, the poet has the option of cutting peristasis down to metonym, if they feel the full detail can be readily derived from the thematic context. So, in something which has established itself in the reader’s mind as a ‘post-funeral arboreal poem’, a man standing below a mournful, whispering oak may simply ‘stand in the shadow of a mournful whispering’. In Sonnet 122, Shakespeare riffs on the idea of a notebook (one he appears to have been given as gift, and subsequently lost; the poem feels like a guilty contrivance) as a memory-substitute. This he compares unfavourably to his own excellent memory, claiming that his beloved is, besides, much too memorable and precious to require such a lowly means of record. Nine lines in, he is able to shorten his reference to the notebook to ‘that poor retention’. Without the carefully established thematic domain of ‘notebooks as inferior means of remembering the beloved’, the metonym would be impossible, and the phrase incomprehensible.

The employment of intra-domain device as default in poetry is a profound technique which forces the reader into an intimate relationship with the text, and so with the poet. The rules of poetry-rereading mean readers must construct the thematic domain by which the metonym or peristasis will be correctly interpreted. ‘Why is that tree so oddly described? Aha …’ It’s through this quiet pursuit of the half-said thing that the reader enters into a state of co-authorship – and what makes poetry a more interactive art form than just about every other, bar the videogame. Often working far more quietly than inter-domain tropes, peristasis and metonymy are the main strategies by which a space is made for the reader, into which they can bring their own interpretation, feeling, ideas and experience. (Their discreet operation is one reason it took a couple of thousand years for them to be properly acknowledged; Jakobson can be credited with restoring metonymy to its rightful place, though neuroscience would have taken care of it eventually.) The half of the poem the reader completes will never wholly coincide with the half the poet intended; the thematic domain they construct will never quite be the one whose spell the poet was under. But this interpretative slippage is necessary for ownership of the poem to be transferred, and its meaning personalised. What, and how much, we are left to deduce from a poem’s metonymic or peristatic evidence is very a matter of style. All good poems supply enough evidence for a complete domain to be circumscribed, but some poets prefer the ‘expert Sudoku’ approach of leaving four clues from which a far larger, interlocked pattern can be deduced or a unifying tone intuited, while at the other extreme, others ask that – as complete and transparent as their poem may seem – it be regarded as a synecdoche of a larger experience. The intra-domain contraction of the poem accomplishes two of our main poetic goals, and makes its speech both brief and original – but it also leaves space for the reader to act as co-creator. Metonymy is a door held open at an inviting angle – showing the reader neither too little, where the poem closes the reader out, nor too much, where the reader can see no reason to enter.

Metaphor

To return to our bad funeral poem: summoning our miserable tree with a peristasis or a metonymy might be insufficient to our meaning. We might want to describe it as if it were something else – a bared nerve, a woman wild with grief, or some kind of negative, inverted lightning. This means that we have to go inter-domain.

Metaphors are tropes of correspondence, and find family relationships between things. Taking our cue from I. A. Richards,13 we generally term the literal subject the tenor (which I’ll abbreviate to T), and the (most often) non-literal thing we compare it to the vehicle (V). Metaphors work through the shared content of T and V, which we call the metaphor’s ground. The amount of this shared content determines the size of the ground, and the similarity of the thing compared. In semiotic terms, metaphors come somewhere between icons, or ‘motivated signs’ (which share so much content with the things they represent that they are instantly recognisable, e.g. a drawing of a sofa for the sofa itself), and true symbols, which share no content with their referents, and whose connection with them is completely arbitrary and learned.

In literature, metaphor is the act of connecting two apparently unconnected things. We say ‘X is Y’, and overcome the patent untruth of the statement by simultaneously demonstrating that X and Y are, at least, surprisingly alike; the metaphor is a lie that is then proven to be partly true. As usual, we naturally incline towards a ‘pink noise’ mean: if a metaphor is to be effective, it will usually steer a path between uncorrelated leap and overcorrelated connection. This means that we rarely make the mistake of creating a metaphor from two things that are either too close or too far apart. ‘This table is a banana’ is a pretty lousy metaphor, and could only be qualified into some sense by a very specific and accommodating context. The two things are too unlike, and the ‘ground’ is too thin; they don’t share enough properties for the statement to declare anything but their difference. This, if you like, is a random, ‘white-noise’ metaphor. At the other extreme, a ‘brown noise’ metaphor is where tenor and vehicle are too close and predictable; it isn’t worth the trouble of pointing out their similarity.14 To say an apple looks like a bit like a peach, or a can of 7-Up looks a hell of a lot like a can of Sprite isn’t making us see those things afresh, which is generally our aim: to have an opportunity to reconsider ‘the known’ via an unexpected correspondence. (For this reason, metaphor is one of the key ways in which language instinctively reinvents itself, and makes itself adequate to a changing reality; this is a point to which I’ll return.)

I’ve mentioned that it’s our instinct to connect any two things that happen to be thrown at us. Where we cannot forge a direct relation or a context (in poetry, a ‘thematic domain’), we will resort to metaphor: if we can’t connect by rule, we will do so by finding some family resemblance between things. This is neatly demonstrated in the workshop game where a noun is randomly given the definition of a completely different one. It can almost invariably be ‘made to fit’: we ransack the noun’s internal properties, aspects and connotations until we have found several points of coincidence or imaginative correspondence – and it transpires that a keyhole is, in a sense, a square box in which one watches moving pictures, or that a TV is a hole into which one places a key that opens a room, or that a train is a small, quick-moving mammal of which some are irrationally frightened, or that a mouse is a fast kind of public transport that runs on rails … Or at least possibilities that can be entertained. (Most of us understand that in this life the imagination is not a negligible reality.) What’s astonishing is the degree to which this process is instinctive and instantaneous – and the extent to which everything really does seem to be secretly related.

Metaphors can take many forms, and I’ll explore some of these in due course. For now, keep in mind a couple of simple facts: there are an infinite number of ways metaphor can be presented in a text, and some of them (especially the ‘simile’, where one thing is compared to another, rather than claimed to be another) have traditionally been given undue prominence. These presentational differences are interesting: they will often imply alternative aspects of the same mapping and present different cognitive experiences, but they do not, I sense, represent important distinctions of formal structure. Far more crucial is the text-status of the vehicle and the tenor. For now: if V and T are both present in the text (and ‘the moon’s a balloon’), I’ll refer to this as a ‘metaphor’; if only V is text-present and the T is to be inferred (by ‘the white balloon in the night sky’), I’ll refer to this as an ‘enigma’ – Greek for ‘riddle’, which is all it is. Some authors refer to this formula as that of ‘symbol’; however I use ‘symbol’ to refer to purely to a ‘vehicle with an abstract tenor’ (although I believe these are not tenors at all, since symbol is a wholly different tropic category).

There is a third category, which I refer to as the ‘isologue’, where V and T are text-present, but the dominant component (i.e. the tenor) cannot be established – usually because both are syntactically unconnected, and often both present in the literal frame of the poem. It may sound like a minor trope-type – but it is extremely common, if monumentally under-identified. (In the isologue, there really is no V or T, nor any limit on the number of mapped elements.) ‘Tonal isologue’ – details which overlap in shared tone – may be the most common trope of all. Isologues which share formal or dynamic properties are also common:

We’ll now look at a metaphor ‘in the field’, and give it a more technical consideration.

I wandered lonely as a cloud

That floats on high o’er vales and hills,

When all at once I saw a crowd,

A host, of golden daffodils16

I think most readers would agree that Wordsworth’s is a reasonably effective simile; at least it seems a uncontentious example to work with. A secondary connotation linking men (tenor) and clouds (vehicle), lonely, is declared in the text; as we’ve seen, original mappings have to be explicitly declared, and often these hinge on shared secondary connotations. Secondary connotations have a degree of subjectivity, are not central to the definition of either term, and are peripheral within their broad domains. Not that clouds are really lonely – this is just the poet’s pathetic fallacy – but a core attribute of ‘lonely’ is ‘isolated’, which is surely the image we’re presented with here. By contrast, ‘I wandered like a cloud’ wouldn’t narrow the sense in the way the author requires.

To assume a text-absent shared secondary attribute is self-evident or even ‘obvious’ is, one of the classic late-modernist and postmodern procedural errors in metaphor, and assumes that the connections between discontinuous statements can be clearly inferred.17 The frequent claim is that such tactical elisions permit the reader a more democratic means of interpretation, or that the poet is graciously crediting the reader with the intelligence to divine these implicit connections. Too often this just overestimates the reader’s powers of ESP, and results in ‘reader’s paranoia’ (or more precisely a localised and temporary version of clinical ‘cryptosemia’, the identification of signs not perceptible to others). The situation has arisen because poets underestimate a sign’s ability to generate a great many secondary attributes in the absence of any explicit narrowing context or fondement – here, a clear thematic domain of the kind often absent in the late modernist poem – and the results are invariably polysemy gone berserk. (Beware poets who, despite their having intended a specific meaning, will nonetheless justify the confusion their incompetence has generated by telling you that no interpretation can be ‘wrong’.)18

Returning to our man-cloud: guided by the additional information provided by the text-present ‘lonely’, we can probably all agree to identify attributes of ‘solitariness’, ‘freedom’ and ‘roaming’ as the common ground in this metaphor. (‘Lonely’ is really a form of transferred epithet from ‘I wandered’; the cloud is not ‘lonely’, but we reach a more cloud-native attribute – ‘solitary’ or ‘single’ – by the usual process of making rapid metaleptic links in the abstract domain, where we jump from concept to concept until we land on one closely linked to our target element.) But the ground tells us nothing additional, and in a sense merely focuses our minds on an attribute already present within either domain. A man and a cloud can both be, in their ways, solitary and alone. The real poignancy comes with something I call the ‘concealed tension’.

The ‘manifest tension’ is that between the big central denotations, if you’ll allow me that happy fiction, of ‘cloud’ and ‘man’; a vaporous, floating rain-giving thing, and a solid, walking bloke. This gives us the requisite difference between T and V that we need to make the metaphor arresting and surprising: ‘Wow! Who’d have thought that these two things were in some sense the same?’ is the kind of reaction we often hope to provoke in the reader. However, the concealed tension resides in the attributes they either almost share but do not, or do share – but possess in a radically different quality, kind or degree. It’s at the very edge of their overlap, in the fuzzy area between ground and tension, that we unconsciously sense them (hence ‘concealed’). The man cannot quite rise, although the poet seems to express an abstract yearning to; a cloud is almost weightless; a man is not. The cloud is unconstrained in 3D space; a wandering man is confined by his plane-bound paths and routes, even though he might desire it otherwise. Poetic metaphor expresses a desire that something were otherwise, and gains its poignancy through the literal impossibility of that desire being realised – while still appearing to hold out the imaginative possibility that it might be. (In a sense, it is expressive of ‘a rage against category’.) Isolated textbook examples won’t help us here, since the concealed tension is often generated by the thematic domain. In ‘the dinner plate is a moon’ our manifest tension is, well, that plates are small and close-at-hand and ceramic, and the moon is huge and distant and rocky; it’s hard to say what the concealed tension could be. But suppose that the poem is about dining alone after a failed marriage: the plate and moon are round and white; though the plate is not remote, isolated, freezing or barren, it conceivably could be – and these hidden poignancies are, to some degree, capable of being drawn forth by the miserable context. (The word ‘lonely’ in ‘I wandered lonely as a cloud’ accomplishes just that.) The concealed tension is an area of conceptual flux, one in which a core or secondary attribute of the V which is not shared by the T is vestigially transferred to it through a determined, context-directed act of the imaginative will. Rich thematic domains and contextually apposite metaphors generate these unstable and poignant associations; simplistic domains and inapposite metaphors do not.19

The ground is not just the place where V and T share attributes. The ground opens up a conduit whereby the tenor can be infected with aspects of the vehicle, and vice versa. It is therefore an active field, and directs sense from the V towards the T and the T towards the V, and is the emergent structure of an original and dynamic conceptual blend. Too often we forget that this dynamic is bidirectional. Let’s flip our earlier example, and take ‘the moon is a dinner plate’. The moon is round and white, and these qualities will have directed our choosing ‘plate’ as a vehicle; but via the ground, the plate returns its own connotations to reshape the idea of ‘moon’ itself, and might draw out quieter aspects of the T like ‘something that comes out in the evening’, as well as more secondary attributes such as ‘shininess’. Releasing these possibilities is a matter of focusing the reader’s attention and intrigue in the right way, but also encouraging them to treat the metaphor as a compound sign.

Such metaphors can be very rich, and form their own emergent domains. This is why the current use of the terms ‘source and target’ in semiotics to replace V and T respectively makes very little sense – and none at all in a poetic context: it’s a lousy metaphor. Things just aren’t that simple. In a compound sign, sense flows in both directions from either component, not in a simple vector from one to the other. (This can be most clearly seen in the sub-trope of isologue.) In Wordsworth’s poem, ‘man’ is mapped to ‘cloud’, and ‘cloud’ to ‘man’. However, whether man or cloud is tagged V or T is an issue of a context-dependent polarity, i.e. whether cloud or poet belongs to the ‘literal domain’ (this is a matter of deixis, discussed elsewhere). This is, of course, crucial, and the metaphor cannot be inverted without changing its sense; how we present a metaphor tells us which conceptual frame is dominant. Nonetheless, the various syntactic forms the metaphor might take don’t eliminate the backwash and undertow of meaning running from V back to T. Each term colours the other. The overinterrogation of the sign implicit in the poetic ‘reading contract’ amplifies and complicates sense-flow massively, and the result is a conceptual blend.

However, I’d suggest that we have become inured to the effects of our presentational defaults and the power-relations they propose:

Here, the tenor of the sloe gin perhaps also contributes a little tipsiness and deliriousness to the vehicle of Betelgeuse – but primarily the old red giant is invoked just to lend its eerie colour, and perhaps a hint of its remote strangeness; it does so by seeing its own sense-potential dramatically constrained. (Heaney being Heaney, he may well have had the final raising of the glass in mind too: Betelgeuse means ‘the hand of Orion’.) As superbly effective as I find this example, I do sometimes worry if we can persist exclusively with this kind of metaphor – the kind where we turn the world into a handy means of lending our human concerns some shape, flavour, colour, drama, dignity; at times it seems yet another way of indulging our own hegemonic delusions.

Nativity and alienity

Too often we fail to take account of the metaphor’s place within the context of the poem itself. A metaphor’s success depends on more than its originality; it must also be apposite, and like any other detail, serve the emergent, rule-based environment of the thematic domain. This all hinges on the nativity or alienity of its vehicle relative to that domain. As poets, we generally try to maximise the metaphor’s effect by placing the vehicle at the optimal distance from the poem’s thematic concerns. A vehicle imported from too distant a sphere relative to the thematic domain might be flashy and impressive – but there’s a very good chance that, having nothing much to do with the poem, it will then rip it apart. The deliberate use of this practice was the aesthetic of ‘Martian poetry’, a movement associated with the literary critic Craig Raine.21 The school only ever had a couple of adherents, but was briefly influential; this was a tribute not just to the moribund state of poetry in England at the time, but to the atrophying of its visual imagination, and ‘Martianism’ at least performed the service of reminding poets that part of their job was to look afresh at what was in front of them. The term ‘Martian’ is still a useful way of describing a flamboyant comparison, often in the form of a ‘visual pun’, that comes at the expense of the poem’s larger integrity. To give an example from Raine’s own writing: in ‘In The Kalahari Desert’, a poem about the fate of the English missionaries, we find the clever and striking image ‘a glinting beetle on its back / struggled like an orchestra / with Beethoven’.22 However, the fact that the image is strenuously justified by the lines that flank it only raises our doubts over its authenticity: the line is preceded by the rather nervous ‘the wilderness was full of home’, and then followed by the suspiciously well-timed corroboration, ‘The Hallé, / Isabella thought and hummed’. Such rickety accommodations seem to point to the metaphor being hatched outside the poem’s own thematic domain, and then let loose into the text. The missionaries may well have been lovers of classical music back in Blighty, but to claim that ‘orchestra’ is an indexically representative member of the conceptual domain ‘English home’ is stretching it. (This shoehorning is further given away by the partly transferred quality ‘struggled’, used here to increase the ground of the metaphor: the upturned beetle is not only like an orchestral violin section, but also even more like an incompetent one; in the larger context of the poem, this seems an odd thing to get nostalgic for.) The image would have been just as happy or unhappy in a thousand other poems, and the vehicle has been imported from too distant a domain; relative to the theme, it suffers an uncomfortable degree of ‘alienity’. (Tonal dissonance between theme and tenor can be just as much of a problem, e.g. ‘Roger spat into the fire, / leaned back and watched his phlegm / like a Welsh rarebit / bubbling on the brands …’ Ah, home.) There is nothing intrinsically bad about Raine’s poetic thesis here; indeed there is a decent point to be made about the colonial mindset and its imaginative hegemony – but because the thematic and tonal consistency of the images is disregarded, the evidence looks merely ‘planted’, and the argument can’t build. With such effects, the reader must leave the spell of the poem in order to applaud them, and then somehow fight their way back in – which can be as difficult as finding your way back into a dream after the dog has woken you up.

The middle-way, similarity-and-difference rule says the comparison should be close enough to the poem’s concerns to draw on its argumentative or thematic circuitry, and distant enough to arrest the reader with its felicity and originality. The following deliberately simplified example will illustrate the point. Let’s propose a theme, and say … We have a poem in which a poor man who has just cleaned up his act – he’s quit drinking, and saved a little money – is taking a holiday on a ferry with his kid; the tone of the poem is a happy and redemptive one. The guy is standing by the guard-rail, looking out to the horizon as his young child falls asleep on his shoulder. He is watching the sun go down. Now: suppose that you have decided to use a metaphor to describe the sun, and present it in the form of an enigma (i.e. where you omit the tenor and just state the vehicle). Which of the following fits the poem best? (a) ‘Derek watched the great, suppurating, angry boil sink into the sea’; (b) ‘Derek watched the plump satsuma sink into the sea’; (c) ‘Derek watched the huge gold coin sink into the sea’; (d) ‘Derek watched the giant happy beachball sink into the sea’.

Setting aside the fact that they’re all pretty bad, this exercise will nonetheless supply us with a little cline for metaphor-fail. Example (a) clearly dissonates; it’s wholly alien to the thematic domain, and imports a tone which clashes badly with the redemptive, serene one we’ve carefully established. Example (b) is better, but the tone and content of the vehicle is really incidental to the thematic domain, and does little to enhance or deepen it. Many poets know to avoid type (a) but cannot resist type (b) (see our ‘Martians’) because it forms a potentially rich category of striking and effective comparison – but only if no attention is paid to context; these tropes may not actually dissonate, but neither are they particularly apposite. Nonetheless, we might still argue that such a metaphor is effective: it gives local pleasure, it lightly disrupts the predictability of things, and it performs a decorative function. (This kind of trope is regularly favoured by ostentatiously clever novelists of the sub-Martin Amis school. Amis himself uses it skilfully.) Example (c) is our best bet; it is not only consonant with both the subject and tone of the thematic domain, it’s quietly propositional of it. It certainly avoids the sin of example (d), a type of metaphor we see far more than we should. It is less supportive or propositional of the thematic domain than it is crudely directive of it: it lacks any naturalism and sounds implausibly ‘convenient’; we might say that it was simply too native to the theme.

Staying with ‘over-nativity’: when the V of the metaphor is drawn from concerns very close to the poem’s theme it runs the danger of being misread as a literal detail, especially in a symbolic construction. In a poem about ‘fear in airports’, ‘I gripped my ticket like my boarding pass’ is a catastrophic metaphor, but making it an enigma – ‘I gripped my boarding pass’ when you meant your ticket – is just idiotic. You might think this trope would never come up, but milder versions certainly do, and too regularly. I recall a poem describing someone having an epileptic fit on a boat. They were described as flopping around like a fish – but the metaphor was presented as an enigma, i.e. ‘the landed fish flopped about the deck’, by which the poet had intended ‘the man’. But since the V was consistent with the scene’s own literal domain, there was no cue to read it figuratively – and it was simply a line about a fish. However, though deictic clarity is crucially important, it’s often good default practice to find our metaphors in the poem’s own imaginative circumstances; this has the effect of deepening the idiom of its thematic domain, and plugging the V straight into the poem’s own circuitry:

In the last line Longley hints lightly at a beautiful metaphor, ‘water is like cream’, or more specifically something like ‘water skimmed by the wind from a water-butt is like cream being skimmed from the top of the milk’; ‘cream’ is drawn from and consistent with the ‘rural, farming community’ aspect of the thematic domain. We may even vaguely sense ‘water is like cream’ is a symbol of something like ‘the paradoxical richness that can be found in apparent poverty’.

These stories must have been inside my head

That day, falling in love, preparing this

Good life; and this, this fly, verbosely buried

In ‘Bliss’, one dry tear punctuating ‘Bliss’.

(DOUGLAS DUNN, ‘Re-reading Katherine Mansfield’s
Bliss and Other Stories’)24

In this poem from Douglas Dunn’s book of elegies for his first wife, the astonishing appositeness and nativity of the vehicle ‘dry tear’ used to describe the wing of an accidentally ‘pressed fly’ hardly needs pointed out; and in the usual non-linear way, it’s possible that this very association prompted the entire poem, or was at least a major factor in the development of its theme.

Incidentally, most poets – whether or not they are making a more specific thematic point – will stick to the safe, good practice of drawing on natural-world or domestic-world vehicles (moon, grass, rose, sunlight, table, lamp, door, etc.), whose presence will rarely clash with the thematic domain. Vehicle-raiding the natural and domestic domains is something of an undiagnosed default.

Poetic metaphors are more often complex than the ones I’ve adduced here, and don’t compare substantives but peristases. They’ll rarely compare the moon and a balloon; simple examples are often handy for analysing the structure of trope, and tend to be the kinds of metaphor discussed by theorists, but the compositional reality is vastly more complex, dynamic and chaotic. Metaphors will be spun by the poet within the nascent, inchoate thematic domain of a hunch, a feeling, a tone, a ‘half an idea’ – and elaborated through the exercise of what Jung used to call the ‘active imagination’. In other words, that moon of ours will more likely be seen through scudding clouds while lodged in the branches of a bare elm, and be watched by a recently bereaved man who has lost his way home while taking a short-cut over the fields. If that image does not already constitute the poet’s objective correlative – i.e. if it does not satisfactorily symbolise the emotion guiding the poem – we may well feel the need to elaborate its metaphor; balloons and plates will not suffice. The man will look at the tree-trapped, cloud-swept moon the way he once did that big silver coin he saw deep down in that murky stream when he was a child, knowing if he could only reach it … and so on. Moreover, peristases often have attributes that are text-absent and derived purely from context, an area often entirely ignored by textbook analysis.

The act of poetic composition is a messy process, the object of which is to find out ‘the truth’; but, from a subjective perspective, ‘poetic truth’ is just the point at which the poet arrives at something they feel to be true. The method by which they get there is unique to the art of poetic composition. In poetry, language is placed under an excessive degree of formal pressure and emotional urgency, until it undergoes a kind of phase shift, as water does at freezing or boiling point. At that point of turning, the poet knows they ‘have something’.25 We pursue that solvent truth, the sudden fusing of sense, music and measure, by necessarily chaotic means – means that our neat formulae and clean textbook examples inevitably betray. The final poem is usually the result of far more radical revisions, cuts, pastes and switches than the reader or critic could ever suspect. Since the good poet tends to work by vague hunch and process, not clear idea and operation, they will often allow the felicity of a comparison itself to propose a thematic domain very different to the one they may originally have had in mind. Actually: poets sometimes have very little in mind. They generally write to find out what they think, not to ‘commit a thought to poetry’. This is because they know that the thoughts they have outside the discipline of poetry are pretty much the same as everyone else’s, and poetry provides their one opportunity to have an original one. In poetic composition, all ‘effects’ are verbs. Metaphor and metonymy are how we write poems, not how we populate them.

1 Giambattista Vico, New Science: Principles of New Science Concerning the Common Nature of Nations, 3rd edn (London: Penguin, 1999), 404.

2 In the admirably sensible scheme used by the Princeton Wordnet, nouns can often be sensibly designated as either meronyms and holonyms in their relation to one another, i.e. part of a whole, or a whole which has parts. Meronyms report to a hypernym (‘zipper’ and ‘turn-ups’ are parts of ‘trousers’), and holonyms are such if they have meronyms (‘trousers’ is the holonym of ‘zipper’; it gets messy when things do not necessary entail a hypernym, i.e. ‘trousers’ may be the meronym of ‘lounge suit’, but mostly aren’t). If words share the same hypernym they are ‘coordinate terms’, but note that this does not make them synonyms (‘Chihuahua’ is not a synonym of ‘wolf’. If you believe that words are continuous with ‘reality’ then there are no true synonyms, of course; a different word insists on representing a different concept, however slight the difference is.) The equivalent of the meronym in verbs is the troponym, which means that troponym X is a manner of the hypernym X (‘mumbling’ and ‘shouting’ are ‘kinds of speaking’); the causal equivalent of the meronym is ‘entailment’, meaning that activity X is unthinkable without Y (‘whispering’ necessarily entails ‘speech’; ‘dreaming’ implies ‘sleep’). See George A. Miller, Christiane Fellbaum, Randee Tengi, WordNet: A Lexical Database for English (Princeton, NJ: Princeton University Press, 2015), https://wordnet.princeton.edu/wordnet.

3 Any time spent in New York will reveal irony as a modality, not a trope. So fundamental is the ironical stance to the cultural life of that great city that it amounts to a dialect; within it one can still be moving, funny, open, generous, sincere, and – pricelessly – ironic. I confess this is one reason I love the place, as this elevation in the phatic irony-floor strikes me as real human progress.

4 Roman Jakobson, ‘The Metaphoric and Metonymic Poles’ (1956), in Metaphor and Metonymy in Comparison and Contrast, eds René Dirven and Ralf Pörings (New York: Mouton de Gruyter, 2003), 41–8.

5 Steven Pinker, Words and Rules: The Ingredients of Language (London: HarperPerennial, 2011).

6 Gideon O. Burton, ‘Silva Rhetoricae’, http://humanities.byu.edu/rhetoric/Figures/P/peristasis.htm.

7 Despite the fact that the category of ‘the image’ has been under assault for several decades (P. N. Furbank’s 1970 Reflections on the Word ‘Image’ being one notable critical intervention) it retains a wholly underserved currency, and is as cheerfully thrown around in academic criticism as it is in the poetry workshop. It tends to forge a false link with the idea of ‘the imagination’, with which it had little connection even for the Romantics. Even its use in describing visual effect is hopelessly narrow, and it reflects neither how we now understand our brains to process visual stimuli (i.e. through a complex process of interpretation and symbolic distillation) nor the fact that in poetry, ‘the visual’ is a highly dynamic and sensually porous domain. Even though ‘image’ and ‘metaphor’ are often discussed as if they can be cleanly distinguished (the first is too poorly defined to be distinguished from anything), the hegemony of ‘the image’ has meant that too many poets conceive of metaphor visually by default – even though it is anything but. If you need a handy example of a perfect sonic metaphor – take the British jazz slang name for the vibraphone, ‘the haunted milk-float’. Lychees taste like elderflower, gorse smells like coconut, and drizzle on our skin feels like pins-and-needles. Moreover – if more subjectively – being in love might feel like stigmata, and the airport check-in remind you of the Bardo states. Not only do the non-visual senses also have their own axes of selection, but those senses – including those which deal in ‘tone’ and ‘abstract concept’ – are all wildly interfused. The image not only lies about this fact; it has the habit of censoring it.

8 This is true of an ‘asymbol’, i.e. a pure symbol, too: the generic domain will generally leave ‘red light’ meaning ‘warning’ or ‘danger’– but in the specific qualifying context (sometimes called the ‘fondement’ in semiotics) of traffic, of studio recording, of prostitution, of weather, etc., it will take on a narrower and sometimes different sense. Specific domains tend to narrow generic senses, but through cultural convention they often establish their own symbolical rules as well.

9 George Mackay Brown, ‘Hamnavoe Market’, The Collected Poems of George Mackay Brown (London: John Murray, 2005).

10 For the most part, I have avoided making much reference to the ‘tropes and schemes’ of traditional rhetoric. In the vicious logical rigour of their various classes and divisions (I jest), they recall nothing so much as Borges’s infamous ‘certain Chinese encyclopaedia’, the ‘Celestial Emporium of Benevolent Knowledge’ (‘[…] (g) stray dogs, (h) those that are included in this classification, (i) those that tremble as if they were mad, (j) innumerable ones, (k) those drawn with a very fine camel’s hair brush, (l) others […]’, etc.) – Jorge Luis Borges, ‘The Analytical Language of John Wilkins’, Other Inquisitions (1937–1952), trans. Ruth L. C. Simms (Austin, TX: University of Texas Press, 1964). One might argue that the identification of such effects in one’s own writing and in the work of others is the first step toward their effective use – but as far as poets are concerned, all techniques should be learned with a view to forgetting them. This is especially the case with the dreaded Figures of Speech, as their self-conscious deployment tends to sounds stagy in the extreme. A ‘reasonably comprehensive list’ seems not to exist; just when you think you’ve covered one area and are leaning in more closely to confirm as much, its elements proliferate before your eyes like the Mandelbrot set. It’s therefore difficult to draw the line, and decide when one has slipped from the identification of useful effect and into the geeky pleasures of pointless taxonomic distinction. (In his highly entertaining 60 Ways to Turn a Phrase [New York: Routledge, 1995], Arthur Quinn deliciously lists hypozeuxis: ‘the refusal to use a zeugma when you could have’.) Should anyone care to pursue this matter further, please consult endnote 10, where I have placed a list of most of the classical figures, schemes and tropes still in regular use; though the reader would be better turning to the standard contemporary reference, Richard A. Lanham’s splendidly sane and useful A Handlist of Rhetorical Terms (San Francisco, CA: University of California Press, 2013).

11 Ray S. Jackendoff, Semantic Structures (Cambridge, MA: MIT Press, 1992).

12 One of the most difficult rules to follow as a poet is ‘always be prepared go where you’re least comfortable or most afraid’. There are always dragons in front of the good stuff.

13 I. A. Richards, The Philosophy of Rhetoric (Oxford: Oxford University Press, 1936).

14 I have just berated a student for describing three mojitos as being lined up ‘like a row of shots’, having refused to accept his defence of ‘but they were’. I realised afterwards that I’d have been fine if, say, John Ashbery had written as much; but then I’d have assumed he was being funny, parading his awareness of the bad metaphor – and communicating a certain louche exhaustion in his indifference to the business of finding a better one, since his point was merely to indicate the speed at which he intended to down his cocktails. However, I decided those were not quite my student’s intentions, and so felt justified in censuring him. An excellent parlour game to play among the poetically literate is to take a poem and pretend another poet wrote it, then watch how interpretations drastically change. (Some poems almost gain from the exercise – Frank O’Hara’s ‘Reasons for Attendance’ is pretty storming, and Dickinson’s ‘The Jumblies’ rounds out her personality delightfully. Eliot’s ‘maggie and milly and molly and may’ is perhaps another matter.) The style of a famous poet constitutes a specific domain, one in which we learn to read everything they write, usually in a way that confirms – even by anomalous contradiction – what we already know about them. As readers we are very loath to allow poets to change their styles, much to their frustration.

15 Paul Muldoon, Quoof (London: Faber & Faber, 2001).

16 William Wordsworth, ‘Daffodils’, The Collected Poems of William Wordsworth, ed. Antonia Till (London: Wordsworth Editions, 1994).

17 Indeed, one might say enthymeme- and anapodoton-fail are the hallmark errors of neo-modernism; though one should probably decline to.

18 In a sense this statement is true, of course, but not when it’s used to abdicate one’s responsibly to ‘as clear expression as one can manage’, regardless of the complexity of the thing expressed. My own line is that all interpretations are valid, but some can sensibly be declared stupid, erroneous or mere projections on the reader’s part: one can misread an obvious and plain sense, or narrow a sense without justification, or import a sense for which the text has given no cue.

19 I am aware that in this attempt to explain the slippery effects that lie behind the poignancy of certain tropes I have been reduced to qualifying somewhat outmoded terms into some kind of accuracy. Literary terminology does not describe cognitive processes particularly well. Old-fashioned literary analysis may be in its last throes, but ‘last throes’ are an essential part of any process: they are, for one thing, a caution against the too-hasty adoption of fatally flawed new terminology (see my comments on source and target).

20 Seamus Heaney, Station Island (London: Faber & Faber, 1984).

21 See Craig Raine, A Martian Sends a Postcard Home (Oxford: Oxford University Press, 1979).

22 Craig Raine, ‘In the Kalahari Desert’, Collected Poems 1978–1998 (London: Picador, 2000).

23 Michael Longley, Collected Poems (London: Jonathan Cape, 2007).

24 Douglas Dunn, Elegies (London: Faber & Faber, 2001).

25 Naturally, poems written ‘from life’ – from its immediate frustrations, angers, joys and tragedies – will often have greater emotional urgency than those we merely ‘make up’. Emotional speech has a different musical and rhythmic quality to non-emotional speech (and a quantifiable, characterisable one at that); there are, to be cold about it, some good technical reasons why ‘the confessional mode’ might sometimes make for a superior poem. Alas, ‘life-poems’ – however necessary they feel to the poet, or genuine the assuagement they bring – are often blighted by the tear-blinded distortions that accompany great emotion: an unconscious tendency to self-censorship, a coyness with the facts or the factual context that helps the reader makes sense of them, and a loss of editorial perspective. We often assume the events by which we have been most moved to have an intrinsic emotional resonance that requires only our plainest description, when in fact their intensity depended on the intricate harmonic circumstances of our lives. We either need to provide the reader with a little of that context (what we merely remember of a sad event, say, is generally not the best metonymic evidence of it; this must be sought out), or alter the facts to fit the truth. Writing with your genuine feeling but otherwise just ‘making it all up’ might often be the poet’s best strategy.