The Divine Comedy 
by Dante Alighieri, translated by Allen Mandelbaum.
Everyman, 798 pp., £14.99, May 1995, 1 85715 183 6
Show More
The Inferno of Dante. A New Verse Translation 
by Robert Pinsky, illustrated by Michael Mazur.
Dent, 427 pp., £20, February 1996, 9780460877640
Show More
Dante’s Hell 
translated by Steve Ellis.
Chatto, 208 pp., £15.99, March 1994, 0 7011 6127 2
Show More
Show More

There are several different things one can be aiming at in a verse translation, leaving aside the genre known as ‘Imitation’, in which poets like Samuel Johnson, Ezra Pound and Robert Lowell have done such marvellous things. A verse translation may aim to be an independent modern work in its own right. Or, I ought rather to say, this is what some famous and admired translations have in fact been. If you took Pope seriously as to the degree of fidelity required of a translator of Homer (‘to copy him in all the variations of his style and the different modulations of his numbers ... not to neglect even the little figures and turns on the words, nor sometimes the very cast of his periods’), his own Iliad would give you rather a turn. Then again, a verse translation may plead to be read purely as a translation: as a compromise and a substitute, offered as such and hoping occasionally, by some good fortune of language, to reach transparency. (The transparency will, of course, be illusory, since the translator will of necessity be working within a different verse-form from the original; verse-forms do not transplant. But then, illusion is all any reader need ask for.) But thirdly, you can have a translation which is intended neither as a work of art nor as a substitute for a work of art, but as a form of exposition – in the way that you might teach rifle-drill with a dummy rifle. This is what Ezra Pound meant when he said of his brilliant and bizarre version of Guido Cavalcanti’s canzone ‘Donna mi priegha’: ‘As to the atrocities of my translation, all that can be said in excuse is that they are, I hope, for the most part intentional, and committed with the aim of driving the reader’s perception further into the original than it would without them have penetrated.’ Of these three new translations of Dante (not all of them quite new, for Allen Mandelbaum’s was first published ten years or so ago), Mandelbaum’s and Pinsky’s belong firmly in the second class, whilst Ellis’s, which makes a point of the modernity of its idiom, aspires perhaps a little to the first class.

The 20th century has been awash with Dante translations: Laurence Binyon’s in the Thirties, Dorothy Sayers’s in 1949, John Ciardi’s in 1955, Mark Musa’s in 1971, C.H. Sisson’s in 1980, not to mention an Inferno by assorted poets broadcast by the BBC’s Third Programme in 1961, the prose versions by John Sinclair and Charles Singleton, and an admirable translation – again in prose – by Robert Durling, which will be published in this country by Oxford next year. Behind this we may perhaps see the influence of Eliot’s famous Dante essay of 1929 and of his own verse, which was haunted by Dante more than by any other poet. There was also the impossible challenge offered by the Dantesque episode in Little Gidding. Eliot wrote: ‘This section of a poem, not the length of one canto of the Divine Comedy, cost me far more time and trouble and vexation than any passage of the same length that I have ever written.’

There are poets whom one should probably not try to translate at all, for instance Mallarmé and Racine. Dante is certainly not in that category, for occasionally someone, just for a moment, brings it off. Shelley comes near to doing so in his version of ‘Matilda Gathering Flowers’ from the Purgatorio; at least he achieves one or two very beautiful effects.

Yet were they [the branches] not so shaken from the rest,
But that the birds, perched on the utmost spray,
Incessantly renewing their blithe quest,

With perfect joy received the early day,
Singing within the glancing leaves, whose sound
Kept a low burden to their roundelay.

Moreover, in The Triumph of Life Shelley not only matched the fulgorous prosopaeia of the Divine Comedy, he managed to create a sustainable Dantesque verse style.

It is necessary to be clear about where the main difficulties for a translator of Dante lie. The hendecasyllabic (11-syllable) line, which is the staple of Italian verse, has very little in common with the English iambic pentameter. It is true that, as with English metre, it is based on stress, but, unlike the pentameter, it has no feet. Dante makes this point explicitly in his Latin treatise the De Vulgari Eloquentia. ‘We must not omit to mention,’ he writes, ‘that we take feet in a sense contrary to that of the regular poets [i.e. classical writers], because they said that a line consisted of feet, but we say that a foot consists of lines.’ The hendecasyllable line has a heavy stress on the tenth syllable, and it has a caesura, but beyond that it is a very fluid and flexible affair. Often one’s ear detects three main stresses (‘Nel mézzo del cammín di nostra víta’), but all sorts of other stress-patterns are also possible. Italian words could be said to fall more easily into verse because of their lilting stress-pattern, and Dante mentions the word sovramagnificentissimamente as one which makes a perfectly acceptable hendecasyllable line in itself.

On the subject of the hendecasyllable, it is worth noticing something that Eliot did in his Dantesque passage in Little Gidding. In Dante’s terza rima one from time to time finds a ten-syllable line, or rather a group of three such lines (versi tronchi), rhyming on a single rather than the normal double syllable. Eliot, who is writing terza rima without rhyme, takes a hint from this, and makes it a formal rule to alternate an 11-syllable line with a ten-syllable one. This, as an attempt to domesticate Italian verse, is cunning; for often his ten-syllable lines, in contrast to his 11-syllable ones, will read more or less like an English iambic pentameter (for instance: ‘The eyes of a familiar compound ghost’; or ‘We trod the pavement in a dead patrol’), and this helps to bring home the quite different wave-pattern, the more springy quality, of the 11-syllable lines.

Unrhymed terza rima is of course a paradox, though it is the solution adopted by two of these translators, and unquestionably a lot is lost by abandoning rhyme. Dante appears to have invented terza rima – a form in which a new set of rhymes has begun before the preceding one has been completed – and it has, in his hands, a wonderful knitting or weaving effect, a combination of stasis and forward motion. Often, too, rhyme gives his lines the effect of a charm or incantation, a necromantic quality which arouses superstitious awe.

Eliot, according to his own account, adopted the unrhymed form in the Little Gidding passage because of the obtrusiveness of rhyme in English and because of the ‘shifts and twists’ to which translators are reduced in their attempts to reproduce Dante’s rhymes. Indeed, he claimed that his reason for alternating 11-syllable and ten-syllable lines was that it was ‘the nearest way of giving the light effect of the rhyme in Italian’.

Ezra Pound, writing about translating Dante and Cavalcanti’s canzoni, argues somewhat differently. ‘It is not,’ he says, ‘that there aren’t rhymes in English; or enough rhymes or even enough two-syllable rhymes, but that the English two-syllable rhymes are of the wrong timbre and weight. They have extra consonants at the end, as in flowing and going, or they go squashy; or they fluff up as in snowy and goeth.’ The point, according to Pound, is not that one language cannot be made to do what another has done, but that ‘it is not always expeditious to approach the same goal by the same alley.’ He was perfectly aware of what was lost by rejecting rhyme, and when Laurence Binyon was translating Dante, using rhyme, he was happy to encourage him. ‘MAGNIFICENT FINISH!’ he wrote to Binyon in May 1938 about his Purgatorio. ‘Utterly confounds the apes who told you terza rima isn’t English.’

It is worth observing that Steve Ellis, in the Introduction to his Hell, says that H.F. Cary’s early-19th-century blank verse translation and Dorothy Sayers’s rhymed version are the only ones he enjoys reading. He did not choose to use rhyme himself, he says disarmingly, mainly because, unlike Tony Harrison and other of his contemporaries, he has never been very good at it.

Robert Pinsky, who feels all the value of Dante’s interlocking rhymes, takes the problem more to heart. His solution is to adopt a more flexible definition of rhyme, employing a strict system of consonantal rhymes (‘tell/feel/well’, ‘sleep/stop/up’ etc.), while deliberately eschewing assonantal or vowel rhymes, like ‘claim/feign’ or ‘state/raid’. Disyllables he rhymes on the same consonantal principle (‘bitter/enter/blunder’), remarking, rightly, that true disyllabic rhyme tends to sound comic in English and to smack too much of the limerick. He says that he actually prefers consonantal rhymes like ‘swans/stones’ to ‘hard-rhyme’ combinations, and claims – but here I think he is stretching things – that his consonantal rhymes are ‘as “like”, perhaps, in the context of English and its great sprawling matrix of sounds, as are “terra/guerra” or “belle/stelle” in the tighter Italian fabric’. These are the reasonings of someone with an ear, and who sees the need to give himself rules; and within his limits Pinsky is a distinctly impressive translator.

For all recent Dante translators, the question of vocabulary has been a burning one. C.H. Sisson wrote that Binyon’s translation, and even more Dorothy Sayers’s, were not only not written in the English of their own time, they were not really written in any language. Steve Ellis takes much the same view. Most versions of Dante, he writes, ‘employ no particular language at all: rather an odd mix of the bookish and the self-consciously demotic, a strange hybrid that lives nowhere off the page (nor frequently indeed even on the page)’. They ‘convert a lively and fast-flowing original into something much more plodding, formal and prolix, keeping Dante, as a venerable “classic”, safely within the purlieus of the academy’. This is why, he says, he has chosen to produce a colloquial version, which tries to recapture some of the ‘vigour and directness’ of Dante’s original.

This horror of archaic or stock-poetic diction goes back, in part, to Ezra Pound – the Pound, shall we say, of the Confucian Odes – but in fact he was very open-minded on the subject of diction. He teased Binyon about his use of ‘even as’ and ‘doth’ but said he was only ‘swatting’ them when they particularly hindered him. ‘In the long run you flow sufficiently to carry one over them.’ In his own case, the question of diction had been a matter of a long development. There was a nice scene, he liked to remember, from the time when he was first trying to teach himself to use modern speech. He had gone to see Robert Bridges, the Poet Laureate, and Bridges had gone through his Personae and Exultations, praising (to his horror) every archaism, exclaiming gleefully, ‘We’ll git ’em all back; we’ll git ’em all back.’ Pound’s own watchword in translation was resourcefulness. The thing was to find the right means, whatever this might be – it might be totally unexpected or even outrageous – for the peculiar task in hand.

It became an obsession with Pound that Milton, by his ‘latinisation’ – that is to say using an uninfected language as if it were an inflected one, where each word would have ‘a little label or postscript telling the reader at once what part it takes in the sentence’ – had done great harm to English verse. ‘Not only does such usage – with remnants of Latin order – ruin the word order in English,’ he wrote, ‘but it shows a fundamental mis-comprehension of the organism of the language.’ One might have supposed from this that he would have come down very heavily on Binyon for using inversions or disrupting normal English word order; but not so. It was, he said, a question of choices, and should not be a matter of rules: ‘A younger generation, or at least a younger American generation, has been brought up on a list of acid tests, invented to get rid of the boiled oatmeal consistency of the bad verse of 1900, and there is no doubt that many young readers seeing Binyon’s inversions etc, will be likely to throw down the translation under the impression that it is incompetent.’ But this, he said, was to be short-sighted. ‘The fact that this idiom, which was never spoken on sea or land, is NOT fit for use in the new poetry of 1933-4 does not mean that it is unfit for use in a translation of a poem finished in 1321.’ In the detailed comments he sent to Binyon, he often lambasted him for his inversions, but only because these particular ones were pointless, not because natural word order was sacrosanct.

The good sense of this becomes clear on studying these three new translations. I think it could be said of all of them that they disregard, or at least downplay, something of the greatest importance in Dante. Let us think of a line in the great Paolo and Francesca episode in Canto 5 of the Inferno (Francesca is speaking): ‘Amor, che a nullo amato amar perdona.’ The meaning of this (‘Love, which pardons no loved one for loving’) is subtle and, in its context, very poignant; but the particular force and beauty of the line lies in the particular way in which the words have been arranged, or rearranged and held in tension together – for instance, the riddling effect of the squeezing together of amato and amar. To a lover of Virgil and Horace this would hardly seem worth mentioning, it being so essential to what those poets are doing; and it is essential with them because of the extreme freedom of word order in Latin. Dante’s Italian, being an uninflected language, does not have the same freedom of word order, though it has distinctly more than modern Italian; and Dante, as a passionate admirer of the Latin poets, very naturally prizes and emulates this aspect of their art. I am not speaking of some trick performed occasionally or in a spirit of antiquarianism, but of what is almost the main staple of his verse: the juggling or rearranging of words and clauses, and the remodelling of syntax (together of course with exact acoustic shaping) to the end of adding extra meaning beyond the literal.

Allen Mandelbaum renders Francesca’s line as ‘Love, that releases no beloved from loving’; Robert Pinsky, as ‘Love, which absolves/None who are loved from loving’; Steve Ellis, as ‘Love insists the loved loves back.’ These all seem to leave out half of Dante’s meaning: perdona, I would have thought, is meant both in the sense of ‘absolves from’ and of ‘pardons for’, an ambiguity which seems to me most suggestive and touching. But, even leaving that aside, the test arouses the gloomiest fears about the project of translating Dante. Could anyone even have hoped to match the grace, precision of effect and acoustic charm of the original line? Pinsky at least achieves something acoustically passable – as often, I like his version best. As for Ellis’s, one either has to say that it is perfectly dreadful, or, which would be fairer, that he has deliberately renounced any attempt to reproduce the plastic qualities of Dante’s verse.

One might have thought that allegory, and the fourfold system of levels of meaning, as described by Dante in his famous letter to Can Grande, would, though obviously important, have no particular bearing on translation. But this would be a mistake. For the process I have described as ‘adding extra meaning beyond the literal’ is at work all the time in the Comedy, in a way that may or may not correspond to any of Dante’s three higher ‘levels’ but certainly counts enormously. Let us think of the great canto of the suicides or ‘Violent against themselves’ (Canto 13). The conception could not be more powerful. The spirits of those who kill themselves are tossed like seeds into the ‘seventh gulf’ of hell, growing up into withered trees, nested in by Harpies, and unable to speak until someone lacerates their foliage. Nor is this the end of their punishment: even at the Last Judgment, when they have dragged their bodies back from earth to the sorrowful wood, they will not, because of their sin, be allowed to reassume them. The bodies will merely be hung on their branches: ‘ciascuno al prun dell’ombra sua molesta’ (each on the thorn of its own molesting shade). The extra emblematic meaning of this line is not difficult to spell out – someone so self-hating as to kill himself can be imagined as having already spent his life ‘on the thorn of his own molesting shade’ – and one gets something out of it even in prose, though the cunning syntax of Dante’s verse is needed for the idea to blossom. But Mandelbaum renders this as ‘each on the stump of its vindictive shade’, which seems simply to miss the point; and Pinsky’s version, ‘Fixed on the thornbush of its wounding shade,’ though much better, is just a bit smudged and lacks the epigrammatic quality of Dante’s wonderful line. Ellis’s ‘each one on his own soul’s thorns’ catches this quality better.

In the opening of the next canto, part of the punishment of the ‘violent against themselves’ is that their tree form is at the mercy of all comers, and we see a pair of sinners, pursued by black hell-hounds, crash into one of them (a Florentine), drawing blood from him and scattering his branches. The opening words of the next canto are ‘Poichè la carità del natio loco/mi strinse, raunai le fronde sparte,/e rende’ le a colui ch’ era gia fioco’ (Because love of my native place constrained me, I reunited the scattered leaves, and returned them to him who was already hoarse), words which seem curiously more moving, more expressive of a ceremonious tenderness, than the translations. I think it may be because the translations render ‘raunai’ merely by ‘gathered’, whereas ‘reunited’ echoes the note that has been quietly sounding thoughout the previous canto: that these sinners are the self-divided and disunited.

A similar little point arises later in the same canto. The travellers come to a terrifying red stream, and Virgil explains to Dante that this and other rivers of hell are formed from the tears that fall from the broken body of the Old Man of Crete (which, made of different metals in descending order from gold to iron, allegorises the history of fallen mankind). Dante then asks why, if this stream derives (deriva) from our world above, he and his companion have not come across it till now – his words tacitly relating these tears of blood to their human origin. The translators lose this by rendering deriva merely literally, as ‘follow such a course’, ‘fall’ or ‘comes’.

Another passage earlier in this canto gives the translator every right to despair. In a way it seems tantalisingly easy, for one can form a fairly clear idea of what one is aiming at, but it is the kind of effect for which, it seems, there simply cannot be a substitute.

D’anime nude vidi molte gregge,
  che piangean tutte assai miseramente,
  e parea posta lor diversa legge.

Supin giaceva in terra alcuna gente,
  alcuna si sedea tutta raccolta,
  ed altra andava continuamente.

Quella che giva intorno era il piu molta,
  e quella men che giaceva al tormento,
  ma piu al duolo avea la lingua sciolta.

The thing to do is get it all entirely uncluttered, so that simple adverbs (miseramente, continuamente) can, by their placing and plangency, do all the work. Pinsky’s version, though it is the best of the three, gets nowhere near.

                     It was a great assemblage
Of naked souls in herds, all of whom mourned
Most miserably and seemed to be subject
To different laws. Some lay upon the ground,

Supine; some sat hunched up; while others walked
Restlessly about. It seemed that those who moved
Were the most numerous, those who lay abject

In torment, fewest – but it was they who grieved
With tongues most loosened by pain.

To begin with: ‘hunched up’ is too physical for raccolta, which conveys a moral and psychological turning-in as much as a bodily one. But more important, one has simply lost the extreme shapeliness and mournful, musical, sighing plangency that Dante wanted, which is the prerequisite for various hints of additional meaning. From Pinsky’s version one can in a matter-of-fact way grasp how the three types of sinner are distinguished by their behaviour – the violent against God lying on their backs (facing the heaven they have defied), the violent against art (i.e. usurers) grudgingly turned in on themselves, and the violent against nature (i.e. sodomites) endlessly in motion. But what one loses is the way that the sheer length of continuamente, and its placing in the line, seem to extend a special kind of sympathy to the sodomites. (It is borne out by the heroic treatment received by the sodomite Brunetto Latini in the next canto.) Similarly, one loses the flick of irony in the placing of sciolta (loose), its hint that it is characteristic of the violent against God to shoot their mouths off.

Steve Ellis’s version runs:

Different herds of naked spirits
  were all weeping wretchedly,
  apparently under different laws:

some were lying along the ground;
  some sat, their legs drawn up;
  others paced about all the time.

These last were the more numerous,
  and those least who were lying,
  though pain was on their lips more.

I suppose one could make out a case for this, as a perfectly prosaic summary, but it can’t have cost Ellis much to write; you would not say he had exactly wrestled with Dante’s original. Also, ‘and those least who were lying/though pain was on their lips more’ is lazy translatorese.

Ellis has thought through what he is doing, and he expounds his aims very clearly and reasonably. The trouble is that – for me, anyway – his theory simply doesn’t work. ‘One of the features I have been especially keen to reproduce,’ he writes, ‘is Dante’s concision and economy, his ability to compress much meaning into a few words.’ Nothing could be more true than that this is one of Dante’s great qualities; but it can only exhibit itself in a context where such an effect will be noticed. It is only one effect among many; and unless all other effects, too (for instance, certain kinds of expansiveness, and what Osip Mandelstam saw as Dante’s supererogatory and non-functional similes), are exactly brought off – unless, that is to say, the whole verse fabric is in a state of tension – there will be nothing for this pregnancy and concision to define itself against. There is a famous example of Dantesque concision in the episode of Count Ugolino in the tower (Canto 33 of the Inferno):

ed io sentii chiavar l’uscio di sotto
  all’orribile torre: ond’io guardai
  nel viso a’ miei figliuoli senza far motto.

Io non piangeva, si dentro impietrai;
  piangevan elli.

The effect of the last two lines (‘I did not weep, so had I turned to stone within; they wept’) impresses every reader; but how can such a supreme rhetorical effect hope to work in a context of the kind Ellis has given it – so relaxed, so evocative of ordinariness and homely human values?

and then I heard that hideous tower
  having its door below nailed up:
  and I just looked at my boys.

I didn’t cry, I was stone inside,
  but they cried: my little Anselmo
  says, ‘Dad, what are you thinking?’

This ordinary-man tone has been established from the opening cantos. The relationship of Dante and Virgil is altogether chatty and chummy. DANTE: ‘Poet and leader/make sure I’m really up to it/before entrusting me to eternity.’ VIRGIL: ‘So now what? Why, why dither,/why feed your heart with fears?/Where’s your go, your passion?’ Ellis, speaking of Dorothy Sayers, says that ‘nothing dates more quickly than contemporary slang’; but of course he falls into the trap himself. ‘Who’s that he-man seeming to scorn/the fire?’ he asks Virgil – and ‘he-man’ is pure Thirties’ Hollywood. Moreover, he doesn’t stick to the chatty and prosaic, which would at least have been a convention, but slips disconcertingly into other registers. For instance: ‘There was no declining the dance/of their poor hands, here, there,/fending away the fresh firefalls’ – which is not bad, but is sub-Gerald Manley Hopkins.

The shining virtue of Robert Pinsky’s translation is that he has created a verse-line which catches the rhythmic pulse – grave and at the same time springy, even at times dancing – of Dante’s original. His version carries one along with a purposeful buoyancy. It is also almost always shapely, its clauses hang together easily and coherently and in close relation to verse-structure. Notice how beautifully the parenthesis is managed in: ‘Lethe you shall see, but out of the abyss:/There where, repented guilt removed, souls gather/To cleanse themselves.’ And he never drops into bathos, nor into translatorese.

His volume gives the original Italian side-by-side with the translation, and there are workmanlike notes by Nicole Pinsky; also a perceptive Foreword by John Freccero, which makes a good point about the gap between the pilgrim’s notion of the good and the author’s: ‘Irony in the Inferno arises from the discrepancy between the perspective of the pilgrim, which is much like ours, and that of the poet, who, by the journey’s end, claims to share God’s view.’ Michael Mazur’s tenebrous illustrations are sometimes most evocative and impressive. His bird’s-eye view of the infernal abyss is too murky to make much of, but he has had the neat idea of superimposing an explanatory overlay of the kind you get in satellite weather-maps.

The new ‘Everyman’ Divine Comedy, with Allen Mandelbaum’s translation, is, physically speaking, a most handsome and attractive volume; and the bibliography, chronology and notes (rather more lavish than in Pinsky’s version) are on the whole well done. Also, as a bonus we are given 42 extremely curious engraved illustrations by Botticelli, done for Lorenzo de’ Medici. Eugenio Montale’s Introduction, first delivered at a Dante conference in 1965, is a bit too plummy. ‘Dante brought the Middle Ages to a close.’ (What could that mean?) Ezra Pound’s Cantos ‘contain all that can be known about a disintegrating world’.(All?)

As for Mandelbaum’s translation, in un-rhymed terza rima, it is sober and decent, free from all archaisms and pretentiousness, and makes a virtue of directness. Indeed, it suffers somewhat from too much directness. The lines, despite their abundance of feminine endings, are essentially just iambic pentameters, and this, combined with such very straightened-out syntax, makes in the long run for a rather plodding effect. And Mandelbaum is inclined to slip into translatorese. To take an example from Canto 14: ‘but these had looser tongues to tell their torment’ is a very sketchy rendering of ‘ma più al duolo avea la lingua sciolta’, neither a natural English idiom nor an effective verbal invention. Again, in the same canto, ‘when their repented guilt is set aside’ is really not good for ‘quando la colpa pentuta è rimossa’: ‘set aside’ suggests that their guilt is being not so much absolved as simply ignored. Nevertheless his version also has many successes and felicities and – what counts especially – a sustained and consistent tone.

Send Letters To:

The Editor
London Review of Books,
28 Little Russell Street
London, WC1A 2HN

Please include name, address, and a telephone number.

Read anywhere with the London Review of Books app, available now from the App Store for Apple devices, Google Play for Android devices and Amazon for your Kindle Fire.

Sign up to our newsletter

For highlights from the latest issue, our archive and the blog, as well as news, events and exclusive promotions.

Newsletter Preferences