Writer’s Writer and Writer’s Writer’s Writer

If you go to the website of the restaurant L’Huîtrière (3, rue des Chats Bossus, Lille) and click on ‘translate’, the zealous automaton you have stirred up will instantly render everything into English, including the address. And it comes out as ‘3 street cats humped’. Translation is clearly too important a task to be left to machines. But what sort of human should it be given to?

Imagine that you are about to read a great French novel for the first time, and can only do so in your native English. The book itself is more than 150 years old. What would/ should/do you want? The impossible, of course. But what sort of impossible? For a start, you would probably want it not to read like ‘a translation’. You want it to read as if it had originally been written in English – even if, necessarily, by an author deeply knowledgable about France. You would want it not to clank and whirr as it dutifully renders every single nuance, turning the text into the exposition of a novel rather than a novel itself. You would want it to provoke in you most of the same reactions as it would provoke in a French reader (though you would also want some sense of distance, and the pleasure of exploring a different world). But what sort of French reader? One from the late 1850s, or the early 2010s? Would you want the novel to have its original effect, or an effect coloured by the later history of French fiction, including the consequences of this very novel’s existence? Ideally, you would want to understand every period reference – for instance, to Trafalgar pudding, Ignorantine friars or Mathieu Laensberg – without needing to flick downwards or onwards to footnotes. Finally, if you want the book in ‘English’, what sort of English do you choose? Put simply, on the novel’s first page, do you want the schoolboy Charles Bovary’s trousers to be held up by braces, or do you want his pants to be held up by suspenders? The decisions, and the colouration, are irrevocable.

So we might fantasise the translator of our dreams: someone, naturally, who admires the novel and its author, and who sympathises with its heroine; a woman, perhaps, to help us better navigate the sexual politics of the time; someone with excellent French and better English, perhaps with a little experience of translating in the opposite direction as well. Then we make a key decision: should this translator be ancient or modern? Flaubert’s contemporary, or ours? After a little thought, we might plump for an Englishwoman of Flaubert’s time, whose prose would inevitably be free of anachronism or other style-jarringness. And if she was of the time, then might we not reasonably imagine the author helping her? Let’s push it further: the translator not only knows the author, but lives in his house, able to observe his spoken as well as his written French. They might work side by side on the text for as long as it takes. And now let’s push it to the limit: the female English translator might become the Frenchman’s lover – they always say that the best way to learn a language is through pillow talk.

As it happens, this dream was once a reality. The first known translation of Madame Bovary was undertaken from a fair copy of the manuscript by Juliet Herbert, governess to Flaubert’s niece Caroline, in 1856-57. Quite possibly, she was Gustave’s lover; certainly, she gave him English lessons. ‘In six months, I will read Shakespeare like an open book,’ he boasted; and together they translated Byron’s ‘The Prisoner of Chillon’ into French. (Back in 1844, Flaubert claimed to his friend Louis de Cormenin that he had translated Candide into English.) In May 1857, Flaubert wrote to Michel Lévy, the Parisian publisher of Madame Bovary, that ‘an English translation which fully satisfies me is being made under my eyes. If one is going to appear in England, I want it to be this one and not any other one.’ Five years later, he was to call Juliet Herbert’s work ‘a masterpiece’. But by this time it – and she – were beginning to disappear from literary history. Though Flaubert had asked Lévy to fix Juliet up with an English publisher, and believed he had written to Richard Bentley & Sons about the matter, no such letter from Paris survives in the Bentley archives (perhaps because Lévy objected to the idea and declined to act on it). The manuscript was lost, and so – more or less – was Juliet Herbert, until her resurrection by Hermia Oliver in Flaubert and an English Governess (1980).* If Juliet’s ‘masterpiece’ were ever to resurface, we should hope it comes attached to the lost Flaubert-Herbert correspondence, and to a photograph of Juliet, of whom no known image has survived. The translation of Candide has also been lost. As for Flaubert and Shakespeare: it’s not clear if he ever achieved fluency in English – with or without Juliet – but his library contained the works of ‘le grand William’ in both English and French.

So the British reader had to wait another three decades – until 1886, six years after the author’s death – for the first published translation of Madame Bovary. It too was made by a woman, Eleanor Marx Aveling (Marx’s daughter – a quiet irony, given Flaubert’s caustic views on the Commune), as is the very latest, by the American short-story writer – and Proust translator – Lydia Davis. In between, most of the 15 or more versions have been made by men. The best-known of them are Francis Steegmuller and Gerard Hopkins; and though Steegmuller did write some fiction – including mysteries under the name of David Keith – it’s a fair bet that Davis is the best fiction writer ever to translate the novel. Which suggests a further question to the opening list: would you rather have your great novel translated by a good writer or a less good one?

This is not as idle a question as it seems. That perfect translator must be a writer able to subsume him or herself into the greater writer’s text and identity. Writer-translators with their own style and worldview might become fretful at the necessary self-abnegation; on the other hand, disguising oneself as another writer is an act of the imagination, and perhaps easier for the better writer. So if Rick Moody tells us that Lydia Davis is ‘the best prose stylist in America’, and Jonathan Franzen that ‘few writers now working make the words on the page matter more,’ does this make her better or worse equipped to render the best prose stylist of 19th-century France into 21st-century American English? Davis’s stories, typically from two or three lines to two or three pages, are decidedly unFlaubertian in scope and extent; they vary from the wry episode and rapt reverie to the slightly cute aperçu; and if there is French influence around it is from a later date (thus Davis’s ‘The Race of the Patient Motorcyclists’ seems to owe a debt to Jarry).† Her own life is clearly the basis for some of the stories, whereas Flaubert’s aesthetic was famously based on self-exclusion. On the other hand, Davis’s work shares the Flaubertian virtues of compression, irony and an extreme sense of control. And if Flaubert in his monasticism and exemplary pertinacity is a writer’s writer, Davis was described to me recently by an American novelist as a ‘writer’s writer’s writer’. That her translation of Madame Bovary was deemed worthy of serialisation by Playboy magazine – which puffed it as ‘the most scandalous novel of all time’ on the cover – is a noisier irony of which Flaubert might well have approved. The publicity sheet for the US edition calls Emma ‘the original desperate housewife’, which, cheesy though it sounds, isn’t far off the mark. Madame Bovary is many things – a perfect piece of fictional machinery, the pinnacle of realism, the slaughterer of Romanticism, a complex study of failure – but it is also the first great shopping and fucking novel.

At least none of those 15 translators has needed to recast its title; problems start, rather, with the subtitle, ‘Moeurs de province’. You can have ‘Provincial Manners’ (Marx Aveling), ‘Life in a Country Town’ (Hopkins, 1948), ‘A Story of Provincial Life’ (Alan Russell, 1950), ‘Provincial Lives’ (Geoffrey Wall, 1992), or ‘Provincial Ways’ (Lydia Davis). No one, as far as I can see, has adopted the cousinly subtitle of Middlemarch, ‘A Study of Provincial Life’. Several translators simply delete it – even, surprisingly, Steegmuller (1957); though this could have been the publisher’s initial dereliction, since more recent editions contain his nicely emphatic ‘Patterns of Provincial Life’. Subtitles can seem fussy and old-fashioned (thus the cover of the current Penguin Middlemarch dispenses with Eliot’s five subsidiary words), but omission seems a little perverse. Many translators (or publishers) also omit the next words in the novel – the dedication to Maître Sénard, who got Flaubert off the charge of outraging morality and religion when the novel, still in serial form, was prosecuted. Davis, an impressive completist, includes both this and the other, and more important, dedicatory page of the first edition, to Flaubert’s partner-in-literature Louis Bouilhet. Though – to enter the world of micro-pedantry – she (or her publisher) prints the two dedications in reverse order to the original French edition.

But then translation involves micro-pedantry as much as the full yet controlled use of the linguistic imagination. The plainest sentence is full of hazard; often the choices available seem to be between different percentages of loss. It’s no surprise that Davis took nearly three years to translate Madame Bovary: some translations need as long as the book itself took to write, a few even longer. John Rutherford’s magisterial version of Leopoldo Alas’s La Regenta – a kind of Spanish Bovary – cost him, according to his calculation, five times as many hours to translate as it had taken Alas to write. ‘Translation is a strange business,’ Rutherford noted in his introduction, ‘which sensible people no doubt avoid.’ Take a simple sentence from the first pages of Flaubert’s novel. In his early years, Charles Bovary is allowed by his parents to run wild. He follows the ploughmen, throwing clods of earth at the crows; he minds turkeys and does a little bell-ringing. Flaubert awards such activities a paragraph, and then summarises the consequences of this pre-adolescent life in two short sentences which he pointedly sets out as a separate paragraph: ‘Aussi poussa-t-il comme un chêne. Il acquit de fortes mains, de belles couleurs.’

The meaning is quite clear; there are no hidden traps or false friends. If you want to try putting this into English yourself first, look away now. Here are six attempts from the last 125 years to translate yet not traduce:

1) Meanwhile he grew like an oak; he was strong of hand, fresh of colour.

2) And so he grew like an oak-tree, and acquired a strong pair of hands and a fresh colour.

3) He grew like a young oak-tree. He acquired strong hands and a good colour.

4) He throve like an oak. His hands grew strong and his complexion ruddy.

5) And so he grew up like an oak. He had strong hands, a good colour.

6) And so he grew like an oak. He acquired strong hands, good colour.

All contain the same information, but only the words ‘he’, ‘like’ and ‘strong’ are consistent to all six. Some of the matters these translators would have considered (on a scale from pertinent reflection to gut feel) would include:

Whether to lay the paragraph out as two sentences or one; if the latter, then whether the break should be marked by a comma or a semicolon.

Whether, indeed, to lay it out as a separate paragraph anyway: thus 1) chooses to run it on at the end of the previous paragraph, which makes its summarising effect less pointed.

Whether poussa implies more vigour than the English ‘grew’: hence 4)’s ‘throve’ and 5)’s addition of the intensifying ‘up’.

Whether acquit is best rendered by a neutral word like ‘had’ or ‘was’; or whether it is a verb indicating a kind of action, intended to parallel poussa. Hence ‘acquired’ or ‘grew’ – though if you have ‘grew’ here, you need a different verb in the first sentence: hence ‘throve’.

Whether you need to – or can – keep the balance of ‘de fortes mains, de belles couleurs’. Only 1) does this by putting them both in the singular; the rest introduce an imbalance of number.

What to do about belles couleurs. All five translators agree that there is no way of preserving the plural form. But a) do you need to unpack this a little, and indicate that the young lad is acquiring a ‘fresh’ or ‘ruddy’ colour, or indeed ‘complexion’ (which decides that couleurs is limited to the face – though reference has already been made, on the novel’s first page, to his ‘red wrists’); or b) is it self-evident where the lad is, and what is happening to his skin, so a non-specific ‘good’ echoes a non-specific belles?

All these six versions – given in chronological order – have their virtues; none is obviously superior. 1) is Marx Aveling, a version which, as Davis notes in her introduction, caused Nabokov ‘much indignation in his marginal notations but to which he resorted in teaching the novel’; 2) is Russell; 3) Hopkins; 4) which, even on this short evidence, looks freer than the others, is Steegmuller; 5) is Wall; and 6) Davis. Wall and Davis are the two who stick closest to the original sentence structure and are least ‘interpretative’.

There is a slightly pretentious term in wine tasting and wine writing called ‘mouthfeel’. (It is also slightly baffling: where else might you feel wine if not in your mouth? On your foot?) TheOxford Companion to Wine calls it a ‘non-specific tasting term, used particularly for red wines, to indicate those textural attributes, such as smoothness, that produce tactile sensations on the surface of the oral cavity’. There is similar mouthfeel about translation. Its general development over the last century and more has been away from smoothness and towards authenticity, away from a reorganising interpretativeness which aims for the flow of English prose, towards a close-reading fidelity – enjoy those tannins! – which seeks to echo the original language. We no longer use the verb ‘to English’ – it sounds proprietorial, even imperialist – but when Flaubert was first being translated it was still current: thus the first London and New York edition of Salammbô – published in 1886, the same year as Marx Aveling’s Madame Bovary – is described on its title page as having been ‘Englished’ by (wait for it) ‘M. French Sheldon’. This progress away from ‘Englishing’ can be seen in the six versions of Charles’s growing (up) quoted above. Similarly, in Chekhov translation, Constance Garnett has been succeeded by Ronald Hingley. Succeeded, and yet not supplanted: some of us continue to read the Garnett translations. Mainly because they do the time-travelling work instantly, and give a better illusion of being a reader back then, rather than a reader now inspecting a text from long ago through precision optical instruments. It may be, however, that something different, or additional, is going on: a kind of imprinting. The first translation we read of a classic novel, like the first recording we hear of a piece of classical music, ‘is’ and remains that novel, that symphony. Subsequent interpreters may have a better grasp of the language, or play the piece on period instruments, but that initial version always takes some shifting.

The authentic rendering of every last nuance of meaning cannot be the sole purpose of translation. Because if it becomes so, it leads to the act of eccentric defiance that is Nabokov’s Eugene Onegin. In his 1955 poem ‘On Translating Eugene Onegin’, Nabokov, addressing Pushkin, writes of turning ‘Your stanza patterned on a sonnet,/Into my honest roadside prose –/All thorn, but cousin to your rose.’ When Nabokov’s version of the poem came out in 1964, it was prose laid out in stanza form, and more woody stalk than thorn. Readers of the poem in English are best advised to have the two volumes of Nabokov’s headmasterly commentary to hand while apprehending the poem’s dance and flow through, say, Charles Johnston’s version. An even weirder example of fidelity leading to perversity is Dillwyn Knox’s 1929 translation of Herodas for the Loeb Classical Library. Knox’s brilliant niece Penelope Fitzgerald describes the outcome in The Knox Brothers with a sympathetic glee:

The language of the Mimes is precious, with unpleasant affected archaisms, and an honest translation, it seemed to Dilly, must be the same. Cloistered in his study . . . Dilly worked out his English equivalent to Herodas. ‘La no reke hath she of what I say, but standeth goggling at me more agape than a crab’ is a typical sentence, while ‘Why can’t you tell me what they cost?’ comes out as ‘Why mumblest ne freetongued descryest the price?’ Satisfied, Dilly corrected his proofs; he read the reviews, all of which praised the accuracy of the text but considered the translation a complete failure, with indifference. ‘If I am unintelligible,’ he wrote, ‘it is because Herodas was.’

Davis, in her introduction, notes that Gerard (not, as she has it, ‘Gerald’) Hopkins’s version has ‘added material in almost every sentence’; while Steegmuller produced a ‘nicely written, engaging version, smoother than Flaubert’s, with regular restructuring of the sentences and judicious omissions and additions’. Does this sound a trifle patronising to America’s greatest Flaubertian? Here is a typical addition (or rather, substitution) which will act as a good test of what a reader requires. When Léon goes to meet Emma inside Rouen cathedral, he first has to get past a verger standing in the left-hand doorway beneath a statue that Flaubert refers to as ‘Marianne dansant’. This was the popular nickname for a carving of Salome dancing on her hands before Herod. What do you do about this? Almost all translators render it as ‘Marianne Dancing’ or ‘The Dancing Marianne’. If you leave these words unannotated, readers will naturally imagine some cheery folkloric image. If you annotate them, then you divert the reader away for a guidebook moment – as elbow-tugging as the intrusive verger will prove to be to Léon and Emma. (You can half-solve it, as Davis does, by having notes at the back but without indication in the text of their existence; so readers may find the solution, but perhaps not at the right time.) Or, as Steegmuller alone does, in his unannotated version, you can cut to the chase and write: ‘The verger was just then standing in the left doorway, under the figure of the dancing Salome.’ This is instantly comprehensible, and has the additional virtue of pointing up this image of lasciviousness beneath which Léon passes on his way to the tryst. (Inside the cathedral, this theme is continued: when the verger reaches the tomb of the Comte de Brézé, he solemnly points out Diane de Poitiers as a grieving widow, while the rest of us know her as a king’s mistress – and also, as Emma is soon to be, the lover of a younger man.) Given that there is probably no one in Rouen who still refers to the statue as ‘Marianne dansant’, there is much to be said for Steegmuller’s solution. But some would find it overly interventionist.

The root feature of Davis’s translation is a close attention to Flaubert’s grammar and sentence structure, and an attempt to mirror it in English. For instance, observance of the ‘comma splice’ – where two main clauses are connected by a comma rather than an ‘and’ – or of subtle tense changes imperceptible to others (and sometimes imperceptible in English). In the earlier example (‘Il acquit de fortes mains, de belles couleurs’), she writes the boy had ‘good colour’ where Wall has the boy had ‘a good colour’: dropping the article retains the original plain adjective-noun balance. In her introduction, Davis castigates some of her predecessors for wanting ‘simply to tell this engrossing story in their own preferred manner’. Interviewed by the Times, she expanded on this: ‘I’ve found that the ones that are written with some flair and some life to them are not all that close to the original; the ones that are more faithful may be kind of clunky.’ This is the paradox and bind of translation. If to be ‘faithful’ is to be ‘clunky’, then it is also to be unfaithful, because Flaubert was not a ‘clunky’ writer. He moves between registers; he cuts into the lyric with the prosaic; but this is French prose whose every syllable has been tested aloud again and again. Flaubert said that a line of prose should be as rhythmical, sonorous and unchangeable as a line of poetry. He said that he aimed only at beauty, and wrote Madame Bovary because he hated realism (an exasperated, self-deluding claim, but still). He said that prose was like hair: it shines with combing. He combed all the time. As for those imprecise translators who nevertheless bring ‘flair’ and ‘life’ to the novel: where does that flair and life come from, if not the novel itself? Davis concludes: ‘So what I’m trying to do is what I think hasn’t been done, which is to create a well-written translation that’s also very close, very faithful to the French.’ This is a high claim; though I doubt any of those previous translators would have thought they were trying to do anything very different.

Davis’s quest to be ‘very close, very faithful’ to the French works best when the Flaubertian sentence is plain and declaratory. Take that great moment of delinquent self-awareness: ‘Emma retrouvait dans l’adultère toutes les platitudes du mariage.’ Davis’s ‘Emma was rediscovering in adultery all the platitudes of marriage’ exactly reproduces the French, and has exactly the same effect. You might think hers is the obvious translation until you compare other versions. Both Steegmuller and Hopkins diminish the line by recasting it, and even Wall, who is closest to Davis, misses out the necessary, intensifying ‘all’. On the other hand, a page or two later there is this equally key sentence: ‘Tout et elle-même lui étaient insupportables.’ This is an unusual sentence. A usual sentence might be ‘Tout lui était insupportable; elle-même comprise’ or ‘y compris elle-même’. Flaubert specifically links the ‘Tout et elle-même’ and it is a mistake to decouple them, as Davis does, into ‘Everything seemed unbearable to her, even herself’ (which adds the clunk of a repeated ‘her’). Wall also goes awry here: ‘It was quite unbearable, beginning with herself.’ Hopkins unpacks it perhaps too much: ‘She hated everything and everyone, including herself.’ Steegmuller is best, with: ‘She loathed everything, including herself.’ But even this doesn’t convey the full effect of that simple et – which is to indicate a separation of self from world which will culminate in the deed which finally does separate Emma’s self from the world.

So Davis’s division of previous translators into flair-bringers and clunkheads doesn’t really hold; nor does her claim to offer the best of both worlds. Two further examples. After Emma’s seduction by Rodolphe, there is a paragraph describing her post-coital, semi-pantheistic experience of the forest surrounding her, with which she is for the moment in harmony. But with the last sentence, Flaubert cuts this mood brutally: ‘Rodolphe, le cigare aux dents, raccommodait avec son canif une des deux brides cassée.’ This great anti-romantic moment has Rodolphe turning both to another physical pleasure (as Gurov will with his watermelon in Chekhov’s ‘The Lady with the Little Dog’), and to masculine, practical matters. All the versions cited here begin, unsurprisingly, with ‘Rodolphe, a cigar between his teeth . . .’ Wall goes on:

was mending one of the two broken reins with his little knife.

Steegmuller:

was mending a broken bridle with his penknife.

Hopkins:

was busy with his knife, mending a break in one of the bridles.

Davis:

was mending with his penknife one of the bridles, which had broken.

Rein or bridle? Knife, little knife or penknife? The difference is slight; all the versions contain the same information. Flaubert’s sentence does its business by not drawing attention to itself; its very downbeatness is the point, after the more rhapsodic prose that has preceded it. Wall, Steegmuller and Hopkins all get this. Davis doesn’t. Instead, she ‘faithfully’ sticks to Flaubert’s sentence structure. But English grammar is not French grammar, and so the quiet cassée (which for all its quietness also hints at Rodolphe’s ‘breaking’ of Emma) has to be unpacked into a ‘which had broken’ – a phrase which now seems pretty redundant, as what would he mend that wasn’t broken? The sentence has a clunkiness which is imported, rather than faithfully transmitted, and quite unFlaubertian.

The second example comes during Charles and Emma’s visit to the opera in Rouen – that greatest of the three great antiphonally constructed scenes in the novel – with Emma’s inner emotional life, her hopes and memories, being played off against the extravagantly exteriorised emotions of Lucia di Lammermoor. As her thoughts and feelings swirl, Emma comes to recognise that both art and life are inadequate in their different ways. This is a key sentence in the novel: ‘Elle connaissait à présent la petitesse des passions que l’art exagérait.’ It is a calm, balanced sentence, in three parts, with a triple alliteration, the part containing the second and third ‘p’ making up the central phrase. The choice for petitesse lies normally between ‘paltriness’ and ‘pettiness’, neither of which is perfect, as they have a slightly more disapproving tinge than petitesse. Wall’s version has the weight and progress of the original:

For now she knew the pettiness of the passions that art exaggerates.

Hopkins takes the alliteration elsewhere:

She knew now the triviality of those passions which art paints so much larger than life.

Steegmuller retains the triple alliteration:

Now she well knew the true paltriness of the passions that art painted so large.

Davis has:

She knew, now, how paltry were the passions exaggerated by art.

She needlessly turns the first major noun into an adjective, then reverses the grammar of the final phrase. But the main failing of the sentence lies in those first three words: ‘knew’, ‘now’, ‘how’ – both a wail of assonance and a stuttering of rhythm far from the original.

A translation can’t be read before the period in which it is written: this is both obvious, and a kind of brute annoyance. And it can’t – or at least shouldn’t – be written in a pastiche of the original work’s period. It must be written for the contemporary reader, yet give that reader the same, or a similar, ease or difficulty as an original reader would have had. And just as there can be delinquent looseness, so there can be misguided over-accuracy. It is very difficult to suggest (except in footnotes and introduction) the general literary context in which a book is written, which is often central to the writer. Books are often written against – against romanticism, against authorial ego and intervention, against the notion that some themes are ‘higher’ than others. There is a linguistic context too: thus Michael Hofmann, introducing his translation of Metamorphosis, quotes Klaus Wagenbach to the effect that ‘the characteristic purity of Kafka, the sober construction of his sentences and the paucity of his vocabulary are not understandable without his background in Prague German.’ Further, no two languages’ grammars match, and their vocabularies diverge (English having many more words and choices than French). Even punctuations have different weights: thus the English exclamation mark is shoutier than the French, so some of Flaubert’s have to be excised; Wall cuts more than Davis, who, at times, adds extra ones. Nor do languages develop over time at the same rate. So, in the case of Madame Bovary, you are having to juggle the French writer then, the French reader then and now, the English translator now, the English reader then and now.

At least nowadays some books are translated with less of a time-lag (La Regenta was first published in 1884-85, but not ‘Englished’ until 1984). Translators can quiz writers about what they mean, by email, or even in person: Don DeLillo had a London conference for his European translators of Underworld, who not surprisingly faced a few problems in a novel which opens with a 60-page baseball game. But translation is not always, or necessarily, about managing loss. When my novel Flaubert’s Parrot was being translated into German, my editor in Zurich modestly suggested some additional flourishes: for instance, a pun on Flaubert as a ‘flea-bear’, and a German slang phrase for masturbation which literally means ‘to shake from the palm tree’. Since Flaubert, at this point of my novel, was being masturbated in Egypt, this felt like a happy improvement on the English text. Adding something extra for German readers seemed a kind of fair-trade translation. But this advantage encloses a new danger: that of the writer anticipating this amiable to and fro with translator or foreign editor. I remember hearing a well-known British novelist admit in a radio interview that he had paused at one point in his writing, thought of the pain he might be inflicting on his Scandinavian translators, and decided to make things easier for them. Apart from this being a denial of your own language, it can easily lead to the sort of international prose that is like an airline meal: it feeds all, doesn’t actually poison anyone, but isn’t noticeably nutritious.

To compare several different versions of Madame Bovary is not to observe a process of accumulation, some gradual but inevitable progress towards certainty and authority (except in the occasional discarding of error); rather, it is to gaze at a sequence of approximations, a set of deliquescences. How could it be otherwise when almost every word of the French can be rendered in several different ways? Consider the moment when Emma, Charles and Léon are eating ice-cream in a café by the harbour, having walked out before the final scene of Lucia. Charles naively suggests that his wife stay on in the city to catch the next performance – an action which precipitates her affair with Léon. Charles addresses his wife (the banality of phrase contrasting with the recent flourishes of Donizetti) as ‘mon petit chat’. Marx Aveling has ‘pussy’, Mildred Marmur (1964) ‘my kitten’, Wall ‘my pussy-cat’, Hopkins ‘darling’, Steegmuller ‘sweetheart’, Russell and Davis ‘my pet’. Marx Aveling’s endearment would work then but sadly not now; Marmur’s is good; Wall’s brings in the slightly unwanted flavour of a bad Dean Martin movie; Steegmuller and Hopkins deliberately duck the felinity (you could argue that the French is already drained of it anyway); while Russell and Davis, mixing banality and distant animality, have found the best solution. Probably. At least for the moment. You can understand why Rutherford called translation a ‘strange business’ which ‘sensible people’ should best avoid.

Davis’s Madame Bovary is a linguistically careful version, in the modern style, rendered into an unobtrusively American English. At its best, it conveys the precision – which some think dryness – of Flaubert’s prose in this novel, while its syntactical mirroring of the French sometimes brings us closer to Flaubert. At its worst, it takes us too far away from English, and makes us less aware of Flaubert’s prose than of Davis being aware of Flaubert’s prose. And such defects come from something very old-fashioned: a lack of love for the work being translated. In her Times interview, Lydia Davis explained:

I was asked to do the Flaubert, and it was hard to say no to another great book – so-called. I didn’t actually like Madame Bovary . . . I find what he does with the language really interesting; but I wouldn’t say that I warm to it as a book . . . And I like a heroine who thinks and feels . . . well, I don’t find Emma Bovary admirable or likeable – but Flaubert didn’t either. I do a lot of things that people don’t think a translator does. They think: ‘She loves Madame Bovary, she’s read it three times in French, she’s always wanted to translate it and she’s urging publishers to do another translation, and she’s done all the background reading . . .’ but none of that is true.

Perhaps some of this is the translator’s equivalent of being demob-happy – three years slogging across Occupied France, it’s no wonder she throws her cap in the air at having come out alive. What does Davis mean by saying that Emma Bovary doesn’t think or feel? The novel is all about the perils of (wrong) thinking and (false) feeling. Perhaps she means ‘doesn’t think or feel in a way that I approve of’. As for complaining that Emma isn’t ‘admirable or likeable’ – this sounds like the most basic book-group objection. Lydia Davis’s Madame Bovary shows that it’s possible to produce a more than acceptable version of a book with which you are profoundly out of sympathy. In that sense, it confirms that translation requires an act of the imagination as well as a technician’s proficiency. If you want a freer translation, Steegmuller is best; for a tighter one, go to Wall. And perhaps one day Juliet Herbert’s lost ‘masterpiece’ will turn up, and we shall be able to compare it to its successors – and see a new way of necessarily falling short.