‘traduttore, traditore’ (‘translator, traitor’)

‘Translations are like women: when they're beautiful, they're not faithful, and when they're faithful, they're not beautiful.’ (Carl Bertrand, also attributed to Yevgeny Yevtushenko) [with apologies to women]

‘Translation is at best an echo.’ (George Borrow)

‘What is lost in the good or excellent translation is precisely the best.’ (Schlegel)

‘Poetry is what is lost in translation. It is also what is lost in interpretation.’ (a remark attributed to Robert Frost, the precise wording disputed and not found in his writings)

Nobody then seems to have a good word to say about translation — a necessary evil at best. The truth is a little different.

Somewhat ironically, the expression ‘lost in translation’ has become so familiar to us in most of the contexts in which it is used these days that it has lost the words that originally preceded and followed it when used by the American poet, Robert Frost (see above). Frost used the expression ‘lost in translation’ of poetry, and maintained that it gets lost not just in translation but also in interpretation.

This article is concerned primarily with translation and not with interpretation (though translation often involves interpretation, of course). It is, however, concerned with prose as well as verse translation, since there is no reason to suppose that prose is not lost in translation as well as poetry. The article seeks to identify what it is that is lost in the translation of Greek and Latin texts, verse and prose. It also questions whether what is lost is of such a kind and of such a quantity that engaging with texts in translation has real value, mindful of the fact that many people who study Classics these days experience Greek and Latin texts only through the medium of translation. I am also mindful that the same question can be asked of texts in other languages and of other periods. (What does a monoglot Chinese person get from a translation of John Donne or Shakespeare's sonnets?) One must also bear in mind that a translation may have positive literary or other merit in its own right, whatever it loses of its original text (the King James Bible springs to mind, as does FitzGerald's translations from the Rubaiyát).

Some genres may be more affected by losses in translation than others. Poetry generally suffers more than does prose. Lyric poetry suffers more than epic or didactic poetry. Of course people whose access to these texts is by way of translation only will not be aware of what they are missing in these different genres. Does this lack of exposure to the original text invalidate or detract from their experience? Is an exercise in literary criticism of Horace's Odes, based on a translation of the text, a criticism of Horace's Odes? Again, people go to translations of texts for different purposes. So people who consult translations of a historical work in prose for factual information only may ‘lose’ less in translation than people who read translations of Horace's Odes.

I cannot read Russian. I think my life would be impoverished at some level if I had been discouraged for that reason from reading a translation of Anna Karenina or The Brothers Karamazov. As a Classicist, I would encourage anyone to read a translation of Oedipus Tyrannus, Herodotus, or Lucian. Whether my or their experience could be regarded as ‘authentic’ or not seems to me to be a secondary (or lower) consideration compared with the value (subjective though it must be) of my exposure to the translations of the texts. And the fact that I cannot comment on features of the original texts that require a knowledge of the language of the texts does not preclude me from experiencing and commenting on many other features that are not tied to the original language. The effect of Oedipus Tyrannus on any audience does not depend on its having been written in a language that it cannot understand. The fact that it is not engaging (directly) with what Sophocles actually wrote is a minor consideration, except for an editor or a textual critic or a Thomas Gainsford who commended the Oedipus Coloneus on the grounds that it was a ‘veritable treasure-house of grammatical peculiarities’.Footnote ¹

So what are the main features of a Greek and Latin text that are impossible or very difficult to reproduce in an (English) translation. The first of these features has in fact already been dealt with in the opening paragraph: terms in the original language that do not have semantic equivalents in the language of translation. A good example is the Greek word eudaimonia and the English word ‘happiness’ or ‘wellbeing’. Others might be hubris, atē, aidōs, aretē, dēmokratia etc (‘democracy’ is OK as long as it is used to denote the ancient Greek governmental system). As for Latin, one can cite pius, plebs, furor, auctoritas, religio, provincia etc., for none of which English equivalents are readily available. In 1960 A.W.H. Adkins (see his book Merit and Responsibility: A Study in Greek Values) declined to attempt to translate certain culturally-laden value terms on the grounds that any translations would be fundamentally misleading Instead he transliterated such terms wherever they occurred in the book. The prime target was agathos, which he never translated as ‘good’. How effective this was in dispelling misunderstanding, especially for Greekless readers, I am not sure. It did not set a trend for subsequent translators. Whether transliteration would be more effective in other contexts I do not know.Footnote ²

Related to the cultural diversity of languages is the difference of idiom. By ‘idiom’ I mean the particular way in which a language expresses a given idea. Here the language of the original has to be recast into a form in which it can be rendered meaningfully in the language of translation. The idiom of both Greek and Latin often differs noticeably from the idiom of English. Actually, this is more of an issue when translating from English into Greek or Latin — so less of an issue these days. For example, ‘their courage won the day’ would be meaningless if rendered literally in either Greek or Latin. Both Greek and Latin would express the same idea much more prosaically as ‘on account of their courage they were victorious’. Anyone who has had to translate a piece of English replete with metaphorical and other figurative language into idiomatic Greek or Latin will be only too familiar with differences of idiom. This is not to say that Greek and Latin do not employ idiom (including their own metaphors and other figurative language, if used more sparingly) that is alien to English and which does not carry over into an English translation. (Housman's Fragment of a Greek Tragedy is a parody, which means that it is an exaggeration of an actual type of composition.) But generally more is lost of English than of Greek or Latin. It would seem that as far as idiom is concerned a translation may be literal or literary, but not both.

The sounds of the original language, and the effect they are designed to produce on the reader or listener (the ‘sound-effects’) can only occasionally, and inconsistently, be reproduced in the language of translation. Two examples from Virgil will suffice to make this point:

sunt lacrimae rerum et mentem mortalia tangunt (Aeneid 1.462). I will not attempt a translation, but if I did it would be very difficult to convey the effect of the elision of rer(um) suggesting a momentary faltering in speech under the pressure of emotion; and this is followed by the subdued, almost muted, and mumbled sounds of the second half of the line (en, em, an, un). Surely something would be lost in translation here.

quadripedante putrem sonitu quatit ungula campum (Aeneid 8.596). You can hear the horses galloping across the plain (see below on the rhythm of verse).

Differences of prosody are a sub-species of sound-effects as far as reproducing the audible rhythm of the language is concerned, as we can see from the Virgil extracts just quoted. Rhythm can be highly suggestive of sense, as sound-effects generally can be. The same effect cannot be achieved consistently by most translators in an English translation, especially in the case of complex metrical systems. There are very few people who could carry off the equivalent of Horace's adaptation of Greek lyric metres for his composition of the Odes. In fact many/most translators do not even try, using prose instead of verse for their translations. The more or less total avoidance of the rhythmic effects of the original language is a significant loss in translation.Footnote ³

Words can denote and connote. A word in the original language may have the same denotation (referent) as the word in the language of translation. However, either may have a different connotation from that of the other or have no connotation at all. Think of connotations as associated meanings. The English word ‘home’ has associated meanings that the word ‘house’ does not have. ‘There's no place like the house’ does not have the same evocations as ‘There's no place like home’. What may be lost in translation here is the semantic richness of the connotation(s) of words in the original language. The Greek word xenos has connotations that its various English translations do not have: ‘host’, ‘guest’, ‘stranger’ etc., even though any of these words may denote in context the same as the Greek word. These connotations of words modulate the response to the actual words for people who are aware of them. If the language of translation does not have the same connotations then the reader will lose something in translation.

Word order, and the literary and rhetorical effect of word order, is a feature that almost inevitably gets lost in translation, and yet was regarded by writers to be of great importance (see, for example, Dionysius of Halicarnassus, On The Arrangement of Words 20, Quintilian 9.4.14). The typical word order of a Greek and Latin sentence (especially Latin) is different from that of an English one. Importantly, word order plays a different role in an English sentence, one that curtails the possibilities for the positioning of words in the sentence or in a line of verse. This is because word order determines meaning in a way that it does not in Greek or Latin. The result is that Greek and Latin can exploit the stylistic possibilities of word order more, often for good effect that cannot be captured by an English translation. (Think, for example, of the first and last words of the opening line of the Iliad. ‘The wrath, sing, goddess, of the son of Peleus, Achilles’ preserves the order of words, but at a cost.) But I don't think I need to dilate on this here as it will be well known to all or most readers.

Last, but not least, is a limitation, if not an actual loss, this one affecting a person who is not so obviously dependent on translation. This has to do with the distinction between reading and translating. Reading a text, in the full sense of ‘reading’, as one reads a text in one's own language, does not involve conversion from one language (L2) to another (L1) in order to facilitate understanding; translating, by definition, does. Only if a person is fully bilingual, equally competent in both languages, does translation play no part in understanding. But can a person be truly bilingual if one of the languages is a written, not spoken, form of a language which is neither a L1 nor learned, or rather acquired, as if it were a L1? This is the case with Greek and Latin texts, which have always been learned as a L2 or as if they were a L2 if their L1 is Latin. Some people who engage in ‘conversational’ Latin may have taught themselves to be fluent in oral composition over a range of topics; this does not mean that they can read Classical texts as they read texts in their own language. Does there exist, has there ever existed, a person who can actually read, in the fullest sense of ‘read’ any text you like in Greek or Latin? Erasmus perhaps, or Heinsius fils, but certainly not any mere mortal, nor any living person.

The upshot of this is that even a person who is highly competent in Greek or Latin is often engaged in translation when attempting to read a text. And often such a person will have recourse to an actual published translation when their own resources fail them. The difference between a person who can read Greek or Latin and one who can't is not as great as one might think. We are all translators, not readers, of Greek and Latin when it comes down to it. It is claimed that Comprehensible Input (CI) enables its students to become readers, not translators, of learning materials in Latin, to acquire Latin as if it were a L1 rather than learn Latin as a L2. Whether this is equally true of reading (say) Horace or Persius as if they were reading their own language I don't know. It would be an outstanding achievement if it were true.

Providing a translation along with the original Greek or Latin is now becoming almost mandatory for authors of books and journal articles, even for the most scholarly and specialised. Whether this is mandated on the grounds of accessibility or ‘inclusivity’, is not clear. It may be an acknowledgement of the dwindling competence of most readers these days to get along without a translation. For example, the successor journal to the august Proceedings of the Cambridge Philological Society (The Cambridge Classical Journal) requires translation of all Greek and Latin (as Greece & Rome has always done). And of course the introduction to OCT texts has no longer been solely in Latin for over 30 years now.

These seem to me to be the most important things that get lost in translation. So there is no doubt that a lot is lost in translation, especially of verse originals. There is equally no doubt that a person who is competent in the original language will be exposed to many more features of a text than a person who can only experience texts in translation that cannot mirror, or only imperfectly mirror, features of the original. It is just plain wrong to claim that a reader can get more or less the same from a translation as from the original, or that ‘to all intents and purposes’ a translation is ‘as good as’ an original. It is even said by some that a translation may be an improvement on the original! No: if the original is bad then the translation will be bad or it is not a good translation. As I said earlier, the person who is dependent on a translation does not really know what he/she is missing. But this does not mean that such a person is not missing something, quite a lot, as it happens. Translations are what they are: substitutes; and substitutes can be very useful. If everyone were literate in the original there would be no need for them. Again, as I said earlier, everyone (including a Regius Professor) has recourse to translations to a greater or lesser extent when engaging with a text (some Regius Professors have even composed them). Everybody who is highly competent in Greek and Latin uses a Loeb for one purpose or another. It is hypocritical then to make use of them and at the same time look down on them as inferior substitutes for ‘the real thing’. (They may be inferior to other substitutes, of course.)

If the translation is as faithful to the original as possible in all possible respects then translations are to be encouraged. In any case, what is the alternative for people who do not have an alternative? We should be as zealous in producing faithful translations as we have been in producing flawless critical editions. The former is no more of an unrealisable ideal than the latter. Given that translations are here to stay and are likely to be needed more and more, I think that we should also encourage more commentaries and other aids designed specifically for translations or to accompany translations, i.e. in the same volume as the translation. These might usefully include the transliteration (of Greek) and explication of hard to translate terms and expressions.

Finally, it is important to remember that people who can access the original language, however accomplished they may be, are not native speakers of the language in any of its various registers. It is possible that the gap that necessarily exists between such people in experiencing the language may be as wide as that which exists between people who can access the language and those who rely on translations. There may be no winners and losers, only different categories of losers. Nobody really knows ancient Greek or Latin for whom it is not a first language; and nobody can be a native speaker of a written language. As Michael ClarkeFootnote ⁴ revealingly says:

Actually, Homer's mother tongue may have been rather different from the words of Homer that Clarke has been engaging with for many years.