Classics in Translation? A Personal Angle (Part I)

Translation in Antiquity and the Present

Wolfgang de Melo

By now I have written a fair few pieces for Antigone, all of them on linguistic topics. But they have one other thing in common: they all started as tangential conversations in the context of philology tutorials at the University of Oxford. When I set tutorial topics, I focus on issues that could potentially come up in the exams. Often, however, students have questions which I find just as interesting, but which I can’t answer properly during the actual tutorial because we are under time pressure. The topic of this piece also arose from a conversation with an undergraduate; our final tutorial had finished, and he had recently found out that Princeton had abolished its requirement for Classics students to learn Latin and Greek; he wanted to know what I, as his language tutor, made of this.

Of course, the issue goes beyond Princeton, or the United States, for that matter. The calls that we should teach Classics in translation are growing louder and louder, as are the calls that we should stop using translation into and out of the languages as a method to teach them or to assess competence in them. Such demands are not restricted to Classics as a minor (I am thinking of Ancient Civilisation options as part of an altogether different degree); they are also made for full-time students of Greek and Latin literature. Although the two issues – Classics in translation and translation as a teaching tool – are connected, and often discussed as if they were one and the same, at least in principle they are separate. Today, then, I would like to give my student a much-belated reply on both fronts.

I will do so wearing two hats at the same time. On the one hand, I am a linguist and a philologist, and as such my entire academic life has been dedicated to advancing our understanding of ancient languages, and to teaching these languages to as high a standard as is possible within the narrow time limits of an Oxford degree; nothing gives me more joy than seeing my students learning how to read their languages with ease, and how to use them actively, for instance in prose and verse composition. But on the other hand, I have almost unwittingly ended up as a translator, too.

Some useful books on Greek and Latin verse.

As a student, I enjoyed entering various contests for Latin and Greek translation, and had some success. I found such translations deeply satisfying because they had an artistic feel to them; during periods of intense academic pressure, there was nothing more relaxing or reassuring than having afternoon tea and biscuits with my verse tutor, Donald Russell, and comparing our own translations of beautiful English poetry into various Latin and Greek metres.

But it was not until I was a postdoc that I started to translate into English at a larger scale; Richard Thomas from Harvard was a visiting fellow at Oxford at the time; it was he who approached me with the request for a new Loeb edition of Plautus. Later, when Oxford University Press asked me to make a critical edition of Varro’s De lingua Latina, I insisted that a highly technical and fragmentary text of this sort needed a translation as well. I hope that my two perspectives, as a teacher and as a translator, can lend the discussion a bit of the nuance it so often lacks.

“An Oxford Don”, a print issued in London, 1808.

I want to cover four topics, which I will divide into two articles. First, I wish to point out just how much Latin literature is translated or adapted from Greek models; I intend to show how ancient writers operated and how they dealt with their sources. In order to understand this type of literature at a deeper level, we really need some experience in translating ourselves. Next, we will be looking at translation as a tool for language teaching. Is it efficient or overrated? What, if any, should its place be in a modern, evidence-based curriculum whose main goal it is to get students to read and understand ancient texts competently?

The second article will begin with the third question: do all students of the Greco-Roman world even need to acquire linguistic competence in Latin and Greek? Are there some groups for whom other activities are a better use of their time? And fourth and finally, I will turn to some fashionable schools of thought, if that is not too generous a term, which have set themselves not only in opposition to translation as a tool for language teaching, but which insist that the subject should be taught purely in translation, if at all. How did such ideologies arise in the first place? Why are they so popular? And how should we engage with them? Let us begin with the topic of Latin literature as translation literature.

The Harley Trilingual Psalter, in Greek, Latin and Arabic, produced for Roger II of Sicily (1095–1154) (British Library, London).

1 ) Latin literature as translation literature

I had Latin at school, but not Greek. When I first started to learn Greek, it came as a shock to me just how important Greek is for any student of Latin. From the earliest literary texts to the last ones before the beginning of the Middle Ages, Roman authors translated and adapted Greek models; and many of the subliterary documents betray bilingual interference or show deliberate switches between Latin and Greek. In this section, I want to look at Plautus, then at general strategies for rendering foreign terminology; we will also have a quick look at Bible translations, and finally we will turn to Latin influence on Greek.

1.1 Plautus and Menander

Plautus (died 184 BC) and Terence (died 159 BC) are the earliest Latin playwrights of whom we have complete comedies; all their works are adaptations of Greek originals, by a host of authors such as Menander and Diphilus. A direct comparison is, for the most part, no longer possible because the Greek originals have been lost. Scholars such as Eduard Fraenkel (1888–1970) have devoted painstaking research to the question of which elements were simply translated and which were added to by Plautus (and Terence). When an important papyrus find was published in 1968, it allowed scholars to compare, for the very first time, a longer section of a Greek original with its Plautine adaptation. The Greek fragment belongs to Menander’s Δὶς ἐξαπατῶν (Dis exapatōn, “The Double Deceiver”). Here is a portion from this text:

(A young man has trickd his father out of money in order to pay a prostitute; he now believes that she has deserted him, and so he wants to pay the money back to his father and own up to his wrong-doing.)

            Πᾶν ἀποδώσω τῶι πατρὶ
τὸ χρυσίον, πιθανευομένη γὰρ παύσεται
ὅταν ποτ’ αἴσθηται, τὸ τῆς παροιμίας,
νεκρῶι λέγουσα μῦθον. (26–9)

“I will pay back all the gold to my father. Then she will stop beguiling me, when she realises, as the proverb goes, that she is telling a tale to a corpse.”

And here is how Plautus rendered Menander in his Bacchides (“The two Bacchises”):

                                                Nam mihi
decretum est renumerare iam omne aurum patri.
igitur mi inani atque inopi subblandibitur
tum quom mihi illud nihilo pluris referet,
quam si ad sepulcrum mortuo narret logos. (515–19)

“I have decided to pay back all the gold to my father immediately. Then she will coax me when I am empty and poor, at a time when this has no more effect on me than if she were prattling to a dead man at his tomb.”

This is quite a literal translation, but elsewhere Plautus is happy to change metres or the names of the characters, and he inserts jokes and puns of his own. Note that here he uses a Greek word, logōs (“words”), which is not taken from the original.

Menander contemplates his masks, Roman copy of a Hellenistic original, 1st cent. BC/AD (Princeton University Art Museum, NJ, USA).

Actually, Plautus uses quite a few Greek words in his plays. Interestingly, it can be shown that these are never taken from his Greek originals, but are his own insertions intended to create a ‘Greek’ atmosphere. For example, for “banker” Plautus uses both argentarius and tarpezita.

Argentarius is a Latin word and occurs in Roman contexts, when Plautus talks about the bankers in the city of Rome; but tarpezita is adapted from Greek and is found when he speaks of bankers elsewhere. Tarpezita cannot be taken directly from the originals, where the form is τραπεζίτης; the adaptation of the ending would be unproblematic, but Attic Greek has trap-, whereas Plautus has tarp- (as proven by the metre), so that we are dealing with a dialect form, one which was probably common in southern Italy at the time.

Similarly, for “sword” Plautus has Latin gladius and Greek machaera. Gladius is employed when Plautus talks about the swords of Roman soldiers, but elsewhere he uses machaera. In this case we know that machaera came from the Greek spoken in Italy because in the Greek originals the word for sword is always σπάθη (later borrowed into Latin as spatha, compare Spanish espada and French épée). Μάχαιρα does occur in the Greek originals, but not as a sword: it is a cook’s knife!

There are a few instances where Plautus slips up, but not because he fails to translate Greek words; on the rare occasions that we can detect direct Greek interference, it is always in syntax.[1] This is understandable: when we translate, we can always consult a dictionary, but occasionally our thoughts get stuck in the source language. In my many years of marking Latin-English translations, I have seen dozens and dozens of cases where native speakers of English produce texts that are barely intelligible until I look at the Latin constructions. But let us dwell a little longer on individual words.

Comic fresco of the 1st cent. AD (Casa dei Quadretti Teatrali, Pompeii, Italy).

1.2 Loanwords, calques, loan shifts

How do translators deal with new concepts for which their native language still lacks established terminology? Cicero discusses such issues at length in his philosophical works, where he needs to render Greek terminology.[2] In principle, there are three options: we can take on a foreign word, possibly with adaptations in pronunciation and morphology; we can translate a complex word element by element; and we can use a native word and give it a further meaning.

In the first case we speak of loanwords. Over time, these can become indistinguishable from native words; hardly anyone today knows that kitchen and cheese started as Latin loans, from coquina and caseus. The second strategy is called calquing; French gratte-ciel is a direct translation of English skyscraper, while German Wolkenkratzer “cloud-scraper” is less literal. If the third method is used, we say that a loan shift has taken place. For instance, English realise is originally a loanword from French réaliser; the original meaning was “to make real, to bring about”, as in “to realise a dream”. In English, the word developed a further meaning, “to understand”. In French, this second meaning was then adopted from English, a clear instance of a loan shift.

Which option we go for depends on a number of factors: the perceived prestige of the source language; the morphological means of the borrowing language; and various other socio-cultural norms. Earlier English often preferred calques, such as heathen (from heath) based on Latin paganus “inhabitant of the countryside, non-Christian”; later, loans were preferred, hence pagan. Similar considerations apply to Greek influence on Latin. On this note, let us turn to the Bible.

The Doctors of the Church, engraving by Cornelius van Dalen after Peter Paul Rubens, early 17th cent.

1.3 Translating the Bible

Here is the opening of St John’s Gospel:

Ἐν ἀρχῇ ἦν ὁ λόγος, καὶ ὁ λόγος ἦν πρὸς τὸν θεόν, καὶ θεὸς ἦν ὁ λόγος.

“In the beginning was the Word, and the Word was with God, and God was the Word.”

The Vulgate renders this as follows:

In principio erat Verbum, et Verbum erat apud Deum, et Deus erat Verbum.

This translation is much more literal than what Plautus does, and this is unsurprising: a sacred text requires an exact rendition. Nevertheless, there are choices to be made: St Jerome chooses verbum for “word”, following earlier European Bible translations, while the North African translators picked sermo.

Translating the Bible had a lasting influence on Jerome, but also on many other Christian writers. For example, following his Greek model, Jerome translates many clauses introduced by ὅτι (“that”) after verbs of speech with quod-clauses instead of the accusative and infinitive. In his letters before the Bible translation, this alternative is very rare, just as it is rare in Jerome’s pagan contemporaries. But after his Bible translation, Jerome uses these quod-clauses much more freely. St Augustine also follows pagan practice before his conversion, but Jerome’s practice after becoming a Christian. Translation had a lasting influence on Latin; but not just on Latin.

Saint Jerome in his study, Pieter Coecke van Aelst, 1530; Walters Art Museum, Baltimore, USA).

1.4 Latin influence on Greek

To a more limited extent, Latin also influenced Greek. It is true that there are few highly literary texts that are translations from Latin into Greek, but when Greece was ruled by the Romans, many administrative documents were translated into Greek, sometimes idiomatically, sometimes less so.

Over time, many Latin words stuck, especially those associated with the military, government or money: hence κουστωδία from custōdia (“guard”), δικτάτωρ from dictātor (“interim dictator”) or ἀκτουάριος from āctuārius (“official in charge of pay”); or, with more significant adaptations of the endings, δηνάριον from dēnārius (a currency term) or κοδράντης from quadrāns (another currency term). Papyri also show calques and mixed words like σηλλοποιός (“saddle-maker”), with its first part from Latin sēlla (“saddle”) and its second from Greek ποιεῖν (“to make”).

Latin influence was so deep that it even affected personal names. The element –poulos in names like Papadopoulos comes from Latin pullus (“chick”) and functions as a patronymic (“descendant of”)! Much later, when Greece came under Ottoman rule, Turkish exerted a comparable influence, seen for example in names starting with Hatzi-, originally referring to those who had done the Hajj, the pilgrimage to Mecca. In a name like Hatzopoulos, we can see a Turkish element welded to a Latin one!

With so much translation going on between Latin and Greek, we really need some translation practice, too, if we want to understand this kind of literature. But how good a tool is translation for language learning?

A partridge with chicks, detail from the 3rd-cent. AD ‘Lod Mosaic’, discovered in 1996 in Lod (ancient Lydda), Israel (Lod Mosaic Center, Israel).

2 ) Translation as a tool for teaching and examining

Much of our language teaching and examining is still done through translation: students are required to translate Latin and Greek into English, or vice versa. What do I as a linguist think about these practices? Are they effective tools for acquiring language competence? In these pages, Melinda Letts has beautifully argued for a greater emphasis on teaching Latin in Latin; and both Judy Nesbit and John Claughton have made excellent points in favour of the more traditional methods. I am not going to come down firmly on either side, as I believe that both methods have their place, but I would like to offer some further thoughts.

2.1 Different goals

In the first place it needs to be said that the two directions of translation are not created equal. Translation from the ancient languages into English is normally meant to be very literal, almost like a crib; in other words, it is normally treated as shorthand for parsing. The teachers asking for such literal translations into English do not care for literary merit or even idiomatic English, but want to make sure that the students have understood the grammatical structures. On the other hand, translations into Latin and Greek are always meant to sound good; often, they are supposed to imitate a specific style, for instance, that of Cicero or Virgil for Latin, or that of Xenophon or Sophocles for Greek.

At first sight, then, translation into Greek or Latin may seem the bigger challenge, and for beginners it normally is. However, in my experience, more advanced students tend to fare better in their translations into the ancient languages. There are good reasons for this. First, in order to understand 90% or more of the words of any text in any language, one needs to know approximately 4,000 words; the most frequent 100 words will give you 50% of any text, because half of any text consists of words like the or forms of be, but thereafter it is a case of diminishing returns, and every further set of 100 words will give you fewer and fewer percent.

Learning 4,000 words takes time, and beginners and intermediate students most often stumble because they have not yet memorised enough Latin and Greek words. But if you are translating from your native language, you can normally operate with a smaller set of ancient words; you won’t have as much lexical variation, but you can get by. And second, if you are stumped by a Latin or Greek construction, there is often little you can do in an exam situation, but if you are translating into these languages, you can again manage with a smaller set of constructions. But back to the question: is translating an efficient method of learning or teaching?

A sample section of Cornelius Schrevelius’ Lexicon Manuale Graeco-Latinum et Latino-Graecum, first published in 1654.

2.2 How unnatural is translation?

One argument advanced against translation as a teaching tool is that it is ‘unnatural’, and that is true insofar as children growing up in monolingual households acquire their native language with ease and yet without ever having to render one language into another. But second-language acquisition is very different from first-language acquisition, and even in a first-language setting, young children growing up in bilingual households often turn to (oral) translation when there is something they don’t understand in one of their languages. They compare and contrast grammatical constructions and lexical meanings quite happily.

I once read of a French teacher whose worry about unnatural methods of acquisition went so far that she eschewed bilingual vocabulary lists; she would introduce a word like pomme (“apple”) with a picture of the fruit so that her charges would not associate pomme with the word apple, but would learn it the way a French child would. An extreme case, perhaps; the reality is that once a first language is in place, learners cannot bypass it, or avoid establishing some translation equivalents; nor should they try to.

“King Pippin” (mother and daughter in the apple house), an illustration by George John Pinwell to ‘Wayside Posies’, edited by R. Buchanan (George Routledge, London, 1866).

Second-language acquisition is different from first-language acquisition for several reasons. Motor skills and memory capacity have already decreased, meaning that pronunciation needs to be taught more explicitly and that morphology and vocabulary typically require input from grammar books and vocabulary lists. The differences thus do not simply arise from reduced exposure to the second language, although that plays a part, too: first-language learners are constantly surrounded by language, at a time when their brains are very adaptive and receptive, while second-language learners normally have to make do with a couple of hours of teaching per week. So, at the end of the day, perhaps we are asking the wrong question: we should not worry whether our teaching methods are ‘natural’ or not; what matters is whether they work, and work better than other tools. People may perceive these questions to be connected, but the connection is marginal at best.

There is one scenario where ‘translation’, or rather, establishing equivalents, works particularly well: this is when you learn a language that is closely related to your own, either because the languages stand in a mother-daughter relationship or because they are close siblings.

For me, this situation arose when I as a budding Latinist took Italian classes, and later in life when I moved to Belgium, where I was expected to teach in Dutch. With a newborn at home, I didn’t have time to attend a language course until half a year after my arrival; but because German and Dutch are so close, I could easily establish equivalents, figuring out that Dutch d typically corresponds to German t (dag and Tag, both meaning “day”), or that German subordinate clauses end with infinitive + modal verb, while in Dutch the order is typically reversed.[3] After half a year of listening to native speakers in the faculty and a further three-week course at the language centre, I was able to teach in Dutch; such fast progress was only achievable because of my method of establishing equivalents.[4]

The family tree of West Germanic languages.

One concern that I sometimes hear is that too much focus on translation prevents students from learning to think in Latin or Greek. There is some truth in that, especially when the only direction of translation is from the ancient languages into the native tongue. This kind of translation, when it is not accompanied by other training tools, breeds an attitude that makes students create their crib-like translations in order to discuss the content of the ancient texts in English, but without acquiring a deeper understanding or appreciation of the source language. To some extent this situation can be mitigated by translating into Latin and Greek, because that forces students to look beyond translation equivalents and to try to polish their products, thereby acquiring a certain feel for Latin and Greek idiom.

But good language teaching should go further than the translation method. There should be exercises that manipulate ancient texts, whether this is done orally or in writing: for instance, a piece of direct speech could be turned into indirect speech; or a passage in primary sequence could be turned into secondary sequence; and with a bit of creativity, a plethora of similar exercises can be devised. Such exercises are not done nearly often enough in classroom settings, usually because there is not enough time; but the higher the students’ active competence is, the more efficiently they can go through the actual ancient texts they are meant to read. However, opting for more active methods should not mean abandoning translating altogether.

We need to get away from the idea that translation is ‘unnatural’ or that adolescent or adult learners can approach a new language as if they were a tabula rasa, as if they would not automatically compare and contrast, as if they would not at first understand the new language through the lens of the languages they already know. And given just how important translation was for the ancients, students need to acquire some translation skills.

“Academical Costume”, Thomas Uwins’ watercolour illustration for Rudolph Ackermann’s History of the University of Cambridge (London, 1815).

2.3 Teaching methods should depend on goals

The tests that we use to gauge prospective university students’ English skills assess four components: speaking, writing, listening and reading.[5] We expect high achievement in all four areas, but if one skill may be a bit lower, it is normally speaking. Why would that be? Students need to be able to do their reading efficiently; they must be able to follow lectures; and they need to be able to write good essays. If their spoken English is a little less fluent when they arrive in this country, they will often improve through their interactions in English with tutors and fellow students, something which they typically lacked in their home countries, where they had easier access to books and recordings. And if their spoken English continues to lag a bit longer, it is typically of little detriment to their results, which are based on written answers.

When I oversaw graduate admissions for the Faculty of Linguistics, Philology and Phonetics, there were often overseas applicants who fell short in one or two of these four areas, and I always had an online conversation with them in order to come up with a study programme tailored to their needs, so that they would fare better when they retook their tests.

I don’t see why it should be any different for Latin and Greek. Our general training should be focused on what our broad goals are, and our special support should be geared towards the individual student’s weak points. For Latin and Greek, acquiring the ability to speak or to understand spoken language can be hugely motivating. If a student wanted to attend such courses, I would always encourage it. But realistically, we don’t learn these languages because we need them to communicate with our contemporaries; being able to do so is a bonus, but never the main goal.

The main goal is reading ancient texts with ease and confidence as well as analysing them with attention to detail. For these goals, text-based methods are key: active reading, with exercises that manipulate the text, as suggested above; translation into English as a means of testing accurate understanding; and translation into the ancient languages as a means to acquire stylistic competence. Ideally, there should also be some speaking and listening so as to become well-rounded, but this is the icing on the cake, not the cake itself. I prefer my cake with icing, but some people like it plain.

17th-cent. drawing of a funerary relief, depicting the bust of a young scholar, flanked by actors, Muses and poets, from the ‘Paper Museum’ of Cassiano dal Pozzo in Rome (British Museum, London).

2.4 Teaching methods should also depend on the students’ level

But what should the ratio be between parsing / translating and other, more active methods? This partly depends on how far advanced the students are. Beginners who have not yet mastered morphology and the most important rules of syntax will not acquire a high level of Latin or Greek through simple exposure to spoken language alone. They need to have that painstaking training that translation can give – and here I always mean translation in both directions. Learning how the languages differ by contrasting them in translation can give fast initial progress.[6] On the other hand, a more advanced student who has mastered these basics will benefit much less from translation practice; such a student needs to read, read, read, and do active exercises.

At this level, acquisition through reading or listening works much better, because we are now trying to learn collocations and idiomatic expressions. Of course these can to some extent be learned from phrase books, but exposure to real texts is key. Those of us who are teaching and writing in languages that we did not learn in early childhood have all had the same experience: in the early stages, vocabulary lists, grammars, and at least occasional translation gave us the fastest progress, but once we reached a certain level of fluency, most of what we acquired in addition came from reading and listening.

The somewhat uncomfortable truth is that language is much more than grammar and vocabulary; John Sinclair (1991) came up with the concept of the ‘idiom principle’: large portions of what we say and write is not constructed from scratch, but consists of ready-made, smaller chunks, chunks which can in principle be analysed, but which in practice are most often learned and processed as a whole. Such ready-made chunks, like on the other hand, are best acquired through exposure to written or spoken language, and although grammar and vocabulary can give us an initial shortcut to an intermediate level of language competence, we can only achieve true mastery through long hours spent with original texts or native speakers; otherwise, our language will be ‘correct’, but will not sound natural.

A speculative reconstruction of the Library of Alexandria, Egypt.

This brings me to the end of this first article. I hope that readers can see how deeply Latin literature depends on translation from Greek; translation from Latin into Greek was also common, especially at a more mundane level. We can only fully grasp this kind of literature if we have some practical translation experience ourselves. In the second part of this article, I discussed what place translation practice should have in language teaching. There is no one-size-fits-all answer to this question; the exact place that translation should occupy depends on our specific teaching goals, but also on the linguistic proficiency of the students.

My next article will be more restricted in its scope, insofar as it focuses on the teaching of Classics in universities in the English-speaking world. The situation in continental Europe, for instance, is significantly different. The problems and challenges I shall outline in my second piece are thus by no means universal in nature.

Wolfgang de Melo is Professor of Classical Philology at Oxford. He has published on early Latin, especially Plautus and Roman comedy, and on Varro. He teaches linguistics and comparative philology and has a special interest in linguistic typology. He has previously written for Antigone on grammatical gender, Latin spelling, Latin accents, and linguistic irregularity.

Part II of this article may be read here.

Further Reading:

Fraenkel’s brilliant study of which elements in Plautus are his own creative inventions and which go back to the Greek originals was originally published in Germany in 1922; now we have an English translation of the work, which I recommend more than the German version or its Italian translation because it has a thoroughly updated bibliography and further notes: E. Fraenkel, Plautine Elements in Plautus (Plautinisches im Plautus), trans. by T. Drevikovsky and E. Muecke (Oxford UP, 2007). For a comparison of the Menander papyrus with Plautus’ Bacchides, the classic work remains E.W. Handley’s Menander and Plautus: A Study in Comparison (Inaugural lecture, University College London, 1968).

A straightforward introduction to the study of second-language acquisition is Rod Ellis’ Second Language Acquisition (Oxford UP, 1997). For the ‘idiom principle’, see especially John Sinclair’s Corpus, Concordance, Collocation (Oxford UP, 1991).


1 One such instance in Plautus is Asinaria 63–4, where we find a Greek-style nominative-and-infinitive construction instead of the expected accusative-and-infinitive type. When Catullus does such a thing (4.1–2), it is in deliberate imitation of Greek, but in Plautus’ time, Greek was not yet a prestige language to be copied.
2 A good example of such a discussion can be found in Cicero’s Academica at 2.17–18, which can be read in English and Latin.
3 Compare English “I am glad that he could come to the meeting”; German “ich bin froh, dass er zum Treffen kommen konnte”; and Dutch “ik ben blij dat hij naar de vergadering kon komen”.
4 Such a contrastive approach also makes us aware of ‘false friends’: German anrufen means “to call someone on the phone”, while Dutch aanroepen means “to call upon the Lord”, a religious meaning which also still exists in German, where it is, however, marginal. If you want to call someone on the phone in Dutch, the verb is bellen (which in German means “to bark”!).
5 Translation is not among them, partly because the tests are used for international students whose first languages differ vastly; it would be difficult to standardise the tests if translation were part of them.
6 The contrastive approach is also particularly useful for teaching pronunciation; learners of English benefit from general discussions of the English sound system, but often they need that explicit contrast with their native sound systems in order to improve their pronunciation.