Reading Rome in Translation

1. The arrangement

The Roman Archive presents each work in translation, and, where it survives in the corpus, beside its original. Cicero's letters to Atticus appear as parallel text: the Latin of L. C. Purser's Oxford edition (1903) on one side, the English of E. S. Shuckburgh (1899–1900) on the other. Caesar's commentaries and Seneca's letters carry their Latin likewise; Augustus's Res Gestae stands beside the Latin recovered from its inscriptions. Marcus Aurelius's Meditations, written in Greek, reaches you for now in the English of Meric Casaubon alone.

The arrangement looks neutral (two columns, one meaning), but no translation is a transparent pane of glass, and a reading interface that retrieves and quotes from the English column owes its readers a clear account of what that column is. This essay gives that account: what the English gives, who each translator was, what each rendering loses, and what sits beneath the Latin column itself.

2. What the English gives

For a reader without fluent Latin or Greek, the translation is the difference between a work being legible and being a locked room. It carries the sense: who is being asked what, which deed is being claimed, which consolation is being reached for. And where the original sits alongside, a reader with a little of the language can check the English against it phrase by phrase, the slow, comparative reading these texts reward, and the reading the parallel display is built to invite.

There is a further, structural point. The Archive's retrieval runs over the English (the embeddings are of the translation, not the original), so the translation is not merely what you read; it is what the system searches. The translator's choices set the boundary of what a query can find. That makes the question "whose English is this?" a technical question as well as a literary one.

Translation studies has a name for the tell-tale residue this leaves. Translated text is not linguistically identical to text originally composed in the target language: it carries systematic traces of the source, and of translation as a process, that make "translationese" measurable: the tendency to normalise, to explicitate what the original left implicit, and to shadow the source's structure (Baker 1993; Volansky, Ordan & Wintner 2013). For a reader this is a matter of texture; for a retrieval system it is a matter of recall. A query embedded in ordinary modern English is being matched against text whose vocabulary and phrasing were pulled, to a measurable degree, toward a Latin or Greek original and toward the idiom of the translator's own century. The mediation the rest of this essay describes in literary terms is, from the system's side, a bias in the embedding space itself.

3. Each translator has a date and a temperament

The Archive's translations were not chosen by literary judgement; they were chosen by copyright. What is freely available is the public-domain stratum, and that stratum has a shape: it runs from the 1630s to the 1920s, with one modern exception. It is worth knowing who these people were:

• Meric Casaubon (1634) made the first English translation of the Meditations, a generation before Pepys. It is a magnificent period piece: expansive, sermonic, often closer to paraphrase than translation, with seventeenth-century theology audible under the Stoicism. Casaubon's Marcus is further from the Greek than any other rendering in the Archive is from its Latin, and his prose is also the finest thing in it.
• William Melmoth (1746, revised by F. C. T. Bosanquet) turned Pliny's polished letters into the polite English of the Georgian essay. Famously elegant, famously free: Melmoth smooths, expands, and "improves," and the revision only partly reins him in.
• William Duncan (whose Caesar the Archive carries in an 1856 printing) and W. A. McDevitte & W. S. Bohn (1870–1872) render the commentaries in the workmanlike English of the schoolroom crib: accurate at the level of the clause, tone-deaf to Caesar's studied plainness, which was itself a rhetorical performance.
• E. S. Shuckburgh (1899–1900) translated the whole surviving correspondence of Cicero in four volumes: dignified, reliable Victorian prose, still the standard free English of the letters.
• R. M. Gummere (1917–1925) and H. E. Butler (1920–1922) are early Loeb Classical Library volumes (Seneca and Quintilian respectively), from the era when the Loebs aimed at serviceable facing-page English rather than literary character.
• Thomas Bushnell (1998), the one modern voice, translated the Res Gestae and released it under a free licence, which is the only reason the Archive's newest translation is newer than 1925.

Each writes with the idiom of their own century: when Marcus Aurelius answers a question in this Archive, the words carry a faint accent of 1634, and Pliny arrives sounding like 1746. The Archive names the translator on every citation because the Englishness a reader hears is Casaubon's or Melmoth's, and it must never be mistaken for the author's.

4. What the English loses

The losses differ by author, because each author's Latin (or Greek) does different work:

• Cicero's letters are the most informal Latin prose to survive antiquity: elliptical, allusive, full of Greek code-switching. He drops into Greek for irony, intimacy, or discretion, much as a bilingual correspondent does today. Shuckburgh renders the Greek and the Latin into one continuous English, and that seam, visible in the original as a change of alphabet, disappears. The puns mostly evaporate too. And in his speeches and treatises Cicero composed for the ear, ending periods on rhythmic clausulae so regular that they can be studied statistically (Zieliński 1904); no translation carries prose rhythm across.
• Caesar's plainness is a performance. The third person ("Caesar ordered…"), the spare vocabulary, the unhurried syntax in the middle of catastrophe: this is a general writing his own dispatches as though an impartial historian were following him around (Riggsby 2006). A flat English translation reproduces the plainness but not the artfulness of it; it reads as mere report, which is exactly what Caesar wanted readers to think it was.
• Augustus's Res Gestae is lapidary in the literal sense, composed to be cut into bronze and stone. Its Latin is terse, official, exact with numbers and offices, first person throughout: "I built… I gave… I refused…". English that runs smooth softens the granite; the numerals and the refusals are the style.
• Marcus's Greek is private. The Meditations were notes to himself: compressed, repetitive, sometimes ungrammatical, thinking rather than prose (Hadot 1998; Brunt 1974). Casaubon unfolds these knots into rolling seventeenth-century periods. What is lost is not accuracy so much as texture: the reader of Casaubon cannot tell which sentences Marcus laboured and which he jotted.
• Seneca's epigrams travel best. His pointed, end-stopped sentence (built to be quoted) survives translation better than any other style in the Archive, which is partly why Seneca has always been the most quoted Roman in English.

5. The text beneath the translation

The Latin column is not bedrock either. Works that survived by copying carry manuscripts that disagree; editors chose; a different editor would print a different word in places. A parallel display can make the original look more fixed than it is, so it is worth knowing how each text reached its editor:

• The letters to Atticus vanished for most of a millennium. Petrarch found a manuscript at Verona in 1345; that codex is lost, and the modern text descends from copies made after his find. We are reading a copy of a copy of a rediscovery, and only one side of it, since Atticus's replies do not survive.
• The Res Gestae survives as no manuscript at all: the original bronze at Rome is gone, and the text is assembled from stone copies in provincial cities, chiefly the bilingual inscription at Ankara (the Monumentum Ancyranum), with gaps restored by editorial conjecture from Mommsen (1883) to Cooley (2009). Some words a reader meets in the Archive's Latin column are a scholar's reconstruction of what the stone once said.
• The Meditations rests on one complete manuscript (Vaticanus graecus 1950) and one lost one, from which Xylander printed the first edition in 1559. Casaubon translated a text that was itself barely a century old in print.
• Quintilian circulated for centuries in a mutilated text until Poggio Bracciolini found a complete manuscript at St Gall in 1416, during a break in the Council of Constance. Seneca's letters travelled the Middle Ages split into two halves (1–88 and 89–124), reunited by print (Reynolds 1965).

The texts themselves deserve confidence: their overall fidelity is one of the triumphs of the manuscript tradition (Reynolds & Wilson 2013). What needs precision is the phrase "parallel original". What stands in that column is an editor's best reconstruction, with a history, beside a translator's interpretation, with a date.

6. How the Archive should be read

Three practices follow:

1. Treat the English as a guide to the original rather than a replacement. When a passage matters, look across the columns; the friction between them is where the act of translation becomes visible, and the parallel corpus is most useful precisely there.
2. Keep the translator's name attached. The Archive shows it on every citation. "Casaubon's Marcus" and "Shuckburgh's Cicero" are accurate descriptions; "Marcus" and "Cicero" alone, for an English quotation, are not quite.
3. Expect the seams, and value them. A smooth surface would be a false one. The Archive does not try to hide the seams: it names every edition and translator, shows the original where it has one, and lets the reader see that reading Rome in English is reading Rome through other centuries' eyes. Read knowingly, that becomes a second subject of the encounter rather than a defect of it.

References

• Baker, M. (1993). Corpus Linguistics and Translation Studies: Implications and Applications. In M. Baker, G. Francis & E. Tognini-Bonelli (eds.), Text and Technology: In Honour of John Sinclair, 233–250. John Benjamins.
• Brunt, P. A. (1974). Marcus Aurelius in his Meditations. Journal of Roman Studies 64, 1–20.
• Cooley, A. E. (2009). Res Gestae Divi Augusti: Text, Translation, and Commentary. Cambridge University Press.
• Hadot, P. (1998). The Inner Citadel: The Meditations of Marcus Aurelius. Harvard University Press.
• Hutchinson, G. O. (1998). Cicero's Correspondence: A Literary Study. Oxford University Press.
• Mommsen, T. (1883). Res Gestae Divi Augusti ex monumentis Ancyrano et Apolloniensi. 2nd ed. Berlin: Weidmann.
• Reynolds, L. D. (1965). The Medieval Tradition of Seneca's Letters. Oxford University Press.
• Reynolds, L. D. & Wilson, N. G. (2013). Scribes and Scholars: A Guide to the Transmission of Greek and Latin Literature. 4th ed. Oxford University Press.
• Riggsby, A. M. (2006). Caesar in Gaul and Rome: War in Words. University of Texas Press.
• Venuti, L. (1995). The Translator's Invisibility: A History of Translation. Routledge.
• Volansky, V., Ordan, N. & Wintner, S. (2013). On the features of translationese. Digital Scholarship in the Humanities 30(1), 98–118.
• Zieliński, T. (1904). Das Clauselgesetz in Ciceros Reden. Leipzig: Dieterich.

Translations discussed: Casaubon (1634); Melmoth, rev. Bosanquet (1746/1878); Duncan (1856 printing); McDevitte & Bohn (1870–72); Shuckburgh (1899–1900); Gummere (1917–25); Butler (1920–22); Bushnell (1998). Latin text of the letters: Purser (1903), Oxford Classical Texts.