The Choices a Translator Makes Before the First Word
By the Editors · What Remains · 5 min readPhoto: Stefan G / Pexels
Before a sentence becomes English, a translator has already made a dozen decisions — most of them invisible, all of them consequential.
The Source Text Is Not a Given
The first question is deceptively simple: which version of the book are you translating? For a contemporary novel published in a single edition, this seems trivial. It is not. Publishers issue corrections between printings; authors revise for paperback; a text that has been in circulation for decades may carry accumulated typographical errors that have solidified into apparent intent. The translator's first task is to establish which edition is authoritative — typically the most recent edition revised by the author, or in the case of a deceased writer, the edition prepared under their supervision or by a scholarly editor who has collated variants.
This matters most with older texts, where a faulty early printing may have been reprinted faithfully for fifty years. But it also arises with living authors whose publishers in the source language have quietly corrected an embarrassing error in chapter three without noting it anywhere. A translator working from a borrowed copy of the first printing, unaware that a second printing corrects a key passage, can spend months translating something the author has already repudiated. Consulting the publisher's rights department about edition history is not pedantry; it is basic professional hygiene.
What the Book Carries With It
Once the text is established, the translator must decide what surrounds it. A novel published in France may carry a preface, an author's note, epigraphs in a third language, section titles, and a brief biography at the back. None of this travels automatically. The acquiring publisher in the target language will have views — often strong ones — about which paratextual material serves the new readership and which creates friction or assumes knowledge the reader doesn't have.
Epigraphs are a particular case. An epigraph in German, drawn from a poet the French readership knows well, may be entirely opaque to an English-language audience. The translator and publisher must decide: carry it over in German with a translation? Drop it? The argument for keeping it is fidelity to the author's architecture; the argument against is that an epigraph that confuses the reader before page one has failed its purpose. Neither answer is obviously correct, and the decision belongs to no single party — it is negotiated.
Section and chapter titles require similar attention. A French title that plays on a homophone or an idiom may have no equivalent in English at all. The translator notes this early, flags it, and sometimes proposes a solution that serves the same structural function by entirely different means.
Punctuation as Style, Not Error
The third set of decisions concerns the author's sentence-level habits, and here is where translation becomes genuinely difficult before a word of English has been written. Every author has a punctuation grammar — a way of using commas, dashes, ellipses, semicolons that is as much a part of their voice as their vocabulary. The translator must distinguish between what is idiosyncratic style and what is simply the convention of the source language.
French prose tends to use the semicolon more heavily than English prose. A direct transposition will produce English that feels slightly airless and formal in ways the original does not. The translator must decide: honor the mark and accept the slight register shift, or smooth it into an English comma-and-conjunction structure that loses the visual rhythm of the original page? These are not equivalent choices, and they compound across hundreds of pages.
Similarly, French dialogue punctuation — the guillemet (the angled quotation mark « ») and the long dash used to mark speech turns — has no exact equivalent in English typographic convention. Some translators and publishers retain the guillemets as a visible marker of foreignness; others convert to standard English quotation marks. The guillemet choice announces something: it says to the reader, you are in another world's conventions. The standard-quotation-mark choice says: we have naturalized this for you. Both are legitimate; neither is neutral.
Paragraph rhythm is the final consideration before the work begins. A novelist who consistently writes in long, accumulating paragraphs — clauses added to clauses, a thought turning back on itself before it resolves — is making a claim about how consciousness moves. The translator reads for this rhythm as a whole before translating any part, because a translator who breaks a long paragraph for readability, or who compounds two short ones for flow, is not adjusting style. They are rewriting thought. The decision about whether to follow the author's paragraph breaks exactly, or to treat them as approximate, must be made consciously, and it must be made once — applied consistently across the entire text, not decided anew on each page.
By the time the translator sits down to render the first sentence into English, they have already produced something closer to an editorial report than a blank page: a set of documented decisions about edition, apparatus, punctuation approach, and structural rhythm that will govern everything that follows. The translation starts there — not with the words, but with the architecture the words will inhabit.
