Every translator confronts, on roughly the third page of every book, the same small problem: a word in the source language that has no clean English equivalent and that the reader cannot be expected to know.
The translator has three options. She can footnote the word and let the reader interrupt her reading to learn. She can gloss it inside the sentence, slipping a short explanation between commas. Or she can leave it untranslated, italicised or not, and trust the reader to either know the word or to learn it from context.
Each choice is a small bet on what kind of reader the book is for. Three recent books show what each bet looks like.
The first is Edward Gauvin's 2025 English version of the Belgian fabulist Bernard Quiriny's collection Histoires assassines, published by Wakefield Press in Cambridge, Massachusetts. The book is full of culturally specific French references: a particular brand of cigarette, a long-dead radio host, an obscure Belgian beer.
Gauvin uses footnotes. Forty-three of them across 178 pages. They are short, at the bottom of the page, in a smaller font, and they explain only what the reader needs to keep reading.
The footnotes work because Wakefield Press is a small independent that publishes books for readers who like footnotes. The Quiriny edition is a hardcover with French flaps. The reader who buys it is, by the act of buying, accepting the footnote contract.
The same translator, the same publisher, the same approach would not work at Penguin. The trade paperback reader expects to read without interruption.
The second case is Megan McDowell's 2024 translation of the Chilean novelist Alejandro Zambra's Childish Literature, published by Viking. The book is full of Chilean idioms, slang, and references to Chilean television and football that the American reader will not know.
McDowell uses in-text glosses. She does not footnote. She does not leave words untranslated. When she needs to explain, she slips a clause into the sentence: "his cousin Patricio, who had played one season for Colo-Colo before his knee gave out," where the original simply says su primo Patricio, ex-Colo-Colo.
The gloss is small. The American reader, having never heard of Colo-Colo, now knows it is a football club. The Chilean reader, reading McDowell's English, might find the explanation patronising. But McDowell's English is not for the Chilean reader. It is for the American who picked up Zambra at the airport bookshop.
Her glosses are, on the whole, deft. Occasionally they sit awkwardly. The Chilean term cuico, which roughly translates as a wealthy class-marked person, is glossed in three different ways across the book: the rich kid Patricio, the boy from Las Condes, and once, untranslated, the cuico. The variation is, in context, doing different work each time.
The third case is the most demanding. The translator Aaron Robertson's 2025 English version of the Cameroonian writer Imbolo Mbue's French novel La saison des marabouts, published by Random House, leaves a large number of French and Pidgin English terms untranslated and unitalicised.
Robertson's preface defends the choice with reference to a 2003 essay by the translator Lawrence Venuti on what Venuti calls the foreignising approach. The reader who picks up an African novel, Robertson argues, should be allowed to feel the friction of an unfamiliar language. To gloss every word is to assume the reader cannot do the work.
The book is, by ordinary trade standards, harder to read than a McDowell-style translation would be. It is also closer to the texture of the original.
Reviews were mixed. The New York Times called the translation "brave and frustrating in equal measure." The Times Literary Supplement called it "the right kind of difficult." Sales were modest. The book was nominated for the National Book Award in Translated Literature.
What these three books make visible is that the footnote, the gloss, and the trust are not equal options. They produce different books. The Quiriny in footnotes is a different reading experience than the Quiriny would be in glosses. The Zambra in glosses is a different book than the Zambra would be in footnotes.
The decision is, in most cases, made by the publisher before the translator begins. Penguin will not accept forty-three footnotes. Wakefield will not accept thirty in-text glosses that smooth a foreign word into invisibility. The translator's job is to choose between the bets the publisher has already placed.
What this means, in practice, is that the same book translated for two different houses would emerge as two different books. The English-language reader, encountering only one of them, would never know.
There are translators who think this is a scandal. There are others who think it is just the work.
Both groups keep translating. The footnotes accumulate, or they do not. The glosses get sharper. The trust gets tested, book by book, in the small economy of how much friction a reader will tolerate before she puts the book down.
The honest translator names the choice in a translator's note. The dishonest one pretends the choice was not a choice. The reader, if she pays attention, can tell which is which by page thirty.