The problem with translating dialect is that dialect is geography, and geography does not translate. A character who speaks Andalusian Spanish in Granada is a character from a particular place. Render her into English and she is from no place, unless the translator gives her one.
Three recent books, each translated by a different working translator, take three different positions on the question. None is fully successful. All three are interesting in their failure.
The first is Lisa Dillman's 2024 English version of the Andalusian writer Pablo García Casado's novel El verano de Chana, published by And Other Stories in Sheffield. The novel is set in a village outside Antequera in 1979. Roughly half the dialogue is in a thick Andalusian Spanish that drops final consonants, swallows syllables, and substitutes a particular set of provincial idioms.
Dillman, who has translated nine Latin American novels and is a professor of Spanish at Emory, chose to render the Andalusian as a generic American rural English. Not Appalachian, not Southern, not anywhere specific. She uses contractions, drops the occasional g, and writes the dialogue as if the speakers were tired and not in a hurry.
The result is readable. The characters sound like rural people. They do not sound like Andalusians, because no English can do that, but they also do not sound like Mississippians, which would have been the wrong solution.
Dillman's translator's note is candid. She considered Appalachian English. She considered Yorkshire English. She rejected both because they would have imported a specific American or British place into a Spanish village, and the reader would have spent the book in two places at once.
The second case took the opposite risk. Aniruddhan Vasudevan's 2025 translation of Perumal Murugan's Tamil novel Pyre, published by Pushkin Press, renders the rural Kongunadu Tamil of the original into a specific English: the English of rural Tamil Nadu, with Tamil words preserved, English syntax slightly bent, and rhythms that mirror the original.
The book is harder to read than Dillman's. It is also, to a reader willing to do the work, much closer to the source. Murugan's village is on the page. The English does not pretend to be neutral.
Vasudevan's preface defends the choice with reference to the Nigerian novelist Chinua Achebe's well-known 1965 essay on English as an African language. The English used to render an Indian village, Vasudevan argues, should be an English that has been shaped by India.
Some reviewers found the result mannered. The book sold modestly. It is also, four years in, the standard English Murugan. New readers will encounter Tamil rural life in the voice Vasudevan made.
The third case is the most idiosyncratic. The Scots writer and translator Kirsty Logan, in her 2025 English version of the Quebec novelist Heather O'Neill's Les filles perdues de la rue Sainte-Catherine, took the radical step of rendering the Joual passages, which fill roughly a third of the novel, into Glasgow Scots.
The book is published by Charco Press in Edinburgh and is divided. Some readers find the Scots brilliant. Others find it bizarre. The Quebec critic Marie-Hélène Voyer, in a 2025 essay for Le Devoir, called the translation "a Glaswegian novel about Montreal," which Voyer meant as both compliment and complaint.
Logan's defence is that O'Neill's Joual is a specific working-class urban dialect with a particular relationship to a standard French it cannot quite escape. Glasgow Scots has the same relationship to standard English. The two are, in Logan's view, structurally parallel.
Whether that parallel survives the cultural transposition is the question the book leaves open. Glasgow is not Montreal. Scots is not Joual. But, Logan argues, no English is Joual, and the choice was between a specific English and a falsely neutral one.
The translator who takes the dialect problem seriously has, in practice, only three options.
One: render the dialect as a generic non-standard English and accept the flattening.
Two: render it as a specific English dialect and accept the false geography.
Three: invent a hybrid English that signals difference without claiming a real place.
Each translator, each book, picks one. The picks accumulate, and the field's working theory of dialect, never written down, slowly shifts.
In 2026 the working theory is roughly this: option one for languages with many English translations already in circulation, where the reader will not notice the loss; option two for languages where a specific English dialect maps closely enough to the source to risk it; option three for languages where the translator is willing to be unpopular.
All three approaches will be defended in the next decade, and all three will produce books that some readers love and others put down at page forty.
The dialect that does not translate is, in the end, a permanent loss. The translator who admits this in her preface is the translator who has done the work.
Filed under