Gorka corrected eleven of my chapters without telling me

12 January. Eduvigis Sanz, my editor for the past nine years, rings me at nine in the morning to say that the Nell Harrower novel runs to six hundred pages and goes to press in six weeks. One person cannot do it. There will be two of us: the odd-numbered chapters mine, the even ones Gorka Ruiz de Azúa's, a translator with two published books, whom I have never met. I send over a glossary of fourteen words, the ones that appear more than twenty times in the book. I counted them.

13 January. Gorka sends his: a hundred and nine entries. They are well made. A glossary of a hundred and nine entries gets consulted twice and then abandoned; one with fourteen is memorised in an afternoon and never opened again. I do not put it to him like that.

16 January. The dog. In my chapter one he is called Botas; in his chapter two, Boots. He argues that proper names are not translated, which is the first thing they teach you. I send him page fourteen, where the mother explains why she gave him that name: for his white paws, which look like outdoor boots. If you leave it in English, a Spanish reader gets an explanation that explains nothing. He replies eleven minutes later: "You're right about chapter one. In chapter two I'm leaving it, nobody explains it there."

17 January. Chapter seven, the worst in the book to translate: the grandmother talks for nine pages and does it all in diminutives. In English they are not diminutives, they are small adjectives placed in front, and if I translate them one by one the woman sounds twee rather than old. I solve it by loading the diminutives where the grandmother talks about things that no longer exist and stripping them out everywhere else. It is the decision I am happiest with so far.

19 January. My chapter five comes back typeset and two paragraphs I had joined are separate again. I put it down to copy-editing and carry on. I am on chapter thirteen.

21 January. Chapter twenty-one. Harrower sets a seven-word sentence alone in its own paragraph, three times in the book, always when somebody decides something without saying so. I leave it alone too. I note in the glossary that those three sentences are not to be touched, and I send the warning round.

23 January. Eduvigis on the phone: "Sort it out between you, I've got the printers on my back." I ask her what needs sorting out. She says Gorka will tell me.

24 January. Gorka does tell me, in writing, and it has to be said that he does not hide: when he signed, he was also given the final harmonisation, going over all thirty-eight chapters at the end to unify criteria. He did not wait for the end. He has done eleven of mine, chapters one to twenty-one. He has put Boots in all of them. He has taken out the grandmother's diminutives. He has changed "the ironing room" to "the utility room".

25 January. There is no utility room in that house. There is a room with an ironing board and a window onto the courtyard, and it is drawn on the plan Harrower put on the endpaper, the folded sheet at the front. I write to him with the page number. I add that the mother's sentence, with Boots, comes out as "I called him Boots for his white paws", which is a sentence that means nothing. Eleven times. And I ask him what he has done with the grandmother's nine pages, which is the only part I really care about.

26 January. He does not reply all day. That night I look at my chapter seven in the shared file: the diminutives are back in all nine pages, every one of them, including where the grandmother is talking about the present. The woman sounds twee. I go to bed.

27 January, 6:41 in the morning. A four-line message: he has spent the whole night undoing his own work chapter by chapter and, on reaching his chapter fourteen, he stopped at page two hundred and eleven. The daughter takes the dog to Leeds and that night, in the flat on Marsden Street, she calls him Boots for the first time. And the dog answers just as he always did, after nine years responding to another name.

He is the one who writes it, not me: the author changes the dog's name halfway through the book. For two hundred and ten pages the mother calls him by a name that means something, and from there on the daughter calls him by a sound that means nothing, because the daughter was not there on the day of the white paws and does not know where it comes from. That is why there is a plan on the endpaper. That is why the grandmother speaks in diminutives. That is why chapter thirty-eight is called "What Is Said Out Loud". The whole book is about who gives out names and who inherits them without the manual.

Same day, afternoon. I have read it twice to be sure and it is true. And I had read the whole novel before starting, in December, plan and all. I did not see it. I have thirty-one translated books behind me and I read looking for the voice: how a sentence breathes, where the narrator's accent shows. Names I take as given. A translator with two books, who still reads counting the words one by one, stopped at the one I had gone over thirty times.

28 January. I propose the only thing that works: the dog is called Botas up to page two hundred and eleven and Boots from there on. Both of the things we were each defending, each in its place. He replies that it is going to look like a mistake. I tell him we will explain it to Eduvigis together, with the page in front of us.

29 January. Eduvigis says yes, says she wants it in writing in the translation report, and says not to tell the copy-editor until the end. She also says something I apologise for noting down, because I have not heard it from her in nine years: "So you've actually read it."

4 February. We are on chapter twenty-four and we are going to make it. Today he sent me a one-line query, the first he has sent without solving it himself first. Underneath, with no comment, he had pasted the sentence from page two hundred and eleven: "The dog, who had spent nine years answering to another name, lifted his head just the same."