Translate Your Own Draft to Find Its Skeleton

Chapter nine of my second novel sat unfinished for six weeks because I could not work out what was wrong with it, so at midnight I gave up editing it in French and translated it into English instead, and by the third page I could see the hole.

The chapter was a visit. A woman goes to see her uncle to ask for money. In French the scene read as tense, the dialogue clipped, the pauses doing what I assumed was serious work. In English, stripped of the French sentence rhythm I had leaned on for six weeks, the visit turned out to contain no actual conflict. The uncle never really refused. The woman never really pushed. Two people discussed a sum of money for four pages and then one of them left, and the tension I had felt while writing it in French was coming entirely from the sentences, not from anything the characters were doing to each other.

The skeleton pass

I now do this on purpose whenever a scene refuses to resolve after two or three drafts in its home language. I call it the skeleton pass. Take the stuck section and translate it, not beautifully, not for publication, just accurately enough to preserve who does what to whom. Read the result cold. A scene with a real skeleton, an actual want opposed by an actual obstacle, survives this badly written pass and still pulls you forward. A scene that only felt alive because of its prose goes limp the moment the prose is gone, and you will feel the limpness within a paragraph.

Once I saw the uncle scene stripped down, the fix took twenty minutes rather than another six weeks. The uncle needed to say no. Not eventually, not implicitly, but at a specific line, on a specific excuse, so the woman had something to argue against instead of a conversation to simply outlast. I rewrote that one line in French, restored the register and the pauses I had cut for the test, and the whole chapter that had defeated me for a month and a half worked on the next read.

Why translation surfaces this and rereading does not

Rereading your own draft in its own language fails at this because you are reading with the sentence rhythm still loaded in memory. Your ear supplies the tension the words no longer contain, the way you can still feel the weight of a bag you set down five seconds ago. Moving into a second language breaks that. You lose your own cadence, the one that was doing the scene's emotional labor for it, and what remains on the page is only what the scene itself is built from. That is uncomfortable to look at, which is exactly why it is useful.

The obvious pushback

The obvious pushback is that most writers do not have a second language sitting ready for a midnight pass, and telling someone to translate their own manuscript to find its problems is advice only a bilingual novelist could give with a straight face. That is fair, and I do not think the second language is actually the load-bearing part of the method. What matters is losing your own sentence rhythm long enough to see the scene without it. A writer with one language can get most of the same effect by paraphrasing the scene aloud from memory, in plain flat sentences, without looking at the page, then comparing that paraphrase to what the draft actually contains. The gap between the two is usually exactly where the skeleton is missing a bone.

What it costs you to skip this

Writers who skip this step tend to discover the missing conflict at the worst possible time, from a reader, months later, in a note that says the middle dragged. By then the fix is not twenty minutes. It is a chapter rebuilt around a character who has already been established as passive for two hundred pages, and every scene after the broken one now needs to be checked for the same fault. A skeleton pass costs an evening. Finding the same problem from a reader costs a rewrite of everything downstream of it, which is the actual argument for doing this early and doing it to every chapter that feels harder to finish than it should.