Three Languages, Three Different Novels

I have three unfinished manuscripts about roughly the same family, one drafted in French, one in English, and one in a French thick with Darija, and the father in all three has never once said the same thing in two of them.

In the French draft the father is ironic, evasive, answers questions with other questions, and the scenes around him run long because nobody in his house says what they mean on the first attempt. In the English draft, written years later after I moved to Marseille and then spent a year working in London, the same father says what he wants inside two lines of dialogue and the scenes are half the length. I did not decide to write him differently. I sat down each time to continue the same story and a different man showed up, wearing the same name.

The language ledger

For a while I treated this as a curiosity, then as a problem to fix, then finally as information worth tracking on purpose. I keep what I call a language ledger now: a page, updated every few months, noting which kinds of scenes come out alive in which language and which come out flat. Not opinions about the languages themselves, just a record of results. Confrontation scenes between men of the same generation come out sharper in the Darija-inflected French, because the register options for showing contempt without raising your voice are richer there than in my English. Scenes of a character explaining themselves to an institution, a clerk, a doctor, a judge, come out cleaner in English, because English gives me shorter, flatter declarative sentences that a bureaucratic voice can hide behind convincingly.

The ledger is not mystical. It is a log of which tool did which job well the last five times I used it, the same kind of record a carpenter keeps about which chisel chips a particular grain of wood and which one splits it. Once I had eight months of entries, the pattern was obvious enough that I now draft certain scenes directly in the language the ledger recommends, rather than writing them in whichever language the manuscript happens to be in and fighting the mismatch for a week.

What this changed for the family novel specifically

The three drafts of the family novel are not actually three different attempts at one book anymore. They have become, without my planning it, a testing ground for which language the finished book should be written in scene by scene, before I commit to a single language for the real draft. The confrontation between the father and his brother over the inheritance is staying in the register-rich French. The scene where the daughter explains to a French housing officer why her documents do not match is staying in flat, declarative English, translated back into French only at the very end, once the bureaucratic flatness is locked in place and cannot be softened by accident.

The sceptic's version of this

A sceptic would say none of this is about the languages at all, that I am simply a more practiced writer now than I was in the earliest French draft, and that the improvement I am crediting to English is really just five more years of practice showing up in whichever draft happened to be open at the time. I take that seriously, because it is probably half true. Skill accumulates regardless of which language you are writing in, and some of what the ledger records is really a timeline of my own development wearing three different disguises.

But it cannot be the whole explanation, because the pattern holds even in scenes I draft in parallel, the same week, working from the same outline. A confrontation drafted in French on a Tuesday and again in English on the Wednesday, with no improvement in craft between the two days, still comes out sharper in French. Something about which options a language makes cheap and which it makes expensive is doing real work here, independent of how good I have gotten at using either one. The ledger cannot separate the two effects perfectly. It does not need to. It only needs to tell me, before I waste a month, which language is going to fight me on a given scene and which one is going to help.