Repetition Means Something Different When Spoken
Dona Epifania said the word "sola," alone, three times in the space of one paragraph describing the year her sons left for the coast, and the first editor who read the transcript circled all three and wrote "repetitive, cut two."
On a page, a repeated word usually is a fault. It signals a writer who has not found a synonym, or has not noticed the echo. In speech, repetition rarely means the speaker forgot she already used the word. It means the word is doing something the sentence around it cannot do alone, and the three instances of "sola" in Epifania's testimony are not the same repetition performing the same function. They are three different jobs wearing the same word.
The three functions
I sort spoken repetition into three functions, and once you have the categories, a transcript stops looking careless and starts looking engineered, because in practice it usually is, even when the speaker has never studied rhetoric a day in her life.
The first function is underline: the word repeated immediately, back to back, to raise its weight, the way Epifania said "sola, sola" when describing the first night after her younger son left, no pause between them, the second instance simply louder in the memory of how she said it. The second function is the gathering pause: the word repeated after a break of several sentences, used to return to a thought and check that the listener is still holding it, which is what her second "sola" did, twenty seconds later, after she had drifted into a story about her husband's illness and needed to pull herself back to the point she was making about being alone in the house. The third function is the ritual marker: a word repeated at a fixed structural position each time a related topic recurs across the whole two hour testimony, which is what her third "sola" did, appearing at the close of a section the same way it had closed two earlier sections, a signal that this part of the account was finished, functioning the way a paragraph break functions on a page, except spoken.
Writing it down without erasing the job
I kept all three instances of "sola" in the final text, but I did not treat them identically on the page, because treating them identically would have hidden the fact that they were doing different work. The underline pair, "sola, sola," I kept exactly adjacent, no punctuation between them, because separating them with a comma would have turned an emphasis into a list. The gathering pause I set off with a line break before it, since on the page a reader needs some visual signal that time and a digression have passed between the first sola and the second. The ritual marker I left to close its paragraph exactly where it closed hers, resisting the urge to add a summarising sentence after it, because the repetition itself was the signal that the section had ended, and adding a sentence after would have stepped on the very thing that made it a marker.
The objection
An editor could reasonably object that most readers will not notice, let alone credit, this much structural care taken over one recurring word, and that from the outside my text simply looks like it failed to cut a repetition a competent line edit would have caught. I do not disagree that most readers will not consciously register it. But readers register rhythm even when they cannot name it, and a testimony with its repetitions intact reads as spoken, present, alive in a way that the same account with every echo removed does not, even to readers who could not tell you why.
The cut version of Epifania's paragraph, with two of the three solas removed as the first editor suggested, is grammatically fine and rhythmically dead. It reads like a report of loneliness rather than an instance of a woman performing, in real time, three different ways of being alone: sharply, in a gathering thought, and as a marker she had learned to place at the end of a subject she needed to close before she could go on to the next one.
I sent both versions, cut and uncut, to a translator colleague who had never met Epifania, without telling her which was the original. She picked the uncut version as the one that sounded like an actual person remembering something, and could not say why beyond that the repeated word felt placed rather than sloppy. She had located the function without having the category for it, which is usually the best evidence that a category is describing something real rather than something I invented to justify a preference.