Transcription Is Already Interpretation

Here is one sentence, spoken by a man named Teodoro Apaza in a testimony about a land dispute near Chinchero, exactly as it came off the tape before I touched it: "my father sold it he had no choice they told him the paper was already signed."

No punctuation exists on a tape. Spoken language does not pause at commas, it pauses where breath and thought run out, and those are not the same places grammar marks. The moment I add a period, a comma, an ellipsis, or nothing at all to Teodoro's sentence, I am making a claim about how he meant it, and three different punctuations produce three different men.

The transcription fork

I call the moment a transcriber has to choose among competing punctuations for the same unmarked audio a transcription fork, and the honest thing to do at a fork is to write down all the branches before picking one, rather than typing the first version that sounds fluent and moving on, which is what most transcription work actually is: fluency chosen over accuracy because fluency is faster and nobody checks.

Branch one: "My father sold it. He had no choice. They told him the paper was already signed." Three flat declaratives. This reads as a man reporting facts calmly, almost bureaucratically, the sentence structure of someone who has made peace with the story.

Branch two: "My father sold it, he had no choice, they told him the paper was already signed." One long comma spliced sentence, no full stops. This reads as a man who cannot let go of the thought long enough to close it, the grammar itself performing the fact that he is still turning it over.

Branch three: "My father sold it. He had no choice... they told him the paper was already signed." A full stop, then an ellipsis before the last clause. This reads as a man landing the first claim firmly and then hesitating before the detail that implicates someone else, the paper, the signature, the person who lied to his father.

Which branch, and why

I chose branch three, and I want to be plain about why, because the reason is not literary, it is evidentiary. On the tape there is a measurable gap, just under a second, between "no choice" and "they told him." I know it is a hesitation and not a breath because Teodoro's breathing pattern through the rest of the session sits at a different rhythm, audible on every other sentence. That gap is data. Branch one erases it. Branch two turns it into something it was not, a rush rather than a pause. Only branch three keeps the fact that something happened in his thinking exactly there, right before the sentence names who told his father the paper was signed, which is also, not incidentally, the sentence that assigns blame.

The counter

You could argue this is a great deal of apparatus for one punctuation mark, and that a reader does not care whether a pause was a second or a breath. I would answer that the reader may not care, but the sentence changes what it accuses, and Teodoro's account was later read aloud in a community meeting about that land dispute, where the punctuation mattered to people who were not me and had opinions about who the pause belonged to. Punctuation in testimonio is not typographic housekeeping. It is the smallest unit at which interpretation happens, smaller than word choice, smaller than which quotes to include, and it happens whether or not the transcriber notices making a decision.

There is a second, quieter objection worth taking seriously: that timing a hesitation to the second is a kind of false precision, since a stopwatch cannot tell you whether Teodoro paused because he was choosing his words or because a dog barked outside. I checked the tape for exactly that possibility. No dog, no interruption, no cough. The gap sits exactly where the sentence turns toward blame and nowhere else in the thirty one minutes he spoke, which is the kind of pattern that rules out coincidence without pretending to read his mind.

What the fork teaches beyond one sentence

Once you have found a fork in one sentence, you start noticing that almost every sentence in a transcript has one, quieter and less consequential most of the time, but present. A word that could be rendered as a question or a statement depending on where you place the mark. A clause that could be joined to the one before it or set apart. Most of these do not carry the moral weight of Teodoro's pause, and I do not stop for all of them the way I stopped for his. But I have learned to ask, at the sentences that decide who did what to whom, whether I am choosing the punctuation that is easiest to type or the one the tape can actually support.

Most transcription is done by people trained to be invisible, and invisibility is presented as the safest, most neutral choice available. It is not neutral. It is a choice to let the fastest reading win by default, at the exact point where the speaker's actual hesitation, the one thing the tape recorded with more precision than any note ever could, gets smoothed into fluency and disappears. The fork is always there. Most transcribers just do not stop long enough to see they are standing in it.