What the Tape Recorder Cannot Keep

Don Anselmo Huaman spoke for thirty eight minutes about the year the mining company moved the boundary stones above Q'eros, and his sister Paulina, sitting against the far wall, corrected him twice. Neither correction is on the recording. The microphone was pointed at Anselmo, the way it always is, and a microphone pointed at one mouth cannot hear the room it is sitting in.

This is the part of the work nobody warns you about before you buy your first recorder. You imagine the device as a net that catches an interview. It catches a voice. If four other people are present, which in Andean households they usually are, the recorder gives you the illusion of a single speaker delivering a monologue, when what actually happened was a conversation with one person doing most of the talking and several others doing the correcting, the laughing, the standing up and leaving when a name came up they did not want said in front of them.

The room track

I now keep what I call a room track, a second column of notes taken during the session, not about what the speaker is saying but about what the room is doing while he says it. Who moved. Who left. Who spoke without being recorded because they were behind the microphone rather than in front of it. The room track is written in real time, in shorthand, on the same page as the transcript notes but never mixed into them, because the two are different kinds of evidence and mixing them is how a false confidence gets built into a text.

For the Anselmo session the room track reads: Paulina, seated left, corrects the year, 1987 not 1986, no reaction from Anselmo. Second correction, the name of the engineer, Paulina again, and this time Anselmo stops mid sentence for three seconds before continuing with her version. Nephew present, silent the entire thirty eight minutes, left once during the part about the boundary stones and returned after.

Why the correction is the story

When I later wrote the testimony, I did not smooth Paulina out of it. The published version reads: "Anselmo said 1986. His sister, from across the room, said 1987, and he did not argue." That single line does more work than the thirty eight minutes around it, because it shows the reader something the recorder could never have shown on its own: that this is a memory under negotiation inside a family, not a fact delivered by one authority. A transcript that only has Anselmo's voice would have printed 1986 as settled. It was not settled. It was contested in front of me, and the contest belongs in the text as much as the date does.

I have gone back to old recordings, single speaker, no room track, and I can hear a rustle, a cough that is not the speaker's, evidence that someone else was there and reacting, and I have no way to recover what that reaction meant. Those testimonies are permanently thinner than the ones where I bothered to write down the room, and the gap cannot be fixed after the fact. It has to be caught while the room is still doing it.

The objection

The obvious reply is that a room track turns the writer into an unreliable narrator of someone else's story, adding stage directions the speaker never authorised and cannot check. That risk is real, and I manage it by keeping the room track strictly observational: who moved, who spoke, in what order, never what I think it meant. I do not write "Paulina looked skeptical." I write "Paulina looked at the floor when he said 1986." The first is my interpretation smuggled in as fact. The second is a fact I can defend if anyone in that room ever reads the page and asks me to.

What this leaves me with is not a cleaner transcript. It is a messier one that tells the truth about how many people were actually inside the account, which for testimonio work is usually the entire point. A single voice on a single tape suggests a single authority remembering alone. Almost nobody remembers alone, and a recorder that cannot hear the rest of the room will let you publish that lie by omission unless you write the room down yourself, while it is still happening, in a column the recorder never touches.