The Only Source You Can Interrogate
Oral history is the one source that answers back. Every other source is fixed — you ask a charter a question and it says what it always said. An interview can be pressed, redirected, asked why, asked what it felt like, asked about the thing no document ever recorded because everyone present already knew it.
And it reaches people the archive structurally excludes. The three filters — recorded, kept, survived — tilt toward the literate and powerful; oral history goes to the people the paper never described. For the history of the twentieth-century working class, of women's work, of migration, of colonial rule from below, it is not a supplement. It is often the only source there is.
Memory Is Reconstructive, Not Retrieval
The obvious objection is that memory is unreliable. The obvious objection is correct, and it is worse than most people think — because memory does not degrade like a photocopy. It reconstructs.
| Effect | What it does |
|---|---|
| Reconstruction | Each recall rebuilds the memory — and re-stores the rebuilt version |
| Retrospective coherence | The past is edited to make sense of how things turned out |
| Source confusion | What you were told, read, or saw on television becomes what you remember |
| Social conformation | Memory reshapes toward what the group agrees happened |
| Present concerns | Today's questions determine what surfaces and how it is framed |
Note what this does to the interview itself: a witness recalling 1943 in 2025 is not opening a file from 1943. They are reconstructing it now, through eighty years of subsequent telling, reading, film, family narrative, and the knowledge of how it all came out — and through the fact that you are sitting there asking. The interviewer is part of the source.
Portelli's Reversal
Alessandro Portelli investigated a worker killed in Terni in 1949 and found that many locals dated the death to 1953, moved it into the middle of mass layoffs, and attached it to a different set of events. Factually, the memories were wrong.
His argument — and it is the founding insight of the field — is that the error is the finding. The community had relocated the death to where it meant the most: into the story of the layoffs, where it became a symbol of what was done to Terni. That is not a defective recollection of 1949; it is accurate testimony about what the killing came to mean, how the community made sense of its own defeat, and what it needed the death to be about.
The formulation: "Oral sources tell us not just what people did, but what they wanted to do, what they believed they were doing, and what they now think they did." A factual error in testimony is a datum, not a disqualification — as long as you are clear which question you are answering with it.
What This Means in Practice
- Never use testimony alone for verifiable facts — dates, numbers, sequences. Cross-check against documents; that is what documents are good at.
- Do use it, essentially alone, for meaning — experience, motivation, atmosphere, what it was like, what people thought they were doing. No document is good at that.
- Record the interview conditions. Who asked, what was asked, when, in whose presence. A widow's account differs with her son in the room.
- Preserve the recording, not just a transcript. Hesitation, laughter, the refusal to answer, the long pause before a name — the transcript discards exactly the parts where the interview is most informative.
The symmetry worth holding on to: documents are strong on the verifiable and weak on the meant; testimony is weak on the verifiable and strong on the meant. Neither is a degraded version of the other, and a historian who dismisses oral history for unreliability has simply misunderstood which question it answers.