Where were you when you heard about the attacks of September the eleventh, 2001? If you're too young for that one, take whatever news stopped the day in your own life, even a phone call about someone in the family.
You have an answer. It came quickly, I expect, and it came with a room in it: a kitchen, a car, a television, a face. You'd be surprised to hear that any of it was wrong. Keep that memory by you. Before long it's going to meet a questionnaire.
I ask because of where the last lecture stopped. A Belgian woman in her eighties or nineties told researchers about a day in the war, some seventy-five years earlier. She was a child, carrying detonators on her bicycle, and a German sentry stopped her and took hold of it. In her family they say, "She always says there was a big hairy hand on her handlebars." She called the memory very vivid. I asked you whether there was a hand.
I can't settle that directly. Nobody kept a record of the sentry's hands. But the question underneath it is the one you'll face across the table from your own narrator. How much of what an old person remembers can you trust?
The answer comes in pieces. What an old memory has most of, and what brings it out. What happens to the details of a vivid one. What telling a story does to it. What oral historians do when a recorded story turns out to be wrong. And along the way, things for you to do: before the interview, in the room, and afterwards.
Before asking whether to believe what comes back, look at what comes back, because it's nothing like a fair sample of a life.
People remember the last few years best, as you'd expect. The surprise is further back. Past forty, people have more memories from between about ten and thirty than from the years just before or just after. Psychologists call that pile-up the reminiscence bump.
The bump moves with the cue, by which I mean whatever sets a memory off: a word, a question, a smell, a photograph. In one small experiment, people aged seventy-three were given a word and asked for a memory to go with it. One group was told it could come from as far back as they could remember, like a trip to a little country store with their grandfather at the age of five. A second group was told it could be as recent as this morning, like a trip to the hardware store yesterday. The first group's memories piled up between ten and twenty, and the second group's in their twenties. A memory is an answer to something. Change what you ask with, and a different part of the life answers.
What I draw from that is my own, and untested as advice. The teens and the twenties will arrive without much help. Early childhood, and the long middle of a life, may need questions of their own. And watch the example you offer when you ask. It picks the decade.
A photograph is a cue you can carry in your hand, and John Collier looked closely at what one does to an interview. He was a documentary photographer with no scientific training, working with a university research team in the Canadian Maritimes in the 1950s. The team interviewed some people with questions alone. With others they put photographs on the table, pictures Collier had just taken of the places where those people worked and lived.
One man had worked for years in a fish plant, and Collier describes him as silent, with strain in his face. Put a photograph in front of him and he talked. Take it away and he sat and said nothing. A younger man from the same plant was asked, with no pictures, where the people he worked with came from. "Oh, my God," he said, and reeled off a few towns. Shown a photograph of the cannery line, he went along it face by face, with names and villages. At that session his wife put down her copy of True Romance, for the first time, and pulled her chair up to the table.
The pictures didn't get more answers in all. On whatever they showed, they got longer and fuller ones, exact, with names and places. They also drew the whole family round the table, competing like contestants on a quiz show, so what got said was fit for all of them to hear. And they decided the subject. The men interviewed without pictures said more about themselves. It was four households, with brand-new photographs, and Collier called it an exploration. He didn't try the old family album.
That gives you the first job, for before the interview: collect your cues. Photographs, letters and newspaper clippings is the list that Densho, an oral history archive, gives to families. Bring the pictures out one at a time, for names and places, and again when the talk runs dry. Put them away for the stretch where your narrator is to lead.
Back to the memory I asked you to keep.
Psychologists call it a flashbulb memory: your memory of where you were and what you were doing when you heard a piece of shocking news. The name, from 1977, was meant to catch the impression that a person takes something like a photograph of that moment. For years the impression couldn't be checked: there was the vivid memory and nothing to compare it with, which is our position with the sentry's hand.
Ulric Neisser, a psychologist at Emory University, proposed the way round it: take a first account at once, and come back later for a second. The space shuttle Challenger exploded in January of 1986. The next morning Neisser and his co-author handed a questionnaire to an introductory class of 106 students, asking how they'd heard. Two and a half years later, forty-four of those students were found and asked again. Only eleven of them remembered ever filling in the first questionnaire, so the forgetting was under way before the test began.
One of the forty-four is known only by her initials, R.T., and by 1988 she was certain. She'd been in her dorm room with her roommate, watching television, when a news flash came on, and afterward she'd called her parents. What she had written the morning after was different. She'd been in her religion class and heard some students talking about an explosion. There was a schoolteacher aboard the shuttle, and R.T. wrote that the teacher's students had all been watching, "which I thought was sad," as she put it. After class she went to her room and watched the news on television, by herself. There was a television, you'll notice, and there was a dorm room. They've moved.
R.T. had company. On lenient scoring, out of seven, three of the forty-four got full marks, fifteen got one or nothing, and the rest fell between. How sure a student felt was a poor guide to how well the second account matched the first. That match is all a study like this can measure, since it has no way back to the day itself.
Half a year on, forty of the students came in to be interviewed, and Neisser's co-author has described how those interviews ended. Each student was shown the first questionnaire, in his or her own handwriting. Among those whose story had changed, not one said, oh yes, that's right. One read that she'd first heard the news at a travel agency, and said, "I don't even now have a recollection of being at the Travel Agency." Another said, "God, I don't remember having a radio!" A third decided the paper was at fault: he must have filled in that first questionnaire in a rush. Shown his own handwriting, he doubted the handwriting.
That student is the reason for the first rule in the room. When your narrator says it's as clear as yesterday, take that as a true report of how the memory feels, and don't produce a document to prove otherwise. A letter on the table as a cue is one thing. A letter as a correction is another. The Holocaust Memorial Museum tells its interviewers not to argue, as a rule, and not to correct mistakes in the middle of an interview. A question is different. The museum allows one when an answer seems mistaken, and Linda Shopes, the historian from the last lecture, thinks the best interviewers do query a contradiction.
When does the change happen? After the attacks of September the eleventh, a team of psychologists asked people how they'd heard: within days, about a year later, at three years, and at ten. Most of the change came in the first year. After the third there was none. What had changed stayed changed.
One person was asked, each time, for the place and what they'd been doing there. In the first days the answer was "Kitchen, making breakfast." A year later their answer was "In dorm room, folding laundry." At three years they wrote "Ironing in dorm room." And at ten they wrote "In dorm room, ironing." The laundry got ironed some time before the third year, and it has stayed ironed. Through all ten years, people's certainty stayed very high.
How does a first version turn into a second? Neisser and his co-author had a suspicion. The morning after the explosion, about one student in five said the news had reached them by television. Two and a half years on, nearly half had a television in the story. The later memories, they suggested, were real ones: memories of some later time in front of the screen, which had taken the place of the first hearing. That is R.T.'s television exactly.
Elizabeth Phelps is one of the psychologists who lead the September the eleventh study, and I'll put her account of such memories into two words of my own. The center is what happened. R.T. wrote on the first morning that the shuttle had exploded, and she never lost that. The edges are where you were, who was with you, and how you heard. On Phelps's account the center is held better than an ordinary memory and the edges worse, and the strength of the center lends its confidence to the edges. "You just feel you know it better," she says.
There's a fair complaint to make about all this. These are mostly students, remembering where they were when they heard some news. Your grandmother is eighty-five, and what she'll tell you happened to her.
Neisser put that complaint to the test himself. In 1989 an earthquake struck California. He and his colleagues collected accounts from people who had been in it, and from people in Atlanta who had only heard the news, and went back to all of them a year and a half later. The Atlantans did much worse. The Californians' accounts of their own earthquake were almost perfect. Being in it had made the difference, the researchers concluded, and they couldn't say why.
Look at what that does to my two words. If you hear about an earthquake, where you were standing is an edge. If you're in one, where you were standing is the earthquake. What happened to you is the center, and the center holds.
A year and a half is not a lifetime, and for that the nearest test is Danish. In Denmark, people in their seventies and eighties were asked about the day the Germans invaded, some sixty years earlier. Some of the questions could be checked against the record, like what the weather had been. These people were being scored against the day itself, where the students were scored only against their own first account. Of a hundred and forty-five, a hundred had the weather right. Of sixty-five younger Danes, born too late to remember, three did. And there was a lean. The weather on the day of the invasion tends to be remembered as worse than it was, and the weather on the day of liberation as better.
Now, weather is an edge, even of a day you lived through. Going by the students, it should have been gone within the year, and most of these people still had it. The Danish authors make the point themselves: their old people did far better than the short studies would predict. So the two bodies of evidence don't fit, and I can't make them fit. It is one study, of two days that changed life for everyone in Denmark. I'd put the difference this way. The edges of a piece of news you heard are soon gone. The edges of a day you were in can last a lifetime, and still bend toward what the day meant.
So, for the person across the table. A long life comes back unevenly, and in answer to the cue. For news she heard, trust the center, and treat her certainty about the edges as a poor guide. For what happened to her, the evidence is kinder, though the earthquake researchers couldn't say why. Neisser had a guess about his Californians. They had told their earthquake stories again and again.
The family in Belgium said as much in three words. "She always says." The hand on the handlebars reaches us as a line, delivered many times.
If Neisser's guess is right, the telling is part of why she still has the story at all. But telling does a second thing, and the clearest picture of it is an experiment with a story about a chaotic night in a bar. It includes a customer who can't stop winking at a woman because, he explains, he has an eye fungus. People of around seventy read it. Some then retold it as if to amuse a group of friends. Some gave an exact account, of the kind you'd give a policeman. Twenty-five minutes later, everyone was asked for the story exactly as written, and the ones who had told it for laughs remembered less of it, in worse order, than the ones who had told it carefully. It's one experiment, with a made-up story, and it suggests that what you say becomes what you remember.
So a story told to children and grandchildren for seventy-five years has been kept by the telling, and shaped by it, toward whatever holds a room.
The listener has a share in that. In another experiment, people told more to a listener who paid attention than to one who was distracted, and remembered it better afterward.
That makes two more things to do in the room. Let your narrator see you listening: Densho's advice is to be patient, not to interrupt, and to resist displaying how much you know. And let the story come first, told their way. Afterwards, if a year or a name matters, ask for it once, plainly, and make not being sure an easy answer. When the experimenters with the bar story asked for it exactly as written, they also asked people not to guess, and I'd borrow that. You might say: what year was that, as near as you can tell? And if you can't tell, that's fine.
Some of what you record will be wrong all the same, and the most cited case of it in oral history may be the one from Terni, a steel town in Italy.
In 1949 a steelworker named Luigi Trastulli was killed there, at a demonstration against Italy's joining NATO. He was twenty-one. The story is told by Alessandro Portelli, an oral historian who grew up in Terni, and by his account the police fired shots along with the tear gas, and Trastulli was hit. More than twenty years later, Portelli was interviewing people in the town, and many of them were sure that Trastulli had died in 1953, in the clashes over mass layoffs at the steelworks. They had changed the manner of it as well. Some had him crushed by a jeep. Some had him machine-gunned against a wall.
What do you do with interviews like those? Linda Shopes describes two opposite mistakes, which she credits to the historian Michael Frisch. One is to take the story as beyond question, because it's vivid and the people were there. Then you write down 1953. The other is to treat an interview as raw data and nothing more. Then all you can do with these people is mark them wrong, and in a family that mistake sounds like this: Grandpa had the year wrong, so there's no use asking him.
Portelli did neither. He asked why 1953. The layoffs were what had torn the town apart, and a death in that fight made a kind of sense that a death in 1949 did not. The man against the wall, he thought, came out of pictures of martyrdom. The lesson, in Portelli's own sentence, is that "errors, inventions, and myths lead us through and beyond facts to their meanings."
Now put Terni beside R.T.'s television and the Danish weather. The death in Terni was real, and moved. The television was real, and moved. The weather leaned the way the day did. One error to expect from your narrator is a real thing in the wrong place, and the direction it moved may tell you what mattered. The arrangement is mine, and it doesn't cover everything. The jeep and the wall were made, and not moved.
Meaning comes second, though. Checking comes first, and it's the job for afterwards, at your own table. Pull out of the interview what can be checked: names, dates, places. Shopes says narrators often get names and dates wrong. Check each against other people's accounts, and against whatever paper the family has. When they disagree, the paper doesn't win automatically, because a document can be limited or slanted too. Then take the difference back to your narrator, as a question, and write down what you found. The note might say only this: she gives 1953, and the newspaper gives 1949. The correction goes in the note. The recording stays as it was said.
And so back, one last time, to the memory I asked you for. It came when it was called, by a question built to fetch it. It's a piece of news you heard, so the center is sound and the room, the person who told you and the laundry are edges. Being sure of them is poor evidence. You've told it more than once, which has kept it and shaped it. If a piece of it has moved, look at which way.
How much of what an old person remembers can you trust? The center, a good deal. The edges of news she only heard: little, however sure she is. The edges of a day she was in: more than the students would make you fear, and for something like the weather, across sixty years, though bent. Names and dates want checking either way. And all of it has been told before, to somebody, which is why the next lecture asks who is listening.
That leaves the hand, and one gap. The sixty-year test was of the weather, on a day a whole country remembers. Nobody checked this one child's afternoon. But it happened to her, which puts it with the earthquake and not with the shuttle. I'd call the hand the center of that moment, and I'm deciding that after the fact. If I'm right, it has better odds than the date or the road. The words big and hairy have had seventy-five years of being told to children. None of it can be checked. It goes into the recording exactly as she says it.