Four Labs Played It Backwards. They Heard Four Different Things.
We sent the same 4.1-second file to four forensic phonetics laboratories with no context and identical fees. Two heard nothing. Two heard something, both marked their own confidence low, and neither heard the famous sentence.
LONDON — The claim rests on a sentence. Play a particular four seconds backwards, the story goes, and a voice says something specific about a specific person being dead.
Millions of people have heard it. This newspaper wanted to know what people hear when they have not been told what to hear.
The test
We sent the same 4.1-second reversed file to four forensic phonetics laboratories, in four countries, with no context at all: no artist, no year, no claim, no mention of the other three.
The commissioning letter, identical in every case, asked for a transcription of anything intelligible and specified that NOTHING INTELLIGIBLE was an acceptable and expected answer.
All four were paid the same fee before any result was returned.
The four results
| Laboratory | Transcription returned | Analyst confidence |
|---|---|---|
| Lab A (Netherlands) | "…nothing intelligible. Phonetic sequence consistent with reversed English but not resolvable to lexical items." | n/a |
| Lab B (Canada) | "…miss him miss him…" (marked LOW) | Low |
| Lab C (South Africa) | "…was a mess, wasn't he…" (marked VERY LOW) | Very low |
| Lab D (Japan) | "…nothing intelligible. Two plosives and a nasal; no vowel structure supporting a lexical reading." | n/a |
Two laboratories heard nothing. Two heard something, both marked their own confidence as low or very low, and the two things they heard are not the same thing as each other and are not the famous sentence.
Lab C's analyst added an unprompted note: "If the commissioning party has a candidate transcription in mind, I would ask them not to tell me, and I would treat any subsequent agreement between us as worthless."
Why anybody hears anything
Dr. Ekaterina Nwosu-Hallberg asked for room rather than a sentence, and got it.
Reversed speech is not speech. It is a stream of acoustic events with roughly the right statistics — the right rhythm, the right spectral density, the right rate of transient — and none of the structure. The auditory system is very good at finding words in that, because finding words in degraded signal is most of what it does all day.
The decisive part is priming. Told in advance what a passage says, listeners hear it, at rates approaching 90 per cent. Not told, they hear almost nothing, and what they do hear varies by listener.
"This is the most replicated finding in the whole area," she said. "It is not controversial and it is not new. What is unusual is that somebody bothered to run the unprimed condition on this particular recording, because everybody who cares about this recording has been primed since 1969."
The man who was in the room
Alistair Redgrave-Boakye is 88 and worked those sessions as a second engineer.
He was played the file — forwards, then backwards — and asked what he remembered.
"There's a reversed cymbal in that section," he said. "We did that. That was deliberate and it was a nuisance to do, because you had to physically flip the tape."
Asked about the passage itself: "I have been asked this for fifty years and my answer has not changed and nobody prints it because it is boring. Nobody said anything backwards. Nobody had time. We were in that room for eleven hours and we were trying to finish."
He then said the thing our correspondent wrote down twice.
"I'll tell you what did happen, and nobody ever asks. Somebody put a message in one of those records on purpose, once, as a joke, and it was so obvious that everybody found it in a week and nobody remembers it now. The one they found is the boring one. The one they can't hear is the exciting one. That's people, isn't it."
The six o'clock edition
Three stories a morning, in your inbox before the coffee is made.
Sources & Method
A blind test, designed before the first file was sent. Four laboratories, one 4.1-second file, no context, identical fees, and an instruction to transcribe whatever they heard and to write NOTHING INTELLIGIBLE if that was the answer. We told none of them about the others. We published all four results including the two that are closest to the famous sentence, and we asked a psycholinguist to explain why any of them heard anything at all.
Who we spoke to
- Four forensic phonetics laboratories, Named individually in the story, all four with consent. Commissioned blind transcription; each paid the same fee ($1,900) and given the same 4.1-second file with no context March–May 2026 None was told what the recording was, where it came from, or that three others had it.
- Dr. Ekaterina Nwosu-Hallberg, Psycholinguist, speech perception. Interviewed by video call 2 June 2026 Explains priming in reversed-speech perception and asked to be quoted at length.
- Alistair Redgrave-Boakye, Recording engineer, worked the original sessions. Interviewed in person 19 May 2026 Aged 88. Describes what was actually done in the room.
What we could not confirm
- What, if anything, is on the tape as an intentional artefact. Four laboratories heard four things and no two agree.
- Whether the passage was reversed deliberately at any stage. Our engineer says a reversed cymbal was, and that he does not remember anything else being.
How Others Covered This
The same events, as reported elsewhere on the same day. We list what each outlet had that we did not, as well as what we had that they did not — including where we come off worse. Why we print this.
-
ApertureWhat We Hear When We Are Told What To Hear
A long essay on priming that used our results with permission and credited them.
Had that we did not
A far better account of the psycholinguistics than ours, and an interview with a researcher we could not get.
Left out
The engineer who was in the room, who is 88 and had never previously been asked.
-
They Buriedthis newspaperFour Labs Played It Backwards. They Heard Four Different Things.
Ran the unprimed condition nobody had run, and published all four transcriptions including the two closest to the famous line.
Had that we did not
The blind test and the engineer.
Left out
We paid four laboratories $1,900 each and did not disclose the total ($7,600) in the first edition. It is disclosed now. — V. Ashcombe-Doyle, standards editor
How was this story?
We publish the result, whatever it is. Reader verdicts appear on the front page and in our newsroom metrics.
Readers' Letters 0
Printed at once under the name you give and read by the desk afterwards; anything unfit is removed, with a note saying so, and nothing else is ever deleted — only corrected. Letters that changed something in the story carry a mark saying so, and there are 5 of those across the archive.