The public record does not appear to establish through forensic voice identification that Mohamed Atta was the person speaking during the three famous American Airlines Flight 11 radio transmissions on September 11, 2001.
The transmissions themselves are well documented. The 9/11 Commission concluded that they came from Flight 11 and were made by a hijacker who apparently pressed the radio-transmission button instead of, or while attempting to use, the aircraft’s passenger-address system. But in the Commission materials reviewed by sherafy.com, the speaker is variously described as “American 11,” “one of the hijackers,” or the “hijacker pilot”—not as a speaker whose voice had been scientifically matched to Atta.
That distinction has become newly relevant because forensic linguist Natalie Schilling, a professor emerita at Georgetown University, published an expert report in April 2026 arguing that several features of the recorded voice are unexpected for the Arabic-accented English she would anticipate from Atta.
Her analysis is interesting enough to justify scrutiny. It is not strong enough to prove that the speaker was someone else.
That may sound like a narrow distinction. It isn’t.
The real question is no longer simply, “Does this sound like Mohamed Atta?”
It is:
How certain was the identification of the speaker as Atta in the first place—and does the linguistic evidence justify reopening that attribution?
What the Flight 11 recordings actually contain
At approximately 8:24 a.m. on September 11, Boston air traffic controllers began receiving transmissions from American Airlines Flight 11 that were apparently intended for passengers aboard the aircraft.
The first recorded message included:
“We have some planes. Just stay quiet, and you’ll be okay. We are returning to the airport.”
Seconds later came another transmission instructing passengers not to move.
At approximately 8:34, another transmission told passengers not to move and said the aircraft was going back to the airport. The 9/11 Commission reproduced all three transmissions and played recordings during its June 17, 2004 public hearing. In that hearing, the audio was introduced simply as “American 11 (from audiotape).”
The Commission’s final report similarly says that “one of the hijackers” made the announcement heard by air traffic control. Elsewhere, the Commission refers to the person making the transmissions as the “hijacker pilot.”
Those formulations establish something important:
The official investigation clearly attributed the transmissions to the hijacking of Flight 11. It did not, in the passages documenting the recordings, present a forensic voice identification demonstrating that the individual speaking was Mohamed Atta.
That does not mean the Atta attribution was unreasonable.
It means we should distinguish an inference from an identification.
Why everyone assumes the speaker was Mohamed Atta
The reasoning is straightforward.
Atta was the only member of the five-man Flight 11 hijacking team whom the 9/11 Commission identified as having the flight training and FAA qualifications necessary to fly the Boeing 767 after the takeover. The Commission therefore concluded that Atta likely moved from his business-class seat into the cockpit during the hijacking.
Atta had also spent years in Germany, spoke German well, and belonged to the Hamburg group whose members had significant familiarity with the West and English.
If Atta took control of Flight 11, and somebody in the cockpit then made announcements over the radio, Atta is an obvious candidate.
That is strong circumstantial reasoning.
But “Atta was probably flying the aircraft, therefore Atta probably made the announcements” is still not the same proposition as “the recorded voice was compared with a verified sample of Atta’s speech and identified as his.”
The distinction was visible even in contemporary reporting.
When the 9/11 Commission publicly played the recordings in 2004, The Washington Post reported that the voice was believed to be Atta’s, while noting that the Commission itself had not identified the individual speaker in either its hearing or the report released that day.
That is much closer to the evidentiary record than the shorthand that developed later: the recording of Mohamed Atta.
Was the Flight 11 voice ever scientifically matched to Atta?
Sherafy.com could not find a speaker-comparison procedure described in the public 9/11 Commission materials we reviewed.
That does not prove that no government agency ever conducted additional nonpublic analysis. It means that the official materials commonly cited to establish what happened aboard Flight 11 do not themselves document a forensic match between the radio voice and Mohamed Atta.
There is another practical problem: a forensic comparison requires something to compare against.
Schilling states in her report that she was not aware of any authenticated recording of Atta speaking that could serve as a reference sample. She emphasizes that even if one were found, the quality and quantity of speech in the Flight 11 transmissions would make definitive speaker identification difficult.
A previously unknown video of Atta and fellow hijacker Ziad Jarrah surfaced publicly in 2006, showing the men at an al-Qaeda gathering before the attacks. But the surviving footage was reported to have no soundtrack, eliminating what might otherwise have been a valuable comparison sample.
So the attribution appears to rest primarily on circumstances: Atta was the trained pilot, investigators concluded he entered the cockpit, the transmissions came from the cockpit environment, and the speaker had a foreign accent.
Again, that may be a perfectly reasonable conclusion.
It is simply a different kind of conclusion from forensic voice identification.
What the new linguistic analysis found
Natalie Schilling is not simply someone listening to the recording on social media and declaring that the accent “sounds wrong.”
She is a Georgetown University professor emerita whose academic work includes sociolinguistics, dialect variation, phonetics and forensic linguistics. Georgetown says she has worked with law enforcement on forensic-linguistic matters, and the Linguistic Society of America elected her a fellow in 2022.
Her April 2026 analysis examined all three transmissions, totaling approximately 57 words, only 34 of them unique. She concluded that the recordings appear linguistically consistent with a single speaker.
Some characteristics, she says, are entirely compatible with a native Arabic speaker using English.
Others surprised her.
The two most important involve the pronunciation of R sounds and the speaker’s pitch and intonation.
The unusual R sounds
One of Schilling’s strongest observations concerns words such as airport.
Arabic varieties generally use an /r/ that is produced as a tap or trill—the sound English speakers often describe as a “rolled R.” Arabic also ordinarily pronounces /r/ in positions where speakers of non-rhotic English dialects may weaken or eliminate it.
Yet Schilling hears weakening or dropping of /r/ in portions of the Flight 11 recording rather than the conspicuous trill or tap she expected from an Arabic-speaking learner of English.
That is not an invented linguistic distinction.
Independent phonetic research has documented Arabic /r/ as commonly realized through taps and trills and has found that Arabic-English bilingual speakers can carry those realizations into English.
But this is also where a viral version of the argument can become too confident.
Arabic speakers do not all pronounce English exactly the same way. The research documents individual variation, bilingual adaptation and exposure to the surrounding English dialect. Some Arabic-English speakers can acquire approximant or non-rhotic English R patterns.
Therefore:
The absence of a conspicuously rolled R is evidence worth examining. It is not a linguistic fingerprint proving that the speaker could not have been Arabic.
Schilling herself is considerably more cautious than some summaries of her work.
The “sing-song” intonation
The other striking feature comes through particularly clearly in the final transmission.
Schilling describes portions of the speaker’s intonation as having a comparatively wide pitch range and almost “sing-song” quality.
She considers that unusual because second-language English spoken by native Arabic speakers has often been observed to retain prosodic characteristics of Arabic.
There is independent evidence behind this part of the argument.
A 2025 study published in The Journal of the Acoustical Society of America compared native Jordanian Arabic speakers reading both Arabic and English with native British English speakers. The researchers found a significantly narrower pitch span in the Arabic speakers’ English, with their English pitch range closely resembling their Arabic pitch range—a result consistent with first-language transfer.
That gives Schilling’s observation some independent scientific support.
But it does not settle the Flight 11 question.
The study involved only 12 Jordanian Arabic speakers, involved read speech rather than speech during an aircraft hijacking, and examined Jordanian Arabic rather than Mohamed Atta’s Egyptian linguistic background. Emotional state, radio transmission, individual speaking style and years of exposure to other languages can all matter.
The responsible conclusion is therefore not that an Arabic speaker “couldn’t” sound this way.
It is that the feature appears sufficiently unusual to deserve investigation.
Not everything in the recording points away from Arabic English
This is an important part of Schilling’s report that can disappear when the research is reduced to a social-media clip.
She identifies several features that are compatible with Arabic-influenced English, including vowel behavior, stress placement and other pronunciation patterns.
In other words, her analysis is not:
Nothing about this man sounds Arabic.
It is closer to:
The speaker has a mixture of characteristics, several of which fit Arabic-influenced English and several of which do not fit as neatly as expected.
That is a much more interesting linguistic problem—and a much weaker basis for declaring the speaker definitively non-Arabic.
Does the voice actually sound Caribbean?
Schilling argues that the overall combination of features bears similarities to some varieties of Caribbean English.
This is probably the most likely part of the report to be exaggerated online.
The analysis does not identify the speaker as Caribbean.
Schilling explicitly acknowledges that many second-language and regional English varieties share phonetic features. The sample is tiny, the recordings are noisy, and the dialect comparison is not exhaustive. She ultimately says that the observed combination appears more similar to certain Caribbean varieties than to the Arabic English she expected—but she cannot determine the speaker’s native dialect from the recording.
That difference matters enormously.
A linguistic resemblance is not an identification.
Otherwise, the argument simply replaces one unsupported certainty with another.
The recording itself is a difficult forensic sample
There are only about 57 words across all three transmissions.
Only 34 are unique.
The audio traveled through early-2000s aircraft radio equipment and air traffic control recording systems. Background noise is substantial. Some individual words are disputed even at the transcription level.
Those limitations eliminate many of the features a forensic linguist would normally want when comparing speakers.
Schilling specifically warns against thinking of a voice recording as a “voiceprint” comparable to DNA or a fingerprint. Speaker identification is probabilistic and becomes especially difficult with short, noisy recordings and no known reference sample.
That warning should apply equally to both sides.
The recording cannot responsibly be used to say:
- The voice is definitely Atta.
- The voice definitely is not Atta.
- The speaker was definitely Caribbean.
- The recording therefore proves that the transmissions were staged or that some alternative account of 9/11 is true.
What the evidence can support is narrower:
There are identifiable linguistic features in the recording that a qualified linguist considers unusual for the Arabic-influenced English expected from Atta, and the widely repeated identification of the voice as Atta’s does not appear to rest on a publicly documented forensic voice comparison.
That is enough to justify further scrutiny without pretending that the mystery has already been solved.
The source of the new report matters
There is another fact readers deserve to know.
Schilling’s report was prepared for the International Center for 9/11 Justice, or IC911.
IC911 is not a neutral academic institution that happened to become curious about the recording. Its own mission and organizational material make clear that it believes the accepted official account of the September 11 attacks is false and campaigns for a new investigation.
That creates a legitimate potential for selection and confirmation bias.
An organization already convinced that the official account is wrong has an obvious incentive to commission research investigating anomalies in that account.
But the source of funding does not answer the phonetic question.
Schilling’s academic credentials are independently verifiable. The correct approach is therefore neither to accept the report because it challenges the government nor to dismiss it because of who commissioned it.
The appropriate question is:
Can independent linguists reproduce the observations?
IC911 says another forensic linguist reviewed the report before publication. But the document is an expert report published through IC911 rather than a conventional peer-reviewed journal article, and the reviewing linguist is not identified in the report itself. That makes genuinely independent replication particularly valuable.
What is verified, what is inferred, and what remains unresolved?
The evidence is easier to understand when the categories are kept separate.
Verified: Three transmissions were received from American Airlines Flight 11 during the hijacking. Air traffic controllers heard a foreign-accented speaker. The 9/11 Commission regarded the transmissions as coming from a hijacker aboard Flight 11. Atta was the member of the Flight 11 hijacking team with the training necessary to pilot the aircraft.
Reasonable inference: Because Atta was believed to have taken control of the cockpit, he is an obvious and arguably likely candidate for the person making cockpit announcements.
Not established by the public evidence reviewed here: That the voice was forensically compared with Mohamed Atta and scientifically identified as his.
Newly disputed: Whether the linguistic characteristics of the speaker are what specialists should expect from Atta’s Arabic-influenced English.
Not established: That the speaker was Caribbean, that somebody else was piloting Flight 11, that the transmissions were fabricated, or that the accent analysis demonstrates a broader alternative account of September 11.
Keeping those categories separate is what prevents investigation from turning into either conspiracy advocacy or reflexive debunking.
Why the attribution question may be more important than the accent question
The most consequential finding here may not be any individual R sound.
It is how easily an inference can harden into a fact through repetition.
The Commission had good reasons to believe Atta was controlling Flight 11. If a hijacker in the cockpit then made radio transmissions, people naturally connected those two facts.
“Probably Atta” became “Atta’s voice.”
Twenty-five years later, most listeners encountering the recording are given the identity of the speaker before they hear him. That matters because human perception is highly susceptible to expectation. If a recording is introduced as Mohamed Atta speaking, listeners are primed to interpret what they hear within that framework.
Schilling raises essentially this problem in her report.
None of this means the original inference was wrong.
It means that when a specific empirical claim can be tested independently, the inference should not be permitted to substitute permanently for the test.
That principle should apply whether the subject is 9/11, a criminal investigation, an intelligence assessment or any other historically consequential event.
What would actually make the evidence stronger?
The next step should not be another viral video declaring the mystery solved.
The useful work would be comparatively straightforward: obtain the highest-quality surviving versions of the original transmissions; have multiple independent forensic phoneticians analyze them without being told the desired conclusion; compare the relevant features against appropriately matched speakers of Egyptian Arabic rather than treating all Arabic speakers as one accent group; model the effects of radio transmission and extreme stress; and continue searching for authenticated recordings of Atta speaking English, German or Arabic.
If an authenticated Atta recording were located, modern forensic speaker-comparison techniques could then examine whether the Flight 11 voice is consistent with it—while still accounting for the severe limitations of the 57-word sample.
That would move the discussion from competing interpretations toward testable evidence.
So, was Mohamed Atta the voice on the Flight 11 recording?
Possibly, and there are strong circumstantial reasons investigators would have assumed that he was. But the publicly available record reviewed by sherafy.com does not show that the voice was ever forensically established as Mohamed Atta’s.
Natalie Schilling’s 2026 analysis provides a legitimate reason to revisit that assumption. Her observations about R pronunciation and intonation have grounding in established linguistic research, and some characteristics of the recording are indeed unusual relative to common descriptions of Arabic-influenced English.
But the recording is short and noisy. Arabic-accented English varies considerably. No authenticated Atta voice sample is known to have been available for comparison. Some characteristics of the speaker actually do fit Arabic-influenced English. And the new report was commissioned by an organization already committed to challenging the accepted account of 9/11.
Taken together, the evidence does not justify saying that linguistics has proved the Flight 11 speaker was someone other than Atta.
It justifies something considerably more modest—and more intellectually defensible:
The identification of Mohamed Atta as the voice on the Flight 11 transmissions appears less directly established than decades of repetition might lead people to assume. A qualified linguistic analysis has now raised specific, testable questions about that attribution. Those questions deserve independent examination rather than automatic belief or automatic dismissal.
References and Further Reading
Official 9/11 Records
The 9/11 Commission Report — Chapter 1: “We Have Some Planes” — The Commission’s primary narrative of the four hijacked flights, including the Flight 11 radio transmissions, their timing and the description of the speaker as a hijacker.
9/11 Commission Twelfth Public Hearing — June 17, 2004 — Official hearing transcript in which the Flight 11 recordings were publicly played and introduced as audio from “American 11.”
9/11 Commission Staff Monograph on the Four Flights and Civil Aviation Security — Detailed reconstruction of the hijacked flights based on air traffic control, airline, military and investigative records; notes that controllers heard a voice with a foreign accent from Flight 11.
The 9/11 Commission Report — Chapter 5: Al Qaeda Aims at the American Homeland — Background on Mohamed Atta, the Hamburg group, flight training and the operational development of the 9/11 plot.
The 2026 Linguistic Analysis
Natalie Schilling, Ph.D. — “Linguistic Analysis of American Airlines Flight 11 Transmissions Purported to be Made by Mohamed Atta” — The April 2026 expert report analyzing the Flight 11 speaker’s pronunciation, R realization, stress, vowels and pitch. The report also lays out important limitations involving recording quality, sample size and the absence of a known Atta comparison sample.
Georgetown University — Linguistics Faculty Information: Natalie Schilling — Independent institutional source confirming Schilling’s academic background in sociolinguistics, dialect variation and forensic linguistics.
Independent Linguistic Research
The Pitch Range of L2 English Read by Native Speakers of Jordanian Arabic Compared With L1 Speakers of English and Arabic — Journal of the Acoustical Society of America — 2025 experimental research finding a narrower pitch span among native Jordanian Arabic speakers using English, relevant to Schilling’s discussion of first-language influence on English intonation.
Ghazi Khattab — “/r/ Production in English and Arabic Bilingual and Monolingual Speakers” — Phonetic research documenting common Arabic tap/trill R realizations while also demonstrating speaker variation and adaptation in bilingual English, an important qualification to categorical claims about how an Arabic speaker “must” pronounce R.
Contemporary Reporting and Source Context
The Washington Post — “‘We Have Some Planes. Just Stay Quiet…’” — Contemporary coverage of the Commission hearing noting that the Flight 11 voice was believed to belong to Atta while the Commission did not specifically identify the speaker in the material released that day.
International Center for 9/11 Justice — Who We Are — The commissioning organization’s own description of its mission and its explicit position that the accepted official account of 9/11 is false. Included so readers can independently evaluate potential institutional and selection bias surrounding the commissioned report.
Editorial note: This article distinguishes the authenticated existence and official interpretation of the Flight 11 transmissions from the narrower question of individual speaker identification. If additional authenticated Mohamed Atta recordings, government forensic analyses, independent replications or new archival records become public, the evidentiary assessment should be updated accordingly.
